(Management Case Study) Development and Implementation of a Novel Standardized Algorithm-Based Automation Optimization Strategy in a Mid-Sized Community Hospital

Purpose: Literature on standardized methodologies for optimizing automated dispensing cabinets (ADC) is sparse and often vague, offering non-specific recommendations and few actionable insights for practical application. At a newly affiliated institution, inefficiencies in ADC optimization and management were identified, characterized by time-consuming, non-streamlined optimization processes resulting in suboptimal inventory turns – a key indicator of inventory management efficiency. This management case study explores these inefficiencies and introduces a novel algorithmic approach to enhance ADC management. It details the methodology adopted to optimize inventory turns, presents quantifiable improvements, and offers key lessons to inform future advancements in ADC optimization strategies.

Methods: To address automation challenges, we engaged with frontline staff and operations pharmacy leadership to assess automation goals. We initiated the project by analyzing standard reports from the ADC system to develop a preliminary algorithm, which we implemented in a pilot area. After a one-month evaluation, the algorithm was refined based on performance metrics and then rolled out across all patient care unit ADCs. Adjustments were data-driven, focusing on optimizing stock levels according to usage patterns and criticality of the medications. The refined algorithm stratified inventory into three cost-based percentile groups: Category A (top 15% by cost), Category B (70th to 85th percentile by cost), and Category C (bottom 70th percentile by cost). The algorithm adjusted periodic automatic replacement (PAR) values, reorder (RO) levels, and critical low (CL) thresholds based on various factors, including medication cost, maximum daily utilization, average daily utilization, and stockout occurrences during the review period. For medications dispensed via patient-specific bins (PSB), the algorithm further assessed whether to add them to the ADCs. Following the initial optimization, a comprehensive review of all ADCs was conducted three months later to fine-tune the PARs, ROs, and CLs. This included retrieving overstocked items, removing unnecessary items, and adjusting PAR levels during restocking processes. Data collection continued for six months post-implementation to monitor outcomes and refine processes.

Results: Overall, maximum PAR levels were reduced by 30.6%. By stratifying medications into cost-based percentile categories—Category A (greater than $29.60 per unit), Category B (between $29.60 and $8.92 per unit), and Category C (below $8.92 per unit)—we achieved a total reduction of $63,610.09 in stored medication costs, a 25% decrease. Of this reduction, Category A medications accounted for 42.5%, Category B for 16%, and Category C for 41.5%. Inventory turns improved markedly, with actual turns increasing from 13.61 to 19.61, a 44.1% increase. This rise illustrates enhanced inventory efficiency and faster utilization. ADC refill dispenses from central pharmacy increased 24.5%, expected with an increase in inventory turns, adding 132 minutes to daily technician workloads (i.e., 10.1 minutes per non-IV technician position per day). This improvement in inventory turns decreased expiration costs by 11.8%, from $24,429.57 to $21,549.79, reflecting reduced waste and improved medication management. The average stockout rate initially rose from 0.39 to 0.83 but decreased to 0.5 after six months. Total doses dispensed remained stable, averaging 166,544 per month pre-implementation and 167,579 post-implementation. Dispensing from ADCs stayed consistent at approximately 95.8% to 96%, while override percentages showed a slight increase from 1.76% to 1.81%.

Conclusion: This novel standardized algorithm for ADC optimization proved to be successful at a mid-sized community hospital. This approach enabled efficient analysis of ADCs in a consistent, data-driven, and actionable manner, reducing inventory on hand and increased inventory turns, without negatively impacting workload or stockout rates. Further algorithm refinement is needed to address the bottom 10% of medications by unit, which, while only accounting for 1% of the cost impact, contributed to 26% of the increase in workload. This promising initiative lays the groundwork for broader application and continuous improvement in pharmacy automation systems, ultimately enhancing operational efficiency and patient care.