Long-Term Cognitive Impairment

8/6/2019 Long-Term Cognitive Impairment

1/8

CARING FOR THECRITICALLY ILL PATIENT

Long-term Cognitive Impairmentand Functional Disability Among Survivorsof Severe SepsisTheodore J. Iwashyna, MD, PhDE. Wesley Ely, MD, MPHDylan M. Smith, PhDKenneth M. Langa, MD, PhD

C

OGNITIVE IMPAIRMENT ANDphysical disability are majorhealth burdens anddrivers of health care costs. The onset

ofdisability isassociatedwithworsenedmortality 1 and substantial increases inmedical costs over subsequent years, 2including a disproportionate strain onMedicaid and Medicare. Both cogni-tive and physical disability impose yetfurther burdens on families and infor-mal caregivers. 3 Irreversible cognitiveand physical impairment followingacute illnesses are particularly fearedoutcomes andweighheavilyonpatientdecision making. 4

Hundreds of thousands of patientsendure severe sepsis each year in theUnited States. 5 It has been suspectedthat many aredischarged with a newbut poorly definedconstellation of cognitive and functional impair-ments, 6 which may explain their re-duced quality of life. 7 Even hospital-izations for less severe illness oftenresult in a period of functional dis-ability 8 and may hasten the progres-sion of dementia. 9,10 Long-term cogni-tive and functional declines have beenshown among survivors of other criti-cal illnesses, but these declines maybe

partially preventable.11-14

Although se-vere sepsis is the most common non-cardiac cause of critical illness, 5,15 thelong-term impact of severe sepsis oncognitive and physical functioning isunknown.

We studied whether an incident epi-sode of severe sepsis increasedtheoddsofsubsequentworsenedcognitiveimpair-mentandfunctionaldisabilityamongsur-

vivors.Wetookadvantageof anationallyrepresentative ongoing cohort study of olderAmericansthatincludeddetailedin-formation from personal surveys and

See also p 1833 and Patient Page.

Author Affiliations: Department of Internal Medicine,Universityof Michigan MedicalSchool, AnnArbor(DrsIwashyna andLanga); Department of Medicine, Vander-bilt University, and the VA Tennessee Valley GeriatricResearch and Education Clinical Center, Nashville (Dr Ely); Department of PreventiveMedicine, Stony BrookUniversity MedicalCenter, Stony Brook, NewYork (Dr Smith); Institutefor Social Research,University of Michi-gan, and Ann Arbor Veterans Affairs Health Services

Research and Development Service Center of Excel-lence, Ann Arbor (Dr Langa).Corresponding Author: Theodore J. Iwashyna, MD,PhD, Pulmonary andCritical CareMedicine,3A23 300NIB, SPC5419,300 N Ingalls, AnnArbor,MI 48109-5419 ([email protected]).Caring for theCriticallyIll Patient Section Editor: DerekC. Angus, MD, MPH, Contributing Editor, JAMA([email protected]).

Context Cognitive impairment and functional disability are major determinants ofcaregiving needs and societal health care costs. Although the incidence of severe sep-sis is high and increasing, the magnitude of patients long-term cognitive and func-tional limitations after sepsis is unknown.Objective To determine the change in cognitive impairment and physical functioningamong patients who survive severe sepsis, controlling for their presepsis functioning.Design, Setting, and Patients A prospective cohort involving 1194 patients with1520 hospitalizations for severe sepsis drawn from the Health and Retirement Study,a nationally representative survey of US residents (1998-2006). A total of 9223 re-spondents had a baseline cognitive and functional assessment and had linked Medi-care claims; 516 survived severe sepsis and 4517 survived a nonsepsis hospitalizationto at least 1 follow-up survey and are included in the analysis.Main Outcome Measures Personal interviews were conducted with respondentsor proxies using validated surveys to assess the presence of cognitive impairment andtodeterminethe numberof activities ofdaily living (ADLs)andinstrumental ADLs (IADLs)for which patients needed assistance.Results Survivors mean ageat hospitalizationwas76.9 years.Theprevalence of mod-erate to severe cognitive impairment increased 10.6 percentage points among patientswho survived severe sepsis, an odds ratio (OR) of 3.34 (95% confidence interval [CI],1.53-7.25) in multivariable regression. Likewise, a high rateof newfunctional limitationswas seen following sepsis: in those with no limits before sepsis, a mean 1.57 new limi-tations (95% CI, 0.99-2.15); and for those with mild to moderate limitations before sep-sis, a mean of 1.50 new limitations (95% CI, 0.87-2.12). In contrast, nonsepsis generalhospitalizations were associated with no change in moderate to severe cognitive impair-ment (OR, 1.15; 95% CI, 0.80-1.67; P for difference vs sepsis=.01) and with the devel-opment of fewer new limitations (mean among those with no limits before hospitaliza-tion, 0.48; 95% CI, 0.39-0.57; P for difference vs sepsis .001 and mean among thosewith mild to moderate limits, 0.43; 95% CI, 0.23-0.63; P for difference= .001). The de-clines in cognitive and physical function persisted for at least 8 years.Conclusions Severe sepsis in thisolder population was independently associated withsubstantial and persistent new cognitive impairment and functional disability amongsurvivors. The magnitude of these new deficits was large, likely resulting in a pivotaldownturn in patients ability to live independently. JAMA. 2010;304(16):1787-1794 www.jama.com

2010 American Medical Association. All rights reserved. (Reprinted) JAMA, October 27, 2010Vol 304, No. 16 1787

by guest on July 13, 2011 jama.ama-assn.orgDownloaded from
http://jama.ama-assn.org/http://jama.ama-assn.org/http://jama.ama-assn.org/http://jama.ama-assn.org/


2/8

Medicareclaims.Thisprovidedanoppor-tunity to examine the long-term impactofseveresepsisbeforeandupto8yearsafterincident disease.

METHODSData SourceTheHealthandRetirementStudy(HRS)is an ongoing cohort nationally repre-sentative of community-dwelling USresidents older than 50years.Begun in

1992, more than 27 000 individualshave contributed 200 000 hoursofdata-collection interviews.Every 2 years, thecohort is reinterviewed. The HRSachieved a very high follow-up rate,routinely exceeding 90% to 95% in-cluding proxies. 16 Furthermore, 16772participants have consented for link-age of their study data with Medicare.

This work was approved by the Uni-versityofMichiganInstitutionalReviewBoard. Patients providedinformedcon-sentonenrollmentintheHRSandagain

for linkage to Medicare claims. We studied all respondents with atleast 1 interview during 1998-2004 inwhich cognitive andphysical function-ing were assessed and for whom therewere subsequent claims-based dataon a hospitalization for severe sepsisduring 1998-2005 ( F IGURE 1 ). Allpatients were followed up through

death or the 2006 survey. Our pri-mary analyses focus on hospitaliza-tions thatpatientssurvived long enoughto participate in at least 1 follow-upinterview.

Characteristics of the hospitaliza-tions for severe sepsis were abstractedfrom theMedicare claims, includinganorgandysfunction score (the sumof thenumber of organ failures of cardiovas-cular, neurologic, hematologic, he-

patic, renal, or respiratory origin).5,17

Self-reportedraceandethnicitywerein-cluded only in the descriptive statis-tics because they may be of interest tosome readers.

Definition of Severe Sepsis We relied on a claims-based defini-tion of severe sepsis, which has beenwidely used and clinically validated. 5This definition requires evidence of both an infection and new-onset or-gan dysfunction duringa single hospi-

talization. If a patient had more than 1distinct septichospitalization,eachhos-pitalization was included.

As a comparison, we conducted par-allel analyses in a cohort of 5574hospi-talizations.Thesewerefirsthospitaliza-tionsfor membersof thelinked Medicarecohort, which included neither severesepsisnor criticalcareuseand for which

a baseline survey and at least 1 fol-low-up interview were available.

Definition of Functional StatusAt each wave of the survey, we askedrespondents if they required assis-tance with any of 6 activities of dailyliving (ADLs: walking, dressing, bath-ing, eating, getting into andoutof bed,and toileting) or 5 instrumental ADLs(IADLs: preparing a hot meal, shop-ping for groceries, making telephonecalls, taking medicines, and managingmoney). We totaled the number of ADLs and IADLs to create a total defi-ciency score (range, 0 requiring no as-sistance to 11 requiring assistance forall categories). 18 Thesurveyasked prox-ies to evaluate the functional status of patients who could not answer forthemselves; proxiescouldanswer thesequestions with high reliability. 18 Forsome analyses, a baseline of function-ing was defined, using the last surveyprior to severe sepsis. It was decided apriori that patients would be dividedinto 3 groups based on their baselinefunctioning:no limits, 0; mild to mod-erate, 1 to 3; and severe limitations, 4or more deficiencies.

Definition of Cognitive ImpairmentThe survey assessed cognitive func-

tion in 2 ways during biennial per-sonal interviews. For those aged 65years or older, a 35-point scale was ad-ministered that included tests of memory, serial 7 subtractions, nam-ing, and orientation. 19,20 For self-respondents younger than65 years, thesurvey tool administered a more lim-ited 27-point scale that excluded theorientation measures.

For patients 65 years or older whowere unable to be interviewed them-selves, the validated Informant Ques-

tionnaire on Cognitive Decline in theElderly 21 was administered to proxies.For proxies representing respondentsyounger than 65 years, the followingquestionswere used to determinecog-nitive function: How would you rate[the respondents] memory at thepresent time? and How would yourate [the respondent] in making judg-

Figure 1. Patient Cohorts

623 Hospitalizations included in the analysis(516 surviving respondents) a

4517 Hospitalizations included in theanalysis (4517 surviving respondents)

9223 Underwent baseline cognitive and functionalassessments (1998-2004)

16 772 Respondents to the Health and Retirement Studyagreed to link their data with Medicare (1992-2006)

35 Hospitalizations with loss to follow-up(32 respondents)

172 Hospitalizations with loss to follow-up(172 respondents)

862 Hospitalizations with death beforefollow-up (720 respondents) a

885 Hospitalizations with death beforefollow-up (885 respondents)

1520 Hospitalizations for severe sepsis(1194 respondents)

5574 Hospitalizations without sepsis(5574 respondents)

a A single respondent with severe sepsis might contribute a hospitalization to the survivor cohort in one hos-pitalization but be lost to follow-up after a future hospitalization. The categorizations of hospitalizations asincluded vs excluded are mutually exclusive, but the categorizationsof respondents are not. Comparisons wereall first hospitalizations.

COGNITIVE IMPAIRMENT, DISABILITY, AND SEVERE SEPSIS

1788 JAMA, October 27, 2010Vol 304, No. 16 (Reprinted) 2010 American Medical Association. All rights reserved.



3/8

ments anddecisions?Theresponse op-tions for both of these questions wereexcellent, verygood,good, fair, orpoor.

We defined cut points on the cogni-tive assessments for mild and moder-ate to severe cognitive impairment

based on prior studies with the HRSdata, 3,22,23 as well as the methods usedfor the Aging, Demographics, andMemory Study (ADAMS), a supple-mental study of dementia in theHRS. 24These cutpoints defined a levelof cog-nitive impairment that was generallyconsistent with mild and moderate tosevere dementia in the ADAMS. Fur-ther detail on the cognitive measuresis available. 25

AnalysesFor analyses of functional status, ourprimary outcome was measured by acombined ADL and IADLscore. For un-adjusted analyses, we grouped pa-tients by the number of surveys theyhad completed since severe sepsis oc-currence; forexample, we compared allpatients at their last survey before hos-pitalization with severe sepsis with pa-tients at their first survey after severesepsis. For multivariable models, weused longitudinal models to examinethe association between the timing of severe sepsis and the timing of func-

tionalchanges. Thesemodels used onlywithin-person variation over time infunctional status to estimate the im-pact of severe sepsis, and thereby con-trol for all characteristics of the pa-tient that did not change over timein essence, patientsserved as their owncontrols. 26 Specifically,we constructedlatentgrowth curvemodels usinga hos-pitalization-level fixed effect, some-timescalledconditionalmodels. 26 Theseresults controlled fornotonly the func-tional status of the patient before his/

her sepsis episode but also for func-tional trajectory.Allof these sequentialevaluations were included in theanaly-sis. In these models, time from admis-sion with severe sepsis to survey inter-view was measured to the day as acontinuous variable. Additional infor-mation about the statistical approachis presented in the eMethods appen-

dix, including alternative specifica-tions (available at www.jama.com).Fixed effects models were estimatedusing xtreg, fe in Stata 10.1 (StataCorpLP, College Station, Texas). Theseanalyses were not conducted accord-

ing to a fully prespecified protocol.For analyses of cognitive function-ing, our primary outcome was level of cognitive impairment. Unadjustedanalyses were conducted as for func-tional status. For multivariable analy-ses, we used conditional logistic regres-sion to analyze the impact of severesepsis on moderate to severe cognitiveimpairment among survivors, usingclogit inStata10.1.As for functional sta-tus, these analyses used only within-person variation over time to estimatethe effect of severe sepsis, controllingfor time-invariant characteristics of therespondent.

All analyses were conducted withhospitalization as the unit of analysisunless otherwise indicated. Two-sided significance testing was usedthroughout, and a P value of .05 wasconsidered statistically significant.

RESULTSThere were 1520 identified episodes of severe sepsis among 1194 respondentsfor the years 1998-2005, from a cohort

of 9223 respondents (Figure 1). Detailabout theentirepopulationofsevere sep-sishospitalizationsispresentedin eTable1.Ninety-daymortality after severe sep-sis was 41.3%(95% confidence interval[CI], 38.8%-43.8%); 5-year mortality,81.9% (95% CI, 79.8%-84.0%). Five-year survivalcurves arepresented in theeFigure (available at www.jama.com).Five hundred sixteen individuals sur-vived 623 episodes of severe sepsis andhadatleast 1 follow-up survey;thesehos-pitalizations by survivors are our pri-

mary cohort for analysis (T ABLE 1

). Pa-tients were followed up for up to 4surveys (7.8 years) of data prior to se-veresepsis and upto4 surveys(8.3years)afterward.

Cognitive OutcomesIncident severe sepsis was associatedwith a clinically and statistically sig-

nificant increase in moderate to se-vere cognitive impairment among sur-vivors. For example, 6.1% (95% CI,4.2%-8.0%) of eventual survivors hadmoderate to severe cognitive impair-ment at the survey just before severe

sepsis, and the prevalence increasedto 16.7% (95% CI, 13.8%-19.7%) atthe first survey after severe sepsis(F IGURE 2 , P .001 by 2 test). In fixed-effects regression, with each patientserving as his/her own control, the in-cidence ofseveresepsisremainedhighlyassociated with progression to moder-ate to severe cognitive impairment(odds ratio [OR], 3.34; 95% CI, 1.53-7.25; T ABLE 2 ). No association existedbetween severe sepsisandthe net preva-lence of mild cognitive impairment inadjusted or unadjustedanalyses;nearlyequal numbersofpreviouslynormal pa-tientsdevelopedmildcognitive impair-ment after severesepsisaspatients withpresepsismildcognitive impairmentde-veloped postsepsis moderate to severecognitive.

Functional OutcomesSurvivors of hospitalization for severesepsis were at greater risk of addi-tional functional limitations at theirnext survey. This was a substantialworsening in their trajectory relative

to before their sepsis hospitalization.The negative effects of severe sepsiswere greater in those patients withbetter baseline physical functioning(F IGURE 3 ). The new functional defi-cits were not concentrated in any par-ticular subset of the functioning mea-sures ( F IGURE 4 ).

The independent effects of severesepsis on long-term disability per-sisted in multivariable analyses witheach patients presepsis functional trajectory serving as his/her own control.T ABLE 3

shows that severe sepsis wasassociatedwith thedevelopment of1.57(95% CI, 0.99-2.15) new limitationsamong patients who had none beforesepsis. Patients with mild to moderatelimitations before sepsis had a similarincrease of 1.50 (95% CI, 0.87-2.12)new IADL and ADL limitations. Forsuch patients, not only was sepsis





4/8

associated with an acuteincrease in thenumber of functional limitations, butsepsis also heralded a more rapid rateof developing further limitations there-after, at 0.51 new limitations per year(P=.007 for difference vs baseline). In

contrast, patients with already poorfunctioning experienced no statisti-cally significant change in function-ingwith severe sepsis, although the re-gressions may be limited by ceilingeffects in measurement of functioning.

Of the hospitalizations involving se-vere sepsis, 59.3% (95% CI, 55.5%-63.2%) were associated with wors-ened cognitive or physical function,orboth, amongsurvivors at the first post-sepsis survey. The association of se-vere sepsis with increased functionallimitations remained clinically mean-ingful and statistically significant in re-gression when controlling for changesin level of cognitive impairment aftersevere sepsis: 1.30 (95% CI, 0.86-1.74) new limitations for those with nolimitations at baseline; 1.20, (95% CI,0.62-1.79) new limitations for thosewith mild to moderate limitations atbaseline. The increased risk of moder-ate to severe cognitive impairment re-mained clinically meaningful but wasattenuatedin the regressionswhencon-trolling for contemporaneous changes

in levels of physical functioning aftersevere sepsis (OR, 1.73; 95% CI, 0.83-3.6).

Comparisonto Other HospitalizationsThe changes in physical and cognitivefunctioning noted after severe sepsiswere worse than those seen after non-sepsisgeneral hospital admissions in acohort of 4517 survivors of 5574 hos-pitalizations.Thus, patientswhodidnotdevelop severe sepsis and who had nofunctional limitationsprior to their hos-

pitalization developed an average of 0.48 (95% CI,0.39-0.57; n=2852; P fordifference vs sepsis .001, eTable 3)new functional limitations. Patientswith mild to moderate functional limi-tations at baseline developed 0.43(95% CI, 0.23-0.63; n= 1124; P fordif-ference=.001; eTable 3) new func-tional limitations after a nonsep-

Table 1. Demographics of Study Cohort of Survivors, by Baseline Physical Functioning (n = 623) a

Functional Class at Baseline by Limitations

None Mild to Moderate SevereNo. 269 195 159Men, No. (%) 143 (53) 92 (47) 46 (29)Race/ethnicity, No. (%)

Black 49 (18) 41 (21) 38 (24)Hispanic 19 (7) 12 (6) 13 (8)

