A nurse rostering algorithm for a district hospital in South Africa
Treutnicht, M. J.
Lane-Visser, T. E.
Van Dyk, L.
Friedrich, S. S.
MetadataShow full item record
ENGLISH ABSTRACT: An acute shortage of healthcare professionals is the rule rather than the exception in South Africa. Effective scheduling of nurses is critical to ensure good quality of care, while limiting sta related healthcare costs and abiding by labour laws. South African district hospital nurses are presently scheduled through the manual production of duty rosters on a monthly basis. In this paper, two related nurse rostering problems (NRP) are formulated for a district level public hospital in Stellenbosch (South Africa). The first problem addresses the scheduling of the months that nurses are on night shift duty. The other problem addresses the scheduling of the days that nurses are working night or day shifts within a month, respectively. A hierarchy of four levels exists among the nursing staff at Stellenbosch Hospital. Distributed over the four levels, the nursing staff totals ninety employees. Due to hospital policy, no casual nurses are employed. Fluctuations in demand are met by scheduling overtime shifts that are limited by current labour legislation. The hospital consists of seven wards, each with separate staff requirements. The NRP for Stellenbosch Hospital is solved using the genetic algorithm. The algorithm is adapted to speci cally adhere to the requirements and constraints given by the formulated NRP. The algorithm outputs optimal feasible rosters for each problem and provides data required to evaluate the performance of the algorithm. Roster results are interpreted and verified using the initial nursing requirements of the hospital. The robustness of using such an algorithm for sustainable use is also discussed. The paper ultimately aims to promote the use of operations research in healthcare on a practical level in South Africa.
Please cite this item using this persistent URLhttp://hdl.handle.net/10019.1/20465
The following license files are associated with this item: