Medical Outcomes Study 36-Item Short-Form Health Survey (SF-36)

The SF-36 is a generic, multipurpose health status survey to be used with persons aged 14 and older. Currently, there are two versions available for use. The SF-36 contains 36 items and yields a physical and mental health summary score, as well as eight individual scales: physical functioning (PF), role-physical (RP), bodily pain (BP), general health (GH), vitality (VT), social functioning (SF), role-emotional (RE), and mental health (MH). The SF-36 has been widely reviewed elsewhere; therefore, the focus of this review is on the application of the SF-36 in the veteran population. For more detailed information concerning the history and development, psychometric properties, and normative data of the SF-36, please refer to the SF-36 user’s manual. (Ref: 1) Information regarding the two summary measures and applicable normative data is presented in a second manual. (Ref: 2) Additional publications regarding the SF-36 are available at http://www.sf-36.com. The SF-36 has also been modified for use among veterans (see the SF-36V Instrument Review.)

Anyone wishing to reproduce the survey must register or obtain a license; see URL below

5-10 minutes. (Ref: 5)

1993 (version 1), 1996 (version 2).

The SF-36 appears to use simple, clear language with simple grammatical structure. The developers suggest that administrators determine the reading ability of a person before giving them the self-administered version. (Ref: 1)

Response options and scale scoring vary by item. (Ref: 1)

Self-administered, questionnaire, or computer-administered. (Ref: 5)

Items are summed and rescaled to a range of 0 to 100; lower scores denote poorer health. (Ref: 1) One item (self-reported health transition) is not scored. The Physical Health and Mental Health summary scales are scored with weights, and are expressed as t-scores with mean=50 and standard deviation=10. (Ref: 2)

The SF-36 has been translated into multiple languages by the International Quality of Life Assessment (IQOLA) project. (Ref: 5)

John E. Ware, Jr., PhD; Mark Kosinski, MA; Barbara Gandek, MS

The Health Institute; New England Medical Center; Boston, MA.

www.sf-36.com

top

Factor analysis produced a two-factor orthogonal solution that accounted for 80-85% of variance in the subscales. All items have correlated at or above 0.40 with their respective scales, except for rare occasions. (Ref: 1) Information on factor analysis performed with veteran samples was not found.

Norms are provided for the general U.S. population in 1998, by age, gender, and chronic condition. (Ref: 2) Means on the physical functioning (PF) and role physical (RP) scales have been provided for veterans with chronic lung disease, chronic low back pain, and osteoarthritis of the knee. (Ref: 3)

An article utilizing Veterans Health Study data found that veterans’ mean scores on each of the eight scales were at least 50% of one standard deviation worse compared to patients of non-VA health systems. They also found that seven of the eight scales showed worse health (lower scores) for the younger veterans, compared to the older cohort (p<0.001). Physical function (PF) scores were lowest in the older groups. The physical health summary score showed slightly worse health in the older cohort (p<0.05), while the mental health summary score showed significantly worse health in the younger cohort (p<0.001). The physical health summary score in this sample (36.9) is greater than 140% of one standard deviation worse than the score in the general population (50). (Ref: 4)

Patients enrolled in VA general medicine clinics were surveyed with the SF-36 as part of the Ambulatory Care Quality Improvement Project (ACQUIP). Results yielded a mean physical health component score of 34.1 and a mean mental health component score of 45.4. Self-reported health status was significantly lower than U.S. population means. (Ref: 7)

Deployment to the Persian Gulf, compared to those not deployed, was associated with a 1.7 point decrease in mean physical component scores and 2.0 point decrease in mean mental component scores. These results indicated that deployed veterans generally had health profiles slightly lower than U.S. Norms, however, the mental health component score was slightly above U.S. Norms. (Ref: 8)

Test-retest across scales ranged from 0.43 to 0.90, as determined by studies utilizing varied (nonveteran) samples. (Ref: 1) Test-retest, as assessed with Gulf War veterans, showed that kappa ranged from 0.39 to 0.79, with 90 to 97 percent agreement. (Ref: 8)

Non-applicable.

Most studies providing reliability statistics have shown internal consistency for the scales to be at least 0.70 in various populations. (Ref: 1) Using data from the Veterans Health Study, researchers found that Cronbach’s alpha was 0.92 on the physical functioning (PF) scale and 0.88 to 0.91 on the role physical (RP) scale. (Ref: 3) A study of Gulf War era veterans found alpha ranged from 0.81 to 0.90 for the subscales. (Ref: 8)

Alternate forms (other equivalent forms of the test, e.g. odd vs. even items) reliability data exists (see McHorney CA, 1995, Med Care 33(1):15-28), but no information was found regarding alternate forms reliability in veteran samples.

For detailed convergent and discriminant validity information utilizing multiple samples, refer to the user’s manual. (Ref: 1) It is reported that mean subscale and summary scores varied as hypothesized among veterans, supporting the construct validity of the SF-36. (Ref: 8)

For detailed criterion-related and predictive validity information utilizing multiple samples, refer to the user’s manual. (Ref: 1) Veterans who reported current physical or mental health disorders had significantly lower scores than those without (p<0.001). (Ref: 8)

The eight health concepts used in the SF-36 were selected from 40 concepts evaluated in the Medical Outcomes Study (MOS). These eight represent the health concepts most affected by disease and treatment, and the most frequently measured in health surveys. (Ref: 1-2)

Researchers have assessed the responsiveness of the SF-36 to the development of diabetes complications in elderly veterans, measured at two times (mean interval=3.1 years). They found that the SF-36 was responsive on six of the eight scales: GH decreased 6.1 points (effect size=0.24), PF decreased 9.7 points (ES=0.38), SF decreased 5.8 points (ES=0.19), PR decreased 14.7 points (ES=0.38), BP decreased 4.0 points (ES=0.14), and VT decreased 4.5 points (ES=0.16). An increase of greater than one diabetes complication over time was related to a decrease of 7.2 to 11.8 points on these six scales. (Ref: 6)

Yes. (Ref: 3-4,6-8)

Yes. (Ref: 1-2)