Interpreting SF-36 summary health measures: a response

Area of Science:

• Health Outcomes Research
• Psychometrics
• Biostatistics

Background:

• The accuracy of extreme scores on the SF-36 physical component summary (PCS) and mental component summary (MCS) has been questioned.
• Concerns exist regarding the interpretation of very high and very low PCS and MCS scores.
• The development, scoring, and factor content of the SF-36 subscales are relevant to score validity.

Purpose of the Study:

• To investigate the empirical validity of extreme SF-36 PCS and MCS scores.
• To address hypotheses suggesting extreme scores are artifacts of scoring weights.
• To provide evidence-based recommendations for interpreting SF-36 summary scores.

Main Methods:

• Analysis of published cross-tabulations between SF-36 items and summary scores.
• Reanalysis of public datasets (N = 5919).
• Examination of preliminary data from the Medicare Health Outcomes Survey (HOS) (N = 172,314).

Main Results:

• Little to no evidence supports the hypothesis that extreme SF-36 scores are invalid artifacts.
• In the HOS, individuals with "unexpected" PCS scores (e.g., <20) had a 25% higher mortality risk compared to the next highest score category.
• All empirical tests supported the validity of extreme PCS and MCS scores in predicting outcomes.

Conclusions:

• Extreme SF-36 PCS and MCS scores demonstrate empirical validity.
• Recommendations are made against interpreting average score differences smaller than one point for detecting "false" measurement.
• Thorough comparison of summary measures with SF-36 profiles is recommended before drawing conclusions.