The Ziggy Theorem: Toward an Outcomes-Focused Health Psychology

Robert M. Kaplan

The purpose of health care is twofold: to make people live longer and to enhance quality of life inthe years before death. These goals are consistent with a Ziggy cartoon that emphasized that themeaning of life was "doin' stuff. .... Doin' stuff" requires being alive (survival) and having thecapability to perform activities. These objectives are quantifiable and can be represented in indicesthat combine life expectancy with health-related quality of life. This article emphasizes patient-oriented outcomes as a focal point for health care. This outcomes orientation is referred to as theZiggy theorem. Examples demonstrate that emphasis on patient-oriented outcomes may redirectconceptualizations of public health indicators and may change the way medical subspecialists makeclinical decisions. Furthermore, the Ziggy Theorem may suggest new approaches to the allocationof public health resources. Y

Key words:outcomes research, quality of life, medical decision making, resource allocation

To be asked to give a presidential address is an honor, an is defined by behavioral functioning, or being able to "do stuff"opportunity, and a challenge. Because I had no training for (Kaplan 1984, 1990, 1993b; Kaplan & Anderson, 1988a, 1990;becoming president or for giving the address, I confronted a Kaplan & Bush, 1982).close colleague and asked, "How do you give a presidential These behavioral conceptualizations of outcome help placeaddress?" His glib response serves as a blueprint for my the objectives of health care into different focus (Kaplan,comments. His advice was this: First, pick an obscure title. 1990). The purpose of this article is to suggest that thisSecond, do not tell anybody what the talk is really about conceptualization can redirect researchers' thinking aboutbecause they may not show up. Third, use a few esoteric several different problems. Specifically, the Ziggy theorem mayhistorical references. Fourth, discuss a lofty philosophical reshape how researchers describe the health of areas or

principle, such as "What is the meaning of life?" countries (health indicators), outcome measurement for clini-My obscure title, "The Ziggy Theorem," is derived from a cal trials, clinical decision making, and resource allocation. To

conversation with David Orenstein, a friend and collaborator review these contributions, it is necessary to present the model.

from the University of Pittsburgh. Orenstein told me that hehad read about my workmin a joke. Actually, it was not in ajoke, but in a Ziggy cartoon by Tom Wilson. In the cartoon General Health Policy ModelZiggy confronts a wise man and asks him, "Tell me, old wise To understand health outcomes it is necessary to build a

one, what is the meaning of life?" The wise man replies, "Ah, comprehensive theoretical model of health status. This modelyes.., the meaning of life. Life, my boy, is doin' stuff!" The includes several components. The major aspects of the modelshocked Ziggy responds, "Life is doin' stuff?. That's it?" The include mortality (death) and morbidity (health-related qual-wise man then reflects, "As opposed to death, which is not ity of life). Diseases and disabilities are important for twodoin' stuff." Ziggy reacts, "It's a more elementary theory than I reasons. First, illness may cause the life expectancy to beexpected, but one you can't argue with!" (See Figure 1) shortened. Second, illness may make life less desirable (health-

For the past 20 years, my colleagues and I have been related quality of life) at times before death (Kaplan &presenting these arguments. We have suggested that the Anderson, 1988a, 1990).purpose of health care is twofold: to make people live for a Over the past 2 decades, a group of investigators at thelonger period of time and to improve the quality of their lives University of California, San Diego (UCSD), has developed aduring the years before death. Quality of life, to a large extent, general health policy model (GHPM). A general conceptualiza-

tion of health status is central to the general health policyThis article is based on a presidential address presented to the model. The model separates aspects of health status into

Division of Health Psychology at the 101st Annual Convention of the distinct components. These are life expectancy (mortality),American Psychological Association, Toronto, August 1993. The functioning and symptoms (morbidity), preference for ob-research was supported in part by Grant PO1 AR 40423 from the served functional states (utility), and duration of stay in healthNational Institute of Arthritis, Museuloskeletal, and Skin Disorders ofthe National Institutes of Health. states (prognosis).

Correspondence concerning this article should be addressed to A model of health outcomes necessarily includes a compo-Robert M. Kaplan, Division of Health Care Sciences 0622, Depart- nent for mortality. Death is an important outcome that must bement of Family and Preventive Medicine, University of California, San included in any comprehensive conceptualization of health. InDiego, La Jolla, California 92093-0622. Ziggy's terms, death is the most extreme and most permanent

Health Psychology 1994,Vol. 13 No. 6, 451-460Copyright ! 994 by the American Psychological Association. Inc., and the Division of Hen th Psychology10278-6133194153.00

451

452 ZIGGY THEOREM

Figure1. Zi_gycartoonabout doin' stuff,©Ziggyand Friends,Inc. distributedbyUniversalPressSyndicate.Reprintedwithpermission.All rightsreserved.

state of "not doin' stuff." Thus, death serves as an anchor been developed (Kaplan et al., 1976). Cross-validation studiesagainst which to evaluate levelsof wellness, have shown that the model can be used to assign weights to

