Table 1 Comparison of abortion incidence estimation methodologies
Expected direction of bias | Expected magnitude of bias | Relative precision | Ability to capture self-induced abortions known only to woman | Potential to yield contextual information | Key data sources | |
|---|---|---|---|---|---|---|
Methods reviewed in this paper | ||||||
Abortion incidence complications method | Unknown | Unknown | Moderate | Only if led to PAC | From PAC facilities and knowledgeable informants | Health facility survey, survey of knowledgeable informants |
Direct questioning | Underestimation | Very high | Very low | Yes | From women who admit to having an abortion | Population-based survey of women |
List experiment | Underestimation | Low | Low | Yes | No | Population-based survey of women |
Confidante method | Underestimation | Moderate | High | No | Limited | Population-based survey of women |
CM with visibility factor | Unknown | Low | High | Yes | Limited | Population-based survey of women |
Modified AICM | Unknown | Unknown | Low /moderate | Yes | From PAC facilities and women who admit their abortions | Health facility survey, population-based survey of women |
Other methods | ||||||
Preceding birth technique | Underestimation | High | Low | Yes | From women who admit to having an abortion | Suvey of antenational clinic patients |
Sealed envelope | Underestimation | Inconsistent | Low | Yes | No | Population-based survey of women |
Randomized response technique | Unknown | Unknown | Low | Yes | No | Population-based survey of women |
Residual method | Unknown | Unknown | Very low | Yes | No | Secondary data |
Best friend approach | Underestimation | Low/moderate | Moderate | No | Limited | Population-based survey of women |
Sealed envelope | Underestimation | Moderate | Very high | No | Limited | Population-based survey of women |
Network scale-up | Unknown | Unknown | Very high | Yes | No | Population-based survey of women |