Social inequalities in a population based colorectal cancer screening programme in the Basque Country

This study reveals socioeconomic inequalities in a number of quality indicators for the Basque Country's colorectal cancer screening programme between 2009 and 2011.

Men in the most disadvantaged socioeconomic class evidence the lowest participation rate, but also the highest rates of premalignant and malignant lesions, between 23 and 55 % higher than the least disadvantaged quintile. Among women, those in the groups with the lowest and highest socioeconomic status participate the least, and no differences are observed in the rate of people with lesions in the most disadvantaged groups. No differences in PPV are detected between socioeconomic groups, either in men or in women, and neither are there any differences between groups in the percentage of colonoscopies performed after a positive FIT result.

Overall participation rate (64.3 %) is close to the 65 % target set by the 2011 European Guidelines [26], although it is lower in men than in women, and lower in the youngest and oldest groups. A similar pattern is found in other CRC screening programmes [15, 19, 27‐29].

Attempting to explain this differential behaviour by sex and age, some studies postulate that men might be less interested about their health, as well as being afraid of the diagnostic test, while women might assume the role of caregiver, leading more of them to worry about their health, for the sake of those around them [15, 16]. The influence of a dominant societal perception of masculinity has been described as an important factor in explaining inferior participation among men, since CRC screening entails the risk of having to undergo an invasive procedure - colonoscopy- that might conflict with normative “male” beliefs [30, 31]. In younger people, feeling healthy and less vulnerable to the disease could be a barrier to participation. Finally not being aware of the importance of screening or not having had their doctor recommend the screening process are some factors that might reduce participation in the older age groups [15, 16]. These hypothetical explanations should be treated with caution, however, since it is widely recognised that knowledge and health beliefs have a limited capacity to explain people's actual behaviour, and underlying motives are particularly difficult to ascertain [32].

The rate of people with premalignant and malignant lesions detected in colonoscopy is three times higher in men than in women. In the older group, it is 2.6 and 1.7 times higher than the younger group, for men and women respectively (5.02 and 2.4 times higher in the older group if we only take into account rates of people with malignant lesions). These figures corroborate the strong association of age and sex with the probability of detecting a premalignant or malignant lesion that has been reported in other studies [33‐36].

The PPV for detection of premalignant and malignant lesions increases with age and is significantly higher in men than in women. These patterns are already known and are probably due in large part to differences in prevalence between certain subgroups [37‐39].

The proportion of colonoscopy performed following a positive FIT result was not associated with age in either sex. Similar results were found by other authors (36-steel, 44-Dupont-Lucasa). The high compliance of the procedure in all the age groups (above 91.5 %) could have played a role in the lack of association.

In both sexes the socioeconomic groups with the lowest participation were, paradoxically, both the least and the most disadvantaged. The former could be due to greater access to private healthcare among the more privileged social classes. In fact public health services are available for all population, but the least deprived normally contract a private insurance as well, where colonoscopy is offered as an opportunistic screening. As already noted above, in reference to the research by Williams, explaining low participation among the most disadvantaged social classes is not a simple task. Lifestyles are composed mostly of actions performed automatically without forethought, with habits and the pursuit of social distinction being the key factors for explaining health and lifestyle, which accounts for the substantial gap between health knowledge and behaviour [32]. Caution is required, therefore, when assessing specific explanations postulated in other articles, such as greater fear of screening, the perception that it is harder to perform, doubts over whether screening is beneficial, lack of knowledge about the test, difficulty in understanding written information and lack of social support [15, 16, 19, 20].

While the published results of other screening programmes show that a lower proportion of those from more disadvantaged social strata attended for colonoscopy after a positive FIT result, it is not the case here [35, 40‐42]. Participation among disadvantaged socioeconomic groups is a sign of good programme implementation. Dupont-Lucas et al., in a study that also showed no socioeconomic differences in the percentage of colonoscopies after a positive test, suggest that the voluntary nature of the programme (which is also the case in the present example) allows those members of disadvantaged social groups who would not have been willing to undergo colonoscopy to opt out [43]. On the contrary, the least deprived quintiles show the lowest percentage of colonoscopy for diagnostic confirmation, possibly because a greater proportion are able to access private clinics, whose data is not available to the current study.

