In this post, we’re offering summaries and comments on articles from BMJ Quality & Safety’s Top 10 of 2020. To check out the full list of our 20 finalists, click here. The editors and the Editorial Board used data such as citation rates and social media engagement in addition to their own judgement to create these lists. Overall, they demonstrate the wonderful breadth of valuable articles published in BMJ Quality & Safety.
- Impact of providing patients access to electronic health records on quality and safety of care: a systematic review and meta-analysis – AL Neves (@ana_luisa_neves) et al. BMJ Quality & Safety 2020;29:1019-1032. Accompanying editorial available here.
In identifying 20 randomised studies with > 17,000 patients, Neves et al. sought to systematically review the effect of providing patients access to electronic health records (EHRs) on the six Institute of Medicine domains of healthcare quality: patient-centeredness, effectiveness, efficiency, safety, timeliness, and equity. Across the studies, the degree of EHR access granted to patients varied substantially, with the majority allowing patients to view test results and securely message clinicians, but far fewer allowing patients to review clinical documentation. EHR access had the greatest impact in the ‘effectiveness’ domain. Three other domains showed more mixed results. Within the ‘safety’ domain, reporting of medication discrepancies to providers significantly increased with access to medication lists, while overall medication adherence did not. ‘Patient-centeredness’ results such as reported patient satisfaction scores, patient activation scales, and self-efficacy measures improved in only ~50% of high-quality studies. Similarly, ‘efficiency,’ as noted by a reduction in healthcare utilisation, improved in some studies but remained unchanged in others. No measures related to the ‘timeliness’ and ‘equity’ domains were directly addressed as primary outcomes. Ultimately, efforts are needed to offer more complete EHR access to patients and to better study the effects of such access on domains outside effectiveness. In particular, equity must be addressed – both as a disparity in providing patient access and as an outcome.
- ‘Immunising’ physicians against availability bias in diagnostic reasoning: a randomised controlled experiment by S Mamede et al. BMJ Quality & Safety 2020;29:550-559. Accompanying editorial available here.
Errors in clinical reasoning may be present in up to 75% of diagnostic error scenarios. Mamede et al. studied a simulation intervention aimed at reducing ‘availability bias’ by improving content knowledge for gastrointestinal diseases in 91 second-year residents in Brazil. This approach differs from previous work by focusing on a knowledge-oriented intervention rather than on reducing bias by prompting recognition of bias-inducing cues. The authors randomised the residents to learn from cases for a set of conditions related to either chronic diarrhea or jaundice by actively comparing and contrasting alternative diagnoses (the ‘immunisation’ phase), before having them return one week later to review either a case of acute viral hepatitis or a case of inflammatory bowel disease (the biasing phase). Immediately thereafter, the residents reviewed three vignettes resembling inflammatory bowel disease and three resembling acute viral hepatitis (the testing phase). Some residents, therefore, were ‘immunised’ against the case presented during the biasing phase, while others were not. Overall, after encountering one case in the biasing phase, non-immunised physicians provided that same biasing diagnosis incorrectly for similar but different diseases in the testing phase twice as often as those immunised against that case. This reduced diagnostic accuracy by 40% in the non-immunised group relative to the immunised group. This novel approach illustrates how a knowledge intervention may reduce availability bias, although such interventions may be limited to distinguishing diseases with very similar clinical presentations.
- User-testing guidelines to improve the safety of intravenous medicines administration: a randomised in situ simulation study by MD Jones (@MatthewJonesUoB) et al. Published Online First: 30 June 2020. Accompanying editorial available here.
More than one-third of intravenous (IV) medication doses may be associated with an error in preparation or administration completion, prompting efforts such as the creation of the National Health Service’s Injectable Medicines Guide (IMG). However, the IMG may include ambiguous information or be difficult to navigate for staff administering meds in the midst of busy clinical shifts. In this simulation study, Jones et al. randomized 273 practicing nurses and midwives to prepare and administer a medication infusion of voriconazole (blinded using the name ‘bathicillin’) using either the standard IMG or a version that had been revised by iterative user-testing to promote comprehension. Errors were identified using observation. Overall, 98% of simulations had at least one error, with 49% of user-tested IMG simulations and 59% of standard IMG simulations having at least one IMG-related error categorized as moderate to severe (RR 0.82, 95% CI 0.66-1.02). Significantly more user-tested simulations had zero IMG-related errors (48% vs. 20%, RR 2.46, 95% CI 1.68-3.60). User-tested simulations were also significantly faster and more likely to be fully completed than those using only the original IMG guidance. Subjectively, users were more confident using the user-tested IMG and preferred this version after viewing both options at the conclusion of their simulation. When analysed further, dose and rate discrepancies were the types of errors most reduced by user-testing, suggesting areas where additional local guidance may be most valuable to improve patient safety.
- Use of telecritical care for family visitation to ICU during the COVID-19 pandemic: an interview study and sentiment analysis by F Sasangohar (@Prof_Sasangohar) et al. Published Online First: 07 October 2020. Accompanying blog post available here.
Physical distancing measures have been implemented widely during the COVID-19 pandemic, including in healthcare environments. In this article, Sasangohar et al. detail work enabling online family visitation within a closed intensive care unit (ICU). The paper itself describes how their health system pivoted an existing platform used by physicians in providing telecritical care services to also include family members outside the hospital. Prior to the pandemic, each ICU room had already been outfitted with a camera with 360-degree pan, tilt, and zoom capabilities that connected with their software platform. Patient families were allowed to schedule a virtual visit in concert with the bedside nurse, who remained with the patient to help facilitate the conversation. During the study period, the team facilitated an average of 21.7 visits per day of an average 30 minutes’ duration. Families were overwhelmingly positive about the experience, with > 86% expressing positive sentiments and many noting feelings of happiness, joy, gratitude, and relief at being able to see their loved one. Concerns with the virtual visits included an inability to communicate with loved ones in certain situations due to patient clinical status, technical difficulties with the platform, lack of touch and physical presence, and the frequency and clarity of communications with the care team. One possible additional benefit of such virtual visits is increasing trust between patient families and the care teams. While many other institutions can likely learn from these efforts, future work will hopefully also improve the technical performance of similar platforms and further integrate such communication into clinical workflows – even after the pandemic.
- Incidence, nature and causes of avoidable significant harm in primary care in England: retrospective case note review by AJ Avery (@TonyAvery1) et al. Published Online First: 10 November 2020.
In many high-income countries, more than 90% of care is delivered in community settings; however, these settings remain understudied for healthcare-associated errors. With a team of 13 general practitioner researchers, Avery et al. performed an important large review of the incidence of avoidable harm in the primary care setting. Overall, the team reviewed > 14,000 primary care charts over a one-year period across 12 general practices in England. They identified and characterized safety incidents, their avoidability, and contributing factors. Over the study period, these patients had over 2,100 new significant health problems, of which 96.4% were deemed to have been managed with an adequate standard of care. The remaining 3.6% of cases included a few that were considered completely unavoidable. However, a larger number representing 1.5% of cases (or a rate of 35.6 per 100,000 patient-years) were deemed to have caused significant harm and were considered to be at least probably avoidable, and 2.4% (arate of 56.8 cases per 100,000 patient-years) to be at least possibly avoidable. Diagnostic errors comprised 61% of the avoidable errors, medication errors 26%, and delayed referrals 11%. In 80% of cases, the resultant harm could have been identified sooner or prevented with better adherence to evidence-based guidelines. Most contributing factors were patient-related (72%), although organisational factors were identified in 21%. Future safety improvement opportunities may exist within the two-thirds of organisational factors leading to harm that were related to insufficient continuity and coordination of care.