key: cord-0966742-z80nt15c authors: Mandrik, Olena; Chilcott, James; Thomas, Chloe title: Modelling the impact of the coronavirus pandemic on bowel cancer screening outcomes in England: A decision analysis to prepare for future screening disruption date: 2022-05-06 journal: Prev Med DOI: 10.1016/j.ypmed.2022.107076 sha: e58fa644136f01d802a3c6a775e263c15f6d3519 doc_id: 966742 cord_uid: z80nt15c The English Bowel Cancer Screening Programme invites people between the ages of 60 and 74 to take a Faecal Immunochemical Test every two years. This programme was interrupted during the coronavirus pandemic. The research aimed: (1) to estimate the impact of colorectal cancer (CRC) Faecal Immunochemical Test screening pauses of different lengths and the actual coronavirus-related screening pause in England, and (2) to analyse the most effective and cost-effective strategies to re-start CRC screening to prepare for future disruptions. The analysis used the validated Microsimulation Model in Cancer of the Bowel built in the R programming language. The model simulated the life course of a representative English screening population from 2019, by age, sex, socio-economic deprivation, and prior screening history. The modelling scenarios were based on assumptions and data from screening centres in England. Pausing bowel screening in England due to coronavirus pandemic is predicted to increase CRC deaths by 0.73% within 10 years and 0.13% over the population's lifetime, with excess deaths due to peak in 2023. More deaths are expected in men and people aged over 70. Pausing screening for longer would result in greater additional CRC cases and deaths. Postponing screening for everyone would be the most cost-effective strategy to minimise the impact of screening disruption without any additional endoscopy capacity. If endoscopy capacity can be increased, temporarily raising the Faecal Immunochemical Test threshold to 190 μg/g may help to minimise CRC deaths, particularly if screening programmes start from age 50 in the future. The English Bowel Cancer Screening Programme (BCSP) has been very successful in detecting cancer early and so decreasing colorectal cancer (CRC) mortality in England. [1] [2] [3] Faecal Occult Blood Testing (gFOBT) for people aged 60 to 69 years began in England in 2006. 1 Currently, the BCSP invites people between the ages of 60 and 74 to take a Faecal Immunochemical Test (FIT) every two years. It is expected that the future screening programme will reduce the starting age to 50 years. 1 The coronavirus disease 2019 (COVID-19) pandemic resulted in routine diagnostic delays and suspension of screening for CRC in England for around three months, similar to other countries. 4 , 5 A modelling study, analysing an 84% decrease in referrals via the 2-week-wait urgent pathway, concluded that delays in symptomatic diagnosis in England could result in around 180 to 540 additional deaths over a three-month lockdown period. 6 Sud et. al. (2020) estimated that a fourmonth diagnostic delay would result in more than 20% reduction in CRC Stage 3 survival over the year. 7 Other research assessed that diagnostic delays (screening plus symptomatic detection) in England could result in around 1,500 additional deaths within five years. 4 While no study specifically analysed the impact of the CRC screening pause in England, a modelling analysis predicted around 320-440 additional deaths in the Netherlands, 1,000 in Australia, and 800 in Canada in 30 years with a three-month screening disruption. 5 While a national-level interruption of cancer screening programmes occurred for the first time due to the pandemic, according to clinical experts short-term screening disruptions are not uncommon in individual screening centres in England. These screening disruptions put additional pressure on the endoscopy capacity required to address patient backlogs. Thus, policy makers should have a well thought-through and evidence-based strategy for managing routine public health programmes J o u r n a l P r e -p r o o f during short-term issues delaying screening and national emergencies, such as the COVID-19 outbreak. Previous modelling studies demonstrated long-term benefits and cost-effectiveness of the CRC screening programme in England 8, 9 . This decision modelling study aimed to quantify the impact of screening pauses of different duration, and the actual COVID-19-related screening pause on future cancer outcomes in England. The research also analysed the most effective and cost-effective strategies to re-start CRC screening.. For this analysis, we used a validated model: Microsimulation Model in Cancer of the Bowel (MiMiC-Bowel), developed for previous research. [10] [11] [12] MiMiC-Bowel is an individual patient simulation model built in the R programming language. The model simulates the life course of patients representing the population of England. Each person in the model has a set of individual characteristics, which determines their cancer risk and response to screening and surveillance. The model has a lifetime horizon and takes an English National Health Service (NHS) perspective. The detailed description of the model, its calibration, and validation is reported online. 