Our analyses of age- and sex-specific incidence and mortality of CRC in the US and 10 other large countries from different parts of the world indicate that the lower incidence and mortality among women quite consistently translates to an age difference of approximately 4–8 years at which comparable levels of risk are reached. Colorectal cancer incidence and mortality at various ages are closely related to potential benefits of screening, which have to be weighed against costs and potential adverse side effects in choosing the age of screening initiation. Our analysis suggests that the balance in favour of screening is likely to be reached several years later among women than among men. This finding is supported by a recent study from Poland among more than 50
000 participants of a colonoscopy-based screening programme, where prevalence of advanced adenomas was higher at each age among men than among women, prompting the authors to the conclusion that gender-specific CRC-screening recommendations may be warranted (Regula et al, 2006
). Furthermore, there are indications both from our analysis and from the literature (Fernandez et al, 2001
) that the gender difference in the epidemiology of CRC has steadily increased during the last few decades. These results may therefore have important implications for the offer of CRC screening programmes and their optimisation in terms of cost effectiveness.
The choice of different, risk adapted ages at initiation of screening is well accepted and established for CRC risk factors other than gender, in particular a history of CRC before age 60 in a first degree relative (Winawer et al, 2003
; Schmiegel et al, 2004
; Smith et al, 2006
). Although the relative risk of CRC among people with such a family history compared to those without is larger than the relative risk of men compared to women (Johns and Houlston, 2001
), the prevalence of the former risk factor in the population is much lower than the ‘prevalence' of male gender. These patterns suggest that appropriate differentiation of age at initiation of CRC screening by gender might be similarly or even more relevant from a public health point of view than the widely practiced differentiation by family history.
Another important epidemiological aspect that might be of importance in the choice of age at initiation of CRC screening is the differential distribution of CRC location among women and men. The proportion of cancer in the distal colon and rectum is considerably lower among women than among men (Stewart et al, 1983
; Bonithon-Kopp and Benhamiche, 1999
; McCashland et al, 2001
). Therefore, the sex difference in distal CRC occurrence is even larger than the sex difference in overall CRC occurrence. These patterns suggest that age differences may even be more relevant for initiation of screening programmes primarily based on sigmoidoscopy than for screening programmes primarily based on stool tests or colonoscopy.
Epidemiological data on the occurrence of colorectal neoplasms are important, but are not the only factors to be considered in the choice of the age range at which screening is offered. Another important factor is remaining life expectancy (Ko and Sonnenberg, 2005
; Lin et al, 2006
), which is of primary relevance for a potential upper age limit for CRC screening. Given that life expectancy is generally higher among women than among men, there are further reasons to define gender-specific age ranges for CRC screening.
Considering potential gender differences in recommended age ranges for CRC screening, nonepidemiological criteria, such as complexity of guidelines, also have to be taken into account (Lieberman, 2005
). One might argue that gender-specific recommendations might add another layer of complexity, which could be a barrier against use of CRC screenings. However, from the patients' point of view, schedules for cancer screening are gender-specific anyway, given that some of the most widely used screening measures refer to female (breast, cervical) and male (prostate) cancers. From the point of view of health-care providers, initiation of CRC screening at different ages for women and men (e.g. 5 years apart) would not seem to be too much of a challenge either.
In the interpretation of our results, the following limitations should be kept in mind. Given the lack of national CRC incidence data for many countries, CRC-mortality data were used along with CRC incidence data in our comparative analyses. As already mentioned, CRC mortality should be a good surrogate parameter given that survival of female and male CRC patients is essentially the same (Bossard et al, 2007
). In fact, exactly the same age differences by sex were seen for incidence and mortality in our analyses for the US, where both measures were looked at. Another potential limitation is that in some of the countries included in this analysis, some form of CRC screening has already been practiced during the calendar years included in the analyses. This may have affected CRC incidence and mortality to some extent. Furthermore, differential participation in CRC screening among women and men might have affected the gender differences reported in our analysis. However, in the years under investigation, the overall impact of CRC screening is likely to have been limited, and gender differences in screening utilisation quite small compared to the major gender difference in CRC incidence and mortality. In particular, participation in endoscopic screening examinations was slightly lower rather than higher among women compared with men in the US (Seeff et al, 2004
; Lieberman et al, 2005
; Meissner et al, 2006
), and could thus not explain the lower CRC mortality among women.
Our analysis only considers the ‘net differences' in CRC incidence and mortality between women and men, which might be owing to a variety of reasons. There are suggestions that hormonal effects, both up to menopause and through hormonal replacement therapy (HRT) may protect from or delay development of CRC (Beral et al, 2002
). To the extent that the recent reduction in HRT use following publication of the results of the Women's Health Initiative Randomised Trial (Rossouw et al, 2002
; Hersh et al, 2004
) and the Million Women Study (Beral, 2003
; Faber et al, 2005
) might increase CRC incidence among women, the previous trend of an increasing gender gap in CRC incidence and mortality might be slowed down or possibly reversed.
Our analyses do not allow a general recommendation regarding the best age for initiation of CRC screening. The latter is likely to vary between populations, owing to between-population variation in CRC incidence and mortality, and in CRC screening and treatment costs. Our results suggest, however, that the optimal age for screening initiation is likely to be around 5 years higher for women than for men within populations, assuming that screening is equally effective in women and men. This could imply either postponement of age at initiation of screening among women or advancement of age at initiation among men compared to nongender-specific-screening schemes. It should also be kept in mind that our results pertain to people at average risk of CRC, and they should therefore not be generalised to specific screening strategies for high-risk groups.
In summary, our results suggest that gender specific differentiation of age at initiation of CRC screening by about 5 years might help to utilise screening resources in a more efficient manner. Gender specific screening schedules should therefore deserve careful attention in the design and evaluation of CRC screening programmes.