Outcomes of community-based social service interventions in homebound elders

Outcomes of community-based social service interventionsin homebound elders

Mary E. Charlson1*, Janey C. Peterson1, Beth L. Syat2, William M. Briggs3, Robin Kline4,Mary Dodd5, Velda Murad5 and William Dionne5

1Center for Complementary and Integrative Medicine, Weill Cornell Medical College, New York, NY, USA2Boston Health Economics, Waltham, MA, USA3Department of Mathematics, Central Michigan University, Mt. Pleasant, MI, NY, USA4Columbia Law School, New York, NY, USA5The Burden Center for the Aging, Inc., New York, NY, USA

SUMMARY

Objective To document within-client change in function and quality of life over 6 months, and determine whether socialservice interventions, comorbidity, depressive symptoms, social support and stress are predictors of within-patient change.Method Assess homebound elderly referred for social services on depressive symptoms measured by the GeriatricDepression Scale (GDS), comorbidity with the Charlson Index, and stress and support with the Duke instrument. Functionwas measured by the Functional Autonomy Scale (FAS), measuring Activities of Daily Living (ADL), IndependentActivities of Daily Living (IADL), mobility, communication and mental function. The SF-36 measured quality of life.Results Among 56 new homebound clients with an average age of 82, 33% had depressive symptoms at baseline (>7 onthe GDS). At baseline clients were at or below 25th percentile for five of eight domains of the SF-36, and mental and physicalsummary scores. Further at baseline, 90% had difficulties with mobility and IADLs; 61% had ADL limitations. At 6-monthfollow-up overall, 26% had depressive symptoms at follow-up. Greater comorbidity was associated with more depressivesymptoms at both baseline and follow-up. By 6 months, 18% had deteriorated on the FAS, while 11% improved. More clientshad changes in quality of life; regarding the physical component score, 13% had important deterioration, while 63%improved. Similarly, 33% declined on the mental component while 27% improved.Conclusion Among newly homebound elders, those with significant depressive symptoms are more likely to experiencedeterioration in function and quality of life over 6 months. However, those with more support showed significantimprovement in the SF-36 mental component scale at 6 months. Copyright # 2007 John Wiley & Sons, Ltd.

key words—homebound persons; frail elderly; social welfare

INTRODUCTION

Cross-sectional studies of homebound elders reportmore comorbidity, worse self-reported health status,greater impairment in activities of daily living (ADLs),more depressive symptoms and less social supportamong the homebound (Bruce and McNamara, 1992;Ganguli et al., 1996; Herbert et al., 1997; Pasternaket al., 1998; Korten et al., 1999). Homebound statusimplies social isolation and lack of support, important

independent predictors of poor outcomes in elderlypatients (Seeman and Berkman, 1988; Parkersonet al., 1989). Longitudinal outcome studies ofcommunity dwelling elderly show that about 20%experience functional decline each year, while 7%improve (Berkman and Syme, 1979; Hebert et al.,1988; Hanson et al., 1989; Herbert et al., 1997; Kortenet al., 1999). Though one might anticipate thathomebound elderly fare worse than non-homebound,there is little literature on longitudinal outcomes ofhomebound elderly. The only study that followedclients receiving home care services found that 15%had major depressive disorder. Patients with more

INTERNATIONAL JOURNAL OF GERIATRIC PSYCHIATRY

Int J Geriatr Psychiatry 2008; 23: 427–432.

Published online 4 October 2007 in Wiley InterScience

(www.interscience.wiley.com) DOI: 10.1002/gps.1898

*Correspondence to: Dr M. Charlson, 1300 York Avenue, Box 46,New York, NY 10065, USA. E-mail: [email protected]

Copyright # 2007 John Wiley & Sons, Ltd.Received 19 May 2006Accepted 31 July 2007

recent stress and Independent Activities of DailyLiving (IADL) limitations were more likely to remaindepressed (Orth-Gomer and Johnson, 1987).Among newly homebound elderly serviced by the

Burden Center, a model community service agency,the objective of this prospective study was todocument within-client change in function (primaryoutcome) and quality of life (secondary outcome) over6 months and determine whether social service inter-ventions, comorbidity, depressive symptoms, socialsupport or stress were predictors of change in functionand quality of life. Function was measured with theFunctional Autonomy Scale, (FAS) including ADLs,IADLs, mobility, communication and mental func-tion. Quality of life was measured by the SF-36.Depressive symptoms were measured with theGeriatric Depression Scale (GDS) (Sheik and Yesa-vage, 1986), support and stress with the Duke scale(Parkerson et al., 1991) and comorbidity with theCharlson index (Charlson et al., 1987).

