HIV Epidemiology Research Study Research Articles

The influence of human leukocyte antigen class I alleles and their population frequencies on human immunodeficiency virus type 1 control among African Americans

Populations of African ancestry continue to account for a disproportionate burden of the human immunodeficiency virus type 1 (HIV-1) epidemic in the United States. We investigated the effects of human leukocyte antigen (HLA) class I markers in association with virologic and immunologic control of HIV-1 infection among 338 HIV-1 subtype B-infected African Americans in 2 cohorts: Reaching for Excellence in Adolescent Care and Health (REACH) and HIV Epidemiology Research Study (HERS). One-year treatment-free interval measurements of HIV-1 RNA viral loads and CD4+ T cells were examined both separately and combined to represent 3 categories of HIV-1 disease control (76 controllers, 169 intermediates, and 93 noncontrollers). Certain previously or newly implicated HLA class I alleles (A*32, A*36, A*74, B*14, B*1510, B*3501, B*45, B*53, B*57, Cw*04, Cw*08, Cw*12, and Cw*18) were associated with 1 or more of the endpoints in univariate analyses. After multivariable adjustments for other genetic and nongenetic risk factors of HIV-1 progression, the subset of alleles more strongly or consistently associated with HIV-1 disease control included A*32, A*74, B*14, B*45, B*53, B*57, and Cw*08. Carriage of infrequent HLA-B but not HLA-A alleles was associated with more favorable disease outcomes. Certain HLA class I associations with control of HIV-1 infection cross the boundaries of race and viral subtype, whereas others appear confined within one or the other of those boundaries.

Bacterial Vaginosis and the Natural History of Human Papillomavirus

Objective. To evaluate associations between common vaginal infections and human papillomavirus (HPV). Study Design. Data from up to 15 visits on 756 HIV-infected women and 380 high-risk HIV-uninfected women enrolled in the HIV Epidemiology Research Study (HERS) were evaluated for associations of bacterial vaginosis, trichomoniasis, and vaginal Candida colonization with prevalent HPV, incident HPV, and clearance of HPV in multivariate analysis. Results. Bacterial vaginosis (BV) was associated with increased odds for prevalent (aOR = 1.14, 95% CI: 1.04, 1.26) and incident (aOR = 1.24, 95% CI: 1.04, 1.47) HPV and with delayed clearance of infection (aHR = 0.84, 95% CI: 0.72, 0.97). Whereas BV at the preceding or current visit was associated with incident HPV, in an alternate model for the outcome of incident BV, HPV at the current, but not preceding, visit was associated with incident BV. Conclusion. These findings underscore the importance of prevention and successful treatment of bacterial vaginosis.

Association between Semen Exposure and Incident Bacterial Vaginosis

Objective. To identify correlates of incident bacterial vaginosis (BV) diagnosed with Nugent scoring among high-risk women. Study Design. We conducted both cohort and case-crossover analyses, stratified by HIV infection status, based on 871 HIV-infected and 439 HIV-uninfected participants in the HIV Epidemiology Research Study, conducted in 4 US sites in 1993–2000. Results. BV incidence was 21% and 19% among HIV-infected and -uninfected women, respectively. Fewer correlates of BV were found with case-crossover than with cohort design. Reporting frequent coitus (regardless of consistency of condom use) was correlated with BV in cohort analyses but not in case-crossover analyses. The sole correlate of BV in both types of analyses was the detection of spermatozoa on Gram stain, which is a marker of semen exposure. Conclusion. The inconsistent association between condom use and BV in prior studies could be from reporting bias. We found evidence of a relationship between semen exposure and incident BV.

