Dodge HH, Belle SH, Morycz RK, Rodriguez EG, Lytle ME, Ganguli M. Functional and demographic predictors of health and human services utilization: a community-based study.  Journal of the American Geriatrics Society, 47: 1271-1276, 1999.

LETTER:

To the Editor: With anticipated increases in both the number and the proportion of older adults in the population, clinicians and policymakers need to be able to predict better the likely concomitant increase in need for services. Current risk adjustment strategies generally explain little of the variance in utilization.^1-3  The Balanced Budget Act of 1997⁴ calls for the Health Care Financing Administration to introduce risk adjustment into capitated Medicare payments by January 2000. Many studies using summary (global or total) or grouped measures of Activities of Daily Living (ADL) and Instrumental ADL (IADL) have found functional impairment to be a strong indicator for health service use. Less is known about the relationships between specific IADLs and use of specific health services.

We studied 766 noninstitutionalized adults aged 65 and older who completed two consecutive interviews approximately 2 and 4 years after being selected randomly from the population for a longitudinal epidemiological survey, the MoVIES Project, in the mid-Monongahela Valley, a mostly rural, low socioeconomic factors community in Southwestern Pennsylvania. We examined whether disabilities in individual IADLs predicted the use of specific health services over a subsequent 2½ year period within the conceptual framework of the behavioral model proposed by Andersen.^5,6 This model includes users' predisposing, enabling, and need characteristics, the latter being the proximate cause of service use and encompassing illness, self-rated health, and disability.

We examined the following predisposing characteristics, i.e., variables potentially reflecting individuals' underlying propensity to use health and human services: age, sex, education (high school and additional schooling vs less than high school education), and living arrangements (alone vs with one or more others). Enabling characteristics are factors that allow individuals to use health services and, theoretically, include two dimensions: individual resources and community resources. We included Medicaid eligibility with individual financial resources. We included no variables with community resources (usually measures such as physician-to-population ratio, hospital-to-population ratio, and geographic location) because our study was conducted in a stable population in two adjacent counties similar in population density and available health and human services. Need characteristics are factors broadly encompassing illness and disability and representing the proximate cause of service utilization. We included self-rated health (poor vs fair, good, or excellent), general mental status (MMSE categorized as <23 vs >=24), and IADL from the modified OARS⁷: getting to places out of walking distance; going shopping for groceries or clothes (assuming transportation is available); preparing meals; doing own housework; taking own medicine; and handling own money. Responses were categorized as "independently able" versus "needs some help or is totally unable".

We examined four categories of health and human services (outcome variables): hospitalization, emergency room visits, home healthcare use, and social services. Home healthcare included in-home services provided by visiting nurses, physical therapists, speech therapists, occupational therapists, and home health aides. Social services included a range of services (Meals on Wheels, homemaker/chore service, social worker/case worker, personal care worker, attendant care, congregate meals) provided by formal as well as informal/volunteer agencies. Table 1 describes our sample with respect to predisposing, enabling, and need factors and health and human services utilization.

Table 1. Distribution of Variables Examined (n = 766)

Logistic regression models were used (see Table 2). All IADL items were associated significantly with one another. Therefore, we fit separate models for total IADL score and for each individual IADL item to predict subsequent use of each service. After adjusting for age, gender, education, living arrangement, Medicaid coverage, self-rated health, and cognitive impairment, we found that total IADL disability scores were significant predictors of subsequent hospitalization, home health care, and social services utilization. Individual IADL disabilities also predicted use of specific health services. Inability to do housework independently predicted hospitalization (odds ratio = 2.3) and use of home health care (odds ratio = 2.6) and social services (odds ratio = 6.2). Inability to shop independently predicted use of home health care (odds ratio = 2.3) and social services (odds ratio = 4.2). Inability to go independently to places out of walking distance predicted home healthcare use (odds ratio = 3.7) and social services use (odds ratio = 3.2). Inability to prepare meals independently predicted social services use (odds ratio = 4.7).

Table 2. The Results of Logistic Regressions: IADL Limitations and Other Factors Predicting Health Services Utilization (N = 766)

Individual IADLs predict the use of home health and social services, but not the use of acute care services, after adjusting for other predisposing, enabling, and need characteristics. Identifying global and specific IADL disabilities can help predict future utilization of different health and human services. Measurement of functional ability may help to target interventions more precisely to those with specific needs and disabilities.

Hiroko H. Dodge, PhD

Steven H. Belle, PhD

Richard K. Morycz, PhD, ACSW

Eric Rodriguez, MD

Mary Lytle, MSW, LSW, ACSW

Mary Ganguli, MD, MPH

University of Pittsburgh; Pittsburgh, PA

REFERENCES: