PIT16 ACCURACY OF RECALL IN HEALTH-RELATED QUALITY OF LIFE ASSESSMENT AMONG TRAUMA PATIENTS: A PROSPECTIVE LONGITUDINAL COHORT STUDY WITH RETROSPECTIVE ASSESSMENTS

Abstract

To investigate the accuracy of recall in health related quality of life (HRQL) assessment among trauma patients. Adult trauma patients who attended the Emergency Department and were admitted between March 2016 and November 2016 were invited to participate. Postal surveys were sent one week (T1), three months (T2) and 12 months (T3) post trauma. The EQ-5D-3L and Visual Analogue Scale (EQ-VAS) were used to assess current and recalled HRQL. A total of 463 patients completed the EQ-5D at T1, T2 and T3. The mean T1 EQ-5D summary score was 0.484, whereas recalled T1 EQ-5D at T2 and T3 was on average 0.413 (p<0.001) and 0.368 (p<0.001), respectively. At T2, the mean EQ-5D summary score was 0.719 and recalled T2 EQ-5D at T3 was 0.692 (p<0.001). The accuracy of recall was fair for T1-T2 (intraclass correlation coefficient (ICC)=0.558) and T1-T3 (ICC=0.507) and good for T2-T3 (ICC=0.615). Most distortion was seen on the domain pain (ICC=0.377-0.478). Mean T1 EQ-VAS was 56.3, whereas mean recalled T1 EQ-VAS on T2 and T3 was 54.8 (p=0.100) and 53.5 (p=0.003), respectively. At T2, mean EQ-VAS was 71.6 and T2 recalled EQ-VAS-score on T3 was 67.0 (p<0.001). Inter-temporal agreements were fair on T1-T2 (ICC=0.553), T1-T3 (ICC=0.555) and T2-T3 (ICC=0.593). The accuracy of recall was lowest in patients with a major trauma (injury severity score≥16). Recalled HRQL measured by the EQ-5D-3L and EQ-VAS is systematically lower compared to the originally reported HRQL. Recall bias may lead to distortion and misinterpretation of the outcome. Researchers should therefore seriously consider recall bias when using retrospective assessments.