P123: Measuring health-related outcomes: is social desirability bias an issue we should be exploring while conducting emergency department research?

Abstract

Introduction: Social desirability bias is a systematic error in self-report measures resulting from the desire of respondents to avoid embarrassment and project a favourable image of themselves to others. This bias may decrease the accuracy of self-reported health outcomes collected in health research compromise the validity of research findings. This study compared outcomes obtained by patient self-report vs. the same outcomes after undergoing verification and external adjudication, in trial involving patients with acute asthma. Methods: Cross-sectional analysis of outcome data obtained in a randomized controlled trial conducted in 6 Canadian emergency departments (ED). Adult patients were allocated to receive usual care (UC), opinion leader [OL] guidance to their primary care provider (PCP), or OL guidance+nurse case-management [OL+CM] for patients (NCT01079000). Asthma relapses and PCP follow-up visits were blindly assessed through patient self-report 30 and 90 days after their ED presentation for acute asthma. Each reported event was verified through the provincial electronic medical record, the ED Information Systems, and by calling the PCPs’ offices. Two study investigators, blinded to the study interventions, independently reviewed and adjudicated the verified outcomes. Disagreements were resolved by consensus prior to un-blinding. Results: Overall, 367 patients were enrolled; more were female (64%) and the median age was 28 years. Overall, patient follow-up was obtained in 85% of cases. The proportion of asthma relapses occurring within the first 90 days were lower when considering patient self-report than when considering the adjudicated outcomes (17%[39/227] vs. 19%[70/367]). The proportion of PCP follow-up visits occurring within the first 30 days were higher when considering patient self-report than when considering the adjudicated outcomes (47%[139/290] vs. 40%[146/367]). The pattern was similar, regardless of the arm of the study (UC vs. OL vs. OL+CM arms); outcome disagreement did not influence the direction of magnitude of the treatment effect. Conclusion: Social desirability bias could have influenced the outcomes obtained by patient self-report in this ED-based study. The direction of the bias was the same for both outcomes; however, the variation did not change the study results. This bias may play a role in studies with smaller sample sizes.