Method (ranked by validity) | Feasibility and validity considerations | Example |
---|---|---|
1 Observational assessments of quality of care | ● Most accurate information for those births observed | Done in MCHIP Quality of Care assessments and now included as an optional Demographic and health Survey (DHS) Service Provision Assessment (SPA) module (done in Kenya; planned for Malawi) |
● Not commonly done | ||
● Expensive to conduct | ||
● When done, not likely to be on a large and nationally representative sample | ||
● Likely excludes home births | ||
2 Facility readiness assessments | ● Need to extrapolate from availability of commodity/personnel to actual use of uterotonic | DHS SPA, Service Availability and Readiness Assessment (World Health Organization) |
● Expensive to conduct on a representative sample on a regular basis | ||
● Likely excludes home births | ||
3 Routine Health Management Information System (HMIS) data | ● Only possible where data are recorded in registers and reported to higher levels | Included in registers in some countries (e.g., Mozambique) |
● HMIS data quality variable | ||
● No additional data collection costs required | ||
● May not include community-level reporting on home births | ||
4 Data from sentinel surveillance sites | ● Only possible where such sites are available | MCHIP used this method in Kenya (results unpublished) |
● No additional data collection costs required | ||
● Question of generalizability | ||
5 Extrapolation from service contact data | ● Estimates require extrapolation with questionable assumptions (i.e., that skilled birth attendant and/or institutional birth implies use of uterotonic in most or all of covered births). | Suggested method – expert panel not aware of previous experience with this |
● No additional data collection costs required | ||
6 Survey of key informants | ● Easy and low cost to interview individuals or group of informants | Suggested method – expert panel not aware of previous experience with this. |
● Likely to be subjective with opinions likely biased and/or based on incomplete information |