Table 1.
Definition of the risk of bias criterion |
||||
---|---|---|---|---|
Risk-of-bias (rating score)
|
Source of data for maternal death in the study (information bias)
|
Completeness of cause-of-death data in the study (missing data bias)
|
Use of ICD-MM in assigning the causes of death in the study (selection bias)
|
Competence level of persons who assigned the causes of death (measurement bias)
|
Low (1) |
Health facility records |
≥90% deaths assigned causes |
Use of ICD-MM definition stated |
Expert panels / MDSR committees |
Medium (2) |
MDSR audits |
75-90% deaths assigned causes |
Not stated but data collected 2015 onwards |
Primary clinical assessments in facility records |
High (3) |
Community surveys with VAs |
<75% deaths assigned causes |
Not stated and data collected 2010 to 2014 |
Not stated |
Rationale for ratings | Health facility records contain all details of events leading to the death. MDSR audits contain summaries, and VA reports lack clinical detail. | The rating looks at increasing the risk of bias. WHO recommends ≥75% completeness to rate the data as reasonable for use in review studies. | Mention of ICD-MM use assures that the standard definitions and classification of the causes of death was followed. Data from 2015 gives near assurance, and data before that may not have used ICD-MM classification. | Deaths assessed by an expert or CEMD/MDSR committees are more comprehensive and accurate in their assessments. Facility clinical assessments may be biased or less competent. |
VA – verbal autopsy, CEMD/MDSR – confidential enquiry into maternal deaths/maternal death surveillance and response