Landscape of mortality during and within thirty days after non-palliative radiotherapy across eleven major cancer types

Ninety-day mortality following curative intent radiotherapy for stage I-III lung cancer in the Netherlands

BACKGROUND AND PURPOSE

The 90-day mortality following lung cancer treatment is a common performance indicator. The aim of this study was to investigate 90-day mortality following (chemo)radiotherapy for stage I-III lung cancer to evaluate the applicability of this outcome indicator in this patient population.

MATERIALS AND METHODS

The Netherlands National Cancer Registry was queried for this retrospective population-based study. Early mortality rates from the start and end of radiotherapy are reported with 95% confidence intervals (CI). The association between clinical characteristics and 90-day mortality was evaluated with multivariable logistic regression analysis.

RESULTS

18,355 Patients treated between 2015 and 2020 were included. The 90-day mortality was 2.56% in stages I-II and 4.60% in stage III, was significantly higher in males, elderly patients and patients with a poor performance status and independent of facility volume. In stage I-II, 90-day mortality was lower after stereotactic versus conventional radiotherapy (2.0% versus 5.25%, OR 0.5 (95%CI 0.4-0.7)). In stage III, mortality decreased from 5.26% in 2015-2016 to 3.73% in 2019-2020 (OR 0.7 (95% CI 0.5-0.9)) and was lower after concurrent versus sequential chemoradiotherapy (3.4% versus 5.9%, OR 1.5 (95%CI 1.2-1.9)). Early mortality increased to 3.20% in stages I-II and 6.70% in stage III when calculated from the end of radiotherapy.

CONCLUSION

Short-term mortality rates following curative intent radiotherapy for lung cancer in the Netherlands are low and independent of facility volume. It was demonstrated that 90-day mortality is an arguable indicator to monitor radiotherapy quality and that standardization of definitions and relevant case-mix factors is warranted.

Nomogram to predict risk of early mortality following definitive or adjuvant radiation and systemic therapy for head and neck cancer

Purpose/Objectives

We sought to create nomograms to predict individual risk of early mortality, which can identify patients who require interventions to prevent early death.

Methods

We included patients in the National Cancer Database with non-metastatic squamous cell carcinoma of the head and neck who received radiation and systemic therapy between 2004 and 2017 in the definitive or adjuvant setting. Early mortality was defined as any death less than 90 days after starting radiation. Multivariable logistic regression was used to assess the relationship between covariates and early mortality. Nomograms to predict the risk of early death were created for both the definitive and adjuvant settings.

Results

Among 84,563 patients in the definitive group and 18,514 patients in the adjuvant group, rates of early mortality were 3.5 % (95 % CI 3.4-3.7 %) and 2.2 %, (95 % CI 1.9-2.4 %), respectively. Patients above the age of 70 had an early mortality rate of 7.8 % (95 % CI 7.3-8.2 %) in the definitive group and 4.4 % (95 % CI 3.6-5.4 %) in the adjuvant group. In the multivariable analysis, age, comorbidity, T and N category, and tumor site were associated with early mortality in both cohorts (p < 0.05 for all). Nomograms including age, comorbidity, T and N category and tumor site performed better than age alone at predicting early mortality (AUC for definitive group: 0.70 vs 0.66; AUC for adjuvant group: 0.71 vs 0.61).

Conclusion

Nomograms including age, comorbidity, T and N category and tumor site were developed to predict the risk of early death following definitive or adjuvant chemoradiation.

Natural Language Processing Methods to Empirically Explore Social Contexts and Needs in Cancer Patient Notes

PURPOSE

There is an unmet need to empirically explore and understand drivers of cancer disparities, particularly social determinants of health. We explored natural language processing methods to automatically and empirically extract clinical documentation of social contexts and needs that may underlie disparities.

METHODS

This was a retrospective analysis of 230,325 clinical notes from 5,285 patients treated with radiotherapy from 2007 to 2019. We compared linguistic features among White versus non-White, low-income insurance versus other insurance, and male versus female patients' notes. Log odds ratios with an informative Dirichlet prior were calculated to compare words over-represented in each group. A variational autoencoder topic model was applied, and topic probability was compared between groups. The presence of machine-learnable bias was explored by developing statistical and neural demographic group classifiers.

RESULTS

Terms associated with varied social contexts and needs were identified for all demographic group comparisons. For example, notes of non-White and low-income insurance patients were over-represented with terms associated with housing and transportation, whereas notes of White and other insurance patients were over-represented with terms related to physical activity. Topic models identified a social history topic, and topic probability varied significantly between the demographic group comparisons. Classification models performed poorly at classifying notes of non-White and low-income insurance patients (F1 of 0.30 and 0.23, respectively).

CONCLUSION

Exploration of linguistic differences in clinical notes between patients of different race/ethnicity, insurance status, and sex identified social contexts and needs in patients with cancer and revealed high-level differences in notes. Future work is needed to validate whether these findings may play a role in cancer disparities.