The Harvard medical practice study trigger system performance in deceased patients.

Area of Science:

• Healthcare Quality and Safety
• Medical Record Analysis
• Patient Safety Systems

Background:

• Retrospective chart review using trigger systems is a method to detect adverse events (AEs).
• Deceased patients' records are crucial for identifying AEs and improving care.
• Sampling deceased patients with triggers may enhance performance and reduce review burden.

Purpose of the Study:

• To determine the performance of a trigger system in a sample of deceased patients.
• To calculate the specificity and sensitivity of the trigger system for predicting AEs.

Main Methods:

• Screened 100 randomly selected medical records of deceased patients (2012-2015) using a trigger system.
• Utilized McNemar's test of symmetry, observed agreement, Cohen's Kappa, and PABAK for statistical analysis.
• Evaluated trigger performance for any trigger present and for individual triggers.

Main Results:

• Observed agreement for any trigger was 75% (Kappa: 0.5, PABAK: 0.5).
• Individual trigger agreement averaged 90% (Kappa: 0.42, PABAK: 0.8).
• Recalculated specificity was 58% and sensitivity was 92%; two AEs were found in cases without prior triggers.

Conclusions:

• The reproducibility of trigger systems is suboptimal, with performance varying among individual triggers.
• Low specificity results in a high false-positive rate, increasing the cost of labor-intensive record reviews.
• Development of more effective and efficient patient safety systems is necessary.