Pitfalls in interpreting flexible sigmoidoscopy screening study results: some lessons from mammography breast screening still not learnt
Judging from some rapid responses, I feel the need to highlight a few important issues regarding this paper so as to emphasise key aspects involved in the interpretation of results.
1. The control group is "No screening"
This is clearly stated in the abstract by the authors themselves: "Individuals were randomised to receive flexible sigmoidoscopy screening once only (SCORE and NORCCAP) or twice (PLCO), or receive usual care (no intervention)."
The paper is an analysis of (general) population based CRC (ColoRectal Cancer) screening by flexible sigmoidoscopy vs "usual care".
The usual care for the three trials are:
US Prostate, Lung, Colorectal and Ovarian cancer screening trial (PLCO) - no organised screening (over 3 - 5 years)
Italian Screening for COlonREctum trial (SCORE) - no screening
NORwegian Colorectal CAncer Prevention trial (NORCCAP) - those not offered screening
Note that each trial has a different set of exclusion criteria (see paper for details).
Therefore the authors' conclusion:
"Flexible sigmoidoscopy is an effective tool for colorectal cancer screening in men and younger women. The benefit is smaller and not statistically significant for women aged over 60; alternative screening methods that more effectively detect proximal tumours should be considered for these women."
is a correct one, as long as the readers understand that the "control" is those with no organised screening.
It is important to remember that the paper does not compare flexible sigmoidoscopy (FS) with other established screening programs found in various parts of the world: for example, FOBT using guaiac test (gFOBT) or fecal immunochemical test (FIT), or colonoscopy.
For countries with an already established screening program (with gFOBT or FIT), the results of this research cannot be used to compare efficacies of FS as a CRC screening tool vs incumbent screening program. Similarly the value of RRR or ARR (relative or absolute risk reduction) is not meaningful if one attempts to use these numbers to compare with the usual screening program currently in place; in the case of using ARR, one must be certain the the control population for FS and standard screening program are comparable (ie, both have a similar incidence of diseases).
An example of the pitfall from using ARR values from different studies is illustrated in a commentary by Belaya on trial comparison on osteoporosis (ref 1).
2. FS does not change mortality rate from proximal CRC
The authors found FS screening has no effect in either men or women on mortality from colorectal cancer due to proximal cancers. Subclass analysis found statistically significant reduction proximal CRC mortality rate in men and young(er) women but not women > 60 years. Given that some 30-40% of CRC are found in proximal colon and the paper does not differentiate the type and aggressiveness of CRC detected at the time of diagnosis, it is possible that FS screening simply results in earlier diagnosis of slower proximal CRC with less aggressive prognosis which would not have necessarily resulted in premature death if it is diagnosed after the cancer become symptomatic.
3. Possible explanation for age/gender differences in mortality rates
Although the study found colorectal cancer mortality was reduced by 27% , 33% in men and 18% in women, it is noted that FS reduced colorectal cancer mortality in the distal colon in men but not significantly in women. Conversely while there was no effect in either men or women on mortality from colorectal cancer due to proximal cancers (as stated earlier) in a subgroup analysis of 55-64 year age group, colorectal cancer mortality was statistically significantly reduced in men and in younger women, but not in women aged 60 years and older.
Various explanations have been proposed: proximal advanced neoplasia without distal lesions are more common in women than in men, technical issues with the procedures for sigmoidoscopy and subsequent colonoscopy (for those women who test positive at screening) relating to women's age, more men (than women) referred for colonoscopy due to findings at screening, and differences in the location of colorectal cancers in younger (distal lesion) versus older women (predominately proximal lesions).
However, keep in mind that the presented data are in terms of relative risk ratios and not absolute values. The reason why FS has made no difference in mortality rates from both proximal and distal CRC may in fact not be due to poorer performance of FS in screened women (especially those in older age groups) but because unscreened older women may behave differently from other groups such that they seek help earlier for vague or poorly differentiated symptoms from CRC. Women, especially those in mature age groups are more likely to see doctors for their health condition earlier than men (Ref 2). Data presented in all 3 papers under review (Ref 3-5) did not list mortality rates by gender stratified according to age group in FS and usual care (no screening). Therefore it is uncertain if the effect of gender behaviour actually reduced the apparent efficacy of FS screening when compared to (better than expected) results in older women who have no organised screening in place but whose proactive behaviour means earlier presentation, diagnosis and management of CRC.
Are these issues anything new?
These points of contention are in part similar lessons learnt from various analysis of screening programs for other conditions; the controversies surrounding the re-evaluation of the efficacy of mammography screening programs certainly stands out when considering issues relating to how to deal with information from screening program data.
More importantly the real question should be: have we learnt the lessons yet?
1. Belaya ZE. Should we really compare absolute risk reduction in different trials on osteoporosis: comment on the article by Ringe JD and Doherty JG. Rheumatol Int. 2011 Dec;31(12):1669-71. doi: 10.1007/s00296-010-1626-8. Epub 2010 Dec 3.
2. Vlassoff C. Gender Differences in Determinants and Consequences of Health and Illness. J Health Popul Nutr. 2007 Mar; 25(1): 47–61. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3013263/
3. Holme Ø, Løberg M, Kalager M, et al. Effect of flexible sigmoidoscopy screening on colorectal cancer incidence and mortality: a randomized clinical trial. JAMA2014;356:606-15. doi:10.1001/jama.2014.8266 pmid:25117129.
4. Schoen RE, Pinsky PF, Weissfeld JL, et al. PLCO Project Team. Colorectal-cancer incidence and mortality with screening flexible sigmoidoscopy. N Engl J Med2012;356:2345-57. doi:10.1056/NEJMoa1114635 pmid:22612596.
5. Segnan N, Armaroli P, Bonelli L, et al. SCORE Working Group. Once-only sigmoidoscopy in colorectal cancer screening: follow-up findings of the Italian Randomized Controlled Trial—SCORE [correction in: J Natl Cancer Inst 2011;103:1903]. J Natl Cancer Inst2011;356:1310-22. doi:10.1093/jnci/djr284 pmid:21852264.
Competing interests: No competing interests