minimod: An Open Source Python Package to Evaluate the Cost Effectiveness of Micronutrient Intervention Programs

Abstract Objectives Micronutrient (MN) deficiencies are a public health problem in many low- and middle-income countries. Given limited funds to invest in MN intervention programs, it is important to identify sets of interventions that are cost-effective. We describe a Python package, minimod, that leverages data on the benefits and costs of alternative intervention programs to find sets of economically optimal interventions over space and time. This software can be used to carry out Monte Carlo simulations and assess the stability of the results to different levels of uncertainty. We illustrate the functionality of this software using examples from the Micronutrient Intervention Modeling Project. Methods minimod is an open-source python package that uses mixed-integer programming to find sets of intervention programs that maximize nutritional benefits subject to cost constraints or that minimize costs subject to a pre-specified benefit constraint. The software is designed for use with estimates of effective coverage (% of a population that achieves adequate MN intake), lives saved, or any other nutritional benefit data coupled with its costs. The package can then run Monte Carlo simulations under reasonable distribution assumptions, and it can be easily programmed to allow other distributions based on users’ needs. Results This optimization procedure can yield the set of MN intervention programs that are cost-effective over time (e.g., over a 10-year time horizon) and space (e.g., across administrative regions). Results are summarized by several visualizations and a report of the frequency with which each intervention appears to be economically optimal across a given set of simulations. When applied to vitamin A programs in Cameroon, the model predicts that investment reallocation can both increase the number of children covered and reduce the cost per child. Conclusions minimod is a tool that can be used by stakeholders and policymakers to identify cost-effective MN intervention programs over space and time. By comparing optimal solutions against current intervention programs, this tool highlights the potential cost savings associated with adopting more cost-effective policies. Funding Sources This research was supported by a grant to UC Davis from Helen Keller International through their support from the Bill & Melinda Gates Foundation.