Gestational exposure to famine has been associated with several chronic diseases in adulthood, but few studies in humans have related prenatal famine exposure to health-related quality of life. We used the circumstances of the Dutch Famine of 1944-1945 (during which official rations were =900 kcal/day for 24 weeks) to assess whether exposure to famine prior to conception or at specified stages of pregnancy was related to self-reported health-related quality of life and depressive symptoms in adulthood.
We studied 923 individuals including persons born in western Holland between January 1945 and March 1946, persons born in the same 3 institutions in 1943 and 1947 and same-sex siblings of persons in series 1 or 2. Between 2003 and 2005 (mean age 59 y), we assessed self-reported quality of life with the Short Form 36 questionnaire and derived mental and physical component scores. Depressive symptoms were assessed with the Center for Epidemiologic Studies Depression scale.
Mean mental and physical component scores were 52.4 (SD = 9.4) and 48.9 (9.0), respectively. The mean depression score was 11.6 (7.4). Age-, sex- and schooling-adjusted estimates for mutually adjusted exposures were -2.48 for the mental component score with exposure before conception (95% confidence interval = -4.46 to 0.50) and 0.07 with exposure during pregnancy (-1.15 to 1.29). Adjusted estimates for the physical component score were 1.26 with exposure before conception (-0.67 to 3.19) and -0.73 with exposure during pregnancy (1.94 to 0.48). Adjusted estimates for the depression score were 2.07 with exposure before conception (0.60 to 3.54) and 0.96 with exposure during pregnancy (0.09 to 1.88). There was no evidence of heterogeneity of effects by specific periods of pregnancy exposed to famine.
A mother's exposure to famine prior to conception of her offspring was associated with lower self-reported measures of mental health and quality of life in her adult offspring.
Obesity is a risk factor for renal cell (or renal) cancer. The increasing prevalence of obesity may be contributing to the rising incidence of this cancer over the past several decades. The effects of early-age obesity and change in body mass index (BMI) on renal cancer have been studied less thoroughly, and the influence of race has never been formally investigated.
Using data gathered as part of a large case-control study of renal cancer (1,214 cases and 1,234 controls), we investigated associations with BMI at several time points, as well as with height. Adjusted odds ratios (ORs) and 95% confidence intervals (CIs) were computed using logistic regression modeling. Race- and sex-stratified analyses were conducted to evaluate subgroup differences.
Obesity (BMI ≥ 30 kg/m2) early in adulthood (OR=1.6 [95% CI=1.1 to 2.4]) and 5 years before diagnosis (1.6 [1.1 to 2.2]) was associated with renal cancer. The association with early-adult obesity was stronger among whites than blacks (Test for interaction, P=0.006), while the association with obesity near diagnosis was marginally stronger in women than men (Test for interaction, P=0.08). The strongest association with renal cancer was observed for obese whites both in early adulthood and prior to interview (2.6 [1.5 to 4.4]); this association was not present among blacks. Estimates of the annual excess rate of renal cancer (per 100,000 persons) attributed to both overweight and obesity (BMI > 25 kg/m2) ranged from 9.9 among black men to 5.6 among white women.
Obesity, both early and later in life, is associated with an increased risk of renal cancer. The association with early obesity appears to be stronger among whites than blacks.
Causal inference; infectious disease; infectiousness; interference; principal stratification; vaccine efficacy
Existing methods for estimation of mortality attributable to influenza are limited by methodological and data uncertainty. We have used proxies for disease incidence of the three influenza co-circulating subtypes (A/H3N2, A/H1N1 and B) that combine data on influenza-like illness consultations and respiratory specimen testing to estimate influenza-associated mortality in the US between 1997 and 2007.
Weekly mortality rate for several mortality causes potentially affected by influenza was regressed linearly against subtype-specific influenza incidence proxies, adjusting for temporal trend and seasonal baseline, modeled by periodic cubic splines.
Average annual influenza-associated mortality rates per 100,000 individuals were estimated for the following underlying causes of death: for pneumonia and influenza, 1.73 (95% confidence interval= 1.53 to 1.93); for chronic lower respiratory disease, 1.70 (1.48 to 1.93); for all respiratory causes, 3.58 (3.04 to 4.14); for myocardial infarctions, 1.02 (0.85 to 1.2); for ischemic heart disease, 2.7 (2.23 to 3.16); for heart disease, 3.82 (3.21 to 4.4); for cerebrovascular deaths, 0.65 (0.51 to 0.78); for all circulatory causes, 4.6 (3.79 to 5.39); for cancer, 0.87 (0.68 to 1.05); for diabetes, 0.33 (0.26 to 0.39); for renal disease, 0.19 (0.14 to 0.24); for Alzheimer disease, 0.41 (0.3 to 0.52); and for all causes, 11.92 (10.17 to 13.67). For several underlying causes of death, baseline mortality rates changed after the introduction of the pneumococcal conjugate vaccine.
The proposed methodology establishes a linear relation between influenza incidence proxies and excess mortality, rendering temporally consistent model fits, and allowing for the assessment of related epidemiologic phenomena such as changes in mortality baselines.
Recently, researchers have used a potential-outcome framework to estimate causally interpretable direct and indirect effects of an intervention or exposure on an outcome. One approach to causal-mediation analysis uses the so-called mediation formula to estimate the natural direct and indirect effects. This approach generalizes classical mediation estimators and allows for arbitrary distributions for the outcome variable and mediator. A limitation of the standard (parametric) mediation formula approach is that it requires a specified mediator regression model and distribution; such a model may be difficult to construct and may not be of primary interest. To address this limitation, we propose a new method for causal-mediation analysis that uses the empirical distribution function, thereby avoiding parametric distribution assumptions for the mediator. In order to adjust for confounders of the exposure-mediator and exposure-outcome relationships, inverse-probability weighting is incorporated based on a supplementary model of the probability of exposure. This method, which yields estimates of the natural direct and indirect effects for a specified reference group, is applied to data from a cohort study of dental caries in very-low-birth-weight adolescents to investigate the oral-hygiene index as a possible mediator. Simulation studies show low bias in the estimation of direct and indirect effects in a variety of distribution scenarios, whereas the standard mediation formula approach can be considerably biased when the distribution of the mediator is incorrectly specified.
Some environmental chemical exposures are lipophilic and need to be adjusted by serum lipid levels before data analyses. There are currently various strategies that attempt to account for this problem, but all have their drawbacks. To address such concerns, we propose a new method that uses Box-Cox transformations and a simple Bayesian hierarchical model to adjust for lipophilic chemical exposures.
We compared our Box-Cox method to existing methods. We ran simulation studies in which increasing levels of lipid-adjusted chemical exposure did and did not increase the odds of having a disease, and we looked at both single-exposure and multiple-exposures cases. We also analyzed an epidemiology dataset that examined the effects of various chemical exposures on the risk of birth defects.
Compared with existing methods, our Box-Cox method produced unbiased estimates, good coverage, similar power, and lower type-I error rates. This was the case in both single- and multiple-exposure simulation studies. Results from analysis of the birth-defect data differed from results using existing methods.
Our Box-Cox method is a novel and intuitive way to account for the lipophilic nature of certain chemical exposures. It addresses some of the problems with existing methods, is easily extendable to multiple exposures, and can be used in any analyses that involve concomitant variables.
Ill-defined causal questions present serious problems for observational studies—problems that are largely unappreciated. This paper extends the usual counterfactual framework to consider causal questions about compound treatments for which there are many possible implementations (for example, “prevention of obesity”). We describe the causal effect of compound treatments and their identifiability conditions, with a special emphasis on the consistency condition. We then discuss the challenges of using the estimated effect of a compound treatment in one study population to inform decisions in the same population and in other populations. These challenges arise because the causal effect of compound treatments depends on the distribution of the versions of treatment in the population. Such causal effects can be unpredictable when the versions of treatment are unknown. We discuss how such issues of “transportability” are related to the consistency condition in causal inference. With more carefully framed questions, the results of epidemiologic studies can be of greater value to decision-makers.
In vaccine trials, the vaccination of one person might prevent the infection of another; a distinction can be drawn between the ways such a protective effect might arise. Consider a setting with 2 persons per household in which one of the 2 is vaccinated. Vaccinating the first person may protect the second person by preventing the first from being infected and passing the infection on to the second. Alternatively, vaccinating the first person may protect the second by rendering the infection less contagious even if the first is infected. This latter mechanism is sometimes referred to as an “infectiousness effect” of the vaccine. Crude estimators for the infectiousness effect will be subject to selection bias due to stratification on a postvaccination event, namely the infection status of the first person. We use theory concerning causal inference under interference along with a principal-stratification framework to show that, although the crude estimator is biased, it is, under plausible assumptions, conservative for what one might define as a causal infectiousness effect. This applies to bias from selection due to the persons in the comparison, and also to selection due to pathogen virulence. We illustrate our results with an example from the literature.
An emerging body of evidence suggests that ambient levels of air
pollution during pregnancy are associated with preterm birth.
To further investigate these relationships we used vital record data
to construct a retrospective cohort of 476,489 births occurring between 1994
and 2004 in five central counties of metropolitan Atlanta. Using a
time-series approach, we examined aggregated daily counts of preterm birth
in relation to ambient levels of carbon monoxide, nitrogen dioxide, sulfur
dioxide, ozone, particulate matter < 10 μm in diameter
(PM10), particulate matter < 2.5 μm in diameter
(PM2.5) and speciated PM measurements. Daily pollutant levels
in five-county Atlanta were characterized using a population-weighted
spatial average of air quality monitors in the study area. We also examined
ambient concentrations at individual monitors in analyses limited to mothers
with residential geocodes within four miles of each monitor. Relationships
between average pollution levels during three gestational windows of
interest were modeled using Poisson generalized linear models. Results were
adjusted for seasonal and long-term time trends.
Although most results were null, there were three positive
associations between ambient pollution levels and preterm birth in the
four-mile capture-area analyses. Daily preterm birth rates were associated
with average NO2 concentrations in the preceding six weeks and
with average PM2.5 sulfate and PM2.5 water-soluble
metal concentrations in the preceding week.
Results provide limited support for late-pregnancy effects of ambient
air pollution on preterm birth.
The causal inference literature has provided definitions of direct and indirect effects based on counterfactuals that generalize the approach found in the social science literature. However, these definitions presuppose well defined hypothetical interventions on the mediator. In many settings there may be multiple ways to fix the mediator to a particular value and these different hypothetical interventions may have very different implications for the outcome of interest. In this paper we consider mediation analysis when multiple versions of the mediator are present. Specifically, we consider the problem of attempting to decompose a total effect of an exposure on an outcome into the portion through the intermediate and the portion through other pathways. We consider the setting in which there are multiple versions of the mediator but the investigator only has access to data on the particular measurement, not which version of the mediator may have brought that value about. We show that the quantity that is estimated as a natural indirect effect using only the available data does indeed have an interpretation as a particular type of mediated effect; however, the quantity estimated as a natural direct effect in fact captures both a true direct effect and an effect of the exposure on the outcome mediated through the effect of the version of the mediator that is not captured by the mediator measurement. The results are illustrated using two examples from the literature, one in which the versions of the mediator are unknown and another in which the mediator itself has been dichotomized.
