Prevalence and potential clinical significance of near miss dispensing errors at a large teaching hospital in the United Kingdom (UK)

Abstract

Abstract Introduction Dispensing is a complex multi-step process where mistakes can arise at any stage leading to a potential to cause patient harm. Published literature identifies near miss dispensing error rates up to 2.7% in UK hospitals.1 Near misses are ‘a dispensing error detected by the checker before it reaches the patient’. Near miss audits are routinely undertaken across two main dispensaries at this teaching hospital where one dispensary (site 1) is automated and the other (site 2) is not. Aim To determine the frequency of near miss dispensing errors, by site, and review the potential clinical significance of near misses observed. Methods Locally adapted data collection tool based on Royal Pharmaceutical Society near miss error codes2 was developed and piloted. Details on prescription type, drug, dose, strength, route, formulation and near misses were included. Details of all medication orders dispensed were also recorded. Prospective observational audit of near misses identified at the checking bench was undertaken, on three separate days, across three consecutive months. Dean and Barber method3 was used for assessing potential clinical significance of near misses. Four independent assessors: two pharmacists; one nurse and one doctor; reviewed near misses for the likelihood to cause harm. The mean severity score attained across assessors was calculated. Approval was obtained by the Trust Pharmacy Research and Audit Committee. The need for ethical submission was waived. Results Overall 3027 items were dispensed; 1539 and 1488 at sites 1 and 2 respectively. There were 177 (5.8%) erroneous dispensed items involving 193 near misses in total (15 items had two near misses and one item had three near misses). Ninety one (5.9%) erroneous dispensed items were captured at site 1 and 86 (5.8%) at site 2 (χ 2, p=0.94). Overall 161 near miss descriptions were assessed for clinical significance: 97 (60.2%) minor, 63 (39.1%) moderate and one (0.6%) severe. Statistically significant difference in severity rating of near misses between prescription type (χ 2(2) = 32.268, p <0.001); mean ranks 80, 57, 125 for discharge, inpatient and outpatient prescriptions respectively. No statistical difference in severity rating of near misses between error type (χ 2(2) = 2.402, p =0.3). Discussion/Conclusion Local prevalence of near misses is higher than in published literature.1 However, the majority of errors were considered to have minor clinical impact on patients. Difficult to make direct comparisons between studies due to differences in research methods, definitions, operating systems and hospital settings. There was no statistical difference noted in prevalence between sites despite presence of automation at one and manual dispensing at the other. Two factors may explain this: Firstly, part-pack robotic dispensing, where generation of medication barcodes is a manual process and one subject to human input error itself; although not explored explicitly as part of this study. Secondly, lack of knowledge and complex prescriptions are known to be key contributory factors associated with dispensing errors, but the categories of drugs dispensed at a specialist tertiary dispensary (site 1) were broader and more complex. Further study is needed on the impact of part-pack robotic dispensing on dispensing errors.