Abstract
IMPORTANCE: Systematically capturing cancer stage is essential for any serious effort by health systems to monitor outcomes and quality of care in oncology. However, oncologists do not routinely record cancer stage in machine-readable structured fields in electronic health records (EHRs). OBJECTIVE: To evaluate whether a peer comparison email intervention that communicates an oncologist's performance on documenting cancer stage relative to that of peer physicians was associated with increased likelihood that stage was documented in the EHR. DESIGN, SETTING, AND PARTICIPANTS: This 12-month, randomized quality improvement pilot study aimed to increase oncologist staging documentation in the EHR. The pilot study was performed at Massachusetts General Hospital Cancer Center from October 1, 2018, to September 30, 2019. Participants included 56 oncologists across 3 practice sites who treated patients in the ambulatory setting and focused on diseases that use standardized staging systems. Data were analyzed from July 2, 2019, to March 5, 2020. INTERVENTIONS: Peer comparison intervention with as many as 3 emails to oncologists during 6 months that displayed the oncologist's staging documentation rate relative to all oncologists in the study sample. MAIN OUTCOMES AND MEASURES: The primary outcome was patient-level documentation of cancer stage, defined as the likelihood that a patient's stage of disease was documented in the EHR after the patient's first (eg, index) ambulatory visit during the pilot period. RESULTS: Among the 56 oncologists participating (32 men [57%]), receipt of emails with peer comparison data was associated with increased likelihood of documentation of cancer stage using the structured field in the EHR (23.2% vs 13.0% of patient index visits). In adjusted analyses, this difference represented an increase of 9.0 (95% CI, 4.4-13.5) percentage points (P = .002) in the probability that a patient's cancer stage was documented, a relative increase of 69% compared with oncologists who did not receive peer comparison emails. The association increased with each email that was sent, ranging from a nonsignificant 4.0 (95% CI, -0.8 to 8.8) percentage points (P = .09) after the first email to a statistically significant 11.2 (95% CI, 4.9-17.4) percentage points (P = .003) after the third email . The association was concentrated among an oncologist's new patients (increase of 11.8 [95% CI, 6.2-17.4] percentage points; P = .001) compared with established patients (increase of 1.6 [95% CI, -2.9 to 6.1] percentage points; P = .44) and persisted for 7 months after the email communications stopped. CONCLUSIONS AND RELEVANCE: In a quality improvement pilot trial, peer comparison emails were associated with a substantial increase in oncologist use of the structured field in the EHR to document stage of disease.