Diagnostic Uncertainty Among Critically Ill Children Admitted to the Pediatric Intensive Care Unit: A Multi-center Study

Background: Uncertainty is inherent in diagnosis and must be managed appropriately. Diagnostic uncertainty in complex and high-stakes settings like pediatric intensive care units (PICUs) may be difficult to manage and place critically ill children at risk of harm from diagnostic error or over-testing. Our objective was to identify and characterize diagnostic uncertainty over time using PICU clinicians’ diagnosis documentation for children non-electively admitted to PICUs.

Methods: We performed a retrospective cohort study of 854 randomly selected patient admissions over one year to 4 tertiary-referral academic PICUs. At each site, 4-5 PICU clinicians conducted standardized chart review to determine diagnostic uncertainty at PICU admission, PICU discharge, and hospital discharge; 35% of charts were reviewed by at least 2 reviewers. We used a published chart review tool to identify diagnostic uncertainty documented in clinicians’ notes, and the validated Safer Dx instrument to identify patients with potential missed opportunities in diagnosis during the first 7 days after PICU admission. The 4 site primary investigators made final diagnostic error determinations by consensus. We performed descriptive data analysis to describe diagnostic uncertainty over time. We used a generalized linear mixed modeling framework to assess site (PICU)-clustered differences in clinical, encounter, and outcome characteristics of patients with and without diagnostic uncertainty at PICU admission.

Results: PICU admission notes for 220 of 854 (25.9%) patients documented diagnostic uncertainty. Inter-reviewer agreement was moderate (kappa=0.51; 95%CI 0.46,0.60). Clinical notes documented uncertainty in 96 (14.6%) patients at PICU discharge and 55 (6.5%) patients at hospital discharge (Table 1). Uncertainty was more common among White (OR 1.54, 95%CI: 1.12, 2.12) than non-White patients. Uncertainty was also more common among patients with neurologic diagnoses (OR 1.95, 95%CI: 1.34, 2.83) and less common among those with respiratory (OR 0.63, 95%CI: 0.47, 0.86) or trauma-related (OR 0.13, 95%CI: 0.03, 0.52) diagnoses. Uncertainty was more common for patients with atypical vs. typical presentations (OR 2.19, 95%CI: 1.08, 4.44). Diagnostic uncertainty was not associated with illness severity, complex chronic conditions, encounter characteristics, or attending intensivist characteristics. However, patients for whom uncertainty was documented had more: 1) imaging tests ordered (OR 1.17, 95%CI: 1.10, 1.26); 2) subspecialties consulted (OR 1.31, 95%CI: 1.14, 1.50); 3) diagnostic discordance between clinicians (OR 3.28, 95%CI: 1.63, 6.61), 4) diagnostic discordance across time (OR 5.39, 95%CI: 2.67, 10.85); and 4) diagnostic errors (OR 9.94, 95%CI: 2.70, 36.52) (Table 2).

Conclusion: Diagnostic uncertainty is common on PICU admission and decreases over time. Uncertainty is associated with increased diagnostic testing, diagnostic discordance, and diagnostic error. Diagnostic uncertainty on PICU admission may identify patients at risk for diagnostic error and should be considered in interventions to prevent diagnostic error-related harm among critically-ill children.