The Quality of Well-Being Scale (QWB; Kaplan & Ander- other states of functioning with a high degree of accuracyson, 1988b) provides a method for estimating some compo- (R2 -- .96). The regression weights obtained in these studiesnents of the general model. The QWB questionnaire catego- are given in Tables 1 and 2. Studies have shown that therizes individuals according to functioning and symptoms, weights are highly stable over a 1-year period and that they areOther components of the model are obtained from other data consistent across diverse groups of raters (Kaplan, Bush, &sources (Kaplan & Anderson, 1990). Applying the QWB Berry, 1978). Finally, it isnecessary to consider the duration ofinvolves several steps. First, patients are classified according to stay in various health states. For example, 1 year in a state thatobjective levels of functioning. These levels are represented by has been assigned the weight of.5 is equivalent to one half of ascales of mobility, physical activity, and social activity. The quality adjusted life year (QALY). The Appendix provides andimensions and steps for these levels of functioning are shown illustrative example of a calculation. Both reliability (Ander-in Table 1. The reader is cautioned that these steps are not son, Kaplan, Berry, et al., 1989) and validity studies have beenactually the scale, only listings of labels representing the scale published (Kaplan et al., 1976; Kaplan & Anderson, 1990).steps. Standardized questionnaires have been developed to The well life expectancy is the current life expectancy ad-classify individuals into one of each of these scale steps justed for the diminished quality of life associated with(Anderson, Bush, & Berry, 1986). In addition to classification dysfunctional states and the duration of each state. Using theinto these observable levels of function, individuals are also system, it is possible to simultaneously consider mortality,classified according to the one symptom or problem that is morbidity, and the preference weights for these observablemost undesirable (see Table 2). Almost 90% of the population behavioral states of function. When the proper steps have beenreports at least one symptom during an average week. Symp- followed, the model quantifies the health activity or treatmenttoms may be severe, such as serious chest pain, or minor, such program in terms of the QALYs that it produces or saves. Aas the inconvenience of taking medication or following a QALY is defined conceptually as the equivalent of a corn-prescribed diet for health reasons. The functional classifica- pletely healthy year of life, or a year of life free of anytion (Table 1) and the accompanying list of symptoms orproblems (Table 2) was created after extensive reviews of the symptoms, problems, or health-related functional limitations.medical and public health literature (Kaplan, Bush, & Berry, In summary, this system combines morbidity (the quality of1976). Over the past decade, the function classification system life) and mortality (the duration of life) with prognosis (theand symptom list were repeatedly shortened until my col- duration in state). An example of an individual patient mightleagues and I arrived at the current versions. Various method- clarify the application of the system. Consider the hypotheticalological studies on the questionnaire have been conducted patient with end state renal disease described in the Appendix.(Anderson, Kaplan, & DeBon, 1989;Anderson, Kaplan, Berry, On the day he was assessed he had general tiredness, weak-Bush, & Rumbaut, 1989). With structured questionnaires an hess, and weight loss. His mobility was confined to the hospitalinterviewer can obtain classifications on these dimensions in 8 because he was on dialysis, and he spent most of the day in ato 15 rain. The classification of functioning formalizes the bed or chair. He performed no major social role but did hismeasurement of"doin' stuff" in the Ziggy theorem, own self-care. The preference weights associated with the

Once observable behavioral levels of functioning have been observable state suggests that peers evaluate the state to beclassified, a second step is required to place each individual on about 0.5 on a 0 to 1.0 scale. If the person remains in this statethe 0--1.0 scale of wellness. To accomplish this, the observable for an entire year, he loses 0.5 well years. Two years in thishealth states are weighted by quality ratings for the desirability state equal the loss of about 1year of health life, or two peopleof these conditions. In other words, the model requires the in this state for 1 year together lose about 1 year of life. If thisquantification of the relative importance of doin' stuff. Human situation was maintained over the course of a decade thevalue studies have been conducted to place the observable person would lose the equivalent of 5 well years of life.states on a preference continuum with anchors of 0 (death) and In summary, the Ziggy theorem emphasizes that the mean-1 (completely well). In several studies, random samples of ing of life is doin' stuff. This is consistent with suggestions thatcitizens from a metropolitan community evaluated the desirabil- the objectives of health care are to make people live longer andity of more than 400 case descriptions. Using these ratings, a to improve the quality of life in years before death (Kaplan &preference structure that assigned the weights to each combi- Anderson, 1990). Methods have become available to quantifynation of an observable state and a symptom or problem has these outcomes. Although very elementary, the Ziggy theo-

ROBERT M. KAPLAN 453

Table 1 traditionallyfocusedonthree major outcomes:lifeexpectancy,Quality of Well-Being-General Health Policy Model: Function infant mortality, and disability days.Scales, With Step Definitionsand Calculating Weights Mortality remains the major outcome measure in most

StepNo. Step definition Weight epidemiologic studies and clinical trials. To make informed

MobilityScale Table2

5 No limitations for health reasons -.000 Quality of Well-Being-General Health Policy Model:

4 Did not drive a car, health related; did not ride -.062 Symptom--Problem Complexes (CPX) With Calculating Weightsin a car as usual for age (younger than 15years),healthrelated,or didnotusepublic CPXno. CPXdescription Weight

transportation, health related; or had or 1 Death (not on respondent's card) -.727would have used more help than usual for age 2 Loss of consciousness such as seizure (fits), -.407to use public transportation, health related fainting, or coma (out cold or knocked out)

2 In hospital, health related - .090 3 Burn over large areas of face, body, arms, or legs -.387

Physical Activity Scale 4 Pain, bleeding, itching, or discharge (drainage) -.349from sexual organs-does not include normal4 No limitations for health reasons -.000 menstrual (monthly) bleeding3 In wheelchair, moved or controlled movement of -.060 5 Trouble learning, remembering, or thinking -.340

wheelchair without help from someone else; clearlyor had trouble or did not try to lift, stoop, 6 Any combination of one or more hands, feet, -.333bend over, or use stairs or inclines, health arms or legs either missing, deformedrelated; or limped, used a cane, crutches, or (crooked), paralyzed (unable to move), orwalker, health related; or had any other broken--includes wearing artificial limbs orphysical limitation in walking, or did not try to braceswalk as far as or as fast as other the same age 7 Pain, stiffness, weakness, numbness, or other -.299are able, health related discomfort in chest, stomach (including hernia

1 In wheelchair, did not move or control the -.077 or rupture), side, neck, back, hips, or anymovement of wheelchair without help from joints or hands, feet, arms, or legssomeone else, or in bed, chair, or couch for 8 Pain, burning, bleeding, itching, or other diffi- -.292most or all of the day, health related culty with rectum, bowel movements, or urina-

tion (passing water)Social Activity Scale 9 Sickor upset stomach, vomiting or loose bowel -.290

movement, with or without chills, or aching all5 No limitations for health reasons -.000 over4 Limited in other (e.g., recreational) role activity, -.061 10 General tiredness, weakness, or weight loss -.259

health related 11 Cough, wheezing, or shortness of breath, with or -.2573 Limited in major (primary) role activity, health -.061 without fever, chills, or aching all over

related 12 Spells of feeling upset, being depressed, or -.2572 Performed no major role activity, health related, -.061 crying

but did perform self-care activities 13 Headache, or dizziness, or ringing in ears, or -.2441 Performed no major role activity, health related, -.106 spells of feeling hot, nervous or shaky

and did not perform or had more help than 14 Burning or itching rash on large areas of face, -.240usual in performance of one or more self-care body, arms, or legsactivities, health related 15 Trouble talking, such as lisp, stuttering, hoarse- -.237

Note. See Appendix for calculation of formulas, ness, or being unable to speak16 Pain or discomfort in one or both eyes (such as -.230

burning or itching) or any trouble seeing aftercorrection

17 Overweight for age and height or skin defect of -.188rern may help researchers to refocus on several problems in face, body, arms, or legs, such as scars,health care and public health. In the following sections, I pimples, warts, braises or changes in colorexplore some of the ways the Ziggy theorem may redirect 18 Pain in ear, tooth, jaw throat, lips, tongue; sev- -.170thinking on these problems, eral missing or crooked permanent teeth

(includes wearing bridges or false teeth);stuffy, runny nose; or any trouble hearing(includes wearing a hearing aid)

Public Health Indicator 19 Taking medication or staying on a prescribed -.144diet for health reasons

U. S. health care is the biggest industry in world history. In 20 Wore eyeglasses or contact lenses -.1011995 the United States is expected to spend an estimated $1 21 Breathing smog or unpleasant air -.101trillion on health care. The United States represents about 5% 22 No symptoms or problem (not on respondent's -.000

card)of the world's population; however, according to estimates by 23 Standard symptom or problem -.257the World Bank and the World Health Organization, more X24 Trouble sleeping -.257than 40% of world health care expenditures occur in the X25 Intoxication -.257United States (World Bank, 1993). One might expect that X26 Problems with sexual interest or performance -.257these greater expenditures would result in better health. X27 Exeessive worryor anxiety -.257

However, the massive health care industry has had no way to Note. X -- weight not available at this time, so standardized weight ismeasure its main product, health. Public health analysts have used.

454 ZIGGYTHEOREM

decisions about the nation's health, the U. S. Congress receives life (World Bank, 1993). Although the disability adjusted lifevarious reports of statistical indicators from the National year is conceptuality and methodologically problematic, it doesCenter for Health Statistics. These include the crude mortality represent a significant advance toward the qualification ofrate, the infant mortality rate, and years of potential life lost. world health needs.Although important, each of these measures ignores dysfunc- The Ziggy theorem can help researchers rethink largetion while people are alive. The National Center for Health problems, such as the conceptualization of health care goals.Statistics reports information on various states of morbidity. However, it can also stimulate new thinking about the practiceFor example, it defines disability as a temporary or long-term of medical subspecialties. I now address the problems ofreduction in activity. Over the past 2 decades, medical and specialty medicine.health services researchers have developed new ways to assesshealth status quantitatively. These measures are often called Beyond the Disease-Specific Focus

quality-of-life measures. Because they are used exclusively to Medical specialization is not a new phenomenon. In fact,evaluate health status, the more descriptive health-related Herodotus described over-specialization in ancient Egypt. Hequality of life is preferred (Kaplan & Bush, 1982). Some suggested that

approaches to the measurement of health-related quality of medicine is practiced amongthe Egyptianson a plan of separa-life combine measures of morbidity and mortality to express tion. Each physiciantreats a singledisorderand no more. Thus,health outcomes in units analogous to years of life. The years the countryswarmswithmedicalpractitioners,someundertakingof life figure, however, isadjusted for the diminished quality of to cure diseasesof theeye,others,of the head,othersagain,of thelife associated with disease or disability (Kaplan & Anderson, teeth, others,of the intestine.(Herodotus, trans. 1942).

1988a). Medicalspecializationhasimportantclinicaladvantages.InThe worldwide pressure to use resources efficiently has particular, it ensures that the patient will see the provider most

created a demand for new health indicators. The QALY serves familiar with his or her problem. However, there are alsothis purpose because it combines morbidity and mortality, disadvantages. Focus on a particular outcome may obscure theFurthermore, the QALY solves the complex problem of total benefit of treatment. This is well illustrated by clinicaltrade-offs between length of life and quality of life. When the trials on arrhythmia suppression following heart attacks.concept of a QALY was introduced by our group at UCSD People who survive a heart attack are at elevated risk formore than 20 years ago, it received neither praise nor criticism, sudden death. One of the established risk factors for sudden

In fact, there was little reaction at all, and it was so foreign that death is asymptomatie irregular heartbeats (asymptoraaticit was difficult to place papers in the peer-reviewed literature, ventricularpremature depolarization in cardiologists' terminol-However, by the late 1970s conceptual advances had been ogy). Several drugs have been shown to suppress these irregu-made by our group (Kaplan et al., 1976) and others (Torrance, lar heartbeats. Two of the most successful ones are Encainide1987). Furthermore, economists and public policy analysts and Flecainide. Although these two drugs were already inbegan using QALYs as a way to appraise clinical treatments widespread use for the treatment of patients with arrhythmia,(Weinstein & Stason, 1977). Nevertheless, the QALY concept the National Institutes of Health (Cardiac Arrhythmia Suppres-was not known outside of a few academic circles until the late sion Trial Investigators, 1989) conducted a clinical trial to1980s. demonstrate the obvious: that medicines that control the

