The Kern Fever in the Emergency Department Study (Kern FEDS): Clinical appearance, serious bacterial illness, and the meaning of uncertainty.
Background: Emergency department (ED) fever management algorithms require the clinician to categorize febrile children as ‘ill’ or ‘not ill’ appearing when determining the risk for serious bacterial illness (SBI). This study describes a natural experiment where an ED pediatric chart allowed clinicians a third option, ‘unsure’. Hypotheses: We hypothesized (1) that chart prompts would improve documentation of clinical appearance, and (2) that exam findings and prevalence of serious bacterial illness in infants categorized as ‘unsure’ would be intermediate between those who were ill and not ill appearing. Design: We conducted a retrospective study of 3005 ED patients aged 0-24 months who had microbiology testing for fever in the ED between 1/1/2006 and 11/30/2009. We modeled overall appearance as the dependent and individual physical findings as the independent variables with ordinal logistic regression to help establish the validity of clinical appearance as a concept. We then compared the prevalence of the components of SBI, bacterial meningitis, pneumonia, urinary tract infection (UTI) and positive blood cultures, between the categorizations, not ill appearing, unsure and ill appearing. Results: Clinical appearance was documented in 60/583 (10.3%) whose encounter was recorded on the template without prompts versus 2036/2420 (84%) with prompts (p<0.001). Age odds ratio (OR) 1.04 (95% CI 1.01, 1.07) weight (quintile) OR 0.81 (95% CI 0.70, 0.95), dehydration OR 9.68 (95% CI 7.17, 13.01), tachycardia OR 1.31 (95% CI 1.04, 1.68), tachypnea OR 2.44 (95% CI 1.61, 3.68), prior antipyretics OR 0.65 (95% CI 0.52, 0.83) and prior antibiotics OR 2.56 (95%CI 1.71, 3.82) were associated with appearance. There was an ordinal relationship between appearance and the prevalence of bacterial meningitis and pneumonia for the categories ill appearing, unsure, and not ill appearing. The prevalence of positive blood cultures among children categorized as ‘not ill appearing' and 'unsure' was similar. Urinary tract infection (UTI) prevalence was similar regardless of appearance. Conclusion: Charting prompts increased documentation of clinical appearance. There was an ordinal relationship between the prevalence of meningitis, and pneumonia, across the categories 'ill appearing', 'unsure' and 'not ill appearing'. This was not the case for blood cultures or UTI.