Complementary Approaches to Identifying Missed Diagnostic Opportunities in Hospitalized Children Free

Pediatric providers recognize that the children and adolescents for whom they care face risks of harm associated with delayed, missed, and erroneous diagnoses.¹  Over one-third of hospital-based pediatric providers estimate committing diagnostic errors (DxEs) resulting in patient harm at least 1 to 2 times per year.²  However, 2 significant barriers prevent pediatric hospitals from identifying, learning from, and reducing harm attributable to DxEs as recommended by the National Academy of Medicine.³  First, clinicians infrequently report DxEs to incident-reporting systems relative to other medical errors.⁴  Second, clinicians hold strong reservations about discussing their own DxEs for system improvement.²  Consequently, novel approaches to identify diagnostic learning opportunities are required for pediatric providers to systematically reduce patient harm from DxE.

This issue of Hospital Pediatrics contains 2 reports detailing complementary approaches to detecting missed diagnostic opportunities among children undergoing diagnostic evaluation in the hospital setting. Although each approach employs a specific trigger to identify episodes of care at high risk of DxE, the methods focus on different clinical environments and practices to promote diagnostic excellence for health care organizations (HCOs) suggested by the Joint Commission.⁵  As HCOs begin incorporating diagnostic safety into their more extensive patient safety portfolios, not every approach for identification and learning from DxEs will suit every condition or clinical care area equally well. Overlapping processes permit a more complete appraisal of an HCO’s diagnostic safety.⁶ 

Michelson et al report the adaptation of a previously validated trigger tool applied to administrative data to predict delayed diagnosis of appendicitis in general emergency departments (EDs).⁷  Having first demonstrated the utility of this method at freestanding children’s hospitals, the authors report similar performance in general EDs, with a positive predictive value for delayed diagnosis of appendicitis of 92% for their algorithm.⁸  Because most children receive emergency care in general EDs, this tool has the potential to promote pediatric diagnostic safety beyond the walls of tertiary and quaternary referral centers. This approach also addresses 1 of the significant barriers to identifying and learning from DxEs by eliminating reliance on providers to report potential DxEs. Additionally, the authors compared their trigger to the criterion standard of consensus diagnostic delay determined through manual chart review, thereby reducing the time necessary for patient safety teams or ED quality and safety directors to aggregate data that can inform diagnostic performance improvement efforts (eg, clinical care pathways).

Using administrative data to identify DxE trends for specific conditions also augments national benchmarks for diagnostic performance. For example, the American College of Surgeons National Surgical Quality Improvement Program tracks negative appendectomy rates, which, in children, lie at about 3.3%.⁹  Tracking negative appendectomies involves comparing 2 distinct events: an operation and a subsequent pathologic determination of whether that operation removed a normal or inflamed appendix. Conversely, monitoring diagnostic delays for appendicitis presents a greater challenge because this requires careful review of the clinical record to identify possible earlier clinical encounters where a clinician should have pursued the diagnosis but did not. Although single-center studies relying on manual chart review from children’s hospitals report rates of delayed diagnosis of appendicitis from 2.6% to 4.8%, scaling this review to a national level is impractical.^10,11  The method developed by Michelson et al provides an opportunity to study diagnostic delays more systematically and efficiently across institutions where most children receive care without relying on providers to report the DxE.

Leveraging large administrative data sets for specific high-frequency, difficult-to-diagnose conditions like appendicitis allows scaling identification of diagnostic performance opportunities across institutions. However, individual institutions can also benefit from identifying local trends in diagnostic performance. Harm from low-frequency DxEs frequently undergoes review in morbidity, mortality, and improvement conferences or during root cause analyses. The detailed analysis characteristic of such reviews may identify unique diagnostic performance opportunities related to clinical knowledge, institutional resources (eg, availability of MRI), and consultant recommendations. However, these reviews are limited insofar as they focus on a single event and the system improvements stemming from them may have limited application to larger diagnostic performance trends.

The study by Congdon et al reviewed pediatric patients readmitted within 15 days of a previous hospitalization (close temporal proximity + unexpected care trajectory).¹²  They found that 6.3% of patients experiencing unplanned readmission suffered a DxE and had an average of 5 areas amenable to improving diagnostic performance, particularly with respect to differential diagnosis generation and prioritization. Furthermore, they demonstrated that patients suffering a DxE had a significantly longer readmission than those without DxE, supporting a patient-centered, as well as financial, imperative for improving diagnostic performance. Although the study did not identify specific admitting diagnoses that accounted for readmissions, this highlights that any admitted patient is at risk for the “hospital-acquired condition” of DxE.

Previous work has demonstrated that specific care events (eg, unplanned ICU transfers, unplanned admissions after a previous ED or primary care visit) often involve DxEs.^13–15  Detecting possible missed diagnostic opportunities in these studies relied on triggers that couple patient encounters closely related in time with unexpected care trajectories. This approach aggregates relatively rare events in a way that permits previously unknown but recurrent risks for DxE to become more visible to clinicians and safety teams. For example, Singh et al found that process breakdowns in patient–provider interactions and referrals were common contributors to DxEs among adult ambulatory patients irrespective of specific diagnoses.¹⁶  Lam and colleagues detected repeated diagnostic reasoning errors among children with unplanned admission after a previous ED visit, including several patients with serious intracranial pathology initially attributed to migraine headaches.¹⁷ 

Although a secondary objective, the study by Congdon et al also addresses the barrier regarding discussion of DxE to improve diagnostic performance. In their discussion, the authors described how using the term missed diagnostic opportunities instead of DxE and a standardized rubric promoted psychological safety and allowed for eventual consensus. These are 2 necessary components for facilitating nonpunitive discussion of sensitive topics.^18,19  The diagnostic process is a high-risk activity, so missed or near-miss diagnostic opportunities need to be discussed in an intentional and collaborative manner without fear of punishment.

These 2 reports demonstrate important characteristics of high-reliability organizations, such as preoccupation with failure and commitment to resilience, which are necessary mindsets needed to improve pediatric diagnostic safety.²⁰  Perhaps, the most encouraging aspect of these 2 reports lies in the fact that not only are pediatric providers recognizing the harm that arises from DxEs, but pediatric researchers and patient safety leaders have begun developing methods to identify and reduce them. Apart from falls and, possibly, medication errors, DxEs are the only medical error to which all our patients are susceptible. Just as our patients are not all 1 size, neither are the solutions necessary to systemically reduce their risk of harm during the diagnostic process.