|
|
|
Animal studies have shown effects of dioxins on sexual development, hormonal levels, and developmental malformations. However, the relationship between dioxin exposure and human sexual development has not been well studied. An ongoing collaborative investigation between HSPH scientists and a Russian team in Chapaevsk, where there is evidence of environmental contamination with dioxin, is evaluating the prevalence of developmental abnormalities and their association with dioxin exposure among a cohort of over 2500 males aged 11-16. Physical examinations were conducted on all subjects and cases of cryptorchidism, hypospadias, and delayed sexual maturation based on Tanner staging and other physical measurements were identified. A case-cohort design was used to identify controls matched by frequency within age groups. Blood samples are currently being collected on both the cases and controls, along with the mothers. Long-term goals of this project include following a subset of cases and controls prospectively to assess later measures of sexual development, such as sperm quality. Paige Williams is the statistician on this study; other HSPH collaborators include Russ Hauser (P.I.), Susan Korrick, Mary Lee, Larisa Altshul, and Jennifer Adibi.
As part of an ongoing investigation, researchers at the HSPH have studied boilermakers and utility workers exposed to fuel oil ash during overhauls of large oil-fired boilers at an electricity generating plant. To evaluate the relationship between exposure to fuel oil ash and respiratory effects, the workers self-recorded peak expiratory flow (PEF) measurements five times per day. PM10 exposures to fuel oil ash were based on personal sampling devices; however, not all workers wore the sampling devices every day. In fact, less than 25% of the person-days at work typically have these ideal exposure measurements. One project developed an approach for estimating exposures for workers who did not wear the personal PM10 monitors based on their work history, work tasks, exposure values for nearby workers, and area PM10 monitors. Another project which is still ongoing has evaluated methods for measurement error that could be applied to adjust for the estimated exposures. Yet another project currently ongoing is investigating the effects of PM10 exposure on cardiac outcomes, such as variability in heart rate. The latter project collects measurements on heart rate and heart rate variability at 5 minute intervals, yielding high-dimensional data on each worker. Statistical methods for accounting for repeated measures and time lags in effects have been developed and are being applied for this data. Paige Williams has been the statistician on these projects, and Sally Thurston and Donna Spiegelman have also played a lead statistical role in the measurement error project. Other HSPH investigators include Russ Hauser, David Christiani, Tom Smith, Mark Woodin, Shannon Magari, and Youcheng Liu.
Ethylene oxide (EtO) is a gas produced in large quantities and used primarily as a chemical intermediate in the production of ethylene glycol and other chemicals. A small percentage is also used as a fumigant for sterilizing medical and dental equipment, and foods, such as spices and nuts. It is well established that ethylene oxide can induce cancer, along with genetic, reproductive, developmental, and acute health effects. The US EPA has developed a cancer potency factor for EtO, but has not developed a reference concentration (RfC) for non-cancer effects. A project has been underway during the last year to conduct a state-of-the-art non-cancer risk assessment for EtO, relying on the rich data base of reproductive and developmental toxicity studies. Traditionally, the EPA and other regulatory agencies have estimated a RfC based on dividing a no adverse effect level (NOAEL) or benchmark dose (BMD) by a series of uncertainty factors, which are used to account for extrapolation from animals to humans, account for differences in sensitivity in humans, and make other conservative adjustments. Factors of 10 are routinely used for each of the adjustment factors. These adjustment factors have been criticized as combining the issues of quantitative risk assessment and the protectionism of risk management in a way that does not allow estimation of the degree of safety provided by any specific RfC. Instead, our approach relies on using sophisticated statistical methods to estimate the effective dose (ED) and its associated distribution, and data-derived distributions for the uncertainty factors used to adjust for animal to human extrapolation and differences in human sensitivity. These results will be submitted to the EPA and the Ethylene Oxide Industry Consortium (EOIC) for use in determining an RfC. Paige Williams has been the statistician on this project, working in collaboration with Sandra Baird, John Evans, Lorenz Rhomberg, and Andrew Wilson.
Risk of lung cancer varies among individuals of similar ages and smoking histories. One reason for this observed heterogeneity is differences in individual genetic polymorphisms. We are examining the relationship between lung cancer risk, and polymorphisms in genes controlling activation (Phase I) and detoxification (Phase II) of tobacco carcinogens, as well as in genes associated with DNA repair. We allow for gene-smoking interactions, and results to date suggest that the effects of several genetic polymorphisms on cancer risk are modified by smoking history. David Christiani is the principal investigator, Sally Thurston is the main statistician, and other HSPH investigators include Geoff Liu, David Miller, Li Su, and Zhou Wei.
