Development of a Prediction Model for Trauma Registry Chart Abstraction Time

Area of Science:

• Trauma Surgery
• Health Informatics
• Quality Improvement

Background:

• Level I trauma centers face challenges meeting data submission deadlines.
• Backlogs in case abstraction limit performance improvement opportunities.
• A validated algorithm is needed to evaluate registry workload.

Purpose of the Study:

• To develop a scientific model for predicting chart abstraction time.
• To evaluate registry productivity on a patient-by-patient basis.

Main Methods:

• Registrars documented time spent on each chart.
• 150 patients were randomly selected from each of 4 trauma registrars (total 600).
• Patient-related variables were identified by trauma program leadership.

Main Results:

• Inhospital mortality, transfer from referring facility, hospital stay, and ventilator days were significant predictors.
• Number of complications, specialty consults, diagnoses, blood products, and procedures also predicted abstraction time.
• A multivariate regression equation was developed to predict time (Y = 38.95 + 31.28 × mortality + 15.33 × referring facility + 4.68 × complications+3.55 × hospital stay + 3.33 × consults + 2.83 × diagnoses + 2.00 × ventilator days + 1.78 × blood products + 1.09 × procedures).

Conclusions:

• The prediction model uses patient-specific variables for accurate time estimation.
• This tool enables trauma centers to assess registry productivity.
• It identifies retrospective opportunities for performance improvement.