Age at sepsis, mean (SD), y 75.8 (7.5) 76.7 (9.5) 79.1 (9.6)Length of stay, mean (SD), d 11.4 (10.7) 11.3 (11.2) 8.5 (6.3)Required mechanical ventilation, No. (%) 64 (23) 32 (16) 27 (17)Required dialysis, No. (%) 9 (3.4) 6 (3.1) 12 (7.6)Used an intensive care unit, No. (%) 137 (51) 75 (38) 57 (36)Underwent major surgery, No. (%) 73 (27) 39 (20) 15 (9)Charlson score, mean (SD) 1.69 (1.42) 1.96 (1.64) 2.11 (1.41)Organ dysfunction score, mean (SD) 1.15 (0.39) 1.16 (0.45) 1.11 (0.34) Acute conditions, No. (%)

Cardiovascular dysfunction 60 (22) 62 (32) 45 (28)Neurologic dysfunction 19 (7) 20 (10) 17 (11)Hematologic dysfunction 61 (23) 34 (17) 27 (17)Hepatic dysfunction 2 (1) 0 (0) 1 (1)Renal dysfunction 103 (38) 79 (41) 60 (38)Respiratory dysfunction 64 (24) 32 (16) 27 (17)

Baseline, No. (%)Cognitve impairment

None 254 (94) 182 (93) 105 (66)Mild 15 (5.6) 9 (4.6) 20 (12.6)Moderate to severe 0 4 (2.1) 34 (21.4)

Physical function deficiencies, mean (SD)Basic ADL 0 1.3 (0.9) 4.0 (1.7)Instrumental ADL 0 0.5 (0.7) 3.0 (1.5)

Proxy respondent, No. (%) At baseline 9 (3) 22 (11) 59 (37) At first postsepsis survey 46 (17) 47 (24) 87 (55)

Abbreviation: ADL, activity of daily living.a Data for the entire cohort of incident severe sepsis hospitalizations are in eTable 1, and risk factors for cognitive im-

pairments are presented in eTable 2 (both available at www.jama.com).

Figure 2. Cognitive Impairment Among Survivors of Severe Sepsis at Each Survey Time Point

15

20

25

10

5

0

Time to sepsis admission,median (IQR), y

No. of patients

Second SurveyBefore Sepsis

3.1(3.7 to 2.7)

484

Last SurveyBefore Sepsis

1.1(1.7 to 0.7)

623

First Survey After Sepsis

0.9(0.4 to 1.4)

623

Second Survey After Sepsis

2.8(2.3 to 3.4)

288

Before sepsis After sepsis

P a t

i e n t s

W i t h C o g n i

t i v e

I m p a

i r m e n

t , %

Cognitive impairmentMildModerate to severe

Error bars indicate 95% confidence intervals (CIs); IQR, interquartile range.Interpretive Example: Compared with stable rates before severe sepsis, the prevalence of moderate to severecognitive impairment increased from 6.1% (95% CI, 4.2%-8.0%) before severe sepsis to 16.7% (95% CI,13.8%-19.7%) at the first survey after severe sepsis ( P .001 by 2 test; Table 2).





5/8

sis general hospitalization. Further-more, nonsepsis general hospitaliza-tions were not associated with aclinically or statistically significant in-crease in the odds of moderate to se-vere cognitive impairment (OR, 1.15;

95% CI, 0.80-1.67; n=4517; P for dif-ference= .01, eTable4 available at www.jama.com)

Subgroup and Sensitivity Analyses We replicated our analyses in severalsubgroups to examine their robust-ness. The effects of severe sepsis weresimilar in the 500 survivors who hadsevere sepsis but who did not requiremechanical ventilation. The regres-sion demonstrated similarly increasedodds (OR, 4.0; 95% CI, 1.71-9.31) of developing moderate to severe cogni-tive impairment after severe sepsisamong patients whowere notmechani-cally ventilated. Similarly, in the 205survivors with no limitations at base-line, severe sepsis without mechanicalventilation was associated with thede-velopment of 1.56 newfunctional limi-tations (95% CI, 0.91-2.22) in multi-variable fixed-effects models. For the163 patients with mild to moderatelimitations, severe sepsis without me-chanicalventilationwasassociatedwith1.65 new functional limitations (95%

CI, 1.01-2.28).A potential threat to the validity of the results is that patients may haveexperienced some other cause of cog-nitive and functional decline betweentheir baseline survey and their sepsishospitalization. Therefore, we reana-lyzed data for the smaller subset of 276 survivors who were never hospi-talized between their baseline surveyand their severe sepsis admission. Wefound consistent results, albeit withlarger standard errors. In this sub-

population, severe sepsis was associ-ated with increased odds of moderateto severe cognitive impairment (OR,2.49; 95% CI, 0.99-6.26). In the 128patients with no functional limitationsat baseline and no intercurrent hospi-talizations, severe sepsis was associ-ated with the development of 1.46new functional limitations (95% CI,

0.76-2.15). In the 86 patients withmild to moderate functional limita-tions at baseline and no intercurrenthospitalizations, severe sepsis wasassociated with the development of 1.34 new functional limitations (95%

CI, 0.34-2.34).Further sensitivity analyses yieldedconsistent results. The associationsbetween severe sepsis and functionaland cognitive impairment were sub-stantively similar in those aged 65years or older at baseline cognitiveassessment and who therefore wereassessed using a single instrumentbefore and after severe sepsis (eTables5 and 6). The patterns observed forfunctional limitations were similar in alarger cohort of 2043 hospitalizations(including 829 hospitalizations among684 survivors) for severe sepsis fol-lowed up for up to 14 years during theperiod 1992-2006 (eTable 7). Examin-ing only the subset of 516 first sepsisadmissions for each survivorso thatno patient appeared in the analysismore than onceyielded nearly iden-tical results (eTables 8 and 9).

