Depressive Symptoms and Hospital Readmission in OlderAdults

Jennifer S. Albrecht, PhD,a,b Ann L. Gruber-Baldini, PhD,a Jon M. Hirshon, MD, MPH, PhD,a,c

Clayton H. Brown, PhD,a,d Richard Goldberg, PhD,d,e Joseph H. Rosenberg, BS,a

Angela C. Comer, MPH,a and Jon P. Furuno, PhDf

OBJECTIVES: To quantify the risk of 30-day unplannedhospital readmission in adults aged 65 and older withdepressive symptoms.

DESIGN: Prospective cohort study.

SETTING: University of Maryland Medical Center.

PARTICIPANTS: Individuals aged 65 and older admittedbetween July 1, 2011, and August 9, 2012, to the generalmedical and surgical units and followed for 31 days afterhospital discharge (N = 750).

MEASUREMENTS: Primary exposure was depressivesymptoms at admission, defined as a score of 6 or more onthe 15-item Geriatric Depression Scale. Primary outcomewas unplanned 30-day hospital readmission, defined as anunscheduled overnight stay at any inpatient facility notoccurring in the emergency department.

RESULTS: Prevalence of depressive symptoms was 19%and incidence of 30-day unplanned hospital readmissionwas 19%. Depressive symptoms were not significantlyassociated with hospital readmission (relative risk(RR) = 1.20, 95% confidence interval (CI) = 0.83–1.72).Age, Charlson Comorbidity Index score, and number ofhospitalizations within the past 6 months were significantpredictors of unplanned 30-day hospital readmission.

CONCLUSION: Although not associated with hospitalreadmission, depressive symptoms were associated withother poor outcomes and may be underdiagnosed inhospitalized older adults. Hospitals interested in reducing

readmission should focus on older adults with more co-morbid illness and recent hospitalizations. J Am GeriatrSoc 62:495–499, 2014.

Key words: hospital readmission; depressive symptoms;older adults

Unplanned hospital readmission has been targeted forquality improvement initiatives to reduce healthcare

costs and improve outcomes. Previous research has sug-gested that 20% of Medicare beneficiaries are readmittedto the hospital within 30 days of discharge and that 34%are readmitted within 90 days.1 Hospital readmissions arecostly as well, accounting for $15 billion in Medicarespending in 2007.2 The Medicare Payment AdvisoryCommission tied 30-day readmission rates to Medicarereimbursements in 2013 with three diagnoses-relatedgroups initially targeted, but the number of targeteddiagnoses groups will expand in 2014.3

Interventions to reduce hospital readmissions havebeen observed to be effective, but cost efficiency dictatesthat they be targeted toward individuals at highest risk ofreadmission.4–6 Risk factors for readmission include olderage, male sex, recent previous hospital admission, longerhospital stay, and greater comorbidity,1,7,8 but thesecharacteristics provide few targets for anything other thanbroad-based interventions.

Depression may affect hospital readmissions inMedicare beneficiaries.9–11 Research examining the effectof depressive symptoms on hospital readmission has founda positive association.12–16 Nonetheless, a focus on illnessessuch as heart failure, age groups that included much youn-ger individuals (≥18), and variable readmission outcomeslimits the generalizability of previous results to an older,more-general population that may be more consistent withcurrent hospital initiatives to reduce readmissions.

This article reports results from a prospective cohortstudy on the association between depressive symptoms atbaseline and unplanned 30-day hospital readmission inindividuals aged 65 and older. It was hypothesized that

From the aDepartment of Epidemiology and Public Health, School ofMedicine, University of Maryland, bDepartment of Pharmaceutical HealthServices Research, School of Pharmacy, University of Maryland,cDepartment of Emergency Medicine, School of Medicine, University ofMaryland, dVeterans Affairs Capitol Healthcare Network, Mental IllnessResearch, Education, and Clinical Center, eDepartment of Psychiatry,School of Medicine, University of Maryland, Baltimore, Mayland; andfDepartment of Pharmacy Practice, College of Pharmacy, Oregon StateUniversity/Oregon Health & Science University, Portland, Oregon.

