Elizabeth L. Ogburn

Demonstrating stratification in a European American population

Population stratification occurs in case-control association studies when allele frequencies differ between cases and controls because of ancestry. Stratification may lead to false positive associations, although this issue remains controversial. Empirical studies have found little evidence of stratification in European-derived populations, but potentially significant levels of stratification could not be ruled out. We studied a European American panel discordant for height, a heritable trait that varies widely across Europe. Genotyping 178 SNPs and applying standard analytical methods yielded no evidence of stratification. But a SNP in the gene LCT that varies widely in frequency across Europe was strongly associated with height (P < 10−6). This apparent association was largely or completely due to stratification; rematching individuals on the basis of European ancestry greatly reduced the apparent association, and no association was observed in Polish or Scandinavian individuals. The failure of standard methods to detect this stratification indicates that new methods may be required.

Marital satisfaction and break-ups differ across on-line and off-line meeting venues

Marital discord is costly to children, families, and communities. The advent of the Internet, social networking, and on-line dating has affected how people meet future spouses, but little is known about the prevalence or outcomes of these marriages or the demographics of those involved. We addressed these questions in a nationally representative sample of 19,131 respondents who married between 2005 and 2012. Results indicate that more than one-third of marriages in America now begin on-line. In addition, marriages that began on-line, when compared with those that began through traditional off-line venues, were slightly less likely to result in a marital break-up (separation or divorce) and were associated with slightly higher marital satisfaction among those respondents who remained married. Demographic differences were identified between respondents who met their spouse through on-line vs. traditional off-line venues, but the findings for marital break-up and marital satisfaction remained significant after statistically controlling for these differences. These data suggest that the Internet may be altering the dynamics and outcomes of marriage itself.

Unconventional natural gas development and birth outcomes in Pennsylvania, USA

Background: Unconventional natural gas development has expanded rapidly. In Pennsylvania, the number of producing wells increased from 0 in 2005 to 3,689 in 2013. Few publications have focused on unconventional natural gas development and birth outcomes. Methods: We performed a retrospective cohort study using electronic health record data on 9,384 mothers linked to 10,946 neonates in the Geisinger Health System from January 2009 to January 2013. We estimated cumulative exposure to unconventional natural gas development activity with an inverse-distance squared model that incorporated distance to the mother’s home; dates and durations of well pad development, drilling, and hydraulic fracturing; and production volume during the pregnancy. We used multilevel linear and logistic regression models to examine associations between activity index quartile and term birth weight, preterm birth, low 5-minute Apgar score and small size for gestational age birth, while controlling for potential confounding variables. Results: In adjusted models, there was an association between unconventional natural gas development activity and preterm birth that increased across quartiles, with a fourth quartile odds ratio of 1.4 (95% confidence interval = 1.0, 1.9). There were no associations of activity with Apgar score, small for gestational age birth, or term birth weight (after adjustment for year). In a posthoc analysis, there was an association with physician-recorded high-risk pregnancy identified from the problem list (fourth vs. first quartile, 1.3 [95% confidence interval = 1.1, 1.7]). Conclusion: Prenatal residential exposure to unconventional natural gas development activity was associated with two pregnancy outcomes, adding to evidence that unconventional natural gas development may impact health. See Video Abstract at http://links.lww.com/EDE/B14.

The role of measurement error and misclassification in mediation analysis: mediation and measurement error.

Methods to estimate direct and indirect effects have been rapidly expanding.1–14 It is now well documented that such mediation analyses are subject to strong no-confounding assumptions and that an unmeasured confounder of the mediator-outcome relationship can lead to substantial bias in direct and indirect effect estimates.1,2,6,7,14,15 Much less attention has been given to the question of how measurement error may bias estimates of direct and indirect effects. le Cessie and colleagues.16 have done a service to investigators interested in direct effects by providing simple correction formulas for direct effects estimates in a variety of mediator measurement-error scenarios. Here we will consider the implications of these and other results as they relate to making inferences, not just about direct effects, but also about mediation and indirect effects.