High-Risk Human Papillomavirus Reactivation in Human Immunodeficiency Virus–Infected Women

To evaluate the presence of and estimate risk factors for reactivation of latent high-risk human papillomavirus (HPV) cervical infection in human immunodeficiency virus (HIV)-infected and HIV-uninfected women. Data from 898 women in the HIV Epidemiology Research Study (HERS) were used to evaluate cervical HPV latency and reactivation. Prior exposure to HPV types (16, 18, 31, 35, and 45) was determined by serologic testing at enrollment, and cervical shedding of HPV was detected by polymerase chain reaction at 6-month intervals. Human papillomavirus cervical shedding and sexual history were used to estimate rates of reactivation and recurrence. Repeated measures survival analysis was used to estimate hazard ratios and 95% confidence intervals for reactivation and recurrence. Rates of total HPV shedding (recurrence and reactivation) during follow-up were assessed by HIV status and rate ratios were calculated. Reactivation of latent HPV infections was observed in HIV-infected women, but few reactivation events were identified in HIV-uninfected women. Factors consistently associated with reactivation in HIV-infected women included CD4 count less than 200/mm and age younger than 35 years. Women infected with HIV had 1.8 to 8.2 times higher rates of viral shedding (reactivation plus recurrence) compared with HIV-uninfected women. Women with a history of cervical HPV infection may be at risk of reactivation of latent viral infection even in the absence of sexual activity, and this risk is higher in women with HIV infection. II.

Bayesian semiparametric regression for longitudinal binary processes with missing data.

Longitudinal studies with binary repeated measures are widespread in biomedical research. Marginal regression approaches for balanced binary data are well developed, whereas for binary process data, where measurement times are irregular and may differ by individuals, likelihood-based methods for marginal regression analysis are less well developed. In this article, we develop a Bayesian regression model for analyzing longitudinal binary process data, with emphasis on dealing with missingness. We focus on the settings where data are missing at random (MAR), which require a correctly specified joint distribution for the repeated measures in order to draw valid likelihood-based inference about the marginal mean. To provide maximum flexibility, the proposed model specifies both the marginal mean and serial dependence structures using nonparametric smooth functions. Serial dependence is allowed to depend on the time lag between adjacent outcomes as well as other relevant covariates. Inference is fully Bayesian. Using simulations, we show that adequate modeling of the serial dependence structure is necessary for valid inference of the marginal mean when the binary process data are MAR. Longitudinal viral load data from the HIV Epidemiology Research Study are analyzed for illustration.

Nutritional biomarkers associated with gynecological conditions among US women with or at risk of HIV infection

Background: Women infected with HIV face a combination of health threats that include compromised nutrition and adverse gynecological conditions. This relation among HIV, nutrition, and gynecological conditions is complex and has rarely been investigated.Objective: Our objective was to investigate nutritional biomarkers associated with several gynecological conditions among US women with or at risk of HIV infection.Design: Data on 369 HIV-infected and 184 HIV-uninfected women with both nutritional and gynecological outcomes were analyzed from a cross-sectional nutritional substudy of the HIV Epidemiology Research Study (HERS). We examined micronutrient distributions comparing HIV-infected with HIV-uninfected participants and both subgroups with the US population. We then modeled the relation of 16 micronutrient serum concentrations to various gynecological conditions, producing partially adjusted odds ratios, adjusted for study site, risk cohort, and HIV status.Results: HIV-infected women’s median antioxidant concentrations were lower than the medians of the US population. HERS women had lower median concentrations for vitamin A, selenium, and zinc irrespective of HIV status. Trichomoniasis prevalence was inversely related to serum α-carotene. Lower concentrations of vitamins A, C, and E and β-carotene were associated with an increased risk of bacterial vaginosis. Higher concentrations of serum zinc were associated with lower risk of human papillomavirus. Candida colonization was higher among women with higher concentrations of total-iron-binding capacity.Conclusion: We identified several significant associations of micronutrient concentrations with the prevalence of gynecological conditions. These findings warrant further investigation into possible causal relations.

Outcomes After Treatment of Cervical Intraepithelial Neoplasia Among Women With HIV

To describe outcomes after treatment of cervical intraepithelial neoplasia (CIN) in women with HIV. Women in two prospective cohort studies, the Women's Interagency HIV Study (WIHS) and the HIV Epidemiology Research Study (HERS), were followed every 6 months after treatment of CIN using human papillomavirus (HPV) testing and cytology with colposcopy as indicated. Identification of CIN or a squamous intraepithelial lesion (SIL) within 6 months was defined as treatment failure and later disease as recurrence. Follow-up was available for 170 HIV-seropositive and 15 HIV-seronegative women. Treatment failed in 84 (45%) women (79 HIV seropositive and 5 HIV seronegative). Failure was more likely in women with lower CD4 counts (CD4 < 200 cells/microL: odds ratio [OR] = 2.96; 95% CI = 1.4-6.2) and detectable HPV DNA (OR 8.20; 95% CI = 1.8-37.4; p = .01). After successful treatment, recurrence-free probabilities at 1,2, 3, and 5 years were .79, .64, .49, and .34, respectively. HIV-seronegative women were less likely to recur than HIV-seropositive women (p = .03). In multivariable analysis of HIV-positive women, recurrence was more likely among women treated for CIN 2,3 (hazard ratio [HR] = 2.4; 95% CI = 1.4-4.8), those with CD4 count of less than 200 cells/microL (HR = 2.9; 95% CI = 1.3-6.5) and those with HPV after treatment (HR 2.9; 95% CI = 1.4-6.1); oncogenic HPV was more strongly associated with recurrence than nononcogenic HPV (p(trend) = .009). Most failures and recurrences were low grade, but one adenocarcinoma was diagnosed 4.2 years after therapy for CIN 1. Treatment failure and recurrence are common in women with HIV but are usually low grade.

Psychological resources protect health: 5-year survival and immune function among HIV-infected women from four US cities

Guided by Cognitive Adaptation Theory, the aim was to determine whether psychological resources (positive affect, positive expectancy regarding health outcomes, finding meaning in challenging circumstances) protect against HIV-related mortality and decline in CD4 lymphocyte counts among women with HIV. The HIV Epidemiologic Research Study, a longitudinal prospective cohort study, with semi-annual interview, physical examination and laboratory assays. A total of 773 HIV-seropositive women aged 16 to 55 years were recruited from four academic medical centers in Baltimore, Maryland; Bronx, New York; Providence, Rhode Island; and Detroit, Michigan. Semi-annually for up to 5 years, the women were interviewed, underwent physical examination, medical record abstraction, and venipuncture. Primary outcomes for these analyses included HIV-related mortality and CD4 cell count slope decline over 5 years. Psychological resources were inversely associated with HIV-related mortality and time to death, beyond the effects of potential confounding variables such as clinical status (e.g., HIV viral load, symptoms, antiretroviral therapy), sociodemographic characteristics (e.g. age, race), and depression at study entry (P < 0.05). Psychological resources also were inversely associated with CD4+ cell count decline (P < 0.01), serving as a possible mechanism linking resources to mortality. Psychological resources may protect against HIV-related mortality and immune system decline. Findings have implications for understanding individual variability in HIV disease progression. Moreover, because psychological resources are potentially amenable to change, results can be applied to clinical interventions aimed at improving the health of women with HIV.

Time to clearance of human papillomavirus infection by type and human immunodeficiency virus serostatus

Persistent infection with high-risk human papillomavirus (HPV) is central to cervical carcinogenesis. Certain high-risk types, such as HPV16, may be more persistent than other HPV types, and type-specific HPV persistence may differ by HIV serostatus. This study evaluated the association between HPV type and clearance of HPV infections in 522 HIV-seropositive and 279 HIV-seronegative participants in the HIV Epidemiology Research Study (HERS, United States, 1993-2000). Type-specific HPV infections were detected using MY09/MY11/HMB01-based PCR and 26 HPV type-specific probes. The estimated duration of type-specific infections was measured from the first HPV-positive visit to the first of two consecutive negative visits. Hazard ratios (HRs) and 95% confidence intervals (CIs) for HPV clearance were calculated using Cox models adjusted for study site and risk behavior (sexual or injection drugs). A total of 1,800 HPV infections were detected in 801 women with 4.4 years median follow-up. HRs for clearance of HPV16 and related types versus low-risk HPV types were 0.79 (95% CI: 0.64-0.97) in HIV-positive women and 0.86 (95% CI: 0.59-1.27) in HIV-negative women. HRs for HPV18 versus low-risk types were 0.80 (95% CI: 0.56-1.16) and 0.57 (95% CI: 0.22-1.45) for HIV-positive and -negative women, respectively. HPV types within the high-risk category had low estimated clearance rates relative to low-risk types, but HRs were not substantially modified by HIV serostatus.

Smoking and Time to Clearance of Human Papillomavirus Infection in HIV-Seropositive and HIV-Seronegative Women

Persistent human papillomavirus (HPV) infection seems central to cervical carcinogenesis. Smoking is associated with cervical cancer in HPV DNA-positive women, but its association with HPV persistence is unclear, particularly with respect to human immunodeficiency virus (HIV) serostatus. The authors evaluated smoking and HPV clearance by HIV serostatus among 801 women from the HIV Epidemiology Research Study (United States, 1993-2000). Type-specific HPV duration was defined as the interval between initial MY09/11 polymerase chain reaction positivity and the first of two consecutive HPV-negative study visits. Hazard ratios adjusted for study site and risk behaviors (sexual activity or injection drug use) were estimated using Cox regression. This analysis included 522 HIV-seropositive and 279 HIV-seronegative women (median follow-up, 4.4 years). Ever smoking was associated with reduced clearance of high-risk HPV in HIV-seronegative women (hazard ratio (HR) = 0.51, 95% confidence interval (CI): 0.30, 0.88) but not in HIV-seropositive women (HR = 0.96, 95% CI: 0.65, 1.42); similar results were found for current smoking. Current smoking was not associated with clearance of any type-specific HPV in HIV-seropositive (HR = 0.99, 95% CI: 0.82, 1.20) or HIV-seronegative (HR = 0.93, 95% CI: 0.68, 1.26) women. HPV clearance did not appear to vary by amount or duration of smoking. Smoking did not modify overall clearance but was associated with lower high-risk HPV clearance in HIV-seronegative women.

Effect of HIV Infection on Atypical Squamous Cells of Undetermined Significance

Detection of atypical squamous cells of undetermined significance (ASCUS) is a cervical cytologic finding that is suggestive but not definitive of squamous intraepithelial lesions (SILs). We examined the risk, characteristics, and progression of ASCUS in women with and without human immunodeficiency virus (HIV) infection. Cervical Papanicolou (Pap) test and colposcopy data were obtained at the first 10 semiannual visits for the HIV Epidemiology Research study of 774 HIV-infected and 480 demographically similar, HIV-uninfected women in the United States. Multiple logistic regression models and Cox proportional hazards models were utilized. ASCUS was more common among HIV-infected women (odds ratio [OR], 1.6 [95% confidence interval {CI}, 1.3-2.0] to 2.6 [95% CI, 1.9-3.6]) after adjustment for human papillomavirus (HPV) infection and other risk factors (e.g., race, condyloma, and prior Pap test result). Among women with normal Pap test results at enrollment, the cumulative incidence of ASCUS was 78% among HIV-infected women and 38% among HIV-uninfected women. HIV-infected and HIV-uninfected women with ASCUS did not differ by prevalence of indices of inflammation (inflammation on Pap test and leukocytes on cervical gram stain). HPV infection, including high risk types, was more common among HIV-infected women with ASCUS. Among women with ASCUS, 60% of HIV-infected and 25% of HIV-uninfected women developed SILs (P < .01). Compared with HIV-infected women with higher CD4+ lymphocyte counts, HIV-infected women with CD4+ lymphocyte counts < 200 cells/microL were more likely to present subsequently with a SIL detected by Pap test (OR, 1.7; 95% CI, 0.8-3.6). Higher risk of SIL following the appearance of ASCUS among HIV-infected women, especially women with low CD4+ lymphocyte counts, supports the need for follow up with colposcopy and histologic examination, as indicated, to allow early detection and treatment of SIL.

Accuracy of Papanicolaou Test among HIV-Infected Women

We examined the accuracy of the Papanicolaou (Pap) test among women with and women at high risk for human immunodeficiency virus (HIV) infection. Pap test results and colposcopy findings were compared for 189 HIV-infected women and 95 HIV-uninfected women from the Baltimore HIV Epidemiology Research (HER) study site who were followed up semiannually from 1993 through 1999. Correlations were examined using data from individual visits that were analyzed using generalized estimating equation methods. Among women with normal results of Pap tests who underwent biopsy, the likelihood of cervical intraepithelial neoplasia was significantly greater for HIV-infected women (14.3%) than for HIV-uninfected women (1.2%) (P < .01). The specificity and negative predictive value of Pap testing were higher among HIV-uninfected women, regardless of whether detection of atypical squamous cells of undetermined significance was considered to be abnormal. Independent predictors of discordant cytologic and histologic findings included detection of human papillomavirus (adjusted odds ratio [OR], 3.1; 95% confidence interval [CI], 1.0-9.8) and a CD4 cell count of < 500 cells/microL (adjusted OR, 6.5; 95% CI, 1.5-29.2). Of the 19 HIV-infected women with normal cytologic findings and cervical intraepithelial neoplasia, 18 had abnormal Pap test findings (most often atypical squamous cells of undetermined significance) within 1 year of discordant cytologic and histologic findings. In this well-described cohort of women with and women at high risk for HIV infection, agreement between cytologic findings and colposcopic and histologic findings was high. A small number of HIV-infected women had cervical intraepithelial neoplasia, despite having normal cytologic findings; 95% of these women would have had cytologic abnormalities detected within 1 year of the discordant results by use of current guidelines for Pap test screening. These data fail to support the need for routine colposcopy in all HIV-infected women.

Extending McNemar's Test: Estimation and Inference When Paired Binary Outcome Data Are Misclassified

McNemar's test is popular for assessing the difference between proportions when two observations are taken on each experimental unit. It is useful under a variety of epidemiological study designs that produce correlated binary outcomes. In studies involving outcome ascertainment, cost or feasibility concerns often lead researchers to employ error-prone surrogate diagnostic tests. Assuming an available gold standard diagnostic method, we address point and confidence interval estimation of the true difference in proportions and the paired-data odds ratio by incorporating external or internal validation data. We distinguish two special cases, depending on whether it is reasonable to assume that the diagnostic test properties remain the same for both assessments (e.g., at baseline and at follow-up). Likelihood-based analysis yields closed-form estimates when validation data are external and requires numeric optimization when they are internal. The latter approach offers important advantages in terms of robustness and efficient odds ratio estimation. We consider internal validation study designs geared toward optimizing efficiency given a fixed cost allocated for measurements. Two motivating examples are presented, using gold standard and surrogate bivariate binary diagnoses of bacterial vaginosis (BV) on women participating in the HIV Epidemiology Research Study (HERS).

Persistence of inconsistent condom use: relation to abuse history and HIV serostatus.

This study longitudinally examines the relation between a history of experiencing childhood and adult physical or sexual abuse, and male condom use by women with or at risk for HIV. Abuse history and prospective condom use data were collected from 214 HIV infected and 189 uninfected women participating in the HIV Epidemiology Research Study (HERS) who were inconsistent condom users at baseline and received two safer sex counseling sessions. Analyses were conducted to assess the association between abuse history and condom use while controlling for sociodemographic variables and other risk factors. HIV-uninfected women with a history of adult physical abuse were five times less likely to report consistent condom use at 1-year follow-up than uninfected women without a history of abuse while holding control variables constant. Expectations of a negative reaction by the partner to suggested condom use did not explain this association. Though in the same direction as in uninfected women, abuse history was not significantly related to consistent condom use among HIV-infected women. These data indicate the need to develop risk prevention strategies tailored to uninfected women with a history of adult abuse. In lieu of specialized interventions, health care providers should assess women's abuse history and supplement HIV prevention counseling with mental health counseling when indicated.

Evaluation of possible effects of continued drug use on HIV progression among women

Data from a prospective, multi-centred study of HIV infection in women (HIV Epidemiology Research Study [HERS]) was analysed to investigate the effect of continued injection drug use behaviours on progression to AIDS. All women enrolled in the HERS had at enrollment and at six-month intervals, a face-to-face interview which included specific injection drug use, a physical exam, and specimen collection that included T-cell subset analysis and HIV plasma RNA detection. Six hundred and thirty-nine HIV-infected women contributed 3021 person years of observation during 7.25 years of follow-up, and 299 of these women progressed to AIDS (46.8%). In multivariable analysis, there was no significantly increased risk of progression to AIDS for women reporting pre-baseline injection drug use [hazard ratio (HR)=1.07 (0.78, 1.47)] or reported injection drug use during follow-up [HR=0.89 (0.66, 1.21)] compared with never injecting. In a separate multivariable-model, comparing women who reported no injection in past six months to active injection drug users, the frequency of injection during the previous six months measured by daily injection [HR=0.97 (0.61, 1.55)] or less than daily injection [HR=0.84 (0.54, 1.33)] was not associated with progression to AIDS. Being in drug treatment was independently associated with a slower progression to AIDS [HR=0.41 (0.28, 0.59)]. Neither injection drug use, nor frequency of injection drug use was associated with progression to AIDS among HIV infected women. Initiation of antiretroviral therapy among drug users should be based on readiness for treatment rather than concern about faster progression.

Mortality rates and causes of death in a cohort of HIV-infected and uninfected women, 1993-1999.

HIV/AIDS-associated and non-HIV/AIDS-associated death rates and causes of death between 1993 and 1999 were examined in 885 HIV-infected women and 425 uninfected women of the HIV Epidemiology Research Study cohort. Causes of death were determined by review of death certificates and the National Death Index. Adjusted hazard ratios were calculated for mortality risk factors. In the 885 HIV-infected women and 425 uninfected women, 234 deaths and 8 deaths, respectively, occurred by December 31, 1999. All-cause death rates in the HIV-infected women were unchanged between the pre-HAART (1993-1996) and HAART eras (1997-1999)-5.1 versus 5.4 deaths per 100 person-years (py). AIDS as a cause of death decreased from 58% of all deaths in 1996 to 19% in 1999, while HAART use increased to 42% by the end of 1999. In spite of the modest proportion ever using HAART, HIV-related mortality rates did decline, particularly in women with CD4+ cell counts less than 200/mm(3). Drug-related factors were prominent: for the 129 non-AIDS-defining deaths, hepatitis C positivity (relative hazard [RH] 2.6, P <.001) and injection drug use (RH 1.7, P = 0.02) were strong predictors of mortality, but were not significant in the Cox model for 105 AIDS-defining deaths (RH 0.9, P >.30 and RH 0.7, P >.30, respectively. The regression analysis findings, along with the high percentage of non-AIDS deaths attributable to illicit drug use, suggest that high levels of drug use in this population offset improvements in mortality from declining numbers of deaths due to AIDS.

Overweight and human immunodeficiency virus (HIV) progression in women: associations HIV disease progression and changes in body mass index in women in the HIV epidemiology research study cohort.

An association of increased weight with a slower progression of human immunodeficiency virus (HIV) disease has been reported in studies that have not included large numbers of women. We evaluated the association of HIV disease progression with body mass index (BMI) in 871 women and present cross-sectional, survival, and longitudinal analyses. A higher baseline BMI was associated with a lower rate of occurrence of the first CD4 cell count <200 cells/mm(3). In analyses that incorporated time-varying BMI, underweight and normal women had an increased risk of clinical acquired immune deficiency syndrome, and underweight women had increased risk of HIV-related death, compared with obese women. The association between change in BMI and CD4 cell count was estimated; increases in BMI were associated with slight increases in CD4 cell counts, even after controlling for prior values of CD4 cell count, viral load, and treatment. Higher BMI and increases in BMI are associated with a decreased risk of HIV progression.

Human papillomavirus type 16 and immune status in human immunodeficiency virus-seropositive women.