Address correspondence to Jennifer S. Albrecht, PhD, Department ofPharmaceutical Health Services Research, University of Maryland Schoolof Pharmacy, 220 Arch Street, 12th floor, Room 01–234, Baltimore, MD21201. E-mail: jalbrecht@rx.umaryland.edu

DOI: 10.1111/jgs.12686

JAGS 62:495–499, 2014

© 2014, Copyright the Authors

Journal compilation © 2014, The American Geriatrics Society 0002-8614/14/$15.00

depressive symptoms would be associated with greater riskof 30-day hospital readmission. If this is the case, screen-ing for and treating depression in this population mayprovide an additional means of reducing hospital readmis-sion, as well as furnishing a target population for currentinitiatives.

METHODS

Study Design and Study Population

This was a prospective cohort study of adults aged 65 andolder admitted to the hospital. Study participants wereenrolled within 72 hours of admission. After hospitaldischarge, participants were followed up by telephone atthree time points (5, 15, and 31 days) after discharge toascertain deaths and unplanned readmission events.

The study population consisted of community-dwell-ing adults aged 65 and older admitted to the generalmedical and surgical services of the University of Mary-land Medical Center (UMMC), a 757-bed, tertiary-carehospital in Baltimore, Maryland, between July 1, 2011,and August 9, 2012. Individuals admitted to psychiatric,obstetrical, and intensive care units; residing in a nursinghome; unable to speak English; or with a Mini-MentalState Examination (MMSE) score of 15 less or wereexcluded.17 The institutional review board at the Univer-sity of Maryland Baltimore approved this study, and allparticipants provided written informed consent beforeparticipating.

Admission and discharge data on individuals meetingthe inclusion criteria were collected daily from the UMMCClinical Data Repository (CDR), a relational databaseincluding administrative, demographic, and outcome infor-mation.

Necessary sample size was calculated a priori basedupon the assumptions of 13% incidence of hospital read-mission, 20% prevalence of depressive symptoms, and10% loss to follow-up. The 13% incidence rate in adultsaged 65 and older was obtained from 5 years of data on30-day hospital readmission to UMMC from the CDR.Based on these assumptions, a sample size of 750 wouldallow 80% power to detect twice the risk of 30-dayunplanned readmission (relative risk = 2) in study partici-pants with depressive symptoms.

Measures

Study participants were administered a baseline question-naire. Prevalent clinically significant depressive symptoms,the primary exposure, were assessed using the 15-itemGeriatric Depression Scale (GDS-15) and defined as ascore of 6 or greater.18 This cut-point has been observedto have a sensitivity of 83% and a specificity of 69% todetect depression in elderly inpatients.19 Disabilities inactivities of daily living were assessed using the Katzscale (range 0–5, 5 indicating the highest level of disabil-ity); cognitive impairment using the 30-item MMSE;and social isolation using the 6-item Lubben Social Net-work Scale,17,18,20,21 which has been validated as a mea-sure of social isolation risk in community-dwellingadults. A score of <12 has been suggested to define

social isolation, which was used in this study.20 Demo-graphic information on race, sex, and education wasself-reported.

In addition to the questionnaire, data on admissionand discharge diagnoses (including the primary diagnosisand up to 15 comorbid medical conditions), number ofmedications prescribed at discharge, discharge instructions,and readmissions to UMMC were collected from partici-pants’ medical charts. Data on Charlson ComorbidityIndex score, a measure of aggregate comorbidity, andmedications prescribed at discharge were collected fromthe CDR.22

Disease categories were created based on participantdischarge diagnosis International Classification of Diseases,Ninth Revision codes.23,24 The following categories werecreated: cancer (140–208, 230–234), heart disease (391–392.0, 393–398, 402, 404, 410–416, 420–429), diseases ofthe digestive system (520–579), diseases of the musculoskel-etal system (710–739), and complications of surgical andmedical care (996–999). The remaining diagnoses weregrouped as “other.”

Study participants were contacted over the telephone 5,15, and 31 days after discharge from the index hospitaladmission to determine the incidence of 30-day unplannedhospital readmission. During these telephone calls, datawere collected on any hospital readmissions that occurredafter discharge, including date, place, reason for readmis-sion, and whether the readmission was planned orunplanned. An unplanned hospital readmission was definedas an overnight stay at any inpatient facility that didnot occur in the emergency department and was not previ-ously scheduled. If the participant was unavailable at thefollow-up call, outcome information was obtained fromthe contact person that the participant had indicated atenrollment.

Data Analysis

The primary exposure (clinically significant depressivesymptoms) and outcome (unplanned 30-day hospitalreadmission) were treated as dichotomous variables.Association between the exposure and outcome variablesand covariates was examined using chi-square tests fordichotomous covariates and Student t-tests for normallydistributed continuous covariates to assess potential con-founding. The Wilcoxon rank sum test was used to assessassociations between nonnormally distributed variables.Statistical significance was defined as P < .05. Covariatesthat were significantly associated with the exposure andthe outcome and covariates known to be associated witheither from the literature were considered for inclusion inthe regression model.

A log-binomial regression model was used to estimatethe relative risk of 30-day unplanned hospital readmissionin the study population. The primary log-binomial regres-sion model contained an indicator variable for prevalentdepressive symptoms and modeled the log relative risk of30-day unplanned hospital readmission. Effect modifiersidentified using the Breslow-Day test (with P ≤ .05) andtheir main effects were also examined in the model. Alldata analysis was performed using SAS version 9.2 (SASInstitute, Inc., Cary, NC).

496 ALBRECHT ET AL. MARCH 2014–VOL. 62, NO. 3 JAGS

RESULTS

During the study period, 3,699 participants aged 65 andolder were admitted to the general medical and surgicalunits at UMMC; 146 of these (4%) were not competent toparticipate, and 61 (2%) did not speak English. Of theremaining 3,492, 750 (21%) were enrolled in the study.Reasons for nonparticipation included refusal (21%);already discharged from hospital (16%); already partici-pating in study (6%); and inaccessibility of the participantbecause of procedures, therapy, or sleep (25%). Individu-als who enrolled in the study were slightly younger, with amean age � standard deviation of 73.2 � 6.5 (vs73.9 � 6.9, P = .01), and more likely to be Caucasian(70% vs 60%, P < .001). Participants did not differ fromnonparticipants according to sex (49% vs 52% female,P = .08) or Charlson Comorbidity Index (2.5 � 2.3 vs2.6 � 2.4, P = .70).

Of the 750 participants enrolled in the study, 13(1.7%) died during their hospital stay, four (0.5%) with-drew from the study, and 17 (2.3%) were lost to follow-up at 31 days. This left 716 (95%) individuals for whomdata on the primary outcome were ascertained.

Mean age was 73.2 � 6.5, and mean MMSE scorewas 27.9 � 8.1 (Table 1). Length of hospital stay wasskewed, with a median of 4 days (interquartile range0–9 days). Nineteen percent of the sample had depressive

symptoms (mean GDS-15 score 3.2 � 2.8). The most-pre-valent discharge diagnosis was heart disease (20%). The“other” category comprised 35% of discharge diagnosesand included (as a percentage of “other”) chronic kidneydisease (17%), diseases of the respiratory system (11%),stroke or transient ischemic attack (7%), septicemia (3%),fracture (3%), diseases of the nervous system (3%), and“symptoms” (15%).

One hundred forty-two (19%) study participantsexperienced an unplanned 30-day hospital readmission; 68of these (48%) were readmitted to UMMC. Prevalentreadmission diagnoses were heart failure (13%), complica-tions (13%), “symptoms” (13%), and digestive problems(6%). Myocardial infarction and respiratory problemsaccounted for 3% and 2% of readmission diagnoses,respectively. Thirty-five percent of readmission diagnosesfell into the “other” category.

