Improving the forecasting of hospital services: A comparison between projections and actual utilization of hospital services

Objectives: To compare projected and observed hospital inpatient use in Belgium and to draw lessons from that comparison. Methods: In 2005, projections for hospital service use were generated up to 2015, based on demographic change, substitution from inpatient to day care, and, the evolution of the average length of stay (LOS). The accuracy of the forecasts was assessed by comparing projected and observed population size, admissions and inpatient days, average LOS and percentage change in case mix. Results: The demographic growth was underestimated. Overall, the baseline projection for hospital admissions was remarkably close to the observed figures but the underlying case mix diverged importantly. With substitution between inpatient and day care, the number of admissions was underestimated by 15%-40%. The number of days was projected to increase in every scenario, whereas a decreasing trend was observed mainly due to the faster decline in average LOS than projected. Conclusion: Hospital capacity planning is an important component of evidence informed policymaking. Projection results benefit from a well-designed methodology: choice of forecast groups, estimation models, selection criteria, and a sensitivity analysis of the results. To cope with the dynamic and continuously evolving context in which hospitals operate, regular updates to incorporate new data and to reassess estimated trends should be an integral part of the projection framework. (C) 2018 Elsevier B.V. All rights reserved.