Table 2

The association between HIV infection and specific cancer types^†

Adjusted prevalence of HIV infection (%)^‡

Estimate

95% CI*

Burkitt lymphoma (n = 269)

9.1

(6.0 to 13.2)

Kaposi sarcoma (n = 29)

81.8

(63.1 to 93.6)

Non-Burkitt non-Hodgkin lymphoma (n = 33)

8.9

(1.8 to 24.0)

All other cancers (n = 164)

3.5

(1.3 to 7.6)

Odds ratio for specific cancer given HIV infection compared to baseline^§ group

Odds ratio

95% CI

p-value

Burkitt lymphoma (n = 269)

2.2

(0.8 to 6.4)

0.132

Kaposi sarcoma (n = 29)

93.5

(26.9 to 324.4)

<0.001

Non-Burkitt non-Hodgkin lymphoma (n = 33)

4.4

(1.1-17.9)

0.038

^†Individual logistic regression models of the risk of having a specific cancer based on HIV seroprevalence compared to the 'all other cancers' group were adjusted for age class and gender; patients diagnosed with nephroblastoma, retinoblastoma and rhabdomyosarcoma have been included in the all other cancers group; patients with missing age, gender or HIV status have been excluded. ^‡Prevalence standardised to a population 50% female and 50% over 5 years old. *Exact binomial confidence limits. ^§Baseline group includes 'all other cancers' and non-malignant conditions.

NOTE: When analyses were repeated restricting to those cases with a laboratory confirmation of diagnosis, the results materially similar: Burkitt lymphoma OR 2.0, 95% CI 0.5-7.7; Kaposi sarcoma OR 46.9, 95% 3.5-631.1; Non-Burkitt non-Hodgkin lymphoma OR 3.6, 95% CI 0.6-21.5.

Mutalima et al. Infectious Agents and Cancer 2010 5:5   doi:10.1186/1750-9378-5-5