The measurement of area-level attributes remains a major challenge in studies of neighborhood health effects. Even when neighborhood survey data are collected, they necessarily have incomplete spatial coverage. We investigated whether interpolation of neighborhood survey data was aided by information on spatial dependencies and supplementary data. Neighborhood “availability of healthy foods” was measured in a population-based survey of 5186 persons in Baltimore, New York, and Forsyth County (North Carolina). The following supplementary data were compiled from Census 2000 and InfoUSA, Inc.: distance to supermarkets, density of supermarkets and fruit and vegetable stores, housing density, distance to a high-income area, and percent of households that do not own a vehicle. We compared 4 interpolation models (ordinary least squares, residual kriging, spatial error regression, and thin-plate splines) using error statistics and Pearson correlation coefficients (r) from repeated replications of cross-validations. There was positive spatial autocorrelation in neighborhood availability of healthy foods (by site, Moran coefficient range = 0.10–0.28; all P < 0.0001). Prediction performances were generally similar for the evaluated models (r ≈ 0.35 for Baltimore and Forsyth; r ≈ 0.54 for New York). Supplementary data accounted for much of the spatial autocorrelation and, thus, spatial modeling was only advantageous when spatial correlation was at least moderate. A variety of interpolation techniques will likely need to be utilized in order to increase the data available for examining health effects of residential environments. The most appropriate method will vary depending on the construct of interest, availability of relevant supplementary data, and types of observed spatial patterns.
Major depressive disorder as well as the use of serotonin reuptake inhibitors in pregnancy have been associated with preterm birth. Studies that have attempted to separate effects of illness from treatment have been inconclusive. We sought to explore the separate effects of serotonin reuptake inhibitor use and major depressive episodes in pregnancy on risk of preterm birth.
We conducted a prospective cohort study of 2793 pregnant women, oversampled for a recent episode of major depression or use of a serotonin reuptake inhibitor. We extracted data on birth outcomes from hospital charts and used binary logistic regression to model preterm birth (<37 weeks’ gestation). We used ordered logistic regression to model early (<34 weeks’ gestation) or late (34-36 weeks) preterm birth, and we used nominal logistic regression to model preterm birth antecedents (spontaneous preterm labor/preterm premature rupture of membranes/preterm for medical indications/term).
Use of a serotonin reuptake inhibitor, both with (odds ratio=2.1 [95% confidence interval=1.0—4.6]) and without (1.6=[1.0—2.5]) a major depressive episode, was associated with preterm birth. A major depressive episode without serotonin reuptake inhibitor use (1.2; [0.68—2.1]) had no clear effect on preterm risk. None of these exposures was associated with early preterm birth. Use of serotonin reuptake inhibitors in pregnancy was associated with increases in spontaneous but not medically indicated preterm birth.
Serotonin reuptake inhibitor use increased risk of preterm birth. Although the effect of a major depressive episode alone was unclear, symptomatic women undergoing antidepressant treatment had elevated risk.
Dichlorodiphenyltrichloroethane (DDT) continues to be used for control of infectious diseases in several countries. In-utero exposure to DDT and dichlorodiphenyldichloroethylene (DDE) has been associated with developmental and cognitive impairment among children. We examined this association in an historical cohort in which the level of exposure was greater than in previous studies.
The association of in-utero DDT and DDE exposure with infant and child neurodevelopment was examined in approximately 1100 subjects in the Collaborative Perinatal Project, a prospective birth cohort enrolling pregnant women from 12 study centers in the U.S. from 1959 to 1965. Maternal DDT and DDE concentrations were measured in archived serum specimens. Infant mental and motor development was assessed at age 8 months using the Bayley Scales of Infant Development, and child cognitive development was assessed at age 7 years using the Wechsler Intelligence Scale for Children.
Although levels of both DDT and DDE were relatively high in this population (median DDT concentration, 8.9 µg/L; DDE, 24.5 µg/L), neither was related to Mental or Psychomotor Development scores on the Bayley Scales or to Full-Scale IQ at 7 years of age. Categorical analyses showed no evidence of dose-response for either maternal DDT or DDE, and estimates of the association between continuous measures of exposure and neurodevelopment were indistinguishable from 0.
Adverse associations were not observed between maternal serum DDT and DDE concentrations and offspring neurodevelopment at 8 months or 7 years of age in this cohort.
