Factors influencing the performance and documentation of clinical interventions by Australian community pharmacists

Introduction Drug-related problems (DRPs) are a major burden on healthcare systems. Once detected, the resolution of these DRPs has the potential to reduce healthcare costs and improve patient outcomes. Community pharmacists are ideally placed to detect and prevent DRPs, with the resolution of a DRP being termed a clinical intervention. Aims Utilising an electronic documentation system, the aim of this project was to determine the number and nature of DRPs detected and clinical interventions performed by Australian community pharmacists. The project also aimed to identify the pharmacy and pharmacist factors that influenced the frequency with which clinical interventions were both performed and documented. Methods An electronic documentation system was designed and integrated into the existing dispensing software of 186 pharmacies in three States of Australia (NSW, Victoria and Tasmania) to allow pharmacists to record details about the clinical interventions they performed in order to prevent or resolve DRPs. Participating pharmacies were randomly allocated to three groups: Group One had documentation software; Group Two had documentation software plus a timed reminder to document interventions; and Group Three had documentation software, timed reminder and an electronic decision support prompt. Pharmacists were trained in the use of the software system and also completed several surveys gathering information about demographics, professional attitude, personality traits and clinical knowledge. Pharmacists classified DRPs, entered recommendations they made, and estimated the clinical significance of the intervention. An observational sub-study, which included 24 pharmacies without any documentation software, was also completed to determine current practice. Results Over 12 weeks, 531 participating pharmacists dispensed 2,013,923 prescriptions for 483,147 patients and documented 6,230 interventions, resulting in a median intervention rate of 2.4 interventions in 1000 prescriptions or 0.24%. Of these 6,230 interventions, 282 were attributed to the electronic prompt in Group Three and were removed prior to analysis. No significant differences were seen in the overall intervention rate between the three groups, however the presence of the prompt in Group Three significantly increased the number of interventions performed on the prompted medications. As expected, the ‘software’ pharmacies had a significantly higher documentation rate compared to the ‘no software’ pharmacies. There was a significant decline in the number of interventions documented over the trial period. Commonly, pharmacists’ interventions were related to drug selection problems (30.7%) and educational issues (23.7%). Recommendations were often related to a change in therapy (40.1%), such as a change of drug or dose, or provision of information (34.7%). When a referral recommendation was made, this was almost uniformly to the prescriber (91.3%). Nearly half of the interventions (42.6%) were classified as having a higher clinical significance by the documenting pharmacists, with these interventions most commonly associated with undertreatment or toxicity DRPs. Drug groups most commonly subject to an intervention included antibiotics, glucocorticoids, and opioids. The antibiotics were commonly associated with DRPs due to allergies, incorrect doses and drug interactions, with the glucocorticoids and opioids often associated with dosing issues. An independent expert panel of 23 experts (5 physicians, 10 GPs and 8 pharmacists) was commissioned to assess the economic value of a sample of 200 interventions. The pharmacist’s assessment of the clinical significance appeared to correlate well with the economic value (p < 0.001), showing that the more clinically significant the pharmacist thought the intervention was, the higher the cost saving to the Australian government. Original prescriptions were associated with significantly more interventions than repeat prescriptions (p < 0.001), most likely due to original prescriptions having a higher incidence of drug selection errors, drug interactions and education requirements compared to repeat prescriptions. As expected, more interventions were performed on older patients (p < 0.001), most likely due to the larger number of medications they were taking. Analysis of the observational sub-study revealed that only 49% of performed interventions were documented within the electronic software system, suggesting that the number of interventions performed may actually be twice the number documented within the software. Multiple regression analysis was used to produce a model to predict the pharmacy’s intervention rate. Two models were created, the ‘pharmacist workload’ model and the ‘prescription volume’ model, however, both model fits were poor and could only explain 10.1% and 11.8% of the variance, respectively. The ‘pharmacist workload’ model had three significant factors: high pharmacist workload; annual financial turnover; and, whether the pharmacy catered for aged care facilities. Pharmacies that had higher pharmacist workload, a higher financial turnover and catered for aged care facilities tended to have lower intervention rates. The ‘prescription volume’ model had five significant factors: high prescription volume; moderate prescription volume; annual financial turnover; location in or near a medical centre; and, participation in other pharmacy trials. Pharmacies with a higher prescription volume, a higher financial turnover and concurrent participation in other trials tended to have lower intervention rates on average, whilst medical centre pharmacies tended to have higher intervention rates on average. Despite the poor model fit, these factors would logically have a significant impact on the pharmacist’s workload, indicating that the busier the pharmacy and pharmacists are, the lower the intervention rate is likely to be as there would be less time to perform and document clinical interventions. This theory was also supported by the bivariate analyses which showed that the intervention rate of the pharmacy was significantly correlated with the workload during the trial, with the busier pharmacies having a lower intervention rate. The observational sub-study also identified workload as a key factor that influenced the pharmacy’s intervention rate. A separate analysis was performed on the individual pharmacist data. The logistic regression model was 65.8% successful in predicting whether a pharmacist would have a high intervention rate using four variables: average number of continuing professional development (CPD) hours completed per year; level of software training; clinical knowledge score; and professional attitude score. The pharmacists who completed more CPD hours per year and who had a higher clinical knowledge score, higher level of training and a more positive professional attitude tended to have higher intervention rates. Conclusions Use of the software, including its electronic prompts, significantly increased the documentation of clinical interventions by pharmacists. Professional development strategies and policies which foster improvements to pharmacy workload and pharmacist clinical knowledge can be expected to further improve pharmacists’ clinical intervention rate, and therefore decrease the healthcare costs associated with DRPs