(S025) Discovery and Validation of Predictive Factors for Safety Incident Reports in Patients Receiving Radiation Therapy: Comparative Results From a Large Safety Variance Database

April 30, 2015

We determined several patient and treatment-specific characteristics that predicted for treatment incident events. Children were more susceptible to incident alerts, possibly related to increased staff vigilance or more frequent use of complex modalities. Patients with head and neck tumors, greater number of fractions, treatment on protocol, and use of IMRT also predicted for incidents.

Shereef M. Elnahal, MD, MBA, Amanda Blackford, SCM, Eric C. Ford, PhD, Annette N. Souranis, Valerie Briner, Koren Smith, PhD, Todd R. McNutt, PhD, Theodore L. Deweese, MD, Jean Wright, MD, Stephanie A. Terezakis, MD; Johns Hopkins University School of Medicine; Department of Radiation Oncology, University of Washington; Department of Radiation Oncology, Johns Hopkins Hospital

OBJECTIVES: Patient safety is a vital concern in radiotherapy, but little is known about the factors that predict patients at risk for safety incidents. We aimed to: a) describe patients who have experienced safety incident reports; and b) compare them with a cohort without events to identify predictive factors and build a nomogram tool to identify patients at risk.

MATERIALS AND METHODS: We used our institution’s radiation treatment safety incident reporting system to build a database of patients who experienced events from January 2011–October 2014. Patient and treatment-specific data were reviewed from our institution’s radiation management software program for a random selection of patients with events in our reporting system. A control group of sequential patients in 2014 was generated for comparison. Summary statistics, likelihood ratios, and mixed-effect logistic regression models were used for group comparisons.

RESULTS: The safety incident group comprised 1050 events. Errors in treatment planning (eg, incorrect planning target volume [PTV] margins submitted) (35.2%) and documentation (eg, wrong patient name denoted) (34.9%) were the most common incident types, followed by communication (17.5%) and treatment delivery (12.4%). Several clinical and treatment-specific factors correlated with safety reports. Incidents were much more frequently reported in minors (age < 18 yr) than in adults (25.3% vs 5.4%; P < .001). Patients with head and neck (16% vs 8%; P < .001) and breast (20% vs 15%; P = .03) primaries were more frequently associated with an incident. Larger tumors (19% vs 11% had T4 lesions; P = .02), a greater mean number of fractions (24 vs 20; P = .02), and cases on protocol (9% vs 5%; P = .005) or with intensity-modulated radiation therapy (IMRT)/image-guided IMRT (IGMRT) (52% vs 43%; P = .001) were associated with incident events. Of note, IMRT was used in 70.5% of minors treated. Other independently associated factors were found in patients with T4 tumors (19.5% protocol, 50.7% head and neck, mean of 24 fractions). On multivariate analysis, dose per fraction, head and neck cases, pediatric patients, protocol enrollment, and average pain score independently predicted for safety events.

CONCLUSIONS: We determined several patient and treatment-specific characteristics that predicted for treatment incident events. Children were more susceptible to incident alerts, possibly related to increased staff vigilance or more frequent use of complex modalities. Patients with head and neck tumors, greater number of fractions, treatment on protocol, and use of IMRT also predicted for incidents. Independently predictive variables have been built into a nomogram that will be validated in a prospective dataset, the results of which will be available at the time of the meeting. 

Proceedings of the 97th Annual Meeting of the American Radium Society- americanradiumsociety.org