SF36

ANALYSIS WORKGROUP
Our mission is to foster the improvements in the analysis of health related phenomena among older minority; encourage the development of methods and measures that better capture the health and determinants of health of diverse elders; promote collaboration between RCMAR sites on analysis issues; and disseminate new knowledge in this area.

SF36


SF-36 in Older Minority Populations

Prepared by the Center for Health Improvement for Minority Elders (CHIME) UCLA

This is a selective bibliography of recent publications on the use of the SF-36 in older and minority populations. The reviewed articles provide examples of a range of applications of the SF-36 in older populations. Some of the papers focus on evaluations (e.g., of measurement equivalence) in specific disadvantaged subgroups of the older population, but a large part of the lessons learned about the limitations of the use of the SF-36 are general issues for older populations.


Confirmatory factor analyses of the SF-36 were conducted on data collected from 3,488 Phase II American Indians participating in the Strong Heart Study (ages 48-81) and compared with an age- and gender-matched sample of 695 people from the National Survey of Functional Health Status. A multiple group model that constrained the factor loadings to be equal in the American Indian and general population samples was found to fit the data significantly less well than a model that allowed the loadings to be freely estimated (? 2 = 107.4, 9 d.f.s, difference in CFI = 0.009). However, the pattern of loadings) looked generally similar (even though significantly different).
The data reported in this paper do not provide new information to support the main conclusion that “use of summary scores assuming a differentiated physical/mental functioning structure is likely improper in at least some populations” (p. 208, abstract). The estimated correlation between the physical and mental health factors in both the American Indian and general population sample from which SF-36 norms were derived was 0.61. This is consistent with prior evidence indicating that physical and mental health scales are positively correlated (e.g., Hays, R. D., & Stewart, A. L., The structure of self-reported health in chronic disease patients. Psychological Assessment, 2, 22-30, 1990; Taft, C., Karlsson, J., & Sullivan, M., Do SF-36 summary component scores accurately summarize subscale scores? Quality of Life Research, 10, 395-404).


The SF-36 was administered in Spanish to 65 Mexican Americans (78% female, average age of 75) by telephone. Internal consistency reliability was 0.80 or above for all scales except social functioning (0.69). Item discrimination across scales (items correlating more highly with other scales than with scales they were supposed to represent) was problematic for some of the items. Full of pep, lot of energy, and tired correlated as high or higher with the general health scale as they did with the vitality scale. Similarly, in general how would you rate your health and my health is excellent correlated as highly with the vitality scale as with the general health perceptions scale. Finally, one item in the social functioning scale (frequency of interference) correlated as high or higher with the general health, vitality and general mental health scales than with the social functioning scale. Bennett and Riegal (2003) cited another study of 100 Hispanic men (Krongrad et al., 1997) reported poor performance for the SF-36 social functioning items.


The SF-36 physical functioning scale was administered to 58 adults 65 to 94 years old. The internal consistency reliability of the scale was 0.82. The physical functioning scale correlated significantly with single limb stance time (r = 0.42), gait speed (r = 0.75) and times up and go test (r = -0.70).


The sample consisted of 3509 individuals from the Blue Moutains Eye Study, a population-based cohort consistently of older urban dwellers in Sydney, Australia. Ninety percent of the sample attempted to complete the SF-36 and 78% of those completed all items (91% completed at least half the items in every SF-36 scale). The odds of completing all items of the SF-36 and half or more of the items in each scale were lower with increasing age (Ors of 0.62 and 0.64, respectively).


Fifty-six people 65-89 years old (78% female, average age = 77) who were new referrals to community based occupational therapy or physiotherapy services in three areas in north west England were interviewed face-to-face before treatment and 6 months later. Participants reported an average of 6 health problems. Comments by respondents revealed problems with the wording of some of the SF-36 items. For example, the “bending, kneeling, or stooping” and “bathing or dressing yourself” items were confusing for respondents who had limitations in some but not all the activities included in the item. Thus, they were not sure whether to focus on one or average across the complete set of activities. Another example of confusion among respondents concerned the questions about walking (walking more than a mile…). Several respondents asked the intended context—was it “hill waking, walking in their street, walking round shops”?
Another kind of problem was related to the absence of don’t know or do not do response choices for the physical functioning scale. The paper includes an example where the interviewer asks a respondent:“How are you for walking more than a mile?” The response is:“Well, I haven’t tried it really. I’m lazy. I get in the car.” Another respondent answers the question about carrying groceries:“Oh well. I haven’t carried a lot of groceries. My husband always comes with me for groceries so he carries them.”


The same 56 people who participated in the study noted above (Millinson, 2002) were mailed a questionnaire that included the SF-36. Forty-five of the 56 (80%) returned the questionnaire, but only 34 (61%) completed all the items in the SF-36. Missing data tended to be highest for the role limitations items. Millinson cites previous suggestions that this could be due to a difference between the focus on problems (have you had any of the following problems with your work or other regular daily activities) in the lead-in to the items and the wording of the items (e.g., “I didn’t do work as carefully as usual”). In addition, she notes that others have commented on the potential difficulty because of the reference to “work” for people who are retired.


The SF-36((R)) Health Survey (Version 2; SF-36) was evaluated among older Vietnamese Americans to determine whether underlying dimensions of physical and mental health were similar to those of other groups in the United States. The study provided support for the reliability and validity of the SF-36. Structural equation modeling provided confirmation of physical and mental health factors. However, the factor loadings for the SF-36 scales were more consistent with previous results from Asian countries than the typical pattern observed in the United States.


The SF-36 was administered to 245 individuals (150 females) 65 years and older (median age = 78) sampled from inpatient, outpatient and community sources. Only 26% of the sample self-completed the SF-36 and these people took a median of 16.5 minutes to complete it. Inability to complete the SF-36 was associated with global functional impairment, cognitive impairment, and manual dexterity impairment. Longer time to complete the SF-36 was related to cognitive impairment, older age, and visuospatial dysfunction.


Internal consistency reliability estimates were 0.76 for the 8 SF-36 scale scores in a sample of 621 older Mexican Americans (subsample from the Hispanic Established Populations for Epidemiologic Study of the Elderly). Confirmatory factor analyses of SF-36 items in a sample of 621 older Mexican Americans provided support for 8 first-order factors consistent with the 8 scales and two second-order physical and mental health factors (comparative fit index = 0.97, root mean square error of approximation = 0.08). The second-order factor loadings for physical health were 0.81, 0.81, 0.86, and 0.83 for physical functioning, role-physical, bodily pain and general health perceptions, and 0.81, 0.56, 0.86, and 0.89 for mental health, role-emotional, vitality, and social functioning. Peek et al. (2004) concluded that the SF-36 was a valid measure of health-related quality of life for their sample of older Mexican Americans.


Last updated July 2010