Causal diagrams for interference

The term “interference” has been used to describe any setting in which one subject’s exposure may affect another subject’s outcome. We use causal diagrams to distinguish among three causal mechanisms that give rise to interference. The first causal mechanism by which interference can operate is a direct causal effect of one individual’s treatment on another individual’s outcome; we call this direct interference. Interference by contagion is present when one individual’s outcome may affect the outcomes of other individuals with whom he comes into contact. Then giving treatment to the first individual could have an indirect effect on others through the treated individual’s outcome. The third pathway by which interference may operate is allocational interference. Treatment in this case allocates individuals to groups; through interactions within a group, individuals may affect one another’s outcomes in any number of ways. In many settings, more than one type of interference will be present simultaneously. The causal effects of interest differ according to which types of interference are present, as do the conditions under which causal effects are identifiable. Using causal diagrams for interference, we describe these differences, give criteria for the identification of important causal effects, and discuss applications to infectious diseases.

High HIV prevalence and incidence among MSM across 12 cities in India.

Objective:To characterize prevalence, incidence, and associated correlates of HIV infection among MSM in 12 cities across India. Design:Cross-sectional sample using respondent-driven sampling from September 2012 to June 2013. Methods:A total 12 022 MSM (∼1000/city) were recruited. Participants had to be at least 18 years, self-identify as male, and report oral/anal intercourse with a man in the prior year. HIV infection was diagnosed using three rapid tests. Cross-sectional HIV incidence was estimated using a multiassay algorithm. All estimates incorporate respondent-driven sampling-II weights. Results:Median age was 25 years, 45% self-identified as ‘panthi’ (predominantly penetrative anal intercourse) and 30.6% reported being married to a woman. Weighted HIV prevalence was 7.0% (range: 1.7–13.1%). In multivariate analysis, significantly higher odds of HIV infection was observed among those who were older, had lower educational attainment, were practicing purely receptive anal sex or both receptive and penetrative sex, and those who were herpes simplex virus-2 positive. Of 1147 MSM who tested HIV positive, 53 were identified as recent HIV infections (annualized incidence = 0.87%; range = 0–2.2%). In multivariate analysis, injecting drugs in the prior 6 months, syphilis, and higher number of male partners and fewer female partners were significantly associated with recent HIV infection. Conclusion:We observed a high burden of HIV among MSM in India with tremendous diversity in prevalence, incidence, and risk behaviors. In particular, we observed high incidence in areas with relatively low prevalence suggesting emerging epidemics in areas not previously recognized to have high HIV burden.

On the nondifferential misclassification of a binary confounder.

Consider a study with binary exposure, outcome, and confounder, where the confounder is nondifferentially misclassified. Epidemiologists have long accepted the unproven but oft-cited result that, if the confounder is binary, then odds ratios, risk ratios, and risk differences that control for the mismeasured confounder will lie between the crude and the true measures. In this paper, we provide an analytic proof of the result in the absence of a qualitative interaction between treatment and confounder, and we demonstrate via counterexample that the result need not hold when there is such a qualitative interaction. We also present an analytic proof of the result for the effect of treatment among the treated and describe extensions to measures conditional on or standardized over other covariates.

High HIV burden among people who inject drugs in 15 Indian cities.