Participants with depressive symptoms had greaterbaseline morbidity than those without. Differences in self-rated health of good or better (30% vs 65%, P < .001),one or more ADL disabilities (31% vs 16%, P < .001),two or more hospital admissions in the past 6 months(36% vs 22%, P < .001), two or more falls in the last6 months (25% vs 12%, P < .001), number of medica-tions prescribed at discharge (11.2 vs 10.0, P = .01), andmean Charlson Comorbidity Index score (3.1 vs 2.4,P = .004) reflected this. A doctor had previously told 48%

Table 1. Characteristics of the Study Population of Hospitalized Individuals at University of Maryland MedicalCenter, 2011–2012 According to Depressive Symptom Status

Characteristic

Total,

N = 750

Depressive Symptoms,

n = 140aNo Depressive

Symptoms, n = 610 P-Valueb

Age, mean � SD 73.2 � 6.5 72.8 � 6.5 73.4 � 6.6 .35Mini-Mental State Examination score,mean � SD (range 0–30)

27.7 � 2.4 27.0 � 2.7 27.8 � 2.4 .002

Length of stay, days, median (IQR) 4.0 (5.0) 4.0 (5.5) 4.0 (5.0) .20Female, n (%) 366 (49) 74 (53) 292 (48) .29White, n (%) 540 (72) 101 (72) 439 (72) .97Married, n (%) 433 (58) 72 (51) 361 (59) .09Self-rated health good or better, n (%) 439 (59) 41 (30) 398 (65) <.001≥1 activity of daily livingdisabilities in, n (%)

141 (19) 43 (31) 98 (16) <.001

≥High school education, n (%) 636 (85) 110 (79) 526 (86) .02≥2 hospital admissions inlast 6 months, n (%)

184 (25) 50 (36) 134 (22) <.001

≥2 falls in last 6 months, n (%) 108 (14) 35 (25) 73 (12) <.001Lives alone, n (%) 199 (27) 35 (25) 164 (27) .62Social isolation, n (%)c 89 (12) 34 (24) 55 (9) <.001Number of medications atdischarge, mean � SD

10.2 � 4.7 11.2 � 5.4 10.0 � 4.5 .01

Primary discharge diagnosis, n (%)Cancer 97 (13) 20 (14) 77 (13) .60Heart disease 151 (20) 27 (19) 124 (20) .78Diseases of the digestive system 93 (12) 17 (12) 76 (12) .92Complications of medical care 83 (11) 19 (14) 64 (10) .29Diseases of the musculoskeletal system 65 (9) 11 (8) 54 (9) .71Other 261 (35) 46 (33) 215 (35) .59Charlson Comorbidity Index,median (IQR) (range 0–37)

2.0 (3.0) 2.5 (3.5) 2.0 (2.0) .001

SD = Standard Deviation; IQR = Interquartile Range.aGeriatric Depression Scale-15 score ≥6.bChi-square for categorical variables, Student t-test for continuous variables, Wilcoxon for length of stay, and Charlson Comorbidity Index.cLubben Social Network Scale-6 score <12.

JAGS MARCH 2014–VOL. 62, NO. 3 DEPRESSION AND READMISSION IN OLDER ADULTS 497

of participants with depressive symptoms that they weredepressed.

Based on the bivariate analyses and previous research,the final regression model included terms for age, sex,Charlson Comorbidity Index, social isolation risk, hospitaladmissions in the past 6 months, and self-rated health.

In the adjusted model, an association between depres-sive symptoms and risk of hospital readmission wasobserved, although it did not achieve statistical significance(relative risk (RR) = 1.20, 95% confidence interval(CI) = 0.83–1.72) (Table 2). Age (3% increase in risk peryear of age), Charlson Comorbidity Index (7% increase inrisk for each point), and hospital admissions in the past6 months (53% increase in risk for ≥2) significantlypredicted 30-day unplanned hospital readmission.