More recently, QALYs have experienced a boom in popular- biological problem will normalize the risk of early death. Theity. This was aided by suggestions in England that QALYs study was known as the Cardiac Arrhythmia Suppression Trialserve as the basis for health public policy (Williams, 1988) and (CAST) and involved a large number of investigators andby greater recognition of these concepts by the United States medical centers. To be eligible for the study, patients had togovernment. In the construction of the United States' health achieve significant suppression of their arrhythmia while oncare objectives for the year 2000, it became apparent that medication. Among the 2,309 patients recruited, 1,727 (75%)researchers could not expect significant changes in either life achieved a benefit from the drug. These patients were thenexpectancy or infant mortality rates. The QALY concept was randomly assigned to a group that used one of the drugs or to aadvanced as a better way to conceptualize health care objec- placebo group. Many observers believed it was unethical totives. When the Department of Health and Human Services even conduct the trial. The logic was that patients withreleased Healthy People 2000, (1991) the blueprint for U.S. arrhythmia were at risk for death--the drugs suppressed thehealth care policy for the year 2000, the overall objective was to arrhythmia--and therefore, not allowing all patients to use theincrease the years of healthy life for the United States medications would condemn those in the placebo group topopulation. The issue of healthy life is conceptually identical death. Because of a meaningful biological criterion, the treat-to the QALY, and the baseline data were ascertained by using merit worked and there was no need to experiment further.estimates from the QWB. The Ziggy theorem demands other data. It is indifferent to

Calculating QALYs requires measurement of health status, the electrophysiological changes resulting from the treatmentOther methods have evolved that allow for these estimates of an asyrnptomatic condition. Instead, the Ziggy theoremfrom less complete data sources. Most notable is the attempt demands that the outcomes be meaningful to patients. In thisby the World Bank and the World Health Organization to ease the meaningful outcome was life expectancy. There was aestimate health needs for the entire world. They have done this significant differenco in mortality in the CAST trial, but it wasusing the disability adjusted life year, which is an attempt to in the wrong direction. Compared with those taking a placebo,combine morbidity and mortality into a unit similar to a year of patients taking Encainide and Flecainide had a significantly

ROBERTM. KAPLAN 455

higher chance of dying of cardiac arrest and to have nonfatal 18 mWaitlngcardiac arrests that disrupted their lives (CAST Investigators, 16 =_dl_ion ]1989).A biologicalmodel that focusedon arrhythmiasuppres- 14 iProltatectg, my[sion would have concluded that the drugs worked. Bythe Ziggy 12

theorem the drugs were dangerous and should be withheld, m 10Focus on specific categories can obscure the most important _ a

behavioral outcomes. Research directed toward specific dis- 6ease categories or aspects of a biological process may not 4capture global concerns about health, That task requires a 2comprehensive behavioral model. 0

Clinical Decision Making 60 e5 70 _'sAGE

The Ziggy theorem may help to refocus some clinicaldecisions. Physicians who emphasize specific biological pro- Figure2. Quality adjusted lifeexpectancy(QALE) under the threetreatment options for men with moderatelydifferentiated tumors atcesses may miss the total health outcome picture. To illustrate different ages (data from Fleming, Wasson, Albertson, Barry, &this issue, consider the problem of prostate cancer in older Wennberg,1993).men. Prostate cancer is an extremely common cancer for menage 70 years and older (Holleb, Fink, & Murphy, 1991). Thetreatment of this disease varies dramatically from country to suggests that the alternatives are equivalent and that thecountry and within regions of the United States. For example, choice should be a matter of patient preference. According tosurgical removal of the prostate gland is done nearly twice as the Ziggy theorem, waiting is a legitimate option.often in the Pacific Northwest as it is in New England. Yetsurvival rates and deaths from prostate cancer are no different Provider Paymentin the two regions (Lu-Yao, MeLerran, Wasson, & Wennberg,1993). The traditional approach to the treatment of prostate Another way that the Ziggy theorem can redirect thinking iscancer was to identify the tumor and to get rid of it through in relation to provider payment. Historically, the United Stateseither surgery or radiation therapy. Success is defined either has operated on a fee-for-service system. Physicians and otherthrough improved life expectancy or through evidence that the health care providers are reimbursed for offering services,man is tumor free. Formal analysis using QALYs can shed a whether those services cause benefit or harm to the patient.different lighton these approaches. One of the problems in the health care system is that there

There are three main options for the treatment of prostate have been incentives to offer higher cost services or to delivercancer: radical prostatectomy (surgical removal of the prostate services when they are unnecessary. Under the growth phasegland), external beam radiation therapy, and "watchful wait- of the Medicare program in the 1960s and 1970s, physiciansing." Both radical prostatectomy and radiation therapy carry were reimbursed on the basis of customary community charges.high risks of complications that may reduce life satisfaction. In effect, this allowed physicians to bill the government forFor example, there are significant increases in the chances of essentially any service they wanted to offer to older patients.becoming impotent or incontinent. Watchful waiting, however, There was a large increase in the number of services offereddoes not require therapy, but only evaluation and supervision and a striking increase in the use of services that wereby a physician. The watchful waiting option has been used the associated with large fees (Enthoven & Kronick, 1990). AI-least because it does not treat the cancer, though physicians and hospitals profited, there was little

