Table. Mean Bias of Accuracy Estimates for 1000 Resampled Studies of 100, 200, 500, and 1000 Participants .
| Mean difference (95% CI), percentage points | ||||||||
|---|---|---|---|---|---|---|---|---|
| Sample size = 100 | Sample size = 200 | Sample size = 500 | Sample size = 1000 | |||||
| Sensitivity | Specificity | Sensitivity | Specificity | Sensitivity | Specificity | Sensitivity | Specificity | |
| Sample-specific optimal cutoff scorea – Population-level optimal cutoff score ≥8b | 6.4 (5.7 to 7.1) | 0.6 (0.0 to 1.2) | 4.9 (4.3 to 5.5) | −0.3 (−0.8 to 0.2) | 2.2 (1.8 to 2.6) | 0.0 (−0.4 to 0.3) | 1.8 (1.5 to 2.1) | −0.6 (−1.0 to −0.3) |
| Sample-specific cutoff score ≥8 – Population-level optimal cutoff score ≥8 | −0.8 (−1.7 to 0.0) | 0.1 (−0.1 to 0.4) | 0.2 (−0.3 to 0.8) | −0.1 (−0.2 to 0.1) | 0.1 (−0.2 to 0.4) | 0.0 (−0.1 to 0.1) | −0.1 (−0.4 to 0.1) | 0.0 (−0.1 to 0.1) |
a
Sample-specific optimal cutoff score refers to the cutoff score maximizing the Youden index in each simulated sample.
b
The optimal cutoff score in the full Patient Health Questionnaire-9 individual participant data meta-analysis dataset is 8 or higher (sensitivity = 80.4%, specificity = 82.0%).