METHODS

The community social service center

The Homebound Unit of the Burden Center is acommunity-based agency specifically structured tosupport homebound clients. The Center provides arange of programs to homebound individuals, includ-ing counseling, case management, benefit assistance,money management and legal assistance. Volunteersprovide friendly visiting, food shopping, medicalescorting, telephone reassurance, transportation andhousekeeping. Two luncheon clubs provide hotlunches, educational workshops andMeals onWheels.The Burden Center has an information systemtracking the social service needs of patients, makingit uniquely positioned to study outcomes of interven-tions.

Screening, eligibility and enrollment

To be eligible, homebound clients (i.e. clients wholeave their home less than once a week) had to be�60years, without dementia or major psychiatric disease.Caseworkers from the Burden Center screened newlyhomebound clients for eligibility between February2002 and February 2003. Of the 106 newly referredhomebound clients, 44 were deemed ineligible bytheir caseworkers because their Mini Mental score was<23 or they required anti-psychotic medicine. Of the62 eligible clients, 57 patients, or 90%, consented tothe study. All procedures were approved by the

Institutional Review Board (IRB) at Weill CornellMedical College. Participants provided informed,written consent as per IRB protocol.

Baseline evaluation

Once the client agreed to participate, a researchassistant accompanied the caseworker to the client’shome, and administered the baseline evaluation, in-cluding demographics, entitlements, income, andservices provided. Clinical data included comorbidmedical and psychiatric problems, medications, andthe Charlson Comorbidity Index score (Charlsonet al., 1987). The Geriatric Depression Scale ShortForm (Sheik and Yesavage, 1986) was employed tomeasure depressive symptoms;

GDS57 indicates important depressive symptoms.The Duke Social Support and Stress Scale (Parkersonet al., 1991) rated support or stress provided by familyand non-family members. FAS (Hebert et al., 1988) isa reproducible, valid measure (Desrosiers et al., 1995)of performance evaluating five domains: ADLs;IADLs; mobility; communication; and mentalfunction. The FAS yields a single score; a changeof 5 points in the overall FAS is acknowledged as aclinically important difference (Hebert et al., 1988).The SF-36 assessed health- related quality of life(Ware et al., 1994), measuring eight domains withscores from 0–100; higher scores indicate betterfunction. The SF-36 has summary scores for mentaland physical quality of life.

Follow-up evaluations

At 6 months following enrollment, each client wasre-assessed by telephone using the same measuresadministered at baseline by the same research assistantwho did the initial interview.

Outcomes

The principal outcome was change in function asmeasured by the FAS at 12 months (Hebert et al.,1988), specifically within-client change in the overallscore FAS score between baseline and follow-up.Clients who had improved by 5 points or more in scorewere designated as improved; change betweenþ4 and�4 was designated as unchanged (Heber et al., 1997).Patients whose score had decreased by 55 pointswere designated as having a functional decline.

The secondary outcome––change in quality oflife––was measured by within-client differencebetween baseline and follow-up in physical and

Copyright # 2007 John Wiley & Sons, Ltd. Int J Geriatr Psychiatry 2008; 23: 427–432.

DOI: 10.1002/gps

428 m. e. charlson ET AL.

mental summary component SF-36 scores (Ware,1994). A change of 55 points indicated a clinicallyimportant difference (Ware, 1994).

Analysis

We calculated separate ordinary linear regressions ofbaseline variables, such as GDS, comorbidity score,Duke support, stress, and social services received inorder to predict within-patient change of both FAS andSF-36. Patients’ age was included as a controllingvariable. Standard regression diagnostics show thatthe variables met the assumptions for our model withreference to normality of residuals.

RESULTS

The 57 homebound clients had a mean age of 82 years(range 67–97), 90% were Caucasian and 77% werewomen. The vast majority, 90%, lived alone.

