More effective and less expensive: lessons from five studies examining community approaches to care

ELSEVIER

Health Policy 34 (1995) 95-112

More effective and less expensive: lessons from five studies examining community approaches to care Ark-am Gafnia’c’d, Gina Browne* a,b*cJacqueline Robertsa>b9C, Robin Wei;b, Susan Watta3e9f,Carolyn ByrneavbTg ‘System-Linked Research Unit on Health and Social Service Utilization, Faculty of Health Sciences, McMaster University, RM 3947, I2tW Main Street, West, Hamilton, Ont. L8N 325, Canadn bSchool of Nursing. McMaster University, RM 3947, 1200 Main Street, West, Hamilton, Ont. L&N 3Z5, Cd ‘Department of Clinical Epidemiology and Biostatistics. McUaster University, RM 3947, I200 Main Street, West, Hamilton, Ont. L8N 325, Canada dCentre for Health Economics and Policy Analysis, M&faster University, RM 3947, 1200 Main Street, West, Hamilton, Ont. L8N 325, Cam& ‘School of Social Work, Mchiaster University, RM 3947, 1200 Main Street, West, Hamilton, Ont. L8N 325, Canada fEducational Centre for Aging and Health, M&faster University, RM 3947, 1200 Main Street, West, Hamilton, Ont. L8N 325, Canada STeaching Health Unit, McMaster University, RM 3947, 1200 Main Street, West, Hamilton, Ont. L8N 325. Cana& Received 21 September 1994; revision received 13 January 1995; accepted 17 January 1995

AhStTllCt Does the nature of community health servicesusedby chronically ill clients and their caregivershave an impacton utilization of services,expenditureand well-beingoutcomes? A seriesof five studies,(four historic cohort and one randomizedtrial) examinedclientssuffering from a variety of chronic conditions in a number of community settingsin different regionsof Southern Ontario. Study samplecompositionand sire varied. Each study was designedto quantify the well-beingoutcomes,and expenditureassociated with different community approaches is coveredundera nation-widesystemof healthinsuranceplans.As a collective, thesestudiesrepresentincreasingmethodologicalrigor. Multiple-perspectiveclient well-beingoutcomemeasures wereused.Caregiverburdenwasalsoanalyzed.A commonap* Corresponding author, Tel: +I 905 5259140 (ext. 22293); Fax: +l 905 5210048. 0168-8510/95BO9.50 0 1995 Elsevier Science Ireland Ltd. All rights reserved SSDI 0168-8510(95)00725-8

%

G. Browne

et al. /Health

Policy

34 (1995)

95-112

preach to quantification and evaluation of expenditure for service consumption was applied across all five studies. The nature of community health services (proactive versus reactive approaches to care) was found to have direct and measurable impact on total expenditure for health service utilization and client well-being outcomes. A recurring pattern of lower expenditure for community health service utilization and equal or better client outcomes was associated with well-integrated proactive services when compared with individual fragmented, reactive approaches to care. The main lesson emerging from examining the five studies on approaches to community care is that it is as, or more, effective and less expensive to offer complete proactive health care services to chronically ill people in the early stages of their illness than to provide services on demand in a piecemeal manner. Keywords: Utilization and delivery of communityhealthservices;Chronic illness;Healthcare

expenditure;Mental health;

1. Introduction

Chronically ill people consume health care resources disproportionate to their numbers. Planners, providers and insurers are looking for community approaches to care to contain expenditure for the chronically ill while maintaining or enhancing client well-being outcomes. In the current economic climate community health services are also experiencing cutbacks in funding and/or limitations on expansion because of a lack of evidence to support their value within the overall system. Initially, research on community health services focused on the expenditure and effects of community care strategies compared with institutionalization for the treatment and on-going management of chronically ill clients. Less is known about the expenditure and effects of alternative community care strategies based on different assumptions about the problem and potential solutions. This paper summarizes the results of five studies designed to quantify the well-being and expenditure outcomes associated with different approaches to community care of clients with select chronic conditions within a national system of health insurance. The scant amount of information available in the literature is fragmented and provides conflicting information. For example, Hedrick et al. [l], reviewed the randomized and non-randomized evaluations of adult day care in the United States. In aggregate, the non-randomized studies provided evidence of improved client functioning and reduced cost with day care in general. These results conflicted with three randomized trials that provided modest evidence of improved patient function with a comparative increase in overall costs associated with the provision of day care. These US studies looked at community care settings where there was some attempt to provide the professional component of institutional care in a community setting during the daytime. These were not proactive, preventative, holistic community approaches and therefore did not represent the full spectrum of strategies that not only are feasible but are actually advocated today. Moreover, these studies did not examine expenditure associated with alternative community strategies for the same client group. They lacked a common approach to the measurement of community

G. Browne et al. /Health Policy 34 (1995) 95-112

91

service expenditure and well-being outcomes of similar clients exposed to different types of community service. Without this type of comparison across studies, samples and settings, one cannot draw meaningful conclusions about the most effective and efficient community strategy. The variety of approaches to community care of chronically ill clients derive from different assumptions about the health care problem and the potential solutions. Some approaches to community care assume a medical perspective of the client as suffering from physical and/or mental impairments. Caregiving goals subsequently focus on repair or replacement of specific deficits on demand. This is a reactive approach to providing care. Clients or their caregivers seek out one or more health care services in an effort to ‘fix’ the impairment. The initial impairment is treated where possible but quality of life issues are generally not addressed. Confinement, immobility and isolation risk factors often impact upon these individuals causing further physical, mental and emotional decline. A corollary of this scenario is that client dependency on the health care system is intensified. The use of health services often substitutes for other non-medical services in a system where only health care is insured. From an individual patient perspective, accumulating evidence illustrates that high use of health services is explained more by emotional, social and attitudinal issues than by issues of disease severity [2,3]. Thus, high utilization of health services is a type of help-seeking behaviour that reflects unmet needs of chronically ill clients. Other approaches to community care are based upon the assumption that the health care problem(s) facing each chronically ill client always poses biopsychosocial sequelae which may accompany or exacerbate the initial impairment(s). Caregiving goals under this perspective take on more holistic parameters. These include proactive measures such as needs anticipation and assessment with appropriate support services. The system is set up to prevent further decline due to the impairment and biopsychosocial sequelae in addition to providing rehabilitation and continued access to medical treatment as required. This preventative and health promotion perspective aims to build on strengths and capabilities of clients, empowering them to minimize the impact of risk factors associated with their limitations. Such community services are shown in our studies to be associated with lower frequency of health care service utilization, equal or better client well-being outcomes and lower overall cost. This paper describes the results of five studies of community services (four cohort analytic and one randomized trial) which were conducted in 1991 and 1992 in Southem Ontario, Canada. In Ontario, insured community service strategies for a variety of chronic problems may be provided in-home, in day programs or in ambulatory clinics. Within each study, a common approach to the measurement of client wellbeing and expenditure for utilization were used to assess the effect of one of these approaches to community care. Across studies a variety of measures of well-being were used yet each study employed the same measure of expenditure for service utilization. Expenditure for services provided to clients engaged or enrolled in these services was compared with that for those referred to but waiting for enrolment or those declining specific services. Studies I-IV compare outcome and expenditure for service utilization patterns

98

G. Browne

et al. /Health

PO/icy 34 (1995)

95- I12

under conditions of self-selection. Study V illustrates the same pattern under conditions of random allocation. The five studies involved day programs, mental health services, speciality pain clinics, internal medicine clinics and community day rehabilitation programs. In all five studies with multiple outcome measures, the more complete approach to community care was associated with equal or higher levels of well-being and less expense to society as a whole than those strategies that were based primarily on the medical perspective. In this era of cost containment and demand for fiscal restraint, there is an opportunity for health care systems to contain or reduce their costs while improving the effectiveness of the services delivered. Scarce resources can be freed and allocated in a more appropriate manner. Our analysis of five studies illustrates that this can be done through the wider use of community services that are based upon a more holistic philosophy of care-giving and a more complete and coordinated plan of action. 2. Methods and measures An historic cohort analytic study design was used in Studies I-IV. Subjects who had been referred to a specific community ambulatory service were categorized as either currently waiting, enrolled or declining this service. Subjects were assessed after a minimum of 2 months of exposure to one of these categories of services. A random allocation procedure was used in Study V which controls for problems of self-selection and non-comparability of groups that may have confounded Studies I-IV. Study I by Milne et al. [4], was a historic cohort analytic study which examined the well-being of 46 of 64 caregivers whose cognitively impaired relative was referred to a Victorian Order of Nurses (VON) day program during the past year for one-half to one full day a week. The 46 caregivers were either of relatives enrolled in the program (N = 24), waiting for the program (N = 14) or having refused the program (N = 8). Caregiver well-being was measured using the Crumbaugh Purpose-in-Life Measure [5,6], PRQ 85 Measure of Social Support (71, the Zarit’s Measure of Caregiver Burden [8,9] and part A (health services used) of the Use of Health Services Utilization Questionnaire [2]. Study II was conducted by Emond and Browne [lo]. It was used as a historic cohort analytic study to examine the mental health of 55 clients referred to a community mental health clinic who were either ‘in-treatment’ (Group 1), ‘completed treatment’ at the mental health clinic (Group 2), ‘rerouted’ (Group 3), or ‘not engaged’ in treatment (Group 4). Mental health was measured using the ‘General Health ~estionnaire’ [l l] and use of health and social services measured by the Browne et al. [2] Health Service Utilization Inventory. Study III by Weir et al. [3], was a historic analytic survey of 571 patients referred to a speciality pain clinic who either ‘completed treatment’ (N = 83), ‘failed to complete advent’ (N = 98), received a ‘consult only’ (N = 180), or ‘declined ~eatment’ (N = 180). A sample of 222 patients were approached to assesstheir current adjustment to illness measured by the Psychosocial Adjustment to Illness Inventory (PAISSR) [ 121 and Use of Health and Social Services [2].

G. Browne

et al. /Health

Policy ‘34 (3995)

95-l

12

99

Study IV conducted by Browne et al. [13] also used a historic cohort analytic design to examine the functional capacity of 152 disabled adults referred to a Community Health Services Centre (CHSC) for rehabilitation. Clients ‘enrolled’ received either 1 or 2 days per week of the day program. The functional capacity and use of services by clients discharged from the program (N = 19), enrolled in the day program funded by the Ministry of Health (MOH) (N = 68), enrolled in the day program funded by the Ministry of Community and Social Services (MCSS) (N = 26) were compared with those not enrolled in the program (N = 34) and in institution (N = 5). Client functional capacity was measured using the Older American Resource Services Questionnaire (OARS MFAQ) [14,15]. Use of services was measured by Browne et al. [2] using the Health Service Utilization Inventory (HSUI). Studies I-IV are studies of patients or their caregivers with comparable illnesses and disease severity who maintain their well-being or adjustment status while selfselecting to one community approach over another. Study V by Roberts et al. [16], was a randomized trial and controls for the methodological problems of self-selection which confound Studies I-IV. In addition, Study V measures client change over time which can be attributed to one service strategy over another. During this 3 year trial, 293 newly referred consenting and representative chronically ill subjects with fair to poor levels of psychosocial adjustment to their physical illness were randomly allocated to receive, in addition to their usual speciality medical clinic care, either an experimental problem-solving intervention (lx-l = 4.6 in l-h sessions), phone call support (lx-l = 5.06 in 6-min calls), or conventional medical clinic care alone over a 6 month period. Approximately 86% of subjects had two or more sessions and 81% of those allocated were retained by Times (Ts 6 months after the period of intervention). Those retained in the study had poorer adjustment than those lost. In Study V, baseline and follow-up measures were taken at trial entry and 6 months following entry into the study, which corresponded to the time immediately following the intervention. Measures were taken again in 6 months after the intervention had ceased (7’s). Follow-up measures included sociodemographic variables, severity of illness, the Derogatis and Lopez [12] Psychosocial Adjustment to Illness Questionnaire, the Crumbaugh [6] Purpose-In-Life Questionnaire and the Browne, et al. [2] Health and Social Service Utilization Questionnaire which was translated into a measure of expenditure for services. The methodological details of each study are summarized in Table 1. 2.2. Outcome measurement In each of the first four studies a variety of ways to measure adjustment to chronic illness and overall well-being at one point in time were used. Factors such as the client’s and/or caregiver’s perceptions about his or her purpose-in-life, social support networks, caregiver burden, mental health of clients, adjustment to illness and overall functional status were measured. The actual degree of improvement (change) in adjustment and well-being was measured in the fifth study. The purpose-in-life measure was used in studies of caregivers of cognitively impaired relatives (Study I) and chronically ill speciality clinic outpatients (Study V).

Sample and size

n Day Program - Enrolled - Waiting - Refused

user groups

-

n Mental Health Program Enrolled Completed Rerouted Not engaged (III) Historic cohort analytic Clients referred to Pain n Pain Clinic (Weir et al., 1992) clinic - Completed N = 222 respondents of 571 - Incomplete referred - Consult only -Non-users (IV) Historic cohort analytic Clients referred to Comn Rehabilitation Day (Browne et al., 1994) munity Rehabilitation Program Day Program - Received (Completed) N = 152 respondents of 255 - Enrolled in MCSS Simureferred lation Program - Enrolled in MOH Physical Rehabilitation prOgrElXll - Not enrolled - Institutionalized (V) Prospective randomized Poorly adjusted chronically n Medical care with trial (Roberts et al., ill medical outpatients of - Problem solving 1995) one of 27 Speciality counselling Clinics - Telephone support N = 293 randomized and - Medical care alone completed 1 year followup of 490 eligible

(I) Historical cohort analytic Caregiven of cognitively im(Mime et al., 1994) paire4I relatives referred for Day Program N = 46 respondents of 64 referred (II) Historic cohort analytic Clients referred to Mental (Bmond et al., 1992) Health Program N = 55 respondents of 163 referred

Study design

Table I Methodological summary of studies

Users of better in activities of daily living

Clinic users with counselling Users with counselling or phone support less exor phone supports (PCS) improved in purpose-inpensive $4298-S&89 vs. life and psychosocial ad$6365 justment to illness scores

Users less expensive Sl792-$2219 vs. $16607

Groups similar - Psychosocial adjustment (PAIS) scores

Enrollees in Day Program less expensive $6779 vs. $8235

Dollars of expenditure for use of all health services per person per annum per group

Current and former enrollees in Mental Health Program less expensive S 1386-$2765 vs. $13756 Users of Pain Clinic services less expensive S3285-$3,801 vs. $6008

Groups similar - Mental health (GHQ scores)

Groups similar - Purpose-in-life scores - Social support scores - Burden scores

Effects

5

G. Browne

et al. /Health

Policy

34 (1995)

95-112

101

It is a 20-item, 7-point scale designed by Crumbaugh and Maholick [6] to measure the degree to which a person experiences a sense of meaning or purpose-in-life. Splithalf correlation co-efficient was 0.85-0.92. Test-retest ranges were between 0.68 and 0.83 [6] and construct validity was assessed for seniors by its relationship to a recent measure of the meaning in life (r = 0.77) and affect balance scale (r = 0.63) [ 171. The score ranges between 20 and 140. Scores below 92 indicate lack of a clear purpose, between 92 and 111 indicates indecisiveness and 112 indicates definite purpose and meaning in life [6]. Social Support was measured in the study of caregivers (Study I) by the PRQ Personal Resources Questionnaire [7] and in the study of patients (Study III) by the Duke-UNC Functional Support Questionnaire [18]. The PRQ is a second generation multidimensional 25-item, 7-point scale measure of social support consisting of five dimensions of support: intimacy, social integration, nurturance, worth, assistance/guidance. The PRQ has been tested on older persons, has a test-retest reliability of 0.72 and it distinguishes anxious and depressed populations in predicted directions [7]. The Duke-UNC Functional Support Questionnaire measures two components of perceived emotional support: confident support, reflecting a contident relationship where important matters in life are discussed and shared and affective support, reflecting a more emotional form of support and caring. This 2-scale, S-item questionnaire has demonstrated construct, concurrent and discriminate validity [18]. Caregiver burden was measured in studies of caregivers of severely cognitively (Study I) or physically impaired relatives (Study IV). It is a 22-item, 5-point scale to evaluate the subjective impact of caregiving [9]. The internal consistency, has been estimated using Chronbach’s alpha = 88 and it has a test-retest reliability of 0.71. Validity was assessed by correlating the total score with a single burden item (r = 0.71) and with the Derogatis et al. [12] Brief Symptom Inventory (BSI), (r = 0.41) [19]. A reverse score provides a ‘without burden’ score. The mental and emotional well-being of clients at an outpatient mental health clinic (Study IV) was studied using the General Health Questionnaire (GHQ), developed by Goldberg [ 111. It is a self-administered screening instrument designed to detect current, diagnosable psychiatric disorders. The GHQ is intended to identify the ‘inability to carry out one’s normal healthy functions and the appearance of new phenomena of a distressing nature’ [20]. Probable cases are identified on the basis of the subject endorsing any 12 or more of 60 items included. It is meant to be a preliminary screening instrument for identifying psychiatric disorder in which the emphasis is placed on change in condition [21]. The main version of the GHQ contains 60 items and Goldberg recommends using this version because of its superior validity. The GHQ score rates problems as either present or absent [21]. Goldberg [20] reported the reliability on the 60-item version to be 0.95 [21]. Validation studies, undertaken in many different countries and using comparable procedures, have consistently shown a high degree of validity. Falsepositives are subjects with substantial physical illness [21]. Chronically ill outpatients attending speciality medical clinics were studied using the Psychosocial Adjustment to Illness Scale - Self-Report (PAIS-SR) to measure

102

G. Browne et al. /Health Policy 34 (1995) 95-112

the outcome of primary interest in Studies III and V. Weisman’s review [22] of psychosocial adjustment cited PAIS as the only appropriate measure for a physically disadvantaged group. Morrow et al. [23] and Derogatis and Lopez [12] developed PAIS-SR to measure psychosocial adjustment, including intrapsychic processes, as well as interactions between the individ~l, others and institutions that represent one’s so&cultural environment. This 46-item self-report assessesadjustment to a current medical illness in seven domains (Health Care Orientation, Vocational Environment, Domestic Environment, Sexual Relationships, Extended Family Relations, Social Relations and Psychological Distress). Each item is rated on a 4point scale, with a score of 35-51 representing fair adjustment and 52 representing poor adjustment to illness [2,12,24]. Since the PAIS measures no change or change for the worse, we added one question to each of the primary domains in Study V and asked patient to rate these for any degree of enhancement on a scale of O-3. Internal consistency of PAIS is 0.81, interrater reliability is 0.86 and the construct validity, judged by factor analysis, is moderate to good [22]. The interrelationship among the seven domains is a reflection of the breadth and sensitivity of the measurement, which should be low. For PAIS, the average was 0.31. Convergent validity, which involved correlating the PAIS Total Adjustment score with the Global Adjustment to Illness Scale, was 0.81. In support of predictive validity, PAIS discriminated between those with a confirmed diagnosis of lung cancer and those without [12]. The widely used multidimensional Older American Resource and Services Questionnaire (Part A) or OARS-MFAQ [14] was used in Study IV with physically impaired clients referred to a community rehabilitation centre. The Multidimensional Functional Assessment Questionnaire (MFAQ) Part A measures functional ability in five domains: social and economic resources, mental health, physical health and self-care capacity (ADL). Part B measures services utilized by the older person. For the purposes of this study, Part A only was used. It was selected over others mainly because of its comprehensiveness as well as other psychometric properties. Moditications were made in the OARS instrument regarding income and expenses to correspond with the 1991 Canadian situation 1151. The validity of this instrument was considered in that items agreed on by ‘experts’ were included. Concurrent validity was assessed by comparing ratings from the instrument with ratings from clinical interviews by clinicians; correlations ranged between 0.67 and 0.89. Intrarater reliability was assessed using the instrument’s predecessor, the longer version Community Survey Questionnaire, on subjects 12-18 months after the original rating. Seven of the original ten raters participated and the correlates of the ratings ranged from 0.47 to 1.0, the majority being at least 0.85. The questionnaire was also rated by different raters with interrater reliability on all domains of function ranging predominantly between 0.80 and 0.97. The tool distinguishes the functional status of elderly in the community, clinic and institution [14,15]*

3. Health service use and expenditure measurement The same approach to the measurement of expenditure for health and social ser-

G. Browne

et 01. /HeoIth

~olicy’34~(1995)

