A modelled evaluation of the impact of COVID-19 on breast, bowel, and cervical cancer screening programmes in Australia
Curation statements for this article:-
Curated by eLife
eLife assessment
This study presents important results on the predicted impact of cancer screening disruptions in Australia during the COVID-19 pandemic based on consultation with public health stakeholders. The evidence presented is solid, as simulations were based on several previously validated breast, cervical, and bowel cancer screening decision models, though the scenarios were based on hypothetical disruptions that do not always match experienced disruptions. The work will be of interest to local policy-makers, public health specialists, and cancer epidemiologists.
This article has been Reviewed by the following groups
Listed in
- Evaluated articles (eLife)
- Epidemiology and Global Health (eLife)
Abstract
Australia introduced COVID-19 infection prevention and control measures in early 2020. To help prepare health services, the Australian Government Department of Health commissioned a modelled evaluation of the impact of disruptions to population breast, bowel, and cervical cancer screening programmes on cancer outcomes and cancer services. We used the Policy1 modelling platforms to predict outcomes for potential disruptions to cancer screening participation, covering periods of 3, 6, 9, and 12 mo. We estimated missed screens, clinical outcomes (cancer incidence, tumour staging), and various diagnostic service impacts. We found that a 12-mo screening disruption would reduce breast cancer diagnoses (9.3% population-level reduction over 2020–2021) and colorectal cancer (up to 12.1% reduction over 2020–21), and increase cervical cancer diagnoses (up to 3.6% over 2020–2022), with upstaging expected for these cancer types (2, 1.4, and 6.8% for breast, cervical, and colorectal cancers, respectively). Findings for 6–12-mo disruption scenarios illustrate that maintaining screening participation is critical to preventing an increase in the burden of cancer at a population level. We provide programme-specific insights into which outcomes are expected to change, when changes are likely to become apparent, and likely downstream impacts. This evaluation provided evidence to guide decision-making for screening programmes and emphasises the ongoing benefits of maintaining screening in the face of potential future disruptions.
Article activity feed
-
-
Author Response
Reviewer 2 (Public review):
A quasi-experimental before and after design as the methodological intention should be stated in the article. Although there are equally powerful alternatives with arguably less-stringent requirements that are appropriate and well-tested for natural experiments such as that intervened by the COVID-19 pandemic given the simulation methods, as of now obtaining the actual stage distribution of cancer and the cancer-specific mortality rates before and after the pandemic is possible for making scientifically valid conclusions based on observed data to support the simulation study.
We agree with the reviewer that a modelled before-and-after analysis would have been informative. However, the required mortality and cancer stage distribution data to inform this analysis is not yet available for …
Author Response
Reviewer 2 (Public review):
A quasi-experimental before and after design as the methodological intention should be stated in the article. Although there are equally powerful alternatives with arguably less-stringent requirements that are appropriate and well-tested for natural experiments such as that intervened by the COVID-19 pandemic given the simulation methods, as of now obtaining the actual stage distribution of cancer and the cancer-specific mortality rates before and after the pandemic is possible for making scientifically valid conclusions based on observed data to support the simulation study.
We agree with the reviewer that a modelled before-and-after analysis would have been informative. However, the required mortality and cancer stage distribution data to inform this analysis is not yet available for Australia. In future, our modelled predictions can be compared to emergent observed national stage and mortality data. The current paper presents estimates that were modelled during rapid-response modelling commissioned by the Australian Government early in the pandemic. Findings therefore demonstrate what could be done with the information available at that time. We have amended, as shown in bold below, the end of the introduction as follows:
“We demonstrate what could be estimated by a rapid response evaluation based on information available early in the pandemic, and discuss how these estimates relate to subsequent observed disruptions to screening. In future, our modelled predictions can be compared to emergent observed national stage and mortality data.”
The screening disruption is the only concerned parameter in modelling the change of cancer progression in this study. But delayed diagnosis after screening as another concern could be possibly affected by the pandemic. This should be taken into consideration in the simulation. The authors also claimed the cancer treatment could also be affected by the pandemic, the evaluation on mortality is therefore not feasible. However, the impacts of COVID-19 pandemic on the delayed treatment and cancer treatment are important issues which should be covered by simulation study.
We clearly state that this is a limitation of the current study. We have added the following sentence to the discussion, lines 377-379.
‘These effects will be incorporated in future modelled evaluations, after careful calibration and validation to observed data, with a view to extending the modelled outcomes to mortality estimates.’
By simulations, the confident intervals for the outcomes should be provided as the requirement to determine the required reliability for the estimates.
The manuscript aims to present indicative estimates for a range of scenarios, with numerous simplifying assumptions as indicated. In this context, generating meaningful uncertainty intervals is not feasible or appropriate.
-
eLife assessment
This study presents important results on the predicted impact of cancer screening disruptions in Australia during the COVID-19 pandemic based on consultation with public health stakeholders. The evidence presented is solid, as simulations were based on several previously validated breast, cervical, and bowel cancer screening decision models, though the scenarios were based on hypothetical disruptions that do not always match experienced disruptions. The work will be of interest to local policy-makers, public health specialists, and cancer epidemiologists.
-
Reviewer #1 (Public Review):
The authors of the current study investigated the effect of the suspension of the Australian breast, bowel and cervical cancer screening program for 3, 6, 9, or 12 months on cancer outcomes and cancer services.
The major strengths of the current study are the usage of the validated Policy1 modelling platform to estimate the effects of delays in the screening program on cancer outcomes. Furthermore, they described a wide range of different scenarios and looked at all three national screening programs together. A clear and detailed description of the screening programs was given. The results are well-described and detailed.