DISCUSSION

Depressive symptoms were not significantly associatedwith risk of unplanned 30-day hospital readmission. Olderage, higher Charlson Comorbidity Index, and previoushospitalizations within the past 6 months were all associ-ated with a significantly greater risk of 30-day unplannedhospital readmission in adults aged 65 and older.

These results contrast with previous studies that havereported positive associations between depression and hos-pital readmission.12–16 There are a number of possibleexplanations for this. Although the sample size was largerthan in most previous studies, the observed effect size(20% greater risk) was smaller than has been reported. Inaddition, this study focused on general medical and surgi-cal patients aged 65 and older and 30-day unplanned read-mission and used the GDS-15, a validated measure inhospitalized elderly adults, to measure depressive symp-toms.19 The more-stringent measures, ability to follow upwith participants readmitted to non-UMMC hospitals, anddefinition of 30-day unplanned hospitalization may haveaffected the size of the effect observed. Previous researchexamining the effect of depressive symptoms on hospitalreadmission has primarily focused on individuals with spe-cific illnesses such as heart failure, age groups thatincluded much younger individuals (≥18), and readmissionoutcomes that varied from 30 days to 1 year and mayhave included planned events.12–16 Furthermore, previousinstruments used to measure depressive symptoms haveincluded somatic measures, which can be confused with

symptoms of comorbid illness in older adults. This vari-ability in previous studies may have resulted in largereffect sizes than observed in this study. Finally, a recentsystematic review concluded that readmission risk predic-tion models generally perform poorly because of the multi-factorial nature of hospital readmission.25 Depressivesymptoms may interact with facility- and individual-levelcharacteristics not found in this study to affect risk.

This study was powered at 80% to detect a doublingof the risk of 30-day unplanned hospital readmission inindividuals with depression based on a sample size of 750and the assumptions described in the methods, but theobserved association suggests a smaller population effectthan assumed in the sample size calculations, leading to aloss of statistical power. Nevertheless, the outcome wasassessed in more than 97% of study participants, and anincidence of 30-day unplanned hospital readmission con-sistent with that reported in a national study of Medicareenrollees was observed, lending validity to the results.1

Individuals with greater depressive symptoms mayhave been less likely to enroll in the study. Nonetheless,the reported prevalence of depressive symptoms (19%) isconsistent with prior research in older hospitalized popula-tions (range 14–25%).26,27 Individuals with a MMSE score<15 and those residing in nursing homes were not eligibleto participate in this study, although it was not thoughtthat these exclusions greatly affect generalizability becausenursing home residents represent only a small percentageof hospitalized older adults, and cognitive impairment hasnot been reported to be a risk factor for hospital readmis-sion.

This single-site study reports incidence of 30-dayunplanned hospital readmission and prevalence of depressivesymptoms consistent with national samples, helping to sup-port its generalizability.1,26,27 Furthermore, this study con-firmed previously observed associations between unplanned30-day hospital readmission and older age, greater comor-bidity, and prior hospital admissions.1,7,8,13,28–30 Only 48%of readmissions were to the index facility. This may haveled to an underestimation of 30-day hospital readmissionrates in the sample size calculations and has implicationsfor institutions attempting to reduce readmissions.

In this large prospective cohort study, a 20% increasein the risk of 30-day unplanned hospital readmission inadults aged 65 and older with depressive symptoms wasobserved but was not statistically significant. Hospitalsinterested in effectively using limited funds to reducehospital readmission in older adults should focus on olderadults with a greater burden of comorbid illness and priorhospital admissions. Nonetheless, the low prevalenceof doctor-recognized depression in people exhibitingdepressive symptoms presents an opportunity for a publichealth intervention to increase screening for depression inolder adults. Further research is needed to determine thecost effectiveness of screening for depression and the effi-cacy of interventions to reduce hospital readmission in thispopulation.

ACKNOWLEDGMENTS

Conflict of Interest: The authors declare no conflicts ofinterest.

Table 2. Risk of 30-Day Unplanned Hospital Readmis-sion (N = 716)

Characteristic

Unadjusted Adjusted

Relative Risk (95% Confidence

Interval)

Depressive symptoms 1.31 (0.93–1.84) 1.20 (0.83–1.72)Age, per year 1.03 (1.00–1.05)Female 0.88 (0.65–1.18)Charlson Comorbidity Index 1.07 (1.01–1.14)≥2 admissions in last 6 months 1.53 (1.12–2.09)Self-rated health good or better 0.90 (0.65–1.24)Social isolation 0.72 (0.43–1.21)

498 ALBRECHT ET AL. MARCH 2014–VOL. 62, NO. 3 JAGS

This work was supported by Agency for HealthcareResearch and Quality (AHRQ) Grant R36HS021068–01(JSA); National Institutes of Health GrantsT32AG000262–14 (JSA), K01AI071015–05 (JPF); andNational Heart, Lung, and Blood Institute Grant R01HL085706 (ALG-B).

Author Contributions: Albrecht: study concept anddesign, acquisition of subjects and data, analysis and inter-pretation of data, preparation of manuscript. Gruber-Baldini,Hirshon, Goldberg, Furuno: study concept and design, inter-pretation of data, preparation of manuscript. Brown: studyconcept and design, analysis and interpretation of data,preparation of manuscript. Rosenberg, Comer: studyconcept and design, acquisition of subjects and data, prep-aration of manuscript.

Sponsor’s Role: The AHRQ had no direct role in thedesign, methods, subject recruitment, data collections,analysis, or preparation of manuscript. No other sponsorshad a direct role in design, methods, subject recruitment,data collections, analysis, or preparation of manuscript.

REFERENCES

1. Jencks SF, Williams MV, Coleman EA. Rehospitalizations among patients

in the Medicare fee-for-service program. JAMA 2009;360:1418–1428.2. Report to the Congress: Reforming the Delivery System. Chapter 4, A Path

to Bundled Payment Around a Hospitalization. Washington, DC: Medicare

Payment Advisory Commission; 2008:89–103 [on-line]. Available at: http://

www.medpac.gov/documents/Jun08_EntireReport.pdf Accessed May 29,

2012.

3. Medicare Payment Advisory Commission. Report to the Congress; Medi-

care Payment Policy: MedPac, March 2012 [on-line]. Available at: http://

www.medpac.gov/documents/Mar12_EntireReport.pdf Accessed May 29,

2012.

4. Mistiaen P, Francke AL, Poot E. Interventions aimed at reducing problems

in adult patients discharged from hospital to home: A systematic meta-

review. BMC Health Serv Res 2007;7:47.

5. Naylor MD, Brooten D, Campbell R et al. Comprehensive discharge plan-

ning and home follow-up of hospitalized elders: A randomized clinical trial.

JAMA 1999;281:613–620.6. Jack BW, Chetty VK, Anthony D et al. A reengineered hospital discharge

program to decrease rehospitalization. A randomized trial. Ann Intern Med

2009;150:178–187.7. Hasan O, Meltzer DO, Shaykevich SA et al. Hospital readmission in gen-

eral medicine patients: A prediction model. J Gen Intern Med

2010;25:211–219.8. Kossovsky M, Perneger T, Sarasin F et al. Comparison between planned

and unplanned readmissions to a department of internal medicine. J Clin

Epidemiol 1999;52:151–156.9. Beekman AT, Penninx BW, Deeg DJ et al. Depression and physical health

in later life: Results from the Longitudinal Aging Study Amsterdam

(LASA). J Affect Disord 1997;46:219–231.10. Penninx BW, Geerlings SW, Deeg DJ et al. Minor and major depression and

the risk of death in older persons. Arch Gen Psychiatry 1999;56:889–895.

