Value of perfect information in home care human resource planning with continuity of care

Home Care (HC) includes medical, paramedical and social services delivered to patients at their domicile. Many resources are involved and the service is delivered in a usually vast territory. Moreover HC providers usually preserve the continuity of care: only an operator for each category follows the whole patient care pathway providing all the visits pertinent to his category. High randomness affects the service delivery, mainly in terms of unexpected changes in patient conditions, making the activity amount highly uncertain and mining the feasibility of plans. Therefore a reliable resource planning, including the estimation of health progression of the assisted patients, is crucial for HC organizations to respect the continuity of care, while avoiding poor quality service level, treatment delays and process inefficiencies. The main purpose is to avoid overloaded operators with consequent organization inefficiencies and operator problems. Aim of this paper is evaluating the value of the perfect information in this balancing problem characterized by high variability and uncertainty of data. A mixed integer linear programming (MILP) algorithm for balancing the workload among the operators of a specific category (physicians, nurses,...) was implemented, satisfying the requests of all the assisted patients and preserving the continuity of care. The peculiarity of the continuity of care makes this algorithm different from the classical literature models. Inputs are patients and operators data. The requests of patients along the time (number of visits provided by the considered operator category) are estimated by means of a stochastic model, developed in a previous work, or directly determined by historical data. This allowed the comparison between the solution under expected, simulated or real data. Patients are divided into already assigned and new arrivals. The first class has the assigned operator which is not changed, the latter must be assigned to an appropriate operator. New patients are assigned considering the operator skills (types of patients he can take care) and the territorial distribution of patients (each operator can only care patients resident in a specific territory); therefore the operators are divided into subgroups. Operators are characterized in terms of availability (maximum number of weekly visits under optimal work conditions). The MILP algorithm allows to balance the ratio between the weekly workload and the availability among the operators, considering a certain number of following weeks. The algorithm was applied to one the largest Italian public HC providers. It was run over a period of 26 weeks, with patient data coming from the stochastic model, the real scenario and a set of simulated scenarios. The capability of balancing the workload among the operators was firstly verified. Then the value of perfect information on future patient requests was estimated by comparing the expected value solution with the wait and see solution, calculated by assuming to know in advance the patient requests generated in each scenario. The results show that a significant higher balancing is obtained using perfect information. Therefore stochastic algorithms for workload balancing can improve the HC human resource planning.