Medication-related adverse events in primary care represent an important cause of hospital admissions and mortality. Adverse events could result from people experiencing adverse drug reactions (not usually preventable) or could be due to medication errors (usually preventable).
To determine the effectiveness of professional, organisational and structural interventions compared to standard care to reduce preventable medication errors by primary healthcare professionals that lead to hospital admissions, emergency department visits, and mortality in adults.
We searched CENTRAL, MEDLINE, Embase, three other databases, and two trial registries on 4 October 2016, together with reference checking, citation searching and contact with study authors to identify additional studies. We also searched several sources of grey literature.
We included randomised trials in which healthcare professionals provided community-based medical services. We also included interventions in outpatient clinics attached to a hospital where people are seen by healthcare professionals but are not admitted to hospital. We only included interventions that aimed to reduce medication errors leading to hospital admissions, emergency department visits, or mortality. We included all participants, irrespective of age, who were prescribed medication by a primary healthcare professional.
Three review authors independently extracted data. Each of the outcomes (hospital admissions, emergency department visits, and mortality), are reported in natural units (i.e. number of participants with an event per total number of participants at follow-up). We presented all outcomes as risk ratios (RRs) with 95% confidence intervals (CIs). We used the GRADE tool to assess the certainty of evidence.
We included 30 studies (169,969 participants) in the review addressing various interventions to prevent medication errors; four studies addressed professional interventions (8266 participants) and 26 studies described organisational interventions (161,703 participants). We did not find any studies addressing structural interventions. Professional interventions included the use of health information technology to identify people at risk of medication problems, computer-generated care suggested and actioned by a physician, electronic notification systems about dose changes, drug interventions and follow-up, and educational interventions on drug use aimed at physicians to improve drug prescriptions. Organisational interventions included medication reviews by pharmacists, nurses or physicians, clinician-led clinics, and home visits by clinicians.There is a great deal of diversity in types of professionals involved and where the studies occurred. However, most (61%) of the interventions were conducted by pharmacists or a combination of pharmacists and medical doctors. The studies took place in many different countries; 65% took place in either the USA or the UK. They all ranged from three months to 4.7 years of follow-up, they all took place in primary care settings such as general practice, outpatients' clinics, patients' homes and aged-care facilities. The participants in the studies were adults taking medications and the interventions were undertaken by healthcare professionals including pharmacists, nurses or physicians. There was also evidence of potential bias in some studies, with only 18 studies reporting adequate concealment of allocation and only 12 studies reporting appropriate protection from contamination, both of which may have influenced the overall effect estimate and the overall pooled estimate. Professional interventionsProfessional interventions probably make little or no difference to the number of hospital admissions (risk ratio (RR) 1.24, 95% confidence interval (CI) 0.79 to 1.96; 2 studies, 3889 participants; moderate-certainty evidence). Professional interventions make little or no difference to the number of participants admitted to hospital (adjusted RR 0.99, 95% CI 0.92 to 1.06; 1 study, 3661 participants; high-certainty evidence). Professional interventions may make little or no difference to the number of emergency department visits (adjusted RR 0.71, 95% CI 0.50 to 1.02; 2 studies, 1067 participants; low-certainty evidence). Professional interventions probably make little or no difference to mortality in the study population (adjusted RR 0.98, 95% CI 0.82 to 1.17; 1 study, 3538 participants; moderate-certainty evidence). Organisational interventionsOverall, it is uncertain whether organisational interventions reduce the number of hospital admissions (adjusted RR 0.85, 95% CI 0.71 to 1.03; 11 studies, 6203 participants; very low-certainty evidence). Overall, organisational interventions may make little difference to the total number of people admitted to hospital in favour of the intervention group compared with the control group (adjusted RR 0.92, 95% CI 0.86 to 0.99; 13 studies, 152,237 participants; low-certainty evidence. Overall, it is uncertain whether organisational interventions reduce the number of emergency department visits in favour of the intervention group compared with the control group (adjusted RR 0.75, 95% CI 0.49 to 1.15; 5 studies, 1819 participants; very low-certainty evidence. Overall, it is uncertain whether organisational interventions reduce mortality in favour of the intervention group (adjusted RR 0.94, 95% CI 0.85 to 1.03; 12 studies, 154,962 participants; very low-certainty evidence.
Based on moderate- and low-certainty evidence, interventions in primary care for reducing preventable medication errors probably make little or no difference to the number of people admitted to hospital or the number of hospitalisations, emergency department visits, or mortality. The variation in heterogeneity in the pooled estimates means that our results should be treated cautiously as the interventions may not have worked consistently across all studies due to differences in how the interventions were provided, background practice, and culture or delivery of the interventions. Larger studies addressing both professional and organisational interventions are needed before evidence-based recommendations can be made. We did not identify any structural interventions and only four studies used professional interventions, and so more work needs to be done with these types of interventions. There is a need for high-quality studies describing the interventions in more detail and testing patient-related outcomes.