Intensive Care Units (ICUs): Clinical Outcomes, Costs, and Decisionmaking

8/14/2019 Intensive Care Units (ICUs): Clinical Outcomes, Costs, and Decisionmaking

http://slidepdf.com/reader/full/intensive-care-units-icus-clinical-outcomes-costs-and-decisionmaking 1/97

Intensive Care Units (ICUs): ClinicalOutcomes, Costs, and Decisionmaking

November 1984

NTIS order #PB85-145928



HEALTH TECHNOLOGY CASE STUDY 28

Intensive Care Units (ICUs)Clinical O utcomes, Costs,

and Decisionm aking

NOVEMBER 1984

This case study was performed as a part of OTA’S Assessment of

Medical Technology and Costs of the Medicare Program

Prepared under contract to OTA by:

Robert A. Berenson, M.D.

1

OTA Case Stud ies are documents containing information on a specific medical tech-nology or area of application that supplements formal OTA assessments. The materialis not normally of as immediate policy interest as that in an OTA Report, nor doesit present options for Congress to consider.

CONGRESS OF THE UNITED STATES

Otlke of Technology Assessment

Washington, D C 20510



Recommended Citation:

Berenson, R. A., Intensiv e Care Units (ICUs): Clinical Out comes, Costs, and Decisionm aki ng(Health Technology Case Study 28), prepared for the Office of Technology Assessment,U.S. Congress, OTA-HCS-28, Washington, DC, November 1984.

Library of Congress Catalog Card Number 84-601138

For sale by the Sup erintendent of Docum entsU.S. Government Printing Office, Washington, D.C. 20402



Preface Intensive Care Units (ICUs): Clinical Outcomes,

Costs, and Decisionmaking, is Case Study 28 inOTA’S Health Technology Case Study Series. Thiscase stud y has been p repared in connection w ithOTA’S project on Medical Technology and Costsof the Medicare Program, requested by the HouseCommittee on Energy and Commerce and its Sub-

committee on Health and the Environment andthe Senate Comm ittee on Finance, Subcomm it-tee on H ealth. A listing of other case stud ies inthe series is includ ed at the end of this preface.

OTA case stud ies are designed to fulfill twofunctions. The primary purpose is to provideOTA with specific information that can be usedin forming general conclusions regarding broaderpolicy issues. The first 19 cases in the H ealth Tech-nology Case Study Series, for example, were con-ducted in conjunction with OTA’S overall projecton The Implications of Cost-Effectiveness Anal-

ysis of Medical Technology. By examining the 19cases as a group and looking for common prob-lems or strengths in the techniques of cost-effec-tiveness or cost-benefit analysis, OTA was ableto better analyze the potential contribution thatthose techniques might make to the managementof medical technology and health care costs andquality.

The second function of the case studies is toprovide useful information on the specific tech-nologies covered. The design and the fund ing lev-els of most of the case studies are such that theyshould be read primarily in the context of the as-

sociated overall OTA projects. Nevertheless, inman y instances, the case studies do rep resent ex-tensive reviews of the literature on the efficacy,safety, and costs of the specific technologies andas such can stand on their own as a useful contri-bution to the field.

Case studies are prep ared in som e instances be-cause they h ave been sp ecifically requ ested bycongressional comm ittees and in others becausethey have been selected th rough an extensive re-view process involving OTA staff and consulta-tions with the congressional staffs, advisory panel

to the associated overall project, the Health Pro-gram Ad visory Comm ittee, and other experts invarious fields. Selection criteria were developedto ensure that case studies provide the following:

q examples of types of technologies by fun c-

q

q

q

q

q

q

q

tion (preventive, diagnostic, therapeutic, andrehabilitative);examples of types of technologies by physicalnature (drugs, devices, and procedures);examples of technologies in different stagesof developm ent and diffusion (new, emerg-ing, and established);

examples from d ifferent ar eas of medicine(e.g., general medical practice, pediatrics,radiology, and surgery);examples add ressing medical problems thatare important because of their high frequen-cy or significant impacts (e.g., cost);examples of technologies with associated highcosts either because of high volume (for low-cost technologies) or high individual costs;examples that could provide information ma-terial relating to the broader policy and meth -odological issues being examined in theparticular overall project; and

examples with sufficient scientific literature.Case studies are either prepared by OTA staff,

commissioned by OTA and performed under con-tract by experts (generally in academia), or writ-ten by OTA staff on the basis of contractors’papers.

OTA subjects each case study to an extensivereview process. Initial drafts of cases are reviewedby OTA staff and by mem bers of the advisorypanel to the associated project. For commissionedcases, comments are provided to authors, alongwith OTA’S suggestions for revisions. Subsequent

drafts are sent by OTA to numerous experts forreview and comment. Each case is seen by at least30 reviewers, and sometimes by 80 or more out-side reviewers. These individu als may be fromrelevant Government agencies, professional so-cieties, consum er and pu blic interest groups, med -ical practice, and academic medicine. Academi-cians such as econom ists, sociologists, decisionanalysts, biologists, and so forth, as appropriate,also review the cases.

Although cases are n ot statem ents of officialOTA position, the review process is designed to

satisfy OTA’S concern with each case study’sscientific quality and objectivity. During the vari-ous stages of the review and revision p rocess,therefore, OTA encourages, and to the extentpossible requires, authors to present balanced in-formation and recognize divergent points of view.

,..111



Health Technology Case Study Seriesa

Case Study Case StudySeries Case study title; author(s); Series Case study title; author(s);number OTA publication number

b

number OTA publication numberb

1

2

3

4

5

6

7

8

9

10

11

12

13

Formal Analysis, Policy Formulation, and End-StageRenal Disease;

Richar d A. Rettig (OTA-BP-H-9(1))C

The Feasibility of Economic Evaluation of Diagnostic Procedures: The Case of CT Scanning;

Judith L. Wagner (OTA-BP-H-9(2))

Screening for Colon Can cer: A TechnologyAssessment;

David M. Eddy (OTA-BP-H-9(3))Cost Effectiveness of Automated MultichannelChemistry An alyzers;

Milton C. Weinstein and Laurie A. Pearlman(OTA-BP-H-9(4))

Periodonta l Disease: Assessing the Effectiveness andCosts of the Keyes Technique;

Richard M Scheffler an d Sheldon Rovin(OTA-BP-H-9(5))

The Cost Effectiveness of Bone Marrow TransplantTherapy and Its Policy Implications;

Stuar t O. Schweitzer an d C. C. Scalzi(OTA-Bp-H-9(6))

Allocating Costs and Benefits in Disease PreventionProgram s: An Application to Cervical CancerScreening;

Bryan R. Luce (Office of Technology Assessment)OTA-BP-H-9(7))

The Cost Effectiveness of Upper GastrointestinalEndoscopy;

Jonathan A. Showstack and Steven A. Schroed er(OTA-Bp-H-9(8))

The Artificial Heart: Cost, Risks, and Benefits;Deborah P. Lubeck and John P. Bunker(OTA-BP-H-9(9))

The Costs and Effectiveness of Neonatal IntensiveCare;

Peter Budetti, Peggy McManus, Nancy Barrand,and Lu Ann Heinen (OTA-BP-H-9(1O))

Benefit and Cost Analysis of Medical Interventions:The Case of Cimetidine and Peptic Ulcer Disease;

Harvey V. Fineberg and Laurie A. Pearlman(OTA-BP-H-9(11))

Assessing Selected Respiratory Therapy Modalities:Trends and Relative Costs in the Washington, D.C.Area;

Richard M. Scheffler and Morgan Delaney(OTA-BP-H-9(12))

Cardiac Radionuclide Imaging and CostEffectiveness;

William B. Stason and Eric Fortess(OTA-BP-H-9(13))

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

Cost Benefit/ Cost Effectiveness of MedicalTechnologies: A Case Study of Orthop edic JointImplants;

Jud ith D. Bentkover and Philip G. Drew(OTA-BP-H-9(14))

Elective Hysterectomy: Costs, Risks, and Benefits;

Carol Korenbrot, An n B. Flood, Michael Higgins,Nora lou Roos, and John P. Bunker(OTA-BP-H-9(15))

The Costs and Effectiveness of Nurse Practitioners;Lauren LeRoy and Sharon Solkowitz(OTA-BP-H-9(16))

Surgery for Breast Cancer;Karen Schachter Weingrod and Duncan N euhauser(OTA-BP-H-9(17))

The Efficacy and Cost Effectiveness of Psychotherapy;

Leonard Saxe (Office of Technology Assessment)(OTA-BP-H-9(18))

d

Assessment of Four Common X-Ray Procedures;Judith L. Wagner (OTA-BP-H-9(19))

e

Mandatory Passive Restraint Systems inAutomobiles: Issues and Evidence;Kenneth E. Warner (OTA-BP-H-15(20))

f

Selected Telecommunications Devices for Hearing-Impaired Persons;

Virginia W. Stern an d Mar tha Ross Redden(OTA-BP-H-16(21))

g

The Effectiveness and Costs of AlcoholismTreatment;

Leonard Saxe, Denise Dougherty, Katharine Esty,and Michelle Fine (OTA-HCS-22)

The Safety, Efficacy, and Cost Effectiveness of Therapeutic Apheresis;

John C. Langenbrunner (Office of TechnologyAssessment) (OTA-HCS-23)

Variation in Length of Hospital Stay: Their

Relationship to Health Outcomes;Mark R, Chassin (OTA-HCS-24)

Technology and Learning Disabilities;Candis Cousins and Leonard Duhl (OTA-HCS-25)

Assistive Devices for Severe Speech Impairments;Judith Randal (Office of Technology Assessment)(OTA-HCS-26)

Nuclear Magnetic Resonance Imaging Technology:A Clinical, Indu strial, and Policy Analysis;

Earl P. Steinberg and Alan Cohen (OTA-HCS-27)

Intensive Care Units (ICUs): Clinical Outcomes,Costs, and Decisionmaking;

Robert A. Berenson (OTA-HCS-28)

available for sale by the Superintendent of Documents, U.S. Government dBackground pavr #3 to The Implications of Cost-Effectiveness Analysis of

Printing Office, Washington, D. C., 20402, and by the National Technical Medical Technology.

Information Service, 5285 Port Royal Road, Springfield, Va., 22161. Call ‘Background Paper #5 to The implications of Cost-Effectiveness Analysis of OTA’S Publishing Office (224-8996) for availability and ordering infor- Medical Technology.IBackground paper #1 to OTA’S May 1982 report Technology and lfandi-ation.

borigina l publication numbers appear in Parenth=s. capped People.‘The first 17 cases in the series were 17 separately issued cases in Background gBackground paper #2 to Technology and Handicapped People.

Paper ,#2: Case Stu dies of Medical Technologies, prepared in conjunction

with OTA’S August 1980 report The Implications of Cost-Effectiveness Anal-

ysis of Medical Technology.

iv



OTA Project Staff for Case Study #28

H. David Bantaland Roger C. Herdman,

2 Assistant Director, 0TA Health and Life Sciences Division

Clyde J. Behney, Health Program Manager

Anne Kesselman Burns, Project Director

Pamela Simerly, Research Assistant

Virginia Cwalina, Administrative Assistant

Beckie I. Erickson,3Secretary/Word Processing Specialist

Brend a Miller, PC Specialist

Jennifer Nelson,3Secretary

Mary Walls,’ Secretary

1Until August 1983.

‘Since December 1983.3Since Janu ary 1984.‘Until January 1984.



Medical Technology and Costs of the Medicare Program Advisory Panel

Stuart Altman, Panel ChairDean , Florence Heller School, Bran deis University

Frank BakerVice PresidentWashington State Hospital Association

Robert Blendon

Senior Vice PresidentThe Robert Wood Johnson Found ation

Jerry CromwellPresidentHealth Economics ResearchChestnut Hill, MA

Karen DavisChair, Departm ent of Health Policy and

ManagementSchool of Hygiene an d Public HealthJohns Hopkins University

Robert DerzonVice PresidentLewin & AssociatesSan Francisco, CA

Howard FrazierDirectorCenter for th e Analysis of Health PracticesHarvard School of Public Health

Clifton GausPresident, Foundation for Health Services

ResearchWashington, DC

Jack HadleyDirectorCenter for H ealth Policy StudiesGeorgetown University

Kate IrelandChair, Board of GovernorsFrontier Nu rsing ServiceWend over, KY

Jud ith LaveProfessorDepartment of Health EconomicsUniversity of Pittsburgh

Mary MarshallDelegateVirginia H ouse of Delegates

Walter McNerney

Professor of Health PolicyJ. L. Kellogg Graduate School of ManagementNorthwestern University

Morton MillerImmediate Past PresidentNational Health CouncilNew York, NY

James MorganExecutive DirectorTruman Medical CenterKansas City, MO

Seymour PerryDeputy DirectorInstitute for Health Policy AnalysisGeorgetown University Medical Center

Robert SigmondDirector, Commu nity Program s for

Affordable Health CareAdv isor on Hosp ital Affairs Blue Cross/ Blue

Shield Association

Anne SomersProfessorDepartment of Environmental and

Community and Family MedicineUMDNJ—Rutgers Medical School

Paul TorrensSchool of Pu blic HealthUniver sity of California, Los Angeles

Keith WeikelGroup Vice PresidentAMIMcLean, VA

v i



Contents—continued

6.

7.

8.

9.10,

Page

Use and Percentage of Hospital Charges Incurred in ICUs and CCUs for MedicareBeneficiaries Discharged From Short-Stay Hosp itals, 1979 . . . . . . . . . . . . . . . . . . . . . . . 26Use and Percentage of Hospital Charges Incurred in ICUs and CCUs for MedicareBeneficiaries Discharged From Short-Stay Hospitals, by Geographic Region, 1979.. 27Use and Percentage of Hospital Charges Incurred in ICUs and CCUs for MedicareBeneficiaries Discharged From Short-Stay Hospitals, by Age, 1980 . . . . . . . . . . . . . . . 27Retrosp ective Ou tcome Stud ies of ICU Care . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35Estimated Number of Special Care Days by Primary Diagnosis Basedon HCFA 20-Percent Sample of Medicare Discharges, 1980. . . . . . . . . . . . . . . . . . . . . . 49



——

Glossary of Acronyms

AHA –APACHE –

CBA –CCC –CEA –

CON –D H H S –

DRG –HCFA –

HMO –ICU –

American Hospital AssociationAcute Physiology and Chronic HealthEvaluation Scalecost-benefit analysiscoronary care unitcost-effectiveness analysis

certificate-of-needU.S. Department of Health andHuman ServicesDiagnosis Related GroupHealth Care Financing Administration(U.S. Department of H ealth andHuman Services)health maintenance organizationintensive care unit

IvLOSNIH

OR

PSRO

SCUTISS

UC R

UR

— intravenous– length of stay– National Institutes of Health (U.S.

Department of Health and HumanServices)

— operating room

— Professional Standards ReviewOrganization— special care unit— Therapeutic Intervention Scoring

System— usual, customary and reasonable

physician charges for paymentpurposes

— utilization reviewIRB – institutional review board

OTA Note

These case studies are authored works commissioned by OTA. Each author

is responsible for the conclusions of specific case studies. These cases are not state-

ments of official OTA position. OTA does not make recommendations or endorse

particular technologies. During the various stages of review and revision, therefore,

OTA encouraged the authors to present balanced information and to recognize

divergent points of view.



1 .

Introduction andExecutive Summary

“The patient’s recovery will be watched not only by nurses but by electric eyestoo. Sensing devices will constantly monitor his heart rate, his temperature, his respi-ration rate, his electrocardiogram, and the blood pressure both in his veins and in hisarteries. The nurses will not rouse the patient early in the morning to poke a glassthermometer between his gums and then spend much of the day checking up on hisand the other patients’ conditions. They will simply push a button at the console oftheir station to get as many readouts as they want. The patient will not have to hopethat if he enters a crisis somebody may spot it.

If any single bodily function or combination of functions deviates beyond the fixedlimits the patient’s Physician has programmed into a computer, lights will flash anda buzzer will sound the-alarm

plete array of equipment will

“Physicians tend to be

Within seconds, nurses, technicians, doctors, and-

be in action at his bedside.”

–Life Magazine, December 2, 1966unun

unimpressed with the published descriptions of units andtheir working. It often seems to them that the assessment of the results is naive, sur-vival being taken as equivalent of a life saved. They suspect that, however expert thehandling of the apparatus, there is often a shallow understanding of the disease andan over-readiness to employ the most dramatic treatment;. . . One is tempted to saythat treatment is often more intense than careful . . .

I believe, therefore, with many of my colleagues, that the attempt to segregateall medical emergencies on a basis of apparatus need will prove to have been an

aberration.”

—Professor A. C. Dornhorst, April 1, 1966



1

Introduction and Executive Summary

INTRODUCTION

Intensive care units (ICUs) exemplify the bestthat Am erican m edicine ha s to offer—teams of dedicated professionals using the latest technol-ogy to save lives that in the past would havealmost surely been lost. Formally developed onlyin the late 1950s, ICUs are p resent in almost 80percent of hospitals in th e United States. They areestimated to consume between 15 and 20 percentof the Nation’s hospital budget, or almost 1 per-cent of the gross national p rodu ct. Yet, despitesuch large expenditures of public and privateresources, there has been remarkably little criticalevaluation of the effectiveness of ICU care by

either the public or the med ical profession.

In recent years, however, there has been grow-ing pu blic and professional awareness of the emo-tional torment su ffered by the pa tients and theirfamilies related to the use of “lifesaving” medicalcare which does not really benefit the patient.Correspondingly, there has been increasing sup -port for the notion that patients have the rightto reject measures that w ill prolong their liveswithout improving their condition.

Along with the increasing public recognition

that there are times when extraordinary m edicalcare should not be employed, three key develop-ments have mad e this an opportun e time to ana-lyze the costs an d benefits of ICU care. First, thePresident’s Commission for the Study of EthicalProblems in Medicine and Biomedical and Behav-ioral Research issued a comp rehensive report inMarch 1983 on the medical, ethical, and legalissues underlying decisions on whether to foregolife-sustaining trea tmen t for seriously ill pa tients(191), The recommendations of the expert com-

mission have d irect bearing on decisionm aking formany ICU patients.

Also in March 1983, a Consensus Development

Conference sponsored by the National Institutes

of Health (NIH) formally evaluated the efficacy

and appropriatenessthe first time (176).

of critical care medicine]

The Conference Reportforex-

amines the evidence for efficacy of critical care.

med icine for various clinical problems an d pro-vides recommendations for organization andadministration of ICUs.

Finally, in April 1983, Congress en acted a pro-spective payment system for Medicare in theSocial Security Act Amend men ts of 1983 (PublicLaw 98-21). This new paym ent system , wh ichbegan to be ph ased in over a 3-year period begin-ning in October 1983, will dramatically alter pay-ment for services provided in ICUs by placing a

limit on the am ount of reimbursem ent availablefor different categories of illnesses. These limitsmay have a significant impact on the servicesavailable for critically ill patien ts.

This case study has two purposes. The first isto present what is currently known about ICUsin terms of the distribution of ICU beds, the costsof maintaining ICUs, the utilization of ICUs, thecharacteristics of ICU patients, and the outcomeof ICU care. There are still imp ortan t gaps in thedata, bu t a su bstantial body of know ledge existsabou t the techn ical aspects of ICU care. The ICU

is examined as a discrete medical technology.The second purpose of the study is to establish

a framework for considering some of the clinical,moral, and legal issues that arise with respect toICU care. The study explores, for example, thefactors un ique to the ICU that som etimes leadph ysicians to continue life-supp ort for patientswh o have minimal hop e of imp roving. It discussesways in w hich p atients can make known theirwishes about foregoing or discontinuing life-support if their condition deteriorates and howphysicians and family members can decide w heth-er to terminate life-sup port w hen the patient isnot capable of making such a decision. It also con-

IThis case stud y d efines both “’intensive care” and “critical care”as care provided in separate hospital units generally known as “in-tensive care units. ” See ch. 2 for a discussion of definitions.

3



4 . Health Case Study 28: intensive Care Units: Costs, Outcome, and Decisionmaking

siders how ICU treatment migh t be rationed inthe futu re if it becomes necessary to d o so.

As is shown in the review of d ata on costs andbenefits of ICU care, the ICU is often an effec-tive, lifesaving technology. However, it is effec-tive at a h igh cost. Ind eed, partially because of its success in many clinical situations, it will not

be easy to simply find and eliminate the “w aste”in ICUs. Changing the economic incentives forprovision of ICU care, as under Medicare’s newhospital payment system, has not m ade it anyeasier for patients, families, and ICU staffs whofrequently face difficult decisions about how ag-gressively to treat individu al patients. Ind eed, asthe case study explores, the new prospective pay-ment system may m ake ICU decisionmaking evenmore difficult and contentious than in the past.

EXECUTIVE SUMMARYThe ICU has been called the h allmark of the

modern hospital but has come into existence onlyover th e last 25 years. Initially, the ICU was anexpansion of the surgical recovery room and wassubsequently an outgrowth of the respiratory careunits made possible by the development of the me-chanical ventilator.

Today, almost 80 percent of short-term generalhospitals have at least one ICU. Overall, 5.9 per-cent of total hospital beds in non-Federal, short-

term commu nity hosp itals in 1982 were beds inICU and coronary care units (CCUS). Beds inother types of special care units, includingpediatric, neonatal, and burn units, add another1 percent to the total complement of special carebeds.

ICU beds are reasonably evenly distributedamong all sizes of hosp itals, regions of the coun-try, and typ es of hospital sponsorship. Over thelast 6 years, the nu mber of ICU beds has risenabouts percent a year, compared to a rise of gen-eral hospital beds of only 1 percent a year. A major rise of ICU beds occurred between 1979 and1981, particularly in hospitals of greater than 500beds. Federal and State policy, particularly cer-tificate-of-need laws an d Medicare reimbursement

The case stud y focuses on ad ult ICUS and notneonatal, burn , or cardiac units. While some of the issues raised h ere are app licable to these otherspecialized care units, these other units generallypresen t different clinical, ethical, and pu blic pol-icy issues. Certainly, all units treat seriously illpatients. However, the moral, ethical, and legal

problems raised by withhold ing care for seriouslyhand icapp ed newborns, for examp le, differ fromthe problems raised by withholding care for anelderly person with a terminal condition. Theissues related to treatment of such infants, whichhas been the center of the recent “Baby Doe” con-troversy, deserve separate attention. Likewise, asthe stud y emp hasizes, coronary care pa tients areclinically different from general intensive carepatients.

policy until 1982, probably contributed to thecontinued expansion of ICU bed s and ICU u tili-zation.

For a number of technical and conceptualreasons, an accurate estimate of the cost of ICUcare is difficult to make. For example, there isdisagreement on whether consideration of ICUcosts should includ e the room and board costs of ICU care only, the room and board an d ancillarycare costs of patients while in the ICU, or the in-

cremental costs of ICUs above that w hich the hos-pital would have to bear in any case for seriouslyill patients. The national average per diem chargein 1982 of an ICU bed was $408 compared to aregular bed per d iem charge of $167, a ratio of about 2.5:1. However, it is likely that the true costratio is closer to 3-3.s:1. In add ition, ICU pat ientsconsume a greater proportion of ancillary serv-ices, particularly laboratory and pharmacy serv-ices, than regu lar floor p atients.

Based on these and other considerations, it is

estimated that the costs of adult ICU and CCU

care—the cost to the hospital patients while they

are in the special care unit—represents about 14to 17 percent of total inpatient, community hos-

pit al costs, or $13 billion to $15 billion in 1982.



Ch. l–Introduction and Executive Summary q 5

Inclusion of the oth er types of specialized an d Fed-eral hospital ICUs would bring the percentage upto abou t 20 percent.

Utilization of ICUs

According to 1979 Medicare data, 18 percent

of Medicare d ischarges included a stay in inten-sive care (including coronary care) in that year.Unfortunately, similar d ata are not av ailable forthe entire population. From reports from individ-ual hospitals, however, certain general utilizationpatterns do emerge (these reports are weightedtowards large and teaching hospitals). The rep-resentation of the elderly in ICUs seems to be thesame or slightly more than in the hospital as awhole. Poor chronic health status, rather thanage, appears to be a p redominant factor limitinguse of ICUs in individual cases in the UnitedStates. In comparison to the Un ited States, ICU

patients in other coun tries have a significantlylower mean age.

There is no accepted classification scheme th atdescribes the clinical char acteristics of ICU pa-tients, largely because ICU patients are a hetero-geneous population who have multiple underlyingmedical problems and who exhibit varying phys-iologic disturbances. ICU patients range fromthose who are in the ICU primarily for monitor-ing for potential disturbances to those who arecritically ill and receive life-supporting treatmentand continuous intensive nursing and physician

care.

Outcomes of Intensive Care

Unfortunately, it is difficult to separ ate the in-tensity of care from the setting in w hich it is pro-vided, and therefore, to know w hether intensivecare would be as effective if provided on the gen-eral hospital floor as in the physically and ad-ministratively separate ICU. Many believe thatrandomized clinical trials of ICUs, at least forunstable patients, are currently un ethical, becauseICU care has become the accepted an d stand ard

mod e of treatment in the Un ited States for mostseverely ill and injured patients.

A recent NIH -sponsored consensus panel foundthat it is impossible to generalize about wh ether

ICU care improves outcome for the v aried ICUpatient p opu lation. The panel felt that ICU in-tervention is u nequ ivocally lifesaving for som econditions, particularly w here there is an acute,reversible disease such as d rug overd ose or major trau ma. There is less certainty abou t the ef-fectiveness of ICU care in other conditions, par-

ticularly in the presence of a severe, debilitatingchronic illness, such as cancer or cirrhosis of theliver. Investigators believe that underlying diseaseis probably the most significant predictor of theoutcome of ICU care, although patient age and

severity of illness are also important.

Recent da ta have emphasized th e inverse rela-tionship betw een the cost of ICU care and sur-vival. At this time, however, there are no acceptedmethods for determining ahead of time which pa-tients will benefit from additional ICU care. Froma nu mber of stu dies, it is clear that the sickest ICU

patients, many of whom do not survive, consumea highly d isproportionate share of ICU charges.Two recent stud ies, for examp le, found that 17and 18 percent, respectively, of the ICU patientpopulation generated half of the ICU charges.Moreover, charges d o not accoun t for the substan -tial cross-subsidization of costs between ICU pa-tients. It is likely, then, that the true proportionof costs consumed by the sickest ICU patients aresubstantially greater than even the charge datasuggest.

At the other end of the ICU patient spectrumare p atients in th e ICU primarily for monitoringof the d evelopment of a life-threatening complica-tion. Some of these patients m ay be able to becared for safely and mor e cost effectively outsideof the ICU, either in intermed iate care un its oron regular medical floors. On the other hand,there may be a population of ICU patients whoare discharged prematurely from ICUs. Researchhas only recently begun to better define which pa-tients should be rou tinely monitored in an ICUand wh ich w ould d o as well or even better if caredfor on other floors in the hosp ital.

Another consideration in deciding whether a

patient shou ld be cared for in the ICU is the realityof adverse effects of ICU care, so-called iatrogenicillness. A list of major iatrogenic complicationsof prolonged ICU care has been identified. Noso-



—.

