Access to diagnostic testing services is a critical element of public health systems in countries with high prevalence of communicable diseases such as the Human Immunodeficiency Virus (HIV) and Tuberculosis (TB). In resource-limited settings, diagnostic services are frequently provided through hierarchical networks comprised of health facilities that collect samples (e.g. blood, sputum, etc.) from patients, and centralized laboratories that process these samples.
The design and implementation of a new optimized sample transportation (OST) system …with the ultimate goal of reducing result turnaround times in Malawi’s diagnostic network.
This strategy provides low cost, high-volume testing while ensuring that patients in remote areas have access to diagnostic services within their own communities (Peter et al. 2017, Alemnji et al. 2014). Sample transport (ST) systems enable the physical movement of samples and test results between health facilities, district hubs, and laboratories within these multi-stage networks, and contribute significantly to the efficiency of diagnostic service delivery (Lecher et al. 2015).
In many countries in sub-Saharan Africa, ST systems are centrally operated with fixed weekly or monthly transportation routes necessitated by a lack of real-time ST demand information from health facilities. Consequently, ST operators are unable to dynamically allocate transportation resources to adapt to daily variability in ST demand, leading to unnecessary travel (when health facilities do not have samples for collection) and unnecessary delays (when samples at facilities have to wait for the next scheduled visit). In this paper, we describe the design and implementation of a new optimized sample transportation (OST) system that addresses these challenges, with the ultimate goal of reducing result turnaround times in Malawi’s diagnostic network.
The OST system comprises two components: (i) a novel data sharing platform to monitor incoming sample volumes at healthcare facilities, and (ii) an optimization based solution approach for generating daily transportation schedules and courier routes, in response to current demand at each site.
First, we design the information sharing platform in collaboration with a local telecommunications firm in Malawi (§4). The platform is a text-based Unstructured Supplementary Service Data (USSD) application that allows health facility staff to report how many samples are awaiting This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=3712556 Preprint not peer reviewed Gibson: Optimized Sample Transportation 00(0), pp. 000–000, c 0000 INFORMS 3 collection on a daily basis.
Second, we develop an optimization based approach for dynamically scheduling and routing motorcycle couriers to clinics—assuming the information sharing platform provides daily updates about ST demand. To this end, we formulate a general multi-stage version of the dynamic multi-period vehicle routing problem (which we refer to as M-DMVRP and define in §5), develop an optimization-based solution approach (§6), and numerically evaluate the performance of this solution approach for scheduling and routing couriers on synthetic multi-stage transportation networks (§7).
As part of our numerical experiments, we compare our solution approach to a lower bound on the optimal solution, which can be generated for small instances of the problem (§7.1). We also benchmark our method against other scalable solution methods on networks of a realistic size (§7.2).
Finally, we conduct sensitivity analysis to illustrate how the performance of dynamic scheduling policies depends on the accuracy of the information collected through the information sharing platform (§7.3). We implement and test this system in collaboration with Riders For Health Malawi (RFH), a nonprofit organization that operates a national sample transportation system reaching approximately 700 healthcare facilities.
We assess the impact of this approach in a pilot study conducted in three districts in Malawi (each comprising 15-18 health facilities served by two RFH couriers) from August–October 2019 (§8 provides details on the field trial design).
Based on analysis of over 20,000 samples and results transported during this study, we show that the implementation of OST routes reduced average ST delays by approximately 25%. The ongoing implementation of this system also resulted in a 55% decrease in the proportion of unnecessary trips by ST couriers, demonstrating that optimized ST routes improve turnaround times without increasing unnecessary travel (§9).