Abstract:
Antimicrobial resistance (AMR) is an increasing global health concern, especially in
developing countries. Intervention programs are being utilized widely to combat the spread
of AMR. For many resource limited countries, this is an expensive exercise with limited
funding. Therefore, before any intervention is implemented, it is crucial to estimate its
impact because it enables investigators to inform policy and planning. This can also
provide information on what risks/protective factors are significant for the region and
have the intervention tailored to the population.
In this thesis we propose a novel method using weighted likelihood (WL) that provides
estimates of the impacts of proposed interventions as well as confidence intervals. Our
research project was divided in three parts, first we conducted a literature study to find
out what models are currently used in estimating impacts of intervention studies, then
we used simulation to test our model and finally we applied the model to data from a
baseline study from a commune in Ha Nam Province, Vietnam.
Most intervention studies use applied time series analyses or two-group tests to evaluate
complete data (Cebotarenco & Bush, 2007; McNulty, Nichols, Boyle, Woodhead, & Davey,
2010; Mol et al., 2005; Willemsen et al., 2010). For proposed interventions, the standard
method is a mathematical model which provides only estimates without a way to quantify
precision (Pagel et al., 2009).
To test our proposed methods, we simulated 2000 data to generate and estimate the
impact of an intervention. Subsequently, we applied the model to the data from Vietnam.
Calibration was applied to the model using auxiliary information in order to gain efficiency.
The proposed model performed well in estimating the impact of the intervention
variable on the outcome. It also provided results for protective/risk factors that agree
with other published studies. This shows that, our methodology works well and can be
used to estimate the impact of intervention programs.