Delays in diagnosis represent a major cause of morbidity in the emergency department (ED) setting. Children are likely at increased risk of delayed diagnosis in the ED because symptoms and signs of serious disease may be nonspecific, and because as many as 95% of children visit EDs that primarily evaluate adults. Rates of delayed diagnosis among children in the ED are unknown, but likely vary substantially across settings. Because delayed diagnosis represents a major threat to patient safety, focused research is needed to determine its incidence, risk factors, and outcomes in children visiting EDs. The major objective of this project is to determine patient and hospital attributes that increase the risk for delayed diagnosis of serious conditions among children visiting EDs, and to determine the differences in patient outcomes when delayed diagnosis occurs. The project?s specific aims are (1) to refine and validate an algorithm for accurately detecting delayed diagnosis of three representative serious conditions (appendicitis, new-onset diabetic ketoacidosis, and sepsis) in billing claims data; (2) to determine the incidence, between-hospital variability, and risk factors for delayed diagnosis; and (3) to determine condition-specific outcomes of delay.
Aim 1 will refine the claims detection algorithm and compare its performance with manual record review; it will be conducted at multiple collaborating centers.
Aims 2 and 3 will use the Agency for Healthcare Research and Quality (AHRQ) Statewide ED and Inpatient Databases for multiple states to apply the algorithm to identify delayed diagnoses. Multilevel models will be constructed to assess predictors of delay and related health and utilization outcomes including need for surgical procedures, critical care interventions, prolonged hospital stays, and death. This study will directly address a top AHRQ research priority, improving health care patient safety through identification of risks, hazards, and harms. We will focus exclusively on children, an AHRQ priority population. The results of this research will directly allow more widespread detection and surveillance of diagnostic delays for critical high-risk conditions, and hone methods to expand to others. This approach utilizing a claims detection method would allow the kind of monitoring of error that is a fundamental property of a learning health system. The Principal Investigator, Dr. Kenneth Michelson, is an early career physician-scientist with a strong clinical background in Pediatrics and Emergency Medicine. This award will foster his development as a researcher with content expertise in diagnostic error, sophisticated outcomes analysis using multilevel modeling, large database analytics, and implementation science. A strong mentorship team composed of experienced biostatistical, quantitative, emergency medicine, and diagnostic error experts support the project and foster Dr. Michelson?s career development toward independent research.
Delayed diagnosis of serious emergencies is a major threat to patient safety and public health among children who visit emergency departments, and reducing the incidence of these errors is a top public health priority. Improving diagnostic accuracy requires generalizable knowledge of the settings, patient characteristics, and diseases most associated with delayed diagnosis. This career development award will directly address this knowledge gap by detecting cases of delayed diagnosis of serious pediatric conditions in emergency departments, and determining the risk factors and outcomes of delay across a broad, diverse population of children.
