Computer protocol finds undetected patients

By combining automated protocols with clinical protocols, a large teaching hospital was able to better identify infectious patients who were not put in isolation, researchers report in the March issue of Infection Control and Epidemiology.

The infection control program at Columbia-Presbyterian Medical Center in New York has utilized medical logic modules to automate its respiratory isolation policy since 1995. The computerized system used automated natural language processing rather than data-entry personnel to encode chest radiograph reports of patients and algorithms, and consequently determine whether patients meet respiratory isolation criteria.

Evaluation includes 214 patients

A retrospective evaluation was developed for a cohort of 171 adult culture-positive TB inpatients to determine whether the automated protocol combined with a clinical protocol worked better than the clinical protocol alone. In addition, 43 adult inpatients were evaluated prospectively.

The study found that automated decision support is a useful adjunct to clinical management, particularly at a hospital such as Columbia-Presbyterian, where a large volume of chest radiographs are performed.

"As the number of TB cases continues to decline in New York City and clinicians have less experience with TB, ensuring isolation of infectious patients may depend increasingly on automated methodologies and the willingness to isolate an increasing percentage of patients with criteria for respiratory isolation but without TB," the authors note.