9S-112

103

vices utilized was employed in all five studies of diverse community service strategies for chronic populations. The utilization of the different types of health and social services (measured as resources used) and the unit costs were obtained in order to calculate the total expenditure for services used. (Expenditure refers to both direct and indirect charges for services used by the chronically ill. Direct expenditure includes those charges for the use of health services such as professional medical care, hospital care, emergency room care, medications, physiotherapy, social work, nursing. Indirect expenditure refers to the time lost from working while waiting for or receiving treatment.) This variable was measured by a modified version of an inventory developed by Browne [2,13]. It consists of questions about the respondent’s use of categories of direct health and social services: primary care, emergency room and specialists; number of hospital admissions, length of time for hospitalizations and the use of seven types of other health professionals and in-home workers such as nurses, homemakers, physiotherapists, adult day care providers and social workers. Inquiries are ‘restricted to the reliable duration of recall space, i.e. 6 months for remembering a hospitalization, 2 weeks for a visit to physician and 2 days for the consumption of a prescription medication [25]. The various health service use time spans are extended to yield an annual rate of utilization per category of health ser‘vice. Annual utilization rates are then multiplied by service dollar values to yield a measure of expenditure on health services which include home/community-based program costs. (The detailed method of calculating expenditure for each category of health and social service use is available [28]. In summary, these 1991 charges for direct services used in these five studies were taken from the 1991 Ontario Ministry of Health Schedule of Benefits where, for example, the average charge in Canadian Dollars for a family physician visit was $33.50, or specialist physician consultation was $104.40; an emergency room physician fee was $26.30 plus use of the emergency room was $34.78 for a total of $61.08 per emergency visit. The average charge for a hospital day in Ontario in 1991 was $505.22; for a physiotherapy treatment was $52.50; for an occupational therapy visit was $50.00; for social work visits was S67.50; nutritionists was $17.50; community nurse (including salary, overhead and car allowance) was $35.83 per visit; chiropractor was $34.00 per visit; homemakers was $45.36 per visit; meals on wheels was $6.20 per meal; psychologists was $90.00 per visit; chiropody was $26.00 per visit; acupuncture was $33.65 per visit; massage therapy was $50.30 per hour; naturopath was $44.50 per visit; geriatric day care was $39.00 per day; muse counsellor was $26.00 per hour; volunteer agency ‘support service’ was $4.79 per visit; optometrist was $54.10; ambulance was $49.99 per trip. Diagnostic test charges were also set using the Ontario Ministry of Health 1991 Schedule and drug charges from the 1991 Ontario Ministry of Health Drug Formulary. More information regarding the calculation of indirect costs and cash transfers is available.) Questions were added to the health service utilization inventory in order to assess direct, out-of-pocket expenditure including number and type of medications (prescription and non-prescription) used in the preceding 2 days; type of medical devices used in the previous 2 weeks; dollars spent on outside babysitting services (due to illness or treatment) per episode (times the number of episodes in a 2 week span); dollars spent on transportation to and from the health services (times the number

104

G. Browne

et al. / Health

Policy

34 (1995)

95-112

of trips in a 2 week span); dollars spent on parking (if applicable) at the health facility (times the number of visits over a 2 week span). Indirect costs during the preceding 2 weeks also are assessed by this inventory. Questions are asked about the amount of lost income by the patient and by their caregivers due to the illness or its treatment; amount of days absent from work by patient and by their caregivers due to illness or its treatment (measured because many have the benefit of paid sick leave). Cash transfer effects of illness are assessed by multiplying the amount per check by the number of checks received in the previous 2 weeks, from sources such as unemployment insurance, worker’s compensation, disability pension, mother’s (family) allowance and private insurance. As with the direct health services, these out-of-pocket expenditures, indirect costs and cash transfer effects of illness are extended to represent the total annual expenditure. This measure has been previously tested [2,13] and assessed for reliability and validity. High levels of observed agreement (0.72-0.99) between the patient’s report and the clinic record have recently been reported [2,13]. 4. ResuIts The results of each study are presented separately in Figs. 1 and 2. Thereafter, a comparison is made of findings across studies. Although the results of all studies are consistent, some of the non-random studies (Studies II and IV) show much bigger differences in expenditure than the randomized trial (Study V). 4.1. Study results In Study I [4], caregivers of relatives with severe cognitive impairment, receiving one of three approaches to community care, exhibited equivalent and high levels of perceived support, purpose and realistic views of the burden imposed by giving care to their relative. Caregivers of relatives enrolled and waiting for the day program exhibited more unfavourable meaning given illness (F = 2.7-3.4; P = O.OS-0.02), compared to the meaning held by caregivers who had declined the use of the day program for their relative. Caregiver unfavourable meaning consisted of views that their relative’s illness had ‘adversely affected their lives’, ‘was harmful’ and/or a ‘disability’. It appeared that the program reached those caregivers who most needed respite but, on the whole, caregivers could maintain their optimistic outlooks through different types of community services. In spite of these equivalent outcomes for caregivers who self-selected to different approaches to community care, there was an administratively important difference in expenditure for caregiver relief and the relatives’ care depending on the approach to community care. Those households with a person waiting to enrol or refusing the day program used disproportionately more distinct types of community service. Those waiting or refusing the program used more visiting nursing, homemaking, meals on wheels, general practitioner and helping hands resources compared with those enrolled. The annual per person expenditure for the use of all direct health services by those enrolled was $6779.00 compared with $7318.00 for those waiting versus $8235.00 for those who refused the day program. The direct expenditure associated with expan-

Outcomes

of Caregivers

STUDY 1 Impaired Relatives Enroled

STUDY IV Outoomw of Clients Receiving Types of Community Rehabilitation Day Programmes

of Cognitivdy

In Day Programme

Outcomes Outpatlent

STUDY II of Clients Rwxivlng Mental Health Servloes

STUDY V of Chronically Ill Attending One of 27 MedIcal Clinics Racdvlng One of 3 Types of Medical Outpatient Services

Outcome Community

Fig. 1. The well-being outcomes of clients and/or caregivers receiving/not receiving different types of community care.

STUDY Ill of ChronIcally Ill Attending Specialty Pain Clinic

Outcomes

Fig. 2. The health and

Thous.nds

EXPENDITURES COMM”NlM

social

S

service

FOR CLIENTS REHABlUTATlDN

EXPENDITURES FOR COGNITIVELY IMPAIRED RELATIVES ENROLLED IN DAY PROGRAMME

WPES

OF

t

FOR MENTAL

use (% of expenditure)

RECEMNG DAY PR-ES

Thouvnda

EXPENDITURES COMMUNITY Thwnrnde

EXPENDITURES ATTENDINQ S

Thousands

S

receiving

specified

services.

