| Article ID: | iaor2007272 |
| Country: | Netherlands |
| Volume: | 9 |
| Issue: | 1 |
| Start Page Number: | 71 |
| End Page Number: | 85 |
| Publication Date: | Feb 2006 |
| Journal: | Health Care Management Science |
| Authors: | Bard Jonathan F., Purnomo Hadi W. |
| Keywords: | programming: integer, personnel & manpower planning |
In many service organizations, rosters must be constructed weekly or monthly as demand and available personnel change. Once the permanent workforce is fixed, it may not be possible to alter its composition easily, implying that expensive contract labor may be the only option to cover shortages. With respect to nursing resources, this means calling in part-timers, casuals, or agency nurses on a daily basis, or hiring travelers for up to several months at a time. This paper addresses the latter option and presents two models that can be used to solve what we call the nurse addition problem. The first was originally developed to solve the midterm preference scheduling problem and is based on a pattern-view formulation. The second is derived from a shift-view formulation and is solved with a branch-and-price algorithm. In either case, the objective is to hire up to some predetermined number of nurses and assign them midterm schedules that minimize the maximum amount of uncovered shifts per day in the planning horizon. Each roster selected for a new nurse must satisfy a set of hard constraints related to the total working hours, workstretches, time between shifts, and weekend requirements, and a set of soft constraints related to days-on and days-off patterns and transitions from one shift type to another. Extensive testing with data provided by a 400-bed hospital indicated that most instances could be solved in a matter of minutes.