Recruitment and Procedure
This 12-month prospective study was conducted between January 2001 and April 2004 in the San Francisco Bay Area. The study compared a convenience sample of newly enrolled participants from 16 ADHC programs in six counties of the San Francisco Bay Area with community-dwelling older adults from the same geographical area, who did not attend an ADHC but would have met eligibility criteria for ADHC attendance if referred.
Comparison group recruitment sites included a nonprofit provider of health and social services for seniors, an in-home social services provider, subsidized senior housing sites, and primary care physicians. They were chosen on the basis of their likelihood of having clients with similar socioeconomic status, age, level of functioning, health status, and health service utilization as ADHC clients. Potential comparison subjects were informed about the study by their social workers or primary care providers, who then provided research study staff with a list of interested candidates. Other comparison subjects self-referred after learning about the study through fliers and letters. A research staff member conducted a telephone eligibility screen with interested study candidates to confirm eligibility and schedule the first home visit for informed consent procedure and baseline interviews.
ADHC centers (n = 16) who were members of the San Francisco Adult Day Services Network were invited to refer eligible participants. All invited ADHCs offered the same daily services provided by a multidisciplinary team of professionals, including nursing services, therapeutic activities (targeted to social, physical, and cognitive functioning), occupational therapy, speech therapy, dietician counseling, one full meal, personal care services, social services, and transportation to and from the center. ADHC staff informed new ADHC enrollees about the study within the first 2 weeks of enrollment. Participants who were interested were then contacted by study staff for a telephone eligibility interview. All participants who met study inclusion criteria were included in the study
The ADHCs included in this study met the Title 22 ADHC licensing requirements, which stipulate that participants must have physician-prescribed treatment or rehabilitation needs, a mental or physical disability that affects daily living but does not require 24-hr care, and a high potential for further deterioration and probable institutionalization if ADHC services were not available. For those who qualify for Medicaid (California’s version is Medi-Cal) and qualify for ADHC services, Medicaid paid for the services (U.S. Department of Health and Human Services, 2005
Data were collected at three measurement points: at baseline, 6 months, and 1 year after enrollment in the study. ADHC and comparison group subjects received $25 in cash after completion of each of the three interviews. The University of California, San Francisco granted human subjects approval.
Inclusion and Exclusion Criteria
Inclusion criteria for both groups were: age 55 or older; fluency in English, Cantonese, or Mandarin; insurance through Medicare, Medi-Cal, a Health Maintenance Organization, or private health insurance; dependency in at least one category of the Physical Self-Maintenance Scale (PSMS); ability to sit up for six hours at a time, and answer basic questions about his/her age; living situation; and functioning. Exclusion criteria were dependency in all seven PSMS categories; severe cognitive impairment defined as a score of 6 or more errors on the Short Portable Mental Status Questionnaire; a history of chronic schizophrenia; residence in a Board and Care home, a Residential Care Facility, an Assisted Living Facility, a nursing home, or homeless; enrollment in Adult Day Health Care, Adult Day Care, or Adult Day Support within the previous 6 months; and neurological or sensory deficits severe enough to interfere with the completion of study instruments.
ADHC subjects were newly enrolled in an ADHC program for 2 or more days per week. ADHC participants were excluded from participation if they had already attended the center for eight or more regular visits before study enrollment or if they attended the center for fewer than 2 days/week. Comparison subjects lived in the community and did not attend a day center or an ADHC within the prior 6 months.
Between January 2001 and March 2003, 127 ADHC clients and 143 community-living older adults who did not attend ADHC were referred to the study. Of the 127 clients referred to the ADHC group, 34 did not meet the screening criteria, 13 refused, and for 5 clients the first interview could not be scheduled before their eighth day of ADHC attendance. Of the 143 clients referred to the comparison group, 49 did not meet the screening criteria, 13 refused, and for 2 the first interview could not be scheduled. Seventy-five were enrolled in the ADHC group and 79 in the comparison group. Among the 75 ADHC participants, 7 refused follow-up interviews, 6 were lost to follow-up, 5 died, and 57 completed 12-month interviews. Among the 79 comparison group participants, 4 refused follow-up interviews, 4 were lost to follow-up, 4 died, and 67 completed 12-month interviews ().
For ADHC participants, the first interview was conducted before their eighth day of ADHC attendance (typically within 3 weeks of enrollment in ADHC). Comparison subjects were not enrolled in a social program and thus were interviewed within 3 weeks of being identified as eligible and willing to participate in the study by referring centers and social workers. Interviews were conducted in participants’ homes by trained and experienced interviewers. Interviewers read the survey questions to the participants in participants’ primary language and participants responded either verbally or by pointing to a response choice on a card printed in a large font. In the vast majority of cases, the ethnicity of the interviewers matched the ethnicity of the subject. In all cases, the interviewer spoke the primary language of the subjects (English, Cantonese, or Mandarin).
