Estimating the impact of the COVID-19 pandemic on diagnosis and survival of five cancers in Chile from 2020 to 2030: a simulation-based analysis

Abstract

Background: The COVID-19 pandemic has strained health system capacity worldwide due to a surge of hospital admissions, while mitigation measures have simultaneously reduced patients' access to health care, affecting the diagnosis and treatment of other diseases such as cancer. We estimated the impact of delayed diagnosis on cancer outcomes in Chile using a novel modelling approach to inform policies and planning to mitigate the forthcoming cancer-related health impacts of the pandemic in Chile.

Methods: We developed a microsimulation model of five cancers in Chile (breast, cervix, colorectal, prostate, and stomach) for which reliable data were available, which simulates cancer incidence and progression in a nationally representative virtual population, as well as stage-specific cancer detection and survival probabilities. We calibrated the model to empirical data on monthly detected cases, as well as stage at diagnosis and 5-year net survival. We accounted for the impact of COVID-19 on excess mortality and cancer detection by month during the pandemic, and projected diagnosed cancer cases and outcomes of stage at diagnosis and survival up to 2030. For comparison, we simulated a no COVID-19 scenario in which the impacts of COVID-19 on excess mortality and cancer detection were removed.

Findings: Our modelling showed a sharp decrease in the number of diagnosed cancer cases during the COVID-19 pandemic, with a large projected short-term increase in future diagnosed cases. Due to the projected backlog in diagnosis, we estimated that in 2021 there will be an extra 3198 cases (95% uncertainty interval [UI] 1356-5017) diagnosed among the five modelled cancers, an increase of nearly 14% compared with the no COVID-19 scenario, falling to a projected 10% increase in 2022 with 2674 extra cases (1318-4032) diagnosed. As a result of delayed diagnosis, we found a worse stage distribution for detected cancers in 2020-22, which is estimated to lead to 3542 excess cancer deaths (95% UI 2236-4816) in 2022-30, compared with the no COVID-19 scenario, among the five modelled cancers, most of which (3299 deaths, 2151-4431) are projected to occur before 2025.

Interpretation: In addition to a large projected surge in diagnosed cancer cases, we found that delays in diagnosis will result in worse cancer stage at presentation, leading to worse survival outcomes. These findings can help to inform surge capacity planning and highlight the importance of ensuring appropriate health system capacity levels to detect and care for the increased cancer cases in the coming years, while maintaining the timeliness and quality of cancer care. Potential delays in treatment and adverse impacts on quality of care, which were not considered in this model, are likely to contribute to even more excess deaths from cancer than projected.

Funding: Harvard TH Chan School of Public Health.

Translations: For the Spanish and Portuguese translations of the abstract see Supplementary Materials section.

Figure 1

Conceptual model

Figure 2

Estimated cases diagnosed per month in the COVID-19 vs no COVID-19 scenarios, 2015–30 The COVID-19 scenario describes the status quo, whereas the no COVID-19 scenario is a counterfactual scenario in which the impacts of COVID-19 on excess mortality and cancer detection were removed. Shaded regions indicate 95% uncertainty intervals. GES=Garantías Explícitas en Salud.

Figure 3

Estimated stage distribution at diagnosis and 5-year net survival by cancer, 2005–30 The COVID-19 scenario describes the status quo, whereas the no COVID-19 scenario is a counterfactual scenario in which the impacts of COVID-19 on excess mortality and cancer detection were removed. Shaded regions indicate 95% uncertainty intervals. Datapoints for empirical mean estimates from the medical literature are shown, with reported 95% CIs.

Figure 4

Estimated cancer deaths per month, 2015–30 Shaded regions indicate 95% uncertainty intervals.