Vaccination of one person may prevent the infection of another either because the vaccine prevents the first from being infected and from infecting the second, or because, even if the first person is infected, the vaccine may render the infection less infectious. We might refer to the first of these mechanisms as a contagion effect and the second as an infectiousness effect. In the simple setting of a randomized vaccine trial with households of size two, we use counterfactual theory under interference to provide formal definitions of a contagion effect and an unconditional infectiousness effect. Using ideas analogous to mediation analysis, we show that the indirect effect (the effect of one person’s vaccine on another’s outcome) can be decomposed into a contagion effect and an unconditional infectiousness effect on the risk-difference, risk-ratio, odds-ratio and vaccine-efficacy scales. We provide identification assumptions for such contagion and unconditional infectiousness effects, and describe a simple statistical technique to estimate these effects when they are identified. We also give a sensitivity-analysis technique to assess how inferences would change under violations of the identification assumptions. The concepts and results of this paper are illustrated with hypothetical vaccine-trial data.
Stochastic transmission models are highly important in infectious disease epidemiology. The quantity of data produced by these models is challenging to display and communicate. A common approach is to display the model results in the familiar form of a mean or median and 95% interval, plotted over time. This approach has drawbacks, however, including the potential for ambiguity and misinterpretation of model results. Instead, we propose two alternative approaches for visualizing results from stochastic models. These proposed approaches convey the information provided by the median and 95% interval, as well as information about unexpected outcomes that may be of particular interest for stochastic epidemic models.
Arsenic exposure has been linked to epigenetic modifications such as DNA methylation in in vitro and animal studies. This association has also been explored in highly exposed human populations, but studies among populations environmentally exposed to low arsenic levels are lacking.
We evaluated the association between exposure to arsenic, measured in toenails, and blood DNA methylation in Alu and Long Interspersed Nucleotide Element-1 (LINE-1) repetitive elements in elderly men environmentally exposed to low levels of arsenic. We also explored potential effect modification by plasma folate, cobalamin (vitamin B12), and pyridoxine (vitamin B6). The study population was 581 participants from the Normative Aging Study in Boston, of whom 434, 140, and 7 had 1, 2, and 3 visits, respectively, between 1999-2002 and 2006-2007. We used mixed-effects models and included interaction terms to assess potential effect modification by nutritional factors.
There was a trend of increasing Alu and decreasing LINE-1 DNA methylation as arsenic exposure increased. In subjects with plasma folate below the median (< 14.1 ng/ml), arsenic was positively associated with Alu DNA methylation (β=0.08 [95% confidence interval = 0.03 to 0.13] for one interquartile range [0.06μg/g] increase in arsenic) while a negative association was observed in subjects with plasma folate above the median (β=-0.08 [-0.17 to 0.01]).
We found an association between arsenic exposure and DNA methylation in Alu repetitive elements that varied by folate level. This suggests a potential role for nutritional factors in arsenic toxicity.
Although the association between exposure to particulate matter (PM) mass and mortality is well established, there remains uncertainty about which chemical components of PM are most harmful to human health.
A hierarchical approach was used to determine how the association between daily PM2.5 mass and mortality was modified by PM2.5 composition in 25 US communities. First, the association between daily PM2.5 and mortality was determined for each community and season using Poisson regression. Second, we used meta-regression to examine how the pooled association was modified by community and season-specific particle composition.
There was a 0.74% (95% confidence interval = 0.41%–1.07%) increase in nonaccidental deaths associated with a 10 μg/m3 increase in 2-day averaged PM2.5 mass concentration. This association was smaller in the west (0.51% [0.10%– 0.92%]) than in the east (0.92% [0.23%–1.36%]), and was highest in spring (1.88% [0.23%–1.36%]). It was increased when PM2.5 mass contained a higher proportion of aluminum (interquartile range = 0.58%), arsenic (0.55%), sulfate (0.51%), silicon (0.41%), and nickel (0.37%). The combination of aluminum, sulfate, and nickel also modified the effect. These species proportions explained residual variability between the community-specific PM2.5 mass effect estimates.
This study shows that certain chemical species modify the association between PM2.5 and mortality and illustrates that mass alone is not a sufficient metric when evaluating health effects of PM exposure.
A major portion of influenza disease burden during the 2009 pandemic was observed among young people.
We examined the effect of age on the transmission of influenza-like illness associated with the 2009 pandemic influenza A (H1N1) virus (pH1N1) for an April–May 2009 outbreak among youth-camp participants and household contacts in Washington State.
