[Paper Review] Optimization of Distribution Network Configuration for Pediatric Vaccines using Chance Constraint Programming

Zahra Azadi, Sandra D. Ekşioğlu+1 more|arXiv (Cornell University)|Jun 9, 2020

Supply Chain and Inventory Management35 references4 citations

TL;DR

This paper proposes a stochastic optimization model using chance constraint programming to improve pediatric vaccine distribution in low- and middle-income countries. By modeling demand uncertainty and evaluating supply chain design, vial sizing, and new vaccine technologies, the study finds that removing regional tiers and adopting thermostable vaccines significantly increases immunization coverage and vaccine availability while reducing waste.

ABSTRACT

Millions of young people are not immunized in low- and middle-income (LMI) countries because of low vaccine availability resulting from inefficiencies in cold supply chains. We create supply chain network design and distribution models to address the unique characteristics and challenges facing vaccine supply chains in LMI countries. The models capture the uncertainties of demand for vaccinations and the resulting impacts on immunization, the unique challenges of vaccine administration (such as open vial wastage), the interactions between technological improvements of vaccines and immunizations, and the trade-offs between immunization coverage rates and available resources. The objective is to maximize both the percentage of fully immunized children and the vaccine availability in clinics. Our research examines how these two metrics are affected by three factors: number of tiers in the supply chain, vaccine vial size, and new vaccine technologies. We tested the model using Niger's Expanded Program on Immunization, which is sponsored by the World Health Organization. We make many observations and recommendations to help LMI countries increase their immunization coverage.

Motivation & Objective

• To address inefficiencies in cold supply chains that limit vaccine availability and immunization coverage in low- and middle-income (LMI) countries.
• To evaluate the impact of supply chain structure, vaccine vial size, and new vaccine technologies on immunization outcomes.
• To maximize both the percentage of fully immunized children (FIC) and vaccine availability in clinics under demand uncertainty.
• To provide actionable policy recommendations for LMI countries using a data-driven, stochastic optimization framework.

Proposed method

• Formulates a stochastic optimization model with chance constraints to handle uncertain demand for pediatric vaccines.
• Models a four-tier supply chain (national, regional, district, clinics) and evaluates a three-tier alternative by removing regional stores.
• Incorporates open vial wastage (OVW) and vaccine expiration into the model using probabilistic constraints.
• Evaluates trade-offs between vial size (1, 5, 10, 20 doses) and their impact on OVW and resource utilization.
• Analyzes the effects of new technologies, including dual-chamber vaccines and thermostable vaccines, on storage needs and coverage.
• Uses real-world data from Niger’s Expanded Program on Immunization (EPI) to calibrate and validate the model.

Experimental results

Research questions

• RQ1 How does reducing the supply chain from four to three tiers affect vaccine availability and immunization coverage?
• RQ2 What is the impact of using mixed vial sizes (1, 5, 10, 20 doses) on open vial wastage and overall vaccine availability?
• RQ3 How do dual-chamber vaccines affect storage capacity and immunization coverage in resource-constrained settings?
• RQ4 To what extent do thermostable vaccines improve immunization coverage and reduce waste in LMI countries?
• RQ5 What is the optimal trade-off between vaccine availability, waste reduction, and storage capacity when adopting new vaccine technologies?

Key findings

• Streamlining the supply chain by removing regional stores increases average fully immunized children (FIC) and supply ratio (SR), especially when cold storage capacity is reallocated to clinics.
• Using a combination of different vial sizes reduces open vial wastage (OVW), but does not always lead to higher FIC or SR due to inventory and handling trade-offs.
• Dual-chamber vaccines increase storage demands per dose and reduce overall vaccine availability, despite eliminating OVW and simplifying administration.
• Thermostable vaccines reduce waste and increase immunization coverage by freeing up cold storage space, making them a strategic investment for high-volume vaccines like DTP-Hep-Hib.
• Reallocating cold storage capacity from regional to clinic levels significantly improves vaccine availability and FIC, highlighting the importance of infrastructure reallocation.
• The model demonstrates that stochastic optimization with chance constraints provides more accurate and effective solutions than deterministic or simulation-based approaches for vaccine supply chain planning.