Impact of HIV-related immunosuppression on invasive cervical cancer risk among women with low-grade squamous intraepithelial lesions

Invasive cervical cancer (ICC) is an important cause of morbidity and mortality among women despite the preventable nature of the disease. Human immunodeficiency virus (HIV) appears to alter the natural history of oncogenesis, but its impact on ICC is not clearly defined. The studies described herein address the impact of HIV infection on the progression to ICC among Massachusetts Medicaid recipients with documented low-grade squamous intraepithelial lesions (LSIL). The studies were restricted to women with LSIL to improve study efficiency and focus on disease progression.; The first study compared the risk for progression to ICC in women with LSIL and documented HIV infection to women with LSIL but without documented HIV infection. Women with documented HIV infection had an increased relative hazard of developing ICC compared with women without documented HIV infection (RR = 2.49; 95% CI 1.12–5.51). These results are consistent with the theory that the progression of LSIL is promoted by HIV infection. The risk of ICC was particularly increased among the subset of women with documented HIV infection who had an AIDS diagnosis compared with women without documented HIV infection (RR = 3.48, 95% CI 1.44–8.39).; The second study used a nested case-control design within the same cohort to evaluate the frequency of documented HIV infection in women with ICC compared with women with LSIL. Women with ICC were more likely to be HIV-infected than women without invasive disease (OR = 3.86; 95% CI 1.44–10.33). In this study also, the effect was more pronounced in HIV-positive women with advanced immunosuppression, as indicated by a diagnosis of AIDS or a CD4 cell count under 200, than in HIV-positive women without severe immunosuppression or women without documented HIV infection.; The third study assessed the reliability of Medicaid data by comparing the agreement of data collected through medical record review and from Medicaid claims files. Observed agreement was excellent for HIV status (agreement 99.4%, k = 0.94). Agreement was moderate for procedures related to cervical cancer screening, diagnosis and treatment. Observed agreement was fair for ICC (agreement 77%, k = 0.37), with high sensitivity but poor specificity. Due to the potential for misclassification, the validity of most variables should be verified via medical record review in a sample of electronic records.