FOR CHRONICALLY ILL SPECIALTY PAIN CUNIC

EXPENDITURES FOR CUEMS AlTENDlNG 1 OF 27 MEDICAL D”TPATlENT CUMCS RECEMNQ ONE OF 3 TYPES OF MEDICAL OUTPATIENT SEWICES

RECEIVING SERVICES

of clients receiving/not

CLIENTS HEALTH

G. Browne et al. /Health Policy 34 (ENS) 95-112

107

ding the program to accommodate the waiting list would be less than the expenditure associated with waiting for the program. Respondents in Study II (compared with non-respondents) [lo], were more often female, married, English-speaking and referred for follow-up and treatment rather than assessment/consultation or counselling. As a group, respondents were about 38 years old, female, secondary-school educated, married, English-speaking, had an estimated household income of $20 000-$30 000 per year, unemployed and had purchased their own homes. Thus, statements about the mental health of Study II respondents may be an underestimate of the mental health of the referral population; however, statements about the use of health services by these same respondents maybe an overestimate considering that females and English-speaking clients are generally higher users of services. Respondents who were ‘non-engaged’ in the mental health service examined in Study II, were significantly more worried about their health, more often unemployed, from non-intact marriages, yet better educated and financially better off. A sizeable proportion of respondents who were ‘rerouted’ to other services, had poor mental health, less meaningful support and greater involvement with other agencies and probably should have been retained for treatment. Respondents in Study II who ‘completed’ treatment had, on average, a more favourable general health score, less cash transfer effects of illness and a higher estimate of their current health indicating that this group had improved in their mental health status as a result of treatment. Respondents in the treated groups were more often married and employed, had purchased their own homes and had been referred for follow-up and treatment. On balance, they would be easier to engage and treat with greater benefits from treatment. Treated clients were statistically similar in their mental health scores to nonengaged and rerouted clients. However, treated respondents consumed far less health services ($1386~$2768 vs. $13 755 per person per annum for those nonengaged). Thus, mental health treatment appears to maintain clients’ mental health while containing their use and cost of other services. Lower expenditure for all health and social services was associated with mental health treatment. In Study III [3], the four user groups were similar in characteristics to each other at referral and follow-up on all the major variables with the exception of two factors: non-users lived further from the clinic than users and users were rated as psychologically more vulnerable than non-users. The best predictors for attending the clinic were the presence of referral information from the referring physician and the geographic location of the patient’s referring physician. The prevalence of poor psychosocial adjustment in Study III was 55.7%, high by comparison with other speciality clinics [16,24]. Seventy percent of the variance in psychosocial adjustment to chronic pain was explained by social and cognitive variables. In addition, users of speciality pain clinic services generated less total costs in the use of all health services ($3285~$3801 per person per annum) when they were compared with non-users ($6008 per person per annum). The importance of social support and meaning of illness variables in predicting psychosocial adjustment to chronic pain is corroborated by Studies III and V as is

108

G. Browne

et al. /Health

Policy

34 (1995)

95-112

the tilevance of the cognitive behavioural approach for these problems. In addition, compared with other chronic pain sufferers with similar characteristics and levels of adjustment to illness who self-select to other community services, it appears that the use of the pain clinic helps to curtail the use of other services and thus has an important economic impact. In Study IV [ 131, clients enrolled in the CHSC approach to community care were similar in their severe level of physical impairment and moderate levels of social/ emotional impairment compared with eligible clients not yet enrolled. This supports the need for such programs by both those enrolled and those waiting. Only the clients discharged from the centre exhibited improved Activities of Daily Living (ADL) function. Clients discharged and enrolled in the CHSC approach to community care, consumed fewer health services ($1792-$2219 per person per annum) when compared with those not enrolled and receiving homecare ($16 607 per person per annum). No significant differences apart from mix and intensity of resources distinguished the MOH from MCSS approaches to CHSC care. Approximately 109 consenting curegivers of CHSC clients receiving one of four different types of community service are remarkably similar in their ‘good’ physical health yet report only ‘fair’ social and emotional health. On the whole, caregivers reported low use of health services and moderate levels of purpose, social support and realistic burdens imposed by caregiving. Perceived burden by caregivers was lowest in the discharged group (N = 10) and equivalent for caregivers of relatives enrolled (N = 68) or not enrolled (N = 23) and highest for caregivers of those in the institutional group. Perceived burden was related to caregiver use of health services. With respect to clients and curegivers, the CHSC program represents a ‘win/win situation’ where an equivalent to higher average of client functional (ADL) capacity is achieved by a lower average use and expenditure for services for both clients and their caregivers. While clients in the CHSC enrolment group were equivalent to those ‘not enrolled’ in their poor health and poor social, emotional and physical function they consumed one-half the amount of health care resources. Across the approaches to community health service delivery, caregivers were remarkably effective and equivalent on all the measured variables, in maintaining their health and outlook in spite of the burden of caregiving. Nevertheless, perceived caregiver burden was positively associated with caregiver use of health services. In Study V (161, the prevalence of poor adjustment to illness was 34% and was the variable most associated with per person expenditure for health services. The proactive problem solving intervention and phone call support for the poorly adjusted chronically ill clients were associated with improvements in purpose-in-life, psychosocial adjustment (particularly less physical distress) and lower rates of health service utilization and lower expenditure. This represents administratively and economically important reductions in expenditure for health and social services associated with the more complete approach to ambulatory care and are commensurate with improvements in a person’s adjustment to illness. Detailed secondary analysis indicated that the effectiveness and efficiency of three types of services differed as a function of a person’s characteristics. Poorly adjusted outpatients with certain characteristics benefited from problem solving counselling while others

G. Browne

et al. /Health

Policy

34 (1995)

