Volume 3, Issue 1

In this article we review current evidence on strategies to evaluate diagnostic error solutions, discuss the methodological challenges that exist in investigating the value of these strategies in patient care, and provide recommendations for methods that can be applied in investigating potential solutions to diagnostic errors. These recommendations were developed iteratively by the authors based upon initial discussions held during the Research Summit of the 7th Annual Diagnostic Error in Medicine Conference in September 2014. The recommendations include the following elements for designing studies of diagnostic research solutions: (1) Select direct and indirect outcomes measures of importance to patients, while also practical for the particular solution; (2) Develop a clearly-stated logic model for the solution to be tested; (3) Use rapid, iterative prototyping in the early phases of solution testing; (4) Use cluster-randomized clinical trials where feasible; (5) Avoid simple pre-post designs, in favor of stepped wedge and interrupted time series; (6) Leverage best practices for patient safety research and engage experts from relevant domains; and (7) Consider sources of bias and design studies and their analyses to minimize selection and information bias and control for confounding. Areas of diagnostic error mitigation research identified for further attention include: role of competing diagnoses, understanding the impacts of organizational culture, timing of diagnosis, and sequencing of research studies. Future research will likely require novel clinical, health services, and qualitative research methods to address the age-old problem of arriving at an accurate diagnosis.

Despite differing target audiences and scope it is possible to compare the Fifth Edition of the Diagnostic and Statistical Manual of Mental Disorders (DSM-5) [American Psychiatric Association: Diagnostic and Statistical Manual of Mental Disorders, 5th ed. Arlington: American Psychiatric Association, 2013] and the Second International Working Group for New Research Criteria for the Diagnosis of Alzheimer’s Disease (IWG-2) [Dubois B, Feldman HH, Jacova C, Hampel H, Molinuevo JL, Blennow K, et al. Advancing research diagnostic criteria for Alzheimer’s disease: the IWG-2 criteria. Lancet Neurol 2014;13:614–29] diagnostic criteria for both Alzheimer’s disease (AD) and dementia with Lewy bodies (DLB). With regard to the diagnosis of AD the principal difference is the inclusion of biomarkers in the IWG-2 diagnostic criteria for this condition. This creates a number of difficulties including a lack of regulatory approval, cultural and other objections to the collection of cerebrospinal fluid (CSF), and a lack of facilities for collection and analysis restricting analysis of CSF proteins to larger tertiary centres [Dubois B, Feldman HH, Jacova C, Hampel H, Molinuevo JL, Blennow K, et al. Advancing research diagnostic criteria for Alzheimer’s disease: the IWG-2 criteria. Lancet Neurol 2014;13:614–29]. With regard to diagnostic criteria for DLB, IWG-2 research criteria designate the co-occurrence of AD and DLB as ‘mixed AD’. However, Alzheimer’s type pathology (ADTP) and Lewy body pathology frequently occur together rendering a separate ‘mixed AD’ category superfluous. The reality is that routine clinical diagnosis of AD and DLB will continue to be based on a thorough general and neurological examination indicating a preponderance of signs and symptoms for one or other of these conditions [Seeley WW, Miller BL. Alzheimer’s disease and other dementias. In: Hauser SL, Josephson SA, editors. Harrison’s neurology in clinical medicine, 3rd ed. New York: McGraw Hill, 2013]. Similarly, AD and DLB research will continue to primarily depend on clinically focussed DSM-5 criteria, making DSM-5 superior to IWG-2 in both clinical and research settings.

The goal of the Choosing Wisely campaign launched in 2009 by the American Board of Internal Medicine (ABIM) Foundation is to promote dialog on avoiding wasteful or unnecessary medical tests, treatments and procedures. Originating in 2009, the Choosing Wisely initiative involved three primary care specialties in a project aiming to develop “Top Five” lists, which were to be “specialty-specific enumerations of achievable practice changes to improve patient health through better treatment choices, reduced risks and reduced costs. The initiative soon became global, many specialty societies outside the US joining. Some time later, however, data collected demonstrated that a reduction had been achieved only for a few low-value tests and procedures, thus highlighting the need for a more evidence-based approach for identifying low-value practices and for evaluating the efficacy of this initiative over time.

Background: Diagnostic errors pose a significant threat to patient safety but little is known about public perceptions of diagnostic errors. A study published in BMJ Quality & Safety in 2014 estimated that diagnostic errors affect at least 5% of US adults (or 12 million) per year. We sought to explore online public reactions to media reports on the reported frequency of diagnostic errors in the US adult population. Methods: We searched the World Wide Web for any news article reporting findings from the study. We then gathered all the online comments made in response to the news articles to evaluate public reaction to the newly reported diagnostic error frequency (n=241). Two coders conducted content analyses of the comments and an experienced qualitative researcher resolved differences. Results: Overall, there were few comments made regarding the frequency of diagnostic errors. However, in response to the media coverage, 44 commenters shared personal experiences of diagnostic errors. Additionally, commentary centered on diagnosis-related quality of care as affected by two emergent categories: (1) US health care providers (n=79; 63 commenters) and (2) US health care reform-related policies, most commonly the Affordable Care Act (ACA) and insurance/reimbursement issues (n=62; 47 commenters). Conclusion: The public appears to have substantial concerns about the impact of the ACA and other reform initiatives on the diagnosis-related quality of care. However, policy discussions on diagnostic errors are largely absent from the current national conversation on improving quality and safety. Because outpatient diagnostic errors have emerged as a major safety concern, researchers and policymakers should consider evaluating the effects of policy and practice changes on diagnostic accuracy.