One recent analysis estimated QALYs under these three evidence that patients benefited.options. The analysis assumed a cohort of 68-year-old men and This problem is not specific to the contemporary Unitedused published studies to estimate outcomes under different States. In fact, George Bernard Shaw in the preface to his 1911treatment options. For men diagnosed at this age, the risk of play The Doctor's Dilemma suggesteddistant metastases is relatively rare: These outcomes occur in 5 "that anysamenation havingobservedthat youcould provideforof every 100 patient years of observation. The median time to the supplyof breadbygivingbakersapecuniaryinterestinbakingmetastasis is 14years, and during this period 58% of the men for you, shouldgo on to givethe surgeon a pecuniary interest in

cuttingoff your leg, is enough to make one despair of politicalwould be expected to die of other causes. For those men who humanity."(p. v)do develop metastases, hormonal therapy is available and canefficaciously delay disease progression until a larger percent- In contemporary medicine it is clear that pecuniary interestsage of the men die of other causes (Fleming, Wasson, come to influence many medical decisions. Perhaps this is bestAlbertson, Barry, & Wennberg, 1993). Figure 2 summarizes illustrated in the study by Hillman, Pauly, and Kerstein (1989),quality adjusted life expectancy under the three treatment who studied the mean cost per episode for the evaluation ofoptions for men with moderately differentiated tumors at four clinical decisions by family physicians (low back pain,different ages. The figure shows the quality adjusted life difficulty in urinating, upper respiratory infections, and preg-expectancy is essentially equivalent under the three options, nancy). The independant variable in Hi,man et al.'s study wasThe small differences between approaches are not statistically whether the family doctor owned his or her own X-raysignificant (Fleming et al., 1993). The traditional view empha- equipment. For each of these conditions, X-rays are commonsizes tumor eradication and aggressive treatment. This analysis evaluative tests, and physicians who do not own their own

456 ZIGGY THEOREM

equipment can refer the case out to a radiologist. Cases in the survival for the patient groups as close to the general popula-study were carefully matched for severity. The results demon- tion or actuarial curve as possible.

strated that there was significantly more expense in each of the Figure 4 summarizes the same analysis but with quality-four diagnostic categories if the doctor owned his or her own adjusted survival replacing survival on the y-axis. There is a

x-ray equipment. It is often argued that the main reason significant difference in QALYs between medical and surgicalphysicians order unnecessary tests is to protect themselves options, with surgical treatment coming closer to the actuarialfrom malpractice lawsuits. However, in the Hillman et al. line. Figure 5 summarizes a similar analysis for a differentstudy, the fear of being sued should have been equal in the two patient. For this patient a medical option appe2rs superior togroups of providers because the patients they cared for were surgery. Diamond et al. (1993) suggested that the physicianmatched. What differed was the pecuniary interest in provid- should be reinforced (or reimbursed) for selecting the surgicaling the test. According to the Ziggy theorem, physicians should option for the patient who will benefit most from surgery andbe reimbursed for helping people to live longer and for helping the medical option for the patient who will benefit most fromthem to do stuff. This may or may not be related to how much medicine. In other words, provider payment should be strut-providers are now paid for their services, tured to maximize patient outcome.

Recently, Diamond, Denton, and Matloff (1993) suggested On average, more patients are expected to benefit from

that the fee-for-service model be replaced by a fee-for-benefit medicine than from surgery. However, surgical options aremodel. The fee-for-benefit model estimates the QALYs pro- now used more than medical treatments. Thus, the implemen-duced by different treatment options for individual patients, tation of this system might lead to lower surgical rates. ADiamond et al. suggested that the expected benefit for each computer simulation of cost-effectiveness using these datapatient can be estimated on the basis of available clinical suggested that quality adjusted survival for coronary heartinformation. Physicians should then be given maximum reim- disease patients would increase 12%. Payments to providers

bursement only if they select the option most likely to benefit would decrease 22%, but cost-effectiveness would increase bythepatient. 55%(Diamondet al.,1993).

The system proposed by Diamond et al. (1993) has been The fee-for-benefit model is both innovative and bold.worked out most clearly for patients with coronary heart However, it is not necessarily new or more extreme than ideasdisease. For this illness, a computer model estimates the risk of proposed in the past. Perhaps the most extreme version was

death for each patient. The model considers measured vari- suggested by Hammurabi in ancient Babylon. According toables including risk factors for heart disease, such as age, Hammurabi,gender, blood pressure, and cholesterol. In addition, the

Ira surgeon has made a deep incision in the body of a man with amodel uses common physiologic perimeters based on labora- lancet of bronze and saves the man's life or has opened an abscesstory tests and clinical examination, in the eye of a man and has saved his eye, he shall take ten shekels

For each patient, it is possible to estimate the life expectancy of silver. If a surgeon has made a deep incision in the body of aunder various treatment options. Life expectancy data are man with the lancet of bronze and so destroys the man's eye, theysummarized in Figure 3. The top line for actuarial survival shall cut off his forehand. (cited in Rosser, 1993,p. 315.)

gives the probability of survival for the general population at The relationships between benefit and provider payment arethe same age. The other two lines summarize the probability of summarized in Figure 6. The flat line in the figure (labeledsurvival expected for those who receive medical or surgical hospital) shows the relationship between payment and benefittreatment. The goal of therapy is to bring the probability of from the hospital's perspective. Hospitals would prefer to be

paid the same rate whether or not they benefit the patient. The

1.0

% .6 .8

Alive.4 _ .6.2 Medical -- QWI_'4 u i -

.0 .2 Medical

012345678910 II 12.0

YEARS 0 1 234567891011 12

Figure3. Outcomes of treatment options for coronary heart disease. YEARSThe top line for actuarial survival gives the probability of survival forthe general population of the same age. The other two lines summarize Figure 4. Outcomes of treatment for coronary heart disease forthe probability of survival expected for those who receive medical or specific patient likely to benefit from surgery as analyzed usingsurgical treatment without adjustment for quality of life. quality-adjusted survival. QWB -- Quality of Well-Being Scale.

