Using Electronic Data to Improve Care of Patients With Known or Suspected Cancer

Early diagnosis, management, and treatment of cancers confer improved survival, thus making timely diagnosis imperative. Delays in cancer diagnosis, however, are common and may arise when abnormal cancer screening results are missed by providers. These delays may result in a reduced opportunity for early and potentially curative cancer therapy.

This project used health information technology (IT) to identify patients for whom a diagnosis of prostate, lung, or colon cancer had been delayed, with the delay then communicated to providers. The specific aims of this project were to:

• Identify patients with cancer-related diagnostic delays using trigger-based data mining of an electronic health record (EHR) repository.
• Determine the effectiveness of a health IT-based intervention to facilitate cancer diagnosis as compared with usual care.

The project team developed electronic triggers to detect delays in diagnosis of the three cancers using information from literature reviews, expert opinion, and clinical logic. The triggers were then validated using manual chart reviews to determine the rate of a positive trigger in patients with cancer, i.e. the positive predictive value (PPV). Each of the triggers achieved PPVs of 58.3 to 70.2 percent. Based on this, use of the four triggers at the study sites could detect an estimated 1,048 instances of delayed or missed followup of abnormal findings and 47 high-grade cancers annually.

Using a randomized controlled trial design, the triggers were applied to all patients cared for by study providers over the course of 15 months. Manual chart reviews were conducted on records identified by the triggers to confirm and analyze the presence of diagnostic breakdowns. The patient’s treating provider was notified of the delay, and electronic tracking was used to monitor the patient. The study team compared the amount of time to followup action with the intervention versus the control patients.

Compared to the control group, patients cared for by providers in the intervention group experienced shorter time to followup care for the prostate and colorectal cancer triggers, while no difference was seen in the lung cancer triggers. Only 21 records with delayed followup were identified for lung cancer, and no difference in the time to follow up between intervention and control groups was observed.