Skip to main content

Table 3 Summary of key limitations related to data inputs and model methodology

From: Estimating the contribution of a service delivery organisation to the national modern contraceptive prevalence rate: Marie Stopes International's Impact 2 model

Area Limitation Ability to minimize
Data-related limitations
Service provision data Service data must reflect services provided directly to clients; potential challenges include ensuring data quality, and inflation when counting commodities further back in supply chain. Model is best positioned to be used by organisations delivering services directly to clients. Organisations should ensure their data accurately reflect services provided.
Client profile data High potential to over- or underestimate CPR contribution depending on accuracy of client profile data. High and low estimates created based on alternative client profiles.
National assumptions (e.g. mortality rates, discontinuation rates, population projections) Limited potential to over- or underestimate results. Best available data have been pre-loaded, with potential error minimised as much as possible.
Model methodology-related limitations
Converting services to short-term users Potential to either over- or underestimate number of users due to uncertainty around how many of the delivered short-term commodities are actually used. For coitus-based methods, there are limited data on units needed in a year, and the potential that dual protection double counts users. Remove condom users from calculations to reduce greatest uncertainty. For pills and injectables, the approach minimises overestimation as much as possible given current data on consistency of use.
Treatment of provider changers Potential to underestimate increases in CPR as model assumes provider changers are not replaced, and that women leaving the organisation (e.g. women who stop using contraceptives) are not counted in the CPR. High and low estimates created based on alternative client profiles.