The impact of automation on pharmacy staff experience of workplace stressors.

OBJECTIVE: Determine the effect of installing an original pack automated dispensing system (ADS) on staff experience of occupational stressors. METHODS: Pharmacy staff in a National Health Service hospital in Wales, UK, were administered an anonymous occupational stressor questionnaire pre- (n = 45) and post-automation (n = 32). Survey responses pre- and post-automation were compared using Mann-Whitney U test. Statistical significance was P ≤ 0.05. Four focus groups were conducted (two groups of accredited checking technicians (ACTs) (group 1: n = 4; group 2: n = 6), one group of pharmacists (n = 17), and one group of technicians (n = 4) post-automation to explore staff experiences of occupational stressors. Focus group transcripts were analysed according to framework analysis. KEY FINDINGS: Survey response rate pre-automation was 78% (n = 35) and 49% (n = 16) post-automation. Automation had a positive impact on staff experience of stress (P = 0.023), illogical workload allocation (P = 0.004) and work-life balance (P = 0.05). All focus-group participants reported that automation had created a spacious working environment. Pharmacists and ACTs reported that automation had enabled the expansion of their roles. Technicians felt like 'production-line workers.' Robot malfunction was a source of stress. CONCLUSION: The findings suggest that automation had a positive impact on staff experience of stressors, improving working conditions and workload. Technicians reported that ADS devalued their skills. When installing ADS, pharmacy managers must consider the impact of automation on staff. Strategies to reduce stressors associated with automation include rotating staff activities and role expansions.

The impact of automation on workload and dispensing errors in a hospital pharmacy.

OBJECTIVES: To determine the effect of installing an original-pack automated dispensing system (ADS) on dispensary workload and prevented dispensing incidents in a hospital pharmacy. METHODS: Data on dispensary workload and prevented dispensing incidents, defined as dispensing errors detected and reported before medication had left the pharmacy, were collected over 6 weeks at a National Health Service hospital in Wales before and after the installation of an ADS. Workload was measured by non-participant observation using the event recording technique. Prevented dispensing incidents were self-reported by pharmacy staff on standardised forms. Median workloads (measured as items dispensed/person/hour) were compared using Mann-Whitney U tests and rate of prevented dispensing incidents were compared using Chi-square test. Spearman's rank correlation was used to examine the association between workload and prevented dispensing incidents. A P value of ≤0.05 was considered statistically significant. KEY FINDINGS: Median dispensary workload was significantly lower pre-automation (9.20 items/person/h) compared to post-automation (13.17 items/person/h, P < 0.001). Rate of prevented dispensing incidents was significantly lower post-automation (0.28%) than pre-automation (0.64%, P < 0.0001) but there was no difference (P = 0.277) between the types of dispensing incidents. A positive association existed between workload and prevented dispensing incidents both pre- (ρ = 0.13, P = 0.015) and post-automation (ρ = 0.23, P < 0.001). Dispensing incidents were found to occur during prolonged periods of moderate workload or after a busy period. CONCLUSION: Study findings suggest that automation improves dispensing efficiency and reduces the rate of prevented dispensing incidents. It is proposed that prevented dispensing incidents frequently occurred during periods of high workload due to involuntary automaticity. Prevented dispensing incidents occurring after a busy period were attributed to staff experiencing fatigue after-effects.

Exploratory study to identify the process used by pharmacy staff to verify the accuracy of dispensed medicines.

OBJECTIVES: To determine the common stages and strategies involved in the dispensing accuracy-checking process used by pharmacy staff and to determine the training activities used by these staff to gain the knowledge and skills for accuracy checking. METHOD: Face-to-face tape-recorded ethnographic interviews (n = 28) were undertaken in 2009-2010 at two large teaching hospitals with a purposive sample of pharmacists and accredited checking technicians qualified to undertake the final accuracy check on dispensed medicines. Participants described their accuracy-checking process, strategies used to aid checking using anonymised prescriptions and accurate dispensing of medicines to aid discussion. The range of training activities undertaken to develop this skill were discussed. Qualitative data were analysed in accordance with the principles of grounded theory to identify themes. KEY FINDINGS: The accuracy-checking process was described as a cognitive and systematic process. The order in which accuracy checking was executed was found to follow two pathways, with all participants checking the prescription first before verifying either the label or dispensed product. Various physical and sensory aids were used to assist in this verification process. There were inconsistencies in the level of accuracy-checking training received by pharmacists and accredited checking technicians, with many pharmacists reporting no training. CONCLUSION: Although an important medication-error prevention strategy, until this study little was known about the process used by pharmacy staff when verifying the accuracy of dispensed medicines. Accuracy checking is a complex cognitive task involving verification of the product and label with the prescription. Strategies obtained during past experience and in training were used to aid checking. The study highlighted that pharmacy staff training to undertake this task was variable. Application of strategies identified in this study may allow individuals to adopt further safeguards to improve patient safety.

