top of page
Research and Data Analysis
Translating complex data into actionable insights for inpatient diagnostic safety
Role: Researcher
_edited.png)
Time: 2 years
Company: Brigham and Women's Hospital
Projects
Predictive Analysis of EHR Data to Identify and Mitigate Diagnostic Errors
Background
Diagnostic errors (DEs) pose significant risks to hospitalized patients. Identifying factors that predict these errors early in the diagnostic process can help prevent DEs from occurring.
Methods
Trained clinicians completed structured chart reviews of 569 patients to identify DEs.
Synthesizing clinicians’ findings and data from the electronic health record (EHR), we used statistical models to uncover key predictors of DE across three groups of high-risk patients.
Findings
1
Multiple doctor visits before hospitalization were associated with a higher likelihood of diagnostic error, possibly indicating diagnostic uncertainty from clinicians.
2
Patients transferred from other hospitals were less likely to experience a diagnostic error. This may be because transfers were made for specific treatments or better care. Care teams may have also been more aware of potential health complexities.
3
Factors such as sex and being admitted to the advanced practice provider (APP) service may also influence diagnostic errors. However, a larger sample size is needed to evaluate this further.
Conclusions
This model helped identify some risk factors for DE, informing more targeted intervention development for improved outcomes in patient safety over shorter timelines.
However, more research is needed to refine predictions and better target at-risk patients.
Identifying Failures in the Diagnostic Process to Enhance Safety Interventions
Background
Diagnostic errors (DEs) in hospitals can lead to preventable harm, affecting about 22.7% of patients. Identifying key areas where the diagnostic process breaks down could help improve patient safety and reduce errors.
Methods
-
Analyzed 100+ cases with DEs and 200+ without DEs, as determined by trained clinician EHR chart reviewers.
-
Identified cases and conducted bivariate analysis for the number of diagnostic process failures between error-positive and error-negative cases, as well as a population estimate and weighted odds ratio.
-
Ranked the most common failures to highlight areas for improvement.
Findings
Clinical Data Gathering Is Fraught With Failures
DEs were significantly linked to failures in key six areas of the diagnostic process, including physical exams, patient history, and diagnostic testing.
Patients' Health History Is Inadequately Considered
The most common failure was improper evaluation of patient history, highlighting critical areas for improvement in reducing diagnostic errors.
Conclusions
The diagnostic process is complex but has high-yield intervention targets for mitigating risk of DE.
Potential targets include improving gaps in physical exam assessment, history, and diagnostic testing and interpretation.
bottom of page