Table 1 Criteria for evaluating methodologies to estimate maternal mortality in refugee and IDP camps in low resource settings
Criterion | Definition | Example | Relevance |
|---|---|---|---|
Statistical training | The level of statistical knowledge required to implement the methodology and analyze results. Referent is an individual with basic arithmetic skills | Bayesian statistics, multilevel modeling | Camp settings: Statisticians with advanced knowledge and skills may not be readily available |
Bias | The extent to which the methodology introduces bias that may affect internal or external validity | Internal: misclassification bias that results in some deaths being counted or not counted as maternal deaths and/or abortion-related; selection bias that prevents certain individuals’ deaths from being counted External: generalizability that prevents results from being widely interpreted | Abortion: Informants reporting on a maternal death must know about a pregnancy and circumstances of a death to be able to report accurately if it is abortion-related. Due to abortion stigma, deaths that are known to be due to abortion may not be reported as such |
Cost | The total cost to implement the methodology, including human resources' level of effort and data collection and analysis tools. This cost excludes transportation and incidentals for human resources | X% of total maternal health costs, x total US$ | Camp settings: There may not be sufficient budgets in humanitarian response for expensive surveillance systems for monitoring and evaluation |
Data collection training | The amount of training required for an individual with no background in data collection | Survey methodology, interview techniques, disease identification | Abortion: Due to stigmatization of abortion, enumerators must be sensitized to the topic and trained on how to collect data without introducing bias |
Data sources | Whether a single or multiple data sources are required. The type of data source used for the estimation of maternal mortality | Facility records, death records, verbal autopsy | Camp settings: Data sources may be incomplete or missing Abortion: Due to stigmatization of abortion, deaths due to abortion complications in facilities may be coded incorrectly. Reports about deaths from informants may be inaccurate for the same reason |
Definitions | Whether the data source includes a definition of maternal mortality that is consistent with the World Health Organization (WHO) definition [2] or a different maternal death definition | Deaths of pregnant individuals according to International Classification of Diseases (ICD) 10 [33] or ICD 11, [34] any death of a reproductive age individual, etc | Abortion: To accurately capture abortion-related deaths, definitions must specify abortion as a cause of death |
Digitalization | Whether the methodology can be digitized into an existing open-source platform for data collection, management, and analysis | Google forms, [35] DHIS2 program, [36] Open Data Kit [37] or KoboToolbox [38] questionnaire | Camp settings: Digitalization of methodologies may simplify data collection during chaotic humanitarian responses Abortion: Using digital data collection methods, may improve ability to collect information confidentially through informants' direct interaction with questionnaires or other data collection forms, protecting them from stigma associated with abortion |
Human resources | The number of individuals and their time required to implement the methodology relative to sample size | X clinicians, x health facility managers, x community health workers, etc. for x days | Camp settings: Due to competing priorities, human resources may not be readily available |
Sample size | The minimum number of maternal deaths and/or abortion-related deaths required by the methodology to produce a precise estimate | X total maternal deaths, x maternal deaths attributable to abortion complications | Camp settings: These settings may not have large sample sizes needed to generate estimates using certain methodologies Abortion: Since the contribution of abortion-related complications to the structure of causes of maternal mortality in humanitarian settings is unknown, methodologies may need to be able to utilize small numbers to produce estimates |
Time needed for implementation | The amount of time required to implement the methodology and produce results | X days, weeks, months, years | Camp settings: In rapid responses, it may be important to implement methodologies quickly, in order to establish a baseline for monitoring and evaluation of sexual and reproductive health service provision |
Timing of point estimate relative to data collection | The amount of time between the end of the time period used for analysis and the estimate of maternal mortality | Methodology produces an estimate of mortality in last x months/weeks/days after x months/weeks/days of methodology implementation | Camp settings: In rapid responses, it may be important to implement methodologies quickly, in order to establish a baseline for monitoring and evaluation of sexual and reproductive health service provision |