ROBERT M. KAPLAN 457

1.0 but mightbe paid increasingamountsas treatment outcomes

improve.

.8 Thefee-for-benefitnotionhasimportantphilosophicalroots.For example, many argue for an egalitarian system in which the

QWI_6 objectiveis the right to access. Indeed, many proposals forhealth care reform focus exclusively on the rights to all services

.4 __ (Kaplan, 1993a, 1993b). Another philosophical approach is thelibertarian position that argues that the major objective is the

.2 Surgical right to choice. Diamond et al. (1993)emphasizedthat right tobenefit improves health status and encompasses many of the

.0 ideasof right to choiceand right to access.The Ziggynotion

0 1 2 3 4 5 6 7 8 9 10 11 12 that doin' stuffis a high priority suggests that physicians shoulduse their resources to help people get better. Change in

YEARS payment for individual clinical decisions may provide a re-sponse to this challenge. In the next section a similar strategy

Figure 5. Outcomes of treatment for coronary heart disease for at the public policy level is considered.specific patients likely to benefit from medicine as analyzed usingquality-adjusted survival. QWB = Quality of Well-Being Scale.

Resource Allocation

stepped line summarizes the relationship from the payer's The Ziggy theorem has also been proposed to help resolveperspective. The payer might prefer to offer no reimbursement the United States' current health care crisis. There are at leastwhen treatment does not work and pay a standard low rate if three major problems with the current health care system.payment is effective. The line labeled patient summarizes these Elsewhere I have described these as the three As: affordability,relationships from the patient's perspective. Patients may not access, and accountability (Kaplan, 1993b). The system costswant to pay for services that provide no benefit, but may be too much, and the accelerating costs have the potential to ruinwilling to pay increasing amounts as outcomes improve. The the entire economy (affordability problem). Despite these highfinal line shows a potential provider perspective. Providers costs, there are significant numbers of people who have tooshould be paid something for their efforts in all circumstances, little or no health insurance (access problem), and the United

22 ' s,_,$

20 £ Hospital s"SS ,_' S

18 - Provider ss a?

''' Payor sSS ._16 ==,====-' Patient s S $"

14 s SJ

S

E 12 /

10 ••s L/*"_ s SO. 8 s"

J, /6 sl S ,"

4 ss S _s2 sS _

I IIA

U • I • I • I • l • I • I • I • I ' I " I • I •

0 2 4 (1 8 10 12 14 16 18 20 22 24

Outcome

Figure 6. Relationship between payment and outcome from multiple perspectives. Hospitals preferpayment independent of outcome. Payers may want to avoid payment if there is no benefit and may bewilling to pay a fixed amount for effective treatment. Patients may not want to pay for ineffectivetreatments, but may pay an increasing amount for more effective treatments. Providers may wantcompensation for their services, but may accept higher fees for more effective treatments.

458 ZIGGY THEOREM

States is unable to demonstrate that its high expenditures on medical treatment, which was a generic term for treatmenthealth care result in better outcomes for patients (accountabil- with medication.ity problem). In particular, a significant number of procedures, The Health Services Commission obtained several sourcesas many as 30--50%, may have nonsignificant effects on health of information. They held public hearings to learn aboutoutcome (Brook & Lohr, 1986). Denying payment for these preferences for medical care in the Oregon communities.valueless procedures would save a significant amount of These meetings helped clarify how citizens viewed medicalmoney, perhaps enough to provide basic care for a greater services. Various approaches to care were rated and discussednumber of people without increasing the cost. To solve the in 48 town meetings attended by more than 1,000 people. Fromhealth care crisis the three As must be addressed simulta- these meetings, 13 community values emerged. These valuesneously. The state of Oregon offered one such proposal, included prevention, cost-effectiveness, quality of life, ability

Oregon, like nearly all states, rations health care. Yet to function, and length of life. One major lesson from theOregon is different because it recognizes that rationing is community meetings was that citizens wanted preventiveimplicit and not open to public scrutiny. Medicaid has a finite services. Furthermore, the people consistently stated that thebudget, and the costs of the program have grown much faster state should forgo expensive heroic treatments for individualsthan the available resources. In the late 1980s, the costs of the or small groups to offer preventive services for everyone.Oregon Medicaid program grew at a rate of 18% per year. In To pay for preventive services, it was necessary to reduceresponse to financial pressures, the eligibility criteria needed spending elsewhere. A major portion of the commissioners'to be revised. In other words, people were being rationed activity was to evaluate services with the QWB from therather than serviced. Medicaid recognizes only some catego- GHPM. The commissioners could not have possibly conductedries, such as Aid to Families With Dependent Children and clinical trials for each of the many condition-treatment pairs.those elderly, blind, or disabled individuals receiving supple- Furthermore, estimation of treatment benefit using the QWBmental security income. Many individuals in need of care cannot be left to laypersons. Therefore, the commissionreceived none because they were in the wrong category. A formed a medical committee that had expertise in essentiallyyoung woman employed as an hourly worker, for example, may all specialty areas and had the participation of nearly all of thebe ineligible for health care, whereas an unemployed twin major provider groups in the state. Working together, thesister would become eligible if she became pregnant. Thus, the committee estimated the expected benefit of 709 condition-system creates incentives to become pregnant to have a regular treatment pairs.source of health care. Because of a change in the criteria for The QWB, which was used to estimate the effect of services,children born after 1983,the system offered Medicaid to poor requires subjective judgments to score the importance orfamilies with young children but disallowed coverage for poor desirability of health states. These weights are not medicalfamilies with older children. Oregon, like many other states,defined Medicaid eligibility for the Aid for Families With expert judgments and should be obtained from communityDependent Children as 50% of the poverty line. In 1989 that peers. The Oregon citizens were particularly concerned aboutpolicy set the criterion income at about $5,700 per year for a using weights from California to assign priorities in their state.family of three. A hard-working independent carpenter earn- Thus, 1,001 Oregon citizens participated in a separate weight-ing $11,000 per year might have been excluded by the policy ing experiment. The weights were obtained in a telephoneeven though he or she was at high risk for injury, survey that was conducted by Oregon State University. In 1990

