Relating the Patient Health Questionnaire (PHQ-9) to the Montgomery-Asberg Depression Rating Scale (MADRS) with equipercentile linking in older adults.

Introduction The Patient Health Questionnaire (PHQ-9) and Montgomery-Asberg Depression Rating Scale (MADRS) are commonly used scales for measuring depression. It is useful to be able to relate their scores when comparing treatment outcomes. We used data from the Optimizing Outcomes of Treatment-Resistant Depression in Older Adults (OPTIMUM) clinical trial, which was a randomized clinical trial examining the effect of different pharmacotherapy strategies in older adults with treatment resistant depression. Methods For this late-breaking abstract, access to the trial dataset was granted in November 2022. Random sampling stratified by pharmacotherapy strategy was performed to split the sample 4:1 into training and validation samples. Equipercentile linking was performed on the training sample to identify corresponding scores of the PHQ-9 and MADRS using raw and change scores from baseline to 10 weeks. Bland-Altman analysis was performed to validate the linked scores on the validation sample. Results 555 participants were included in the training sample and 187 in the validation sample. PHQ-9 total scores of 1, 15, and 27 corresponded approximately with MADRS scores of 1, 24, and 48; PHQ-9 change scores of -13, 10, and 22 corresponded approximately with MADRS change scores of -23, 13, and 32 (see Tables). In the validation sample, the correlations between the PHQ-9 and MADRS scores were: rho = 0.76, p < 0.001 for week 10; rho = 0.66, p < 0.001 for change; the correlations between PHQ-9 original scores and PHQ-9 scores created from the MADRS scores were: rho = 0.77, p < 0.001 for week 10; rho = 0.64, p < 0.001 for change; similarly, correlations between MADRS original scores and MADRS scores created from the original PHQ-9 scores were: rho = 0.76, p < 0.001 for week 10; rho = 0.65, p < 0.001 for change. The Bland-Altman analysis did not show a proportional bias. Conclusions The correlations between the PHQ-9 and MADRS scores were moderate to high. Researchers could use our conversion tables to compare late-life depression studies that only used one of these scales. This work will benefit from replication in a mid-life sample. This research was funded by The Optimizing Outcomes of Treatment-Resistant Depression in Older Adults (OPTIMUM) clinical trial was funded by award TRD-1511-33321 from the Patient-Centered Outcomes Research Institute (PCORI).