The observed relationship between in utero exposure to polychlorinated biphenyls (PCBs) and later cognitive development depends, in part, on how PCB exposure is measured. In a cohort of children living adjacent to a PCB-contaminated harbor in New Bedford, Massachusetts, in utero PCB exposure was measured both as the concentration in the umbilical cord serum of the newborn, and for a subset, as the concentration on a per fat basis in the mother's breast milk. PCBs readily cross the placenta so that maternal serum and cord serum levels correlate. However, because PCBs are lipophilic, their concentrations in cord blood (which has a lower lipid content than adult blood) tend to be low and are therefore more difficult to measure than in relatively lipid rich matrices such as breast milk. Thus, the PCB concentration in a mother's breast milk is usually considered a more accurate measure of in utero exposure than cord blood PCBs. Using serum PCB instead of milk PCB for mothers who do not breastfeed can introduce measurement error bias. We take a Bayesian approach to correct for this bias in separately considering two outcomes: the Fagan test of infant cognitive function given at age 6 months, and the multiple outcomes of the Wide Range Assessment of Memory and Learning test given at ages 5-6 years. For the latter analyses we use a mixed effects model to allow correlation between outcomes within person, while allowing the effect of PCB exposure to differ across outcomes. Susan Korrick is the project principal investigator on this project, and Sally Thurston is the statistician.
ARDS is a devastating respiratory condition with a mortality rate of 40-50%. Major risk factors have been identified, and much current research has focused on the role of inflammatory cells and the mediators they produce in lung injury. However, genetic predisposition in humans has not been investigated adequately. It is likely that polymorphisms in several genes are important in mediating the inflammatory response, and are thus in themselves important predisposing factors for ARDS in humans. The purpose of this ongoing project is to evaluate whether there is an increased risk of ARDS in patients with a heritable polymorphism in several genes responsible for inflammatory cytokine expression and surfactant production. This objective will be addressed via a case-control study involving patients with ARDS and comparable patients without ARDS to compare the rates of occurrence of certain genetic polymorphisms. The role of polymorphisms in the prognosis and clinical outcome among ARDS cases will also be examined. Some of the genetic polymorphisms to be considered include TNF-alpha, TNFB, IL-1ra, and IL-10. Paige Williams is the statistician on this study, collaborating with David Christiani (P.I.) and Michelle Gong.
In occupational epidemiology it is typically assumed that the relationship between exposure, possibly transformed, and the risk of an outcome is linear in the parameters. Alternatively, exposure is catgorized into quantiles, although risk estimates are sensitive to the arbitrary choice of cut-points. Penalized splines in the Cox proportional hazards model may provide a more robust method for examining the linearity assumption. We have applied this approach to the study of prostate and brain cancer mortality in a cohort of autoworkers exposed to several metalworking fluids. The shape of the dose-response curve for soluble metalworking fluids and each cancer was approximately piecewise linear, with a small increase in risk at low exposures, and a larger increase at exposures above a critical point. Penalized splines will also be used to examine the relationship between silica exposure and lung cancer mortality, while allowing different curves for each form of silica. A related issue we are addressing is how best to summarize annual exposure measurements into a single relevant metric. Investigators on this project include Ellen Eisen (principal investigator), Joel Schwartz, and Sally Thurston (lead statistician).
The banning of DDT in the United States in 1972 heightened public concern regarding the abundant use of pesticides and their adverse health effects, especially among those who are chronically exposed. There are many different classes of pesticides. Although no longer widely used, organochlorides such as DDT continue to pose serious risk to human and animal populations because of their tendency to accumulate in fatty tissue. Organophosphate pesticides are still produced in large quantities worldwide and are widely used in agriculture. Although thought to be safer than the older style organophosphates, many scientists believe that these compounds also pose serious health risks. Effects on reproduction and development are of particular concern for pesticides due to their structural similarity to many natural hormones such as estrogens. There are several projects currently underway at the School, addressing the effects of pesticide exposure on reproductive health. One project (Principal Investigator, Russ Hauser) is examining the association between sperm quality and PCE exposure among men seen at the infertility clinic at Massachusetts General Hospital. The Program in Population Genetics (Director, Xiping Xu) has a number of projects underway which involve prospectively following women until conception. Louise Ryan is the statistician on these reproductive studies.
In 1995, the Government of the Russian Federation initiated an Environmental Management Project with a general aim to improve environmental conditions and associated human health (Jaakkola et al. 2000). An Environmental Epidemiology Component of the Russian Environmental Management Project was initiated to perform policy-oriented environmental epidemiology and risk assessment, and to establish systems for information transfer to decision-makers in environmental policy. Harvard School of Public Health, representing a broad group of international investigators, was chosen as a consultant to the Environmental Epidemiology Component. As part of the scientific program, cross-sectional and panel studies of the relations between exposure to urban air pollution, specifically particles, SO2 and NO2, and the occurrence of asthma, respiratory symptoms and infection, and allergic diseases in primary school children were conducted. Helen Parise is the statistician representing the Harvard School of Public Health on these studies.
 |
Copyright 2008, President and Fellows of Harvard College |