Human papillomavirus (HPV) type 16 is etiologically associated with approximately half of all cervical cancers. It is important, therefore, to determine the characteristics that distinguish HPV16 from other HPV types. A preliminary result based on cross-sectional baseline data in the Women's Interagency Human Immunodeficiency Virus (HIV) Study (WIHS) suggested that the prevalence of HPV16 might have a weaker association with immune status in HIV-seropositive women than that of other HPV types. To address this issue, we examined HPV test results from repeated study visits in the WIHS and from an independent study, the HIV Epidemiology Research Study (HERS). HIV-seropositive women in the WIHS (n = 2058) and in the HERS (n = 871) were assessed semiannually. HPV DNA was detected in cervicovaginal lavage specimens by using polymerase chain reaction assays. Prevalence ratios were used to compare the prevalence of each HPV type in women with the lowest CD4+ T-cell counts (<200 T cells/mm3) with that of women with the highest CD4+ T-cell counts (> or =500 T cells/mm3). A summary prevalence ratio for each HPV type (i.e., across visits and studies) was estimated using generalized estimating equations. The association of CD4+ T-cell stratum with type-specific HPV incidence was measured using multivariable Cox regression models. All statistical tests were two-sided. The prevalence ratio for HPV16 was low compared with that of other HPV types at every study visit in both cohorts. The generalized estimating equation summary prevalence ratio for HPV16 (1.25, 95% confidence interval [CI] = 0.97 to 1.62) was the smallest measured, and it was statistically significantly lower than that of all other HPV types combined (P =.01). The association of CD4+ T-cell stratum with HPV16 incidence was also among the smallest measured (hazard ratio = 1.69, 95% CI = 1.01 to 2.81). The prevalent and incident detection of HPV16 is more weakly associated with immune status in HIV-seropositive women than that of other HPV types, suggesting that HPV16 may be better at avoiding the effects of immune surveillance, which could contribute to HPV16's strong association with cervical cancer.

Estimating causal treatment effects from longitudinal HIV natural history studies using marginal structural models.

Several recently completed and ongoing studies of the natural history of HIV infection have generated a wealth of information about its clinical progression and how this progression is altered by therepeutic interventions and environmental factors. Natural history studies typically follow prospective cohort designs, and enroll large numbers of participants for long-term prospective follow-up (up to several years). Using data from the HIV Epidemiology Research Study (HERS), a six-year natural history study that enrolled 871 HIV-infected women starting in 1993, we investigate the therapeutic effect of highly active antiretroviral therapy regimens (HAART) on CD4 cell count using the marginal structural modeling framework and associated estimation procedures based on inverse-probability weighting (developed by Robins and colleagues). To evaluate treatment effects from a natural history study, specialized methods are needed because treatments are not randomly prescribed and, in particular, the treatment-response relationship can be confounded by variables that are time-varying. Our analysis uses CD4 data on all follow-up visits over a two-year period, and includes sensitivity analyses to investigate potential biases attributable to unmeasured confounding. Strategies for selecting ranges of a sensitivity parameter are given, as are intervals for treatment effect that reflect uncertainty attributable both to sampling and to lack of knowledge about the nature and existence of unmeasured confounding. To our knowledge, this is the first use in "real data" of Robins's sensitivity analysis for unmeasured confounding (Robins, 1999a, Synthese 121, 151-179). The findings from our analysis are consistent with recent treatment guidelines set by the U.S. Panel of the International AIDS Society (Carpenter et al., 2000, Journal of the American Medical Association 280, 381-391).

A prospective study of adherence and viral load in a large multi-center cohort of HIV-infected women.

To examine the relationship between antiretroviral adherence and viral load, and to determine the predictors of adherence over time in HIV-infected women. Prospective observational study. One-hundred sixty-one HIV-infected women who were taking antiretroviral therapy for a median of 3.0 years were recruited from the HIV Epidemiology Research Study, a multicenter cohort study of HIV infection in women. Antiretroviral adherence (percent of doses taken as prescribed) was measured over a 6-month period using MEMS caps. At baseline and follow-up, CD4 lymphocyte count and viral load were measured, and a standardized interview was administered to elicit medication history and drug use behaviors. To examine changes in adherence over time, the mean adherence to all antiretroviral agents was calculated for each monitored month. Adherence varied significantly over time (P < 0.001), ranging from a mean of 64% in month 1 to 45% in month 6. Nearly one-fourth of the participants had a 10% or greater decrease in adherence between consecutive months. Virologic failure occurred in 17% of women with adherence of > or = 88%, 28% of those with 45-87% adherence, 43% of those with 13-44% adherence, and 71% of those with < or = 12% adherence. In multivariate analysis, factors predicting lower adherence included active drug use, alcohol use, more frequent antiretroviral dosing, shorter duration of antiretroviral use, younger age, and lower initial CD4 lymphocyte count. Antiretroviral adherence is not stable over time. Interventions aimed at monitoring and improving long-term adherence in women are urgently needed.