Care Team Model and Diagnostic Error Risk in Medical Patients Who Transferred to the ICU or Died.

Abstract

Diagnostic errors are a large source of harm among hospitalized patients. Understanding how medical team composition influences diagnostic safety is needed to design care models that support accurate and effective diagnostic processes. To evaluate whether direct attending care, resident teaching teams, or advanced practice provider-led (APP) services are associated with differences in the risk of diagnostic errors. Retrospective observational study using average treatment effect and inverse probability of treatment weighting to estimate marginal rate ratios (mRRs) for diagnostic errors across care models. Diagnostic errors were previously adjudicated by two-physician chart review methods. One thousand five hundred forty-four general medicine patients who transferred to the ICU or died in hospital between January 1 and December 31, 2019, at 29 hospitals across the USA. None. Primary outcomes were any diagnostic error and harmful diagnostic error. Any diagnostic error was defined as a missed opportunity for a timely and accurate diagnosis, regardless of whether harm occurred, whereas harmful diagnostic errors were limited to those resulting in temporary harm requiring intervention, permanent harm, or death. Of 1544 patients, 969 (63%) were cared for by teaching teams, 442 (29%) by direct care, and 133 (9%) by APP services. Direct care was associated with a higher risk of any diagnostic error compared with teaching teams (mRR, 1.36; 95% CI, 1.04-1.68). Care team structure may influence diagnostic error risk. Models involving residents, often working with attending physicians, may offer diagnostic safety advantages through team-based diagnostic decision-making and workload distribution.