Full Citation

Title: Estimating modern contraceptive supply shares at national and subnational administration levels using Demographic and Health Survey data, with an application to the calculation of estimated modern contraceptive use

Citation Type: Dissertation/Thesis

Publication Year: 2023

ISBN:

ISSN:

DOI:

NSFID:

PMCID:

PMID:

Abstract: Contraceptive method supply shares reflect the contributions of the public and private sectors to the distribution of a given method each year. Quantifying these public/private-sector supply shares of contraceptive methods within countries is vital for effective and sustainable family planning delivery. They are useful to Family Planning officials as they show where contraceptive users have obtained their most recent supplies. Unfortunately, due to the cost, many low- and middle-income countries (LMIC) are not in a position to carry out the national-scale surveys necessary to collect this data regularly. Therefore, they evaluate the contraceptive supply market using out-of-date data. Using out-of-date supply share estimates for family planning monitoring has significant knock-on effects. They may lead to inaccurate conclusions on the stability of the contraceptive supply market. In addition, the estimation of other family planning indicators that depend on method supply shares will also be inaccurate and distorted. To date, neither Bayesian nor frequentist methods have been used to estimate this important family planning indicator. In this thesis, a methodology using Bayesian hierarchical penalised spline models for estimating modern contraceptive method supply shares, at national and subnational administration levels, for LMICs participating in the global Family Planning 2030 (FP2030) initiative is proposed. A series of Bayesian models that evaluate method supply shares using both large multi-country datasets and computationally efficient single-country datasets are described. Lastly, the impact of using national-level annual contraceptive method supply shares with uncertainty in the calculation of another key family planning indicator, estimated modern use (EMU), is evaluated. To begin, an approach for estimating the proportion of modern contraceptive methods supplied by the public and private sectors at the national administration level using a multi-country dataset is described and evaluated. The proposed approach utilises Bayesian hierarchical modelling techniques, taking advantage of the geographic nature of the data, in combination with penalised splines, capturing the complex shape of the data over time, to produce annual estimates with uncertainty for these supply-share proportions. Global-level correlations between rates of change in method supply shares are estimated and incorporated into the modelling approach to promote more precise estimation, even in the absence of data. This modelling approach is compared and validated against simpler modelling alternatives. Next, the mcmsupply R package is presented. This R package combines 4 variations of the multi-country national model. The national modelling approach is extended in two directions. Firstly, the model is extended to estimate method supply shares at the sub-national administration level using a multi-country subnational dataset. The motivation for this model extension is an ever-growing interest in subnational family planning indicators due to the decentralisation of family planning services in countries participating in FP2030. Utilising the knowledge gained from modelling the national-level estimates, hierarchical modelling structures of the multi-country national model are incorporated into the multi-country subnational model. Secondly, the model is extended again to estimate the method supply shares (at either national or subnational administration levels) using only the data for a single country. Similarly, informative priors informed by the multi-country model parameter estimates are used in the single-country modelling approach. This reduces the uncertainty of survey estimates, making the estimates more precise and reliable. The single-country modelling approach is computationally efficient, without a loss of model accuracy. Lastly, an alternative application of annual method supply share estimates in the calculation of another family planning indicator, estimated modern use (EMU) derived from family planning service statistics, is demonstrated. Presently, EMUs are a stand-alone estimate without any associated uncertainty. This chapter considers the calculation process for EMUs. An updated methodology is proposed that calculates EMUs using annual method supply shares with uncertainty, rather than the existing approach that uses the method supply shares observed in the most recent DHS survey for a given country. The impact of this potential update is considered and compared against the existing approach. The benefits and strengths of the proposed update are illustrated using case studies in Country A and Country B.

Url: https://mural.maynoothuniversity.ie/18132/1/Hannah_Comiskey_Thesis_2023_V2.pdf

User Submitted?: No

Authors: Comiskey Leontine Alkema Niamh Cahill, Hannah

Institution: Hamilton University

Department: Statistics

Advisor:

Degree:

Publisher Location:

Pages: 1-410

Data Collections: IPUMS Global Health - DHS

Topics: Health, Methodology and Data Collection, Population Health and Health Systems

Countries: