Format

Send to

Choose Destination
Int J Cancer. 2016 Sep 15;139(6):1209-16. doi: 10.1002/ijc.30154. Epub 2016 May 18.

Record linkage to correct under-ascertainment of cancers in HIV cohorts: The Sinikithemba HIV clinic linkage project.

Author information

1
National Cancer Registry, National Health Laboratory Service, Johannesburg, South Africa.
2
Graduate School for Cellular and Biomedical Sciences, University of Bern, Bern, Switzerland.
3
Institute of Social and Preventive Medicine (ISPM), University of Bern, Bern, Switzerland.
4
Centre for Infectious Disease Epidemiology and Research (CIDER), School of Public Health and Family Medicine, University of Cape Town, Cape Town, South Africa.
5
Health Economics HIV And AIDS Research Division (HEARD), University of KwaZulu-Natal, Durban, South Africa.
6
McCord Hospital, Overport, Durban, South Africa.

Abstract

The surveillance of HIV-related cancers in South Africa is hampered by the lack of systematic collection of cancer diagnoses in HIV cohorts and the absence of HIV status in cancer registries. To improve cancer ascertainment and estimate cancer incidence, we linked records of adults (aged ≥ 16 years) on antiretroviral treatment (ART) enrolled at Sinikithemba HIV clinic, McCord Hospital in KwaZulu-Natal (KZN) with the cancer records of public laboratories in KZN province using probabilistic record linkage (PRL) methods. We calculated incidence rates for all cancers, Kaposi sarcoma (KS), cervix, non-Hodgkin's lymphoma and non-AIDS defining cancers (NADCs) before and after inclusion of linkage-identified cancers with 95% confidence intervals (CIs). A total of 8,721 records of HIV-positive patients were linked with 35,536 cancer records. Between 2004 and 2010, we identified 448 cancers, 82% (n = 367) were recorded in the cancer registry only, 10% (n = 43) in the HIV cohort only and 8% (n = 38) both in the HIV cohort and the cancer registry. The overall cancer incidence rate in patients starting ART increased from 134 (95% CI 91-212) to 877 (95% CI 744-1,041) per 100,000 person-years after inclusion of linkage-identified cancers. Incidence rates were highest for KS (432, 95% CI 341-555), followed by cervix (259, 95% CI 179-390) and NADCs (294, 95% CI 223-395) per 100,000 person-years. Ascertainment of cancer in HIV cohorts is incomplete, PRL is both feasible and essential for cancer ascertainment.

KEYWORDS:

HIV; South Africa; cancer; record linkage

PMID:
27098265
PMCID:
PMC5084785
DOI:
10.1002/ijc.30154
[Indexed for MEDLINE]
Free PMC Article

Supplemental Content

Full text links

Icon for Wiley Icon for PubMed Central
Loading ...
Support Center