COMMENTIn this nationally representative co-hort, we have demonstratedfor thefirsttime that severe sepsis is indepen-dently associated with enduring cog-nitive and functional limitations. Se-

veresepsis is independently associatedwith a tripling in the odds of moderateto severe cognitive impairment. Fur-thermore, severe sepsis was indepen-

Table 2. Severe Sepsis and Moderate toSevere Cognitive Impairment AmongSurvivorsa

Odds Ratio(95% Confidence

Interval)P

ValueBefore sepsis

(per additionalyear)

1.35 (1.11-1.65) .002

Effect of sepsis 3.34 (1.53-7.25) .002

After sepsis(per additionalyear)

1.68 (1.28-2.21) .001

a Results of latent growth curve regression with individual-level fixed effects, controlling for all time-invariant char-acteristicsof thepatient. Theabsence ofassociationwouldbe indicated by an odds ratio of 1.

InterpretiveExample: With each passingyear,patientsweremodestly more likelyto develop moderate to severecog-nitive impairment. Afterseveresepsis,survivorshad a 3.3-fold greateroddsof havingmoderateto severe cognitiveimpairment than before sepsis (Figure 2).

Figure 3. Functional Trajectories by Baseline Functioning

SevereLimitations at baseline

NoneMild to moderate

9

8

4

5

7

6

3

2

1

0

Time to sepsis admission,median (IQR), y

No. of patients, by baselinephysical functioning

Severe limitsMild to moderate limitsNo limits

Third SurveyBefore Sepsis

142105

5.2(5.6 to 4.7)

87

Third Survey After Sepsis

5628

5.2(4.5 to 5.5)

14

Second SurveyBefore Sepsis

206151

3.1(3.7 to 2.7)

127

Second Survey After Sepsis

15393

2.8(2.3 to 3.4)

42

Last SurveyBefore Sepsis

269195

1.1(1.7 to 0.7)

159

First Survey After Sepsis

269195

0.9(0.4 to 1.4)

159

M e a n

N u m

b e r o f

A D L

a n d I A D L L i m

i t a t i o n s

Before sepsis After sepsis

The unadjusted mean number of functional limitations of surviving cohort members is shown. Error bars in-dicate 95% confidence intervals.Interpretive Example: Groups that had no functional or mild to moderate limitations before sepsis had a stabletrajectory beforesepsisbut developedabout2 newlimitations after sepsis. Butpatientswith severelimitationsat baseline had a modest increase from a baseline of 6.99 to 7.98 at their first survey after sepsis (Table 3).





6/8

dently associated with the acquisitionof 1.5 new functional limitations in pa-tients with no, mild, or moderate pre-

existing functional limitations. Thesenewdisabilities weresubstantially largerthan those seen after nonsepsis gen-eralhospital admissions. Cognitiveandfunctional declines of the magnitudeseen after severe sepsis are associated

with significant increases in caregivertime, nursinghome admission, depres-sion, andmortality. 3,27-30 These data ar-gue that the burdenof sepsis survivor-ship is a substantial, underrecognizedpublic health problem with major im-plications forpatients, families,andthehealth care system.

Our findings,and thenationallyrep-resentativedata from theHRS,allow usto make an estimateof theoverall pub-lic health burden of sepsis on brainhealth among older adults in theUnitedStates. Given publisheddemen-tia31 andsepsis 5 incidence rates for thoseaged 65 years or older in the UnitedStates, our results suggest that nearly20000 new cases per year of moderateto severe cognitive impairment in theelderly may be attributable to sepsis.Thus, an episode of severe sepsis, evenwhen survived, may represent a senti-nel event in the lives of patients andtheir families, resulting in new and of-ten persistent disability, in some caseseven resembling dementia. 3,22,32,33

The levelof severecognitive impair-

ment found in these patients has beenassociated with an additional 40 hoursper week of informal care provided byfamilies, 3 analogous to an additionalfull-time job. If causally related, thisrepresents a substantial public healthburden of accelerated or de novo braindysfunction, and one that has re-ceived almost no attention, even in theface of the dramatically increasing in-cidence of severe sepsis. 15 In markedcontrast to Alzheimerdisease andsomeother forms of dementia, onset andac-

celerationof cognitive impairment dueto sepsis is likely partially preventablein many patients. These benefitsmightbe achieved by raising the standard of care forpatients whodevelop sepsisbothsepsis-specificcare aswell asotherintensivecare unit practices such as se-dation management and early physi-calandcognitiverehabilitationandby

avoiding sepsis altogether. 34 Improv-ing theprevention and managementof sepsis may warrant a place in thebroader brain health and disabilityagendas.

Although an observational study can

never prove causation, there are mul-tiple plausible causal pathways bywhich sepsisandits treatmentmayleadto significant declines in physical andcognitive function.The literatureon in-tensive care unitacquired weaknessandchronic illness myopathy andpoly-neuropathy suggests that there is a di-rect inflammatoryand hypoperfusion-mediated degradation of muscle fibersand neurons, 35-37 which may be exac-erbated by prolonged immobility 38 andlack of physical therapy. 39 Similarly,frank hypotension or relativehypoper-fusion maydirectlycontribute to braininjury and subsequent cognitive im-pairment. 40-42

Inflammationa cardinal compo-nent of the pathophysiology of sep-sisis hypothesized to contribute toboth vascular dementia and Alzheimerdisease. 6,10,43 Delirium,an acute form of brain dysfunction characterized by in-attention, is commonin sepsis,prevent-able, andtreatable. 44,45 Deliriumhasbeenassociated with increased cognitive de-cline among patients with Alzheimer

disease9,32

aswell aswith increasedratesof long-term cognitive impairment inmechanically ventilated patients. 33 Ba-sic biological research to understandthese mechanisms is clearlywarranted.Equally pressing is the need for inno-vative clinical trials of both sepsis-specific therapy and improved life sup-port.Ourresults suggest that such trialsshould lookbeyond short-termmortal-ity to long-term cognitive and func-tionaloutcomes ofcrucial interest topa-tients. 46

We conducted analyses that addressseveral possible limitations.The regres-sions usedonlywithin-personvariationto estimate the association with severesepsis; thus, characteristicsof thesurvi-vors that did not change over time can-notexplain thetimingofchangesin cog-nitiveandfunctionalstatus.Thedifferentcognitive and physical function out-

Figure 4. Change in Individual ADLs andInstrumental ADLs by Baseline Functioning

0 0.2 0.4 0.80.6 1.0Fraction of Patients With Difficulty

Walk

Dress

Bathe

Eat

Get into bed

Toilet

Prepare meal

Grocery shop

Use telephone

Take medications

Manage money

Severe limitations (n = 159)

Walk

Dress

Bathe

Eat

Get into bed

Toilet

Prepare meal

Grocery shop

Use telephone

Take medications

Manage money

Mild to moderate limitations (n = 195)


Before sepsis After sepsisWalk

Dress

Bathe

Eat

Get into bed

Toilet

Prepare meal

Grocery shop

Use telephone

Take medications

Manage money

No limitations (n = 269)


InterpretiveExample: Nosingleactivityofdailyliving(ADL)or instrumental ADLaccounted forthe worsened func-tional status among survivors of severe sepsis. Instead,there wasa widerange ofnew difficulties acrossthe ar-ray of activities.





