Table 2.
Multiple-bias analysis results for Study A: proportion of maternal deaths due to unsafe abortion adjusted for selection bias, misclassification, and random error, after 50,000 simulation trials per scenario
| Bias model | Scenario (probability distribution/s) | Median | 2.5, 97.5 percentiles | Ratio of limits |
|---|---|---|---|---|
| None (reported) | NA | 0.206 | 0.206, 0.206 | 1.00 |
| None (conventional, with estimate of precision) | NA | 0.206 | 0.196, 0.316 | 1.61 |
| Adjusted for misclassification only, no random error | 1 (W1&W2 narrow) | 0.369 | 0.248, 0.523 | 2.11 |
| 2 (W1&W2 medium) | 0.368 | 0.245, 0.530 | 2.16 | |
| 3 (W1&W2 wide) | 0.372 | 0.242, 0.550 | 2.27 | |
| Adjusted for misclassification and selection bias, no random error | 4 (W1&W2 narrow, W3&W4 narrow) | 0.305 | 0.209, 0.431 | 2.06 |
| 5 (W1&W2 narrow, W3&W4 medium) | 0.308 | 0.210, 0.438 | 2.08 | |
| 6 (W1&W2 narrow, W3&W4 wide) | 0.302 | 0.206, 0.432 | 2.10 | |
| 7 (W1&W2 medium, W3&W4 narrow) | 0.305 | 0.207, 0.437 | 2.11 | |
| 8 (W1&W2 medium, W3&W4 medium) | 0.306 | 0.207, 0.443 | 2.14 | |
| 9 (W1&W2 medium, W3&W4 wide) | 0.302 | 0.204, 0.437 | 2.11 | |
| 10 (W1&W2 wide, W3&W4 narrow) | 0.308 | 0.206, 0.452 | 2.19 | |
| 11 (W1&W2 wide, W3&W4 medium) | 0.310 | 0.206, 0.458 | 2.22 | |
| 12 (W1&W2 wide, W3&W4 wide) | 0.305 | 0.203, 0.452 | 1.96 | |
| Adjusted for misclassification and selection bias, random error included | 13 (W1&W2 narrow, W3&W4 narrow) | 0.307 | 0.175, 0.462 | 2.64 |
| 14 (W1&W2 narrow, W3&W4 medium) | 0.310 | 0.176, 0.469 | 2.66 | |
| 15 (W1&W2 narrow, W3&W4 wide) | 0.304 | 0.173, 0.461 | 2.66 | |
| 16 (W1&W2 medium, W3&W4 narrow) | 0.307 | 0.174, 0.467 | 2.68 | |
| 17 (W1&W2 medium, W3&W4 medium) | 0.309 | 0.173, 0.473 | 2.78 | |
| 18 (W1&W2 medium, W3&W4 wide) | 0.304 | 0.170, 0.466 | 2.74 | |
| 19 (W1&W2 wide, W3&W4 narrow) | 0.311 | 0.173, 0.480 | 2.77 | |
| 20 (W1&W2 wide, W3&W4 medium) | 0.313 | 0.172, 0.485 | 2.81 | |
| 21 (W1&W2 wide, W3&W4 wide) | 0.308 | 0.169, 0.480 | 2.84 |
W1: Selection probability for abortion-related deaths
W2: Selection probability for non-abortion-related deaths
W3: Sensitivity of cause of death classification
W4: Specificity of cause of death classification