One-fifth of patients reported previous myocardialinfarction (Table 1). Twenty-one percent reportedhistory of diabetes; 30% had a cancer history.Twenty-eight percent of patients had a comorbidityscore of >4, and 36% had a score of 2–3.

The mean GDS was 6.3 at baseline. A third ofclients reported they felt helpless, worthless, bored orthat their ‘life is empty’. Overall, 33% had GDS 57,indicating clinically important depressive symptomsat baseline, 12% had scores 510, indicating majordepressive symptoms (Sheik and Yesavage, 1986).

The Duke Social Support and Stress Scale(Parkerson et al., 1991) was used to rate the amountof support or stress from family and non-family. Thebaseline mean Duke support score was 29, indicatinglow support. This is not surprising, since 90% livedalone. At baseline, 81% of clients had someone totrust; 53% trusted a family member, 38% a friend, and13% a social worker. Of note, 29% indicated theydistrusted someone; almost all distrust involved familymembers. The Duke baseline mean stress score was 11,indicating low stress from family and non-familymembers––potentially indicative of social isolation.Stress scores were higher for family members, asmany clients felt neglected by or estranged fromchildren.

FAS scores rate independence to complete depen-dence in five sectors; more negative scores indicateworse function (Hebert et al., 1988). Sixty-one percentof clients reported ADL difficulties, including wash-ing, dressing, and grooming. Ninety-five percent ofclients reported difficulties with mobility, commonlystairs and walking outdoors, and 11% were home-

bound because of severely impaired mobility. Fiftypercent of clients had communication difficulties,16% reported they were only able to see objectoutlines and 4% were blind. Sixty-four percent ofclients reported memory problems; 91% reportedIADL difficulties, including telephone use, medi-cations and finances. The mean score at baselinewas �17.Clients had an SF-36 mean score of 13 for social

function, significantly lower than the 25th percentilescore of 50 for age-matched controls. Clients alsoindicated that their physical and/or emotional healthinterfered ‘extremely’ with normal social activities.Homebound clients had a mean score of 22 forphysical function, near the 25th percentile forage-matched controls. Their mental health, pain andenergy/vitality scores were all near the 25th percentilefor controls. However, role-emotional, role-physicaland general health scores were closer to populationmedians. The average physical and mental componentsummary scores were 27 and 45, respectively, at orbelow the 25th percentile for age appropriate norms.

Table 1. Baseline clinical characteristics of new homeboundclients (n¼ 57)

Clinical features

History of Myocardial Infarction 21%History of Congestive Heart Failure 9%History of Hypertension 60%History of Peripheral Vascular Disease 34%History of Peripheral Bypass Operation 10%History of Stroke (with residua) 12%History of Diabetes 21%History of Cancer 30%

Charlson Comorbidity Score0–1 36%2–3 36%54 28%

MedicationsAnti hypertensives 9%ASA 11%Beta blockers 9%Channel blockers 9%Cholesterol 9%Coumadin 2%Diabetes 12%Digoxin 2%Diuretics 4%Estrogen receptor blockers 2%Glaucoma 4%H2 blockers 2%Immunosuppressives 7%Nitrates 4%Sedatives or Hypnotics 5%Thyroid 7%Vitamins 9%


DOI: 10.1002/gps

social service interventions in homebound elders 429

Social services provided

Clients received an average of 1.2 services from theBurden center. Seventeen percent received luncheonservices through the Carter Burden Luncheon Club orthe Knickerbocker program. Twenty-nine percentreceived Meals on Wheels, 23% received assistanceshopping, housekeeping, or transportation services,and 18% received telephone reassurance or friendlyvisiting. An additional 37% had entitlement evalu-ation, financial counseling, and end-of-life planning.The Luncheon Club and Knickerbocker programsinvolve daily social contact, making them the mostintense interventions, followed by Meals on Wheels.

Follow-up at 6 months

At 6 months, six clients had died, one had moved outof the area, one was in a nursing home, and four werelost to follow-up, leaving 45 clients.About 19% of clients suffered a decline (�5 points)

in overall FAS, but 16% of clients reported animprovement (�5 points). In the subscales (Table 2),

most showed little change, although 12% showedimprovement in IADL.

