Provider decision regret - a useful method for analysis of palliative thoracic re-irradiation for lung cancer?

Abstract

Background - The overall usefulness of palliative thoracic re-irradiation depends on the balance between efficacy, survival, and toxicity, and is difficult to judge from previous studies. In the absence of patient-reported data, we developed a method for provider decision regret that addresses the question “would we re-irradiate this patient again in light of the known outcome?” Furthermore, we analyzed different reasons for decision regret and defined a subgroup at increased risk.<p>

<p>Patients and methods - A retrospective analysis of 33 patients with lung cancer re-irradiated with 17–45 Gy was performed. Reasons for decision regret included re-irradiation within the last 30 days of life, immediate radiological progression after re-irradiation (as opposed to stable disease or objective response), radiation myelopathy, any grade 4–5 toxicity, grade 3 pneumonitis, and other grade 3 toxicity in the absence of a symptomatic benefit or a time period of at least 3 months without worsening of the treated tumor.<p>

<p>Results - Median survival time was 5.2 months (95% confidence interval 3.4–7.0 months). Symptomatic and radiological responses were observed. Provider decision regret was declared in 12 patients (36%): 2 patients with grade 3 pneumonitis, 3 patients with a short survival (radiotherapy during the last 30 days of life), and 7 patients with progression. Decision regret was declared only in patients with Eastern Cooperative Oncology Group (ECOG) performance status (PS) 2 or 3 and was associated with a time interval to re-irradiation <6 months.<p>

<p>Conclusion - Our data support the usefulness and acceptable side effects profile of palliative re-irradiation for lung cancer. Patients with reduced PS are at increased risk of futile treatment. Future research should aim at prediction of immediate disease progression (the prevailing cause of decision regret). Evaluation of provider decision regret has the potential to improve the way we learn from retrospective databases and should also be considered for other scenarios where high-quality prospective outcome data are lacking.