The results of our model, indicate that the Portuguese context, the most cost-effective strategy for CRC screening is colonoscopy every 10 years, with the first follow-up colonoscopy performed 3 years after completion of the baseline colonoscopy. The colonoscopy based-strategies (colonoscopy 3/10y and colonoscopy 10/10y) absolutely dominate all stool-based strategies as well as CT-colonography, with ICERs clearly below the WTP threshold of 30,000€ per QALY and the largest net monetary benefits (NMB). Choosing a different strategy resulted in lower effectiveness and higher costs, making such an option not rational and undesirable. The biennial FIT strategy, although positive in reducing CRC incidence and mortality compared with no screening, is the least cost-effective choice among the 5 strategies modelled.
Our findings are consistent with previous findings in the literature, although there are studies with different results [12]. Teldford et al, used a Markov probabilistic model to estimate the most cost-effective strategy for CRC screening in Canada; colonoscopy every 10 years was the most cost-effective strategy with an ICER of 6,133€ per QALY when compared to FIT performed annually [13]. In a study by Barzi et al evaluating 13 screening strategies, including FIT, colonoscopy was the most effective strategy with the highest effectiveness, but also the lowest total cost per patient ($2861) [14]. They showed that annual FIT was dominated by colonoscopy every 10 years. In our study, we found similar results, as annual FIT is absolutely dominated by colonoscopy, with an ICER of – 44,461.97€ per QALY (FIT 1/1y versus colonoscopy 3/10y). The absolute dominance of colonoscopy over all other screening strategies in Portugal, observed in our model, is probably explained by two main factors acting simultaneously in the Portuguese context: (1) relatively low cost of anaesthetic-assisted colonoscopy (compared to other developed countries), and (2) the high reimbursement received by NHS hospitals during the first two years of CRC treatment.
In our model colonoscopy 3/10y is more cost-effective than colonoscopy 10/10y, with an ICER of 802.7€ per QALY, in the base case scenario. Colonoscopy every 10 years is the strategy generally evaluated in CRC models. As far as we know, this is the first time, a colonoscopy 3/10 y strategy has been included in a model, aiming to decrease the number of missing lesions in the baseline colonoscopy[2].
Our results are reinforced by the perfect adherence scenario. Colonoscopy-based strategies continue to dominate all other strategies and the cost-effectiveness ranking remains unchanged. These results are further supported by PSA, in which colonoscopy 3/10y is the most cost-effective strategy in more than 80% of the iterations. The only situation in which FIT is more cost-effective than colonoscopy is for initial adherence for colonoscopy below 39.8% ceteris paribus, reinforcing the importance of health policies to increase adherence.
The colonoscopy adherence rate used in the cost-effectiveness models for base case estimates is variable, from 22–60%[12, 15, 16]. In our study, we assumed the upper limit of this range, as colonoscopy was modelled under anaesthesia assistance (in Portugal, almost all colonoscopies requested by NHS primary care physicians are performed under anaesthesia in an open access basis, free of charge at the point of care), which will very plausibly increase the adherence rate. In 2015, a US survey found that 45.3% of adults aged 50 to 54 years reported screening with either colonoscopy or sigmoidoscopy[2].In our model, the assumption of adherence for initial screening with colonoscopy-based strategies was 60%. Individual adherence can be influenced by many factors, such as, education about screening, test characteristics and physician recommendation. Regarding colonoscopy adherence, two major factor influencing patient compliance is ensure that they will not have of pain and safety.
Our results are in line with previous studies on the clinical impact of different screening strategies on the natural history of CRC. Screening colonoscopy 3/10y, prevented 64% of CRCs (129.7 CRC cases in the unscreened cohort) and gained 337.5 QALYs (per 1000 screened patients); improving screening for a perfect adherence scenario would prevent 80% of the CRCs cases and gain 451.9 QALYs. In a study by Ladabaum et al, the number of cases of CRC per 1000 persons was 105 in the unscreened cohort; colonoscopy every 10 years, averted 60 cases of CRC (57.2% reduction)[17]. Again as expected, the least effective strategy, biennial FIT avoided only 55% of CRCs, and gained only 77.2% of the QALYs obtained by the colonoscopy 3/10y strategy, being more than six times more expensive compared to colonoscopy 3/10y.
As far as we know, our study is the first to estimated cost-effectiveness analysis for the Portuguese context, using the NHS perspective, using a Markov model replicating the natural history of colon cancer, as well as its current treatment according to CRC staging, where the five strategies were superimposed. Areia et al compared only 2 strategies in the Portuguese context: biennial FIT and colonoscopy every 10 years (using no screening as baseline) [18]. They concluded that biennial FIT was the most cost-effective strategy, as colonoscopy presented an ICER (103,633€ per QALY) above the WTP threshold. The study used a societal perspective, which despite strong economic theoretical arguments for its use, is not the perspective used and recommended by most government agencies, such as NICE or the Infarmed - National Portuguese Authority of Medicines and Health Products, due to its lack of consistency and objectivity, as it introduces a large number of discretionary costs[19–21]. Their analysis is limited to 2 strategies, as they argue that including more strategies in the model would compromise the evaluation of the 2 strategies. This argument has no practical or theoretical support, as the quality of the model analysis is not dependent on the number of strategies included, as each strategy is an independent node in a Markov model. In our opinion, FIT 1/1 y should have been included, not least because FIT 2/2y is not endorsed by most medical societies[22]. Moreover in their model they only considered a very low compliance rate for colonoscopy, which seems not realistic given that in Portugal, colonoscopy is performed under anaesthesia, free of charge. The 38% adhesion rate used in the base case estimate was also used as the upper limit in the model's sensitivity analysis; it does not seem plausible that the baseline value is, at the same time, the upper limit in a sensitivity analysis. Furthermore, using the initial adherence rate value of 38%, as the adherence rate for subsequent colonoscopies is surprising. In most studies, the rate of adhesion used for subsequent colonoscopies is higher between 60%-80%[12]. The cost per patient of the 2 modelled strategies is very low: 11.9€ in the FIT strategy, 199.4€ in the colonoscopy strategy and 7.9€ in unscreened patient. This is difficult to understand as the cost of the cancer treatment, is the main driver of the total costs in any CRC screening program, give the price of new drugs and treatments. They should have used, as we did in our model, for objectivity and simplicity, the fixed reimbursement for the first 2 years of CRC treatment that it is being progressively implemented in Portugal. In their work, the cost of CRC per stage is not explicitly reported, as only cost data is showed in a supplementary table, with unit cost for several tests and procedures.
We should note that this study is a a modelling exercise, using multiple health states, patients’ pathways, inputs and assumptions, and using published data. The costs are derived from the Portuguese NHS, so care must be exercised when applying these results to other countries. Despite the limitations, we believe that the study can contribute to the design of better and more efficient screening policies.
Our results, strong suggest, that biennial FIT, the screening strategy used in Portugal, should be abandoned as it is the most expensive and less effective strategy for the NHS, and replaced by one of the two colonoscopy strategies. Finally, the implementation of an organized screening strategy should include a cost-effectiveness analysis for the governments to take more rational decisions and better allocate resources in healthcare.