For the SF-36 (Table 3), the mean within-patientchange in the physical component score (PCS) wasabout 1 and 2 for the mental component score (MCS).Five or more is a clinically significant change. About63% of clients’ physical component scores improved,while 13% reported a negative change of at least thesame amount. For the mental component score, about27% of clients improved and 33% declined.

Greater comorbidity was associated with moredepressive symptoms at baseline and follow-up.Overall, 26% still had depressive symptoms atfollow-up. About 7% of clients developed newsymptoms at follow-up, with a baseline GDS lessoverall mean than 7 and a GDS >7 at 6 months. Atfollow-up, the Duke Support score dropped from 26 to14, indicating decreased support. Correspondingly,clients reporting no one to trust increased from 19% to36%. However, the mean Duke Stress score decreasedfrom 21 to 5, indicating less stress. Depression wasassociated with support and stress; specifically, higherGDS scores were indicative of lower support and

Table 2. Functional autonomy scores at baseline and follow-up for new homebound clients

Component Baseline Follow-up Within patient difference Percent improved �5 Percent worse �5

ADL �1.60 �1.60 0.065 2 2Mobility �3.70 �3.30 �0.033 2 0Communications �0.75 �0.93 0.217 0 0Mental �0.65 �0.43 �0.196 2 0IADL �0.93 �0.84 �0.641 12 4Overall FAS �16.0 �14.6 �0.587 18 11

Score of 0 is best; negative scores are worst. For example, with meal preparation 0¼ ‘prepares their own meals’;�1¼ ‘prepares own mealsbut needs guidance to maintain adequate nutrition’;�2¼ ‘only prepares light meals or heats up pre-prepared meals’;�3¼ ‘does not preparemeals’.

Table 3. Mean SF-36 Scores at baseline and follow-up for new homebound clients

Domain Baseline* Follow-up* Improved 5 points Declined 5 points Age matched population norms

25th Percentile 50th Percentile

Physical function 22 26 23% 19% 26 55Role physical 23 48 35% 11% 0 25Role emotional 77 72 11% 19% 0 100Social function 13 21 40% 21% 50 88Pain 45 55 35% 16% 41 62Mental health 50 53 23% 18% 64 80Vitality 36 36 30% 32% 35 50General health 52 57 25% 19% 42 59Physical component 29 34 44% 9% 33 47Mental component 45 41 18% 21% 48 66

*Scores range from 0–100, higher is better.


DOI: 10.1002/gps


higher stress on the Duke scales. There was norelationship between GDS and age.

Higher scores on the baseline GDS were associatedwith worse overall Functional Autonomy Scores at 6months (Table 4). Not surprisingly, higher GDS scoresat baseline were highly correlated with the mentalcomponent summary score of the SF-36 (p¼ 0.0003)and with all domains of the SF-36 except physicalfunction.

Multivariate analyses explored the predictors ofwithin-patient change in FAS (primary outcome) andthe physical and mental component scores of theSF-36 (secondary outcome). Baseline FAS or baselineSF-36 dominated the predictivemodel. Controlling forFunctional Autonomy Score at baseline, only olderage had borderline significance in predicting FASat 6-months (p¼ 0.08). With regard to secondaryoutcomes, baseline mental component scores onthe SF-36 (p¼ 0.04) and social support (p¼ 0.02)predicted mental summary component scores onthe SF-36 at 6 months. With regard to physicalcomponent scores on the SF-36, only the baselinephysical component score predicted 6 month scores(p< 0.0001).

DISCUSSION

Epidemiologic studies of community-based popu-lations show that 10% of people >75 are homebound(Ganguli et al., 1996). Homebound elderly are morelikely to be women, have IADL impairment and

greater depressive symptoms (Fredman et al., 1989;Seeman et al., 1989; Ganguli et al., 1996); those withcomorbid illnesses are more likely to have depressivesymptoms (Seeman et al., 1989). In the EpidemiologicCatchment’s Area study, homebound patients werethree times more likely to have a major depressiveepisode and twice as likely as non-homeboundpatients to have a dysthymic disorder (Fredmanet al., 1989; Bruce and McNamara, 1992). Depressionis a major predictor of mortality in frail commu-nity-dwelling elders (Seeman et al., 1989; Yaffe et al.,2003), and therefore for homebound elders as well.Previous longitudinal studies provided context for

