An analysis of cause-specific and sex-specific schedules of cancer mortality in an urban complex: Evidence from routine records in Ghana

Background: In countries without functioning cancer registries, population-based cancer estimates are problematic, and the burden of all-sites and site-specific cancer mortality is hard to gauge. Objective: The study aimed to estimate cause-specific, age-and sex-specific cancer mortality burden using data from the Vital Registration System (VRS) in an urban complex with no cancer registry. Methods: The Ghana Vital Registration System (GVRS) is designed to be coterminous with the national political-administrative units, allowing for mortality reporting at the household level. Death records are stored centrally at the National Office in Accra, the capital of Ghana. We collected and analysed, using Binomial/Poisson, Hazard and Cox regression models, mortality records from Ghana’s VRS over a 14-year (1998-2011) period in order to estimate and study the relationship among sex, age, and the risk of dying of cancer. Results: Overall, the results showed an increase in cause-specific hazard rates with increasing age. The results also showed strong evidence of a difference in all-cause relative hazard rate between males and females (1.12, 95% CI 1.07-1.19, p < 0.001). Finally, further cause-specific analyses showed that, while stomach cancers and lymphomas were more prevalent in males than females, the converse was true for bladder and pancreatic cancers (11.4% and 12.5%, respectively). Conclusion: The consistency of the findings reported here with results from similar analyses using data from standard cancer registries supports the view that routine data from vital registration systems can be used validly to estimate the burden of cancer mortality in resource-poor settings where cancer registries are unavailable.