The Short-Form (SF-36) Health Survey

National Center for Interprofessional Practice and Education's picture
Submitted by National Center... on Oct 21, 2016 - 11:43am CDT

Instrument
Authors: 
Brazier, J.E.
Harper, N.M.
Jones, A.
O'Cathain, K.J.
Thomas, T.
Usherwood, T.
Westlake, L.
Overview: 

The SF-36 is a multi-purpose survey designed to capture adult patients' perceptions of their own health and well-being.  Based on a much longer survey developed in the 1980's by Ware, J.E., the SF-36 has 36 items grouped in 8 dimensions: physical functoning, physicial and emotional limitations, social functioning, bodily pain, general and mental health.  It is a generic measure, as opposed to one that targets a specific age, disease, or treatment group.  It can be administered in a range of settings (e.g., primary care, community clinics, specialty care serving patients with chronic conditions).  Over the years, the SF-36 has been used widely and proven useful in many studies of general and specific populations, comparing the relative burden of diseases, and in differentiating the health benefits produced by a wide range of different treatments.  It could also be used as an outcome measure for studies of organizational or education interventions involving interprofessional collaborative practices.  In the one of the first validity studies (Brazier, 1992, reported here), the instrument was found to have acceptable to good estimates of reliability, support for construct validity, and support for convergent and divergent validity for 4 of the 8 dimensions.

Link to Resources
Descriptive Elements
Who is Being Assessed or Evaluated?: 
Individuals
Instrument Type: 
Self-report (e.g., survey, questionnaire, self-rating)
Source of Data: 
Patients, clients
Notes for Data Sources: 

In general, the survey is designed for persons 18 years and older.  In the Brazier validation study, the survey was sent to 1980 patients, randomly selected from two general practice lists in Great Britain.  For test-re-test reliability, the survey was re-sent to a subset of 250 patients randomly selected from the same lists.  The demographics of the sample largely replicated that of Britain's general household survey.

Instrument Content: 
Patient / client health status
Notes for Content: 

Eight dimensions related to functioning and well being are measured (see below).  An additional item regarding change in health status is not scored.

Physical Functioning:

  • Physical functioning
  • Role limitations: physical
  • Role limitations: emotional
  • Social functioning

Well-Being

  • Bodily pain
  • General health
  • Vitality
  • Mental health
Instrument Length: 

36 items, about 5 minutes to complete.

Item Format: 
Combinations of 5-point scales, 3-pt scales, and dichotomous (yes/no) items.
Administration: 
The Brazier study in 1992 reported results from a mailed survey. Currently, as supported by the distributor, Optum, Inc., the survey can be administered in a variety of modalities, in-person paper and pencil, interview, online and over mobile devices.
Scoring: 
Scoring was not discussed in the Brazier validation study. As supported by Optum, the instrument's scores are norm-based: a score of 50 = average. Two summary measures are derived: a physical component summary (PCS) and a mental component summary (MCS). (See link to the instrument manual, above.)
Language: 
English; as supported by Optum, the survey has been widely translated and in available in 170 languages.
Norms: 
Not available.
Access: 
License required
Notes on Access: 

Multiple versions of the SF-36 can be found on the internet, but distribution rights appear to be held by Optum, Inc. 

Psychometric Elements: Evidence of Validity
Content: 
Survey items were taken and adapted from a much longer survey designed by the Rand Corporation in the U.S. (Ware, J.E., 1980).
Response Process: 
To test the British version of the original American survey, the authors held face-to-face interviews and made some minor adjustments to 6 items. To test the feasibility of the mailed survey booklet, they piloted the instrument with 120 people (40% response, no reminders). The response rate for the final survey (with three reminders) was 83%, and the response rate for the test-re-test sample was 75%. Very little missing data was reported.
Internal Structure: 
Item correlations with their specified dimensions were good (above 0.50 for 33 of 36 items). Cronbach's alpha was good: 0.85; reliability estimates for each of the 8 dimensions were above 0.50 for all dimensions except for social functioning, which had only 2 items in the scale. Test-re-test reliability estimates (obtained two weeks after the first survey) were in the low to moderate range (0.60 - 0.81).
Relation to Other Variables: 
To assess construct validity, the authors looked for associations between scores on the SF-36 and demographic characteristics of responding patients that are expected to affect health and well being. The claim of construct validity was largely supported; that is, for example, lower physical function and increased pain were correlated with patient age (p = <0.001); socioeconomic class correlated with 7 of the 8 dimensions (p = <05); and persons with chronic conditions perceived their health as worse on all dimensions (p<0.001) except mental health. The authors also explored convergent and discriminant validity by comparing results of the SF-36 with an established patient health survey. Correlation coefficients for the four comparable dimensions of the SF-36 questionnaire and Nottingham health profile were reported to be higher than correlations between non-comparable dimensions.
Consequential: 
No information.
9