6 q Health Case Study 28: Intensive Care Units: Costs, out come, and Decisionmak ing

comial infections—i. e., infections that were notpresent or incubating at th e time of hospital ad-mission—and various serious psychological re-actions are particular complications of ICU care.

Payment for ICU Services

To the extent that insurers distinguish ICU care

from other hospital care for purposes of payment,the result has been to rew ard ICU care relativeto care in intermediate level special care units oron gen eral floors of the hospital. For examp le, in1980, Medicare tightened the existing paym entlimits on routine bed costs but n ot on ICU bedcosts—the so-called “section 223 limits. ” Further-more, u tilization review efforts generally have notconsidered the app ropriate level of care within th ehospital.

Medicare’s inpatient hosp ital payment p olicies,however, have now changed dramatically as a re-

sult of the passage of the Social Security ActAmend men ts of 1983 (Public Law 98-21). Und erthe relatively new system, hospitals receive a fixedpayment per d ischarge based on the patient’s prin-cipa l diagn osis. The classification system , wh ichidentifies 467 different clinical conditions calleddiagnosis-related group s (DRGs) appears ill-suitedfor describing certain types of patients cared forin ICUs. DRG payments are based largely on asingle diagnosis. Yet, ICU patients often havemu ltiple serious underlying illnesses. For these pa-tients, designation of a single, principal diagno-sis is likely to be arbitrary, and the resources used

du e to the presence of add itional diagnoses wouldnot be accoun ted for.

In add ition, the DRG scheme does not take se-verity of illness into account. For some d iagno-ses, particularly noncardiac medical conditions,the DRG category does not reflect the use of ICUsfor the more severely ill patients with that prin-cipal diagnosis. For example, only 3.5 percent of the average total hospital stay for Medicare pa-tients with cirrhosis (DRG 202) represent ICUdays. Yet, the sickest patients with cirrhosis areamon g the highest cost ICU patients.

Furthermore, the outlier policy that the HealthCare Financing Administration has implementedpays hospitals less than the marginal costs of car-

ing for the sickest ICU patients. In sh ort, it ap-pears that un der Med icare’s DRG payment sys-tem, the sicker ICU patients w ill be su bstantial

financial “losers” to the hospital.

Decisionmaking in the ICU

The new incentives of the DRG paym ent systemmay conflict with an ICU decisionmaking envi-ronm ent in many h ospitals in w hich the cost of care has been of minor concern in the p ast. In-deed, a number of factors, some of which aresomew hat u nique to the ICU, have led to a d eci-sionmaking process that often has led physiciansto provide life-support care in the ICU after theinitial rationale for doing so no longer exists. Fac-tors that have created an ICU treatment im-perative includ e:

q

q

q

q

q

q

q

The highly technological nature of ICU care,wh ich often resu lts in focus on the techn ical

details of treatment rather than the rationalefor continued treatment.The nature of ICU illnesses, which often re-quire “technologically oriented” treatmenteven when the primary intent is to providecomfort rather th an cure to a desperately illpatient.Traditional moral d istinctions in med icinethat in some cases result in more care thanthe patient would choose if able to do so.Diffusion of decisionmaking responsibility,especially in relation to decisions to foregoor terminate life-support.

Problems of informed consent in the ICUwhere many patients are temporarily or per-manently incompetent.The practice of defensive medicine by physi-cians, wh ich involves taking or n ot takingcertain actions more as a defense against po-tential legal actions than for the p atient’s ben-efit. Defensive med icine may be a par ticularproblem in the ICU, because of the life-and-death natu re of ICU care, the relative visi-bility of ICU decisions, and great uncertaintyabout likely court decisions on these kindsof cases.

A p ayment environment wh ich, un til 1982,provided financial rewards to hosp itals andph ysicians for p rovision of ICU care. Physi-



Ch. I—Introduction and Executive Summa ry q 7

cian paym ent method s continue to pay gen-erously for the procedure-oriented ICU care.

qThe absence of a data base for the comm onICU cond itions on w hich to make reliableclinical predictions of individual ICU pa-tients’ chances of immediate and long-termsurvival.

Foregoing Life=Sustaining Treatment

The Critical Care Consensus DevelopmentConference sponsored by NIH has concluded thatit is not appropriate to devote limited ICUresources to patients without reasonable prospectof significant recovery or to simply prolong thenatu ral process of death .

In general, a terminally ill patient’s right toforego or discontinue life-sustaining treatment hasbeen established and is usually protected by theconstitutional right to privacy. Practical dif-ficulties arise when the patient is not competentto decide, and when other decisionmakers, in-cluding physicians, families; and patient sur-rogates, do not agree on w hat medical treatmentto pu rsue. State courts have d iffered on the d eci-sionmaking procedures to use when a patient isnot able to choose for himself.

Recent court decisions differ even over whena patient is considered “terminal” and over whatconstitutes “medical” treatment. Likewise, manycourts have continued to invoke a d istinction be-tween ordinary and extraordinary care, while

some have explicitly rejected the distinction.

Possible Future Steps

Because of the increasing burden of medical carecosts on ind ividuals and on society as a whole,it is likely that the fun ds available for intensivecare will be much m ore strictly limited than inthe past. Because Medicare’s DRG payment sys-tem in general makes m any ICU Medicare patientsfinan cial losers for the hosp ital it may, therefore,alter the prevailing provider attitudes about theappropriateness and extent of ICU care in indi-

vidual situations.

In recent years, the number of ICU beds h asexpanded to meet increased demand for beds, ex-cept in public hospitals in financial distress or at

times when there was a shortage of ICU nursesto staff available beds. In the futu re, there w illneed to be greater attention paid to how to ra-tion ICU beds. The DRG system used by Medicareis a form of “implicit” rationing, because the pay-ment limitations place greater pressures on physi-cians and hospitals to make resource allocation

choices without setting “explicit” limitations onservices or eligible patients. Under this form of rationing, there will be a need to consider expand -ing the procedur al safeguard s used on behalf of patients wh o become m ajor financial losers forthe h ospital. ICU d ecisionm aking w ill becomeeven more d ifficult than it has been in the p astdu e to potential financial conflict between p a-tients, physicians, and hospitals.

A number of steps might improve the environ-ment for intensive care decisionmaking:

Research on developing accurate p redictorsof survival for patients with acute andchronic illnesses could be expanded in orderto permit better informed decisions based onthe likelihood of short- and long-term sur-vival. In the absence of valid and reliabledata, hosp itals could consider formalizing aninstitutional prognosis committee whosefunction would be to advise physicians, fam-ilies, and p atients on the likelihood of sur-vival with ICU care.

The suitability of the cur rent DRG methodof payment for ICUs shou ld be tested and

modified if necessary to take sufficient ac-count of severity of illness.

The legal system may need to recognize thepossible conflict between malpractice stand-ards which assume quality of care that meetsnational expert criteria, and a d ecisionmak-ing environment in w hich resources may beseverely limited.

Health professionals who are involved in de-cisionmaking on critically ill patients mightbenefit from more education in med ical ethicsand relevant legal procedures and obli-gations.

The actual decisionmaking process for criti-cally ill patients may need greater attention.For example, hospitals might explore formal-izing decisionmaking committees to lessen the



8 . Health Case Study 28 : ]ntensive Care Units : Costs, Outcome, and Decisionmaking

burd en on ind ividuals faced w ith difficult through formal hospital committees, throughchoices about terminating life-support. More government-imposed procedures which cangenerally, society will need to decide how it follow fixed rules and regulations, or other,wishes conflicts over decisions on terminating perhaps more decentralized, mechanisms.life-support to be resolved—i.e., in courts,



2.

Evolution, Distribution, andRegulation of Intensive Care Units



2 s

Evolution, Distribution, and

Regulation of Intensive Care Units

THE DEVELOPMENT OF THE ICU

The intensive care unit (ICU) has been calledthe hallmark of the modern hospital (205), yet itis a recent d evelopment, having come into ex-istence only in the last 25 years. The developmentof ICUs was preceded by the rapid grow th of post-operative recovery rooms (115) following WorldWar II. As early as 1863, however, FlorenceNightingale had foreseen the u tility of a separatearea for observing patients recovering from theimmediate effects of surgery (172).

To a large extent, the initial stimulu s for a sep-arate recovery area for specialized care was amanagerial response to overwhelming medicaldemands. The Massachusetts General Hospital,for examp le, wh en sud denly faced w ith treating39 survivors of the Boston Coconut Grove Firein 1942, set up a makeshift “burn unit” which itmaintained for 15 days, until the majority of pa-tients had been sent home (115). In the NorthAfrican and Italian camp aigns of World War II,shock ward s w ere established to resuscitate bat-tlefield casualties and to care for injured sold iersbefore and after surgery (115). After the war, an

acute shortage of nur ses provided much of the im-petus for the spread of recovery rooms in theUnited States.

Although recovery rooms were established ini-tially as a means of managing large numbers of patients more efficiently, the medical benefits of better postoperative nursing care soon becameapparent, and recovery rooms flourished. In 1951,only 21 percent of community hospitals had re-covery rooms; a decade later, virtually all hospi-tals had them (205).

During the 1950s, using the recovery room as

a model, a few ICUs began appearing on bothsides of the Atlantic. An early version of w hathas become known as a respiratory ICU, for ex-ample, was set up in Denmark d uring the 1952polio epid emic in Scandinavia. After 27 of 31 pa -tients suffering from respiratory or pharyngeal

paralysis at Copenhagen’s Blegdam H ospital died,the hospital’s senior anesthetist performed a tra-cheotomy on a 12-year-old girl and inserted acuffed endotracheal tube. The patient underwentprolonged manual ventilation and survived.

With th is new lifesaving, if laborious, technol-ogy in han d, a separate area to care for polio vic-tims was established in the hospital. “At an earlystage the following measu res were ad opted : 1) pa-tients who w ere likely to develop respiratory com-

plications were transferred to special wards forobservation and recording vital signs, etc.; 2)tracheotomies were done under general anesthe-sia and cuffed tubes w ere used; 3) man ual, inter-mittent positive-pressure ventilation was used in-stead of or to supplement respirators; and 4)secondary shock was treated” (121).

In add ition, the hospital developed an elaboratepersonnel system, involving anesthetists, epidemi-ologists, nurses, medical students, and hospitalworkers, to provide continu ous care for patientsand to maintain the machinery being used. As a

result of these measu res, the mortality rate forpolio victims w as reduced from 87 to 40 percent.

With the exception of Danish experience, ICUs,like recovery r ooms, w ere established initiallymore for managerial than for med ical reasons. Amajor factor in th eir early d evelopm ent w as theneed to relieve nurses who were so busy caringfor a few critically sick p atients that they w ereneglecting the remaining patients on the wards(30). In addition, ICUs were even seen as a meansof reducing the cost of medical care (115).

By the late 1950s, the rapid development of the

mechanical ventilator provided the medical ra-tionale for establishing ICUs. This life-supportingtechnology needed to be monitored too closelyto be dispersed throughout the hospital (136,200).

In a nu mber of hospitals, the general ICU was adirect outgrow th of a respiratory ICU set up to

11



——.

12 q Health Case Study 28: ]ntensive Care Units : Costs, Outcome, and Decisionmaking

care for patients suffering respiratory paralysis ing an ICU. By the last half of the 1960s, mostcaused by polio (36) or tetanus (155). U.S. hospitals had established at least one ICU

In 1958, only abou t 25 percent of community(205).

hospitals with more than 300 beds reported hav-

ADVANTAGES AND DISADVANTAGES

Early advocates of ICUs identified a number of advan tages for establishing a separate intensivecare unit (frequently called an “intensive therapyunit” in England and Europe) (25,47,178,208,231):

q

q

q

q

q

q

q

maintenance of high standar ds of care forseriously ill patients by using specially trainedphysicians and nurses;prov ision of more continuous observationand frequent measurements of relevant in-dicators of clinical condition;

concentration of technologies in one locationto avoid duplication of equipment and per-sonnel;direct access to p atients for major procedu resand therapies, including resuscitation;avoidance of upsetting the regular ward rou-tine and disturbing less ill ward patients;fostering h igh staff morale and team w ork;andopportunities for concentrated education and

.research.

From the outset, there was disagreement on

which p atients wou ld benefit from ICU care. Earlyunits attemp ted to exclud e “terminal care cases,chronic cases, and disturbed or disturbing pa-tients” (23). Some emphasized that intensive ther-apy should be provided to support vital functionsuntil the underlying disease process could be cor-rected or r un its course (200). Other early com-mentators saw the ICU simply as the place for the“critically ill” (187), or advocated the use of theICU as a last resort for a “final desperate attemp t”to save a life (36). Lack of agreem ent p ersists onwh ich p atients should h ave priority access to ICUcare.

While the advan tages of the ICU were recog-nized early, so were the potential disadvantages(25,64,178):

OF ICU CARE

a noisy, intrusive environment for seriouslyill patients;interrupted continuity of m edical responsi-bility;mental and physical strain on the ICU staff;overenergetic treatment—for both hopelessand less serious cases;decreased n ursing skills on the general wardsas the sickest patients are removed;poten tial for high cost with u nfair claims onthe hospital bud get; andincreased cross-infections among seriously illpatients in the same area.

Stated another way, in some situations, applica-tion of intensive care maybe unnecessary becausethe condition is not serious enough; unsuccessfulbecause the condition is too far advanced; unsafebecause the risk of complications is too great; un-sound because it serves no useful purpose for thepatient; or un wise because it ut ilizes too manyresources (125).

Despite recognized patient care problems and,

more recently, cost concerns, ICU beds have con-tinued to p roliferate. There is substantial evidencethat, at least for some types of patients, care pro-vided in ICUs is extremely effective. For manymedical problems, care of patients outside an ICUwould be unthinkable to the modem clinician. Atthe same time, it is remarkable that such an all-pervasive and cost-generating innovation has de-veloped p rimarily because of “a p riori” considera-tions, with few critical evaluations of its effective-ness (198). The grow th of ICUs has been fosteredby a highly favorable reimbursement system (6o),by the d evelopm ent of professional med ical andnursing critical care societies which constitute astrong constituency for continued expansion of ICUs (166), and by Federal policies which either



.

14 • Health Case Study 28 : Intensive Care Units: Costs, Outcome, and Decisionmaking

REQUIREMENTS OF AN ICU

A detailed consideration of the design, orga-nization, staffing levels, skills, personnel policies,and other components of an ICU is beyond thescope of this stud y. Yet in general, all intensivecare units meet these requirements:

care for severely ill or potentially severely illpatients;employ specially trained registered nu rses ona one-nur se to one- to three-patient basis;identify a ph ysician as the director of patientcare and adm inistrator of the un it;have 24-hour acute care laboratory sup port;andprovide a w ide range of technological serv-ices, with the help

-of expert medical sub-

specialists and ancillary person nel (51,166).

While the availability of physicians in ICUs varieswith the size and type of hospital, all ICUs com-bine intensive nu rsing care and constant patientmonitoring (116). In comm un ity hosp itals, theICU medical director is frequently not full-timeand shares patient care responsibilities with otherstaff ph ysicians w ho also h ave m ajor non-ICUresponsibilities. In these units, day-to-day man-

agement and administrative decisions are madeby the head nurse of the ICU (283). Large hospi-tal ICUs tend to have full-time medical directors.

The NIH Consensus Panel has identified the

minimal technological capabilities that an ICUshould provide, regardless of the type of facilityin w hich it is located (176):

A. card iopulm onary resuscitation;B. airway management, including endotracheal

incubation and assisted ventilation;C. oxygen d elivery systems an d qualified res-

piratory therapists or registered nurses todeliver oxygen therapy;

D. continual electrocardiographic monitoring;E. emergency temporary cardiac pacing;F. access to r apid and comprehensive, speci-

fied laboratory services;G. nutritional support services;H. titrated therapeutic interventions w ith in-

fusion pumps;I. additional specialized technological capa-

bility based on the particular ICU patientcomposition; and

J. portable life-sup port equipm ent for u se inpatient transport.

SPECIALTY V. MULTISPECIALTY ICUs

Since their development two decades ago, hos-pitals have differed on wh ether to establish oneor more mu ltispecialty ICUs to treat the range of seriously ill med ical and surgical patients or toset up separate ICUs for patients with similarproblems (208). For reasons of efficiency andeconomy, smaller hospitals generally have a com-bined medical and surgical ICU. The smallest hos-pitals also combine coronary care with intensivecare in a single unit (4)0

Larger hospitals, particularly teaching hospi-tals, often have separate general medical and sur-

gical units as well as separate subspecialty unitsfor specific types of medical problems, e.g., car-diac surgery and respiratory care. The Massachu-setts General Hospital, for example, has nine sep-

arate subspecialty ICUs (248). However, evenhospitals of similar size and type have adopteddifferent ap proaches to the issue of mu ltispecialtyv. separate specialty ICUs (136).

The major ra tionale for mu ltispecialty ICUS isa medical one, namely, that regardless of theund erlying disease, many life-threatening p hysi-ological disturbances are quite similar in seriouslyill patients (43,208,265). Thus, a basic purpose of ICU care is to support general physiologic re-sponses to stress in ord er to provide time for aspecific therapy for the underlying illness to take

effect (89,116,199,222). At times, ICUs primar-ily treat physiologic disturbances, not diseases;they save lives primarily by supporting oxygena-tion, often with respirators (209), and by prevent-



Ch. 2–Evolut ion, Distribution, and Regulation of intensive Care Units q 15

ing circulatory collapse and shock (222). Sincephysiologic complications are similar regardless

of precipitating factors, there is a strong medicalrationale for multispecialty intensive care pro-vided by comprehensively trained generalists (8).

Increasingly, concerns about efficiency and ris-ing costs have supported maintaining multispe-cialty units rather than separate subspecialty un its.With mu ltispecialty units, there maybe less du pli-cation of expensive equipm ent, although ICUsgenerally d o not utilize “big ticket” technologies(6). More importantly, because of highly variableclinical demands for ICU care, ICU occupancy canvary d ram atically, and combining medical andsurgical specialty an d subspecialty u nits perm itsgreater efficiency in the use of personnel, particu-larly nurses, which is a major cost factor in ICUs(212).

Traditionally, however, demand for ICUs hasdeveloped along subspecialty lines, usually in re-sponse to the availability of new medical technol-ogy. For example, the m echanical respirator ledto the respiratory ICU, and the advent of cor-onary artery bypass surgery led to the p ostcar-diac surgery ICU. In add ition, specialists often feelthat physicians trained in other fields d o not havesufficient u nd erstanding and skill to care for pa-tients with particular “subspecialty” p roblems. In-deed, some have advocated a separate surgical

DISTRIBUTION OF ICU BEDS

It is difficult to estimate precisely the numberof ICUs and ICU beds in this country because of the ways in which hospitals report their bed ca-pacity. This is particularly a problem with smallerhospitals, which may designate their ICUs asCCUs or mixed ICU/ CCUs in the ann ual Ameri-can Hosp ital Association (AHA) survey. In ad -dition, the annual AHA survey includes multipleICUs reported from single hosp itals. From 1981AHA su rvey tapes, it can be estimated that 78 per-cent of short-term general hospitals have at leastone ICU or CCU, and that 93 percent of hospi-tals larger than 200 beds have a separate ICU(106). Overa ll, in 1982, 5.9 percen t of the total

ICU for each surgical specialty in a large hospi-tal (81). Others feel that nursing personnel skilledin one subspecialty, such as card iology, may beunsuited by temperament, motivation, and train-ing for work in other subspecialties (147).

In short, the debate over the desirability of generalists v. specialists which exists in m edicine

generally is also being waged in the intensive careworld. The trend, which is supported by theSociety for Critical Care Medicine, is to crosstraditional departmental and specialty lines andto create a “m ultidisciplinary specialty” equallyskilled at caring for med ical and surgical prob-lems (95,274). An attem pt to d efine th e boun d-aries of critical care medicine by examination andprescribed training has recently been developedby the American Board of Medical Specialties (8).In 1980, the Boards of Internal Medicine, Pedi-atrics, Anesthesiology, and Surgery joined to-gether to offer a certificate of special competencein critical care med icine (95). This examina tionhas yet to be given. In 1982, some 50 fellowshipprog rams in critical care med icine in the Un itedStates were training approximately 150 physiciansto become critical care generalists (91,92). Another36 program s were training fellows in p ediatriccritical care m edicine. Despite the n ew cadre of critical care generalists, however, many hospitalscontinue to maintain separate specialty and su b-specialty ICUs along departmental lines.

hospital beds in non-Federal, short-term commu-nity hosp itals were ICU and CCU beds. This fig-ure d oes not include pediatric ICU beds, neonatalbeds, or burn care bed s, wh ich ad d an other 0.2percent, 0.7 percent and 0.1 percent, respectively,to the total nu mber ICU beds (4).

Table 1 shows the distribution of reported ICUbeds by size of hospital. In general, ICU beds ar efairly evenly d istributed across all sizes of hospi-tals. In 1982, for examp le, hosp itals larger th an500 beds, which account for 22.6 percent of totalshort-term general hospital beds (4), have 24.8percent of reported ICU beds. Table 2 show s the



-——

16 q Health Case Study 28: intensive Care Units: Costs, Outcome, and Decisionmaking

Table 1 .- Distribution of ICU Beds in Short-Term, Non-Federal Hospitals, by Size of Hospital, 1982

Hospital bed size Total hospital beds Percent of total Total ICU/CCU beds Percent of total

<100 . . . . . . . . . . . . . . . . . . . . . 146,706 14.5 5,889 9.9100-199 . . . . . . . . . . . . . . . . . . . 195,425 19.3 10,677 17.9200-299 . . . . . . . . . . . . . . . . . . . 179,312 17.7 11,302 18.9300-399 . . . . . . . . . . . . . . . . . . . 144,012 14.2 9,312 15.6400-499 . . . . . . . . . . . . . . . . . . . 120,682 11.9 7,692 12.9>500 . . . . . . . . . . . . . . . . . . . . . 229,043 22.6 14,826 24.8

Total . . . . . . . . . . . . . . . . . . . 1,015,180 99.3 59,698 100.0

SOURCE: American Hospital Association, Hospital Statistics, 1983 edition.

Table 2.-lCU/CCU Beds as Percent of Total Bedsby Hospital Size for Short-Term Nonfederal

Hospitals, 1982

Hospital bed size Percent ICU/CCU beds

<1oo . . . . . . . . . . . . . . . . . . . . . . . . 4.0100-199 . . . . . . . . . . . . . . . . . . . . . . 5.5200-299 . . . . . . . . . . . . . . . . . . . . . . 6.3300-399 . . . . . . . . . . . . . . . . . . . . . . 6.5400-499 . . . . . . . . . . . . . . . . . . . . . .>500 . . . . . . . . . . . . . . . . . . . . . . . . : : :

Total . . . . . . . . . . . . . . . . . . . . . . 5.9SOURCE: American Hospital Association, Hospital Statistis, 1983 edition,

percent of ICU/ CCU beds as a percentage of totalbeds by hosp ital size in 1982. For h ospitalsof200beds or m ore, the ICU/ CCU bed percentage isvery consistent.

Table 3 indicates the distribution of combined,non-Federal intensive and coronary care bedsbyregion as of 1981. (Coronary care beds makeup

about 25 percent of the total.) There are somevariations in the number of these beds as a per-cent of total beds, with the Pacific, East NorthCentral and Mou ntain States having the highestpercentages. However, as Russell pointed out, thedistribution of ICU/ CCU beds is much more uni-form wh en considered in relation to pop ulation,rather than to h ospital beds (205).

Finally, as shown in table 4, the distributionof ICU beds varies somewhat according to hos-pital sponsorship.

EXPANSION OF ICU BEDSWhile the number of community hospital beds

increased only about 6 percent between 1976 and1982, reported ICU and CCU beds in community

Table 3.–Distribution of ICU and CCU Beds,by Region, 1981

Per 10,000 Per 100Region population hospital beds

New England . . . . . . . . . . . . . . 5.8Middle Atlantic . . . . . . . . . . . . ; : :South Atlantic . . . . . . . . . . . . . 2.9 : : :East North Central . . . . . . . . . 3.3 6.7East South Central . . . . . . . . . 5.3West North Central . . . . . . . . ; : ; 5.0West South Central . . . . . . . . 2.6 5.2

Mountain . . . . . . . . . . . . . . . . . 2.6 6.2Pacific . . . . . . . . . . . . . . . . . . . 2.7 7.0

Total . . . . . . . . . . . . . . . . . . . 2.9 5.9a

aHospital data in this table includes Federal hospitals and specialty service short-

term hospitals.

SOURCE: American Hospital Association, Hospital Statistics, 1982 edition; andU.S. Department of Commerce, Bureau of the Census, State and Metropolita Area Data Book, 1982,

Table 4.—Percentage of ICU/CCU Beds in Short-TermHospitals, by Hospital Sponsorship, 1976 and 1982

Percent of hospital

beds that are ICUor CCU beds

hospitals increased by 29 percent, or an averageof almost 5 percent a year. Moreover, over h alf of that rep orted increase occurred between 1979



Ch. 2–Evolut ion, Distribution, and R egulation of Int ensive Care Units q 17

to 1981. In this 2-year span, reported ICU bedsincreased 14.3 percent and reported CCU bedsgrew 15.4 percent (4), despite the absence of anydramatic medical breakthroughs that would ex-plain such a sharp rise. While the nu mber of cor-onary artery bypass graft surgery procedures per-formed in the country was increasing by perhaps

20 percent a year d uring these years (257), the in-crease in the num ber of such operations could ex-plain only a very small increase in ICU beds.

One can speculate, therefore, that the Medicarepolicy implemented in 1980 (73) that tightenedlimits on routine bed charges—commonly knownas the “section 223 limits” —but not on special care

REGULATION OF ICUs

Along with the medical and organizational rea-sons for their expan sion, ICUs and CCUs wereencouraged by the Federal Government in the1960s initially in th e Regional Med ical Programs(205).

In the 1970s, State certificate-of-need (CON)statutes were passed in most States. CON statutesrequire a prior determination by a governmentalagency that certain major capital expenditures orchanges in health care facilities are needed (19).Early evaluations showed that CON programshelped forestall the add ition of general hosp italand long-term care beds (19). However, ICU beds

have generally been approved by CON agencies.

In ad dition, Salkever and Bice (211) foun d thatwhile CON programs controlled expansion in bedsupply to some extent, they stimu lated other typesof hospital investment. Specifically, they foundthat assets per hospital bed, for equipment andother nonlabor p rodu cts, actually increased as aresult of CON. A subsequent, more definitivestudy confirmed the findings that the CON re-quirement generally has been successful in limitingthe nu mber of beds, but not th e intensity of re-source use or costs (188). Ironically, the threat of

CON review may have encouraged hospitals to

bed charges or ancillary services, created a strongstimulus for hospitals to add more ICU beds (60)or, perhap s, to reassign beds to special care wherepossible. The most dram atic rise in ICU/ CCUbeds between 1979 and 1981 occurred in hospi-tals with more than 500 beds, which accountedfor almost 55 percent of the total increase in ICU/

CCU beds in these two years (4). In 1982, thenu mber of ICU/ CCU beds increased 4 percent,wh ile total comm un ity hospital beds increasedonly 1 percent. Thu s, while ICU bed expansionhas continued at a mu ch faster rate than hospitalbeds gen erally, the pace of grow th found in 1980and 1981 has slowed. ,

convert low-asset rou tine care beds into comp ara-tively h igh-asset ICU bed s (166).