An influenza-like illness attack rate of 51% was found among 96 camp participants. We observed a cabin secondary attack rate of 42% (95% confidence interval = 21%–66%) and a camp local reproductive number of 2.7 (1.7–4.1) for influenza-like illness among children (less than 18 years old). Among the 136 contacts in the 41 households with an influenza-like illness index case who attended the camp, the influenza-like illness secondary attack rate was 11% for children (5%–21%) and 4% for adults (2%–8%). The odds ratio for influenza-like illness among children versus adults was 3.1 (1.3–7.3).
The strong age effect, combined with the low number of susceptible children per household (1.2), plausibly explains the lower-than-expected household secondary attack rate for influenza-like illness, illustrating the importance of other venues where children congregate for sustaining community transmission. Quantifying the effects of age on pH1N1 transmission is important for informing effective intervention strategies.
A strength of time-series analyses is the inherent control of individual-level risk factors that do not vary temporally. However, in studies of adverse pregnancy outcomes, risk factors considered time-invariant at the individual level may vary seasonally when aggregated into a pregnancy risk set. To illustrate, we describe the seasonal patterns of birth in Atlanta and demonstrate how these patterns could lead to confounding in time-series studies of seasonally-varying exposures and preterm birth.
The study cohort included all births in 20-county metropolitan Atlanta delivered during the period 1994–2004 (n=715,875). We assessed the seasonal patterns of estimated conception and birth for the full cohort and for subgroups stratified by sociodemographic factors. Based on the observed patterns, we quantified the degree of potential confounding created by (1) differences in the gestational age distribution in the risk set across calendar months and (2) differences in the sociodemographic composition of the risk set across calendar months.
The overall seasonal pattern of birth was characterized by a peak in August–September and troughs in April–May and November–January. Seasonal patterns differed among racial and ethnic groups, maternal education levels, and marital status. As a consequence of these seasonal patterns, systematic seasonal differences in the gestational age distribution and the sociodemographic composition of the risk set led to differences in expected rates of preterm birth across calendar months.
Time-series investigations of seasonally-varying exposures and adverse pregnancy outcomes should consider the potential for bias due to seasonal heterogeneity in the risk set.
A difficult issue in observational studies is assessment of whether important confounders are omitted or misspecified. Here, we present a method for assessing whether residual confounding is present. Our method depends on availability of an indicator with two key characteristics: first, it is conditionally independent (given measured exposures and covariates) of the outcome in the absence of confounding, misspecification and measurement errors; second, it is associated with the exposure and, like the exposure, with any unmeasured confounders.
We demonstrate the method using a time-series study of the effects of ozone on emergency department visits for asthma in Atlanta. We argue that future air pollution may have the characteristics appropriate for an indicator, in part because future ozone cannot have caused yesterday’s health events. Using directed acyclic graphs and specific causal relationships, we show that one can identify residual confounding using an indicator with the stated characteristics. We use simulations to assess the discriminatory ability of future ozone as an indicator of residual confounding in the association of ozone with asthma-related emergency department visits. Parameter choices are informed by observed data for ozone, meteorologic factors and asthma.
In simulations, we found that ozone concentrations one day after the emergency department visits had excellent discriminatory ability to detect residual confounding by some factors that were intentionally omitted from the model, but weaker ability for others. Although not the primary goal, the indicator can also signal other forms of modeling errors, including substantial measurement error, and does not distinguish between them.
The simulations illustrate that the indicator based on future air pollution levels can have excellent discriminatory ability for residual confounding, although performance varied by situation. Application of the method should be evaluated by considering causal relationships for the intended application, and should be accompanied by other approaches, including evaluation of a priori knowledge.
Animal studies have demonstrated that timing of pubertal onset can be altered by prenatal exposure to dioxins or polychlorinated biphenyls (PCBs), but studies of human populations have been quite limited.