7/8

comes between the survivors of severesepsisandofthecomparisongeneralhos-pitalizations suggest that the sepsisresults were notsimplydue to theagingof the cohort or the mere fact of hospi-talization, processes shared equally by

both groups. These different outcomesalso suggest that our results cannot beattributed solely to asymmetric censor-ing, one form of a potential bias knownas truncation by death. 47 However,because patients with worse cognitiveand physical functioning have greatermortality (eFigure, available at www.jama.com), 22,30 there may besomecon-servative bias in our results. This formof truncation by death results if pa-tientswith theworst cognitiveandphysi-cal declines after sepsis do not survivelong enough for a follow-up survey. Totheextent that such truncation bydeathis present, the full effectof severe sepsison cognitive and physical functioningwould be even greater than we mea-sured.

Ourstudyhasseverallimitations. Un-like prior studies that have focused onacute functionaldecline in the perihos-pitalization period, 1,8,48-50 the pres-ent resultsdemonstrate only long-termeffects; short-termdeficits (eg, less than6-12 months) are likelygreater, with atleast some patients recovering some

function prior to their next HRS bien-nialsurvey.Theneuropsychologicalbat-tery that we used provided an assess-ment of global cognitive function, butdid not allow detailed study of indi-vidual cognitive domains nor did it es-tablish a definitive clinical diagnosis of dementia. Importantly, we used cogni-tivecategoriesandcutoff scoresthathaveshowngood correlation with clinicalde-mentia 51 and expected outcomes of de-mentia 3,27 in prior studies. We used aclaims-based definition of severe sep-

sis.Although this isnotthe sameaspro-spective clinical assessment, it is thesame approach used in recent land-mark epidemiological studies. 5,15 Ourdata were restricted to fee-for-serviceMedicare patients aged 65 years orolder. 5 We have shown that these dete-riorations were temporally associatedwith severe sepsis and independent of

other stable patient characteristics, butwehave not conclusively proven that itwas severe sepsis rather than other si-multaneous events that led to these de-clines. Although, to ourknowledge, thisis the largest study todateofsevere sep-sis and our outcomes of interest, ourstudy was not powered to examine in-teractions, such as the extent to whichthe changes after sepsis varied with thenumber of organ failures or type of in-citing organism. Medicare claims lack

the information necessary to disen-tangle whether particular acute inter-ventions are associated with differinglong-term outcomes. Finally, we dem-onstrated the association of severe sep-sis with functioning under the treat-ment regimes in effect in a range of UShospitals at a particular point in time.New treatments for sepsis, or changesin life support or other hospital prac-tices, maymodify the long-term cogni-tive andfunctional effects of severesep-sis, even if these deficits are not an

explicit target of care.In summary, in this large nationallyrepresentativecohortofolder adults,wefound that the odds of acquiring mod-erate to severe cognitive impairmentwere 3.3 times ashighfollowing anepi-sode of sepsis, with an additional meanincrease of 1.5 new functional limita-tions per person among those with no

or mild to moderate preexisting func-tional limitations. Thus, sepsis is of-ten a sentinel event in the lives ofolderpatients, initiating major and endur-ing cognitive and functional declineswith lasting implications for patientsindependence, for their loved ones, andforthesocietal institutionschargedwithsupporting them. Future research toidentify mechanismsleading fromsep-sis to cognitive impairment and func-tional disabilityand interventions to

prevent or slow these accelerated de-clinesis especially important nowgiven the aging of the population.

AuthorContributions: DrIwashynahad full accessto allof the data in thestudy and takes responsibility for theintegrityofthedataandtheaccuracyofthedataanalysis. Study concept and design: Iwashyna, Ely, Langa.Acquisition of data: Langa.Analysis and interpretation of data: Iwashyna, Ely,Smith, Langa.Drafting of the manuscript: Iwashyna, Ely, Langa.Critical revision of the manuscript for important in-tellectual content: Iwashyna, Ely, Smith, Langa. Statistical analysis: Iwashyna, Ely, Smith, Langa.Obtained funding: Iwashyna.Administrative, technical, or material support:Iwashyna, Ely, Langa. Study supervision: Iwashyna, Ely, Langa.Financial Disclosures: None reported.Funding/Support: This work wassupportedby grantsK08 HL091249, R01 AG027010,and R01 AG030155from the National Institutes of Health; the Society ofCritical Care Medicines 2010VisionGrant; andby pi-lot support fromgrant UL1RR024986from theMichi-gan Institute for Clinical and Health Research. GrantU01 AG09740 from the National Institute on Agingprovided fundingfor theHealthand Retirement Study,which is performed at the Institute for Social Re-search, University of Michigan. Dr Ely was supported

Table 3. Acquisition of New Functional Limitations Before and After Sepsis Among Survivorsby Functional Class at Baseline a

Functional Class at Baseline by Limitations

None(n = 269)

Mild to Moderate(n = 195)

Severe(n = 159)

Before sepsis 0.020 0.11 0.84

Per year, CI 0.046 to 0.086 0.01 to 0.21 0.73 to 0.92P value .55 .03 .001

Effect of sepsis 1.57 1.50 0.04Per year, CI 0.99 to 2.15 0.87 to 2.12 0.74 to 0.81P value .001 .001 .93

After sepsis 0.19 0.51 0.16Per year, CI 0.03 to 0.41 0.24 to 0.77 0.19 to 0.50P value .09 .001 .37

Abbreviation: CI, confidence interval.a Results of latent growth curve regression with individual-level fixed effects, controlling for all time-invariant character-

istics of the patient. The within-patient R 2 were 0.25 for the no limitation group, 0.37 for those with mild to moderatebaseline limitations, and 0.45 for those with severe baseline limitations. The absence of association would be indi-cated by the acquisition of 0 new functional limitations.