Equipment u sed in ICUs rarely requires CONapproval. The national threshold for requiringCON approval in the National Health Planningand Resources Developm ent Act of 1974 (PublicLaw 93-641) was $150,000, and most ICU equip-ment is well below that level. The cost per bedof typical ICU cardiac monitoring equipment in1978, for example, ranged from $6,000 to $8,500(6). A new ICU respira tor costs betw een $10,000to $15,000 (87).

The construction costs of each patient unit inthe ICU was estimated to cost between $44,oooan d $75,000 in 1978 dollars (6), Renovation costswere m uch less. Thus, hospitals with sufficientcapital can escape CON review altogether bygradually expanding and upgrad ing already ex-isting ICUs (119,166). As was noted earlier, hos-pitals reported about a 15-percent increase in ICUbeds between 1979 and 1981, a time when CONprograms were functioning in virtually everyState. The current trend toward raising CONthresholds practically assures that CON regula-tion of ICUs w ill remain a minor issue.



3s

Cost of ICU Care



3.Cost of ICU Care

COMPONENTS OF ICU COSTS

The cost of intensive care units (ICUs) can be

divided into the direct costs of operating the ICUand the ind irect costs for central services that areallocated to the ICU (6). Sanders estimates (212)that for Massachu setts General Hospital in Bos-ton about 65 percent of ICU costs (for labor,equipment, etc. ) are direct, and that about 35 per-cent of costs (for h ospital overhead, h ousekeep-ing, etc. ) are indirect.

Direct costs include fixed costs and variablecosts. Fixed costs exist no matter h ow many pa-tients are treated in the ICU and include dep recia-tion for th e cost of construction, renovation, an d

COSTS OF AN ICU DAY

It has become increasingly clear that hospitalcharges do not represent the true costs of provid-ing hospital services (80). Generally, charges aregreater than operating costs, in order to pay forbad d ebts, to support n onreimbursable educa-tional and preventive health programs, and to payfor costs d isallowed by cost-based insurers, in-clud ing many Blue Cross plans, Medicaid, andMedicare (80). For examp le, by analyzing cost andbilling data , the Health Care Financing Ad minis-tration has calculated the national ratio of al-lowable Medicare inpatient operating costs toMedicare inpatient charges at 0.72 (74).

ICUs are d ifferent from m ost hosp ital services(including generaI room and board’), however,in that charges for ICU room and board are oftenset below cost (6,212,240). In on e d etailed econo-metric analysis, ICU charges for room and boardin one hospital were found to be only slightlymore than half of calculated costs (109). ICU datafrom U.S. hospitals consist mostly of room andboard charge data, unadjusted for actual cost. The

1Overall, room and board charges make up slightly less than half

of total hospital inpatient charges; the rest is made up of variouscategories of ancillary services.

equipment, as well as equipm ent maintenance (6).

Variable costs are depend ent on the volum e of services pr ovided. Some va riable costs, such aspersonn el costs, are fixed ov er a specific range inpatient volume, but change when the patient vol-ume exceeds the ran ge. Other var iable costs, suchas nondu rable equipment and oxygen, are de-pend ent d irectly on total patient d ays (6). Datafrom both foreign and domestic ICUs indicate that

50 to 80 percent of direct costs are variable per-sonnel costs, pr imar ily for nur sing (42,101,155,212). On av erage, ICUs use almost th ree times asmany nu rsing hours per p atient day as do gen-eral floors (205).

charges or costs for the ancillary services used byICU patients are not matched to their ICU stays,because hospitals report their charges for thevarious an cillary services by departm ent, not bysite of patient location . If one consider s only ICUroom and board charges in estimating ICU costs,one m ay significantly u nd erestimate th e relativecostliness of ICU care, then, because ICU chargesunderestimate ICU costs and because the costs of ancillary services that are performed w hen pa -tients are in the ICU are not included .

With the exception of certain ad ministrativecosts that support ICU physician staff, the costsof physician services to ICU patients generally arenot included in hospital cost reports or in hospi-tal charges. As will be discussed further in chapter6, there is reason to b elieve that ICU p atients re-ceive a greater intensity of billable physician serv-ices than non-ICU patients.

Cost data from other countries provide an op-portu nity to d etermine relative costliness of ICU

v. non-ICU care, particularly in countries wh erehospitals receive operating budgets. In those fixedrevenue systems, hospitals do not need to chargemore than costs in some departments to make up

21



—.

22 q Health Case Stud y 28: Intensive Care Units: Costs, Outcome, and Decisionmaking

for losses in other dep artments. Estimates of costsfor a day of ICU care comp ared to a d ay of wardcare have ranged from a 2.5:1 ratio in France(182), to 3:1 in Canada and Australia (29,89), andto 4:1 in Great Britain (174). An attempt in theearly 1970s to estimate actual costs (includingancillary services) in the United States yielded an

estimate of 3.5:1 in a large, teaching hospital (97).But anecdotal reports now suggest that relativecosts of ICU to non-ICU care in some institutionsare as much as 5:1 (93).

Nu merou s U.S. studies of the per d iem chargeratio for room an d board in the ICU comp aredto non -ICU floors have show n a range of 2:1 to2.5:1 in small community hospitals (43,140) toabout 3:1 in large commun ity and teaching hos-pitals (140).

The Equitable Life Assurance Hosp ital DailyService Charge Survey of 2,519 hospitals in 1982

(71) showed an average char ge of $408.50 for an

intensive care bed and $167.50 for a p rivate bed ,a ratio of about 2.5:1.

Patients in ICUs have a relatively greater per-centage of their charges attr ibuted to ancillaryservices than to accommodations compared togeneral floor patients. In a recent study of a

large-sized community hospital, for example, 45.7percent of the total charges for ICU patients werefor room and board, while 57.1 percent of thetotal charges for non-ICU patients were for roomand board (175). Generally speaking, the m oreacutely ill the patient, the greater the percentageof the bill attributable to ancillary services(49,67,162,271).

In short, ICU patients consume more directresources, mostly for nursing, than regular floorpatients, as well as a greater pr oportion of ancil-lary services, particularly laboratory and ph ar-

macy services (49,101) than regular floor patients.

TOTAL NATIONAL COSTS OF INTENSIVE CARE

There is a notable lack of precision in estimatesof the portion of hosp ital care costs that can beattributed to intensive care. In a major review of ICUs in Technology in Hospitals (205), LouiseRussell provided a method for indirectly estimat-ing the n ational cost of ICU care. Recent rev iewsusing Russell’s method (described in app. B) esti-

mate that 15 to 20 percent of total costs of hospi-tal care can be at tributed to intensive care (40,136,206).

Before refining and updating this estimate, itis important to present the alternative ways of analyzing the costs of intensive care, includ ingcalculations of: 1) the d irect and ind irect costs of operating an ICU; 2) the total hospital costs, in-cluding the costs of ancillary services as well asICU costs, incurred by patients wh en they ar e inthe ICU; 3) the tot al hosp ital costs attribu tableto patients who spend any time in ICUs; and 4)the incremental cost generated by ICUs above thecost that a hospital would have to absorb fortreating very sick patients who w ould rem ain inthe hospital even if ICUs did not exist. The lastdefinition is particularly relevant to this case

study, since it is consistent with the concept thatthe ICU is a separate technology, ind epend ent of the patients treated in it.

Estimates of the total hospital cost of patientswh en in an ICU (Definition 2) and of the incre-mental costs of operating an ICU (Definition 4)are probably the most relevant in terms of public

policy considerations, but are not easily madefrom available hospital accounting sources (267).The direct and ind irect costs of an ICU (Defini-tion 1) and the total costs of intensive care pa-tients (Definition 3) are more easily estimatedfrom hospital accounting data, but have muchmore limited policy relevance.

Based on these considerations, estimates of thepercentage of total national inpatient hospitalcosts attributable to intensive care according tothe

q

q

different definitions can be made:

Definition 1: The direct and indirect costs of running the ICU, as reflected in charges forICU room and board —8 to 10 percent.Definition 2: The total hospital costs of pa-tients w hen in the ICU—14 to 17 percent.



Ch. 3–Cost of ICU Care q 23

q

q

Definition 3: The total hosp ital costs for pa- costs associated with most physician services,tients who sp end any time in the ICU dur- neonatal, pediatric, or bur n u nits, or the provi-ing a hosp italization—28 to 34 percent. sion of intensive care in Federal hospitals, oper-Definition 4: The incremental cost generated ated m ainly by the Veterans Adm inistration andby ICUs above the cost that a hospital wouldhave to absorb for treating ICU-type patientsif the ICU did not exist—cannot be estimated.

The assumptions underlying the estimates and thecalculations are available in appendix B.

Given these percentages, one can estimate thenational cost of adult intensive care. It should beemphasized that these estimates necessarily in-clud e the costs of coronary care, but not th ose

the Department of Defense. In 1982, total nationalexpenditures for hospital care were $136 billion,of which 84 percent were for acute care in com-

munity hospitals —or $114 billion (87a). Since anestimated 87 percent of community hospital costsare inpatient costs (4), $13 billion to $15 billionwere spent in 1982 for costs associated with pa-tients in adult ICUs and coronary care units,according to Definition 2 above.

25-338 0 - 84 - 3



4Utilization of ICUs



4

Utilization of ICUs

INTRODUCTION

there is little system-characteristics of in-

For a n um ber of reasons,atic information abou t thetensive care unit (ICU) patients, i.e., their age,sex, length of stay, and case mix. Hospitals andphysicians vary considerably, for example, in theway th ey treat patients with the same d isease. Fur-thermore, as w as noted earlier, there is no singlemodel of ICU organization—some hospitals havean ICU combined with a coronary care unit (CCU),wh ile others have separate u nits; some combinemedical and surgical ICUs, and others do not; stillothers have multiple subspecialty ICUs. Commu-nity hospitals, which usually do not have full-time

salaried physicians, may put less sick patients inICUs primarily to provide them with concentratednursing care (67).

There is no national data ba se wh ich d escribesICU utilization in any detail. The American Hos-pital Association (AHA) survey data provides in-formation only on ICU and CCU beds and daysby hospital size and type (see ch. z). A moredetailed pr ofile of ICU patients is based on pu b-lished stud ies from individual hosp itals. A com-

UTILIZATION BY TYPE OF ICU

Surgical ICU patients tend to be younger (49,155,175,227), to have more limited or reversiblediseases with reasonably well-defined therap eu-tic endpoints (50,56,129,175,178), and to be morehomogeneous than medical ICU patients (49).Even so, there are substantial differences amongsurgical ICU patients. The patient profile of sur-gical trau ma pa tients, for examp le, differs signif-icantly from that of postcardiac surgery patients.Trauma patients on average are younger and have

ICU ADMISSION RATESIt is not known what percentage of the popula-

tion, or even how many hospitalized patients areplaced in an ICU for any defined period of time.Relman suggests that 15 to 20 percent of all pa-

pilation of many, but not all, such stu dies is pre-sented in table 5. It should be emp hasized thatthese stud ies are from teaching hospitals and largecommunity hospitals and may not be represent-ative of the ICU care provided in small comm u-nity hospitals.

Recently, the Health Care Financing Adminis-tration (HCFA) has developed a profile of Medi-care hospital utilization, including ICU/ CCU uti-lization, based on its short-stay hospital inpatientstay record file for 1979 and 1980 (111,112). Thisfile, called the MEDPAR file¹, is generated by link-ing information from three HCFA master program

files for a 20-percent sample of Med icare benefi-ciaries. The MEDPAR file is the on ly data basewhich provides population-based rather thanhospital-based ICU utilization data, and, of course,it only profiles the Medicare popu lation.

‘The MEDPAR file also contains billed charge data and clinicalcharacteristics, such as principal diagnosis and principal procedure,

in addition to utilization data.

longer ICU stays than postcardiac surgery pa-tients.

Medical ICU patients tend to be older, havemore progressive, chronic diseases (29,174,248,265) and have more concurrent illnesses (265).These differences must be kept in m ind w hen eval-uating rep orts of utilization and ou tcome fromparticular ICUs.

tients are cared for in an ICU or CCU at somepoint du ring their hospital stay (195).

According to the 1979 MEDPAR sample, 18percent of Medicare patients w ho w ere discharged

25



26 q Health Case Study 28: lntensive Care Units : Costs, Outcome, and Decisionmaking

Table 5.—Summary of Selected ICU Studies

Dates ofType of data Number Mean ICU Percent ICU Percent hospital

Study author* Country ICU collection in study age LOS mort al it y mortality for ICU patients

Safar . . . . . . . . . . . U.S.Bates . . . . . . . . . . . CanadaBoyd . . . . . . . . . . . U.S.Crockett . . . . . . . . G.B.

Callahan . . . . . . . . U.S.BMA a . . . . . . . . . . . G.B.Rogers . . . . . . . . . U.S.Carroll . . . . . . . . . . U.S.Skidmore . . . . . . . G.B.Safar and

Grenvik(1971). . . . . . . . . Us.

Pessi . . . . . . . . . . . FinlandSpagnolo . . . . . . . U.S.Bell . . . . . . . . . . . . G.B.Petty (1974) . . . . . U.S.Nun. . . . . . . . . . . G.B.Turnbull . . . . . . . . U.S.Tagge (1975) . . . . . U.S.Tomlin . . . . . . . . . . G.B.McLeave . . . . . . . . Australia

Vanholder . . . . . . . BelgiumChassis . . . . . . . . . U.S.Byrick . . . . . . . . . . CanadaFedulo . . . . . . . . . . U.S.P o r n o . . . . . . . . , . . U . S .Thibault . . . . . . . . . U.S.Legal . . . . . . . . . . FranceMurata. . . . . . . . . . U.S.Hauser . . . . . . . . . U.S.Franklin . . . . . . . . . U.S.Knaus, et al.

( C C M , 1 9 8 2 )b. . U.S.

M-SR

M-SM-S

M-CM-SR

MM-S

M-S

iM-SM-SM-S

M-S(ca)M-SM-SM-S

M,R

M-SM-CM-SMMM

M-S

1959-19611958-1962

19631963-1965

1964-19661966-19671965-1968

19681965-1969

1965-19701965-19711970-19711966-19721964-19731970-19741971-19741972-19741973-19761975-1976

19761977197819781978

1977-19791978-1979

19791978-19801979-1980

1980-1981

561

336608

1,0005,521200

951,162

4,9181,001

2312,8961,598

4221,0352,8781,718

843

38048958

182558

2,693228149724512

1,408

—

48.0—

44.3

———

54.0—

—

50.056.045.2

— —

63.0

53.0

53.054.059.165.054.7

60.050.0

62.7 — —

54.0

— —

5.0 —

3.94.0 —

—

—

6.24.84.4 — —

5.2 —

3.03.4 —

5.18.0

3.63.4 —

3.9 — —

4.1

30.343.021.018.0

10.714.718.0

29.8

18,520.128.016.625.316.422.38.213.514,4

32.6

21.011.7

6.0

16.719.326.0

— —

26.0 —

—

—

28.947.0

—

38.6 —

19.7

42.614.0 —

29.017.310.034.026.8 — —

16.9

“Full citations found in References section.aweighted average from 14 ICUs.

bWeighted average from 6 ICUs.KEY: M-S Medical-Surgical ICU; M Medical ICU; M-C Medical-Cardiac ICU; R Respiratory ICU; S Surgical ICU.

SOURCE: Office of Technology Assessment.

from the hosp ital used intensive or coronary care.Fifteen percent used both general ward and ICU/ CCU beds, while 3 percent used only ICU/ CCUbeds. As table 6 indicates, use of ICU/ CCU bedsby Medicare patients does not vary significantlyby hospital size, except for hospitals under 100beds. Table 6 also show s that there is little varia-tion in ICU use by Med icare patients by size of hospital when ICU/ CCU use is considered as apercentage of the patients’ total charges. In-terestingly, there was also little variation in ICU/ CCU charges as a percent of total charges by type

of hospital sponsorship (not shown); 7 percent of all charges for Medicare patients in volun tary,

Table 6.—Use and Percentage of Hospital Chargesincurred in ICUs and CCUs for Medicare Beneficiaries

Discharged From Short-Stay Hospitals, 1979

Percent Percent totalusing charges incurred

Hospital bed size lCU/CCU in ICU/CCU

1-99 beds. . . . . . . . . . . . . . . . 12 5100-199 beds. . . . . . . . . . . . . 18 7200-299 beds. . . . . . . . . . . . . 20 8300-499 beds. . . . . . . . . . . . . 19 8>500 beds . . . . . . . . . . . . . . 20 7

All hospitals . . . . . . . . . . . m 7

SOURCE: C. Helbing, “Medicare: Use of and Charges for Accommodation andAncillary Services in Short-Stay Hospitals, 1979,” Office of Research,Health Care Financing Administration, U.S. Department of Health andHuman Services, undated.



Ch. 4–Utilization of ICUs q 27

proprietary, and public, non-Federal hospitalswere room and board charges for ICU/ CCUs.

Given the significant reg ional variations in theconcentration of ICU/ CCU beds (see ch. 2), it isnot surp rising that u tilization of ICU/ CCU bedsby Medicare patients also varied somewhat ac-cording to region (see table 7). Perhaps part of the explanation for the higher per diem costs andshorter lengths of stay in ICUs on the west coastis a result of the greater u se of relatively costlyICU/ CCUs in that region (255).

There are also variations by State in the use of ICU/ CCUs by Medicare patients; with a rangefrom 12 percent of Medicare hospital dischargesin Louisiana, Kansas, and South Dakota, to 27

percent in Connecticut.

Table 7.—Use and Percentage of Hospital ChargesIncurred in ICUs and CCUs for Medicare Beneficiaries

Discharged From Short-Stay Hospitals,by Geographic Region, 1979

RegionNew England . . . . . . . . .Middle Atlantic . . . . . ., .South Atlantic. . . . . . . . .East North Central ., ., .East South Central . . . .West North Central . . . .West South Central . . . .Mountain . . . . . . . . . . . . .Pacific ... , . . . . . . . . . . .

Percent using

I c w c c u

Percent totalcharges incurred

in ICU/CCU

SOURCE: C. Helbing, “Medicare: Use of and Charges for Accommodation andAncillary Services in Short-Stay Hospitals, 1979,” Office of Research,Health Care Financing Administration, U.S. Department of Health andHuman Services, undated.

Studies of ICU patients demonstrate a remark-ably consistent male to female ratio of about 3:2(16,47,56,67,146,175,178,248). Only Chassin re-ports a slight female predominance (40). In gen-eral, the ratio represents th e prevalence of seriouscardiovascular d iseases among m ales and femalesun der th e age of 70. Above that age, female rep-resentation in ICUs increases (248).

A major issue with respect to Medicare is therepresentation of elderly p eople in ICUs. With

aging comes an increase in the incidence of criticalillness. Thus, elderly people might be expected torequire more intensive care than their proportionof the general popu lation (34) and , possibly, morethan their proportion of the hospitalized popula-tion (76,175). On the other hand, to the extentthat ICU beds are in short supply (248,265) or thatpoor patient prognosis is considered (34,54,56,76),elderly patients might receive less intensive carethan younger patients.

In the United States, the representation of elderlypatients in ICUs seems to be the same or onlyslightly more than as it is in the hospital as a whole

(76,139,175). Data from ICUs do not ad dr ess theeffect of screening on the basis of age that maytake place prior to ICU entry. Speculation on theextent of such screening differs (33,76,137). Therecent HCFA MEDPAR data is somewhat helpful

201918

17

1515151823

7

7

7

7

6 7

6 7

10

SEX AND AGE DISTRIBUTION OF ICU USE

on this issue. As table 8 shows, use of ICU/ CCUsby elderly people does not var y from that of thegeneral pop ulation un til age 85. Even for p eople85 and older, however, the decrease in ICU/ CCUuse is slight.

Once in the ICU, elderly patients generally re-

ceive more interventions than younger patients(34). However, wh en an a ttempt is mad e to con-trol for acute severity of illness, the age of ICUpatients does not appear to be a factor in the

amount of resources expended in the ICU (137,140). Rather, health status, independent of age,

Table 8.-Use and Percentage of Hospital ChargesIncurred in ICUs and CCUs for Medicare BeneficiariesDischarged From Short-Stay Hospitals, by Age, 1980

Percent totalBeneficiary age Percent using charges incurredgroup lCU/CCU in ICU/CCU

<65 . . . . . . . . . . . . . 18 7

65-69 . . . . . . . . . . . . 18 8

70-74 . . . . . . . . . . . . 18 7

75-79 . . . . . . . . . . . . 18 7

80-84 . . . . . . . . . . . . 17 7>85 . . . . . . . . . . . . . 15 6

Total all ages . . . 18 7

SOURCE: C Helbing, Supervisory Statistician, Office of Research, Division of

Beneficiary Studies, Health Care Financing Administration, U.S. Depart-ment of Health and Human Services, personal communication, June6, 1983. Data derived from the MEDPAR file.



28 q Health Case Study 28: ]ntensive care Units: Costs, Outcome, and Decisionmaking

seems to be the key factor influencing the useof ICU resources once the p atient is in th e ICU(33,137).

Age does appear to be an important determi-nant of ICU admission in other countries. Whilethe populations are not strictly comparable, table5 clearly demonstrates a you nger m ean age of ICUpatients in foreign countr ies. Knaus compared theICUs in five U.S. teaching hospitals and sevenFrench teaching hospitals and found that 45.5 per-cent of U.S. emergency ICU admissions were 60years or older compared to only 31 percent of theFrench patients (142). Vanholder in Belgium ac-knowledged that w hen th ere is a lack of space inthe ICU, older p atients are less apt to be admitted(265). With many fewer ICU beds per capita avail-able in Britain, age appears to be a primary fac-

ICU CASE

Diagnoses

One characteristic of the ICU, particularly incomparison to other special care units (i.e., cor-onary, burn, and neonatal units), is the wide va-riety of underlying d iseases that are present. AsChassin emph asized, medical ICUs treat a widespectrum of illnesses; any specific disease repre-sents a very small proportion of the total numberof diseases that ar e present (40,238,265). Similarfindings have been described for mixed ICUs and

nonsu bspecialty su rgical ICUs (49,54,129, 139).Even respiratory ICUs treat a variety of primarydiseases (10,29).

In surgical ICUs in major regional centers,trauma patients may represent 40 to 50 percentof the ICU population (129,178). In other surgicalICUs and mixed ICUs, traum a victims r epresenta mu ch smaller percent of the overall ICU popula-tion (139), but are still a large proportion of themost critically ill patients (54). Trauma patientsare mu ch younger th an the overall ICU profile(54,129).

There is no accepted classification scheme th atdescribes the clinical characteristics of ICU pa-tients. Perhaps th e major p roblem with iden tify-ing ICU case mix is the fact that many critically

tor for limiting access to the scarce ICU beds (1).

When they were first developed, use of renaldialysis machines were rationed partly on the basisof age, and it has been suggested th at age was sim-ilarly a factor in the Un ited States in rationingscarce beds in the ear ly days of ICUs (248). In fact,as can be seen in table 5, in the last 15 years orso, there has been no dramatic trend toward olderICU patients even though the mean age of thepopulation has increased. Unfortunately, data onthe age of ICU patients in the late 1950s and early1960s, when ICUs were first opened, are not avail-able. In ad dition, there app ears to be n o consist-ent age d ifference in ICU use based on size or typ eof hospital. Finally, it should be pointed out thatmean ages reported in ICU studies are a few yearslower than th e med ian ages (248).

ill patients have multiple underlying medical prob-lems wh ich interact to p rodu ce severe physiologiccomplications. Vanhold er found , for examp le,that, excluding coronary care patients, each pa-tient in his ICU had an average of 4.39 signifi-cant, distinct diagnoses (265). Questionable diag-noses, disorders n ot likely to have v ital conse-quences, and previous diseases that had beencured at th e time of admission to the ICU were

not included in his calculation. The sicker the pa-tient, the more likely it is that the ICU is treatingfailure of ma jor organ system s, in ad dition to theund erlying d isease or the d isease that p recipitatedthe failure.

Other Case Mix Parameters

Recognizing that the complexity and severityof illness of ICU patients are generally not re-flected by the primary diagnosis, other descrip-tions of ICU case mix have been used . Patientscan be grouped accord ing to those referred d i-rectly from emergency rooms, those transferredfrom the regular hospital floors, and thosetransferred from other hospitals (31). InterhospitalICU tra nsfer of patients is r elatively infrequent



Ch. 4–Utilization of ICUs q29

in the United States, but common in some othercountries (81,142,146).

ICU admissions can be characterized as em er-gency or elective, the latter u sually referring topostop erative adm issions. Medical ICU admis-sions are usually emergencies, whereas the ma-

jority of surgical admissions are elective (49,52,227), unless the hospital is a major trauma cen-ter. Elective, postoperative patients may, never-theless, be critically ill, or at least need closemon itoring and observation (54).

ICU patients can be characterized as those re-quiring close observation and monitoring andthose requiring intensive therapy. As was pointedout earlier, there is no general agreement on howto classify patients into these groups. Some haveemployed subjective medical assessments of sever-ity of illness and treatment needs (42,163,179).Others have employed objective measures of ther-apeutic resource use developed by Cullen andcolleagues at Massachusetts General Hospital inBoston to separate patients into discrete groupsrequiring different personnel and treatment re-quirements (51,129,144). Recent work has at-tempted to ascribe a severity-of-illness score toeach patient and has found a good correlation be-tween scores of severity and treatment require-ments (144,270).

Because authors use varying approaches to de-scribe the intensity of ICU therap y, it is difficultto summarize the data. Nevertheless, it would ap-

pear from the literature —most of which is fromteaching or major community hospitals—that pa-

tients receiving the most concentrated intensivetreatment, involving fairly continuously directphysician involvement and various forms of life

support, represent less than half and sometimesas little as 10 to 20 percent of the ICU patientpopulation (54,129,144). At the other end of thespectrum, patients who receive technical monitor-ing and nu rsing care but only routine ph ysician

care probably represent about 20 to 30 percentof patients in general ICUs (136,137,178,246,269).The remaining 30 to 70 percent of ICU patients

are those that receive actual therapeutic interven-tion to maintain and stabilize one or more p hys-iologic functions, but do not require constantphysician involvement in their care or nurse-to-

patient ratios of greater than 1:1. The percentageof “monitor patients” is much higher in ICUs thatalso serve a CCU function (31,249).

Because most research has come from teachinghospitals, the pattern of case mix in communityhospitals may be d ifferent, although an ecdotal

reports do not indicate a consistent difference be-tween teaching and community hospitals (67,163,175).

Readmission

Recently, attention has focused on the fact thathigh-cost users of hospital care are often patientswith chronic illnesses wh o have r epeated hospi-tal ad missions (161,218). This pattern is being in-creasingly recognized for intensive care as well(231,248). As might be expected, readmission tothe same unit are less frequent for surgical ICUpatients (178). In a 5-year p eriod, almost 19 per-cent of all patients seen in a major teaching hos-pital medical/ cardiac ICU were readmissions, and6 percent were patients readm itted to the ICU du r-ing the same hospital stay (so-called “bouncebacks”) (248).

Length of Stay

The mean length of stay (LOS) in an ICU forall Medicare ICU/ CCU patients in 1980 was 4.2days (112). The LOS in ICUs is about 0.5 dayslonger th an in CCUs (49). The LOS is reported lylonger in non-U.S. ICUs (29,88,142,178), prob-

ably because th ere are fewer m onitor patients inthese ICUs. The average LOS in U.S. hospitalshas been n otably stable over the p ast 15 years (seetable 5).

As expected , mean LOS is significantly longerthan median LOS (42). The mean does not reflectthe great variat ion in LOS of ICU pa tients. In astudy of 1,001 consecutive patients in a surgica]ICU, Pessi (178) found that 27 percent stayed lessthan 2 days, while 15 percent stayed longer than10 days. In one medical ICU, Chassin (40) foundthat 10 percent stayed longer than 10 days. ICU

stays of more than a m onth are not u ncommon(49).

While the mean hospital LOS before the recentchanges in Medicare reimbursement in U.S. hos-



—

30 q Health Case Study 28:: intensive Care Units: Costs, Outcome, and Decisionmaking

pitals was 7.6 days (4) and 10.4 days for Medicare sumably because of case mix differences (40,49,patients (113), ICU patients have significantly 175). Part of the variation in p ublished stud ieslonger total hospital stays. From the few reports may also represent the general pattern of shorterthat present both ICU and total hospital LOS, hospital lengths of stay on the west coast (256).there is significant variation in hospital LOS, pre-



5 .

Outcomes of Intensive Care:Medical Benefits and Cost

Effectiveness



5 m

Outcomes of Intensive Care:

Medical Benefits and Cost Effectiveness

DIFFICULTIES IN ASSESSING EFFECTIVENESSEvaluating the effectiveness of the care provided

in the general adult intensive care unit (ICU)presents a number of problems. Unfortunately,it is difficult to separate the intensity of the carefrom the setting in wh ich it is provided (97,98),

and therefore, to know wh ether the same carewould have been equally effective whether it wasprovided in an ICU or in a general hospital floor.

Theoretically, at least, intensive therap y couldbe provided on regular medical floors (120). Infact, there are institutional differences about whois treated in ICUs and for how long (142). More-over, the level and style of intensive care for simi-lar health problems differ significantly amongICUs (67). These differences have developed be-cause of the particular circumstances of individ-ual hospitals, rather than because established cri-teria were ava ilable (247).

For some complex medical problems, manyphysicians feel that the necessary care can onlybe provided in an ICU (65). In the late 1960s and1970s, admission to an ICU became routine fora number of medical problems, despite the lack

of evidence that ICU care improved outcome.There have been no prospective clinical trials inwhich patients with similar problems were ran-domly allocated to two groups, one of which wastreated in an ICU while the other received inten-sive care ou tside the ICU (98,222). There is gen -eral agreement that such rand omized studies wouldbe unethical (262,279), and it is felt that for manyproblems, treatmen t in an ICU is necessary if apatient is to have a chance of survival (50).

Since, as noted, randomized clinical trials of ICUs are considered by many to be unethical,

most ICU outcome studies have been historicalcontrols and pre-ICU/ post-ICU d esigns (166).These types of studies, however, have been seri-ously flawed by the absence of acceptable criteria

for stratifying ICU patients by diagnosis andseverity of illness to assure comparability of pa-tient p opu lations between different ICUs and inthe same ICU over time (226,248,281).

In the coronary care unit (CCU), for example,it is felt that patients suffering myocardial infarc-tion should be stratified into clinically coherentsubpop ulations based on the type of myocardialinfarction suffered in order to assess outcomeproperly (28). The problem of stratification is

especially complicated in the ICU, because pa-tients often have multiple diagnoses, which makecategorization difficult (16,265), and because theirseverity of illness var ies (136).

There are other practical problems in conduct-ing research on ICU ou tcome, including:

1.

2 .

3 .

4 .

In

the fact that any ind ividu al institution w illhave a relatively small number of patientsin any clinical subset;the lack of a standard format for collectingdata;

the difficult yin obtaining informed consentfrom ICU patients in need of imm ediate, life-saving intervention (176); andthe d ifficulties in conducting stu dies that fol-low patients after their discharge from thehospital.

short, because of the absence of an acceptedclassification scheme for stratifying ICU pa tientsinto accepted subpopulations and because pro-spective clinical trials have not been performed,very little is known about the effectiveness of theICU as a distinct, discrete technology. Investi-

gators wh o report on changes in ICU mortalityrates or lengths of stay can only speculate onwhether their patient populations have changedover time (227,248).

33



34 . Health Case Study 28 : ]ntensive Care Units: Costs, Outcome, and Decisionmaking

Finally, while the primary measure for assess- presence of ICUS may adversely affect the qualitying the effectiveness of ICUs is patient ou tcome, of nursing care on the regular medical and surgi-it should be recognized that the ICU as a discrete cal floors (25,136). As difficult as it is to measureunit w ithin the h ospital may be a focus for edu - the effectiveness of ICU treatm ent for pa tients incation and research activities which have positive the ICU, it is nearly impossible to assess objec-“trickle down ” effects on care for non-ICU pa- tively the benefits or d raw backs of the ICU fortients (55,86,97). At the same time, how ever, the the hosp ital as a wh ole.

CLINICAL OUTCOMES OF ICU CARE

Because of the varied case-mix in ICUs, it is im-possible to generalize about w hether ICU care im-proves ou tcome. The NIH consensus p anel, whichwas asked to assess this issue, concluded thateviden ce of the benefit of ICU care w as un equiv-ocal for a portion of the heterogeneous ICU pa-tient population (176). The NIH panel identifieddifferent ou tcomes for three categories of patients(176):

First is the p atient with acute reversible diseasefor whom the probability of survival without ICUintervention is low, but the survival probabilitywith such interventions is high. Common clini-cal examples include the p atient with acute revers-ible respiratory failure due to drug overdose, orwith cardiac conduction disturbances resulting incardiovascular collapse bu t am enable to pace-maker therapy. Because survival for many of these patients without such life-support interven-tions is uncommon, the observed high survivalrates constitute unequivocal evidence of reducedmortality for this category of ICU patients. Thesepatients clearly benefit from ICU care.

Another grou p consists of patients with a lowprobability of survival without intensive carewhose probability of survival with intensive caremay be higher—but the potential benefit is notas clear. Clinical examples include patients withseptic or cardiogenic shock. The weight of clini-cal opinion is that ICUs reduce mortality for manyof these patients, though this conviction is sup-ported only by uncontrolled or poorly controlledstudies. Often these studies do not allow one todistinguish between ICU effectiveness and/ or dif-ferences in cointerventions that do not require theICU.

A third category is patients admitted to theICU, not because they are critically ill, but be-cause they are at risk of becoming critically ill.The purposes of intensive care in these instancesare to prevent a serious comp lication or to allow

a prompt response to any complication that mayoccur. It is presumed that the prompt responseto a potentially fatal complication made possibleby continuous monitoring plus the concentrationof specialized personnel in the ICU increases theprobability of a favorable outcome. The risk of complication may be h igh (as in the patient w ithan acute myocardial infarction and complex ven-tricular ectopy) or low (as in the patient withmyocardial infarction suspected because of chestpain in the absence of electrocardiograph ic abnor-malities). Also, the differences in probability of a favorable outcome following a complication in-side rather than outside the ICU may be large (asin the patient with postcraniotomy intracranialbleeding) or small (as in the patient with gastro-intestinal bleeding). The strength of evidence su p-porting the effectiveness of the ICU varies withthe probability of a complication and with the dif-ference in expected outcome inside and outside theICU. When the risk of complication is high andthe potential gain large, a decrease in mortality

is likely. Similarly, when the risk is low and thepotential gain small, an observable decrease inmortality is un likely. These p atients are n ot likelyto benefit from ICU care.

The differences in outcomes of ICU care bydiagnosis has been demonstrated in all studies thathave looked at the issue, from the earliest studies(17) to th e most r ecent (248). Table 9 gives ex-amples of specific retrospective outcome studieson the effect of ICU care for certain illnesses.(Note that contrad ictory find ings are sometimesfound for the same condition. ) In general, condi-tions which respond well to ICU care are reversi-ble illnesses without significant underlying chronicillness (e.g., respiratory arrests as a result of d rug

overdoses, major trauma, reversible neuromus-cular diseases such as Guillain-Barre Syndrome,and diabetic ketoacidosis) (198,214). Conditionswh ich generally do not respon d well are exacer-



Ch. .5-Outcomes of intensiv e Care: Medical Benefits and Cost Effectiv eness • 35

Table 9.—Retrospective Outcome Studies of ICU Care

Study Condition

A. Studies showing definite reduction in mortality for condltlon: Petty (1975). . Respiratory failure treated with ventilatorsRogers. . . . . . Respiratory failure treated with ventilatorsBates . . . . . . . Status asthmatics and emphysemaDrake . . . . . . . Non-hemorrhagic strokes

Skidmore . . . Postoperative trauma patientsFeller. . . . . . . Severe burns

B. Studies showing no reduction in mortality for condition: Pitner. . . . . . . StrokesPiper . . . . . . . Drug overdoseJennet . . . . . . Head injuries with comaCasali . . . . . . Postoperative acute renal failureGriner . . . . . . Pulmonary edemaHook . . . . . . . Pneumococcal bacteremia

NOTE Studies are cited in the Reference section

SOURCE Off Ice of Technology Assessment

bations of chronic cond itions for which there hasbeen no definitive treatment (e.g., cirrhosis withgastrointestinal hemorrhaging, and advancedcancer).

FUNCTIONAL OUTCOME

Different investigators have u sed varying meas-ures of functional status to gauge outcomes otherthan mortality. These measurements have beensubjective and depend to a large extent on the pa-tient’s prehospital functional status. For patientswith a chronic disability, posthospital functionalstatus is almost never better than their prehosp italfunctional status (34,40,146), although improve-ment has occasionally been found (29).

Surgical patients su ffering an acute injury orillness have a reasonable chance of returning to

Most stud ies have looked at mortality in theICU or in the hospital as a measure of the efficacyof ICU care. However, for some physiologic con-ditions, such as cardiac arrest, ICU care may belifesaving in the short term but may not affect theultimate course of the und erlying illness (174,214).Indeed, in some instances, patients with severe

underlying illnesses, such as terminal cancer andcystic fibrosis, have not been offered ICU care be-cause of the dismal prognosis associated with theunderlying illness (58,110,252,253).

Investigators have only recently begun to look at posthospital survival. As might be expected,the ability to follow patients for 6 months orlonger after their ICU stay depends to a great ex-tent on the pop ulation being stud ied. In general,chronically ill and medical patients are more likelythan acutely ill and su rgical patients to d ie shortlyafter d ischarge from the hosp ital (29,34,50,129,146,174,175,178,248) .

a normal functional status (54,178). In a followupstudy, Cullen reported that the l-year mortalityrate was similar to the rate in a previous studyof similarly critically ill patients, bu t tha t the p a-tients’ quality of life as measured by the numberof patients who were fully recovered or returnedto full productivity was significantly improved(54,56). This finding suggests that Outcome COme meas-

ures other than survival should also be examinedwhen determining effectiveness of ICU care.

CHARACTERISTICS OF ICU NONSURVIVORS

As noted above, certain diseases and conditions Ageare associated with particularly high ICU mortal-ity rates. Underlying disease is probably the m ost A number of investigators have looked at the

significant single predictor of outcome of ICU care association of age and mortality in ICUS. Most

(54,139). Other factors, includ ing age an d sever- have found a direct relationship between increas-

ity of illness, are importan t as w ell. ing age above 65 and hospital mortality (54,107,



36 q Health Case Study 28: Intensive Care Units: Costs, Outcome, and Decisionmaking

116,178,214,248). In addition, for medical patientsin particular, some have found that patients 70and over who leave the hospital have very highposthospital mortality rates (29,174,248,249).Others, however, have found either a small or noassociation between age and survival (40,50,76,165,265).

When an attempt is made to control for chronichealth status in a multivariate logistic regressionanalysis, age has been foun d to remain a reliableindependent predictor of mortality (268). Thisfind ing suggests that age is not simply a surrogatefor chronic health status. Fedullo (76), on the otherhand, suggests that with the passage of time, eld-erly patients have already gone through a proc-ess of selection, and therefore “healthy” elderlypatients are as able as youn ger patients to sur -vive an acute m ajor illness.

Severity of IllnessVanholder (265) found th at ICU survivors had

an average of 3.13 major diagnoses whereas non-survivors h ad 6.09 diagnoses. LeGall (146) founda strong positive correlation between the numberof organ system failures and the likelihood of notsurviving a stay in an ICU. In a num ber of set-tings, the George Washington University ICU Re-search grou p in Washington , DC (143) has testedan acute severity-of-illness measure based primar-ily on the deviation from normal of certain clini-cal and laboratory measurements. Using their

scoring system, they found a direct relationshipbetween acute severity of illness and ICU mor -tality and conclud ed that acute p hysiologic de-rangement (i.e., acute severity of illness) is sec-ond only to the underlying disease as a risk factorof hospital mortality (139). Less sophisticatedseverity-of-illness classification systems have con-sistently demonstrated a positive relationship be-tween increasing severity of illness and likelihoodof mortality (51,178).

Resource Use

In comparing resource use of ICU nonsurvivorsto sur vivors, it is necessary to look at the p atient’sentire hospitalization, not just the stay in the ICU.In a num ber of studies from different types of hos-

pitals, 25 to 40 percent of ICU pa tients who d iedin the hospital did so after they w ere transferredfrom the ICU to the regu lar med ical floor (seetable 5 in ch. 4). Presumably, many of these non-ICU deaths were anticipated and represented thetransfer of “hopeless” patients out of the ICU.

It is now recognized that a significant n um berof deaths in the ICU occur after “no resuscitation”orders hav e been written. In tw o large medicalcenters, as many as 40 to 70 percent of ICU deathsoccurred un der these circumstances (9,96). In alarge community hospital, 19 percent of ICU non-survivors had no hop e of recovery and were inthe ICU solely for ter minal care (165). In shor t,a substantial portion of ICU care for nonsur-vivors occurs after hope of recovery has beenabandoned.

Some nonsu rvivors have very short and somehave very long ICU stays. Pessi (178) found thatone-third of surgical ICU nonsurvivors died within2 days and 80 percent died within 10 days of ICUadmission. More recently, Cromwell (49) foundthat w hile 20 percent of ICU nonsu rvivors diedwithin 3 days of ICU admission, 10 percent diedafter 2 months in the ICU. On average, nonsur-vivors stay in th e ICU about 1.5 to 2 times longer

than survivors (42,48,76,248)

In 1973, Civetta (42) first d escribed the inverserelationship betw een ICU charges and survival.Since then, whenever it has been examined, thesame relationship has been foun d—ICU nonsur-

vivors accumulate up to two times more h ospi-tal charges than survivors (40,49,61). Byrick (29)found the same correlation in Canad a w hen heconsidered actual ICU costs rather than charges.Furthermore, nonsurvivors have incurred propor-tionately higher charges for ancillary services(e.g., laboratory tests, X-rays, and blood) thansurvivors (61,76). Only Parno (175), in a studyinvolving a large comm unity hospital, found nosubstantial difference in ICU charges between sur-vivors and nonsurvivors.

The inverse relationship betw een charges an d

survival is not as simple as it m ight first app ear,however. Detsky (61) looked at the relationshipbetween charges and patients assigned to varioussubjective p rognostic categories. He found the



Ch. 5—Outcomes of Intensive Care: Medical Benefits and Cost Effectiveness q37

highest per capita charges in two groups: sur-vivors who initially had been thought to have apoor chance of survival, and nonsurvivors whohad initially been felt to have the best chance of survival. Predicted nonsurvivors who died andpredicted survivors who lived consumed fewerresources. The two grou ps w ith highest charges

wou ld logically be the ones who might benefit themost from intensive m edical care.

In another study utilizing a severity-of-illnessmeasure, Scheffler (214) found a nonlinear, U-shaped relationship between the use of resourcesavailable in the ICU and the p robability of sur-vival. The first segment —45 percent of patientsand 19 percent of therapeutic interventions-ex-hibited an overall decrease in th e p robability of death as therapy increased. The second segment,foun d a t the bend of the curve, showed little cor-

relation between p robability of death and resourceuse. However, in the third segment, the rising por-tion of the U-shaped curve, there was an overallincrease in the p robability of death as resourceuse increased. This last segment represented only9 percent of the ICU population, but those pa-tients consumed as mu ch as 30 to 40 percent of

the ICU resources. Thus, many patients, even themost seriously ill, may benefit from ad ditionalICU resources applied to their care. While, in ret-rospect, some resources may prove to h ave been“wasted” in the sense that individuals did not sur -vive desp ite consu ming th ese ICU resour ces, it isclear that many patients d o benefit from increaseduse of ICU resources. The patients who will ben-efit from additional ICU resources cannot cur-rently be identified ahead of time with any cer-tainty.

DISTRIBUTION OF ICU COSTS AMONG PATIENTS

The data demonstrate that a small percentageof the ICU patient pop ulation consumes a substan-tial proportion of total ICU resources. Cromwell’sgroup (49) found that 1 p ercent of all ICU patientsincurred 10 percent of hospital charges, and 5 per-cent of ICU patients incurred 25 percent of thecharges. In Chassin’s ICU stu dy (40), 7.4 percentof the patients incurred 31 percent of the charges,and 17 percent of patients incurred so percent of the charges. The 7.4 percent subgrou p averaged

$63,000 in charges in 1977 dollars. In general, thehigh cost subgroup was broadly representative of the total ICU pa tient popu lation in term s of age,diagnosis, and other patient characteristics. Sim-ilarly, Parno (175) found that 18 percent of theICU p opu lation in his hosp ital generated half of the ICU charges.

In addition, it is likely that within the ICU,there is su bstantial cross-subsidization of charges.As noted in chapter 4, ICU populations includepatients who are there primarily to be observedand monitored for the development of comp lica-

tions as well as patients w ho are receiving com-plex life-sustaining therapy. The nurse-to-patientratio can vary from 1:4 or 1:5 for patients withcardiac arrhythmias to 1:1 or greater for thesickest patients (176). While a portion of fixed di-

rect costs and allocated indirect costs should bedistributed evenly among all patients, the ICUcharge structure does not reflect the substantialdifferences in variable labor costs between p a-tients.

The Therapeutic Intervention Scoring System(TISS) (53,130) is a relative value scale wh ich re-

duces most of the tasks commonly performedwithin an ICU to 75 items which are assigned

varying weights. It has been used as a d irect meas-ure of the use of labor in the ICU. Wagner (270)found that patients recuperating from coronary by-pass su rgery u tilized 2.5 times more TISS points perday than ICU patients recovering from brainsurgery.

The difference in labor resource use appears tobe even greater for other types of patients (51,54).The distribution of TISS points suggests that all ICUpatients receive a minimum amount of treatment be-yond that provided on the regular wards (67). Thedata also suggest, however, that even if indirect andfixed ICU costs are distributed evenly among all pa-tients, perhaps 50 percent of actual ICU resourcecosts—particularly labor costs—vary dramaticallyamong patients.

25-338 0 - 84 - 4



38 q Health Case Study 28: Intensive Care Units: Costs, O utcome, and Decisionmaking

As noted earlier, the sickest ICU patients incursubstantially more total hospital charges than thosewho are relatively less sick. Yet, the actual cost d if-ferences between these two groups is even greater.Under the new Medicare payment system, which

MONITORED PATIENTSIncreased attention has been paid recently to

ICU patients wh o d o not receive active intensivetherapy but rather are monitored and observedfor the d evelopment of p otentially fatal complica-tions which must be responded to promptly (176).Progress has been made in identifying the char-acteristics of coronary patients who do not rou-tinely requ ire coronary intensive care (85,90,141,189,190), and in recognizing CCU patients whocan be d ischarged to the general floor after 24hou rs rather t han the u sual 3 days (163). Simi-larly, national and regional data on intensive carefor patients with burns suggest that a substantialnumber of patients suffering relatively minorburns d o not benefit from treatment in an inten-sive bum unit but receive it nevertheless (78,151).

Researchers at George Washington University(269) found that 513 of 1,148 admissions (45 per-cent) to a mixed medical-surgical ICU in a teach-ing hospital could be considered “monitoringonly” patients. Using a multivariate logistic regres-sion analysis of several variables, including aseverity-of-illness measure, they found that 154patients (13 percent of the total ICU patient p op-

ulation) had less than a 5-percent pred icted risk of requiring active intensive therap y. For thosepatients, the au thors felt that the risks of iatro-genic illness’ migh t ou tweigh the ben efits of ICUmonitoring. In fact, only of the 154 low-risk pa-tients actually received intensive therap y, and inno case did th ose patients require therapy for animmed iately life-endangering cond ition. After up-dating their data base and looking at preliminarydata from other un iversity hospitals, the authorsconcluded that all ICUs have significant propor-tions of predictably low-risk, monitor admissions(141). The conclusions were supported in a recent

‘An iatrogenic illness is an illness that results from clinical ther-apy rather than from the patient’s disease.

pays a fixed price per diagnosis regardless of actualcost of the treatmen t pr ovided , hospitals ‘maybecome more aware of the highly disproportionateshare of ICU resources consumed by the m ostseverely ill, long-term ICU p atients (see ch. 6).

study by Fineberg that looked at p atients with arisk of myocardial infarction that is low, but notlow enough for home care to be d esirable (about5 percent). He calculated that ad mission to anintermed iate care un it, rather tha n a CCU, washighly cost effective (79).

Others who have studied monitored patientsare not as sangu ine about the ability to pred ictlow risk. In a coronary care-oriented ICU, Thibault(248) found that 1 of 10 patients admitted forcareful mon itoring subsequently required a ma-

jor ICU intervention. Using primarily subjectivecriteria, he could not p redict which of the moni-tored patients wou ld d o well.

Teplick, et al. (246), studied pa tients rou tinelyadm itted to a sur gical ICU after uneventful, ma-

jor surgery of various types. Using a fairly con-servative d efinition of benefit, the auth ors foundthat overall, 33 percent of the patients benefitedmedically from an overnight stay in the ICU.There was a broad range in the percentage of pa-tients who benefited from ICU care across typesof surgery, from 44 percent of patients who had

vascular surgery to no patients who had anteriorcervical Iaminectomies. A number of the unan-ticipated complications w ere imm ediately life-threatening. Furthermore, using both a preopera-tive risk assessment and an evaluation of intra-operative problems, the authors were unable toidentify the patients w ithin each surgical categorywho were more likely than others to develop seri-ous postoperative problems.

Another study of the same ICU, however, foundthat less than 1 p ercent of patients routinely ad-mitted overnight to the ICU for certain other sur-gical conditions suffered significant adverse post-operative effects (220). These contrasting findingsdemonstrate the importance of stratifying eventhe monitored ICU patients in order to determine



Ch. 5—Outcomes of lntensive Care: Medical Benefits and Cost Effectiv eness Ž 39

which subgroups of monitor patients do well with-out rou tine adm ission to the ICU.

Attention has also been focused recently on p a-tients wh o may be d ischarged from the ICU pre-maturely. Schwartz (220) found that 15 percentof patients electively d ischarged from the ICU,and 23 percent of patients transferred ou t of the

ICU because of lack of space, suffered a signifi-cant ad verse effect on the surgical floor. Adverseeffects included death, return to the ICU, orresidence in hospital 1 month after completion of the stud y. The researchers also found that app rox-imately one-third of patients undergoing ab-dom inal vascular surgery d eveloped serious res-piratory and / or circulatory conditions after dis-charge from the ICU. They did not speculate onwh ether outcomes for these patients wou ld havebeen d ifferent h ad the comp lications occurred inthe ICU.

In a retrosp ective chart review, Franklin (82)noted that 62 percent of readmission to a mixedICU might have benefited if they had n ot beendischarged from the ICU initially. The au thors d idnot indicate whether the patients readmitted tothe ICU differed in any p redictable manner from

patients who d id not need to be readm itted. Nordid the study address how many lives were lostbecause of early discharge. Mulley (163), who rec-ommended identification of low-risk patients forearly transfer from the ICU, acknowledged that2 percent of the low-risk group had major com-plications during their stay in the ICU that would

have occurr ed after transfer if an early tr ansferpolicy had been in effect.

By stratifying ICU-monitored patients, it maybe possible to reduce or eliminate ICU stays forsome patients with a low risk of resulting adverseeffects. This risk may, in fact, be lower than therisk of iatrogenic ICU illness for some patients.At the sam e time, other mod erately sick ICU pa-tients are probably discharged too soon or not ad-mitted to the ICU at all because of lack of bedspace or recognition that the patients are at risk for serious complications. As a result, they suf-

fer avoidable adverse health effects.

Work is only now beginning on attempts to pre-dict which ICU discharge patients are most likelyto suffer adverse effects on the regular medicalor surgical floor.

ADVERSE OUTCOMES OF ICU CARE

Iatrogenic Illness

The possibility that the adverse effects of ICU

care may outw eigh the potential benefits for somepat ients is being increasingly recognized (176,275).However, the rates of iatrogenic illness and otheruntoward physical and psychological reactions toICU care are not known with any precision (176).

As with the problems of measuring the positiveeffects of ICU care, it is difficult to distinguish be-tween the negative effects that occur among crit-ically ill patients rega rd less of location and thosethat are specific to the ICU.

An iatrogenic illness is any illness or otherharm ful occurrence that results from a d iagnos-

tic procedure or therapy that is not a natural con-sequen ce of the patient’s d iseases (239). The m a- jor iatrogenic complications that result from pro-longed ICU care include nosocomial infections

(defined below), stress-induced gastrointestinal

bleeding, alterations of consciousness a ssociatedwith m etabolic disorders, coagulation disordersassociated with multiple transfusions and infec-tion, dru g interactions, comp lications of intra-vascular catheterization, comp lications of p ro-longed end otracheal and nasogastric incubation,and sleep disorders and psychoses (41,275). Someof these complications, such as drug interactionsand bleeding, would likely occur in seriously illpatients regardless of location. Nosocomial infec-tions and various p sychological reactions are oftena result of the ICU itself.

Recently, Steel found that 36 percent of patientson th e med ical service of a un iversity teachinghosp ital had an iatrogenic illness (239). In 9 per-cent of the cases, the incident was life-threatening



.

40 q Health Case Study 28:: Intensive Care Units: Costs, outcome, and Decisionmaking

or p rodu ced considerable disability. In 2 percentof the cases, the iatrogen ic illness was believedto have contributed to the death of the patient.The authors did not specify which problems spe-cifically occur red within the CCU or ICU sectionof the medical service. Nevertheless, a number of the complications came from drugs, such as lido-

caine, and procedures, such as Swan -Ganz cath-eterization, that are, for the most part, only usedin ICUs.