Prediction of 3-hydroxypyridin-4-one (HPO) log K1 values for Fe(III).

As a means to aid in the design of 3-hydroxypyridin-4-ones (HPOs) intended for use as therapeutic Fe(3+) chelating agents, a novel methodology has been developed using quantum mechanical (QM) calculations for predicting the iron binding affinities of the compounds (more specifically, their log K(1) values). The reported/measured HPO log K(1) values were verified through their correlation with the corresponding sum of the compounds' ligating group pK(a) values. Using a training set of eleven HPOs with known log K(1) values, reliable predictions are shown to be obtained with QM calculations using the B3LYP/6-31+G(d)/CPCM model chemistry (with Bondi radii, and water as solvent). With this methodology, the observed log K(1) values for the training set compounds are closely matched by the predicted values, with the correlation between the observed and predicted values giving r(2) = 0.9. Predictions subsequently made by this method for a test set of 42 HPOs of known log K(1) values gave predicted values accurate to within ±0.32 log units. In order to further investigate the predictive power of the method, four novel HPOs were synthesised and their log K(1) values were determined experimentally. Comparison of these predicted log K(1) values against the measured values gave absolute deviations of 0.22 (13.87 vs. 14.09), 0.02 (14.31 vs. 14.29), 0.12 (14.62 vs. 14.50), and 0.13 (15.04 vs. 15.17). The prediction methodology reported here is the first to be provided for predicting the absolute log K(1) values of iron-chelating agents in the absence of pK(a) values.

Prediction of 3-hydroxypyridin-4-one (HPO) hydroxyl pKa values.

As an aid in optimising the design of 3-hydroxypyridin-4-ones (HPOs) intended for use as therapeutic Fe(3+) chelating agents, various quantum mechanical (QM) and semi-empirical (QSAR) methods have been explored for predicting the pK(a) values of the hydroxyl groups in these compounds. Using a training set of 15 HPOs with known hydroxyl pK(a) values, reliable predictions are shown to be obtained with QM calculations using the B3LYP/6-31+G(d)/CPCM model chemistry (with Pauling radii, and water as solvent). With this methodology, the observed hydroxyl pK(a) values for the training set compound are closely matched by the predicted pK(a) values, with the correlation between the observed and predicted values giving r(2) = 0.98. Predictions subsequently made by this method for a test set of 48 HPOs of known hydroxyl pK(a) values (11 of which were determined experimentally in this study), gave predicted pK(a) values accurate to within ±0.2 log units. In order to further investigate the predictive power of the method, two novel HPOs were synthesised and their hydroxyl pK(a) values were determined experimentally. Comparison of these predicted pK(a) values against the measured values gave absolute deviations of 0.13 (10.18 vs. 10.31) and 0.43 (5.58 vs. 5.15).

Prediction of Absolute Hydroxyl pKa Values for 3-Hydroxypyridin-4-ones.

pKa values have been calculated for a series of 3-hydroxypyridin-4-one (HPO) chelators in aqueous solution using coordination constrained ab initio molecular dynamics (AIMD) in combination with thermodynamic integration. This dynamics-based methodology in which the solvent is treated explicitly at the ab initio level has been compared with more commonly used simple, static, approaches. Comparison with experimental numbers has confirmed that the AIMD-based approach predicts the correct trend in the pKa values and produces the lowest average error (∼0.3 pKa units). The corresponding pKa predictions made via static quantum mechanical calculations overestimate the pKa values by 0.3-7 pKa units, with the extent of error dependent on the choice of thermodynamic cycle employed. The use of simple quantitative structure property relationship methods gives prediction errors of 0.3-1 pKa units, with some values overestimated and some underestimated. Beyond merely calculating pKa values, the AIMD simulations provide valuable additional insight into the atomistic details of the proton transfer mechanism and the solvation structure and dynamics at all stages of the reaction. For all HPOs studied, it is seen that proton transfer takes place along a chain of three H2O molecules, although direct hydrogen bonds are seen to form transiently. Analysis of the solvation structure before and after the proton transfer event using radial pair distribution functions and integrated number densities suggests that the trends in the pKa values correlate with the strength of the hydrogen bond and the average number of solvent molecules in the vicinity of the donor oxygen.

Measuring dispensary workload: a comparison of the event recording and direct time techniques.