Interpretive Example: Patients with mild to moderate limitations at baselinewere acquiring0.11new limitations a yearbefore severe sepsis. They acquired 1.50 new limitations at hospitalization for severe sepsis. Each year after sepsis,they acquired 0.51 new limitations a year, a statistically significant increase relative to their presepsis rates (Figure 3).





8/8

by grant AG027472 from the National Institutes ofHealth, by the Veterans Affairs Merit Award, and bythe TennesseeValleyGeriatric Research, EducationandClinical Center. Consultative support was providedbythe Measurement Core of the Michigan DiabetesCenter through grant P60 DK-20572 from the Na-tional Institute of Diabetes and Digestive and KidneyDisease.Role of the Sponsor: The funders played no role in

the design, interpretation, or decision to publish theanalysis presented herein.Disclaimer: Theviewsexpressed in thisarticle arethoseof the authors and do not necessarily reflect the po-sition or policy of theDepartment of Veterans Affairsor the US government.Online-Only Material: The eMethods appendix,eTables 1-9, and the eFigure are available at www.jama.com.Additional Contributions: We thank Mohamed Ka-beto, MS,RyanMcCammon, AB,Lili Deng, MD,MA,and Tish Shapiro, MA, all at the University of Michi-gan, for their expert programming; they were finan-cially compensated for their work. We further thankRobertHyzy, MD,and RodneyHayward,MD, of theUniversityof Michiganfor insightful critiques; theyre-ceived no compensation for their work.

REFERENCES1. Rozzini R, Sabatini T, Cassinadri A, et al. Relation-ship between functional loss before hospital admis-sionand mortalityin elderly persons withmedical illness. J Gerontol A Biol Sci Med Sci . 2005;60(9):1180-1183.2. Fried TR,Bradley EH,WilliamsCS, TinettiME. Func-tional disability and health care expenditures for older persons. Arch Intern Med . 2001;161(21):2602-2607.3. Langa KM, Chernew ME, Kabeto MU, et al. Na-tional estimates of the quantity and cost of informalcaregiving for the elderly with dementia. J Gen In-tern Med . 2001;16(11):770-778.4. Fried TR, Bradley EH, Towle VR, Allore H. Under-standing the treatment preferences of seriouslyill patients. N Engl J Med . 2002;346(14):1061-1066.5. Angus DC, Linde-Zwirble WT, Lidicker J, ClermontG, Carcillo J, PinskyMR. Epidemiologyof severesep-sis in the United States: analysis of incidence, out-come, and associated costs of care. Crit Care Med .2001;29(7):1303-1310.6. Yende S, Angus DC. Long-term outcomes fromsepsis. Curr Infect Dis Rep . 2007;9(5):382-386.7. Winters BD, Eberlein M, Leung J, Needham DM,Pronovost PJ, Sevransky JE. Long-term mortality andquality of life in sepsis: a systematic review. Crit Care Med . 2010;38(5):1276-1283.8. Boyd CM, Landefeld CS, Counsell SR, et al. Re-covery of activities of daily living in older adults after hospitalization for acute medical illness. J Am Geriatr Soc . 2008;56(12):2171-2179.9. Fong TG, Jones RN, Shi P, et al. Delirium acceler-atescognitivedecline in Alzheimer disease. Neurology .2009;72(18):1570-1575.10. HolmesC, Cunningham C,ZotovaE, etal. Systemicinflammationanddiseaseprogressionin Alzheimerdisease.Neurology . 2009;73(10):768-774.11. HerridgeMS, CheungAM, TanseyCM, et al;Ca-nadian Critical Care Trials Group. One-year out-comes in survivors of the acute respiratory distresssyndrome. N Engl J Med . 2003;348(8):683-693.12. Chelluri L, Pinsky MR, Donahoe MP, Grenvik A.Long-termoutcome of critically ill elderlypatientsre-quiring intensive care. JAMA. 1993;269(24):3119-3123.13. Hopkins RO, Weaver LK, CollingridgeD, ParkinsonRB, Chan KJ, Orme JF Jr. Two-year cognitive, emo-

tional, and quality-of-life outcomes in acute respira-tory distress syndrome. Am J Respir Crit Care Med .2005;171(4):340-347.14. Kaarlola A, Tallgren M, PettilV. Long-termsur-vival, quality of life, and quality-adjusted life-yearsamong criticallyill elderly patients. Crit Care Med . 2006;34(8):2120-2126.15. Martin GS, Mannino DM, Eaton S, Moss M. Theepidemiologyof sepsis in the United States from1979

