Limitations
Study limitations included the biased and confounding factors previously described about retrospective and nonrandomized observational studies that are controlled for during RCTs.15 Electronic health record data may have been incorrectly collected resulting in missing or wrong data points that affect the validity of our conclusion. Recall bias with regard to documentation by health care providers describing reasons why CRT or durvalumab were not initiated or the patient’s ability to recall previous treatments and report ECOG status or toxicities also may have impacted our findings. Comorbidities and poor performance status, frequently occurring among veterans, negatively impact cancer treatment decisions and may result in a detection bias. For example, tobacco use, cardiovascular disease, including heart failure, and chronic obstructive pulmonary disease, are notoriously higher in the US veteran population when compared with civilian cohorts.16-18 Also, veterans with poorly controlled depression and posttraumatic stress disorder resulting in functional impairment are a factor.19 Steps were taken to address some of these biases by performing repeat checks of tabulated data and employing 2 independent reviewers to evaluate all relevant clinical documentation, compare results, and reach a consensus.
Conlcusions
This retrospective analysis of adherence to category 1 NCCN guidelines for durvalumab use among patients at the Birmingham VAMC Oncology Clinic reinforced our practice and identified minor deficiencies in documentation that would impact future clinical visits. More importantly, it depicted the massive disparity in treatment candidacy among Birmingham veterans compared with clinical trial populations. Efforts will be made to address factors impacting a veteran’s candidacy for CRT and explore other variables such as socioeconomic barriers to treatment. Multiple complementary tools to assess patients’ frailty, such as the Charlson Comorbidity Index (CCI), are now being used for a variety of disorders including cancers. More robust data and standardization are needed to validate the use of these assessments in predicting response to immune checkpoint inhibitors.
Immune checkpoint inhibitors are currently being evaluated in stage III NSCLC studies and may be implemented as routine practice in the future.12 It is important to distinguish fit from frail veterans with lung cancer for treatment selection. We would like to see the expansion of the eligibility criteria for clinical trials to include patients with a performance status of ECOG 2 in order for results to be truly generalizable to the real-world population. Our hope is that such work will improve not only the quality of lung cancer care, but also the quality of care across multiple tumor types.