13, 14 Underpinning the model is a CRC natural history module with nine mutually exclusive h ealth states: Normal Epithelium, Low Risk Adenoma, High Risk Adenoma, CRC Dukes Stage A, CRC Dukes Stage B, CRC Dukes Stage C, CRC Dukes Stage D, CRC Death, and Other Cause Death. In the model, around 85% of CRC develops through adenomas and the rest through serrated pathways, reflected by the transition from normal epithelium directly to CRC. The model incorporates personalised relative CRC risk through the combination of individual risk factors. 14 Once an individual develops CRC, they have a probability of progressing to the next stage. At each stage, there is a probability that an individual will be diagnosed either through screening, surveillance or via symptomatic presentation. It was assumed that after diagnosis CRC stops progressing and individuals start following a disease pathway, which includes treatment costs, utility reductions and reduced survival compared to the general population. The model uses Office for National Statistics CRC survival data from 2013 to 2017 to estimate mortality from CRC by age, sex, cancer stage at diagnosis and time since diagnosis. 15 The model incorporated historical screening through gFOBT and flexible sigmoidoscopy (FS) in order to accurately model the current health states of the population. The FIT uptake, sensitivity, and J o u r n a l P r e -p r o o f specificity were based on data from the English FIT pilot. 2 A variety of sources were used to parameterise screening follow-up and surveillance. 14 The model baseline population is composed of individuals eligible for screening, based on data about sex, age, and Indices of Multiple Deprivation quintile composition in England in 2019 (Table 1) . 16 The individual characteristics of the population were retrieved from the Health Survey for England (HSE) 2014, 17 Legend: IMD -Indices of Multiple Deprivation quintiles, where quintile 1 represents the most deprived and quintile 5 represents the l east deprived. Calibration of natural history disease parameters was against pre-screening data from 2005 (see supplementary Table 1 ). 13 The HSE 2014 population includes EQ-5D values for individuals that are used as baseline quality of life estimates in the modelling. 17 The model includes utility decrements for CRC diagnosis, age, and screening-related harms. Costs used in the previously published model 11, 12 were either inflated to 2019/20 values using the National Health Service cost inflation pay and prices index (costs for faecal tests and CRC treatment) 14, 18 or updated using the most recent National Schedule of Reference Costs (2018/19). 19 The model applied 3.5% discounting to both costs and effects. The parameters of the original MiMiC-Bowel model were converted from annual to three-month cycle lengths to run the analysis (online supplementary detailed modelling methodology). All J o u r n a l P r e -p r o o f calibrated natural history disease parameters, CRC mortality, and other cause mortality were converted to rates and then to probabilities by time using logarithmic and exponential equations. 20 The annual costs of CRC were converted from a one-year cycle to the three-month cycle duration linearly. The screening-related costs and harms were assigned to the cycle when the event occurred. MiMiC-Bowel was externally validated to sensitivity of FIT and FS screening tests in England and recent data on CRC incidence and mortality, and cross-validated to the US and German models (see Supplementary detailed modelling methodology). 