This study is the first to use record linkage analysis to estimate cancer incidence among PLWA in West Africa. Among the ADCs, only KS was significantly increased among HIV positive persons and there was no statistically significant increase in incidence of NHL, CC and NADCs. These results for KS and CC are consistent with the findings from studies conducted in other countries where increased risk of KS among HIV positive persons and no association between CC and HIV was found [24–26]. Although several studies suggest that the burden of NHL is increased among HIV positive persons in Africa [12, 13, 27, 28], studies from West Africa have not supported this, just like our results. Similar to Bolarinwa et al., we identified several cases of NHL in the general population but none of these were matched in the HIV registry .
The relatively low incidence of cancers among HIV positive persons in Nigeria is probably multifactorial. An important possibility may be due to lack of cancer diagnosis. Given the size of the HIV/AIDS epidemic and main focus to diagnose and treat HIV or roll out prevention programs, care providers may miss clinical features suggestive of cancer in persons with HIV infection or fail to investigate those clients thoroughly. Although not well documented, our causal impression is that some physicians who are concerned about occupational exposure to HIV may provide limited quality services to HIV positive persons [29, 30]. Finally, the infrastructure to diagnose, refer, and register patients may also have caused low levels of utilization of diagnostic procedures needed to confirm cancer diagnosis in PLWA .
There is some evidence that ascertainment of cancer by the ABCR may not be complete. For example, the Kampala Cancer Registry, which covers a population of about 2.1 million, registered 4235 cancer cases during 2007 - 2009 . ABCR, which covers a population of about 1.4 million, registered 1815 cancer cases from 2009–2011 registered. This may be due to a longer experience with cancer registration resulting in more complete data or a lifestyle with a higher cancer risk profile in Uganda. We also note that the HIV prevalence is higher in Kyadondo compared to Abuja, 15% and 8.6%, respectively [12, 32], which may lead to a higher burden of HIV-associated cancers, such as KS in Uganda. In this regard, it is worth noting that the prevalence of human herpesvirus 8 is lower in Nigeria compared to Uganda , which would also explain the lower prevalence of cancer in ABCR, particularly KS. Our study highlights the opportunity to improve case finding methods of cancer registries in Nigeria. Similar to other countries in Africa, the infrastructure for population-based cancer registries is poorly developed and not fully linked to the HIV screening and registration process. Increased utilization of electronic medical records and databases, improved collaboration between registries and a multidisciplinary approach to data collection and management will improve case ascertainment of HIV and cancer cases, their report to registries and the linkage of registry data. A separate study evaluating the completeness of data obtained by the ABCR is under way.
Other possible explanations for the low cancer incidence include the availability of HAART, which would be associated with improved immune function and reduced risk of AIDS-associated cancers, such as KS [34, 35]. The coverage of HAART in HIV positive persons in Nigeria is relatively low (24% of individuals who should be on HAART ), thus the contribution of this factor is uncertain. Loss to follow-up is another possible explanation , as is competing mortality from common diseases, such as tuberculosis and malaria.
Our study is also limited by the small sample size thus we were unable to compare prevalent versus incident cancers and we lacked sufficient power to detect small differences that may exist between the general population and the HIV positive population. Another limitation is lack of unique individual personal identification numbers, such as social security numbers, which would increase the specificity and accuracy of the match. Nevertheless, thorough clerical review of the linked records increased our confidence in the matched records in study. Additional research is needed to understand and improve the specificity of matches in resource-limited settings, where unique individual identification numbers are generally lacking. As the HIV positive population in Nigeria continues to age, coverage with HAART improves and the registry systems are strengthened, more record linkage studies to examine the growing public health impact of cancers occurring in people living with HIV/AIDS will be warranted.