A Qualitative Assessment of the Determinants of Completing Therapy for Patients with Select Cancers

BACKGROUND: Various attempts have been made to understand how patient choice determines utilization patterns for clinical care and service delivery.  In 2010 the Florida Cancer Data System (FCDS) received funding from NPCR/CDC to become an Enhanced Registry for Comparative Effectiveness Research.  FCDS staff collection detailed first course of treatment for cancers of the breast, colon, rectum, and chronic myeloid leukemia diagnosed in 2011 and patient data across Florida’s five most populous counties.  Both the planned and received treatments were recorded by agent NSC number, dosages, dates received, and reasons for discontinued treatment, as well as biomarker status.

METHODS:   The data capture process consisted of FCDS field staff conducting on-site case collection and follow-back to hospital facilities.  Throughout data collection the FCDS recorded text documentation that supported the coded values for patient treatment data.  Particular documentation also was made for cases where codified treatment values were not complete in detailing treatment decisions.  FCDS conducted a qualitative analysis of these cases to identify contextual factors behind treatment decisions that were not captured through selection of codified values.  Additionally, a qualitative assessment of the abstractors’ experience during data collection provided insight into the pattern of information available on patient medical charts, as well as the trend in types of biomarkers ordered, and the difference between hospital-based and physician-based treatment decisions. 

RESULTS:   In many circumstances reasons that guide treatment decisions were not captured by available codes.  Reason for refusing treatment, for example, was often due to lack of financial resources and insurance, as indicated in the original patient medical chart.  There were no codes in the CER record applicable for these circumstances.  Other patterns observed relate to the type of biomarkers ordered in the physician office compared to treating hospitals. Oncotyping was a common test ordered among private oncology practices that drove breast cancer treatment decisions despite presence of nodes positive in patients.  Due to low recurrence scores among these cases, many patients opted to forgo chemotherapy.

CONCLUSIONS:   Qualitative review of text data provides important contextual information by capturing a more comprehensive profile of treatments planned and eventually received, reasons for refusing treatment, and utilization of specific prognostic biomarkers for treatment decisions.  While evaluation of experience-based knowledge of registrars after data collection may not be feasible on a larger scale, for the purposes of this project in Florida, it added much value to the coded dataset and highlighted emerging patterns that can now be targeted in broader analyses.