OBJECTIVES: To compare dispensary workload, determined using the Welsh benchmarking event recording technique and the direct time technique, at two district general UK National Health Service hospitals within different university local health Boards (hospital A--manual dispensing system; hospital B--automated dispensing system). METHODS: Data on dispensary workload were collected, over a period of 6 weeks (hospital A: 8 May-18 June 2007; hospital B: 1 October-11 November 2007), by a non-participant observer using two simultaneous methods of workload measurement: direct time and event recording. Direct time technique involved timing each task involved in dispensing a sample of prescriptions from receipt to issue of dispensed medicines to patients. Welsh benchmarking event recording involved continuously logging staff activities that deviated from the dispensary rota on a data collection form to enable calculation of total staff time involved in dispensing activities. Data on number of items dispensed were obtained from the pharmacy computer system and also by manual counting of prescription items. The mean dispensary workloads were calculated as the number of items dispensed per person per hour. Two-sample t-tests were used to compare dispensary workload measurements determined using direct time and event recording technique reported by each individual hospital. Mean workloads for hospitals A and B were compared using a two-sample t-test. Statistical significance was taken as P≤0.05. KEY FINDINGS: Hospital A was associated with a lower workload (direct time: 7.27±7.16 items per person per hour; event recording: 9.57±10.6 items per person per hour). In contrast, hospital B gave a higher workload (direct time: 11.93±8.3 items per person per hour; event recording: 12.6±8.80 items per person per hour). There was a significant difference between workload (direct time: P<0.01; event recording: P<0.01) reported for both hospitals. The direct time and event recording techniques produced consistent results at each hospital (hospital A: t=0.02, P=0.99; hospital B: t=0.004, P=0.1). CONCLUSION: The direct time and Welsh benchmarking event recording techniques produced consistent results at both hospitals. Thus the Welsh benchmarking event recording technique is a valid and reproducible method of measuring dispensary workload. Hospital B (automated) had a higher workload than hospital A (manual). Further work is required to investigate the impact of automation on dispensary workload.

Unprevented or prevented dispensing incidents: which outcome to use in dispensing error research?

OBJECTIVES: To compare the rate, error type, causes and clinical significance of unprevented and prevented dispensing incidents reported by Welsh National Health Service (NHS) hospital pharmacies. METHODS: Details of all unprevented and prevented dispensing incidents occurring over 3 months (September-December 2005) at five district general hospitals across Wales were reported and analysed using a validated method. Rates of unprevented and prevented dispensing incidents were compared using Mann-Whitney U test. Reported error types, contributory factors and clinical significance of unprevented and prevented incidents were compared using Fisher's exact test. KEY FINDINGS: Thirty-five unprevented and 291 prevented dispensing incidents were reported amongst 221,670 items. The rate of unprevented (16/100,000 items) and prevented dispensing incidents (131/100,000 items; P = 0.04) was significantly different. There was a significant difference in the proportions of prevented and unprevented dispensing incidents involving the wrong directions/warnings on the label (prevented, n = 100, 29%; unprevented, n = 4, 10%; P = 0.02) and the wrong drug details on the label (prevented, n = 15, 4%; unprevented, n = 6, 14%; P = 0.01). There was a significant difference in the proportions of prevented and unprevented dispensing incidents involving supply of the wrong strength (prevented, n = 46, 14%; unprevented, n = 2, 5%; P = 0.02) and issue of expired medicines (prevented, n = 3, 1%; unprevented, n = 5, 12%; P = 0.002). CONCLUSION: The use of prevented dispensing incidents as a surrogate marker for unprevented incidents is questionable. There were significant differences between unprevented and prevented dispensing incidents in terms of rate and error types. This is consistent with the medication error iceberg. Care must be exercised when extrapolating prevented dispensing incident data on error types to unprevented dispensing incidents.

Incidence, type and causes of dispensing errors: a review of the literature.

OBJECTIVES: To identify, review and evaluate the published literature on the incidence, type and causes of dispensing errors in community and hospital pharmacy. METHOD: Electronic databases were searched from 1966 to February 2008. This was supplemented by hand-searching the bibliographies of retrieved articles. Analysis of the findings explored the research methods, operational definitions, incidence, type and causes of dispensing errors. KEY FINDINGS: Sixty papers were identified investigating dispensing errors in the UK, US, Australia, Spain and Brazil. In general, the incidence of dispensing errors varied depending on the study setting, dispensing system, research method and operational definitions. The most common dispensing errors identified by community and hospital pharmacies were dispensing the wrong drug, strength, form or quantity, or labelling medication with the incorrect directions. Factors subjectively reported as contributing to dispensing errors were look-alike, sound-alike drugs, low staffing and computer software. High workload, interruptions, distractions and inadequate lighting were objectively shown to increase the occurrence of dispensing errors. CONCLUSIONS: Comparison of the reviewed studies was confounded by differences in study setting, research method and operational definitions for dispensing errors, error rate and classification of error types. The World Health Organization is currently developing global patient safety taxonomy. Such a standardized taxonomy for dispensing errors would facilitate consistent data collection and assist the development of error-reduction strategies.