through 2000. N Engl J Med . 2003;348(16):1546-1554.16. Health and Retirement Study. Sample Sizes and Response Rates ; 2008http://hrsonline.isr.umich.edu/sitedocs/sampleresponse.pdf . Accessed 27 October 2009.17. Quan H, SundararajanV, Halfon P, et al. Codingalgorithmsfor defining comorbiditiesin ICD-9-CM andICD-10 administrative data. Med Care . 2005;43(11):1130-1139.18. Fonda S, Herzog AR. Documentation of Physi-cal Functioning Measured in the Heath and Retire-mentStudyandthe Asset andHealthDynamicsAmong the Oldest Old Study. Ann Arbor, Michigan: SurveyResearch Center; December 21, 2004.19. Herzog AR, Wallace RB. Measures of cognitivefunctioning in the AHEAD Study. J Gerontol B Psy-chol Sci Soc Sci. 1997;52(spec No):37-48.20. Welsh KA, Breitner JCS,Magruder-Habib KM.De-tection of dementia in the elderly using telephonescreening of cognitive status. Neuropsychiatry Neu-ropsychol Behav Neurol . 1993;6:103-110.21. Jorm AF. The Informant Questionnaire on cog-nitive decline in the elderly (IQCODE): a review. Int Psychogeriatr . 2004;16(3):275-293.22. Langa KM,Larson EB,KarlawishJH, et al. Trendsin the prevalence and mortality of cognitive impair-ment in theUnited States: is there evidence of a com-pression of cognitive morbidity? Alzheimers Dement .2008;4(2):134-144.23. Alzheimers Association EarlyOnset Dementia:ANationalChallenge,a Future Crisis; 2006. http://www.alz.org/national/documents/report_earlyonset_full.pdf. Accessed on August 26, 2010.24. Langa KM, Plassman BL, Wallace RB, et al. TheAging, Demographics, and MemoryStudy: studyde-sign and methods. Neuroepidemiology . 2005;25(4):181-191.25. Ofstedal MB, Fisher GG, Herzog AR Documen-tationof Cognitive FunctioningMeasures in theHealthandRetirementStudy; 2005 . http://hrsonline.isr.umich.edu/sitedocs/userg/dr-006.pdf.AccessedAugust 25,2010.26. AllisonPD.FixedEffectsRegressionModels. Thou-sand Oaks: Sage; 2009.27. Banaszak-Holl J, Fendrick AM, Foster NL, et al.Predicting nursinghome admission. Alzheimer DisAs- soc Disord . 2004;18(2):83-89.28. Zivin K, Llewellyn D, Lang I, et al. Depressionamong older adults in theUS andEngland [publishedonline ahead of print April 9, 2010]. Am J Geriatr Psychiatry . doi: 10.1097/JGP.0b013e3181dba6d2.29. Yaffe K, Fox P, Newcomer R, et al. Patient andcaregiver characteristics and nursing home place-ment in patients with dementia. JAMA. 2002;287(16):2090-2097.30. MehtaKM,Yaffe K, Langa KM,SandsL, WhooleyMA, Covinsky KE. Additive effects of cognitive func-tion and depressive symptoms on mortality in elderlycommunity-living adults. J Gerontol A Biol Sci Med Sci. 2003;58(5):M461-M467.31. Ziegler-Graham K, Brookmeyer R, Johnson E,Arrighi HM. Worldwidevariationin the doublingtimeof Alzheimers disease incidence rates. AlzheimersDement . 2008;4(5):316-323.32. Jackson JC, Gordon SM, Hart RP, Hopkins RO,Ely EW. The association between delirium and cog-nitive decline:a review of theempirical literature. Neu-ropsychol Rev . 2004;14(2):87-98.

33. Girard TD, Jackson JC, Pandharipande PP, et al.Delirium as a predictorof long-termcognitive impair-ment in survivorsof critical illness. Crit Care Med . 2010;38(7):1513-1520.34. Saint S, Savel RH, Matthay MA. Enhancingthe safety of critically ill patients by reducing urinaryand central venous catheter-related infections.Am J Respir Crit Care Med . 2002;165(11):1475-1479.

35. Sharshar T, Bastuji-Garin S, Stevens RD, et al;Groupe de Re flexion et dEtude des Neuromyopa-thies En Reanimation. Presence and severity of inten-sive care unit-acquired paresis at time of awakeningare associated with increased intensive care unit andhospital mortality. Crit Care Med . 2009;37(12):3047-3053.36. Callahan LA, Supinski GS. Sepsis-inducedmyopathy. Crit Care Med . 2009;37(10)(suppl):S354-S367.37. Schweickert WD, Hall J. ICU-acquired weakness.Chest . 2007;131(5):1541-1549.38. Needham DM. Mobilizing patients in the inten-sivecare unit: improving neuromuscular weakness andphysical function. JAMA. 2008;300(14):1685-1690.39. Schweickert WD,Pohlman MC,Pohlman AS,et al.Early physical and occupational therapy in mechani-cally ventilated, critically ill patients. Lancet . 2009;373(9678):1874-1882.40. KuoH-K,Sorond F, Iloputaife I, GagnonM, MilbergW, Lipsitz LA. Effect of blood pressure on cognitivefunctions in elderly persons. J GerontolA Biol SciMed Sci. 2004;59(11):1191-1194.41. Langa KM, Foster NL, Larson EB. Mixed demen-tia: emerging concepts and therapeutic implications. JAMA. 2004;292(23):2901-2908.42. Mikkelsen ME, Lanken PE, Biester R, et al. Con-servative fluide strategy is associated with neurocog-nitive deficits in survivors of acute lung injury. Paper presented at: American Thoracic Society; May 21,2008; Toronto, ON.43. Kuo HK, Yen CJ, Chang CH, Kuo CK, Chen JH,Sorond F. Relation of C-reactive protein to stroke,cognitive disorders, and depression in the gen-eral population. Lancet Neurol . 2005;4(6):371-380.44. Sprung CL, Peduzzi PN, Shatney CH, et al; TheVeterans Administration Systemic Sepsis Coopera-tive Study Group.Impact of encephalopathy on mor-tality in the sepsis syndrome. Crit Care Med . 1990;18(8):801-806.45. Ely EW, Shintani A, Truman B, et al. Delirium asa predictor of mortality in mechanically ventilatedpa-tients in the intensive care unit. JAMA. 2004;291(14):1753-1762.46. IwashynaTJ. Survivorshipwill bethe defining chal-lenge of critical care in the 21st century. Ann Intern Med . 2010;153(3):204-205.47. McConnell S, Stuart EA, Devaney B. Thetruncation-by-death problem. Eval Rev. 2008;32(2):157-186.48. Boyd CM,RicksM, Fried LP,et al.Functional de-cline and recovery of activities of daily living in hos-pitalized, disabled older women. J Am Geriatr Soc .2009;57(10):1757-1766.49. Boyd CM, Xue QL, Guralnik JM, Fried LP. Hos-pitalization and development of dependence in ac-tivities of daily living in a cohort of disabled older women. J Gerontol A Biol Sci Med Sci . 2005;60(7):888-893.50. CovinskyKE, PalmerRM, FortinskyRH, etal. Lossof independence in activities of daily living in older adultshospitalized withmedical illnesses: increasedvul-nerability with age. J Am Geriatr Soc . 2003;51(4):451-458.51. Plassman BL, Langa KM, Fisher GG, et al. Preva-l e n c e o f d e m e n t i a i n t h e U n i t e d S t a t e s .Neuroepidemiology . 2007;29(1-2):125-132.




Long-Term Cognitive Impairment

Documents

Transcript of Long-Term Cognitive Impairment