Logistics Efficiency Improvement and Waste Reduction using the Appropriate Forecasting Techniques Analysis for Hospital Pharmaceutical Demand Forecasting Error Reduction

Abstract

Effective waste management is guided by the 7R framework (Refuse, Reduce, Reuse, Repair, Repurpose, Recycle, and Recover). This study emphasizes the first two most important principles (Refuse and Reduce) by improving pharmaceutical demand forecasting to reduce waste and enhance logistics efficiency. The objective of this research was to analyze the appropriate forecasting techniques for the case study hospital important pharmaceutical demand forecasting error reduction. Using ABC analysis, 219 products were classified, with Group A items (70.95% of total inventory value) selected for analysis. These were further categorized into cyclical/seasonal demand (31 SKUs) and demand without seasonality (188 SKUs). Randomized Complete Block Design (RCBD) was applied in the Design of Experiments (DOE) using the Analysis of Variance (ANOVA) and multiple comparisons test for the appropriate forecasting techniques analysis in order to reduce the important pharmaceutical demand forecasting error. The forecasting technique was the main factor and Mean Absolute Deviation (MAD) served as the response variable. According to the class A cyclical/seasonal demand pharmaceutical products, the most appropriate forecasting technique was the 12-month seasonal length Winters’ method. The average of MAD obtained by the yearly seasonal length Winters’ method decreased by 7.04 units per month comparing to the 3-month moving average which was the current forecasting method because of the seasonality of pharmaceutical demand. For the class A drugs without seasonality, the most appropriate forecasting technique was single exponential smoothing. The MAD of single exponential smoothing decreased by 38.47 units per month comparing to the 3-month moving average which was the as-is forecasting method of the case study hospital. It can be concluded that Winters’ method with 12-month seasonal length was suitable for cyclical/seasonal demand drugs, reducing MAD by 11% compared to the traditional 3-month moving average. For pharmaceutical demand without seasonality, single exponential smoothing was the most appropriate forecasting method, reducing MAD by 17.5%. The findings demonstrated that selecting appropriate forecasting methods could significantly improve logistics efficiency, reduce pharmaceutical waste, and enhance hospital supply chain performance.