interpreting changes in our study’s participants (Gillet al., 1997a,b; Hebert et al., 1997; Korten et al.,1999). In a study of elders with ADL dependence,28% recovered independence over 1-year, demon-strating that disability does not inexorably lead todeterioration (Gill et al., 1997b). Similarly, our studyshowed functional decline was not inevitable; roughlyequal proportions of homebound elders improved anddeteriorated. These outcomes were associated withservices received.Since there is no comparison group not who had

received social services, we cannot draw conclusionsabout elderly with no social services support.However, any study of homebound elderly wouldinvolve clients receiving services (Brown et al., 2003).Although the sample size in this study was small

and underpowered in definitively addressing outcomepredictors among newly homebound patients, itsuggests that support and depression deserve furtherevaluation as predictors of long-term function andquality of life. Importantly, no homebound clientswere on anti-depressants, suggesting a potentialimportant intervention. Interventions have beenstudied in this vulnerable population (Clarke et al.,1992; Tinetti et al., 1994; van Haastregt et al., 2000),however, few have targeted depressive symptomsspecifically, rather than merely as a potentialconfounder of outcomes. Though there has beenrapid growth of programs designed for the homebound(Clarfield and Bergman, 1991; Kellogg and Brickner,2000), home health providers may not recognizedepressive symptoms among homebound patients,resulting in lack of appropriate services (Brown et al.,2003).

CONCLUSION

Functional decline and worsening health status are notinevitable for homebound frail elderly; however,depressive symptoms identify clients at high risk

Table 4. Relation of Baseline Geriatric Depression Score to func-tion and quality of life at 6 months Regression analysis

Coefficient P-value

FAS at 6 monthsADL 0.04 0.69Mobility �0.13 0.37Communication �0.01 0.75Mental �0.03 0.37IADL �0.16 0.44Overall �0.30 0.40

SF-36 at 6 monthsPhysical function �0.7 0.46Role physical �8.1 0.0004Role emotional �8.0 0.0001Social function �2.9 0.008Pain �4.3 0.007Mental �5.1 0.000001Vitality �3.5 0.006General Health �4.2 0.0006Physical SummaryMental Summary


DOI: 10.1002/gps

social service interventions in homebound elders 431

for poor outcomes. Greater attention is needed oninterventions targeted to such symptoms.

ACKNOWLEDGEMENTS

Supported by the New York Community Trust.Financial Disclosure(s): Mary E Charlson: none;

Janey C. Peterson: none; Beth L. Syat: none; WilliamM. Briggs: none; Robin Kline: none; Mary Dodd:Director of the Homebound Unit, Burden Center forthe Aging; Velda Murad: Director of Social Services,Burden Center for the Aging; William Dionne: Execu-tive Director, Burden Center for the Aging.Author Contributions: Mary E Charlson: design;

Janey C. Peterson: design; Beth L. Syat: concept,enrollment, data collection; William M. Briggs:analysis; Robin Kline: design; Mary Dodd: design,recruitment; Velda Murad: design; William Dionne:design.

REFERENCES

Berkman LF, Syme SL. 1979. Social networks, host resistance, andmortality: a nine-year follow-up study of Alameda County resi-dents. Am J Epidemiol 109(2): 186–204.

Brown EL, McAvay G, Raue PJ, et al. 2003. Recognition ofdepression among elderly recipients of home care services.Psychiatr Serv 54(2): 208–213.

Bruce ML, McNamara R. 1992. Psychiatric status among thehomebound elderly: an epidemiologic perspective. J Am GeriatrSoc 40(6): 561–566.

Charlson ME, Pompei P, Ales KL, et al. 1987. A new method ofclassifying prognostic comorbidity in longitudinal studies: devel-opment and validation. J Chronic Dis 40(5): 373–383.

Clarke M, Clarke S, Jagger C. 1992. Social intervention and theelderly: a randomized controlled trial. Am J Epidemiol 136:1517–1523.

Clarfield AM, Bergman H. 1991. Medical home care services for thehousebound elderly. CMAJ 144(1): 40–45.

Desrosiers J, Bravo G, Hebert R, et al. 1995. Reliability of therevised functional autonomy measurement system for epidemio-logic research. Age Aging 24: 402–406.