13, 21 Stakeholder involvement We analysed the outcomes using probabilistic sensitivity analysis on a population of 18 million people for model runs using three-month model cycles (modelling analysis 1) and nine million people for the analysis using annual cycles (modelling analysis 2). A lifetime horizon was used for all outcomes together with shorter (5, 10, and 20 years) horizons. We also calculated the resource use in the year of the screening restart. Modelling analysis 1. Impact of colorectal cancer disruption on health outcomes We analysed screening disruptions of three, six, nine, and 12 months, plus an approximation of the real screening pause that occurred in 2020. For these scenarios we assumed that screening restart led to a continual postponement of screening for everyone, which is the closest to the actual re-start strategy after the 2020 screening pause. Based on data reporting an increase in other cause mortality during the pandemic (partly due to COVID-19 itself), 22 we also conducted a scenario analysis considering a one percent increased mortality from other diseases during the ten-year period. J o u r n a l P r e -p r o o f The linear trend between the latter two parameters ( Figure 1 ) suggests that for the majority of the centres the delays in screening are ongoing. In most (55%) centres the delays were for roughly three months. To use the data in the state-transition model, we estimated the proportion of the population experiencing approximately zero (less than one-and-half months) delay, three (one-andhalf to four-and-half) months delay, six (four-and-half to seven-and-half) months delay, and nine (seven-and-half to 10.5 months) delay from the total. We assumed that the centres experiencing three to nine months delay had an ongoing impact of the screening pause similar to the modelled hypothetical screening pauses of different lengths described above (i.e. would never catch up with the delayed screening). The centres experiencing less than one-and-half months delay ( Inviting everyone who missed screening within the next cycle, but temporarily increasing the FIT threshold to 190 µg/g during this cycle, which will reduce the positivity compared to the current threshold of 120 µg/g. This will reduce the numbers that will be referred to colonoscopy compared with Scenario 3.1, although will still require some additional colonoscopy capacity compared with undisrupted screening. Pausing screening for three, six, nine, and 12 months increases total CRC cases, stage C & D CRC cases, and CRC deaths ( Table 2) . In general, with longer screening pauses, greater impacts on CRC incidence and mortality are observed. When the screening is paused, CRC incidence initially decreases because of diagnostic delays (Figure 2a Modelling of the actual screening pause based on screening centre data gave results that were slightly larger than the three-months pause screening scenario. The analysis predicted that the screening pause in England will result in more than 270 additional CRC cases (0.13% increase), 880 (Table 3 ). Both Scenarios 3.1 and Scenario 3.2. required additional FIT invites (4 million for both) and colonoscopy capacity: 24,000 for 3.1 and 13,000 for 3.2 in the year of screening restart. The decrement in resource use and delay in diagnosis for some of the patients who have slowly progressing cancer resulted in generally lower discounted costs (online supplementary Figure 4 ). However, since biennial FIT screening is cost-effective, 9, 12 all return-to-screening strategies had negative net monetary benefit for the screened population in England compared with undisrupted screening. The return-to-screening scenarios with smallest negative clinical impact (Scenarios 3) were the most cost-effective ( Figure 5 ). We estimate that the pause in CRC screening that occurred in England due to the COVID-19 outbreak, will lead to around 270 additional CRC cases and 700 additional CRC deaths over the next 10 years, resulting in 6,960 LYs and 3,500 QALYs lost during the lifetime of the screening population. modelling study concluded that a strategy of increasing FIT uptake from 15% to 22% could be an effective approach to improve clinical outcomes. 27 Our study is the first modelling study assessing the impact of COVID-19 specifically on CRC screening, based on actual data, and including multiple CRC outcomes. Maringe et. al. (2020) estimated that delays in CRC screening and diagnosis in England could result in 1,445 to 1,563 additional deaths over the next five years. 4 Loveday et. al. (2021) concluded that if the recommended two-week diagnosis for symptomatic CRC patients is extended to two months, this will result in 653 additional deaths. 28 Our model predicts that within five years 370 CRC deaths could be attributed to the screening pause alone, going up to 700 additional deaths within ten years. The results are generally comparable with those reported for other countries (0.1-0.3% increase over 30-40 years of follow-up without screening catch-up). 5 The study had several limitations. The aim of the project was specifically to assess the screeningrelated impact of COVID-19, so any impact of delay to treatment or symptomatic diagnosis was not incorporated into the model. The subgroup analysis was based on the assumption that screening pauses affected all of the subgroups proportionately, given a lack of data to suggest otherwise, but this may not have been the case. To decrease the computational burden and heterogeneity in predictions of small effects, we modelled the return-to-screening scenarios based upon the 12month screening pause only, although we do not expect conclusions to differ with shorter pauses. Journal Pre-proof The CRC screening pause that occurred due to COVID-19 is predicted to result in a small increase in CRC cases and CRC deaths in the next 10 years, with deaths expected to peak in 2023. Longer CRC screening pauses would be expected to result in greater additional CRC incidence and mortality. Selection of the optimal strategy for screening restart depends on available endoscopy capacity. If no additional capacity is available, postponing screening for everyone is likely to be the optimal strategy for minimising deaths and maximising cost-effectiveness. Report of the independent review of adult screening programmes in England: Appendix D: NHS Bowel cancer screening programme: NHS England Increased uptake and improved outcomes of bowel cancer screening with a faecal immunochemical test: results from a pilot study within the national screening programme in England Outcomes of the Bowel Cancer Screening Programme (BCSP) in England after the first 1 million tests The impact of the COVID-19 pandemic on cancer deaths due to delays in diagnosis in England, UK: a national, population-based, modelling study Impact of the COVID-19 pandemic on faecal immunochemical test-based colorectal cancer screening programmes in Australia, Canada, and the Netherlands: a comparative modelling study Effect of delays in the 2-week-wait cancer referral pathway during the COVID-19 pandemic on cancer survival in the UK: a modelling study Quantifying and mitigating the impact of the COVID-19 pandemic on outcomes in colorectal cancer Optimizing the Design of a Repeated Fecal Immunochemical Test Bowel Cancer Screening Programme With a Limited Endoscopy Capacity From a Health Economic Perspective Optimising Bowel Cancer Screening Phase 1: Optimising the cost effectiveness of repeated FIT screening and screening strategies combining bowel scope and FIT screening: ScHARR, University of Sheffield Calibrating Natural History of Cancer Models in the Presence of Data Incompatibility: Problems and Solutions The Costs and Benefits of Risk Stratification for Colorectal Cancer Screening Based On Phenotypic and Genetic Risk: A Health Economic Analysis Should colorectal cancer screening start at different ages for men and women? Cost-effectiveness analysis for a resource-constrained service. Cancer Rep (Hoboken) 2021:e1344 Calibration and Validation of the Microsimulation Model in Cancer of the Bowel (MiMiC-Bowel), an Individual Patient Simulation Model for Investigation of the Cost-effectiveness of Personalised Screening and Surveillance Strategies. ScHARR HEDS Discussion Papers Development of the Microsimulation Model in Cancer of the Bowel (MiMiC-Bowel), an Individual Patient Simulation Model for Investigation of the Cost-effectiveness of Personalised Screening and Surveillance Strategies Adult Cancer Survival Tables: One and Five Year net survival for adults diagnosed between Populations by sex, age group and Index of Multiple Deprivation (IMD) quintile, England Estimating Transition Probabilities from Published Evidence: A Tutorial for Decision Modelers Changes in all-cause and COVID-19 mortality over time, England and Wales: deaths occurring between 28 December The national FIT-based colorectal cancer screening program in the Netherlands during the COVID-19 pandemic Colorectal Cancer Screening and COVID-19 Impact of the COVID-19 Pandemic on Colorectal Cancer Screening: a Systematic Review The Lancet Gastroenterology H. Resuming bowel cancer screening post-COVID-19 Model-Based Estimation of Colorectal Cancer Screening and Outcomes During the COVID-19 Pandemic Prioritisation by FIT to mitigate the impact of delays in the 2-week wait colorectal cancer referral pathway during the COVID-19 pandemic: a UK modelling study Olena Mandrik: Conceptualization, Methodology, Software, Investigation, Data curation, Funding acquisition, Writing-Original draft preparation We are grateful for the contribution from the following experts who participated in the We are also grateful to the patient experts who consulted on this research: Kim Page, Margaret Johnson, Debjani Chatterjee, Patricia Fairbrother, and Jacqui Gath.We are also grateful to the National Institute for Health Research UK for funding this research (NIHR202316). Institute for Health Research UK (NIHR202316). The funder had no impact on the outcomes of the study, data analysis or interpretation. The authors have no conflict of interest to report.