Search

BackgroundAlthough fracture rates are higher in HIV+ than HIV- women, whether HIV infection increases risk of falls is unclear. We determined the longitudinal occurrence and risk factors for falls in the Women's Interagency HIV Study (WIHS), and explored associations with cognitive complaints.MethodsRecent (prior 6 months) self-reported falls were collected in 1,816 (1,250 HIV+; 566 HIV-) women over 24 months. Generalized estimating equation models using stepwise selection determined odds of any fall (versus none).ResultsHIV+ women were older than HIV- women (median 49 versus 47 years; P=0.0004), more likely to report neuropathy (20% versus 16%; P=0.023), and had greater central nervous system (CNS) medication use. At least one fall was reported in 41% HIV+ versus 42% HIV- women, including ≥2 falls in 25% HIV+ and 24% HIV- (overall P=0.30). Cognitive complaints were associated with falls among HIV+ (odds ratio [OR] 2.38; 95% CI 1.83, 3.09) and HIV- women (OR 3.43; 95% CI 2.37, 4.97); in adjusted models, cognitive complaints remained significant only in HIV- women (adjusted [aOR] 2.26; 95% CI 1.46, 3.48). Factors associated with any fall in adjusted analyses included: depressive symptoms and neuropathy (both HIV+ and HIV-); age, marijuana use, multiple CNS medications, and HCV infection (HIV+ only); and cognitive complaints, quality of life, hypertension and obesity (HIV- only).ConclusionsMiddle-aged HIV+ and HIV- women had similar fall rates. Among HIV+ women, factors affecting cognition such as age, depressive symptoms, marijuana use and multiple CNS medications were important predictors of falls, however, cognitive complaints were not.

ObjectiveThe aim of this study was to determine the associations between visceral adipose tissue (VAT) and abdominal subcutaneous adipose tissue (SAT) mass with homeostatic model assessment-insulin resistance (HOMA-IR) and incidence of diabetes mellitus in women with and without HIV infection.DesignCross-sectional design for associations between abdominal fat and HOMA-IR; longitudinal design for associations between abdominal fat and incident diabetes.MethodsWe assessed associations between dual X-ray absorptiometry scan-derived VAT and SAT with HOMA-IR in a subsample from the Women's Interagency HIV Study (n = 226 with and n = 100 without HIV) using linear regression. We evaluated associations of VAT, SAT and HOMA-IR with incident diabetes mellitus using Cox proportional hazards models.ResultsVAT mass was positively associated with log HOMA-IR in fully adjusted linear regression models stratified by HIV serostatus, including adjustment for SAT. During median follow-up of 10.6 years, incidence of diabetes was 1.63 [95% confidence interval (95% CI) 1.15-2.31] and 1.32 [95% CI 0.77-2.28] cases per 100 person-years in women with and without HIV (P = 0.52). In a fully adjusted model, baseline VAT (hazard ratio 2.64 per kg; 95% CI 1.14-6.12; P = 0.023) and SAT (hazard ratio 1.34 per kg; 95% CI 0.73-2.45; P = 0.35) were associated with incident diabetes, but the latter was not statistically significant.ConclusionVAT mass was independently associated with HOMA-IR in women with and without HIV and was independently associated with future development of diabetes.

BackgroundWe previously reported that fracture incidence rates did not differ by HIV status among predominantly premenopausal Women's Interagency HIV Study participants. We now conduct a follow-up study with 5 additional observation years to further characterize fracture risk associated with HIV infection in women as they age.MethodsWe measured time to first new fracture at any site in 2375 (1713 HIV-infected and 662 HIV-uninfected) Women's Interagency HIV Study participants, with median 10-year follow-up. Fractures were self-reported semiannually. Proportional hazards models assessed predictors of incident fracture.ResultsAt index visit, HIV-infected women were older [median age of 40 years (IQR: 34-46) vs. 35 (27-43), P < 0.0001] and more likely to be postmenopausal, hepatitis C virus infected, and weigh less than HIV-uninfected women. Among HIV-infected women, mean CD4 count was 480 cells per microliter and 63% were taking highly active antiretroviral therapy. Unadjusted incidence rates of any fracture were higher in HIV-infected than in HIV-uninfected women [2.19/100 person-years (py) vs. 1.54/100 py, P = 0.002]. In multivariate models, HIV status, older age, white (vs. black) race, prior fracture, history of cocaine use, and history of injection drug use were significant predictors of incident fracture. Among HIV-infected women, age, white race, prior fracture, smoking, and prior AIDS were predictors of new fracture.ConclusionsMiddle-aged HIV-infected women had a higher adjusted fracture rate than HIV-uninfected women. Cocaine use and injection drug use were also associated with a greater risk of incident fracture. Further research is needed to understand whether the risk of fracture associated with cocaine use relates to increased rate of falls or direct effects on bone metabolism.

In HIV negatives, markers of hemostasis, including D-dimer, factor VIII, plasminogen activator inhibitor-1 antigen (PAI-1), and total protein S are associated with all-cause and cardiovascular disease mortality. In HIV positives, studies of D-dimer and factor VIII with death were limited to short follow-up; associations of PAI-1 and total protein S with death have not been examined. In 674 HIV-infected women from the Women's Interagency HIV Study, markers from the first visit after enrollment were exposures of interest in multivariate analyses of death (AIDS and non-AIDS) in separate models at 5 and 16 years. There were 87 AIDS and 44 non-AIDS deaths at 5 years, and 159 AIDS and 113 non-AIDS deaths at 16 years. An inverse association of total protein S quartiles with non-AIDS deaths was observed at 5 (P trend = 0.002) and 16 years (P trend = 0.02); there was no association with AIDS deaths. The third quartile of PAI-1 was associated with AIDS deaths at 5 [hazard ratio (HR) = 4.0; 95% confidence interval (CI): 1.9 to 8.4] and 16 years (HR = 3.4; 95% CI: 1.9 to 5.9); and with non-AIDS deaths at 5 years (HR = 4.8; 95% CI: 1.6 to 13.9). D-dimer and factor VIII were not associated with AIDS or non-AIDS death at 5 or 16 years. Lower total Protein S was a consistent marker of non-AIDS death. We found no association between D-dimer with AIDS or non-AIDS death, in contrast to previous studies showing increased short-term (<5 years) mortality, which may represent sex differences or population heterogeneity. Given longer survival on highly active antiretroviral therapy, further studies of these markers are needed to determine their prognostic value.

BackgroundCoinfection with HIV and hepatitis C virus (HCV) is common. HIV infection and treatment are associated with hypercoagulability; thrombosis in HCV is underinvestigated. Proposed markers of hemostasis in HIV include higher D-dimer, Factor VIII%, and plasminogen activator inhibitor-1 (PAI-1) antigen and lower total Protein S% (TPS) but have not been examined in HCV. We assessed the independent association of HCV with these 4 measures of hemostasis in a multicenter, prospective study of HIV: the Women's Interagency HIV Study.MethodsWe randomly selected 450 HCV-infected (anti-HCV+ with detectable plasma HCV RNA) and 450 HCV-uninfected (anti-HCV-) women. HCV was the main exposure of interest in regression models.ResultsFour hundred forty-three HCV+ and 425 HCV- women were included. HCV+ women had higher Factor VIII% (124.4% ± 3.9% vs. 101.8% ± 3.7%, P < 0.001) and lower TPS (75.7% ± 1.1% vs. 84.3% ± 1.1%, <0.001) than HCV- women, independent of HIV infection and viral load; there was little difference in PAI-1 or log10 D-dimer. After adjustment for confounders, these inferences remained. HIV infection was independently associated with higher Factor VIII% and log10 D-dimer and lower TPS.ConclusionsHCV was independently associated with higher Factor VIII% and lower TPS consistent with hypercoagulability. Higher Factor VIII% and D-dimer and lower TPS were also strongly associated with HIV infection and levels of HIV viremia, independent of HCV infection. Further investigation is needed to determine if there is increased thrombotic risk from HCV. Studies examining hemostasis markers in HIV infection must also assess the contribution of HCV infection.

An improved First Borel-Cantelli Lemma incorporating dependency structure between subsets of events is presented and applied to an example where the First Borel-Cantelli Lemma can't be used. Analogies to improved Bonferroni upper bounds are given.

Arnold and Groeneveld (1979) presented distribution free bounds for E(X(g:n)) the expected value of the 'gth' order statistic when sampling 'n' possibly dependent observations from populations each with finite expectations mu(i) and finite variances sigma squared(i). Here, the problem of incorporating known dependencies among the distributions of observations in a sample to produce bounds for the order statistics of that sample is considered. Limits are obtained for expectations of the 'gth' order statistic and the range. These limits hold even when the population variances are not finite. An application of the upper bound for E(X(n:n)) is given. Keywords: Order statistics, Statistical samples, Mathematical equations.

Dunnett (1955) developed a procedure comparing m treatments to one control with an exact overall type I error of alpha when all sampling distributions as normal. Sometimes it is desirable to compare m treatments to K > or = 2 controls. In particular, it is often desired to compare m treatments with two controls. For instance, several new treatments (e.g., pain relievers) could be compared to two standard treatments (e.g., Aspirin and Tylenol). Alternatively, a standard treatment could be very expensive, difficult to apply and/or have bad side effects, making it useful to compare each new treatment to both standard treatment and not treatment (Placebo). Dunnett's method is expanded here to give comparisons of mean values for m treatments to mean values for K > or = 2 controls at an exact overall type I error of a alpha when all sampling distributions are normal. Tabled values needed to make exact simultaneous comparisons at alpha =.05 are given for K=2. An application is made to an illustrative example for the literature.

Two related improved Bonferroni methods to obtain upper bounds for the probabilities of unions of events are described. One procedure (bivariate method) uses probabilities of events and pairwise intersections of events. The other superior and more complicated procedure (trivariate method) uses probabilities of pairwise and three way intersections of events. These new methods are applied to multivariate normal hypothesis testing and simultaneous unbiased confidence intervals and are shown to give results superior to those of the currently used procedure (conservative assumption and independence of events) if the number of variables is not too large and the data is highly correlated. Modifications of these new methods to non-normal probabilities and different types of probability regions are also mentioned. Keywords: Statistical processes.

Upper bounds to multivariate normal probability areas outside of n dimensional rectangles centered at the origin are of interest due to their applications in producing conservative simultaneous confidence intervals and hypothesis tests. The current procedure used to compute these upper bounds (Dunn-Sidak method) is based upon making the conservative assumption that the variables are independent. Three new approaches which give tighter (lower) upper bounds for such probability areas have been developed. The first of these (intersection subtraction) is an improved version of the Bonferonni upper bounds. The second of these methods (conditional multiplicative) requires that the multivariate normal distribution have the MTP-2 property. The third method (conservative independent subunit) is a more complicated form of the conservative assumption of independence among variables. These three methods are compared theoretically with the following results: 1) The conditional multiplicative, when it can be applied is better than the other two methods; and 2) The intersection subtraction is better than the conservative independent subunit when n is small, but becomes worse than the conservative independent subunit when n is small, subunit as n becomes larger.

The recent literature contains theorems improving on both the Standard Bonferroni inequality and the Sidak/Slepian inequalities. The application of these improved theorems to upper bounds for non-coverage of simultaneous confidence intervals on multivariate normal variables is explored. The improved Bonferroni upper bounds will always apply, while improved Sidak/ Slepian bounds only apply to special cases. The improved Sidak/Slepian upper bound, if it applies, is always superior to the equivalent improved Bonferroni bound. This improvement, however, is not great when both methods are used to determine upper bounds for Type I error in the range of .01 to .10. It is shown that improved Sidak/Slepian bounds will apply to Normal Markov Processes, a commonly occurring and easily identifiable class of multivariate normal variables.

This paper presents the Subset Complement Addition Upper Bound (SCAUB) procedure which produces upper bounds for probabilities of unions of n events given that probabilities of unions and/or intersections of subsets including up to k events are known. The SCAUB method is an extension of Hunter's (1976) improved Bonferroni bounds. The SCAUB inequality is much simpler to calculate than are other distribution free upper bounds proposed in the past. It is also a distribution free analog of Glaz and Johnson's (1984) product type bounds. We prove that for any fixed n events, the SCAUB inequality monotonically decreases with k. SCAUB upper bounds are applied to the multivariate normal (or t) simultaneous interference interval problem. Keywords: Multivariate normal distribution.

In time to response investigations, such as survival studies, observations are often censored. the censoring may or may not be independent of the response. A particular form of positive dependence between censoring time and death time is defined as response linked (RL) censoring. Many examples are given of response linked censoring. We present a general method for modeling and estimation to be used when both response linked and independent censoring occur. This method involves parametric modeling with (or without) covariates and handles the special cases of all response linked and all independent censoring. We illustrate in detail are response linked censoring and the method with the Standford heart transplant data. Survival data on radio tagged animals provides a second example. The effectiveness of our methods is tested with a lung cancer data set where real cancer deaths have been randomly RL censored. Several other application areas (e.g, study of reformed heroin addicts time to relapse) are mentioned. Keywords: Survival; Dependent censoring; Independent censoring; Response linked censoring; Covariates; Data analysis.