11. Bruce ML, Leaf PJ. Psychiatric disorders and 15-month mortality in a com-

munity sample of older adults. Am J Public Health 1989;79:727–730.12. Song E, Lennie T, Moser D. Depressive symptoms increase risk of rehospi-

talisation in heart failure patients with preserved systolic function. J Clin

Nurs 2009;18:1871–1877.13. Muzzarelli S, Leibundgut G, Maeder M et al. Predictors of early readmis-

sion or death in elderly patients with heart failure. Am Heart J

2010;160:308–314.14. Jiang W, Alexander J, Christopher E et al. Relationship of depression to

increased risk of mortality and rehospitalization in patients with congestive

heart failure. Arch Intern Med 2001;161:1849–1856.15. Smolderen K, Spertus J, Reid K et al. The association of cognitive and

somatic depressive symptoms with depression recognition and outcomes

after myocardial infarction. Circ Cardiovasc Qual Outcomes 2009;2:

328–337.16. Rumsfeld JS, Jones PG, Whooley MA et al. Depression predicts mortality

and hospitalization in patients with myocardial infarction complicated by

heart failure. J Am Heart 2005;150:961–967.17. Folstein MF, Folstein SE, McHugh PR. ‘Mini-mental state’. A practical

method for grading the cognitive state of patients for the clinician. J

Psychiatr Res 1975;12:189–198.18. Sheikh VI, Yesavage VA. Geriatric Depression Scale (GDS): Recent evi-

dence and development of a shorter version. In: Brink TL, ed. Clinical Ger-

ontology: A Guide to Assessment and Intervention. New York: Hawthorn

Press, 1986, pp 165–173.19. Lesher EL, Berryhill JS. Validation of the Geriatric Depression Scale—

short-form among inpatients. J Clin Psychol 1994;50:256–260.20. Lubben J, Blozik E, Gillmann G et al. Performance of an abbreviated

version of the Lubben Social Network Scale among three European

community-dwelling older adult populations. Gerontologist 2006;46:

503–513.21. Katz S, Ford AB, Moscowitz RW et al. Studies of illness in the aged: The

index of ADL: A standardized measure of biological and psychosocial

function. JAMA 1963;185:914.

22. Charlson M, Szatrowski T, Peterson J et al. Validation of a combined

comorbidity index. J Clin Epidemiol 1994;47:1245–1251.23. Centers for Medicare and Medicaid Services. Chronic Conditions Ware-

house [on-line]. Available at http://www.ccwdata.org/chronic-conditions/

index.htm Accessed April 3, 2012.

24. Buie VC, Owings MF, DeFrances CJ et al. National Hospital Discharge Sur-

vey: 2006 summary. Vital Health Stat 13(168) 2010 [on-line]. Available at

http://www.cdc.gov/nchs/data/series/sr_13/sr13_168.pdf Accessed September

25, 2012.

25. Kansagara D, Englander H, Salanitro A et al. Risk prediction models for

hospital readmission: A systematic review. JAMA 2011;306:1688–1698.26. Covinsky KE, Fortinsky RH, Palmer RM et al. Relation between symptoms

of depression and health status outcomes in acutely ill hospitalized older

persons. Ann Intern Med 1997;126:417.

27. Koenig HG, George LK, Peterson BL et al. Depression in medically ill hos-

pitalized older adults: Prevalence, characteristics, and course of symptoms

according to six diagnostic schemes. Am J Psychiatry 1997;154:

1376–1383.28. Gorman J, Vellinga A, Gilmartin JJ et al. Frequency and risk factors associ-

ated with emergency medical readmissions in Galway University Hospitals.

Ir J Med Sci 2010;179:255–258.29. Mudge AM, Kasper K, Clair A et al. Recurrent readmissions in medical

patients: A prospective study. J Hosp Med 2011;6:61–67.30. Marcantonio ER, McKean S, Goldfinger M et al. Factors associated with

unplanned hospital readmission among patients 65 years of age and older

in a Medicare managed care plan. Am J Med 1999;107:13–17.

JAGS MARCH 2014–VOL. 62, NO. 3 DEPRESSION AND READMISSION IN OLDER ADULTS 499