95-112

109

benefited from phone call support. For others this additional care was not necessary to either improve the outcome or lower the expenditure. This information has been detailed more fully elsewhere [ 161. 4.2. Multistudy

comparisons

There is much to say about each study alone and each has appeared as a separate paper [3,4,10,13,16]. We could not do a meta analyses. When meta analysis can handle multi-dimensional outcomes, this may be a useful technique. Unlike clinical trials, our five studies examine both costs and consequences of different approaches to treating chronic illness in the community. In our studies the consequence of each chronically ill client/caregiver circumstance and hence, their well-being outcomes, differ across studies. This report compares expenditure for health service utilization across studies. While absolute expenditure varies with the chronically ill client’s circumstance, the same trend of equal or better outcomes when measuring expenditure for utilization is noted in the five studies. Each study had different samples and sample sizes, which were obtained from different regions, with similar subjects receiving different types of comprehensive community services. Across studies clients were assessed with different outcome measures appropriate to their circumstance, within research designs of increasing methodological rigor. All of the studies employed the same approach to the quantification of expenditure for using health and social services (Figs. 1 and 2).

Studies I-IV used historic cohort analytic designs to illustrate that the expenditure associated with only in-home or non-clinic approaches to the care of some chronic conditions is higher than the seemingly more expensive speciality clinic or in-home plus day program approach. Expenditure associated with waiting for comprehensive, co-ordinated community programs could be spent expanding the spaces in these programs. In addition, more careful study of why people decline these services is needed so that cost-effective programs can be tailored to people’s whole circumstance. Study V employed the randomized controlled trial design which controlled for self-selection, consumer preferences and differences in consumer circumstances all of which may have affected the well-being and expenditure outcomes in Studies I-IV. Study V illustrates that chronically ill outpatients from one of 27 medical clinics, who were comparable in their willingness to receive one of three approaches to clinic care and in their poor adjustment to illness, improved with either the additional psychosocial services added to conventional medical care. In addition, recipients of these more complete services used fewer other health and social services. In summary, clients and/or their caregivers enrolled in a more comprehensive, ambulatory community service exhibited, in comparison with other approaches to ambulatory/community care, equivalent or superior health status or well-being (Fig. 1), while consuming fewer per person annual dollars of health and social services (Fig. 2). While differences in the magnitude of effects appear to favour comprehensive and proactive home/community service models, it is not known whether these differences are significant in each study.

110

G. Browne

et al. /Health

Policy

34 (1995)

95-112

These five studies illustrate a number of important factors when considering the value of community-based programs for chronically ill clients in health care systems. First, one should consider the nature of community programs being evaluated. Are they holistic, proactive and preventative? Are they comprehensive and co-ordinated in design or are they singular, isolated, fragmented and reactive? Secondly, it is important to take a comprehensive approach to examining expenditure associated with different approaches to community care from multiple points of view. Publicly funded or centralized systems of health care require more complete information about total direct and indirect costs to society from all points of view. The comprehensive vantage point is often ignored in financial analysis of community health care services. While the five studies were conducted in Southern Ontario, they reflect similar community service options and client circumstances often encountered in other community-based tertiary centres. More research is needed before definitive conclusions can be made about what is the most economical and efficient approach to community health care for chronically ill people. Randomized trial evidence is needed but we also encourage more prospective studies as well. In the current context of constraint, randomized trials of downsizing health services easily could be conducted. Such trials should be designed to analyze client characteristics such as coping skills and perceptions about purpose-in-life as they interact with the various approaches to treatment. This is done in order to explain outcomes. This type of information can be used to justify priorities for services given to certain target groups. The use of prospective studies would allow for an estimation of change in client status over time. Any biases which may be operative when clients self-select to one service strategy over another could be assessed in part by the use of such studies. While problems of recall and/or client self-selection potentially threaten the validity of cohort analytic study findings, they are cheaper than randomized trials and therefore increase in importance as research resources decline. The main lesson we have learned from examining five studies on approaches to community care is that it is more effective and less expensive to offer complete proactive health care services to chronically ill people in the early stages of their illness than to provide services on demand in a piecemeal manner. It appears that people will attempt to get what they need to maintain their well-being. What type of service and when they get the service can have a positive or detrimental impact on expenditure. The assumption that more comprehensive, co-ordinated services creates dependency in the form of increased utilization is refuted by the data from these studies. We know from prior research [26,27] that home/community-based demonstration programs may not always reduce total program costs because decreases in the use of other health care services may not be large enough to offset the costs associated with the provision of a more comprehensive array of home/community services. Expenditure data in these five studies include home/community-based program costs. Even when these are included in the total expenditure calculations, total

G. Browne et al. /Health Policy 34 (1995) 95-112

111

expenditure was lower when the fuller array of community services was provided proactively. When there are conflicts between individual and societal perspectives (i.e. perceived cost versus actual client need), policy-makers have to find ways of compensating or cross-subsidizing community agencies which bear the brunt of expense in order to achieve savings for the whole health care system. There is a need to subsidize community agencies to maintain the more holistic, less expensive approach to care. Decision-makers must decide how to bring about these initiatives without compromising the gains associated with more effective, economical care. Acknowledgements Funded by Ontario Ministry

of Health.