Fredman L, Schoenbach VJ, Kaplan BH, et al. 1989. The associationbetween depressive symptoms and mortality among older partici-pants in the Epidemiologic Catchment Area-Piedmont HealthSurvey. J Gerontol 44(4): S149–156.

Ganguli M, Fox A, Gilby J, et al. 1996. Characteristics of ruralhomebound older adults: a community-based study. J Am GeriatrSoc 44(4): 363–370.

Gill TM, Robison JT, Tinetti ME. 1997a. Predictors of recovery inactivities of daily living among disabled older persons living inthe community. J Gen Intern Med 12(12): 757–762.

Gill TM, Williams CS, Richardson ED, et al. 1997b. A predictivemodel for ADL dependence in community-living older adults

based on a reduced set of cognitive status items. J Am Geriatr Soc45(4): 441–445.

Hanson BS, Isacsson SO, Janzon L, et al. 1989. Social network andsocial support influence mortality in elderly men. The prospectivepopulation study of ‘Men born in 1914’, Malmo, Sweden. Am JEpidemiol 130(1): 100–111.

Hebert R, Brayne C, Spiegelhalter D. 1997. Incidence of functionaldecline and improvement in a community-dwelling, very elderlypopulation. Am J Epidemiol 145(10): 935–944.

Hebert R, Carrier R, Bilodeau A. 1988. The Functional AutonomyMeasurement System (SMAF): description and validation of aninstrument for the measurement of handicaps. Age Aging 17(5):293–302.

Kellogg FR, Brickner PW. 2000. Long-term home health care for theimpoverished frail homebound aged: a twenty-seven-year experi-ence. J Am Geriatr Soc 48(8): 1002–1011.

Korten AE, Jorm AF, Jiao Z, et al. 1999. Health, cognitive,and psychosocial factors as predictors of mortality in an elderlycommunity sample. J Epidemiol Community Health 53(2): 83–88.

Orth-Gomer K, Johnson JV. 1987. Social network interaction andmortality. A six year follow-up study of a random sample of theSwedish population. J Chronic Dis 40(10): 949–957.

Parkerson GR Jr, Michener JL, Wu LR, et al. 1989. Associationsamong family support, family stress, and personal functionalhealth status. J Clin Epidemiol 42(3): 217–229.

Parkerson GR Jr, Broadhead WE, Tse CK. 1991. Validation ofthe Duke Social Support and Stress Scale. Fam Med 23(5):357–360.

Pasternak R, Rosenweig A, Booth B, et al. 1998. Morbidity ofhomebound versus inpatient elderly psychiatric patients. IntPsychogeriatr 10(2): 117–125.

Seeman TE, Berkman LF. 1988. tructural characteristics of socialnetworks and their relationship with social support in the elderly:who provides support. Soc Sci Med 26(7): 737–749.

Seeman TE, Guralnik JM, Kaplan GA, et al. 1989. The healthconsequences of multiple morbidity in the elderly. The AlamedaCounty study. J Aging Health 1(1): 50–66.

Sheik JI, Yesavage JE. 1986. Geriatric Depression Scale (GDS):recent evidence and development of a shorter verson. In ClinicalGerentology: A Guide to Assessment and Intervention, Brink TL(ed.). Hawthorn Press: New York; 165–173.

Tinetti M, Baker D,McAvay G. 1994. Amultifactorial interention toreduce the risk of falling among elderly people living in thecommunity. N Engl J Med 33: 821–827.

van Haastregt J, Diederiks J, van Rossum E, et al. 2000. Effects of aprogramme of multifactorial home visits on falls and mobilityimpairments in elderly people at risk: randomised controlled trial.BMJ 321: 994–998.

Ware JE, Snow KK, Kosni M, et al. 1993. SF-36 Health Survey:Manual and Interpretation Guide. The Health Institute, NewEngland Medical Center: Boston, MA.

Ware J. 1994. SF-36 Physical and Mental Health Summary Sales: AUser’s Manual. The Health Institute, New England Medical:Center: Boston, MA.

Yaffe K, Edwards ER, Covinsky KE, et al. 2003. Depressivesymptoms and risk of mortality in frail, community-living elderlypersons. Am J Geriatr Psychiatry 11(5): 561–567.


DOI: 10.1002/gps


Documents

Outcomes of community-based social service interventions in homebound elders