Purpose: As patient-reported outcome measures (PROMs) are increasingly used in clinical practice for screening, monitoring, and management, the potential for response bias has been raised (e.g., over-reporting problems for attention, under-reporting to avoid treatment changes/discontinuation). We investigated whether patients systematically bias their responses when they know clinicians will review their PROM results.
Methods: We conducted secondary analyses of three experimental studies evaluating PROMs in adult and pediatric care. Prior to PROM completion, intervention group patients were informed that the results would be shown to their clinicians ("feedback" arm), whereas control group patients were told that their clinicians would not see their responses ("no feedback" arm). Independent sample t-tests compared the "feedback" and "no feedback" arms' PROM scores at baseline. Effect sizes and 95% confidence intervals were estimated using Cohen's d statistics with Hedges' g correction, and effect sizes > 0.50 were considered clinically relevant.
Results: Across the 29 domains assessed in the three studies, no between-arm differences reached an effect size of ± 0.50. Only 3/29 effect sizes exceeded ± 0.30. The confidence intervals for 14 domains included ± 0.50, with 4 favoring the "no feedback" arm and 10 favoring the "feedback" arm. Two domains reached statistical significance, one favoring the "no feedback" arm and one favoring the "feedback" arm.
Conclusion: This study does not support the hypothesis that patients systematically bias their PROM responses if they know that clinicians will see their results. These findings support using PROMs in clinical practice as a valid mechanism to promote patient-centered care.
Keywords: Bias; Clinical practice; Patient-reported outcome measures; Validity.
© 2024. The Author(s).