Impact of the COVID-19 Pandemic on Radiation Treatment Termination Patterns at a Large Multi-Center Institution

PURPOSE/OBJECTIVE(S): Premature discontinuation of radiation therapy (RT) is a significant challenge portending inferior outcomes, particularly with curative intent RT. This study measures the impact of the COVID-19 pandemic compared to prior treatment terminations (TTs) from a multi-institutional health system in the New York Metropolitan area. MATERIALS/METHODS: TT data was collected from the internal quality assurance records of a multicenter radiation oncology department for the September 2019-February 2020 pre-pandemic interval (BP) and the March 2020-August 2020 post-pandemic interval (AP). We measured aggregate and monthly CT simulations (CT sims), TTs, and terminations with zero fractions of radiation given (0Fx), the latter stratified by palliative vs. curative intent (0Fx-P vs. 0Fx-C). AP TTs attributed to COVID-19 anxiety were also obtained. The data were analyzed as aggregate ratios with monthly weighting of TTs/CT sims, 0Fx/CT sims, and 0Fx/TTs. Ratio statistics were employed and 95% confidence intervals generated. 0Fx-P and 0Fx-C were tested for association with BP and AP by chi-square analysis. RESULTS: For BP and AP respectively: TTs/CT sims were 8.1% (95% CI 5.2%-11.0%) and 11.2% (95% CI 9.2%-13.3%); 0Fx/CT sims were 1.4% (95% CI 0.16%-2.70%) and 2.5% (95% CI 1.49%-3.6%); 0Fx/TTs were 15.8% (95% CI 6.35%-25.21%) and 22% (95% CI 16.0%-27.9%). The association of 0Fx-P and 0Fx-C with BP and AP yielded chi-square = 0.43, P = 0.51. TTs attributed to COVID-19 anxiety comprised 3.9% of TTs in AP. CONCLUSION: The endpoints for TTs in the BP and AP periods in our study cohort were suggestive of a COVID impact on patients discontinuing radiation therapy, but without statistical significance. Given the subtleties of TT dynamics and uncertainties such as the count of COVID-19 anxiety-related terminations, along with a relatively small sample size, it is reasonable to infer that this review was underpowered. Moreover, direct statistical attribution of TTs to COVID understates knock-on effects, e.g., the case of a 0Fx TT that would have occurred regardless of COVID but under the stress of the pandemic exacerbates the harm of the waste of resources. At the patient care level, understanding how COVID affects TT patterns may enable interventions that facilitate completion of care as intended by the treating physician. At the health system level, such understanding may elucidate a plausible mechanism of COVID's cascading downstream influence on oncologic outcomes. Nevertheless, it will require a multi-institutional cohort to quantitatively discern the impact of the pandemic on TT dynamics.