Background:Injecting drug use has historically been the principal driver of the HIV epidemic in the northeast states of India. However, recent data indicate growing numbers of people who inject drugs (PWIDs) in north and central Indian cities. Methods:We conducted face-to-face surveys among PWIDs in seven northeast and eight north/central Indian cities using respondent-driven sampling. We used a rapid HIV-testing protocol to identify seropositive individuals and multiassay algorithm to identify those with recent infection. We used multilevel regression models that incorporated sampling weights and had random intercepts for site to assess risk factors for prevalent and incident (recent) HIV infection. Results:We surveyed 14 481 PWIDs from 15 Indian cities between January and December 2013. Participants reported high rates of needle/syringe sharing. The median (site range) estimated HIV prevalence and incidence were 18.1% (5.9, 44.9) and 2.9 per 100 person-years (0, 12.4), respectively. HIV prevalence was higher in northeast sites, whereas HIV incidence was higher in north/central sites. The odds of prevalent HIV were over three-fold higher in women than in men. Other factors associated with HIV prevalence or incidence included duration since first injection, injection of pharmaceutical drugs, and needle/syringe sharing. Conclusions:The burden of HIV infection is high among PWIDs in India, and may be increasing in cities where injecting drug use is emerging. Women who inject drugs were at substantially higher risk for HIV than men – a situation that may be mediated by dual injection-related and sexual risks.

Predictors of Indoor Radon Concentrations in Pennsylvania, 1989–2013

Background Radon is the second-leading cause of lung cancer worldwide. Most indoor exposure occurs by diffusion of soil gas. Radon is also found in well water, natural gas, and ambient air. Pennsylvania has high indoor radon concentrations; buildings are often tested during real estate transactions, with results reported to the Department of Environmental Protection (PADEP). Objectives We evaluated predictors of indoor radon concentrations. Methods Using first-floor and basement indoor radon results reported to the PADEP between 1987 and 2013, we evaluated associations of radon concentrations (natural log transformed) with geology, water source, building characteristics, season, weather, community socioeconomic status, community type, and unconventional natural gas development measures based on drilled and producing wells. Results Primary analysis included 866,735 first measurements by building, with the large majority from homes. The geologic rock layer on which the building sat was strongly associated with radon concentration (e.g., Axemann Formation, median = 365 Bq/m3, IQR = 167–679 vs. Stockton Formation, median = 93 Bq/m3, IQR = 52–178). In adjusted analysis, buildings using well water had 21% higher concentrations (β = 0.191, 95% CI: 0.184, 0.198). Buildings in cities (vs. townships) had lower concentrations (β = –0.323, 95% CI: –0.333, –0.314). When we included multiple tests per building, concentrations declined with repeated measurements over time. Between 2005 and 2013, 7,469 unconventional wells were drilled in Pennsylvania. Basement radon concentrations fluctuated between 1987 and 2003, but began an upward trend from 2004 to 2012 in all county categories (p < 0.001), with higher levels in counties having ≥ 100 drilled wells versus counties with none, and with highest levels in the Reading Prong. Conclusions Geologic unit, well water, community, weather, and unconventional natural gas development were associated with indoor radon concentrations. Future studies should include direct environmental measurement of radon, as well as building features unavailable for this analysis. Citation Casey JA, Ogburn EL, Rasmussen SG, Irving JK, Pollak J, Locke PA, Schwartz BS. 2015. Predictors of indoor radon concentrations in Pennsylvania, 1989–2013. Environ Health Perspect 123:1130–1137; http://dx.doi.org/10.1289/ehp.1409014

Incorporating Contact Network Structure in Cluster Randomized Trials.

Whenever possible, the efficacy of a new treatment is investigated by randomly assigning some individuals to a treatment and others to control, and comparing the outcomes between the two groups. Often, when the treatment aims to slow an infectious disease, clusters of individuals are assigned to each treatment arm. The structure of interactions within and between clusters can reduce the power of the trial, i.e. the probability of correctly detecting a real treatment effect. We investigate the relationships among power, within-cluster structure, cross-contamination via between-cluster mixing, and infectivity by simulating an infectious process on a collection of clusters. We demonstrate that compared to simulation-based methods, current formula-based power calculations may be conservative for low levels of between-cluster mixing, but failing to account for moderate or high amounts can result in severely underpowered studies. Power also depends on within-cluster network structure for certain kinds of infectious spreading. Infections that spread opportunistically through highly connected individuals have unpredictable infectious breakouts, making it harder to distinguish between random variation and real treatment effects. Our approach can be used before conducting a trial to assess power using network information, and we demonstrate how empirical data can inform the extent of between-cluster mixing.