Mobile clinics routing and scheduling in the Witzenberg region of South Africa

Abstract

Despite much literature on operations research applied to various healthcare problems, impactful implementation in public healthcare is limited, which often results in allocative inefficiency. This article uses a mobile clinic routing and scheduling problem in the Witzenberg region of South Africa as a case study to demonstrate the improvement of implementation success through cross-disciplinary collaboration, and also to propose a new three-stage approach for modelling a mobile clinic problem that incorporates continuity of care, fairness, and minimisation of distance travelled. Mobile clinics are used in many countries to improve access to healthcare for rural communities. Decision makers must assign farms or villages to mobile clinics, and determine their monthly visit schedules. To improve implementation success, we follow a collaborative three-phased mixed-methods approach with healthcare professionals to improve workload balance, fairness, and transportation cost. During phase 1, qualitative and quantitative data are gathered through qualitative research methods. In phase 2, fairly distributed optimal routes and schedules are designed using a three-stage model that incorporates a multi-vehicle routing problem to determine daily routes, a knapsack problem to establish a fair allocation of these daily routes between different clinics, and another variation on the vehicle routing problem to determine the monthly visit schedule that minimises the distance between the last farm visited on each consecutive day in the case of having to return to a farm the next day. Different input parameter estimations result in different routes and schedules. In phase 3, AHP is performed with main decision makers to determine their preferred solution. Final routes and schedules are designed based on model results, AHP results, and contextual input from decision makers. In our case study, an improved workload balance, a 23% reduction in total distance travelled, and buy-in to implement the changes, were obtained.

Fig 1. Questionnaire results about work demands.

Results from the questionnaire about (A) most demanding work on the mobile clinic and (B) least demanding work on the mobile clinic.

Fig 2. Patient and travel data for the month of February 2023.

Data for the month of February 2023 depicting (A) time spent travelling for the three mobiles, (B) the number of patients seen by each mobile, (C) the times the three mobiles left the depot and (D) the times the three mobiles returned to the depot for each day of their schedule.

Fig 3. The day lengths for the current schedule with each set of service times.

The lengths of the days for the current mobile schedule based on service times used in (A) E1, (B) E2, (C) E3 and (D) E4 and total travel time for the day, with the yellow dotted line indicating the maximum length of a working day.

Fig 4. Day lengths for the E1 schedule for all 3 mobiles.

The lengths of the days for the E1 mobile schedule consisting of E1 service times and total travel time for the day, with the yellow dotted line indicating the maximum length of a working day.

Fig 5. Day lengths for the E2 schedule for all 3 mobiles.

The lengths of the days for the E1 mobile schedule consisting of E1 service times and total travel time for the day, with the yellow dotted line indicating the maximum length of a working day.

Fig 6. Day lengths for the E3 schedule for all 3 mobiles.

The lengths of the days for the E3 mobile schedule consisting of E3 service times and total travel time for the day, with the yellow dotted line indicating the maximum length of a working day.

Fig 7. Day lengths for the E4 schedule for all 3 mobiles.

The lengths of the days for the E4 mobile schedule consisting of E4 service times and total travel time for the day, with the yellow dotted line indicating the maximum length of a working day.