Our primary outcomes measure was quality of life. In the context of long-term care services, quality of life has been conceptualized as a key indicator to assess if personal capabilities are aligned with environmental demands (Lawton, 1983
). As indicators of participants’ physical, cognitive, and emotional capabilities, we assessed subjects’ comorbidity, cognitive status, mood, and ability to perform activity of daily living. We assessed participants’ demographics (marital status, household income, and living alone) as indicators of social and financial resources. Finally, we collected data on service utilization (including hours of home care) to assess subjects’ in-home environmental support.
Data for these outcomes were collected with the following instruments:
, the primary outcome, was assessed using the Medical Outcomes Survey Form (SF 36), an internationally used multidimensional instrument of quality of life (Ware, Snow, Kosinski, & Gandenk, 1993
). We assessed the following domains physical functioning (extent to which heath-related problems limit moderate or vigorous physical activities), role physical (extent to which physical health problems affect functioning in daily activities), social functioning (extent to which physical and emotional health problems interfere with social activities), role emotional (extent to which emotional problems limit the type and amount of regular daily activities), and mental health (lengths of time experiencing various mood states). Each of these domains is scored on a 0- to 100-point scale with higher scores reflecting better functional-related quality of life. Reliability coefficient for social functioning is .76 and for role physical and role emotional equaled or exceeded .80 (Ware et al., 1993
were assessed using the PSMS, which asks if the respondent needs no help, some help, or is completely dependent on others to perform ADL (e.g., going to the toilet, bathing, grooming, eating, transferring, or dressing). Higher scores reflect greater capacity to perform these tasks independently. Lawton and Brody (1969)
determined the Guttman reproducibility coefficient for this scale to be .96.
was assessed using the Mini-Mental Status Examination (MMSE), a 30-point screening instrument for cognitive impairment with a test–retest reliability of 0.89. Possible scores range from 0 to 30, with higher scores indicating higher cognitive functioning (Folstein, Folstein, & McHugh, 1975
; Tombaugh & McIntyre, 1992
was assessed using the 15-item self-reported Geriatric Depression Scale (GDS) with a reliability coefficient of .75 in functionally impaired community-dwelling older adults. A score of 6–11 points indicate mild to moderate depressed affect, and a score of 11–15 severe depressed affect. (Friedman, Heisel, & Delavan, 2005
; Yesavage et al., 1982
Charlson Comorbidity Index
(CCI), which is based on weighted numbers and seriousness of comorbid conditions, was also determined. Developed with a cohort of 559 patients, the 1-year mortality rates for the different scores were “0,” 12%; “1–2,” 26%; “3–4,” 52%; and “greater than or equal to 5,” 85% (Charlson, Pompei, Ales, & MacKenzie, 1987
To compare ADHC and comparison group subjects in their baseline characteristics, chi-square tests were performed for categorical variables and t tests were computed to compare groups on continuous-valued variables. To assess whether temporal trends in quality of life domains significantly differed between ADHC and comparison group subjects, repeated measures analysis of covariance (ANCOVA) models were computed for each quality of life domain. The repeated dependent variables were the quality of life scores used for each occasion (baseline, 6 months, and 1 year). The time variable indicated the assessment occasion. The independent variable reflected whether or not the subject participated in ADHC, and the time by group interaction assessed whether the trajectories for quality of life differed for the two groups over the three assessments. The reported quality of life scores were adjusted for baseline values of variables that differed at baseline between the two groups with a p value of .10 or less, including whether or not the subject lived alone, number of comorbidities as assessed with the CCI, level of daily functioning as assessed with the PSMS, depressive symptoms as assessed by the GDS, and cognitive functioning as assessed by the MMSE. We conducted Mauchly’s test of sphericity to assess whether the error covariance matrix has variances and covariances close to zero, an assumption of the repeated measures ANCOVA model. The fit of the model was determined by computing an F test that compares the ratio of the mean square for lack of fit with the mean square for pure error. The regression coefficients used to estimate the predicted score for each occasion are presented in to describe the contribution of the covariates to the adjusted scores. For each outcome, we computed an F test to assess the occasion by group interaction. A significant F test would indicate that the groups had differential patterns of change in adjusted quality of life scores over occasions. The covariate-adjusted quality of life scores and the p values for the time by group interaction are presented in . For those domains of quality of life that showed differential trends over time, we conducted secondary analyses to determine whether those changes could be explained by changes in level of daily functioning (PSMS), depressive symptoms (GDS), and cognitive functioning (MMSE). To accomplish this, the longitudinal generalized linear models were rerun using time-varying values rather than only baseline values of the PSMS, GDS, and MMSE.
Regression Coefficients Across Occasions for Quality of Life Outcomes