References VI Hedrick, SC., Rothman, M.L., Chapko, M., Inui, T.S., Kelly, J.R. and Ehreth, J., Adult day health

care evaluation study: methodology and implementation, Health Services Research, 5(6) (1991) 935460. 121 Browne, G., Arpin, K., Corey, P., Fitch, M. and Gafni, A., Individual correlates of health service utilization and the cost of poor adjustment to chronic illness, Medical Care, 8(l) (1990) 43-58. 131 Weir, R., Browne, G., Tunks, E., Gafni, A. and Roberts, J., A profile of users of speciality pain clinic services: predictors of use and cost estimates, Journal of Clinical Epidemiology, 45(l) (199) 1399-1415. 141 Milne, C., Sacco, C., Cetinski, G., Browne, G. and Roberts, J., Correlates of Well-Being among Caregivers of Cognitively Impaired Relatives. Canadian Journal of Nursing Research, 26(l) (1994) 27-39. PI Crumbaugh, J.C., Cross-validation of purpose-in-life test, Journal of Individual Psychology, 4 (1968) 74-81. 161 Crumbaugh, J.C. and Maholick, L.T., Manual of Instructions for the Purpose-In-Life Test, V. Frank1 Institute of Logotherapy, P.O. Box 85, Saratoga, CA, 95070, 1969, 1981. 171 Weinert, C., A social support measure: PRQs 85, Nursing Research, 35(5) (1987) 73-77. PI Zarit, S.H., Reeves, K.E. and Bach-Peterson, J., Relatives of the impaired elderly: correlates of feelings of burden, Gerontologist, O(6) (1980) 649-655. 191Zarit, S.H., Todd, P.A. and Jarit, J.M., Subjective burden of husbands and wives as caregivers: a longitudinal study, Gerontologist, 6(3) (1986) 60-66. [lOI Emond, A. and Browne, G., Prognosis and Cases of Clients Referred to a Community Mental Health Clinic, Working Paper Series 92-11, McMaster University, System-Linked Research Unit on Health and Social Service Utilization, 1992. 1111Goldberg, D., The Detection of Psychiatric Illness by Questionnaire, Oxford Press, London, 197. U21 Derogatis, L.R. and Lopez, MC., PAIS and PAIS-SR Scoring Manual I, John Hopkins University, School of Medicine, Baltimore, 1983. 1131 Browne, G., Roberts, J., Watt, S., Gafni, A., Stockwell, M. and Alcork, S., Community Rehabilitaiton: Strategies, Outcomes, Expenditures, Canadian Journal of Rehabilitation 8( 1) (1994) 9-22. [I41 Fillenbaum, G.G., Validity and Reliability of the MFAQ (OARS) in Multidimensional Functional Assessment: The OARS Methodology. 2nd Ed., Chapter 3, Duke University, Durham, North Carolina; also in Multidimensional Function Assessment of Older Adults, Lawrence Erebaym Associations, New Jersey, 1988. VA Browne, G., Jam&on, E., Hay, I. and Roberts, J., Modifications and Scoring of the OARS MFAQ Measure of Functional Capacity: 1991 Canadian Situation, Working Paper 92-8, Funded by

112

1161 1171 [18] [ 191 [20] [21] (221 [23] [24] [25] (261 (271 [28]

G. Browne

et al. /Health

Policy

34 (1995)

95-112

NHRDP Grant No. 6606-445457C and The Ontario Ministry of Healths System-Linked Research Unit on Health and Social Service Utilization, October 1992. Roberts, J., Browne, G., Streiner, D., Gafni, A. and Pall&r, R., Hoxby, H., Jam&on, E., Eichenbaum, D., The effectiveness and efficiency of health promotion in specialty clinic care, Medical care, 33(9) (1995) 1-14. Warner, S. and Williams, J.I., The meaning of life scale determining the reliability and validity of a measure, Journal of Chronic Disease, 40(6) (1987) 503-551. Broadhead, W., Gehlbach, S.H., DeGruy, F.V. and Kaplan, B.H., The Duke-UNC Functional Social Support Questionnaire, Medical Care, 6(7) (1988) 709-771. Butt, G., A Quasi-Randomized Controlled Study of the Effects of an Educational Intervention on Caregiver Burden for Primary Caregivers of Individuals with Alzheimers Disease. Unpublished MHSc Project, McMaster University, Health Care Practice Program, 1989. Goldberg, D. and Hillier, V., A scaled version of the General Health Questionnaire, Psychological Medicine, 5 (1979) 139-145. McDowell, I. and Newell, C., Measuring Health: A Guide to Rating Scales and Questionnaires, Oxford University Press, Oxford, 1987. Weisman, M., The assessment of social adjustment: an update, Archives of General Psychiatry, 38 (1981) 150. Morrow, G.P., Chiarello, R.J. and Derogatis, L., A new scale of assessing patients psychosocial adjustment to mental illness, Psychological Medicine, 9 (1978) 605-610. Arpin, K., Fitch, M., Browne, G. and Corey, P., Efticacy of Family Clinic Appointments in Promoting Adjustment to Chronic Illness, Ministry of Health Project No. 0174, Final Report. Faculty of Nursing, University of Toronto, August 1988. Spitzer, W., Roberts, R. and Demore, T., Nurse practitioner in primary care V: development of the utilization and financial index methodology to measure the effects of their deployment, Canadian Medical Association Journal, 76 (1976) 1099. Kemper, P., The evaluation of the national long term care demonstration: overview of the findings. Health Service Research,‘3(1) (1988) 161-174. Weissert, W.G. and Cready, C.M., Determinants of hospital-to-home placement delays: a pilot study. Health Service Research, 3(5) (1988) 619-647. Browne, G., Gafni, A., Roberts, J. and Hoxby, H., Approach to the Measurement of Costs (Expenditures) When Evaluating the Efficiency of Health and Social Programs, Working Paper Series 9212, McMaster University, System-Linked Research Unit on Health and Social Service Utilization, 1992.