In a different teaching hospital, Abramson (3)iden tified 145 reports of significant adverse oc-currences in 4,720 ICU admissions during a 4-yearperiod. Ninety-two of these incidents were felt tobe the result of hum an error, and 53 were equ ip-ment malfunctions. However, 43 of the 92 in-cidents linked to human error involved equip-ment, mostly mechanical ventilators. Thus, abouttwo-thirds of the adverse events involved the

technically complex equipment used in ICUs. Theincidence of equipm ent-related adverse occur-rences would probably be much higher if theequipment and the staff operating it were dis-persed throughout the hosp ital (208). On the otherhand , ICU technology may sometimes be used un -necessarily for less sick patients, producing someincidence of avoidable iatrogenic illness (198). Asnoted in chap ter 7, the ICU milieu p rovides a biasto the use of technology, which at times may beof only marginal benefit and can produce adversereactions (242).

Finally, it is clear that the sophisticated care

provided in the ICU requires skilled nu rses andother technicians. Adverse effects in ICUs havebeen particularly noted during periods of nurs-ing shortages (3,136). The ICU environ men t pr o-duces “technology-oriented” treatment protocols(100), and physicians are less apt to tailor thera-py based on the specific skills of the nu rse andtechn icians on du ty or on the par ticular nu rse-to-patient ratios during a particular shift. In otherword s, certain ICU monitoring and therap y pro-tocols may w ork well und er ideal circumstancesbut may be particularly subject to human and me-chanical error under less favorable circumstances.

Nosocomial Infections

Nosocomial infections a re infections occurr ingduring hospitalization that were not present, andnot incubating, at the time of hospital admission(117). All patients in an ICU are at increased risk of developing nosocomial infections (117). The

rate of significant nosocomial infection in an ICUis about 20 percent, or three to four times thatof a patient on a general ward (63,173). This in-creased rate stems in part from unalterable fac-tors, including the severity of the underlying ill-ness; the greater use of invasive procedures; andthe greater use of prior antibiotic therapy, whichmay predispose a patient to a superimposed in-fection (63,117,192). However, at least part of theincreased rate of ICU infection is due to cross-infection betw een very sick patients in the con-fined area of the ICU (63,204). Nosocomial in-fection “outbreaks” in ICUs are not uncommon

(63). Bacterial infections may be spread directlyfrom one p erson to another, often via p ersonnel,or may require an intermediate reservoir, such asrespirator nebulizers or tubing (117). While dif-ficult to estimate precisely, the costs of nosocomialinfections in terms of increased morbidity, mor-tality, and hospital charges are undoubtedly sub-stantial (108).

Psychological Reactions

There is a substantial body of literature on the

psychological reactions of patients in ICUs. It ap-pears that the frequency of psychiatric syndromesis consider ably less in a CCU, where p atients arerelatively stable, than in an ICU, where seriou slyill patients suffer organic impairments of cerebral,renal, and pulmonary function (104,131,156).

The so-called “intensive care syndrome” (156)described a “m adness,” or acute delirium, that hadoriginally been seen in the p ostoperative recoveryroom (168). However, many psychiatric syndromeshave been noted , from acute anxiety, fear, andsustained tension to agitated depression and acutedelirium (132).



Ch. 5–Outcomes of Intensive Care: Medical Benefits and Cost Effectiveness q41

The unique environment of the ICU has beengraph ically implicated as a cause of the varied andoften dramatic psychological reactions:

Immobilized, weak, inhibited from moving bya network of wires and tubes which connect everyorifice in his body with bottles and machines, helies watching the light pattern move from left toright on the monitor, disappear, then start again.He listens to the suction of the draining appara-tus, the on and off of the pulmonary respirator,the hissing sound of the steam from the vaporizedoxygen; steam which sometimes clouds his visionin the tent. He ad ds h is own fantasies to thisbewildering environment. Fear and tension mount. . . . In the ICU, the lights are on constantly, andthere is little or no change in the level or type of sensory inp ut. The activity, in sp ite of its decreasetoward early morning, remains high. Hours anddays merge and blend. Privacy is almost impos-sible. The patient is exposed ; his most p rivate actsbecome p ub lic. . . . Strangers control the m a-chines. Their au thority is absolute. In this seem-ingly irrational environment, he is depr ived of any volitional control. He becomes an object

rather than a participant in the struggle for life(62).

Sleep depr ivation, sensory d eprivation, sensoryoverload, medications, and various emotional fac-tors related to coping w ith serious illness havebeen cited as causes for ICU psychiatric syndromes

(38,104,131,145).Given the d ramatic behavioral responses to ICU

care, it is remarkable that most p atients remembervery little abou t the “terror in the ICU” (216). Insurveys taken both shortly after transfer out of the ICU and man y month s later, ICU patients gen-erally remember few details of their stay (24,29,115,127,162,216). Whether due to the seriousnature of the underlying illnesses (104,127), thelack of sleep, which produces general fogginess(24,127), or a powerful psychological defensemechanism of denial called “psychoplegia” (104,216,217), survivors of ICU care genera lly do not

carry u nique p sychological scars of their ICU ex-perience.

COST-EFFECTIVENESS ANALYSIS OF ADULT INTENSIVE CARE

Cost-effectiveness analysis (CEA) is intendedpr imarily to measu re and comp are the costs of different w ays of arriving at similar ou tcomes(256). This typ e of analysis has not been d one forICUs, because it is considered unethical to denyICU care for most ICU pat ients (see ch. 6). The

few “before ICU/ after ICU” stud ies focused onrelatively small ICU subp opu lations and are clear-ly dated (99,183).

For the low-risk mon itored patient, it may beethically permissible to compare ICU observationwith n on-ICU observation to determine the costeffectiveness of ICU care. Both Mulley (163) andWagner (269) have projected cost savings thatwou ld be generated by more selective adm issionand earlier discharge policies. Using conservativeeconomic assumptions, Mulley found that a moreselective policy w ould result in a 6-percent r educ-

tion in ICU charges. Similarly, Wagner estimateda 4-percent redu ction in total ICU days w ith ear-lier discharge of low-risk patients. Neither authoraccounted for the possibility that earlier transferfrom the unit might either increase or, conceiva-

bly, decrease the rate of major complications,wh ich, in turn , wou ld a ffect costs (163). Finebergestimated that for patients with about a 5-percentprobability of having sustained a m yocardial in-farction, admission to a CCU would cost $2.04million p er life saved and $139, ooo per year of

life saved, as compared to care in an intermediatecare unit (79). Teplick (246) concluded that rou-tine overnight ICU ad mission for p ostoperativepatients at an additional cost of $3OO would re-duce overall patient costs if only 13 of the 88routinely ad mitted patients in their study whobenefited from the ICU were p revented from be-coming critically ill.

Another factor in considering the overall costeffectiveness of earlier discharges of low-risk ICUpatients is the fact tha t the costs of caring for thesepatients on the regular floors would increase,

mostly because of the need for additional nurs-ing, probably from private duty nurses (97,220).There might also be a need for additional monitor-ing equipment on the regular floors. Finally, pro- jecting savings based on charges probably over-



42 qHealth Case Study 28 : Intensive Care Units: Costs, Outcome, and Decisionmaking

estimates the savings from early discharge of low-risk patients because of the cross-subsidiza-tion that is reflected in the ICU charges (see ch. 6).

Attempts have been made to assess averagecharges necessary to achieve one survivor forvarious subpopulations of ICU patients. For ex-

amp le, Parno (175) found that h ospital chargesin 1978 dollars for a survivor alive 2 years afterdischarge averaged $15,000, with a range of $1,650for dru g overd ose patients to $46,000 for renalmedical patients. In a population of the most crit-ically ill surg ical ICU p atients, Cu llen (50) foundthat in 1977-78 dollars, it required $71,000 in hos-pital charges to achieve a sur vivor alive 1 yearafter h ospital discharge. N either add itional post-hospitalization expenses nor physician chargeswere includ ed in this estimate. For the categoryof illness that includes gastrointestinal bleeding,cirrhosis, and portal hyp ertension, Cullen found

that it cost $260,000 to achieve one survivor.An interesting variation on this ap proach is to

look at “life-years” saved (134). The method isnot a true cost-benefit analysis (CBA), however ,since CBA requires that benefits be assigned amonetary value in order to provide a direct com-parison of the costs and benefits of a particulartechnology (256). Assigning monetary values tothe varied and controversial outcomes of the ICUhas not been done. Theoretically, the life-yearssaved method could be extended into CBA. Rec-ognizing that longevity is generally considered abenefit, Bendixen used the life-year saved modelto view th e cost of ICU care in relation to pre-dicted remaining lifespan. He used the followingequation:

cost = (cost per day) X (duration of stay)

(survival fraction) X (predicted remaining lifespan)

This approach assumes not only that survivalis a benefit, but also that su rvival value is a mu lti-ple of survival time, i.e., that 2 years of survivalhas tw ice the valu e of 1 year of surv ival. The ap -proach theoretically permits one to w eigh the fac-tors of a patient’s age and the prognosis associ-ated with chronic disease. The formula, however,does not d iscoun t the futur e value of costs andbenefits into present dollars; in essence, it over-states the importance of predicted remaining life-span (256).

The unavailability of disease-adjusted actuarialdata for d iagnostic subgroups makes p redictionof life expectancy for chronic diseases inexact(215). ICU survival fraction and predicted remain-ing lifespan are the m ajor determinants of cost ef-fectiveness according to this formula. Using thisapproach in 1977, Bendixen estimated a cost-per-

year saved of $84 for barbiturate overdose and$180,000 for h epa toren al failure.

When better estimates of life expectancy for pa-tients with chronic illnesses become available, thiscost-effectiveness approach may be more useful.Nevertheless, application of this app roach d ocu-ments the importance of the underlying diseaseprocess and th e patient’s age in d etermining thecost effectiveness of ICU care (215). The formulacurrently does not p ermit quan titative consider-ation of quality of life, which is obviously impor-

tant for patients w ith d ebilitating chronic illnesses(18). Method s for adjust ing life-years saved forquality of life have been attemp ted (213), but havebeen criticized as representing “bad science” andfor ignoring considerations of justice and equity(7).



6Payment for ICU Service;



6.

Payment for ICU Services

TRADITIONAL HOSPITAL REIMBURSEMENT

Derzon (60) emp hasized that several featuresof the American health financing and p aymentsystem operate to reinforce use of expensive tech-nology, such as intensive care units (ICUs). Thesefactors includ e p ayment certainty, consumer in-surability, government assumption of risk, andbenefits based on “medical” necessity. These fac-tors have provided the major impetus to expan-sion of ICUs in the 1960s, 1970s, and ear ly 1980s.

Most patients are covered for all or par t of theirhospital costs through p rivate or government in-surance. With the exception of small indemnityinsurance comp anies, wh ich p ay a fixed d ollaramount per day or a fixed coinsurance rate basedon hospital charges, most private insurance com-panies pay full hospital charges, sometimes afteran initial deductible. Other major payers, in-clud ing man y Blue Cross plans, Medicare, andMedicaid, have traditionally reimbursed hospi-tals on the ba sis of the actual cost of providingthe service to their beneficiaries.

To the extent that many insu rers distinguishICU care from other hospital care for pu rposesof reimbursement, the result has been both to

reward ICU care and to penalize intermediate levelspecial care units. For examp le, until 1982, Medi-care paid hospitals different per d iem rates foronly two levels of hospital care—routine care and“special care” (ICU and coronary care unit (CCU)care. ) Levels of care below special care w ere reim-bursed at routine care levels. The other cost-basedpayers h ave tend ed to follow Med icare’s d efini-tional guidelines (166). Also, in 1979 and 1980,

as was noted in chapter 2, Medicare tightened ex-isting payment limits on routine bed costs but noton ICU bed costs—the so-called “section 223limits” (73).

Furthermore, two reimbursement mechanismsdesigned, in p art, to curb u nnecessary utilizationof care (includ ing ICU care) have no imp act on

ICU utilization. These mechanisms—patient co-paym ents and utilization review—are d iscussedbelow.

Patient Copayments

Little is known directly about the effect of di-rect patient paymen ts on the u tilization and costof ICU care. Cullen (56) found that only 100 of 189 seriously ill patients in a Boston surgical ICUwere billed directly for any amou nt. The aver-age bill for patients who did get billed was $1,856,

which was equal to 9 percent of their total hospi-tal bills. Cromwell (49) found that 80 percent of ICU patients in a different Boston teaching hos-pital had direct bills of less than $100, and mostof the patients, 42 percent of whom h ad Medi-care coverage, were well covered for the costs of ICU care. Only 2.5 percent of the sample had out-of-pocket bills above $3,000, and they were re-sponsible for 67 percent of all uncovered hospi-tal charges for ICU care. This pattern may differin other parts of the country where private insur-ance coverage is not as extensive.

Finally, Cromwell found little correlation be-tween coinsurance rates and the u tilization of ICUbeds and ancillary services after the completelyuninsured patients were discounted. He did find,however, that p atients with no insuran ce cover-age had hospital and ICU stays about half as longas those patients w ith more extensive insurancecoverage. Uninsured patients may exhibit a dif-ferent case mix that explains at least part of thedifference in utilization.

Utilization Review

Theoretically, hosp ital utilization review (UR)programs have the potential for limiting hospitalreimbursement for ICU care by denying p aymentsto patients “inap prop riately” in the ICU. In re-

45



— -.

46 qHealth Case Study 28: Intensive Care Units: Costs, Outcome, and Decisionmaking

ality, hospital UR programs, administered for the than on app ropriate level of care w ithin the h os-last decade in accordance with the Federal Pro- pital (234). It wou ld be m ost un usu al for a URfessional Standards Review Organization pro- committee or insurance compan y to d eny pay-gram, have focused almost exclusively on wh ether ments for a patient in the ICU or recommendthe admission is appropriate and whether the transfer to a lower cost unit in the hospital.length of the hospitalization is necessary, rather

PROSPECTIVE PAYMENT PROGRAMS

Evidence is accumulating that State-based hos-pital prospective payment programs have beensomewhat successful in reducing hospital cost in-flation (20,45). There is almost no published data,however, on how ICUs have fared relative toother hosp ital services in States with p rospectivepayment systems.

OTA’S analysis shows that between 1976 and

1981, in the eight States that had established hos-pital prospective payment (rate-setting) d emon-stration programs (Connecticut, Maryland, Mas-sachusetts, New Jersey, New York, Rhode Island,Washington, and Wisconsin), increases in ICU/ CCU beds w ere below the n ational average (4).How ever, some of these States, including Con-necticut and Washington, had relatively highlevels of ICU and CCU beds to begin w ith, so adecreased rate of increase may not n ecessarilybe attributable to the State regulatory paymentsystem.

Indeed, as demonstrated by Cromwell andKanak (48), it is difficult to separate the effect of

the p resence of a State’s hosp ital prospective pay-ment p rogram from man y other factors that mayinfluence hospital costs and utilization of specifictechnologies. These factors include the mix of hos-pitals, the effectiveness of a complementary cer-tificate-of-need program, the length of time theprospective payment program has been in effect,the type of rate review, and the baseline level of costs and services. When they reviewed the period

between 1969 and 1978, Cromwell and Kanak (48)did find that in some States, the presence of a pro-spective payment system appeared to retard thediffusion of ICUs. It should be noted, however,that th eir analysis looked at the d iffusion of in-tensive and coronary units, not at ICU/ CCU beds.By 1969, the majority of hospitals already hadestablished ICUs (205).

There is no systematic information available onwh ether ICU length of stay (LOS) is redu ced inStates with prospective payment programs.

MEDICARE’S CURRENT INPATIENT HOSPITAL PAYMENT SYSTEM

Description

Title VI of the Social Security Act Amendmentsof 1983 (Public Law 98-21) provided a dramat-ically new p ayment system for Medicare inpatienthospital services. A full discussion of the impli-cations of Medicare’s prospective payment sys-

tem for ICU care is beyond th e scope of this casestudy.1 Nevertheless, a few preliminary observa-

tions on likely effects of the system on th e provi-sion of ICU care can be offered.

In brief, the current p ayment system is basedon the concept of diagnosis-related groups (DRGs),Und er this DRG system, which began to be phasedin over a 3-year period on October 1, 1983, hos-

pitals receive a fixed p ayment per discharge basedon the patient’s diagnosis. Hospitals that treat pa-

‘See the Office of Technology Assessment’s technical memoran-dum, entitled Diagnosis Related Groups (DRGs) and the Medicare

Program: Implications for Med ical Technology, which d escribes thepotential impact of the new payment system on medical technol-ogy (254).



—

Ch. 6—Payment for ICU Services q47

tients for less than Medicare’s payments are al-lowed to keep th e d ifference. Those hosp itals thatspend more have to absorb the loss.

More specifically, under the DRG payment sys-tem, ra tes are set for each of 470 different DRGs.

2

More complex DRGs, such as kidney transplants(DRG 302), receive much higher payments thansimpler cases, such as hernia repairs (DRG 161).Certain types of cases with complications or a sec-ondar y diagnosis receive a higher payment thancases without complications. For example, heartattacks with complications (DRG 121) receive asomewhat higher payment than uncomplicatedheart attacks (DRG 122).

The DRG classification system, however, doesnot directly take into account severity-of-illnessvariations of patients who have the same primarydiagnosis. For example, in one teaching hospitala grou p of only four p atients in DRG 206 (dis-

orders of the liver, excluding malignancy, cir-rhosis, alcoholism, and h epa titis, age less than 70withou t complications or comorbidities) had arange of char ges from $1,171 to $114,515 (118).

The U.S. Department of Health and HumanServices (DHHS), which proposed the DRG-basedpayment system, has recognized that within someDRGs, some pa tients may be more severely ill(264). DHHS argu es that in DRGs wh ere sever-ity of illness is strongly associated with treatmentcost, most hospitals will have patients w ho ex-hibit a range of severity levels, thus produ cing on

balance only minor financial adv antages or d is-advantages to most general hospitals. In addition,as enacted, the DRG payment system p rovides foradd itional p ayments in “ou tlier” cases—atypicalcases which have p articularly long lengths of stayor w hich are unusually expensive. For those cases,the ad ditional costs, wh ich m ust ran ge between5 an d 6 percent of the total national payments fordischarges in a year, are based on the marginalcost of care beyond established LOS or cost cut-off points.

Regulations implementing the new law werepublished on January 3, 1984 (75). Under the reg-ulations, a discharge could become either a “day”

‘Although there are 467 DRGs for clinical conditions, there are3 additional categories for payment purposes. Two of these cate-gories involve reassigning the original classification and have norates assigned.

outlier or a “cost” outlier. A day outlier is a dis-charge tha t exceeds the m ean LOS for dischargeswithin that DRG by the lesser of 20 days or 1.94standard deviations. The mean LOS for each DRGare included in the regulations. If the dischargeis considered a day outlier, the hospital will bepaid 60 percent of the average per d iem Federal

rate for the excess days considered medically nec-essary. The 60-percent factor is intended to ap -proximate the m arginal cost of care for the ex-cess days. How ever, a hospital will not be paid60 percent of the actual costs of outlier days, butrather 60 percent of the average DRG per diemrate based on the DRG price.

Additional payments will be made for costoutliers if a hospital requests such paym ent andif the cost of a discharge exceeds the greater of 1.5 times the w age-adjusted Federal DRG paymentor $12,000. Additional payment will equal 60 per-

cent of the difference between the hospital’s adjusted cost for the discharge and the cutoff amount.The adjusted cost will be determined by mu ltiply-ing the billed charges for the covered services by72 percent, the charge-to-cost adjustment factor.Importantly, a d ischarge will not be considereda cost outlier if it qualifies as a day outlier.

DHHS estimates th at initially 5.1 percent of alldischarges will qualify as day outliers and only0.9 percent as cost outliers. Ind eed, DH HS inten-tionally established criteria that would result insubstantially more day outliers than cost outliersfor two reasons: the information necessary to de-

termine day outliers is automatically and routinelyavailable in the bill processing system; and pay-ments to hospitals that may simply be high-cost,inefficient providers of care will be minimized.

Another payment decision in the DRG paymentregulations could have specific relevance to ICUcare. Hospitals transferring a p atient to anotherinstitution are paid a p er diem rate based on theaverage LOS for the DRG treated. Full paymentfor the DRG treated is mad e to the h ospital fromwhich the patient is finally discharged. For exam-ple, if hosp ital A treats a pat ient in a DRG with

an average LOS of 10 days for an initial 4 daysand then tran sfers the p atient to hospital B, hos-pital A will receive 40 percent of the DRG pay-men t and hosp ital B will receive a full DRG pay-ment, regardless of the actual LOS in hospital B.



48 q Health • Study 28: Intensive Care Units: Costs, Outcome, and Decisionmaking

Finally, Medicare p reviously reimbursed hos-pitals for the reasonable costs of capital, whichinclude d epreciation, interest, and rent. Und er thecurrent law, capital expenses are specifically ex-cluded from the prospective payment system andcontinue to be reimbursed on a reasonable costbasis until October 1, 1986. At that time, Con-

gress will decide wh ether to continue to pay r ea-sonable costs or to incorporate paym ent for cap-ital into the DRG system.

Medicare Utilization of ICUs by DRGs

Because ICU p atients often have m ultiple diag-noses and suffer serious p hysiologic abnormalitiesthat frequently do not correspond to diseaseentities, the DRG classification scheme may bepoorly suited to d escribing ICU p atients. Never-theless, a p reliminary analysis has been performedof the DRG case mix of Medicare ICU/ CCU

3pa-

tients based on av ailable Health Care FinancingAdministration (HCFA) data for 1979 and 1980(259,260). For t he pu rpose of th is ana lysis, multi-ple DRGs for the same p rimary d iagnosis werecombined . For examp le, DRGs 121 to 123—myo-cardial infarctions w ith d iffering clinical charac-teristics—were considered together.

Of the 15 DRG-based primary diagnoses withthe longest average LOS in special care in 1979,14 involved operating room procedu res. The ex-ceptions were th e DRGs for m yocardial infarc-tions. Another w ay to view DRG case mix is to

consider special care as a percentage of total hos-pital stay. Of the 16 primary diagnoses in whichspecial care represented at least 10 percent of thetotal hospital stay, 9 were medical diagnoses.However, these medical diagnoses were mainlyfor the cardiovascular system—mostly related tocoronary artery d isease. One can conclud e thatfor DRGs involving certain operating room pro-cedures and coronary artery disease, stays inspecial care units are standard and, therefore, cap-tured by th e DRG category. For examp le, 92 per-cent of cases for card iac valve procedur e w ithpu mp supp ort (DRG 105) included special care.

For man y common su rgical procedu res and car-diac diagnoses, special care was u tilized in m ore

than !70 percent of cases. For the remaining,pred ominately med ical diagnoses, the DRG cat-egory does not reflect the use of special care unitsfor the more severely ill patients w ith that p rin-cipal d iagnosis. For examp le, a number of ICUstudies indicate that gastrointestinal bleeding inpatients w ith cirrhosis is one of the ICU pr oblems

associated with both long ICU lengths of stay andhigh cost (40,50). Yet, the DRG for this condi-tion, “cirrhosis and / or alcoholic hep atitis” (DRG202), has a mean special care length of 0.6 days,or only 4.5 percent of the average total hospitalLOS for discharges with this DRG.

A somewhat different picture of ICU use emergeswhen frequency of diagnosis is taken into account.By multiplying th e nu mber of discharges in the20-percent MEDPAR sample

4by the average LOS

in special care, the nu mber of special care daysby d iagnosis can be estimated . Table 10 shows

the 15 diagnoses which use the most special caredays. Again, cardiovascular disease predomi-nates. However, diseases involving operatingroom procedures become less important as ma-

jor special care diagnoses.

Applicability of DRGs to ICUs

As noted above, the current DRG classificationsystem may not be suitable for describing certaintypes of patients cared for in ICUs. DRGs arebased on a principal diagnosis, with some add i-tional categories available for patients with a

single substantial secondary diagnosis, called a“comorbidity,” or a significant “complication. ”Yet ICU patients often have multiple, seriousun der lying illnesses. In one stu dy (265), ICU pa-tients had on average over four major diagnoses,and the high-cost nonsurvivors had over six diag-noses. For these patients, designation of a prin-cipal diagnosis is likely to be arbitrary andunreliable at times. Furthermore, the additionaldiagnoses wou ld not be accounted for.

As d iscussed earlier, many cardiac diseases,particularly those involving coronary d iseases,and man y of those surgical diagnoses involvingoperating room p rocedu res, includ e stays in theICU and CCU as a matter of routine. For exam-

3Available HCFA data combines ICU and CCU patients as special

care patients.4For a description of HCFA’s MEDPAR data base, see ch. 4.



Ch. 6—Payment ICU Services q 4 9

Table 10.-Estimated Number of Special Care (lCU/CCU) Days by Primary DiagnosisBased on HCFA 20-Percent Sample of Medicare Discharges, 1980

a

Special care Routine care

Percent of totalDiagnosis DRG Total days hospital days Total days

1. Myocardial infarctions . . . . . . . . . . . . . . . . . . . . . ..121-123 176,963 33 ”/0 362,0132. Atherosclerosis . . . . . . . . . . . . . . . . . . . . . ........132-133 103,781 14 625,450

3. Heart failure and shock . . . . . . . . . . . . . . . . . . . . . . 127 87,34711

693,4394. Pneumonia and pleurisy . . . . . . . . . . . . . . . . . . . . . 89-91 78,211 13 555,115

5. Unrelated OR procedure . . . . . . . . . . . . . . . . . . . . . 468 66,451 9 734,684

6. Arrhythmia . . . . . . . . . . . . . . . . . . . . . . . . . . .......138-139 54,464 21 200,923

7. Angina . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 140 53,926 22 194,653

8. Ungroupable . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 470 51,100 6 734,684

9. Cerebrovascular accident . . . . . . . . . . . . . . . . . . . . 14 42,120 5 715,668

10. Chronic obstructive pulmonary disease. . . . . . . . 88 41,203 8 467,825

11. Pacemaker implant. . . . . . . . . . . . . . . . . . . . . . ....115-118 37,109 30 83,58612. Coronary artery bypass surgery . ..............106-107 30,169 32 64,968

13. Pulmonary edema and respiratory failure . . . . . . . 87 28,371 25 83,276

14. Major bowel OR procedure . ..................148-149 27,191 10 242,18815. Major reconstructive vascular procedure . ......110-111 20,543 18 94,077aMultiple DRGs for the same primary diagnosis were combined for this analysis.

SOURCE: Office of Technology Assessment.

pie, in the United States it is stand ard to treat allpatients with acute myocardial infarctions (heartattacks) in CCUs or ICUs. The average DRG priceper discharge will reflect the portion of the hos-pital costs consum ed in the h igher cost special careunit.

How ever, the DRG categories for m any m edi-cal diagnoses are so broad that ICU days repre-sent only a small proportion of total hospital days.For examp le, in 1980, hospital stays for chron icobstructive pulmonary disease (DRG 88) and forcirrhosis of the liver (DRG 202) averaged only

0.82 and 0.60 days of intensive care, r espectively(260). Yet, the sick patients within these DRGsmay spend many d ays in the ICU and u se moretotal hospital resources than p atients within DRGsthat includ e a m uch longer average special carestay. In other words, it appears that variationsin severity of illness are particularly great for non-coronary, medical diagnoses that rep resent themed ical patients in m edical or mixed ICUs. Like-wise, the DRG classification system does notsatisfactorily account for patients with a primarysurgical diagnosis who su ffer major m edical com-plications. For example, in a series of critically

ill surgical patients, Cullen (54) found that renalfailure (a costly medical complication) was apow erful pred ictor of ultimate su rvival. Many cli-nicians might agree that renal failure had becomea patient’s major clinical problem, but the DRG

system requires that the operating room proce-du re take precedence in DRG assignment. Thepresence of renal failure, then, wou ld not signifi-cantly affect DRG payment.

Unfortunately, there is no d ata base av ailableto test whether there are systematic differences byhospital type in severity of illness in ICU popula-tions. DHH S’S initial evaluation found that teach-ing hospitals do h ave higher costs per case, sug-gesting, at least in part, that they treat moreseriously ill patients (75). Survey tapes of theAmerican Hospital Association document that

major teaching hospitals do have 50 percent moreICU days as a p ercentage of total hospital daysthan nonteaching hospitals (106). These add itionalICU days p robably explain som e of the highercosts per case in teaching h ospitals.

However, without an accurate severity-of-illness measure, one does not know w hether theadditional ICU use in teaching hospitals representsthe presence of a sicker population or a differentthreshold for transferring and maintaining pa-tients in the ICU. Likewise, differences in resourceuse between ICUs may represent differences in

severity of illness or differences in intensity andstyle of care. Prelimina ry resu lts from 15 tertiarycare hospitals recently surveyed by Knaus’ groupat George Washington University in Washington,DC, suggest that severity of illness, in fact, ac-



50 q Health Case Study 28:: ]ntensive Care Units: Costs, outcome, and Decisionmaking

counts for a substantial portion of the differencesin ICU resource use for patients with the same pri-mary diagnosis (268).

Under Medicare’s DRG payment system, manycostly ICU cases will likely become out liers forwh om only m arginal costs above a day or cost

threshold are p aid. As was d escribed earlier, bydesign, day outliers will predominate over costout liers. Utilizing HCFA’S 1980 MEDPAR data,OTA has estimated that 12 percent of cases in-volving special care w ould be classified as dayoutliers, in comparison to 9 percent of total cases.By definition, the margina l costs for day outliersare calculated based on the DRG price, not theactual cost for that patient. Yet, as was noted inchapter 3, the cost per d ay in the ICU is over threetimes greater than the cost for a general h ospitalday. Thu s, a hospital may receive far less thanthe actual m arginal costs for caring for a long-

term ICU patient. In short, the outlier paymentrules generally favor less severely ill, non-ICU,long-stay patients, such as those with strokes orcertain types of cancer, over more severely ill,long-stay ICU patients.

It would appear, then, that severely ill Medi-care patients, especially if they are in the ICU,will be “revenue losers” to the hospital, even withan outlier policy in effect. This fact, combinedwith the lack of a financial penalty for transfer-ring patients to a second hospital, may result inmore interhospital transfers of the sickest ICU pa-

tients to tertiary care hospital ICUs. A region-alized system of ICU care that is common in someparts of Europe might thereby be stimulated inthe United States, perhaps desirably. It should benoted, however, that unless either a severity-of-illness m easure or a d ifferent outlier policy isadopted, the tertiary care hospital receivingseverely ill transferred patients will be likely tolose financially. These hospitals would then facethe d ilemm a of either not accepting these patientsin transfer or of accepting these patients into theirhigh -quality ICUs at a financial loss. At the ex-treme, tertiary care hospitals could, in effect,

become large ICUs (212). Public hospitals and

some teaching hospitals, however, may simply beun able to sustain the costs of ICU care and beforced to ration care even more strictly than theydo now (212).

The 3-year cap ital cost exclusion in the DRGlaw is not likely to affect ICUs, at least in the short

run. ICU care is relatively costly largely becauseit is so labor-intensive. Common ICU technol-ogies, such as cardiac monitors, respirators, pul-monary artery (Swan-Ganz) catheters, centralfeeding lines, etc., are labor-generating rather thanlabor-reducing technologies, because they requirefairly constant att ention.

As was noted in chapter 5, the mon itor-onlyand other less severely ill ICU patients have beensubsid izing the care of the m ost critically ill ICUpatients. Under the DRG system, there may bea new incentive to treat monitor patients on reg-

ular floors or perh aps in intermediate care units.In addition, hospitals will attempt to pass on tocharge payers the unreimbursed cost of ICU careto Medicare pa tients. The add itional “p ass-on,”combined with nonadmission and earlier dischargeof some of the less sick ICU patients, should re-sult in substantially increased charges for an ICUday. The current 2.5:1 ratio of ICU bed chargesto routine bed charges (71) will correspondinglyrise.

In short, ICU care to Medicare patients will notbe financially reward ing to hospitals und er DRGpayment. Almost all ICU cases are likely to be

“losers” to the hospital—ICU days are about 3 to3.5 times more costly than non-ICU days and ICUpatients have longer hospital stays than non-ICUpatients. The new incentives of the DRG paym entsystem will be imp osed on an ICU d ecisionmak-ing environment in man y hospitals in w hich thecosts of care had previously been a relatively mi-nor concern. The imp lications of the collision be-tween th e hospital’s new interest in red ucing thecost of ICU care and a d ecisionmaking environ-ment that results in expand ing ICU care w ill bediscussed in chapters 7 and 8.



Ch. 6–Payment for ICU Services q 51

PHYSICIAN PAYMENT

In a fee-for-service system that pays on the basisof “usual, customary, and reasonable” standards,“technological and procedural” med icine has beenrewarded (202). The ICU is a focal point for tech-nological and p rocedural med icine within the hos-pital. Incubation, use of respirators, and arterialline placement are among the many ICU proce-du res that generally require ICU adm ission andnumerous followup ICU visits by the patient’s pri-mary and consulting physicians. Payment for ICUprocedures and visits is generally high and israrely questioned by insurers (166).

Patients in ICUs have m ultiple diagnoses andoften multiple organ system failures. It is not sur -prising, therefore, that ICU patients have manyph ysicians. Murata and Ellrodt (164), foun d in a

large community hospital in which the ICU had

full-time housestaff that at least one physicianconsultation was requested in 65 percent of the

private adm issions. In this study, pr ivate ICU pa-tients had an average of nearly 2.5 physicians car-ing for them, in add ition to rou nd-the-clock house-staff coverage.

The situation is somewhat different in teachinghospitals and other large nonteaching hospitals.In these hospitals, there is usually one or m orefull-time staff physicians who help administer theICU, provide staff education and, to varyingdegrees, participate in direct patient care. Al-

though the specific paym ent method adop ted bya particular hospital may be unique, compensa-tion arran gements can generally be classified intoone of four categories: fee-for-service, percentage

of income arrangements, salary only, and com-binations of the first three (77). Straight salaryarrangements represent the only compensationmethod that d oes not includ e a financial incen-tive comp onen t (77). In term s of ICU care, ICUstaff physicians who are not paid on strict salarybasis have a financial incentive to keep the unitfilled and to perform procedures and provide tech-nical services (166). Surveys on the p revalence of various comp ensation method s in U.S. hospitalshave n ot specifically included ICU physicians (77).Similarly, the extent of ICU physician double bill-ing (submitting fees for reimbursement for pro-fessional services while receiving salaries foradm inistrative and edu cational activities, wh ichare reimbursed as a hosp ital cost) is unkn own .

Under DRG payment, ICU staff physicians may

face conflicting payment incentives unless they arepaid on a strictly salary basis. Given the high costsof ICU care, it may be in a hosp ital’s interest toincrease the cost control function of ICU staff phy-sicians and to pay them salaries as their primaryform of remuneration. Hospitals could even pro-vide incentive bonuses for reduced costs or de-creased lengths of stays. In addition, hospitalswhich do not currently have ICU directors mayfind it in their economic interests to hire one tomonitor the costs of care provided by p rivate phy-sicians w ho ad mit p atients to the ICU. Thu s, itis possible that a hospital’s attemp t to redu ce hos-

pital ICU costs, paid u nd er Part A of Medicare,might also indirectly result in a red uction in PartB physician payments.



7The ICU Treatment Imperative



7 s

The ICU Treatment Imperative

INTRODUCTION

Medical decisionmaking involving seriously illpatients is often difficult and uncertain. In manycases, physicians do not know ahead of timewhether the treatment they prescribe will benefittheir patient. Physicians in the intensive care unit(ICU) frequ ently face the similar dilemm a of notknowing whether to employ available life-supportfor critically ill patients and whether or when towithdraw such support when it seems clear thatcontinu ed treatment will merely prolong h is lifewith no improvement in his grave condition. Onereason ICU decisionmaking is so difficult is that

it is so successful; most ICU patients survive. Yet,it is also clear that in some cases ICU care isprovided—at a very high cost—to patients whoare beyond help. In other cases, ICU care maybe immediately lifesaving but results in returningthe patient either to a cond ition that still has avery short life expectancy or to a condition witha severely limited functional status.

At the p resent time, there is no reliable w ay topredict outcome for most critically ill patients, and

therefore, it is usually reasonable and app ropri-ate to initiate intensive care treatment for severelyill patients. However, because of certain factorssomewh at un ique to the ICU, care is sometimescontinued beyond the point of benefit to thepatient.

This chap ter explores those factors—includ ingthe underlying chronic illnesses suffered by manyICU patients, the diffused nature of decisionmak-ing that often p revails in th e ICU, the frequentinability of patients themselves to make informedchoices about continu ing therap y in the face of

a hopeless situation, the concern over the possi-bility of m alpractice lawsu its or even criminalprosecution and the inability to predict out-come—that often lead p hysicians to provide life-sup port after the initial rationale for doing so nolonger exists. Togeth er, these factors create anICU treatment imperative.

THE HIGHLY TECHNOLOGICAL NATURE OF ICU CARE

The “technological imperative,” which has beendefined by Fuchs as the desire of physicians to d oeverything that they h ave been trained to do,regardless of the benefit-cost ratio (84), flourishesin the ICU. ICU technology can drama tically andconsistently sustain life for long p eriods of time.The ICU is a p rototype of what Thomas h as calleda “halfway technology, ” one that attempts tocompensate for the incapacitating effects of cer-tain diseases whose courses one is unable to af-fect. It is a technology designed to m ake up fordisease or to postpone death (250). Many of theindividual technologies used in an ICU, includ ingrespirators, defibrillators, and balloon pumps,

sustain vital functions but do little to correctund erlying disease processes.

In a well-functioning ICU, patients rarely dieimmediately of respiratory failure or circulatorycollapse, because the available technology can de-lay these complications (50). Some pa tients, par-ticularly those with the common ICU problemsof cardiovascular, respiratory, and necrologicfailure (139) have their vital functions sustainedby technology so as to forestall death, but theirbasic disease or d iseases do not improve. For some

disease processes, then, ICU care does not changethe ultimate outcome, but rather results in a p ro-

.55



——

56 q Health Case Study 28: ]ntensive Care Units: Costs, Outcome, and Decisionmaking

longed, yet inexorable course, with death occur-ring sometimes from complications of ICU care(135) or after a decision is finally made to ter-minate the special life-support.

Measurements and monitoring are often pur-sued as ends in themselves in the ICU (198). Pa-

tient care may become depersonalized. As onecritical care specialist noted, the paradox is thatICU staff treasure life highly and go to any lengthto salvage lives, yet often ignore, or actuallydebase, the very qualities that render patientsuniquely human (35). The technological impera-tive, which frequently results in more effectivemethods of managing very sick patients, can leadto the uncritical adoption of harmful therapies onthe assum ption that the m ost critically ill have lit-tle to lose from new app roaches (198). In addi-tion, new ICU therapies that are demonstrably ef-ficacious in expert hands for specific problems

may become widely adopted an d routinely usedin situations and und er conditions where demon-strat ion of their effectiveness is absen t (243).

Physicians who become intensive care special-ists—“intensivists” —by predilection and trainingare generally believers in technological interven-tion (95). ICU-oriented physicians naturally are

THE NATURE OF ICU ILLNESSES

As was noted in chapter 5, diseases of the car-diovascular, respiratory, and neurological sys-tems, both med ical and surgical, predom inate inthe ICU. Failure of these systems often resu lts inacute respiratory distress, which is manifested bysevere smothering or “air hunger, ” and circulatorycollapse or shock, which results in altered statesof consciousness. Even w hen th e impu lse on thepart of the medical professional is simp ly to makea d esperately sick patient more comfortable andnot to initiate heroic measu res in an attemp t toreverse the illness, that impu lse may requ ire theuse of the full panoply of ICU technologies, par-

ticularly the respirator. Some patients with can-cer and other chronic debilitating illnesses may

believers in the highly complex technology thatthey have mastered and often save lives thatwould have been lost under non-ICU conditions.Likewise, some nurses w ho choose ICU-basedcareers tend to be therapeutic activists, not proneto accepting the inevitability of a p atient’s d e-teriorating condition (278).

The highly technical natu re of ICU care itself affects the way in which life and death decisionsare made. The most critically ill patients havemultiple organ systems failure and receive multi-ple interventions. The very exacting nature of thisform of patient management results in standardprotocols of treatment, p erhaps at increased ex-pense (100), and in concentration on the d etailsof treatment.

In such situations, the fund amental considera-tion of the long-term benefits to the patient r e-

ceiving care is often overlooked among the seem-ingly endless technical decisions that ar e mad ethrou ghou t the course of an ICU stay. Yet, themost critically ill patients, who require the mostconcentrated focus on the details of day-to-dayman agement, are precisely those for w hom fun-damental likelihood and quality of survival ques-tions are most appropriate.

be cared for outside the hospital, perhaps inhospices, with appropriate use of pain medicationand emotional supp ort. Many terminal illnesses,however, produ ce symptom s that cause severedistress to the patient and that are frightening totheir families. The need for relief often resu lts inhospitalization and treatment in the ICU. For ex-ample, symptoms of smothering from emphysemacannot be treated with m edication alone—at least,not w ithout the very real p ossibility of depress-ing the patient’s respiration to the point of risk-ing immediate death (102). “Naturalness,” there-fore, may have to be sacrificed for comfort, which

at times can only be achieved w ith ICU man age-ment and technologies (191).



Ch. 7–The ICU Treatment Imperative q57

“ICU diseases” often develop rapidly-some-times in seconds. When a severely ill, perhaps dy-

ing p atient is seen in an emergency room or ona med ical floor, physicians, who are often not fa-miliar with the patient, naturally and appropri-ately attempt resuscitation (179). Frequently, thebasic physiological and other clinical data wh ichare necessary for a med ical jud gment on the sever-it y and likely outcome of an illness cannot be ac-quired before adm ission to the ICU (55).

With some terminal diseases there is time to an-ticipate and plan the degree and nature of in-

tervention in the event of a sud den deteriorationin the patient’s condition. Because end-stageemphysema, severe heart disease, or generalizedarteriosclerosis are, rightly or wrongly, not con-sidered terminal diseases in the same way that can-cer is, patients experiencing a sudden decompen-sation are routinely and responsibly treated withall available technology. Once initiated, however,treatment modalities that have been initiated pri-marily to respond to acute, disabling symptomsmay become difficult to stop for the reasons de-scribed below.

TRADITIONAL MORAL DISTINCTIONS IN MEDICINE

Along with the notion that p hysicians shouldnot “play God, ” the traditional medical ethic has

been to disregard subjective views of quality of life in making life and death decisions. In termsof ICU care, this general ethic has been char-acterized as one in w hich “survival is being takenas equivalent to a life saved” (64).

Underlying the other considerations which playa part in ICU decisionmaking is the generallyactivist attitude of many physicians, who may em-body a fundam ental and somewh at unique at-titude of Am erican culture. The d ecision to pu llback is frequently more difficult to make th an thedecision to push ahead w ith aggressive sup port,

using the comp lex and sophisticated m edical tech-nology available (269). As other reviewers of in-tensive care have observed, this attitude has beencaptured in T. S. Elliot’s The Family Reunion (18):

Not for the good that it will doBut that nothing may be left undoneOn th e margin of the imp ossible.

In situations wh ere patients are in acute d istressand wh ere decisions mu st be made in seconds orminu tes, there are powerful reasons initially toapply a lifesaving technology. Having done so,it is often quite difficult to reverse the course

of treatment in the light of new information orthoughtful judgment because of traditional moraldistinctions in medicine. Specifically, fundamentalmoral and ethical distinctions are frequently mad ebetween actively causing death and allowing it to

occur by declining to intervene; between w ith-holding and withdrawing treatment; and between

ordinary and extraordinary treatment (191).

Many primary decisionmakers in the ICU feel,

for example, that having decided to put a patienton a respirator, one is committed to its continued

use and thus make a fundamental distinction be-

tween intentionally withholding and actively

withdrawing the respirator. Likewise, while some

prominent medical ethicists have abandoned thedistinction between “ordinary” and “extraordi-

nary” obligations to dying patients, physicians

generally continue to use the distinction (160). The

ordinary-extraordinary distinction has been af-firmed consistently in Catholic moral theology,

notably in a m ajor ad dress by Pope Pius XII in1957 (185). Certainly, most of the public considerthat there is a d ifference between care that is com-mon and reasonably simple and care that isunu sual, comp lex, expensive, and uses elaborate“unnatural” technology. Similarly, most physi-cians consider intravenous fluids to be ordinaryand standard therapy and resuscitation an ex-traordinary measure (160). Other physicians,

however, consider even the use of respirators and

other common ICU technologies to be routine, or-

dinary treatment (191). Many physicians do notuse the ordinary-extraordinary distinction at all,

but rather fundamentally consider whether an

intervention, however invasive, is “medicallyindicated. ”



— .

58 q Health Case Study 28: Intensive Care Units: Costs, outcome, and Decisionmaking

As the President’s Commission for the Studyof Ethical Problems in Medicine and Biomedicaland Behavioral Research has observed, invoca-tion of these moral d istinctions is often so mechan-ical that it neither illuminates an actual case norprovides an ethically persuasive argument (191).Nevertheless, the moral distinctions cited abovehave great importance in ICU decisionmaking. Be-cause many different ind ividuals typ ically partici-pate in ICU d ecisions, a ran ge of moral attitudesare represented (278). There is a natural tendencyto defer to the individual, whether physician,nurse, or family member, who firmly holds atraditional moral view.

In ad dition, there is still uncertainty abou t thelegal interpretation of these m oral d istinctions. Forexample, a New Jersey appeals court recentlyreversed a lower court order allowing removal of a feeding tube in an extremely ill, demented pa-

tient with no hope of recovery.¹ The two courtsdiffered in their interpretation of w hether n aso-gastric feedings (nourishment) constituted ordi-nary or extraordinary treatment for this particu-lar patient in question.

A California app ellate court, on the other hand ,in vacating a lower court’s reinstatement of mur-der charges against two physicians for terminatingcertain treatments for a patient they diagnosed ashopelessly comatose, explicitly rejected the dis-tinction between ordinary and extraordinary care.The court, rather, invoked an ethical measure of

“proportion,” writing:Proportionate treatment is that which has at

least a reasonable chance of providing benefits to

‘See In re Conroy, 188 A/.} Super. 523, 532 (Ch. Div. 1983).

the patient, which benefits outweigh the burdensattendan t to treatment. Thus, even if a prop osedcourse of treatment might be extremely painfulor intrusive, it would still be proportionate treat-ment if the prognosis was for complete cure orsignificant imp rovement in the p atient’s condition.On the other hand, a treatment course which is

only minimally painful or intrusive may nonethe-less be considered disproportionate to the poten-tial benefits if the p rognosis is virtua lly hopelessfor any significant improvem ent in condition.

z

Ironically, as was pointed out by the President’sCommission, if there is any reason to draw amoral distinction between withholding and with-draw ing treatment, it generally cuts the op positeway from the usual formulation: greater justifica-tion ought to be required to withhold treatmentrather than to withdraw it (191). Whether a p ar-ticular intervention will have positive effects isoften uncertain until the therapy has actually been

tried (50). If therapy is initiated and it then be-comes clear that the patient is not benefiting fromit, this is actual dem onstration, rather than meresurmise, to sup port terminating that treatment(191). Yet, physicians who believe in a moraldistinction between withholding and withdraw-ing treatment, or who are concerned that anotherindividual or the courts would judge their actionsbased on this distinction, might choose not to uti-lize the lifesaving treatment in the first place outof concern that the treatment could not subse-quently be readily withdrawn.

‘Neil Leonard Barber, Robert Joseph Nedjl v. Sup erior Court ofthe State of California for the County of Los Angeles; Court of Ap-peals of the State of California, Second Appellate District, Civil No.60350; Oct. 12, 1983.

THE DIFFUSION OF DECISIONMAKING RESPONSIBILITY

Because of the nature of ICU care, many pro-fessionals become important decisionmakers, in-clud ing nur ses who attend to the patient full time,

housestaff, consultants, and the patient’s personalphysician. In larger hospitals, there are frequentlyone or m ore ICU-based p hysicians in attendan cewh o also are involved in the d ecisionmaking p roc-ess. In som e ICUs, particularly in large, teaching

hospitals, the primary legal responsibilit y for apatient’s care is transferred to th e ICU or to anICU-oriented specialist (165,244). often, the pa-

tient’s personal p hysician feels intimidated by theclinical complexity and the bureaucracy perceivedin the ICU and gives up an active role in d ecision-making (136). As a result, the patient’s personalphysician, wh o often has the best und erstanding



Ch. 7—The ICU Treatment Imperative . 59

of the patient’s baseline medical condition, qualityof life, and personal values, goals and concerns,does not participate in important decisions aboutthe care the p atient receives in th e ICU (194). Atthe same time, physicians who do not treat manyICU patients may have unrealistic expectationsabout what ICUs can accomplish and do notknow how or when to ad dress fund amental issuesabout terminating particular kinds of care (243,244).

Many ICU patients enter the h ospital throughthe emergency room, often in a hosp ital wheretheir personal physician, if they have one, is noton staff. Victims of acute trauma or sud den severeillness may not have a p reviously established rela-tionship with the physician(s) who is caring forthem.

ICUs in large hospitals utilize a team approach

to ind ividual p atients, which is felt to result ina higher quality of care (207,208). Some havewondered, however, whether a patient cared forby an ICU team in fact has a doctor (212). Withthe team approach, decisionmaking responsibilitymay be diffused, and the difficult issue of ter-minating special care is frequently d eferred or

deflected. Families who are interested in address-ing this painful issue may not know how to engagea d iverse team of busy professionals in d iscussion.

With multiple professionals in decisionmakingroles, there m ay w ell be different m edical andmoral views expressed. Unanimity among profes-

sionals is desirable, especially when the issue iswithdraw ing life-supp ort (233,243). In such situ-ations, there is a natu ral tendency to defer to amember of the group who holds a traditionalmoral view, such as the distinction between with-holding and withdrawing treatment.

Decisions not to tr eat a debilitated p atient ina nursing home (27), not to transfer to the ICUa patient with end-stage cystic fibrosis (58) or can-cer (253), or to choose a hospice rather than a hos-pital, can often be addressed privately by patientsand their doctors. In the ICU, however, such im-

portant decisions are m ore visible and often con-troversial (278). When the responsible physicianaddresses the issue of termination of special life-support with a family or patient, for example,counterpressures from other physicians and nursesmay make d ecisionmaking extremely d ifficult andemotionally charged.

PROBLEMS OF INFORMED CONSENT IN THE ICU

As the President’s Commission has emphasized,

the voluntary choice by a competent and informedpatient shou ld determ ine whether or not a life-sustaining therapy will be undertaken or con-tinued (191). Unfortunately, the ICU environmentis ill-suited to guaranteeing patient competenceand to providing the necessary flow of informa-tion to ensu re fully informed consent.

Case studies have demonstrated that patientsin ICUs, as well as other seriously ill pa tients, donot always act or communicate in their own in-terest (124). As noted in chapter 5, ICU patientsmay undergo acute psychological reactions tosleep deprivation, sensory overstimulation, de-pend ency, and nearness of death. This is true inother life-threatening situations as w ell, but u nlike

the ICU, there is usually sufficient time and a

satisfactory environment in such cases for work-ing w ith the p atient before taking an irrevocabledecision.

Moreover, ICU patients suffer from subtle, butreal, metabolic disturbances which alter their

jud gment. They are frequently in severe, althoughoften reversible, pain or d iscomfort. Furthermore,a patient on a respirator may be reasonably com-petent to give informed consent but unable tosatisfactorily communicate his or her wishes. Thisis due to the extreme difficulty ICU patients ex-perience in commu nicating, often because theyhave an end otracheal tube in their throat or havehad a tracheotomy, which makes verbalizing im-possible (225).



60 qHealth Case Study 28: Jntensive Care Units: Costs, Outcome, and Decisionmaking

The natur al response to some situations is todefer decisionmaking, particularly with respect toterminating care, until the p atient is able to giveinformed consent. Indeed, physicians may haveto salvage the life of a critically ill patient in or-der to obtain h is or her informed consent to stop

care (69). Medical professionals naturally have abias toward supporting patient survival until itcan be determined that a patient is competent andthat the choice to stop treatment is truly informed(245).

When the patient lacks the capacity to give in-formed consent, the family is normally recognizedas having the authority to make a decision on thepatient’s behalf. In practice, this procedure gen-erally works well (191). Yet, in some cases, familymembers may have motivations wh ich d o notnecessarily supp ort the best interests of the pa-tient (152,244) or, they may d isagree among them-

selves. Again, because of the uncertainty of whoshould make life and d eath decisions on behalf of an incompetent patient, physicians naturallyadop t a policy of continuing intensive care u ntilresolution of disputes and roles occurs.

Finally, it has been n oted that patients w ith seri-ous acute illnesses are generally more passive and

distant from treatment decisions than patientswith chronic stable diseases (150). For many m ed-ical decisions, patients and their families can par-ticipate in decisionmaking with full appreciationof the medical issues involved. Because of thehighly technical nature of ICU care, however, pa-

tients and families may not fully un derstand theimplications of the many decisions that mu st beconfronted in the ICU and are m ore prone to de-fer to ph ysicians (280). In the ICU, the doctor’sorientation toward the pa tient is to be active andin control of the situation, while the patient ispassive and d ependent (280). Some even considerit to be the ICU physician’s responsibility to bringa family to the point wh ere it can look at the p a-tient’s situation from the p hysician’s per spective(278).

Whether the p hysician ad opts a controlling at-titude or not, it may n evertheless take some time

for patients or their families to accept th e fact thatcontinued therapy is hopeless, and the process of informing them of the cond ition p laces the ph y-sician in a d ifficult position. “It is extremely d if-ficult to tell a critically ill patient that all is notgoing well” (232).