The premalignant and malignant lesion detection rates show an inverse association with socioeconomic status in men but not in women. The influence of socioeconomic status on the incidence of CRC is not clear in the literature. Recent systematic reviews and large prospective studies show mixed results. While it appears that in the United States and Canada, lower socioeconomic status is associated with higher rates of CRC, especially in the proximal colon, the tendency in Europe seems to be in the opposite direction, i.e., lower incidence of CRC in lower socioeconomic strata [44‐46]. Studies with a higher incidence of CRC among lower socioeconomic strata or those with less education, point to a higher prevalence of modifiable risk factors associated with CRC such as smoking, excessive alcohol consumption, obesity, low levels of physical activity or non-adherence to a Mediterranean diet, as well as greater psychological stress due to socioeconomic status, which may lead to increased susceptibility to disease in general. Moreover, opportunistic CRC screening, which is performed with greater frequency in higher social classes, leads to a higher percentage of cancers avoided in this group, due to lesions being detected in premalignant phases [45, 47, 48]. For the European studies that show a lower incidence of CRC in the lowest socioeconomic strata, it is argued that in addition to a lesser influence of the differential CRC screening effect in Europe, people also have better dietary habits or adhere more closely to a Mediterranean diet, especially in rural areas and southern countries [45‐46].

In the Basque Country, according to the 2013 Basque Health Survey, the proportion of men who are smokers, obese or sedentary increases with decreasing social class or education level. Fruit and vegetable intake is also lower in the lower socioeconomic strata, while it seems that the only risk factor that occurs in greater proportion in the higher strata, is that of dangerously high levels of alcohol consumption [49]. The higher prevalence of risk factors in male members of the most disadvantaged groups in the Basque Country may partly account for the results obtained in this study. But not so in the case of women, where no inverse association between social class and lesion detection exists, findings similar to those published by Oliphant et al. with data from a CRC screening programme in the west of Scotland [50]. The authors of the Scottish study could not provide a clear explanation for this differential behaviour between the sexes, although they raised a number of possible reasons, including the possibility of some differences in risk factors between the sexes, such as excessive alcoholism in men or less physical activity in women. In the present study, the 2013 Basque Health Survey reveals that in the Basque Country while tobacco consumption among men increases as social class decreases, no pattern can be seen in women from the age of 45. Nevertheless, the highest prevalence of other risk factors occurs in the most disadvantaged social groups, meaning that the social inequalities in the screening results cannot be explained through these risk factors alone.

Limitations of this study include the fact that, during the study period (2009–2011), the centres participating in the CRC screening programme were not chosen at random, but by the ability of their corresponding referral hospitals to assume the task of performing colonoscopy for screening purposes. This does not preclude broad representation from all socioeconommic groups, since all three provinces are represented and more than 200,000 eligible individuals were included, representing about 50 % of the target population (not taking exclusion criteria into account), with sufficient numbers of people in all relevant variables. It seems, therefore, that the included target population would have similar differential characteristics to the rest of the hypothetical target population of the Basque Country. Further analysis should nevertheless be performed on the definitive data covering 100 % of the population (2014), to assess whether the behaviour is reproduced throughout the entire target population. Moreover, as mentioned above, the non-participant population would most likely have different clinical, social and cultural characteristics to the participating population, and their inclusion might modify the results of this study. Studies that included this non-participant population would be necessary to clarify this question. Finally, using aggregated census tract data to assign socioeconomic status to each individual may lead to incorrect classification of the socioeconomic level in some cases, altering the results to a greater or lesser extent. Numerous studies have, however, demonstrated the use of socioeconomic data of small areas as an approximation to the socioeconomic status of individuals in order to detect health inequalities [51‐54]. The absence of individual socioeconomic data has thus been overcome by allocating to subjects the characteristics of the census tract of residence.