Investigating the epidemiology of medication errors in adults using electronic health records in Riyadh, Saudi Arabia
Assiri, Ghadah Asaad M
Introduction: The Harvard Medical Practice (Study II) identified medications as the most common source of injury resulting from medical care. The subsequent Institute Of Medicine (IOM; now renamed as the National Academy of Medicine) report “To Err Is Human: Building a Safer Health System” brought considerable national and international attention to the problem of errors in hospitals. Given that medication errors and error-related adverse drug events (ADEs) are overall much more common in ambulatory care, primary care and home (henceforth collectively referred to as community) settings than in hospital settings, it is also important to focus on these hitherto neglected sectors. There is, however, very limited research on the frequency of medication errors and error-related ADEs in Saudi Arabia’s (SA’s) community settings. Aims: To estimate the incidence and prevalence of medication errors and associated ADEs in community settings, and to identify the risk factors for these outcomes, with an emphasis on those that are potentially modifiable. Methods: I have undertaken a phased programme of work. In Phase 1, I undertook a systematic review of the existing research on the epidemiology of medication errors and error-related ADEs, and their risk factors in community settings. Phase 2 was a feasibility study to identify the ambulatory settings and electronic database, evaluate the feasibility of data extraction and data collection from electronic health records (EHRs) and to check the availability and assess the reliability of key outcome measures. Phase 3 was a pilot, retrospective cohort study using clinically important errors in medicine management that were extracted from EHRs. This third phase also focused on assessing the sample size calculations for undertaking a larger cohort study. A random sample of 200 records was selected; a list of all patients who visited the Family Medicine department two weeks before data collection was generated. Each record was given a code number and EHRs were selected using a random number table that was generated using the ‘simple random sample without replacement’ function in STATA. The final study, Phase 4 was a larger retrospective cohort study to estimate the period prevalence of clinically important errors in medicine management, identify risk factors associated with patients at risk of clinically important errors in medicine management and to compare the estimates from this SA-based study with QRESEARCH analysis of secular trends in the United Kingdom (UK). A random sample of 2000 records was selected using a similar process to Phase 3. All research participants were adults aged ≥18 years. Phases 2-4 were based on the methods used by Avery et al. (2012). Phases 3 and 4 were undertaken in randomly selected samples of 200 and 2000 patients, respectively. Statistical analyses in Phases 3 and 4 were undertaken using STATA (version 14) statistical software. Results: For Phase 1, I identified a total of 15,302 potentially eligible studies, of which 60 met the inclusion criteria: 53 studies focused on medication errors, three on error-related adverse events and four on risk factors only. None of these studies was undertaken in SA. The prevalence of prescribing errors was reported in 46 studies with prevalence estimates ranging widely from 2.0-94.0%. Inappropriate prescribing was the most common type of error reported. Only one study reported the prevalence of monitoring errors, finding that incomplete therapeutic/safety laboratory-test monitoring occurred in 73.0% of patients. The incidence of preventable ADEs was estimated as 15/1000 person-years, the prevalence of drug-drug interaction (DDI)-related adverse drug reactions (ADR) as 7.0% and the prevalence of preventable ADE as 0.4%. A number of patients, healthcare professionals and medicationrelated risk factors were identified, including the number of medications used by the patient, increased patient age (≥75 years), the number of multi-morbidities, the use of anticoagulants, cases where more than one physician was involved in patients’ care and care being provided by family physicians/general practitioners (GP). For Phase 2, I selected the EHRs of King Faisal Specialist Hospital & Research Centre (KFSH&RC) Family Medicine and Polyclinics, Riyadh, SA and the findings confirmed that the pilot study was feasible and likely to yield random samples. More specifically, all information needed for the outcome measures were available in one electronic system and were useable in Phases 3 and 4. In Phase 3, a random sample of 200 records was selected. The overall period prevalence of patients with at least one medication error over 15 months was 10.0% (95% confidence interval (CI) 5.8 to 14.2). The overall period prevalence of medication errors over 15 months was 16.0% (95% CI 8.2 to 23.8). Risk factors that significantly predicted the overall patients at risk of medication errors were patient’s age of ≥65 years and using over-the-counter (OTC) medications. In Phase 4, a random sample of 2000 records was selected using a similar process to Phase 3. The overall period prevalence of patients with at least one medication error over 15 months was 5.85% (95% CI 4.8 to 6.9). The overall period prevalence of medication errors over 15 months was 8.1% (95% CI 6.5 to 9.7). The overall period prevalence estimate of the first 12 clinically important errors in medicine management in the cohort study was more compared to the QRESEARCH analysis of secular trends. This may reflect the different types of healthcare services provided and the different methods of data extraction between both countries. Risk factors that significantly predicted the overall patients at risk of errors were patient’s age of ≥65 years, male gender, Saudi nationality and taking five or more concurrent drugs (polypharmacy). In both Phases 3 and 4, the highest risk of prescribing errors was found to be for ‘Outcome 2a: patients with asthma who had been prescribed a β-blocker’. For monitoring errors, the highest risk was in ‘Outcome 7: patients receiving lithium for at least three months who had not received a recorded check of their lithium concentrations in the previous three months’. Conclusions: This is the first study to investigate medication errors in community settings in SA. This research has revealed that clinically important medication errors are common with a period prevalence estimate of 8.1% and are seen both in relation to the prescribing and monitoring of drugs. Future research should replicate this work in different community contexts in SA and other countries, in order to investigate in greater depth the error-related adverse events and develop and evaluate interventions to decrease clinically important errors in medicine management.