During 2018, breast cancer was the leading cause of incidence and mortality in Colombia. Most new cancer cases (~50.00%) occurred in Central and Bogotá D.C. regions as well as in the contributory insurance, conversely, cervical cancer was more frequent in the subsidized. Regards incidence by region, there was a homogeneous pattern for breast and prostate cancer. Some regions showed significantly higher incidence rates than national, varying by type of cancer, as follows: Bogotá D.C. and Central for breast cancer, Bogotá D.C. for prostate cancer and “other departments” for cervical cancer. Consistently, Eastern region had both, incidence and mortality rates, significantly lower than national for all types of cancer. By municipalities, Agua de Dios in Cundinamarca was among the highest incidence rates for breast and cervical cancer and they were significantly higher than national. Mortality patterns by municipality were highly heterogeneous.
There are differences in ASR for all types of cancer between CAC and GLOBOCAN estimations. Generally, GLOBOCAN rates overestimate both, incidence and mortality rates reported by CAC. In 2018, the biggest difference was observed for prostate cancer (CAC: 11.34 vs. GLOBOCAN: 49.80), followed by breast (CAC: 18.69 vs. GLOBOCAN: 44.10) and cervical cancer (CAC: 5.93 vs. GLOBOCAN: 12.70). The same pattern was found for mortality in prostate (CAC: 7.58 vs. GLOBOCAN: 12.00), breast (CAC: 10.48 vs. GLOBOCAN: 11.90) and cervical cancer (CAC: 4.31 vs. GLOBOCAN: 5.70). Differences are consistent with a previous study conducted by the CAC and could be explained by the data sources and methodology used. In the case of GLOBOCAN, data on incidence come from four Colombian city-based registries that have been classified as high quality and the calculation is based on projections, whereas CAC information provided by health insurers is updated yearly (12).
In both sexes combined, the leading causes of new cases of cancer and deaths in Colombia are different from global trends. While in the world, lung cancer was the most commonly diagnosed and the leading cause of death (2), in Colombia was breast cancer. Compared with LAC, breast cancer was also the leading type among new cases and unlike Colombia, mortality trend was similar to the world (2).
Differences in cancer distribution between Colombia and the world reflect the ongoing social, economic, and health care changes in LAC. Such remarkable geographical contrast can be explained by differences in exposure to risk factors (reproductive, dietary, hormonal and environmental) and serious inequalities in timely access to screening and effective cancer treatment (5,13).
Discrepancies in the distribution of incidence and mortality worldwide were also reflected in Colombia’s regions. In terms of incidence, there was a homogeneous trend in breast and prostate cancer, with the highest ASR in Bogotá D.C., Central and Pacific regions, being significantly higher than the national in Bogotá D.C. This pattern could be explained by geographical proximity and similar social and economic development, which from a broad epidemiological perspective, implies a comparative distribution of risk background and access to quality care. Furthermore, domestic differences on prostate cancer incidence could also be explained by the distribution of ethnic or genetic variations across the country that have been linked with a higher risk of this type of cancer (14–16). Indeed, regions with the highest incidence coincide with a high proportion of Afro-Colombian population (17). Otherwise, prostate cancer screening coverage has also been associated with incidence rates and more than other cancer, screening with the prostate-specific antigen (PSA) increases the probability of being diagnosed (18). In fact, countries with a high PSA screening coverage also have higher incidence rates, early diagnosis and lower mortality rates (19). In Colombia, an organized population screening is not recommended. Early detection is focused on men aged >50 years or those aged <50 years with known risk factors and screening interval should not be inferior to 5 years (20,21). According to the national health survey, conducted in 2015, coverage of PSA screening in men older than 50 years was 44.60% and it varied by insurance, geographic location, education and socioeconomic level, being higher in Bogotá D.C while the lowest was identified in “other departments”(22). This coverage corresponds to the magnitude of prostate cancer incidence in those regions.
On the other hand, the incidence pattern for cervical cancer was different, showing a significantly higher ASR in “other departments” region, mainly composed of nonmetropolitan and rural areas. It has been reported that women in rural areas may experience barriers to optimal cervical cancer prevention, screening, and treatment, as well as, a higher frequency of risky sexual behaviors (23,24).
Regarding mortality, its distribution varied widely between regions, being the Eastern region the one that had lower ASR than national for all types of cancer. According to previous studies, disparities in mortality can be explained by diagnosis in advanced stages, limited access to quality health services, and treatment opportunity (25–27). However, in our study population, there were no differences between the proportion of people diagnosed with invasive neoplasms or the first treatment initiation.
As the trend is evaluated at a lower level, such as municipalities, there is more variability in patterns for both, incidence and mortality. Nevertheless, Agua de Dios in Cundinamarca was among the significantly higher incidence rates than the national for all types of cancer. The above may suggest a high prevalence of known risk factors in the municipality, as well as, limited access to screening, early diagnosis, and quality treatment.
Although there is a national policy for cancer attention (28), differences in incidence and mortality trends can be related to local approaches for implementing health programs and emerging social and economic changes typical of each region, department or municipality.
Finally, some studies have documented similar results and remarkable findings. According to the analysis of the cancer situation in 2015 performed by the Colombian Cancer Institute, in the city of Pasto, its population-based cancer registry showed how types of cancer, such as breast and prostate cancer had a higher incidence in urban areas while cervical cancer was commonly diagnosed in rural areas (22). Likewise, a report prepared by the National Health Observatory showed that there was a high health inequality across the country for these type of cancer. For example, in breast cancer, a higher concentration of cases and deaths was observed in regions with the highest wealth per capita, in women belonging to the contributory insurance and with increased access to mammography and specialized centers (29,30).
Regarding cervical and prostate cancer, their findings suggest that higher mortality rates were reported in regions with noteworthy socioeconomic inequality. They found that the lowest cervical cancer mortality rates were observed in the richest municipalities and at departmental/regional settings, they found a correlation between a higher income inequality (measured by the GINI index) and higher death rates. In the case of prostate cancer, the lowest rates were also reported in the richest municipalities, however, there was no evidence of a gradient consistent with municipal poverty levels. Surprisingly, at the departmental/regional level findings were contradictory, with a directly proportional correlation between income per capita and prostate cancer mortality rates, but and an inversely proportional association with the GINI inequality index (30).
Strengths and limitations
This analysis has important strengths, including the large completeness of the NACR, which guarantees the external validity and utility of our findings in the evidence-informed health policymaking process at the national and regional levels. Furthermore, the accuracy and quality of the information of all new cases were verified by a data monitoring process.
On the other hand, some limitations should be discussed. First, the passive case reporting by the health insurers could lead to under-reporting. In any case, it would be a small proportion because the reporting process is mandatory (12). The cross-sectional nature of the analysis does not allow establishing consistency in the trends we observed. Moreover, information bias cannot be ruled out because clinical records are the primary data source and they may be subject to error.