The authors reached their aim. They showed how a disruption of the breast cancer screening program of 12 months led to less screen-detected and interval invasive cancers, and to an increase …
Reviewer #1 (Public Review):
The authors of the current study investigated the effect of the suspension of the Australian breast, bowel and cervical cancer screening program for 3, 6, 9, or 12 months on cancer outcomes and cancer services.
The major strengths of the current study are the usage of the validated Policy1 modelling platform to estimate the effects of delays in the screening program on cancer outcomes. Furthermore, they described a wide range of different scenarios and looked at all three national screening programs together. A clear and detailed description of the screening programs was given. The results are well-described and detailed.
The authors reached their aim. They showed how a disruption of the breast cancer screening program of 12 months led to less screen-detected and interval invasive cancers, and to an increase in the percentage of tumours with a tumour size of more than 15mm or with nodal involvement. In addition, suspension of the bowel screening program for 12 months led to upstaging for 891 tumors. Suspension of the cervix screening program for 12 months let to 27 upstaged tumors, and to 69 extra tumors. On the contrary, suspension of the breast screening for 3 months did not lead to a higher percentage of tumours with a tumor size of more than 15 mm or to a higher percentage of tumors with nodal involvement. Suspension of the bowel screening program for 3 months led to upstaging of 261 tumors, and suspension of the cervical screening program for 6 months led to 21 extra tumors and to 9 upstaged tumors. The conclusion of the authors that 'maintaining screening participation is critical to reducing the burden of cancer at a population level' is therefore not completely correct, as suspension for 3 months might be needed in situations with limited resources and will not have a very large impact on the cancer burden.
This paper predicts upstaging due to the disruptions in the screening program. This information can be used by hospitals so they know what they can expect, and can be used in the future if decisions need to made about suspending the screening program.
-
Reviewer #2 (Public Review):
The report was based on three nation-wide cancer screening programs (breast, bowel, and cervix cancer). This paper attempts to simulate the potential impact of screening disruption on the cancer detection. The authors raised an important concern; that the screening disruption by COVID-19 pandemic would led to an increase in cervical cancer but a reduction in detection of breast and bowel cancer.
There are some issues that must be addressed to ensure the analysis and conclusions can be clearly studied. Importantly, it is not entirely clear if the simulation methodology applied to arrive at a scientific conclusion. The authors could provide more insights on how they will address not only the change of cancer detection but also the driving some improved methods for screening helping return to pre-pandemic levels.
Reviewer #2 (Public Review):
The report was based on three nation-wide cancer screening programs (breast, bowel, and cervix cancer). This paper attempts to simulate the potential impact of screening disruption on the cancer detection. The authors raised an important concern; that the screening disruption by COVID-19 pandemic would led to an increase in cervical cancer but a reduction in detection of breast and bowel cancer.
There are some issues that must be addressed to ensure the analysis and conclusions can be clearly studied. Importantly, it is not entirely clear if the simulation methodology applied to arrive at a scientific conclusion. The authors could provide more insights on how they will address not only the change of cancer detection but also the driving some improved methods for screening helping return to pre-pandemic levels.
1. A quasi-experimental before and after design as the methodological intention should be stated in the article. Although there are equally powerful alternatives with arguably less-stringent requirements that are appropriate and well-tested for natural experiments such as that intervened by the COVID-19 pandemic given the simulation methods, as of now obtaining the actual stage distribution of cancer and the cancer-specific mortality rates before and after the pandemic is possible for making scientifically valid conclusions based on observed data to support the simulation study.
2. The screening disruption is the only concerned parameter in modelling the change of cancer progression in this study. But delayed diagnosis after screening as another concern could be possibly affected by the pandemic. This should be taken into consideration in the simulation. The authors also claimed the cancer treatment could be also be affected by the pandemic, the evaluation on mortality is therefore not feasible. However, the impacts of COVID-19 pandemic on the delayed treatment and cancer treatment are important issues which should be covered by simulation study.
3. By simulations, the confident intervals for the outcomes should be provided as the requirement to determine the required reliability for the estimates.
-
Reviewer #3 (Public Review):
This is an interesting manuscript with an important subject pertaining to the impact of COVID-19 pandemic on various delayed schedules of population-based cancer screening, leading to the reduction of screen-detected cancers and the possible upstaging cancers. The results were assessed by simulation model (Policy I modelling) with the demonstration of Australia scenarios including three major cancers, including breast cancer, colorectal cancer, and cervical cancer.
Assess the impacts of COVID-19 disruption to population cancer screening for three major cancers on short-term and long-term outcomes for policy analysis.
The merit of this study is to provide a series of simulated results under disruption scenarios but the weakness are several-fold including lacking of mortality estimates, inadequate assessments …
Reviewer #3 (Public Review):
This is an interesting manuscript with an important subject pertaining to the impact of COVID-19 pandemic on various delayed schedules of population-based cancer screening, leading to the reduction of screen-detected cancers and the possible upstaging cancers. The results were assessed by simulation model (Policy I modelling) with the demonstration of Australia scenarios including three major cancers, including breast cancer, colorectal cancer, and cervical cancer.
Assess the impacts of COVID-19 disruption to population cancer screening for three major cancers on short-term and long-term outcomes for policy analysis.
The merit of this study is to provide a series of simulated results under disruption scenarios but the weakness are several-fold including lacking of mortality estimates, inadequate assessments and inaccurate reports on missed cancers (interval cancers) and upstaging.
Policy analysis based on disruption scenario through the simulation model would be very informative to guide policy-makers for designing a salvage program to minimize the impacts of COVID-19 disruptions.
Direct reporting data on the empirical disruption scenario instead of relying on the sensitivity analysis of disruption scenario is more transparent and convincing for the public.
-
