METHODS: We conducted an individual patient data meta-analysis. We fit bivariate random-effects models to assess diagnostic accuracy.
RESULTS: 16 742 participants (2097 major depression cases) from 54 studies were included. The correlation between PHQ-8 and PHQ-9 scores was 0.996 (95% confidence interval 0.996 to 0.996). The standard cutoff score of 10 for the PHQ-9 maximized sensitivity + specificity for the PHQ-8 among studies that used a semi-structured diagnostic interview reference standard (N = 27). At cutoff 10, the PHQ-8 was less sensitive by 0.02 (-0.06 to 0.00) and more specific by 0.01 (0.00 to 0.01) among those studies (N = 27), with similar results for studies that used other types of interviews (N = 27). For all 54 primary studies combined, across all cutoffs, the PHQ-8 was less sensitive than the PHQ-9 by 0.00 to 0.05 (0.03 at cutoff 10), and specificity was within 0.01 for all cutoffs (0.00 to 0.01).
CONCLUSIONS: PHQ-8 and PHQ-9 total scores were similar. Sensitivity may be minimally reduced with the PHQ-8, but specificity is similar.
OBJECTIVE: To use an individual participant data meta-analysis to evaluate the accuracy of two PHQ-9 diagnostic algorithms for detecting major depression and compare accuracy between the algorithms and the standard PHQ-9 cutoff score of ≥10.
METHODS: Medline, Medline In-Process and Other Non-Indexed Citations, PsycINFO, Web of Science (January 1, 2000, to February 7, 2015). Eligible studies that classified current major depression status using a validated diagnostic interview.
RESULTS: Data were included for 54 of 72 identified eligible studies (n participants = 16,688, n cases = 2,091). Among studies that used a semi-structured interview, pooled sensitivity and specificity (95% confidence interval) were 0.57 (0.49, 0.64) and 0.95 (0.94, 0.97) for the original algorithm and 0.61 (0.54, 0.68) and 0.95 (0.93, 0.96) for a modified algorithm. Algorithm sensitivity was 0.22-0.24 lower compared to fully structured interviews and 0.06-0.07 lower compared to the Mini International Neuropsychiatric Interview. Specificity was similar across reference standards. For PHQ-9 cutoff of ≥10 compared to semi-structured interviews, sensitivity and specificity (95% confidence interval) were 0.88 (0.82-0.92) and 0.86 (0.82-0.88).
CONCLUSIONS: The cutoff score approach appears to be a better option than a PHQ-9 algorithm for detecting major depression.
OBJECTIVE: To evaluate the degree to which using data-driven methods to simultaneously select an optimal Patient Health Questionnaire-9 (PHQ-9) cutoff score and estimate accuracy yields (1) optimal cutoff scores that differ from the population-level optimal cutoff score and (2) biased accuracy estimates.
DESIGN, SETTING, AND PARTICIPANTS: This study used cross-sectional data from an existing individual participant data meta-analysis (IPDMA) database on PHQ-9 screening accuracy to represent a hypothetical population. Studies in the IPDMA database compared participant PHQ-9 scores with a major depression classification. From the IPDMA population, 1000 studies of 100, 200, 500, and 1000 participants each were resampled.
MAIN OUTCOMES AND MEASURES: For the full IPDMA population and each simulated study, an optimal cutoff score was selected by maximizing the Youden index. Accuracy estimates for optimal cutoff scores in simulated studies were compared with accuracy in the full population.
RESULTS: The IPDMA database included 100 primary studies with 44 503 participants (4541 [10%] cases of major depression). The population-level optimal cutoff score was 8 or higher. Optimal cutoff scores in simulated studies ranged from 2 or higher to 21 or higher in samples of 100 participants and 5 or higher to 11 or higher in samples of 1000 participants. The percentage of simulated studies that identified the true optimal cutoff score of 8 or higher was 17% for samples of 100 participants and 33% for samples of 1000 participants. Compared with estimates for a cutoff score of 8 or higher in the population, sensitivity was overestimated by 6.4 (95% CI, 5.7-7.1) percentage points in samples of 100 participants, 4.9 (95% CI, 4.3-5.5) percentage points in samples of 200 participants, 2.2 (95% CI, 1.8-2.6) percentage points in samples of 500 participants, and 1.8 (95% CI, 1.5-2.1) percentage points in samples of 1000 participants. Specificity was within 1 percentage point across sample sizes.
CONCLUSIONS AND RELEVANCE: This study of cross-sectional data found that optimal cutoff scores and accuracy estimates differed substantially from population values when data-driven methods were used to simultaneously identify an optimal cutoff score and estimate accuracy. Users of diagnostic accuracy evidence should evaluate studies of accuracy with caution and ensure that cutoff score recommendations are based on adequately powered research or well-conducted meta-analyses.