Pooled demand management in Ethiopian Public Hospital Supply Chains: Practices and Issues

Abstract

Introduction: Predicting future trends provides additional value for improved healthcare system management in today’s global business trend and step forward technologies. After all, the healthcare system is going to undergo a huge data revolution, with Artificial Intelligence (AI), predictive analytics, and business intelligence ready to increase efficiency and enhance health outcomes. Thus, developing data analytics mechanisms and capabilities play a crucial role for successful implementation of a committed demand program. Limited health supply chain analytics practice, low supply chain planning performance and frequent disruptions in managing vital pharmaceuticals of committed demand programs reducing the health outcomes of patient’s public health facilities in Ethiopia. Objective: to provide new insights on demand planning practices of vital pharmaceuticals under the committed demand program in 13 federal and university hospitals in Ethiopia. Method: A quantitative, descriptive, and explanatory systematic analysis of 46 vital pharmaceuticals consumed in 13 federal and university hospitals across the country as part of the committed demand program between 2017–2022. Predictive modeling approach used to foresee future occurrences or outcomes, as well as to predict future trends, by searching for patterns that have occurred in the past or by analyzing historical data. Key informant interview was done to identify challenges. Lewis MAPE scale of judgment for forecast errors was used to determine the forecast accuracy. Result: The results of the study showed that there was considerable variation in the number of SKUs issued and the cost of issuance over the years. Specifically, 2019 had the highest number of SKUs issued, while 2018 had the highest cost incurred. The top five pharmaceutical items issued were Sodium Chloride (Normal Saline) 0.009 infusion, Ceftriaxone 1gm injection, Vancomycin 1gm infusion, Carbamazepine 200mg tablet, and Anti-Rho (D) Immune Globulin 300mcg in 2ml injection, which accounted for a significant proportion of the total quantity and cost issued. These findings suggest that certain pharmaceutical items are more commonly issued than others, and that the dosage form and cost can vary considerably between items. The study also found that only 14 out of the 46 pharmaceuticals had a MAPE value less than 50%, which is considered accurate according to the Lewis MAPE scale. Furthermore, the 2, 3, and 4-year moving averages showed that the number of pharmaceuticals with a MAPE less than 50% varied between 14 and 17, indicating that accuracy in forecasting SKUs may be difficult to achieve consistently over time. Conclusion: Overall, this study provides valuable insights into the patterns and challenges of pharmaceutical SKU forecasting and management. These findings can inform healthcare organizations in their efforts to improve inventory management and reduce unnecessary costs. Further research is needed to develop more effective forecasting methods for pharmaceutical SKUs and to identify factors that contribute to the variability in SKU issuance and cost.