Purpose: Recent US-based studies have utilised item response theory (IRT) to equate several self-report scales for depression and anxiety using the PROMIS depression and anxiety common metrics. The current study reports on the validity of the US-based equating procedures for the Patient Health Questionnaire-9 (PHQ-9), Generalized Anxiety Disorder-7 (GAD-7) and Kessler 6 psychological distress scale (K6) to equate scores in a large online sample of Australian adults.
Methods: Data comprised 3175 Australians recruited online. Each participant provided responses to the PROMIS depression and anxiety item banks, the PHQ-9, the GAD-7 and the K6. Two scoring methods were used to convert the scores on the PHQ-9, GAD-7 and K6 to the PROMIS depression and anxiety metrics. The converted scores were compared to the PROMIS depression and anxiety scores using intraclass correlations, mean difference, mean of absolute differences and Bland-Altman limits of agreement.
Results: Statistically significant mean differences were identified in five out of eight equated scores, albeit the effect sizes were small (Cohen's dz ≤ 0.25). The correlations were uniformly high (ICC ≥ 0.86). The mean of absolute differences between observed and equated scores for each metric and across scoring methods ranged between 4.23 and 5.33.
Conclusions: The results demonstrate the validity of generating PROMIS depression and anxiety scores from the PHQ-9, GAD-7 and K6 in an independent sample of Australians. The agreement between equated scores provides some assurance that researchers and clinicians can utilise the converted PHQ-9, GAD-7 and K6 scores on the PROMIS metrics without a substantial decrease in accuracy and precision at the group level.
Keywords: Item response theory; PROMIS; Scale equating; Validity.