Estimating the utilization rates of bilateral tubal ligation services in 2008 and 2011 with missing information on the year when the procedure was undertaken
Date
2014-04
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
Estimation of annual use rates for bilateral tubal ligation (BTL) service is essential for program planning since it serves as basis in determining the resource requirements to ensure availability of BTL service. The standard method in estimating utilization of BTL service is based on the ex-ante decision on what family planning (FP) method to use. However, readily available data from population-based surveys, like the standard Demographic Health Surveys (DHS), provide ex-post information on what FP method women were currently using at the time of the survey. Incidence rate measures the annual utilization of BTL service but population-based surveys yield the prevalence rate, a cumulative estimate of BTL utilization. The prevalence rate when used for annual estimation would overestimate the number of women expected to have undergone the procedure in a given year.
Given the limitations of available data, use of a two-stage model that follows Heckman's approach is proposed to estimate the incidence rate. The first stage estimates the BTL prevalence rate, while the second stage allows estimation of annual BTL utilization using information on the year of BTL operation which is available in the 2008 NDHS. Result of the mean estimates showed that BTL use rate in 2008 was only 0.302 percent, significantly lower than the reported BTL prevalence rate of 6 percent.
However, information on year of BTL operation is not always available. For some DHS-like surveys that have missing information on the year of BTL operation such as the 2011 Family Health Survey (FHS), it is proposed that the year of last pregnancy be used as proxy. To determine the validity of the proposed solution, the two-stage model with information on year of BTL operation and the two-stage model with proxy indicator are implemented using the 2008 NDHS. The models are then compared using the Hausman specification test. Test result showed that the two models have no systematic difference, thus, the two-stage model with proxy variable is a valid alternative of the model with information on year of BTL operation in cases where the latter is not available. Using the most recent data, BTL use rate in 2011 was estimated at 0.167 percent or equivalent to 37,115 BTL clients, significantly smaller than the reported BTL prevalence rate of 5.6 percent (equivalent to 1.3 million women).