America, We Are Confused: The Updated U.S. Preventive Services Task Force Recommendation on Colorectal Cancer Screening

We are surprised by the latest update of the recommendations for colorectal cancer screening from the U.S. Preventive Services Task Force (USPSTF) (1). Contrary to the principles of evidence-based medicine, the guidelines provided equally strong recommendations for tests with very different quality of evidence for benefits and harms. In recent years, most guideline makers have adopted transparent strategies to grade quality of evidence and strength of recommendations. Consequently, in the European Union colorectal cancer screening guidelines, we have graded the strength of recommendations for each screening strategy on the basis of the quality and trustworthiness of the supporting evidence (2). Strategies with limited high-quality evidence have been dismissed or have been recommended with a clear warning about the lower trustworthiness of the recommendation compared with strategies with a more robust evidence base (2). We believe that guideline makers should inform readers about the degree of trustworthiness behind their recommendations. If high-quality randomized trials exist for a screening strategy, there is a clear expectation for strong recommendations. On the other hand, for screening strategies without evidence from high-quality randomized trials, recommendations should be issued with greater caution and should include an explicit justification of why the balance of benefits and harms was considered to be positive. The latest colorectal cancer screening recommendations from the USPSTF also applied a grading strategy (1). However, the USPSTF seems to have graded all screening tests as a single test rather than distinguishing among them. This is surprising because the tests differ greatly in terms of mode of action, invasiveness, and quality of evidence for effectiveness and cost-effectiveness. Despite the fact that 5 of the 7 strategies recommended by the USPSTF (colonoscopy, fecal immunochemical testing [FIT] for occult blood, computed tomography colonography, FIT DNA testing, and the combination of sigmoidoscopy and annual FIT) have not been subjected to clinical effectiveness trials, they are as strongly recommended as the 2 that have performed favorably in such trials (guaiac-based fecal occult blood testing [FOBT] and sigmoidoscopy alone). There may be good reasons for strong recommendations even in cases of low quality of evidence of benefit for a screening strategy, but the USPSTF did not explain them. Of note, the USPSTF recommended sigmoidoscopy in combination with annual FIT screening, citing a Norwegian randomized trial that compared sigmoidoscopy alone versus sigmoidoscopy plus FIT (3). However, the recommendation does not correspond with what the trial showed. Colorectal cancer mortality did not differ significantly between sigmoidoscopy alone and sigmoidoscopy plus FIT (3). There was a trend toward better effectiveness for the combined strategy for colorectal cancer mortality, but there was an opposite trend (in favor of sigmoidoscopy alone) for incidence (3). Also, the trial tested a single FIT (at the time of sigmoidoscopy). No trials have tested sigmoidoscopy plus annual FIT, which is the strategy the USPSTF recommended. In recent years, the medical community has learned tough lessons in other areas of cancer screening (such as prostate-specific antigen testing and mammography) because estimates from observational and modeling studies have not translated into clinical effectiveness for cancer mortality in clinical trials or screening programs. Cancer screening involves a close balance of benefits and harms. The quality of modeling studies depends on the quality of the imputed estimates for benefits and harms. Because few high-quality data are available on the magnitude of benefits and harms for 5 of the tests recommended by the USPSTF, modeling results are naturally uncertain and have low trustworthiness. Modeling has an important role in medical research but should not serve as the main basis for strong recommendations for screening tests or strategies, thus subjecting millions of presumably healthy persons to invasive procedures. Small changes in effectiveness or adverse event rates in real life compared with modeling estimates can have large consequences for the benefitharm ratio of a screening strategy. The European Union guidelines recommend introduction of new cancer screening tests in routine healthcare only after they have been evaluated for efficacy in randomized controlled trials (2). This approach means that of the 7 strategies recommended by the USPSTF, only sigmoidoscopy and FOBT (guaiac-based FOBT or FIT [the modern version]) can currently be recommended for routine use in Europe. The Table shows the 7 strategies endorsed by the USPSTF and their recommended use in our countries. Table. Colorectal Cancer Screening Strategies Recommended by the USPSTF and Their Recommended Use in Italy, Norway, Poland, Spain, and Sweden We encourage implementation of colorectal cancer screening and support to give the strongest recommendations for the strategies that have been found to be effective in clinical trials. We use modeling data in our decision-making process but do not consider them to be equivalent to data derived from well-performed clinical trials. Most important, to overcome the challenge of performing randomized trials in environments of widespread screening activity, we have started to evaluate promising strategies in clinical trials embedded in colorectal cancer screening programs. In the current era of widespread screening activities, performing the traditional, independent randomized trials that we have done in the past to generate the current evidence for FOBT and sigmoidoscopy screening would be difficult, chiefly because of contamination and the lack of valid control groups. To generate high-quality evidence for effectiveness of new screening strategies, we consider the integration of high-quality clinical trials into ongoing screening programs as the only viable solution. In Europe, several population-based clinical trials with cancer incidence and mortality as end points and more than 1000000 total enrolled participants are under way to determine the effectiveness of the most promising screening strategies (410). Some of the most recent (in Finland, Poland, and Norway) are embedded in public screening programs (46). We hope these trials will give some of the answers we lack to provide individuals, decision makers, and health care workers with the evidence base needed for informed decision making. We believe that guideline makers have an important role in facilitating this development and encourage them to participate in the process.