These problems were brought to public attention by a the commission released a draft copy of its first prioritized list.grassroots citizens groups known as Oregon Health Decisions. Unfortunately, many of the rankings seemed counter-intuitive,This group brought attention to the problem by organizing and the approach drew serious criticism in the popular press.more than 300 community meetings throughout the state that As a result, the system was reorganized according to threewere attended by more than 5,000 Oregonians (Crawshaw, basic categories of care: essential, very important, and valuableGarland, Hines, & Lobitz, 1985). The movement gained the to certain individuals. Within these major groupings thereattention of the state legislature and, in particular, that of John were 17 subcategories. The commission decided to place theKitzhaber, the physician president of the state senate. In greatest emphasis on problems that were acute and fatal. Inresponse to this problem, Oregon passed three pieces of these cases treatment prevents death, and there is full reeov-legislation, including the controversial Senate Bill 27, that cry. Examples include appendectomy for appendicitis andmandated that health services be prioritized using a process nonsurgieal treatment for whooping cough. By the Ziggysimilar to the GHPM. The justification for the prioritization theorem, these services would make great differences in thewas that it would eliminate services that did not provide ability to do stuff. Other categories classified as essentialbenefit. The process of creating the prioritized list was ex- included maternity care, treatment for conditions that pre-tremely difficult. The commission began by creating a priori- vents death but does not allow full recovery, and preventivetized list of all health services. However, it soon became care for children. There were nine categories classified asapparent that this was a nearly impossible task. Thus, the essential. Listed as very important were treatments for nonfa-commissioners began searching for combinations of conditions tai conditions that would return the individual to a previousand treatments that could be lumped together. For example, state of health. Also included in this category were acute,the problem of rectal prolapse was paired with the treatment nonfatal one-time treatments that might improve quality ofpartial eolectomy, whereas osteoporosis was paired with life. These would include hip replacements and cornea trans-

ROBERT M. KAPLAN 459

plants. At the bottom of the list were treatments for fatal or strategies for clinically localized prostate cancer. Journal of thenonfatal conditions that did not improve quality of life or American MedicalAssociation 269, 2650-2658.

extend life; progressive treatments for the end stages of Herodotus. (1942). The Persian wars. (G. Rawlinson, Trans.). Newdiseases such as cancer and AIDS or care for conditions in York: Modern Library.which the treatments were known not to be effective were Hillman, A. L., Pauly, M. V., & Kerstein, J. J. (1989). How do financial

included in this category. In other words, services that did not incentives affect physicians' clinical decisions and the financialhelp people do stuff (or live longer) were given low priority. In performance of health maintenance organizations? New Englandthe revised approach, the commission decided to ignore cost Journal of Medicine, 32l, 86-92.information and to allow their own subjective judgments to Holleb, A. I., Fink, D. J., & Murphy, G. P. (1991). Textbook of clinicalinfluence the rankings on the list. Unfortunately, the final oncology. Atlanta, GA: American Cancer Society.exercise in Oregon resulted in many deviations from the Kaplan, R. M. (1982). Human preference measurement for health

decisions and the evaluation of long-term care. In R. L. Kane &GHPM. However, the exercise demonstrated an attempt to R. A. Kane (Eds.), Values and long-term care. Lexington, MA:resolve the health care crisis on the basis of health outcome.

Lexington Books.After nearly 5 years of debate, a revised Oregon plan was Kaplan, R. M. (1984). The connection between clinical health promo-implemented on February 1, 1994. tion and health status: A critical overview.American Psychologist,39,

755-765.

Summary Kaplan, R. M. (1989). Health outcome models for policy analysis.Health Psychology,8, 723-735.

The Ziggy theorem argues that the meaning of life is doin' Kaplan, R. M. (1990). Behavior as the central outcome in health care.stuff (and living long enough to do stuff). In many respects, American Psychologist,45, 1211-1220.Ziggy has summarized the objectives of health care. This Kaplan, R. M. (1993a). Application of a general health policy model inpresentation has extended the Ziggy theorem to include the American health care crisis. Journal of the Royal Society ofquantification. Measurement of quality of life is possible and is Medicine, 86, 277-281.well represented in formal analyses such as the QALY. Kaplan, R. M. (1993b). The Hippocratic predicament: Affordability,

Quality-adjusted survival analysis has important applica- access, and accountability in health care. San Diego, CA: Academictions in clinical decision making, public policy, and clinical trial Press.research. However, this is a new era of investigation. Out- Kaplan, R. M., & Anderson, J. P. (1988a). A general health policy

comes research offers a new paradigm that may stimulate model: Update and applications. Health Services Research, 23,different directions in provider payment, resource allocation, 203--234.and the selection of clinical measures. Kaplan, R. M., & Anderson, J. P. (1988b). The Quality of Well-Being

Scale: Rationale for a single quality of life index. In S. R. Walker &R. Rosser (Eds.), Quality of life: Assessment and application (pp.

References 51-77). London: MTP Press.

Anderson, J. P., Bush, J. W., & Berry, C. C. (1986). Classifying Kaplan, R. M., & Anderson, J. P. (1990). An integrated approach tofunction for health outcome and quality-of-life evaluation: Self quality of life assessment: The general health policy model. In B.versus individual models. Medical Care,24, 454--469. Spilker (Ed.), Quality of life in clinical studies (pp. 131-149). New

Anderson, J. P., Kaplan, R. M., Berry, C. C., Bush, J. W., & Rumbaut, York: Raven.R. G. (1989). Interday reliability of function assessment for a health Kaplan, R. M., Anderson, J. P., Wu, A. W., Mathews, W. C., Kozin, F.,status measure: The Quality of Well-being Scale. Medical Care, 27, & Orenstein, D. (1989). The Quality of Well-Being Scale: Applica-1076-1084. tions in AIDS, cystic fibrosis, and arthritis. Medical Care, 27 (Suppl.

Anderson, J. P., Kaplan, R. M., & DeBon, M. (1989). Comparison of 3), $27-$43.responses to similar questions in health surveys. In F. Fowler (ed.), Kaplan, R. M., & Bush, J. W. (1982). Health-related quality of lifeHealth survey research methods (pp. 13-21). Washington, DC: Na- measurement for evaluation research and policy analysis. Healthtional Center for Health Statistics. Psychology,1,61-80.

Brook, R. H., & Lohr, K. (1986). Will we need to ration effective Kaplan, R. M., Bush, J. W., & Berry, C. C. (1976). Health status: Typeshealth care7 Issues in Science and Technology, 3, 68-77. of validity and the index of well-being. Health Services Research, 11,

Cardiac Arrhythmia Suppression Trial Investigators. (1989). Prelimi- 478-507.

nary report: Effect of Encainide and Flecainide on mortality in a Kaplan, R. M., Bush, J. W., & Berry, C. C. (1978). The reliability,randomized trial of arrhythmia suppression after myocardial infarc- stability, and generalizability of a health status index. In Proceedingstion. New England Journal of Medicine, 321, 406-412. ofthe Social Status Section (pp. 704-709). Alexandria, VA: American

Crawshaw, R., Garland, M. J., Hines, B., & Lobitz, C. (1985). Oregon Statistical Association.health decisions: An experiment with informed community consent.Journal of the American MedicalAssociation, 254, 3213-3216. Kaplan, R. M., Bush, J. W., & Berry, C. C. (1979). Health Status Index:

Diamond, G. A., Denton, T. A., & Matloff, J. M. (1993). Fee-for- Category ratingversus magnitude estimation for measuring levels ofbenefit: A strategy to improve the quality of health care and control well-being. Medical Care,17, 501-525.costs through reimbursement incentives. Journal of the American Kaplan, R. M., DeBon, M., & Anderson, B. F. (1991). Effects ofCollege of Cardiology, 22, 343-352. number of rating scale points upon utilities in a Quality of

Enthoven, A. C., & Kronick, R, (1990). Universal health insurance Well-Being Scale. Medical Care, 29, 1061-1064.through incentives reform. Yournal of the American MedicalAssocia- Lu-Yao, G. L., McLerran, D., Wasson, J. & Wennberg, J. E. (1993).tion, 265, 2532-2536. An assessment of radical prostatectomy: Time trends, geographic

Fleming, C., Wasson, J. H., Albertson, P. C., Barry, M. J., & variation, and outcomes. Journal of the American Medical Associa-Wennberg, J. E. (1993). A decision analysis of alternative treatment tion, 269, 2633-2636.

460 ZIGGYTHEOREM

Rosser, R. (1993). The history of health-related quality of life in 10½ Weinstein, M. C., & Stason, W. B. (1977). Foundations of cost-paragraphs. Journal of the Royal Society of Medicine, 86, 315-318. effectiveness analysis for health and medical practice. New England

Shaw, B. (1911). The doctor's dilemma, getting married, and the Journalo[Medicine, 296, 716-721.

shewing-upofblanco. New York: Brentano's. Williams, A. (1988). The importance of quality of life in policyTorrance, G. W. (1987). Utility approach to measuring health-related decisions. In S. Walker & R. Rosser (Eds.), Quality of life: Assess-

quality of life. Journal ofCt, ronic Diseases, 40, 593-600. ment and application (pp. 279-290). London: MTP Press.U.S. Department of Health and Human Services. (1991). Healthy

people 2000: National health promotion and disease prevention Wilson, T. Ziggy [cartoon]. Universal Press Syndicate.objectives (DHHS Publication No. 91-50213). Washington, DC: World Bank. (1993). Worlddevelopment report1993: Investingin health.Author. NewYork:OxfordUniversityPress.

AppendixExample QWB Calculation

Formula 1 State Description Weight

To obtain the point-in-time well-being score for an individual, W, PAC-1 In bed, chair, or couch for most or all of day, -.077health relateduse SAC-2 Performed no major role activity, health -.061

related, but did perform self.careW = 1 + (CPXwt) + (MOBwt) + (PACwt) + (SACwt), the Wscore iscalculated as

where wt is the preference-weighted measure for each factor, CPXis W = 1 + (-.257) + (-.000) + (-.077) + (-.061) = .605.the symptom-problem complex, MOB is mobility, PAC is physicalactivity, and SAC is social activity. For example, for a person with thefollowingdescriptionprofile, Formula2

State Description Weight To obtain well-years (WI0 or OALY as an output measure, use

CPX-11 Cough, wheezing or shortness of breath, with or -.257 WY = [No. of persons xwithout fever, chills, or aching all over

MOB-5 No limitations -.000 (CPXwt + MOBwt + PACwt + SACwt) x time].