Extended SAFPH℞ (Systems Analysis for Formal Pharmaceutical Human Reliability): Two approaches based on extended CREAM and a comparative analysis. (December 2020)

• Record Type: Journal Article
• Authors: Zheng, Xi; Bolton, Matthew L.; Daly, Christopher
• Is Part Of: Safety science. Volume 132(2020)
• Publication Date: 2020-12
• Subjects: Human reliability analysis (HRA) -- Extended cognitive reliability and error analysis method (CREAM) -- Medication errors -- Formal methods -- Probabilistic model checking; Industrial accidents -- Periodicals; Accident Prevention -- Periodicals; Safety -- Periodicals; Travail -- Accidents -- Périodiques; 363.11
• DOI: 10.1016/j.ssci.2020.104944
• Languages: English
• ISSNs: 0925-7535

Abstract

Highlights: Medication errors from community pharmacies are a serious patient safety hazard. Prior work created Systems Analysis for Formal Pharmaceutical Human Reliability. This method (SAFPHR) used stochastic model checking to predict error rate ranges. We extend SAFPHR in two ways to produce point estimates of medication error rates. We apply extended SAFPHR to pharmacy dispensing and analyze the results. Abstract: Medication errors originating in community pharmacies can cause severe harm. To provide pharmacies with the ability to accurately predict error rates, understand why errors are occurring, and mitigate problems, we developed a new human reliability analysis (HRA) called the Systems Analysis for Formal Pharmaceutical Human Reliability (SAFPH℞). Through the combination of HRA that is based on the Cognitive Reliability and Error Analysis Method (CREAM) and probabilistic model checking (an automated method for proving properties about stochastic systems), SAFPH℞ is able to address the limitations of previous HRAs. The previous, "basic" version of SAFPH℞ was based on the "basic" version of CREAM. With this, we predicted a realistic range of medication error rates for a typical US community pharmacy dispensing procedure. However, basic SAFPH℞ was not capable of providing point estimates except through averaging. In this research, we attempted to address this limitation by making SAFPH℞ compatible with the two variations of extended CREAM, enabling SAFPH℞ to make point predictions about pharmacy dispensing error rates. Then, to determine which of the versions of SAFPH℞ produce the most accurate predictions, we compare results from each approach to aggregate rates published in the community pharmacy literature. In this, arithmetic averages across basic SAFPH℞'s range were consistently the most accurate for the overall error rate and rates of errors originating at different stages of the dispensing procedure. We use this finding to derive recommendations from basic SAFPH℞ for improving the reliability of community pharmacy dispensing with the ambition of improving patient health and safety.