Background: Previous studies have shown that changes in diagnoses from admission to discharge are associated with poorer outcomes. The aim of this study was to investigate how diagnostic discordance affects patient outcomes. Methods : The first three digits of ICD-9-CM codes at admission and discharge were compared for concordance. The study involved 6281 patients admitted to the Western Galilee Medical Center, Naharyia, Israel from the emergency department (ED) between 01 November 2012 and 21 January 2013. Concordant and discordant diagnoses were compared in terms of, length of stay, number of transfers, intensive care unit (ICU) admission, readmission, and mortality. Results : Discordant diagnoses was associated with increases in patient mortality rate (5.1% vs. 1.5%; RR 3.35, 95% CI 2.43, 4.62; p<0.001), the number of ICU admissions (6.7% vs. 2.7%; RR 2.58, 95% CI 2.07, 3.32; p<0.001), hospital length of stay (3.8 vs. 2.5 days; difference 1.3 days, 95% CI 1.2, 1.4; p<0.001), ICU length of stay (5.2 vs. 3.8 days; difference 1.4 days, 95% CI 1.0, 1.9; p<0.001), and 30 days readmission (14.11% vs. 12.38%; RR 1.14, 95% CI 1.00, 1.30; p=0.0418). ED length of stay was also greater for the discordant group (3.0 vs. 2.9 h; difference 8.8 min; 95% CI 0.1, 0.2; p<0.001) . Conclusion : These findings indicate discordant admission and discharge diagnoses are associated with increases in morbidity and mortality. Further research should identify modifiable causes of discordance.

Background: Specific immunoglobulin E to Ara h 2 (sIgE to Ara h 2) is described as an upcoming predicting factor for diagnosing peanut allergy in children. The gold standard for diagnosing peanut allergy is a double blind placebo controlled food challenge, however this is time consuming and potentially harmful. We investigate Ara h 2 as a preliminary less invasive diagnostic tool for diagnosing peanut allergy in a general population of peanut sensitized children. Methods: Children (n=52) with peanut sensitization were retrospectively included. An oral food challenge (OFC) confirmed peanut allergy or tolerance, as primary outcome. Individual candidate predictors were identified by univariate regression analysis and used in a prediction model. Different cut-off values were obtained and receiver operating characteristic curves were plotted. Results: Multivariate analyses resulted in Ara h 2 as best predictor, with a discriminative ability of 0.87 (95% confidence interval, 0.77–0.97). Sensitivity and specificity of 55% and 95%, respectively, were found for a sIgE to Ara h 2 cut-off value of 4.25 kU/L. The highest positive predictive value of 100% was reached at 5.61 kU/L. No absolute relation was found between the value of Ara h 2 and the severity of the reaction during OFC. Conclusion: This study developed a prediction model in which sIgE to Ara h 2 was the best predictor for peanut allergy in sensitized children in a general hospital. Therefore depending on the history and the Ara h 2 results, an OFC is not always needed to confirm the diagnosis.

Background: Acute pulmonary embolism (PE) is frequently a fatal disease. The clinical presentation of PE is variable and frequently nonspecific, and there is commonly a diagnostic delay. We aimed to investigate factors associated with the delay in the diagnosis of PE. Methods: Data from patients with PE were collected from January 2011 to December 2013 in an acute care teaching hospital. Time-to-diagnosis, evaluated by obtaining a diagnostic computed tomography scan, was then analyzed by the Cox proportional hazard model for examining factors associated with time to the diagnosis of PE. Independent variables included age, gender, activities of daily living, means of transport to the hospital, body temperature, hypoxemia, typical symptoms for PE, serum C-reactive protein (CRP) concentrations, infiltration on chest radiograph, Wells score, classification of patients with PE based on early mortality risk, patients referred from other specialties, daytime versus nighttime arrival, diagnosed by an emergency physician, and diagnosed by a medical resident. Results: Sixty patients were included. The time to diagnosis was significantly delayed in low-risk patients (hazard ratio [HR], 2.2; 95% CI, 1.2–4.1) and in patients who did not use an ambulance (HR, 1.9; 95% CI, 1.0–3.7). In an analysis of the latter subgroup, higher serum CRP concentrations were associated with a delayed diagnosis (HR, 1.1; 95% CI, 1.0–1.2). Conclusion: The time to the diagnosis of PE was delayed in low-risk patients and in patients who attended the hospital by means other than an ambulance. In such patients, a delayed diagnosis was associated with higher serum CRP concentrations.