We assessed the association between maternal serum concentrations of dioxins and PCBs and the sons’ age of pubertal onset in a prospective cohort of 489 mother–son pairs from Chapaevsk, Russia, a town contaminated with these chemicals during past industrial activity. The boys were recruited at ages 8 to 9 years, and 4 years of annual follow-up data were included in the analysis. Serum samples were collected at enrollment from both mothers and sons for measurement of dioxin and PCB concentrations using high-resolution mass spectrometry. The sons’ pubertal onset—defined as pubertal stage 2 or higher for genitalia (G) or pubic hair (P), or testicular volume >3 mL—was assessed annually by the same physician.
In multivariate Cox models, elevated maternal serum PCBs were associated with earlier pubertal onset defined by stage G2 or higher (4th quartile hazard ratio = 1.7 [95% confidence interval = 1.1– 2.5]), but not for stage P2 or higher or for testicular volume >3 mL. Maternal serum concentrations of dioxin toxic equivalents were not consistently associated with the sons’ pubertal onset, although a dose-related delay in pubertal onset (only for G2 or higher) was seen among boys who breast-fed for 6 months or more.
Maternal PCB serum concentrations measured 8 or 9 years after sons’ births—which may reflect sons’ prenatal and early-life exposures—were associated with acceleration in some, but not all, measures of pubertal onset.
Highly active antiretroviral therapy (HAART) rapidly suppresses human immunodeficiency virus (HIV) viral replication and reduces circulating viral load, but the long-term effects of HAART on viral load remain unclear.
We evaluated HIV viral load trajectories over 8 years following HAART initiation in the Multicenter AIDS Cohort Study and the Women’s Interagency HIV Study. The study included 157 HIV-infected men and 199 HIV-infected women who were antiretroviral naïve and contributed 1311 and 1837 semiannual person-visits post-HAART, respectively. To account for within-subject correlation and the high proportion of left-censored viral loads, we used a segmental Bernoulli/lognormal random effects model.
Approximately 3 months (0.30 years for men and 0.22 years for women) after HAART initiation, HIV viral loads were optimally suppressed (ie, with very low HIV RNA) for 44% (95% confidence interval = 39%–49%) of men and 43% (38%–47%) of women, whereas the other 56% of men and 57% of women had on average 2.1 (1.5–2.6) and 3.0 (2.7–3.2) log10 copies/mL, respectively.
After 8 years on HAART, 75% of men and 80% of women had optimal suppression, whereas the rest of the men and women had suboptimal suppression with a median HIV RNA of 3.1 and 3.7 log10 copies/mL, respectively.
The Women’s Health Initiative randomized trial found greater coronary heart disease (CHD) risk in women assigned to estrogen/progestin therapy than in those assigned to placebo. Observational studies had previously suggested reduced CHD risk in hormone users.
Using data from the observational Nurses’ Health Study, we emulated the design and intention-to-treat (ITT) analysis of the randomized trial. The observational study was conceptualized as a sequence of “trials” in which eligible women were classified as initiators or noninitiators of estrogen/progestin therapy.
The ITT hazard ratios (95% confidence intervals) of CHD for initiators versus noninitiators were 1.42 (0.92 – 2.20) for the first 2 years, and 0.96 (0.78 – 1.18) for the entire follow-up. The ITT hazard ratios were 0.84 (0.61 – 1.14) in women within 10 years of menopause, and 1.12 (0.84 – 1.48) in the others (P value for interaction = 0.08). These ITT estimates are similar to those from the Women’s Health Initiative. Because the ITT approach causes severe treatment misclassification, we also estimated adherence-adjusted effects by inverse probability weighting. The hazard ratios were 1.61 (0.97 – 2.66) for the first 2 years, and 0.98 (0.66 – 1.49) for the entire follow-up. The hazard ratios were 0.54 (0.19 – 1.51) in women within 10 years after menopause, and 1.20 (0.78 – 1.84) in others (P value for interaction = 0.01). Finally, we also present comparisons between these estimates and previously reported NHS estimates.
Our findings suggest that the discrepancies between the Women’s Health Initiative and Nurses’ Health Study ITT estimates could be largely explained by differences in the distribution of time since menopause and length of follow-up.