Do SF-36 summary component scores accurately summarize subscale scores?

Area of Science:

• Health Outcomes Measurement
• Psychometrics
• Biostatistics

Background:

• Standard scoring algorithms for the SF-36 yield Physical Component Summary (PCS) and Mental Component Summary (MCS) scores.
• Concerns exist regarding the independence of PCS and MCS scores, suggesting potential overlap in measured constructs.

Purpose of the Study:

• To examine the relationships between SF-36 subscales and their summary component scores (PCS and MCS).
• To investigate the interrelationships between PCS and MCS scores.
• To explore the implications of these relationships for interpreting research findings.

Main Methods:

• Simulation analyses to illustrate scoring algorithm contributions to discrepancies.
• Correlation and regression analyses using the Swedish SF-36 normative database.
• Examination of relationships between subscale profiles and component scores.

Main Results:

• Significant negative correlations (r = -0.74, -0.67) were observed between PCS and MCS scores at upper scoring intervals.
• Regression analyses indicated substantial overlap: PCS measured 57% mental health aspects, and MCS measured 65% physical health aspects in these ranges.
• Discrepancies between subscale profile scores and component scores were identified and explained.

Conclusions:

• The current PCS and MCS scoring procedure may inaccurately summarize SF-36 subscale profiles.
• Component scores should be interpreted cautiously and alongside profile scores until the scoring procedure is revised.
• Further research is needed to refine SF-36 scoring algorithms for improved accuracy.