LEGAL PRESSURES: DEFENSIVE

The past decade has seen an explosion in thenu mber of malpractice lawsuits brought againstmedical professionals, particularly suits chargingthat a p hysician was n egligent in his or her d utyto p rovide ad equate m edical care. For this casestudy, malpractice is defined as a wrongful act,committed by one or more parties upon anotherperson; the injured party may seek monetarydam ages from the p erson(s) responsible as com-pensation for an injury. The injured party mustdemonstrate that the injury was caused by con-du ct which failed to conform to the “stand ard of care” for that med ical problem an d that class of

provider (199).While many malpractice claims ultimately are

unsuccessful, they have caused doctors and othermedical personnel to become more cautious and,

MEDICINE

in effect, to practice “defensive medicine,” whichinvolves taking or not taking certain action moreas a defense against potential legal liability thanfor the p atient’s benefit (68). Although the extentto w hich defensive medicine is practiced is notknow n, it clearly has contributed to the pr ovi-sion of costly, unnecessary, and sometimes haz-ardous medical care.

Physicians, under certain circumstances, mayalso be subject to potential criminal prosecution,The criminal law confines people’s freedom of ac-tion in order to protect society, not simply in-dividuals, and therefore, consent is never acceptedas a defense against the crime of murder (191).Taking inn ocent hum an life is seen as a w rongagainst the entire society, not just against the deadperson. As such, criminal prosecution is the ex-



— .

62 q Health Case Study 28: Intensive Care Units: Costs, Outcome, and Decisionmaking

mid-1970s (154,193,243,253) in recognition thatnonresu scitation was app ropriate when a patient’swell-being would not be served by an attempt toreverse a cardiac arrest. Yet a recent stud y of non-resuscitation decisionmaking suggested that phy-sicians frequently for-m op inions about a patient’sdesires for resuscitation withou t involving the pa-

tient or the patient’s family, and often physicianstake actions w hich do not conform to the p atient’spreference (15).

For difficult ICU decisions that d o not involveresuscitation, however, physicians and staff maynot have formal procedu res to follow an d th ere-fore must speculate about their potential legalliability. The President’s Commission surveyfound that only 1 percent of hospitals in this coun-try, and just 4.3 percent of hospitals with m orethan 200 beds, have ethics committees to help doc-tors and families reach decisions about withhold-

ing or withdrawing life-support (191). That num-ber may have increased dramatically—to perhaps20 percent of hospitals—in the 2 years since thePresident’s Comm ission su rvey (21). Among theirother functions, these committees may provideinformation to the hospital’s medical staff on th eirlegal responsibilities in certain situations.

In the absence of an institutional m echan ismfor advising staff on the possible legal implica-tions of their actions, physicians, understand ably,tend to adopt a cautious, “defensive” approachto decisionmaking. This is especially tru e in hos-pitals where legal responsibility for care of a pa-tient in the ICU is not turned over to an ICU-based

physician. It is not unusual, for example, for pa-tients who are “brain dead” to remain on an ICUrespirator for days because of unfound ed p hysi-cian or hospital concern abou t a p ossible malprac-tice suit or criminal p rosecution. Physicians mayfeel more confident to disconnect the respiratorif hospital guidelines indicate when it is appro-priate to do so. Even wh en a h ospital does estab-lish wr itten gu idelines and procedur es for mak-ing life and death decisions, however, they arenecessarily cautious and conservative in content(111). Thus, whether physicians decline to with-hold or withd raw ICU technology for fear of legal

liability or whether the institution pr ovides guide-lines, the result is the sam e: the continu ation of ICU care for a longer time than is often necessaryor desirable for the patient’s well-being.

PAYMENT AND THE TREATMENT IMPERATIVE

Methods of hospital and p hysician paym ent de-

scribed in chapter 6 have tended to be permissivefactors for provision of excessive ICU care. It isunlikely that the care is performed primarily toreceive greater income. Rather, the payment sys-tem has not interfered with the factors describedin this section wh ich d o prod uce an ICU treat-ment imperative.

Until 1982, hospitals in States w ithout prosp ec-tive rate-setting programs were reimbursed bysome insurers for actual costs of ICU care and byother insurers according to charge schedules which

permitted hospitals to recoup the costs of caring

for very high-cost, seriously ill ICU patients. Phy-sician paym ents based on a usu al, customary, andreasonable charge system or even on a fixed feeschedule have tended to amply reward physicianswho provide technical ICU services. While the in-centives for the hospital have now changed, atleast for Medicare patients, physician paym entsystems still perm it physicians to pr ovide con-tinued , high-level ICU care withou t d irect con-sideration of the costs to either the patient orsociety.



Ch. 7–The ICU Treatment Imperative q 63

THE ABSENCE OF CLINICAL PREDICTORS

As with many chronic or terminal illnesses,there is an absence of data for the common ICUconditions on w hich to make p redictions of anindividual ICU patient’s chances of immediate sur-vival, as well as the likelihood of his or her long-term survival. Probabilities based on quantitativeinformation for popu lations of similar patients areused as a reference point on w hich to base deci-sions about treatment of patients (277). This ap-proach is common for cancer, for example, wherethere are defined stages of disease and accumu -lated outcome data based on alternative modesof therapy.

Were it possible to p redict wh ich r isk factorsconsistently yield p oor outcomes, man y p atientsmight be considered u nsalvageable at an earlierpoint in their ICU stay (247). With reliable predic-

tors of ICU su rvival, many of the other factorsthat result in excessive ICU care would becomeless imp ortant. For examp le, physicians wou ldhave less concern about legal liability if reliabledata were available to support their clinical judg-ment that special care should be terminated fora particular individual.

It has been argued that the use of predictivescores shou ld have its greatest app lication to d eci-sions involving grou ps of patients or on h ow toexpend societal resources and may have morelimited app lication to decisions involving ind ivid-

ual p atients (157). Unfortunately, accurate quan-titative approaches to clinical decisions are diffi-

cult. Collecting large, accurate data bases is

expensive and time-consuming; verifying theirrelevance to other patient populations is costly

and sometimes not feasible. Data bases can rapid-ly become obsolete for predictive purposes once

new tests or procedures become available (157).

Collecting data on heterogeneous ICU populations

in which diagnostic monitoring and therapeuticintervention often occur simu ltaneously is particu-larly problematic. Yet, unfortunately, as will bepointed out below, purely subjective prognostica-

tion in the ICU is especially uncertain.In recent years, work has begun on establish-

ing quantitative predictive mod els which w ould

aid in predicting outcome of ICU care (143,223,236,247,270). Up to th is point, n o su ch m odel of clinical predictions has been accepted for generalICU use (176). However, the Acute Physiologyand Chronic Health Evaluation (APACHE) scaledeveloped by Knaus and colleagues has begun toreceive particular atten tion as an objective meas-ure of the severity of illness of ICU patients forresearch an d evaluation pu rposes, mu ch as theTherap eutic Intervention Scoring System scale of Cullen (see ch. 5) has been u sed as an objectivemeasure of ICU resource use. A recent simplifica-tion of the APACHE mod el may make this ap-proach more widely useful to help physiciansmake more p recise treatmen t decisions (138). Bydesign, however, the APACHE scale is moreappropriate for predicting outcomes of popula-tions of ICU patients rather than prognosticatingfor individual patients.

A generally reliable predictive model is avail-able in burn u nits, and has been u sed to makedecisions about ind ividual patients (123). Its usein clinical decisionmaking, however, has not beengenerally accepted by experts in th e field (263).

Recently, a scale of rating the likelihood of sur-vival for patients in coma (149) has been devel-oped an d is used in some ICUs for individual deci-sionmaking. For the great majority of ICU patients,how ever, no p red ictive scale is available. Even if

such scales were available, it would be difficultto app ly a popu lation-based scale to individuals(229), especially where a “wrong” decision canhave such profound implications.

For a patient-care area that is as technologicallybased as the ICU, judgments on outcome havebeen remarkably subjective. Subjective prognosti-cation n ear the end of life is notoriously uncer-tain (201) and varied (177). Some feel that physi-cians tend to maintain overly pessimistic prognosesbecause patients with p oor outcomes claim greaterphysician attention (70). Some physicians employa strategy that has been called the “hanging of

crepe, ” i.e., predicting the worst so that anythingless dire will be viewed as a m ajor achievement(229). Others feel that physicians remember the



64 q Health Study 28:: ]ntensive Care Units: Costs, Outcome, and Decisionmaking

rare “miraculous” recovery, forget the more com-mon failures, and act on that faulty memory(280).

Other p roblems w ith ICU outcome p redictioninclude the fact that recognition of terminal pa-tients during an acute admission is difficult (198);

as noted earlier, an acute illness is often not seenin the context of the patient’s overall condition.Furthermore, in many community hospitals, onlya few p hysicians ever hand le a significant nu m-ber of ICU patients. Most physicians have limitedexperience with the relative progn oses of thesevery sick patients (165). Very few hospitals rec-ognize an institutional responsibility to advisephysicians, patients, and families on likely out-comes of ICU care, even for the group of patientswho might be in a vegetative, nonrecoverablestate (191). Opinion on likely prognosis remainsan individual p hysician’s responsibility and , not

infrequently, dramatically different opinions areoffered by th e various physicians involved in aparticular case.

Another major problem is the lack of mean-ingful predictors of the outcome of chronic illness(215). Many ICU patients su ffer an acute, majorICU episode as par t of a deteriorating chronic con-dition, e.g., emphysema, cancer, cirrhosis, or re-nal failure. Often the issue is not the likelihoodof surv iving the acute episode, but rather w hatthe natu ral course of the illness wou ld be evenwith a favorable acute recovery. As was n oted

in chapter 5, it is generally accepted by ICU ex-perts that ICU care does not favorably affect thecourse of a chronic illness, but r ather r everses anacute deterioration in the illness. Some patients,when given information about relatively poor life

expectancy and quality of life, choose not tound ergo temp orary lifesaving treatment. For ex-ample, cancer patients, relying on population-based ou tcome studies, sometimes choose not tosubmit to active cancer therapy. For the mostpart, prognostic indexes, stratified by disease andseverity of illness, do not exist for most other com-

mon chron ic cond itions (158).

Physicians have demonstrated dramaticallydivergent pred ictions of life expectancy for p a-tients with “end-stage” diseases (177). In the ab-sence of data on acute or chronic outcome, phy-sicians can offer only imprecise, qualitativeassessments, i.e., survival is “unlikely,” “unusual,”or “possible,” rather than the quantitative assess-ments, which have probability ranges attached,i.e., “1O to 20 percent chance of one year su rvival”(277).

A fundamental dilemma is that the rare mirac-ulou s recovery d oes occasionally occur . Describ-ing the dismal outcomes of 18 patients treated inan ICU for acute renal failure after rupture of anabdominal aneurysm, Morgan was one of the firstICU specialists to note the problem of high-cost,low-yield ICU care (162). The pat ients w ere el-derly (mean age 65.2 years), with a high incidenceof obesity, chronic pu lmonary d isease, and arte-riosclerotic heart disease. Desp ite energetic clini-cal efforts and dramatically high cost per patient,17 out of 18 died. Looked at another w ay, how-ever, one survived and was able to return to his

previous fun ctional level. A retrospective reviewof clinical records in these cases did not p ermitsuccess or failure of treatmen t to be pred icted byany means other than actual trial.



8.

Foregoing Life-Sustaining Treatment

INTRODUCTION

Chapter 7 described the various, interrelatedfactors that help prod uce an environment in whichexcessive intensive care u nit (ICU) care is some-times provided . The ICU treatment imperative isnow being moderated by two relatively recent de-velopments.

First, there has been increasing recognition of the emotional torment for the patient, the family,ph ysician, and the h ospital staff, of seeminglyendless ICU stays that ultimately end with thedeath of the patient (243,247,278). A growinghumanistic concern for the patient and his family

supports the need to preserve the dignity of a dy-ing patient, and may require earlier cessation of active life-support (18).

Second, there has been a growing recognitionthat the high costs of treating the most severelyill ICU patients may be too high , particularly if they obviously limit th e resources available totreat mod erately sick patients who are more likelyto benefit from intensive care (54,247).

These two developments were explicitly recog-nized by experts in critical care medicine at theCritical Care Consensus Development Conference

convened by the National Institutes of Health(NIH). The Consensus statement on Critical CareMedicine concludes:

It is not medically appropriate to devote limitedICU resources to patients without reasonableprosp ect of significant recovery w hen p atientswh o need th ose services, and w ho hav e a signifi-cant prospect of recovery from acute] y life-threat -ening disease or injury, are being turned away forwan t of capacity. It is inap prop riate to maintainICU management of a patient whose prognosishas resolved to one of p ersistent vegetative state,and is similarly inappropriate to employ ICUresources where no pu rpose will be served but aprolongation of the natural process of death (176).

The NIH statement is significant not only be-

cause it recogn izes the fut ility of ICU care in som esituations but also because it acknowledges tha tICU care is, in fact, already being rationed tosome extent.

A full discussion of the difficult medical, ethi-cal, and legal issues involved in deciding to foregolife-sustaining t reatment either because of the d e-sire to perm it death w ith dignity or because of a need to ration ICU resources is clearly beyondthe scope of this case study. Read ers are referredto the recently published r eport on this subjectby the President’s Commission for the Study of Ethical Problems in Medicine and Biomedical and

Behavioral Research (191). In this chapter, a fewissues of par ticular relevance to ICUs are brieflydiscussed.

THE NATURAL PROCESS OF DEATH

ICUs are uniquely capable of interfering withthe natu ral process of death, since respirators andother ICU technologies are able to sustain vitalfunctions long after the p atient has any chanceof recovery. As a consequence of these lifesav-

ing technologies and the moral considerations in-volved in their use, man y people today d ie “ICUdeaths” rather than natural deaths (135). For ex-

ample, once a patient has been placed on a respi-rator, death may occur on ly when th e physiciansteps in and discontinues its use. Indeed, someICUs are developing p olicies and procedu res for“terminal weaning” off of a respirator, which pro-

vide for the withdrawal of a respirator in a humaneand efficient man ner for the acknowledged pu r-pose of perm itting the p atient to d ie (94).

67



68 q Health Case Study 28:: ]ntensive Care Units: Costs, O utcome, and Decisionmaking

As chapter 7 pointed out, in some situations,such as wh en the patient’s prognosis and w ishesconcerning treatment are not initially known,there may be greater moral justification for per-mitting a death by w ithdr awing life-sustainingtreatment, than for passively allowing death tooccur by withhold ing ICU care in the first place.

The ability of ICU technologies to intervene inthe natu ral death p rocess is also evident in p a-tients suffering from a severe, debilitating, chronicillness wh o can survive their acute illness but w hocan never improve beyond their original unsatis-factory functional status.

In the past, pneumonia was known as “the oldman ’s friend , ” because it often p rovided a rela-

tively quick and painless death for those facingyears of disability (170). Medical technology,however, has largely removed this escape h atch.As a result, extremely elderly patients and patientswith a severe chronic illness frequently survivetheir ICU stays, often at a great expense, only tobe restored, at best, to their p revious state of ill

health. Similarly, patients who suffer a devastat-ing injury and illness that in the past would havebeen fatal are now frequently saved by excellentmed ical skills and mod ern ICU technology, onlyto exist in a permanent state of profound physi-cal or mental impairment (51).

FUNDAMENTAL ETHICAL, MORAL, AND LEGAL CONSIDERATIONS

For certain categories of patients, there has beenconsiderable discussion in medical circles aboutthe extent to which physicians and hospitals shouldbe obligated initially to provide and then to con-tinue ICU care. Many ICU physicians have takenthe p osition, for examp le, that a necessary p re-requisite to ad mission to an ICU is the potentialsalvageability of the patient (51,176,200,238).Some feel that in cases where the patient is clearlymoribund and has no chance of improving, thephysician’s duty is to make the patient comfort-able and not to impose intensive care (51,152).

Although patients’ families may attempt to pres-sure physicians into using the ICU, the physicianand the institution probably would be on safelegal ground in denying such care, assum ing thefacts of the case sustain th eir position and thatthe decisionmaking p rocess was reasonable (51).

A more difficult situation arises when a patientis terminally or irreparably ill, but is consideredto have a chance of surviving the present acutedeterioration (51). In such situations, the fun-damental decision on whether to use life-sustain-ing technology shou ld, if possible, be mad e bythe p atients after they h ave been fully informedof their options and understand their implications(191). A term inally ill patient’s right to forego ordiscontinue life-sustaining tr eatment h as beenestablished, and is usually protected by the con-

stitutional right to privacy (191). An immediateproblem is that the term “terminal” is not a stand-ard technical term with clear and precise criteriafor its definition (12). Physicians may not agreeon w hen an illness is terminal, and some d o noteven use the term.

In addition, as noted in chapter 7, critically illpatients are frequently not competent to make aninformed decision. This is particularly true of ICUpatients w ho su ffer subtle alterations of conscious-ness and develop psychological reactions to their

illness or to the ICU itself. If the patient’s in-capacity to consent is temporary, the decision toforego the use of life-sustaining treatment mayhave to be p ostponed. If the condition is perma-nent, however, the question arises as to whoshould make the d ecision on the p atient’s behalf and on w hat basis (151).

The President’s Commission recommended thatwhen a patient lacks the capacity to make a de-cision—a common ICU occurrence—a surrogatedecisionmaker should be designated. Ordinarily,this will be the patient’s next-of-kin (191). Prob-

lems arise when family mem bers disagree, arethemselves incapable of good decisionmaking, ordemonstrate family interests that conflict with thepatient’s interest (191). In man y instances, an in-capacitated patient may have no family or even



Ch. 8–Foregoing Life-Sustaining Treatm ent q69

close friends who can act as a surrogate on maki-ng d ecisions abou t life-sustaining treatm ent.

At times, there maybe a fundamental disagree-ment between p hysicians and the patient’s next-of-kin on the appropriate treatment for an in-competent, seriously ill patient. When such dis-

agreements cannot be resolved th rough discussionor through a hospital-based forum such as anethics committee, or when the patient is a wardof the State, the issue may have to be resolvedin court. Sometimes, a physician may agree witha surrogate’s decision to forego life-sustainingtreatment, but, nevertheless, seek a judicial rul-ing for fear of criminal prosecution or civil liability(191). It should be emphasized that cases whichmandate specific procedures for determiningwh ether to continue m edical treatment for an in-capacitated p atient have been decided by Statecourts. Therefore, these court-ordered remedies,

which sometimes have differed in significantways, apply only to th e State in wh ich the casewas brought, unless courts in other States specif-ically adop t the sam e analysis. A d iscussion of the decisionmaking procedures mand ated or ap-proved by the courts in situations where a patientcannot choose for himself is beyond the scope of this case study.

It should be noted, moreover, that th ere is legalconfusion even over when a patient ought to beconsidered terminally ill and over w hat constitutes“medical” treatment. A New Jersey app eals court,

for example, overruled a trial judge’s decision thatwould have permitted removal of a life-sustainingfeeding tube from an 84-year-old woman consid-ered to be terminally ill but n ot facing imminentdeath (241). The appeals court found that removalof a feeding tube would have inflicted new suf-fering from dehydration and starvation on the pa-tient. The court found that the State has a “su b-stantial and overriding” interest in preserving thelives of patients who are not m oribund . It alsoseems to have found a legal difference between“nourishment” and “medical treatment .“ “We holdonly that when nutrition will continue the life of

a patient who is neither comatose, brain dead, norvegetative, and w hose death is not irreversibly im-minent, its discontinuance cannot be permittedon the theory of a patient’s right to privacy, or,

indeed, on any other basis. ”1 A New Jersey Su-preme Court review of this finding is pending.

The California appellate court decision in thecriminal case of People v. Nejdl and Barber, inwhich homicide charges were brought againstphysicians who withdrew nutrition in the form

of intravenous fluids and nasogastric feedingsfrom a comatose patient, significantly departsfrom the reasoning used by the New Jersey Ap-peals Court in the case cited above. The Califor-nia case did n ot distinguish between “ord inary”and “extraordinary” care and instead defined theconcept of “p roportionate treatment .“ The courtwrote:

Proportionate treatment is that w hich . . . hasat least a reasonable chance of prov iding benefitsto the patient, which benefits outweigh th e bur-dens attend ant to the treatment. Thus, even if aproposed course of treatment might be extremely

painful or intru sive, it wou ld still be proportionatetreatment if the prognosis was for comp lete cureor significant impr ovement in the p atient’s con-dition.

z

The reasoning of the New Jersey Appeals Courtdecision, wh ich has been app ealed to the State Su-prem e Court, and that of the California App ealsCourt would appear to be irreconcilable. Thus,considerable legal uncertainty rem ains over p re-cisely which medical therapies, if any, are con-sidered rou tine or ordinary and in which clinicalsituations they must be provided. Treatmentsmight be considered mandatory for some patientsbut not for others. Weaning a terminal patientwho is brain dead off a respirator would appearto be permissible, for example, but removing afeeding tube or intravenous line might not. Like-wise, physicians might be legally required to pro-vide different treatment for patients who are seri-ously ill and have no chance for sustained recoverythan for patients who are p ermanently comatoseor w ho face imminent d eath.

ISee lrJ

re Ccmmy, 188 N.]. Super . 523, 532 (ch. Div. 1983 ) .‘See Neil Leonard Barber, Robert Joseph Nedjl v. Superior Court

of the State of California for the County of Los Angeles; Court of Appeals of the State of California, Second Appellate District, CivilNo. 60350; Oct. 12, 1983.

25-338 0 - 84 - 6



70 q Health Case Study 28: ]ntensive Care Units: Costs, Outcome, and Decisionmaking

The use of life-sustaining technology has alsobeen questioned for the patient who is not ter-mina lly ill but w ho finds the qua lity of life un ac-ceptable and without any reasonable chance forimprovement (51,225). Judgments about qualityof life obviously reflect the values and biases of the person making the judgment (152) and there-fore are relevant only if they represent the viewsof the pa tient (148). While some courts have ven -tured into this area (26), there is much less legalprecedent on which to guide physicians about theobligation to provide ICU care for such patients,particularly where the patients are incompetentto d ecide for themselves (170). Because it took years for even the cur rent level of consensus todevelop regarding the p ossibility of foregoing carefor terminally ill patients, one should expect asimilar evolutionary p rocess on the issue of fore-going life-sustaining care for those with an un ac-

ceptable qu ality of life.Beginning w ith the enactment of the Califor-

nia Natu ral Death Act in 1976, 15 States and theDistrict of Colum bia have enacted statutory au -

thorization for competent individuals to write an“advance directive” wh ich directs their physiciansto forego life-sustaining treatment under circum-stances in which they are both incompetent andsuffering from a terminal condition (273). A “proxydirective” designates a surrogate of the patient’schoice to m ake decisions for the p atient if he orshe is unable to do so; it maybe accompanied byan “instru ction d irective” wh ich sp ecifies the typeof care the p erson w ants to receive. In add ition,42 States have enacted “durable power of attor-ney” statutes, which provide authority to appointa proxy to act after a person becomes incompe-tent. Although developed in th e context of prop-erty law, these statutes may be u sed to p rovidelegal authority for an ad vance directive.

There are a number of unresolved issues abouthow advance directives should be drafted, givenlegal effect, and used in clinical practice. Never-

theless, the President’s Comm ission r ecommendedtheir use as a way of honoring patient self-deter-mination (191)0

PROCEDURES FOR REVIEW OF DECISIONMAKING

The mod els of decisionmaking p rocedu res forincompetent patients derived from court opinionsare quite different. The New Jersey SupremeCourt in the Quinlan³ case invoked the presence

of hospital “ethics committees” to provide con-sultation to an incompetent patient’s guard ian andspecifically rejected judicial review of such deci-sions (191). By contrast, the Supreme JudicialCourt of Massachusetts in the Saikewicz

4case ap-

peared to explicitly reject the New Jersey methodof decisionmaking and instead has established

judicial review of these decisions as the rule ratherthan the exception (191). However, in followupdecisions, the Massachusetts court has seeminglymod ified its Saikewicz opinion such that on ly cer-tain categories of cases wou ld ap pear to require

jud icial review, such as wh en the family or the

family and doctors are in d isagreement or w hen

31n re Quinlan, 70 N.J. 10, 35sA. 2d 647, 699, cert. denied, 429U.S. 422 (1976).

4Superintendent of Belchertow n Stat e School v. Saikewicz , 370N.E. 2cI 417,434-3s (Mass. 1977).

the ph ysicians needs, but cannot otherw ise ob-tain, consent to a course of treatmen t wh en thepatient is a ward of the State (272).

The President’s Commission recommended thatresorting to courts should be reserved for occa-sions when ad jud ication is clearly required byState law or when concerned parties have dis-agreements over matters of substantial importan cethat they cannot resolve. The Commission statedthat ethics committees and other institutional re-sponses can function more rap idly and sensitivelythan jud icial review (191).

As was noted earlier, relatively few hospitalshave such ethics committees, and those in ex-istence serve various functions, ranging from for-mu lating policy and guidelines and serving as a

forum for considering difficult ethical problems,to consulting on p rognosis in individual cases and,finally, to reviewing or even mak ing treatmen tdecisions (191). Because of the lack of general ex-perience with ethics committees, the Commission



Ch. 8—Foregoing Life-Sustaining Treatment q71

called for ad ditional evaluation of various formsof formal and informal institutionally based com-mittees before general adoption in all hospitals(191).

It should be noted that over the p ast 15 years,Institutional Review Board s (IRBs) have been set

up to review in ad vance the ethical considerationsof specific research involving human subjects. Al-thou gh initially controversial, IRBs are now gen-erally accepted in the biomedical research com-munity (191).

RATIONING ICU CARE

Up to now, discussion of withholding andwithd rawing ICU care has focused primar ily onthe perceived or actual interests of the patient.

However, the NIH panel has acknowledged thefact that patients who might benefit from treat-ment in an ICU are denied admission because bedsare occupied by patients who do not have a rea-sonable p rospect of “significant” recovery. Earlyanalysts recognized that a few ind ividuals con-sumed a dramatically disproportionate share of ICU resources and suggested that those resourcesbe increased so as to avoid difficult choices aboutaccess (162). The President’s Commission advisedagainst limitations on access to life-sustaining careas an initial part of any cost-containment strat-egy (191). It argued, instead, that the first step

should be the control of “small ticket” tests andtreatments, such as routine blood test and X-rays,which are believed by some to be less cost effec-tive than more dramatic forms of therapy (153),and wh ich can be d iscussed in relatively dispas-sionate terms. Unfortunately, because marginalcosts of ancillary services are much less than aver-age costs, cutbacks on these services are not likelyto have a m ajor imp act on hosp ital costs (1).

In addition, the care of a typ ical high-cost ICUpatient is, to a large extent, an accum ulation of small ticket items. While some efficiencies in ICU

care can be achieved (227), the fact remains thatthe d ecision to initiate and continue ICU care forpatients for whom recovery is un likely, but p os-sible, is one of the major causes of the increasingproportion of the Nation’s health costs accounted

IRBs or ethics committees have been identifiedin the so-called “Baby Doe” controversy, as a pos-sible appr oach to aiding d ecisionm aking aboutdetermining medical treatment for severely hand-icapped n ewborns. While the situation of severelyhand icapped newborns is somewhat different

from that of seriously ill adults, in part becauseit involves consideration of par ental rights andobligations, it may be that the Baby Doe con-troversy will generate a heightened general interestin the role of ethics committees.

for by ICUs. Despite current efforts to makehealth care m ore efficient, it seems clear that fur-ther attempts at cost-containment will encounter

the reality that a large amou nt of med ical careis consum ed by p atients with highly unfavorableprognoses (219).

“It is a basic tenet of our society that we w illnot give up a life to save dollars, even a greatmany dollars” (111). Yet, to some extent, this“lifesaving imperative” is a m yth, since society’sdevotion to saving lives is greatest where thethreat is to identifiable individuals, such as trapped “miners or the victims of catastrophic disease.Society, however, accepts the loss of many “sta-tistical” lives (111), wh ether from the results of

toxic waste or inadequate preventive health care.Physicians usually follow the same lifesaving

imperative. Many health p rofessionals, lawyers,and philosophers have warned that while societymay choose to limit medical treatments for eco-nomic reasons, it is not app ropriate for ph ysiciansto do it for individual patients (17,83,152,266).They argue that th e doctor-patient relationshiprequires an absolute comm itment to do everythingpossible for the ind ividu al patient, regardless of the effect on society’s resources.

However, most physicians by training and

practice accept the fact that there are limits to theresources that society can expend on an y one in-dividual, and in some circumstances they act associety’s agent in balancing the needs of their pa-tient against the needs of other patients and so-



72 qHealth Case Study 28:: Intensive Care Units: Costs, Outcome, and Decisionmaking

ciety as a whole. For example, a patient with in-tractable gastrointestinal hemorrhage does notreceive limitless supplies of blood (152). At somepoint, a p hysician m akes a decision, sometimesimplicitly, that society's interest in having a sup plyof blood av ailable for the comm unity ou tweighsthe patient’s need for continued transfusions. The

threshold for the decision to discontinue transfu-sions obviously varies, depending on factors suchas the patient’s underlying medical problem, age,and perceived life expectancy and the physician’spoint of view.

In less dire circumstances, physicians commonlyweigh the value of a marginal benefit to their pa-tients against the general cost to society. An ex-ample is the support of preventive health screen-ing based on population-related, cost-effectivenessdata. For instance, differences in the recom-mended intervals for screening for cervical can-

cer throu gh u se of the PAP test (103) are essen-tially based on different views of how manymissed cases are acceptable on a cost-effectivenesscalculation. Physicians who choose one standardover others d o so with an imp licit acceptance thatthere is some level of risk that is acceptable foran individual patient.

Physicians in health maintenance organizations(HMOs) may practice a somewhat different styleof medicine, based on the reality of a fixed poolof resources. HMOs face “either-or” choices andmu st decide w hether pa rticular treatments, such

as for catastrophic diseases, are better investmentsthan others, such as for prenatal care (111). While

the constraint of limited resources is imposed ex-ternally, HMO physicians, perhaps unconsciously,may alter their decisionmaking for individual pa-tients in accordance with the reality of limitedresources. While the HMO may exclude or limitcertain benefits explicity in its contract withsubscribers, it also counts on physicians to prac-

tice “cost-effective” medicine, often at a small butmeasurable r isk to certain individu al patients (22).

The clear bias in ICU decisionm aking is to ini-tiate and continue ICU care even w hen it is ex-tremely un likely that th e patient w ill benefit fromsuch care. Nevertheless, because of limited ICUresources, decisions are made every day to cur-tail the care provid ed to ind ividu al ICU patientsand to restrict access to ICUs. In public hospitals,difficult decisions to ration limited ICU beds havebecome commonplace (186). Even in nonpublichospitals, rationing of ICU resources has occurred,

particularly where there has been a shortage of nurses (220,230).

Up to now, shortages in ICU capacity to treatpatients w ho m ight benefit from intensive carehave resulted primarily from internal hospitaldecisions on allocation of beds and other resourcesbetween the ICU and general floors, and not fromexternal economic restraints. As discussed inchapter 6, that w ill no longer be the case, how -ever, under the Medicare’s DRG hospital paymentsystem. In the n ext few years, therefore, muchmore a ttention w ill have to be given to h ow ICU

care should be rationed.

EXPLICIT OR IMPLICIT RATIONING OF ICU CARE?

Explicit Rationing

Provision of ICU care can involve both explicitand implicit forms of rationing. Explicit ration-ing of medical care generally involves direct ad-ministrative decisions on such issues as exclusion

of certain types of services from insurance cov-erage, limitations on the availability of specificmethods of care, preauthorized and concurrentreview and ap proval for expensive treatments and

procedures, required intervals between provisionof specified services, and limitations on totalbenefits (159). In the context of the ICU, explicitrationing might include the establishment of med-ical criteria for treatment based on predictors of outcome for ICU care as they become available.

In add ition to pred ominately medical considera-tions, factors su ch as life expectancy, family role,and social contribution could also be formallyconsidered (196), although the experience of al-



Ch. 8—Foregoing Life-Sustaining Treatment q73

locating rare renal dialysis machines and selectingpatients for kidney tr ansplants in the 1960s on thebasis of social factors was nearly intolerable tothose involved (2,14) and might not be accept-able to society.

The ethical considerations of how to decide

who should receive lifesaving treatment and whoshould not has received attention by bioethicists(13). It is relevant h ere to note th at to avoid ex-plicit rationing for lifesaving treatments, healthplanners and policymakers have tended either toapprove facilities or financing mechanisms thatwill assure treatm ent for nearly everyone with aparticular illness, e.g. end-stage renal disease, orthey make a decision not to facilitate treatmentfor anyone suffering from a certain condition, e.g.patients n eeding h eart transplants (111) exceptperhaps on an experimental basis (122). SinceICUs are not disease-specific, explicit rationing on

the basis of disease would not seem to be anapp ropriate m eans of limiting ICU care.

Explicit rationing of ICU care might also includelimits on covered benefits beyond a certain amount,or in certain clinical situations, wh ere pat ientscould h ave to bear the costs of ICU care directly.Currently, most patients have insurance cover-age for most ICU costs. Many without coveragehave been subsidized. In public hospitals, ration-ing of limited ICU beds has been based largelyon a combination of medical factors, such aslikelihood of successful intervention, and demo-graph ic factors, such as age, and not on considera-tions of ability to p ay (186). There is the real ques-tion of whether society w ould tolerate explicitdenial of “life and death” ICU care on th e basisof insurance coverage or personal wealth. In re-cent years, Congress has considered several pro-posals for national health insurance that wouldextend coverage to everyone for catastrophic ill-ness in order to avoid den ial of care on the onehand and the possibility of extreme financial hard-ship and bankruptcy on the other (72).

Implicit Rationing

Implicit rationing involves limitations on theresources available to health care providers, suchas fixed budgets and restrictions on sites of careor hospital beds (159). These limitations are im-

plicit because they do not specify what servicesshould be provided to w hom or w hat assessmentsph ysicians should make. Instead, they achievetheir effect by placing greater pressures on phy-sicians and hospitals to make hard allocationchoices. Simple reliance on the price mechanismcan also be a rationing device, since everyone’s

ability to pay is limited at some point; for almostall resources in ou r society, price does “ration”access to goods and services. Cost-based p aymentfor insured med ical services has been a notableexception. The new DRG payment system forMedicare is a form of imp licit rationing since thetotal payments allowed under the system arefixed, regardless of the level of services provided.

Other forms of paym ent limits could also re-quire rationing. Indeed, because many p eople lack insurance altogether or have less than full-cost,open-ended coverage, implicit rationing occurs for

many medical services today, particularly non-hospital care. It would be possible to limit totalsocial spending on ICUs (or anyth ing else) throughthe imp licit rationing device of pa tient cost-shar-ing, which does not requ ire adm inistrative deci-sions. Such price-based allocation of resources canbe troublesome, however, w hen ap plied to cata-strophic medical care for a variety of reasons (see111).

The cost of care for the sickest patients in theICU is currently being subsidized to a great ex-tent by those who are not as ill, and by the hos-

pital. DRG fixed payments, which are not adjusted according to the severity of the illness, willoften make high-cost Medicare ICU patients sig-nificant financial “losers” for the h ospital. In thissituation, p hysicians w ill likely feel institutionalpressures not only to alter the style of ICU carethey p rovide to red uce costs, but also to recon-sider the th resholds for w ithholding and with-drawing ICU care from specific individuals. In ad-dition, hospitals may limit or even reduce thenumber of ICU beds, thus reducing access for pa-tients who w ould h ave received h igher cost ICUcare. This form of implicit rationing of ICU careraises a nu mber of questions:

qWhat protections will patients require toavoid arbitrary decisionmaking to limit care?Will certain categories of patients, such as theelderly, the retarded , or otherwise chronically



74 q Health Case Study 28:: ]ntensive Care Units: Costs, Outcome, and Decisionmaking

q

q

dep endent p ersons who might benefit fromICU care, be systematically excluded onpurely economic considerations?

Will the potential threats of criminal prosecu-tion and malpractice suits act as a sufficientcountervailing force to the new incentivesthat DRGs will bring? More specifically, w illthere be a fundamental conflict betweentraditional malpractice standards and newnorms of practice that may involve limitingcare more strictly? Malpractice law has tradi-tionally judged the behavior of m edical careproviders almost exclusively by the custom-ary p ractice of their p eers, rather than by anindependently determined standard of so-cially appropr iate care (22). Malpractice lawgenerally does not recognize varying stylesof care to su it varying available resources.It remains to be seen whether courts w ill rec-ognize limited available resources as a fac-tor in determining negligence. In fact, hos-pitals and physicians may have new incentivesnot to treat very sick ICU-type patients inthe first place, not only because of the d i-rectly negative economic consequences, butalso because it may place them in legal jeop-ardy under existing malpractice standards.Once care has been initiated, the primaryresponsibility of the p rovider is to m eet ahigh standard of care that may not be reim-bursed sufficiently un der the DRG paymen tscheme. Hospitals may decide systematically

to avoid the respon sibility in the first place bydiverting and transferring patients elsewhere.

Will society tolerate different levels of ICUcare based on willingness and ability to pay?Medicare will prohibit hospitals for the mostpart from seeking direct payments from itspatients above the allowable DRG payments

(Social Security Act Am end men ts of 1983,Public Law 98-21). Can a Med icare patientin a life or death situation be denied the con-tinued ICU care he or she d esires and is will-ing to pay for personally, primarily throughprivate insurance, because Medicare pro-hibits patient payments above the DRG

limit? If not, it is likely that d ifferent typesof ICUs will develop, based largely on theability to pay.

qFinally, what procedures should be used toassist ICU decisionmaking in an era in whichat least some patients become financial“losers” for the hospital? A num ber of pro-cedu ral safeguards h ave been p roposed toprotect the interest of patients who h ave in-sufficient capacity to make particular deci-sions on their own behalf, includ ing: 1) nam-ing an ap propriate surrogate to act on th e

patient’s wishes or in the patient’s interest;2) establishing administrative arrangements,such as ethics committees for review and con-sultation of d ifferent d ecisions; and 3) per-mitting ad vance directives, such as livingwills, through which people designate so-meone to m ake health care decisions on theirbehalf, and/ or give instructions about theircare (191). While initially proposed in thecontext of protecting the interests of in-competent patients, these or other proceduralsafeguard s also appear necessary to protectthe interests of competent patients who might

otherw ise be rationed ou t of the ICU. ICUdecisionmaking has been d ifficult when therewas no th eoretical conflict betw een the in-terests of patient, physician, and institution.Under a prospective payment system, patients,physicians, and hospitals may have differentinterests.



9u

Conclusions andPossible Future Steps



9Conclusions and Possible Future Step;

Until passage of the Social Security Act Amend-

ments of 1983 (Public Law 98-21), intensive care

unit (ICU) expansion was able to proceed with-out major consideration of costs because of the

favorable payment environment. Indeed, tight-

ened section 223 limits on costs of routine hospi-tal beds in 1979 and 1980 may have even stimu-

lated ICU expansion. It would seem clear that

Medicare’s inpatient hospital prospective diagno-

sis-related group (DRG) payment system will

cause hospital administrators and ICU directors

to look differently at the costs of ICU care. Un-

fortunately, they will find no easy solutions tothe cost problem, particularly if Medicare allows

only relatively low rates of annual spending in-

creases.

Under DRG payment, some savings may be

generated by better organization and management

of ICUs, perhaps by centralizing separate ICUs

into larger, more general ICUs (212). Arguably,additional savings may be gained by substituting

lower paid health personnel for nurses or physi-

cians to provide certain ICU functions (162,212).

There may be new efforts to find cost-saving tech-

nologies that can substitute for expensive ICU

labor. One ICU, for example, has demonstrated

a significantly decreased ICU length of stay, at-

tributable in part to the use of computer-assisteddecision algorithms (227).

In ad dition, it may be possible in the near futureto predict more accurately which monitored p a-tients do not need to be in the ICU at all. In-termediate care units or other arrangements couldbe developed to care for these patients, probably,at a somewhat lower cost (141) .

At the same time, however, it is now being rec-

ognized that some ICU patients are dischargedprematurely from the ICU. One can argue that

longer stays in the ICU for these patients wouldnot only represent a more ap prop riate use of theICU but also might even save the hospital moneyby reducing the costs of subsequently treating forthese prematurely discharged patients (246).

Nevertheless, the fact remains that relativelyfew ICU patients are responsible for a substan-

tial portion of ICU costs. This case study has at-tempted to demonstrate the clinical, moral, legal,and economic factors w hich currently make it dif-ficult to decide not to treat even those patientswho show little promise of benefiting from ICUcare. The high-cost subgroup is spread among allages, diagnostic group s, and d isability classes (40).There are as yet no d emograph ic identifiers or ac-cepted general p rognostic indicators wh ich p er-mit systematic exclusion of any of the high-costgroup from ICU care. Public progr ams, privateinsurers, perhaps th e pu blic at large, but almostcertainly hospital managers and providers, will

face increasingly difficult decisions about whoshould be given ICU care and in what manner.The process of ICU decisionmaking will become

even more imp ortant when economics may dic-

tate curtailing or even denying care to seriouslyill patients.

A number of steps might improve the environ-ment for intensive care decisionmaking:

Research on developing accurate predictorsof survival for patients with acute andchronic illnesses could be expanded in order

to permit better informed decisions based on

the likelihood of short- and long-term sur-vival. Since the results of outcome data willalways be incomplete and subject to differ-ing interpretations, especially in relation toan individual patient, hospitals might con-sider formalizing an institutional “prognosiscommittee” whose function w ould be to ad -vise physicians, families, and patients on thelikely survival w ith ICU care in ind ividualsituations. Such a committee or hospitalfunction, perhaps u tilizing a routinely up -dated national data base, obviously couldalso provide a similar function for non-ICUpatients.

The suitability of the current D RG methodof payment for ICUs should be tested. If, infact, the DRG scheme takes insufficient ac-

77



7$ qHealth Case Study 28: Intensive Care Units: Costs, Outcome, and Decisionmaking

q

q

q

count of severity of illness, it is likely thatsome hospitals and, consequently, some ICUpatients may face a d egree of rationing thatCongress did not envision.

The legal system, including legislators andthe courts, may need to recognize the possi-

ble conflict between malpractice standardswhich assume quality of care that meets na-

tional expert criteria, and a decisionmakingenvironment in which resources may be se-

verely limited. At the same time, it mu st bekept in mind that the threat of both malprac-tice suits and criminal prosecution maybecome an even more imp ortant protectionagainst arbitrary or unfair denial and ter-mination of ICU care.

Health professionals who are involved inmaking decisions regarding critically ill pa-tients might benefit from more education on

medical ethics and relevant legal proceduresand obligations. In recent years, the journalCritical Care Medicine, published by theSociety of Critical Care Medicine, has in-cluded articles and editorials on specific ethi-cal and legal issues. Likewise, new textbookson critical care m ed icine (224) have devotedchapters to specific ethical and legal issuesthat frequently arise in the ICU. More for-mal education at the gradu ate and p ostgrad-uate level for all health p rofessionals whowork with critically ill patients might be con-sidered.

The actual decisionmaking process for criti-cally ill patients may need greater attention.

At a time w hen the interests of the ICU pa-tient, physician, and hospital were theoreti-cally the same, i.e., under a full-cost reim-bursement system, the need for formal rulesand procedu res for life and death decisionsmight not have been necessary. Even so,many hospitals found the need to establishformal procedures for “Do Not Resuscitate”orders. With a p ayment system that sets theinterests of at least some very sick ICU pa-tients against the immediate financial inter-ests of the hospital, however, it may benecessary to impose additional formal pro-tections on the d ecisionmaking process. Hos-pitals m ight explore formalizing d ecision-making committees or mandating secondopinions to lessen the burden on individualsfaced with excruciatingly difficult choicesabout terminating life-support. Hospitals

could consider formally separating the ICUtriage function from the direct patient carefunction, particularly with regard to the ICUMedical Director, in ord er to m inimize po-tential conflicts of interest. More genera lly,society will need to d ecide how it wishes con-flicts over decisions on terminating life-support to be resolved—in courts, throughformal hosp ital comm ittees such as ethicscommittees, through government-imposedutilization review procedures which can fol-low fixed ru les and-regulations, or other,

haps more decentralized, mechanisms.per-



Appendixes



Appendix Am —Acknowledgments andHealth Program Advisory Committee

ACKNOWLEDGMENTS

The development of this case study was greatly aided by th e advice of a nu mber of people. The auth or

and OTA staff wou ld like to express their ap preciation to the Med ical Techn ology and Costs of the Med icareProgram Ad visory Panel, to the H ealth Program Ad visory Committee, and especially to th e following individuals:

Randy Bovbjerg Richard Lieberman

The Urban Institute SysteMetrics

Washington, DC Bethesda, MD

Martha Brooks Joanne Lynn

Springfield, VA Cliften, VA

Mark Chassin John E. Marshall

The Rand Corp. Na tional Center for H ealth Services Research

J. Richard Crout Michael McMullan

National Institutes of Health Health Care Financing Administration

Palmer Dearing Lois P. Myers

Blue Cross/ Blue Shield Association University of California, San Francisco

Patricia Donovan Joseph Onek

W a s h Onek, Klein, & Farr. . / ,

Betty DraperWashington, DC

The George Washington University Joseph E. Parrillo

School of Medicine-

Mark GoodhartAmerican Hosp ital Assoc

Andria GrenedierAlexandria, VA

Ake Grenvik University Health Center

Stephanie Hadley

National Institutes of Health

Steven A. Schroed er

ation University of California, San Francisco

Douglas WagnerThe George Washington UniversitySchool of Medicine

of Pittsburgh Donald A. YoungProspective Payment Assessment Commission

National Institutes of HealthRebekah Zanditan

William KnausTakoma Park, MD

The George Washington UniversitySchool of Medicine

81



App. A—Acknowledgments and Health Program Advisory Committee • 83

Dorothy P. RiceRegents LecturerDepa rtm ent of Social and Behavior SciencesSchool of Nu rsingUniversity of California, San FranciscoSan Francisco, CA

Richard K. RiegelmanAssociate Professor

George Washington UniversitySchool of MedicineWashington, DC

Walter L. RobbVice President and General ManagerMedical Systems OperationsGeneral ElectricMilwaukee, W1

Frederick C. RobbinsPresidentInstitute of Med icineWashington, DC

Rosemary StevensProfessorDepartment of History and Sociology of ScienceUniversity of Pennsylvania

Philadelphia, PA



Appendix B. —Cost Estimates

As emp hasized in chap ter 3, there are significanttechnical problems in estimating the actual or even therelative costliness of intensive care u nit (ICU) care, Itis essential to recognize some of the m ost importan tdata p roblems that have had to be confronted. First,only charge d ata is generally available. Assum ptionsabout th e relation of charge to cost have been ma deseparately for room and board an d for ancillary serv-ices. Second, national data on the am oun t of inpatientICU care provided is available for Medicare, but n otfor the general pop ulation. In ad dition, there are con-cerns about the reliability of the MEDPAR data base(254). The national estimates have necessarily had tobuild up from this Medicare data base.

Third, standardized national data exists for ICUbeds bu t not for ICU days. Usually, bed occupan cyrates in ICUs are comparable to hospital bed occupan-cy rates in general. We assume, then, that the prop or-tion of ICU days to total hospital days is nearly thesame as ICU beds to total hospital beds.

Fourth, the relevant data bases combine ICU andcoronary care u nit (CCU) care. No attem pt, therefore,is made to d istingu ish ICU and CCU costs. Further-more, the assum ptions un derlying cost estimating forICU and CCU care may not hold for other types of special un its, such as p ediatric, neonatal, and burnICUs. A data base for intermediate care units is simplynot available at all. Therefore, the estimates presentedhere are for adu lt ICU/ CCU costs which und erstatethe costs of more broad ly defined sp ecial care units.As was noted in chapter 2, adu lt ICU/ CCU beds in1982 made up 5.9 percent of hospital beds, while sep-arate pediatric, neonatal, and burn ICUs togethermad e up another 1 percent of beds.

Definition 1—8 to 10 percent: The percentage of

hospital costs represented by the direct and indirect

cost of running the ICU, as reflected in charges for ICUroom and board. The Health Care Financing Ad min-istration (HCFA) has analyzed th e use of and chargesfor accommodation and ancillary services in short-stayhosp itals for Med icare beneficiaries based on a 20-percent sample of Medicare beneficiaries—the MED-PAR data base (112). In 1980, HCFA’S sample showedthat charges for ICU/ CCU care constituted 7 percentof total hospital charg es. Since Medicare p atients’ uti-lization of ICUs is roughly in the sam e prop ortion asnon-Medicare patients (see ch, 4), we assume then thatabout 7 percent of all hospital charges were forICU/ CCU room and board charges. As discussed inchapter 3, charges generally underestimate actual costsof operating ICUs. In one careful study from a singlehospital, the hospital charge for special care room andboard w as found to be only 65 percent of the marginal

cost of maintaining the bed . In contrast, the mar ginalcost for general floor beds was less than the establishedcharge by ap proximately one-third (110). Thus, basedon this and other anecdotal reports, one can conserv-atively estimate that ICU/ CCU costs represented 8 to10 percent of hospital costs in 1980. The proportionof hospital beds devoted to intensive care has, how -ever, increased since 1980. It is likely tha t the p rop or-tion of ICU bed days has increased as well. Therefore,today, the estimate would be at the high end of the8- to 10-percent ran ge or even slightly higher.

Definition 2—14 to 17 per cent: The percentage of total hospital costs consumed by patients when in theICU. This includes room and board and ancillaryservices.

Method A: The simp le appr oach to this estimate isto double the room and board charges—room andboard makes up about 50 percent of total hospitalcharges—and then m ake a charge-to-cost adjustmen t.As noted in chap ter 3, in general, hospitals mark upcosts for an cillary services by almost a third to d eter-

mine charges. Thus, it would not be appropriate tosimply dou ble the cost estimate d erived from th e cal-culations in Definition 1 above. We simply do notknow precisely the ap prop riate charge-to-cost adjust-ments to make for ICU room and board charges andfor ancillary service charges. In addition, data suggestthat ICU patients use m ore ancillary services per d aythan non-ICU patien ts (see ch. 3). The extent of thisadd itional u tilization is not precisely kn own .

If one assumes that the markup for the ancillaryservices and the markd own for ICU room and boardwere rough ly the same and that ICU patients use thesame amount of ancillary services as non-ICU pa-tients—conservative assump tions—the estimate for

percentage of hospital costs consumed by patientswh en in the ICU wou ld be 14 percent, relying on theMEDPAR data for 1980 presented above. If it is assumedthat ICU patients used 20 percent more ancillary serv-ices than non-ICU patients, the estimate rises to 15 per-cent. The recent expansion in ICU beds since 1980might add another 1 to 2 percent. The estimated range,then, is 14 to 17 percent.

Method B: Louise Russell provid ed a method forestimating the total costs of ICU care by relating thepercentage of the total hospital beds th at were ICU/ CCU beds to the relative costs per d ay in an ICU andin a general hospital ward (205). This method assumedthat days of care are proportional to the number of beds. Russell also used a 3:1 ratio for relative costlinessof an ICU day compared to a regular bed d ay. Hermethod, when applied to 1976 American HospitalAssociation (AHA) bed data, p rovides a conservative

84



86 q Health cue Stucfy 28: Intensive care Units: Costs, Outcome, and Decisionmaking

ated w ith a greater use of ancillary services than rou -tine bed d ays. Using the regression, they found thatICU days on average cost about 56 percent more inancillary services than regular days, holding case mix,surgery, insurance status, and other variables constant.While the case mix measu re used (diagnosis and ur -gency of admission) may not be a p recise measure of severity of illness, the regression did confirm that the

ICU days are associated with a dd itional costs in ancil-lary services above those that can be explained by p a-tient characteristics. Again, it is possible that very sick,“ICU-type” patients would have greater ancillary serv-

ices used for their care regardless of their bed location.The 56-percent increment, however, is substantial and,at least, suggests that the ICU itself may h ave beenpartly responsible for the greater use of ancillaryservices.

Griner’s and Cromwell’s work together suggest thatICUS generate incremental hosp ital costs both in ad-ditional d irect ICU costs and in greater u se of ancil-

lary services to achieve similar outcomes as care onregular medical and surgical floors. An estimate of theamount of this cost cannot be provided.



References



98 qHealth Case Study 28::Intensive Care Units: Costs: Outcome, and Decisionmaking

276<

277<

278.

279.

Medicine—Caveat Emptor,” Arch. Intern. Med.143:1391, 1983.

Weinberg, S. L., “Intermediate Coronary Care:Observations on th e Validity of the Concept,”Chest 73:154, 1978.Weinstein, M. C., and Fineberg, H . V., Clinical

Decision Arudy sis (Philadelphia: W. B. Saunders,

1980).Weiser, B., “AS They Lay Dying, ” WashingtonPost [Apr. 17: Part I, 1 (co]. 1), 1983; Apr. 18:Part II, 1 (coI. 2), 1983; Apr. 19: Part III, 1 (co1.

1), 1983; Apr. 20: Part IV, 1 (coI. 4), 1983; Apr.21: Part V, 1 (co1, 5), 1983. ]Williams, S. V., “Can Rand omized Trials of In-tensive Care Meet Ethical Standards?” in Criti-cal Issues in Medical Technology, B. J. McNeiland E. G. Cravalho (eds. ) (Boston: AuburnHouse Publishing Co., 1982).

280.

281.

282.

283.

Wilson, J. B., “Medical Dilemmas in Terminal

Care,” in Death by Decision (Philadelphia: West-minster Press, 1975).Wilson, R. F., “Development and Cost Effective-ness of Su rgical Intensive Care Un its, ” in Crit.

Care Meal,, program abstract from the NationalHealth Institute of Health Consensus Develop-

men t Con feren ce, Mar. 8, 1983.Young, W. R., “It’s a Miracle That We Save Anyof Them ,” Life Magazine, Dec. 2, 1966.

Zimmerman, J. E., “Administrative Structure of a Critical Care Unit, ” ch. 25 in Major Issues inCritical Care Medicine, J. E. Parrillo and S. M.Ayers (eds.) (Baltimore, MD: Williams & Wil-kins, 1984).

Intensive Care Units (ICUs): Clinical Outcomes, Costs, and Decisionmaking

Documents

Transcript of Intensive Care Units (ICUs): Clinical Outcomes, Costs, and Decisionmaking