More About the 20-Item Short Form Survey (SF-20)

The 20-Item Short Form Health Survey (SF-20) was developed for the Medical Outcomes Study (MOS), a multi-year study of patients with chronic conditions. The resulting short-form survey instrument provides a solution to the problem faced by many investigators who must restrict survey length. The instrument was designed to reduce respondent burden while achieving minimum standards of precision for purposes of group comparisons involving multiple health dimensions.

Description of the 20-Item Short-Form Survey

Physical Functioning
Six items were selected to assess physical functioning, a dimension measured in the HIE by aggregating twenty items measuring physical limitations and capacities, mobility, and self- care (Stewart, Ware, and Brook, 1978; Stewart, Ware, and Brook, 1982.). The goal was to approximate as closely as possible the 6-level scale constructed in the HIE. Response choices and item wording were modified from the HIE version to capture better specific limitations of interest, to describe more accurately the scale level defined by each item, and to facilitate oral administration. One new item (moderate level of limitation in physical functioning) was added to fill a gap in the HIE scale.
Role Functioning
Two items were selected to measure limitations in role functioning due to poor health. These are the two best items from the 3-item HIE role functioning scale (Stewart, Ware, and Brook, 1978, 1981, 1982).
Social Functioning
Social functioning is defined as the ability to develop, maintain, and nurture major social relationships. The single social functioning survey item focuses on whether the respondent's health has limited social activities.
Mental Health
General mental health was assessed using five items derived empirically from the HIE Mental Health Inventory (MHI). This set is the best 5-item predictor of the summary score based on the full 38-item MHI (Davies, Sherbourne, Peterson, and Ware, 1988). The set represents the four major mental health dimensions (anxiety, depression, loss of behavioral-emotional control, and psychological well- being) as confirmed in factor-analytic studies of the MHI (Veit and Ware, 1983). These five items correlated 0.92 with the MHI total score in the HIE sample used to derive the short-form scale. On cross-validation using another HIE sample, the scale correlated 0.92 with the MHI total score (Davies, Sherbourne, Peterson, and Ware, 1988).
Current Health Perceptions
The 22-item Health Perceptions Questionnaire (HPQ) (Davies and Ware, 1981; Ware and Karmos, 1976; Ware, 1976) included six subscales that are substantially intercorrelated. The Current Health subscale was the most reliable and empirically valid of these (Davies and Ware, 1981). That subscale also best represents the overall HPQ concept, accounting for the largest amount of variance common to the HPQ subscales. For these reasons, four items were selected from the Current Health subscale (Davies and Ware, 1981). These items had high correlations with the Current Health subscale, had substantial and roughly equal correlations with other physical and mental health measures, and achieved the balance between favorably and unfavorably worded items necessary to control for acquiescent and opposition response sets.
Pain
The survey includes one measure of pain that asks respondents to rate pain on a scale from none to very severe.

Scale Scoring

Consistent with previous studies, limitations in physical and role functioning were counted regardless of duration and were scored to reflect the number of limitations present (Stewart et al. 1981; Stewart et al., 1982) Scores were reversed so that a high value indicated better functioning. Mental health scales were scored by summing the item responses, after reversing the scoring of some items, so that a high score indicated better health. Before combining items in the health perceptions scale, the response choices of the overall health item (item 1) were recoded to better reflect the unequal intervals of the item. The single-item measures were scored so that high scores indicated better social functioning and more pain. Finally, for all measures, scores were transformed linearly to 0-100 scales, with 0 and 100 assigned to the lowest and highest possible scores, respectively.

Reliability and Validity

Support for the reliability and construct validity of the SF-20 is provided in previously published documents (Stewart et al., 1988, Ware et al., 1992).

Respondent Burden

Self-administration requires approximately three minutes for most respondents. Administering the survey over the telephone requires about three or four minutes. These administration times represent a reduction in respondent burden of approximately ninety percent relative to the full-length versions from which the scales were derived.