Result card

  • EFF4: What is the effect of FIT versus gFOBT for CRC screening on the mortality caused by CRC?

What is the effect of FIT versus gFOBT for CRC screening on the mortality caused by CRC?

Authors: Jesús González-Enríquez, Francesca Gillespie, Stefania Lopatriello, Iñaki Imaz

Internal reviewers: Pseudo125 Pseudo125

Refer to domain search and domain methodology section.

We have not identified any study that compares FIT vs gFOBT in terms of mortality. The identified studies that provide mortality information compared screening using FIT vs no screening. A controlled trial {3} and three observational studies {4, 6, 7} examined mortality effect of the use of FIT for colorectal cancer screening versus no screening. Therefore no direct evidence comparing FIT vs gFOBT in the context of a population based colorectal cancer screening program is available. Because of the lack of direct available evidence to answer the research question a GRADE profile has not been elaborated.

The tables 1 to 5 extract and evaluate the most important results of the four above mentioned studies. Only one of them is a randomised controlled trial{3}, two are case-control studies {6, 7} and one is an observational and comparative retrospective study{4}. The randomised controlled trial compares the use of one round Reverse Passive Hemagglutination (RPHA) immunochemical test along with an individual attribute degree value score versus no screening. This trial, which analyses the screening program conducted in Jiashan (China), presents several bias risks and it is no focused on our specific research question.

First at all, the index test differs from the test consider in our research question because the FIT was used together with a quantitative individual risk assessment method (Attributive Degree Value). On the other hand, the reference standard was initially sigmoidoscopy and colonoscopy only for positive sigmoidoscopies and for positive FIT for those who FIT was repeated. Individuals with a positive FOBT were asked to undergo flexible sigmoidoscopy (FS). If FS failed to detect colorectal lesions, the participants were asked to repeat the FOBT. Those without a lesion found by FS but with a positive repeated FOBT were re-examined by 150-cm colonoscopy to confirm the results. The no use of colonoscopy for all the positive FIT could lead to an underestimation of the colon cancer cases and may explain the better results for detection of rectal but no for colon cancer. Double reference standard can lead to a differential verification bias. That occurs when some of the index test results are verified by a different reference standard. This usually occurs when patients testing positive on the index test receive a more accurate, often invasive, reference standard than those with a negative test result. In this case, the results are overestimated.

The controlled trial is a cluster randomized trial based on townships allocation, with a high risk of bias in several key domains of the Cochrane risk of bias table (randomization, allocation concealment and blinding). Twenty one townships were matched by population size and age distribution into 10 pairs. In each pair, a table of random digits was used to allocate to the screening or to the control group. The unit of analysis was individuals, but there was no inter-cluster correction that would minimize the selection bias risk. On the other hand, there was no participant and personnel blinding, which implies a high risk of performance bias.

This trial obtained no significant differences for colorectal and colon, but significant for rectal cancer mortality, when compared those FIT based screened vs. those no screened. The 8 year cumulative colorectal cancer mortality rate per 1,000 was 2.08 for the FIT group (95% CI: 1.96-2.18) and 2.44 (95% CI: 2.33-2.55) for no screening (p=0.19). For the specific colon cancer mortality the 8 year cumulative mortality rate per 1,000 was 0.90 for the FIT group (95% CI: 0.83-0.97) and 0.83 for the no screening group (95% CI: 0.76-0.90) (p=0.222). The 8 year cumulative rectal cancer mortality rate per 1,000 was significantly different (p<0.05) for the FIT group (1.10; 95% CI: 1.02-1.18) vs the no screening group (1.61; 95% CI: 1.52-1.70).

The rest of the studies are observational retrospective studies with high risk of bias in most of the domains of the Cochrane risk of bias table. Two case-control studies compare, in the context of two different screening programs performed in two Japanese counties, the Odds Ratio of dying due to colorectal cancer for those screened within 1 to 5 years of case diagnosis vs. those no screened. The last one is a retrospective observational study that compares the five years survival rate of 194 screened detected colorectal cancer subjects versus 352 routinely detected (no screening) colorectal cancer at Hirosaki Hospital (Japan). Any of these observational studies provided high quality evidence of the effect of FIT versus gFOBT on the mortality caused by colorectal cancer in the context of a population based colorectal cancer screening program.

We have observed a similar reduction in CRC mortality in RCT of FIT versus no screening and in RCT of gFOBT vs no screening. If we compare the only one randomized clinical trial on CRC mortality in a population invited to FIT versus no screening {3} with the meta-analysis of four RCT on CRC mortality of gFOBT versus no screening {8}, we observe similar results but with a wide confidence interval. The meta-analysis used the Peto method and effect fixed modelization for obtaining an Odds Ratio of 0.84 (IC 95%: 0.78-0.9). Using the same method for estimating the effect of the unique RCT that compared FIT versus no screening the Odds Ratio would be 0.85 (CI 95%: 0.70-1.03). Table 7 {EFF Appendix 3}.

González-Enríquez J et al. Result Card EFF4 In: González-Enríquez J et al. Clinical Effectiveness In: Jefferson T, Cerbo M, Vicari N [eds.]. Fecal Immunochemical Test (FIT ) versus guaiac-based fecal occult blood test (FOBT) for colorectal cancer screening [Core HTA], Agenas - Agenzia nazionale per i servizi sanitari regionali; 2014. [cited 16 June 2021]. Available from: