Selected Research Projects

Air Pollution: Short-Term Effects

Particulate air pollution and daily mortality in Montreal, Quebec, 1984-1993

Granting agency: Health Effects Institute, Cambridge, MA; Toxic Substances Research Initiative, Health Canada, Canadian Institutes of Health Research.

PI: Mark Goldberg

Collaborators: JC Bailar III, Robyn Tamblyn, Richard Burnett, Jeffrey Brook, Yvette Bonvalot, Hélène Riberdy.

Abstract: This submission is in response to the Health Effects Institute’s RFA94-2, Objective 1, that requested applications to develop epidemiologic studies that would identify populations at higher than average risk from dying because of ambient exposure to particulates. The present proposal is designed to investigate cause-specific mortality in relation to sub-groups of the general population with varying levels of preexisting medical conditions. Thus, the principal objectives of the present investigation are 1) to determine whether concentrations of particulates in the ambient air of Montreal, Quebec were associated with daily all-cause and cause-specific mortality in the period 1984 to 1993, inclusive and 2) to determine whether there are sub-groups of the population (individuals with specific health conditions prior to death or those who were hospitalized at the time of death) with higher than average risks from particulates.

This study will cover the Island of Montreal, Quebec (hereafter referred to as Montreal) and will use 10 years of mortality, air pollution, meteorologic, and medicare data. (The population of Montreal is 1,775,871 persons, in an area of 493.5 km²). The ambient air of Montreal is relatively clean (mean PM10 29.5 μg/m3, May to October, 1984-88), so we will be investigating an exposure range that is low, but no so low as to miss effects predicted by extrapolations from other models. The environmental monitoring data from the network of stations in Montreal include particulate pollution (total suspended particulate (TSP), coefficient of haze (COH), PM₁₀, PM_2.5 (and sulfate and nitrate fractions of PM_x)), as well as measurements of NO₂, CO, CO₂, SO₂, and ozone. COH was measured every day but TSP and PM_x were obtained every sixth day, except in the summer months starting in 1992. Thus, we will use imputation algorithms to fill in missing data, but we will also be able to validate these algorithms using daily measurements. The study design depends heavily on linking records in a series of administrative files. Deceased subjects will be identified by merging mortality and medicare data files. Cause of death and other key variables will be obtained through this process. We will examine mortality in potentially susceptible subgroups by developing indices of morbidity based on medical services (all ages) and on drug use in the elderly (>65 years old) and in persons receiving welfare benefits. These indices will be validated within the study by determining their ability to predict mortality. Because of the universal coverage, broad scope, and detail of the Canadian medicare system, we are in an unique position to investigate whether there are differences between susceptible subgroups in risks associated with airborne particulates. The cost of the study is also minimized because investigators do not receive any remuneration as their salaries are paid by their affiliated institutions.

This study will use 10 years of mortality, air pollution, and meteorologic data (1984-93). The ambient air of Montreal is relatively clean, so we will be investigating the low end of the exposure range. Readily available mortality records and data from the network of environmental monitoring stations in Montreal will be used to investigate associations with daily cause-specific mortality and particulates. The environmental monitoring data includes particulate pollution (total suspended particulate (TSP), coefficient of haze (COH), PM₁₀, PM_2.5 (and sulfate and nitrate fractions of PM_x)), as well as measurements of NO₂, CO, CO₂, SO₂, and ozone. Deceased subjects will be identified from administrative mortality and medicare files. Mortality by geographic residence at time of death will be investigated. Moreover, we will create potentially susceptible subgroups by developing indices of morbidity based on medical services (all ages), and drug use in the elderly (>65 years old) and in persons receiving welfare benefits. These indices will be validated within the study by determining their ability to predict mortality. Because of the universality of the Canadian medicare system, we are in an unique position to define susceptible subgroups accurately and at low cost and to determine whether there are differences between groups in the risk of daily mortality associated with airborne particulates.

Publications:

Goldberg MS. Particulate air pollution and daily mortality. Who is at risk? J Aerosol Medicine, 1996;9:43-53.

Goldberg MS, Bailar JC III, Burnett R, Brook J, Tamblyn R, Bonvalot Y, Ernst P, Flegel KM, Singh R, Valois M-F. Identifying subgroups of the general population that may be susceptible to short-term increases in particulate air pollution: A time series study in Montreal, Quebec. Health Effects Institute, Cambridge, MA, July 2000 (see http://http://www.healtheffects.org/).

Goldberg MS, Burnett R, Brook J, Tamblyn R, Bonvalot Y, Singh R, Valois M-F, Bailar III JC. Cause-specific mortality and exposure to particulates in Montreal: A time series study. Epidemiology 1999;10(Number 4 Supplement): S47 (Abstract).

Goldberg MS, Burnett R, Brook J, Tamblyn R, Flegel K, Ernst P, Bonvalot Y, Singh R, Valois M-F, Bailar III JC. The Montreal Air Particulate Daily Mortality Study: Identifying susceptible sub-groups using medicare data. Epidemiology 1999;10(Number 4 Supplement): S94 (Abstract).

Goldberg MS, Burnett R, Bailar JC III, Brook J, Bonvalot Y, Tamblyn R, Singh R, Valois M-F. The association between daily mortality and short-term effects of ambient air particle pollution in Montreal, Quebec: 1. Nonaccidental mortality. Environ Res A86:12-25, 2001.

Goldberg MS, Burnett R, Bailar JC III, Brook J, Bonvalot Y, Tamblyn R, Singh R, Valois M-F, Vincent R. The association between daily mortality and short-term effects of ambient air particle pollution in Montreal, Quebec: 2. Cause-specific mortality. Environ Res A86: 26-36, 2001.

Goldberg MS, Burnett R, Bailar JC III, Tamblyn R, Ernst P, Flegel K, Brook J, Bonvalot Y, Singh R, Valois M-F, Vincent R. The identification of persons with cardio-respiratory conditions who are at risk of dying from the acute effects of ambient air particles. Env Health Perspect 109(Suppl 4):487-494, 2001.

Mark S. Goldberg, Richard T. Burnett, Jeffrey Brook, John C. Bailar III, Marie-France Valois, Renaud Vincent. Associations between Daily Cause-specific Mortality and Concentrations of Ground-level Ozone in Montreal, Quebec. Am J Epidemiol 2001;154:817-826.

Goldberg MS, Burnett RT, Valois M-F*, Flegel K, Bailar III JC, Brook J, Vincent R, Radon K. Associations between Ambient Air Pollution and Daily Mortality among Persons with Congestive Heart Failure. Environmental Research, In press.

Kolb S*, Radon K, Goldberg MS. The short-term influence of weather on daily mortality in congestive heart failure. Arch Environ and Occup Health 2007;62(4):169-76.

Goldberg MS, Burnett RT, Yale J-F, Valois M-F, Brook, J. Associations between Ambient Air Pollution and Daily Mortality among Persons with Diabetes. Environmental Research, 2006; 100:255–267.2. Assessing the Population Health Impact of Short Term Exposure to Urban Air Pollution Using Administrative Health and Environmental Information

A longitudinal study to identify factors that trigger ventricular tachyarrhythmias among persons with implanted cardioverter defibrillators

Granting agency: CIHR

PI: M S Goldberg and Paul Khairy

Collaborators: J. BROPHY, R. BURNETT, V. ESSEBAG, T. HADJIS, N. MAYO, S. MILLER, M. TALAJIC

Sudden cardiac death from heart rhythm disorders (arrhythmias) is one of the leading causes of death in Canada. Although difficult to estimate accurately, a new Canadian study estimated that the incidence rates for cardiac arrest were in the order of 55 per 100,000 and rates appeared to be declining since 1995. To place this in perspective, 69,500 deaths from cancer occurred in Canada in 2005. Ventricular arrhythmias (fast heart rhythms that arise entirely within the lower chambers of the heart (the ventricles), including tachycardia and fibrillation, are the most common initial rhythm documented by emergency response personnel. The need to prevent sudden cardiac death is reflected in the large number of people who are eligible to receive an implantable cardioverter defibrillators (ICD), which is about 92,000 at any one time. Clinical care guidelines for implantation of ICDs restrict the population to those individuals who have serious diagnosed cardiac conditions and who are at high risk of developing a fatal ventricular tachyarrhythmia. This remains a very large population (in the hundreds of thousands in Canada). Sudden cardiac death from ventricular tachyarrhythmias is a serious public health problem.

Unfortunately, there is scant information regarding the association between triggering factors and the frequency of tachyarrhythmias in ICD patients, with a handful of studies suggesting that increases in air pollution and changing weather conditions (e.g., high temperatures) may affect event rates. It is important to identify transient factors, especially those that are modifiable, so that potentially fatal arrhythmias and mortality can be reduced, especially amongst those who are being treated aggressively. Widespread use of ICDs provide an opportunity for investigating these triggering factors, and is the only way that arrhythmias can be investigated. Thus, the primary purpose of this study is to identify personal, pharmaceutical, and environmental factors that may increase the occurrence of from ICDs to treat ventricular tachycardia, including pacing, as well triggering factors for “inappropriate treatment”.

This is a longitudinal, prospective, inception cohort study of about 500 persons, age 50-75 years, living in Montreal with newly implanted “third-generation” dual-chamber, cardioverter defibrillators and followed at the Montreal Cardiology Institute and at the McGill University Health Center during 2007 to 2009 and subjects will be followed for three years. The primary objective of this study is to identify personal, clinical, pharmaceutical, and environmental factors that may increase the occurrence of “appropriate treatment” for ventricular tachyarrhythmias. The study population covers about 65% of the total implanted population in Montreal and comprises individuals with serious heart conditions who have either had a near-death experience or who are at very high risk of sudden death. They are already followed very closely, routinely coming to the clinic every six months and upon receipt of a shock from the ICD. Compliance to this clinical protocol in both our centres is close to 100%. Our clinical experience in both of our centres indicates that this population is motivated, receptive, and compliant.

We will record data at baseline and a questionnaire will be self-administered periodically and at the time of a perceived therapeutic intervention (shock) to describe potential personal triggering factors (e.g., physical activity, psychological stress, physical activity). Subjects will be trained as to how to fill in the questionnaires and follow the protocol. We will institute a series of active procedures to follow subjects closely through time, thereby maximizing response rates. We will also obtain data on medical procedures, drugs, hospitalization, mortality, etc. from provincial medicare files and environmental data (air pollution, weather) from fixed-site monitors.

Our principal health endpoint will be ventricular tachyarrhythmias, which is a recurrent, non-absorbing binary outcome. Analyses will be based on case-crossover as well as random effects binary logistic models. The case-crossover design is analogous to a matched case-control study, in which exposure during a case-period is compared with exposure during one or more control periods sampled from that same individual at different time points. Thus, within-subject comparisons are made across time and then “averaged” across subjects. We will thus be able to investigate a variety of personal and environmental triggers. This study will therefore provide valuable data on aetiology, it will provide insights into mechanisms through a longitudinal follow-up of a cohort of high risk subjects, and it may provide data to suggest strategies to reduce these potentially fatal events and, consequently, to reduce sudden cardiac death.

longitudinal study of Acute effects OF ambient air pollution in susceptible populations

Granting agency: CIHR

PI: M S Goldberg

Collaborators: Richard Burnett, Kenneth Flegel, Nadia Giannetti, Michal Abrahamowicz, Robert Platt

Abstract: Despite the general reduction in pollution levels, short-term exposures to ambient air pollutants generated from the combustion of fossil fuels have been found to be associated with increased hospitalizations and deaths. The evidence derives mostly from parallel time series studies, which are used to answer the question “are there increases in numbers of hospitalizations or deaths on the day or the next few days following an increase in the level of air pollution?”. Studies have been carried out in countries with varied levels of industrialization and air pollution, and the results have been extremely consistent regardless of the extent of pollution or the mix of pollutants. It has been hypothesized that only persons in poor health should be affected by the relatively low levels of air pollution present today. The hypotheses focus on persons with cardiopulmonary disease and those in poor and failing health who because of their illness are susceptible to the adverse effects of air pollution. The identification of mechanisms is essential and, although the toxicology is progressing, epidemiological studies are required to uncover which subgroups of the population are susceptible to the effects of air pollution. This protocol addresses this issue through a longitudinal study of selected subgroups to identify whether the incidence of certain health outcomes are correlated with short-term changes in air pollution.

The principal objective of this longitudinal study is to determine among multiple cohorts of persons with specific health conditions whether short-term changes in air pollution, on the order of days, are associated with the incidence of critical indicators of health, including death, hospitalizations, emergency room visits, and selected “intermediate health events”. Specific objectives are: 1) for selected conditions (e.g., congestive heart failure), to define intermediate health events (e.g., changes in medication signaling an adverse health event) and develop algorithms to identify these from administrative health data and 2) to conduct a series of statistical analyses using a Poisson point process model developed by Dewanji and Moolgakavkar that will provide estimates of associations between short-term changes in air pollution and the above-mentioned indicators of health.

The study is designed as a longitudinal cohort of men and women who are residents of Montreal and are age 65 years and above during the 12 year period 1991-2002. Identification of subjects will be from the files of the Quebec medicare system (RAMQ). Subcohorts of individuals having specific health conditions who have been reported in the literature to be at higher than expected risk will be selected and followed in time for adverse events. The health conditions to be studied are: chronic obstructive pulmonary diseases, asthma, congestive heart failure, hypertension, chronic coronary artery disease, acute myocardial infarction, and diabetes. In addition, we will define a “control population” that has no major chronic health conditions at the beginning of follow-up. The data sets to be used in this study are: daily weather and air pollution data for Montreal; RAMQ billing and prescription files for each subject; and Quebec mortality and hospital discharge data. The new Poisson point process model will be used to estimate associations between these endpoints and daily ambient air pollution, accounting for weather and personal characteristics of subjects.

Publications:

Goldberg MS, Burnett RT. A new longitudinal design for identifying subgroups of the population who are susceptible to the short-term effects of ambient air pollution. J Toxicol. Environ Health A 2005;68(13-14), 1111-1125.

The short-term effects of ambient air pollution on clinical signs and symptoms for congestive heart failure: A pilot study

Granting agency: Canadian Institutes of Health Research

PI: Mark Goldberg

Collaborators: BURNETT, Richard; FLEGEL, Kenneth; GIANNETTI, Nadia; MAYO, Nancy E; TAMBLYN, Robyn; VINCENT, Renaud

Abstract: Although ambient levels of pollution have declined in the past 40 years, recent studies have shown that short-term exposures to ambient air pollutants generated from the combustion of fossil fuels cause exacerbations of cardio-respiratory disease that result in hospitalization and death. An important unanswered question is "which subgroups of the population may be susceptible to the effects of ambient air pollution?". One of the most susceptible subgroups may be individuals with congestive heart failure (CHF): much higher rates of hospitalization and death are found in this group when air pollution increases. There is no information regarding the time course of changes in health status of these individuals nor how these changes are linked to variations in ambient air pollution. We thus seek to determine whether changes in key indicators of health status in subjects with CHF are associated with variations in levels of air pollution. Because of the complexities involved in such a study, we need to first test the feasibility of various aspects of conducting such a study as well as to pilot various aspects of the study.

Publications:

Goldberg MS, Giannetti N, Burnett RT, Mayo NE, Valois M-F, Brophy JT. A Panel Study in Congestive Heart Failure to Estimate the Short-term Effects from Personal Factors and Environmental Conditions on Oxygen Saturation and Pulse Rate, Occupational and Environmental Medicine. 2008;65:659-6.

Goldberg MS, Giannetti, N, Burnett RT, Mayo NE, Valois M-F, Brophy JM. Shortness of Breath at Night and Health Status in Congestive Heart Failure: Effects of Environmental Conditions and Health-Related and Dietary Factors. In press, Environmental Research.

A panel study of the effects of changing lifestyle and environmental conditions on signs and symptoms in heart failure

Granting agency: Health Canada (contract)

PI: Mark Goldberg

Collaborators: Amanda Wheeler, Nadia Giannetti, Eileen O'Meara, Nancy E. Mayo, James Brophy, Richard T. Burnett

Heart failure (HF) is a complex clinical syndrome resulting from any cardiac condition that impairs the ability of the heart to pump blood from the ventricles. It causes serious fatigue, dyspnea on exertion and then at rest and, thus, reduces the ability to conduct activities of daily living. It is a chronic and progressive disease resulting in frequent visits to outpatient clinics and emergency departments, frequent admissions to hospital, and is often fatal. Incidence rates are very high and increase by age: for example, in the US, for ages 65-74, 75-84, and >85 are 15.2 per 1,000, 31.7, and 65.2, respectively. (To appreciate the magnitude of this problem, the incidence rate in Canada in 2005 for lung cancer for ages 65-74 years was approximately 1.96 per 1,000.) In Canada, prevalence increases monotonically with age. In the US, at age 40, the lifetime risk of developing HF for both men and women is 20%, and without a previous myocardial infarction the prevalence in men is 11% and in women it is 17%. About 86,000 persons were hospitalized in 2000-1 in Canada for HF, and there were readmissions for 33% of these people. Five-year survival rates are about 28%; less severe cases have 5-year survival rates from 40-50%. About 5,000 deaths are attributed to HF (rate of ~250/100,000) and this corresponds with about 2% of all deaths in Canada. The annual health care cost in the US is approximately $20 billion.

The purpose of the present study is to determine among patients with HF whether daily changes in selected signs, symptoms, and physiologic indicators are associated with daily changes in selected personal characteristics/lifestyle and with daily variations in environmental conditions. This is a study of acute effects. A motivation for the study is the lack of identified modifiable personal risk factors for decompensated HF and little understanding of the causal pathways from sub-clinical to adverse states of health. A second motivation derives from findings of environmental studies of acute effects that showed that daily hospitalizations and daily mortality among persons with HF increase when ambient temperature and when levels of ambient air pollution increases. Identification of modifiable factors can be used in clinical practice to improve outcomes and these data will also assist in more fully understanding mechanisms by which patients experience decompensated HF. Through the support of CIHR (MOP-49397), Goldberg and colleagues conducted a feasibility study to determine whether potential subjects i) would agree to participate in a study, ii) whether participants and non-participants would be comparable, and iii) whether participating subjects would be able to record data on a daily basis. We have concluded that the main study is feasible: the participation rate was 86%; subjects were able to complete the study protocol (e.g., only 3.2% of data were missing); and we have data to improve the design (e.g., the daily diary) and estimate sample size.

We propose a daily diary panel study and 200 subjects will participate over a two month, consecutive period. They will comprise a consecutive series of outpatients who receive health care at HF clinics in two major teaching hospitals in Montreal. At baseline, we will measure characteristics of subjects, take blood, and measure selected physiological indicators. In addition, we will visit the home to conduct a household inventory and install indoor/outdoor devices to continuously measure and record temperature, relative humidity, and concentrations of fine particles (PM2.5), NOx, and CO. Every day during the observation period, before breakfast, participants will record symptoms, physiologic measures, and potential confounding variables. During the observation period, we will obtain daily measurements of selected signs, symptoms and physiologic indicators (recorded onto a scannable, daily diary) and daily environmental conditions. We will ask them to wear a GPS device when they leave the home and these data will be used to infer exposures to ambient air pollution that will be estimated using land use regression methods. Daily fluctuations in these measures of signs, symptoms, and physiologic measures will then be linked to daily measurements of personal and environmental variables using mixed (random effects) regression models.

Air Pollution: Long-Term Effects

Reanalysis of the Harvard Six-Cities Study and the American Cancer Society Study of Air Pollution and Mortality

Granting agency: Health Effects Institute, Cambridge, MA.

PI: Daniel Krewski (University of Ottawa)

Collaborators: Richard Burnett, Mark Goldberg, Jack Siemiatycki.

Abstract: The overall objective of the project is to conduct a rigorous and independent assessment of the findings of the Six-Cities and ACS studies of air pollution and mortality. The project has two main tasks: 1) Attempt to validate and replicate the published results by: conducting a quality assurance audit on a sample of the original data; attempting to reproduce the original numerical results. 2) Conduct sensitivity analyses to test the robustness of the original findings and interpretations to alternative analytic approaches.

Publications:

Krewski D, Burnett R, Goldberg MS, Siemiatycki J, et al.. Re-analysis of the Harvard Six-cities and American Cancer Society Cohort Studies of Particulate Air Pollution and Mortality, Phase II: Sensitivity Analysis. Health Effects Institute, Boston, MA, August 2000 (see http://www.healtheffects.org/).

Richard Burnett, Renjun Ma, Michael Jerrett, Mark S. Goldberg, Sabit Cakmak, C. Arden Pope III, Daniel Krewski. The Spatial Association Between Community Air Pollution and Mortality: A New Method of Analyzing Correlated Geographical Cohort Data. Env Health Perspect 109(Suppl 3):375-80, 2001.

Richard Burnett, Renjun Ma, Michael Jerrett, Mark S. Goldberg, Sabit Cakmak, C. Arden Pope III, Daniel Krewski. The Space-time Association Between Community Air Pollution and Mortality: A New Method of Analyzing Correlated Geographical Cohort Data. Statistical Society of Canada, 28^th Annual Meeting June 2000, Proceedings of the Survey Methods Section 1375-80, 2001.

Paul J. Villeneuve, Dan Krewski, Richard T. Burnett, Mark S. Goldberg, Yue Chen, Jack Siemiatycki. Fine particulate air pollution and all-cause mortality within the Harvard Six-Cities study: variations in risk by period of exposure. Annals of Epidemiol, 2002;12(8):568-576.

Krewski, D; Burnett, R; Goldberg, M; Hoover, K; Siemiatycki, J; Abrahamowicz, M; White, W. Reanalysis of Harvard Six Cities Study, Part I. Validation and Replication. Inhalation Toxicology, 2005;17(7):343-352

Krewski, D; Burnett, R; Goldberg, M; Hoover, K; Siemiatycki, J; Abrahamowicz, M; Villeneuve, P; White, W. Reanalysis of Harvard Six Cities Study, Part II. Sensitivity Analysis. Inhalation Toxicology, 2005;17(7):335-42

Sabit Cakmak, Richard T. Burnett, Michael Jerrett, Mark S. Goldberg, C. Arden Pope III, Renjun Ma, Timur Gultekin, Michael J. Thun, Daniel Krewski. Spatial Regression Models for Large Cohort Studies Linking Community Air Pollution and Health. Journal of Toxicology and Environmental Health, 2003;66 (Numbers 16-19), 1811-1824.

Jerrett M, Burnett RT, Goldberg MS, Sears M, Krewski D, Catalan R et al. Spatial analysis for environmental health research: concepts, methods, and examples. J Toxicol Environ Health A 2003; 66(16-19):1783-1810.

Jerrett, M., R. Burnett, A. Willis, D. Krewski, M.S. Goldberg, P. DeLuca, N. Finkelstein. Spatial analysis of the air pollution-mortality relationship in the context of ecologic confounders. Journal of Toxicology and Environmental Health, 66 (Numbers 16-19), 1735-1778, 2003.

Daniel Krewski, Richard T. Burnett, Mark S. Goldberg, Michal Abrahamowicz, Jack Siemiatycki, Michael Jerrett, B. Kristen Hoover. Rejoinder: Reanalysis of the Harvard Six Cities Study and American Cancer Society Study of Particulate Air Pollution and Mortality. The Journal of Toxicology and Environmental Health, 66 (Numbers 16-19), 1715-1722, 2003.

Jack Siemiatycki, Daniel Krewski, Mark Goldberg, Yuanli Shi, Louise Nadon, Ramzan Lakhani. Control of Occupational Confounding in the Harvard Six Cities Study and American Cancer Society Study. The Journal of Toxicology and Environmental Health, 66 (Numbers 16-19), 1591-1604, 2003.

Alette Willis, Daniel Krewski, Michael Jerrett, Mark Goldberg, RT Burnett. Selection of Ecologic Covariates in the American Cancer Society Study. The Journal of Toxicology and Environmental Health, 66 (Numbers 16-19), 1563-1590, 2003.

B. Kristin Hoover, Donna E. Foliart, Warren H. White, Aaron J. Cohen, Linda J. Calisti, Daniel Krewski, Mark S. Goldberg. Retrospective Data Quality Audits of the Harvard Six Cities and American Cancer Society Studies. The Journal of Toxicology and Environmental Health, 66 (Numbers 16-19), 1553-1562, 2003.

Krewski D, Burnett RT, Goldberg MS, Hoover K, Siemiatycki J, Jerrett M, Abrahamowicz M, Warren H. Overview of the reanalysis of the Harvard six cities study and American Cancer Society study of particulate air pollution and mortality. The Journal of Toxicology and Environmental Health, 66 (Numbers 16-19), 1507-52, 2003.

Cancer incidence and mortality among adults in relation to long-term exposure to outdoor air pollution: A general population cohort study of Ontario residents

Granting agency: CIHR

PI: Paul Villeneuve

Collaborators: JR. BURNETT, M S Goldberg, R Ma

The purpose of this study is to determine whether long-term exposure to air pollution is associated with the occurrence of cancer and other causes of death in a cohort of adults who were resident in one of 10 urban centres in Ontario. These centres were selected on the basis of having widely different exposure characteristics and the availability of survey data on smoking prevalence, while providing sufficient statistical power to address the research objectives. Although there are now a number of cohort studies that have been used to answer this question, most have used average exposure in the city where a subject resided at the time of enrollment and did not account for mobility, which would lead to errors in assigning levels of exposure. Consequently, there may have been considerable misclassification of the relevant exposure metrics and risks may have been underestimated.

The cohort will be constructed by identifying individuals, 40 years of age and older, from the T1-Family File (T1FF) database. This database contains information from annual income tax T1 forms and includes all persons who filed Federal tax returns as well as any family members identified on the tax returns from 1982 onwards. To be included in the cohort, individuals had to be resident in one of 10 urban centres for at least one tax filing between 1982 and 1986. Internal linkage within the T1FF will allow for the residential mobility of these individuals to be tracked each year from 1982 until 2001. Mortality and cancer incidence within the cohort will be ascertained by linking personal identifying information in the T1FF database to the Canadian Cancer Registry and Vital Statistics Death databases.

Air pollution exposures will be assigned to each individual using data from fixed-site monitoring stations from the National Air Pollution Surveillance network. Annual air pollution indices will be created for O₃, SO₂, and total suspended particulates. An ?extended? Cox regression model will examine associations between air pollution and health outcomes, including: all cancers, lung cancer, and cardiovascular and respiratory mortality. The extent by which socioeconomic status modifies the air pollution risk estimates will be evaluated using reported income from the T1FF.

While individual-level income data are available for our cohort on an annual basis, information for other covariate information is missing (e.g., smoking). Not adjusting for key risk factors may lead to biased estimates if there are large discrepancies in prevalence between the different the cities included in the present study. Using data from the Ontario Health Survey, we will use an indirect method to account for smoking, alcohol consumption, and physical activity on the risk estimates for air pollution.

In summary, the large number of persons exposed to higher levels of air pollution in Ontario, the capability of assembling a general-population based cohort with individual level indicators of socioeconomic status, the ability to trace the residential mobility of this population, and the availability of retrospective air monitoring data, together, provide a unique opportunity to investigate the relationship between long-term exposure to air pollution and chronic health effects.

Publications:

Hong Chen*, Mark S. Goldberg, Paul J. Villeneuve. A Systematic Review of Relation between Long-term Exposure to Ambient Air Pollution and Chronic Diseases. Submitted to Reviews of Environmental Health, Rev Environ Health, 2008.

Traffic-related air pollution and socioeconomic gradients in the incidence of cancer

Granting agency: CIHR

PI: M S Goldberg

Collaborators: Michael Jerrett, Nancy Ross, Nicolas Gilbert, Jeffrey Brook, Renjun Ma, Richard Burnett, Claire Infante-Rivard, Jack Siemiatycki

Abstract: It has been found that short-term exposures to ambient air pollutants generated from the combustion of fossil fuels are associated with increased hospitalizations and deaths. It has been hypothesized that only persons in poor health should be affected by the relatively low levels of air pollution present today. The hypotheses focus on persons with cardiopulmonary disease and those in poor and failing health who because of their illness are susceptible to the adverse effects of air pollution. In addition, it has also been observed from cohort and case-control studies that there may be associations between long-term exposures to air pollution and lung cancer and childhood leukemia. Results from three cohort studies have indicated that the chronic effects may be mediated by socio-economic status; this implies that either local levels of air pollution in socio-economically disadvantaged areas are higher than implied by average city-wide levels or that individuals living in these areas are at higher risk, or both.

We are proposing to determine in Montreal whether 1) chronic health effects are associated with long-term effects of local levels of ambient air pollution, respectively, and are associated with indicators of socio-economic status and 2) whether socio-economic status modifies the effects of air pollution (and vice versa). (Montreal is a particularly interesting urban context to examine these types of relationships because it has a relatively unequal distribution of income and is one of the most economically segregated cities in Canada.) The chronic health effects comprise selected sites of cancer, and we will make use of four cancer case-control studies conducted by us in Montreal. The linkage with traffic-related air pollution is through the address of the residences of subjects in these data sets. For the case-control studies, we have detailed residential histories and we will estimate local levels of traffic-related air pollution (the major component affecting short-range variations) through a monitoring study of nitrogen dioxide. We will then make use of fixed-site air pollution monitoring stations, measurements of traffic density, and spatial interpolation techniques to estimate historical exposures. Enumeration area census data will provide us with a means to identify “natural neighbourhoods” that are indexed by contextual ecological variables. We will then juxtapose these data sets to obtain for selected sites of cancer adjusted estimates of relative risk by levels of traffic-related air pollution and socio-economic status and we will determine whether these factors modify each other.

Publications:

Dan Crouse*, Nancy Ross, Mark S. Goldberg. Neighbourhood-scale Socioeconomic and Environmental Risks in Montreal, Canada. Submitted.

Dan Crouse*, Mark S. Goldberg Nancy Ross, Michael Jerrett Modelling Temporal and Spatial Variability of Ambient Air Pollution in Montreal, Canada. Submitted.

Breast Cancer Research

Gene-environment interactions in postmenopausal breast cancer: A case-control study

Granting agency: CIHR

PI: M S Goldberg

Collaborators: Jacques Simard, Francine Durocher, France Labrèche, Marie-Élise Parent, Bryan Langholz

Abstract: Female breast cancer is a major public health concern in the industrialised world. It is the most commonly diagnosed malignancy among Canadian women, and it accounts for about 30 percent of all new cases of cancer. About one in nine Canadian women is expected to develop breast cancer in her lifetime and one in 25 women is expected to die from the disease. It is the second most common cause of death from cancer, accounting in 1999 for an estimated 18% of all cancer deaths among Canadian women.

We and others have postulated that exposures to certain chemical and physical agents may increase the risk of developing breast cancer, including pesticides, polycyclic aromatic hydrocarbons (PAH), organic solvents, and magnetic fields (EMF) and light-at-night. Common genetic polymorphisms of low penetrance such as those that code for the cytochrome P450 Phase I oxidizing enzymes (the family of CYP genes) and Phase II detoxifying or conjugating enzymes (e.g., glutathione s-transferase (GST) and n-acetyl transferase (NAT)) are only recently drawing researchers’ interest. Cytochrome P450 enzymes are important factors in steroidogenesis and in detoxifying environmental chemicals such as PAHs and benzo(a)pyrene. Research in gene-environment interactions in breast cancer is in its infancy, with only a handful of high quality studies that have been completed.

The objective of the present proposal is to determine whether common polymorphisms for selected genes (e.g., CYP1A1, GSTs, NATs, COMT, HRAS1) combined with lifestyle/demographic characteristics (e.g., alcohol, physical activity, reproductive factors, body size, hormonal replacement therapy, exposure to light-at-night and shift work) are associated with the incidence of postmenopausal breast cancer. A population-based case-control design will be used to enrol 1,800 incident, histologically-confirmed cases of postmenopausal breast cancer from all major hospitals in Montreal. An equal number of control subjects will be selected from among women with other health conditions (excluding cancer) admitted and treated in the same hospital as the cases, and they will be frequency-matched to cases in each hospital by “ethnicity” and age. Detailed interviews will be conducted to obtain information on all accepted and suspected risk factors and detailed lifetime job histories. Known DNA variants will be investigated in genomic DNA extracted from 30cc of peripheral blood lymphocytes from case and control women. Where polymorphisms are not available from the literature, they will be identified by sequencing a small number of individuals. Both coding and non-coding polymorphisms (which may be in linkage disequilibrium with the causative polymorphisms) will be sought. Because of the complexity of the environment and inter-individual variability in cancer susceptibility, we propose to integrate different metabolic pathways in our study. The variants will be selected on the basis of their known or putative variable function and their relatively high prevalence. They will be genotyped  using the most appropriate up-to-date genomic methods.

Breast cancer in women and exposure to occupational and environmental substances

Granting agency: Canadian Breast Cancer Research Initiative

PI: Mark Goldberg

Collaborators: France Labrèche, Jack Siemiatycki, Ben Armstrong, Bruce Case, Richard Margolese, Michel Gérin

Abstract: Although progress has been made in limiting the progression of female breast cancer through surgical and chemical adjuvant treatments, we are unfortunately left with the grim picture that breast cancer incidence continues to increase unabated. In Canada, breast cancer incidence rates have increased by 27.9% over the last 25 years. Increases in incidence have also been observed in other jurisdictions. Secular changes in accepted and suspected risk factors may explain some of the long-term increases, although these factors probably cannot account for a large proportion of cases. While important contributions have been made in identifying potential risk factors, new directions for research are clearly required. One area that has received virtually no attention is the occupational setting. With more and more women entering the workforce over the past 35 years, it is important to ascertain whether certain occupations or exposures increase the risk of developing breast cancer.

The case-control study that we describe herein is the first major investigation of invasive breast cancer and its relationship with occupation. Our underlying hypothesis is that there are chemical and physical agents present in the workplace that may induce breast cancer. In particular, we have developed an hypothesis that exposure to organic solvents is associated with the incidence of breast cancer. The study uses cutting-edge methodology to assess past occupational exposures. Coupled with this and other design features of the study, we will test the hypothesis that organic solvents are associated with postmenopausal female breast cancer. As well, we will assess risk for a wide range of occupational risk factors that have not been hitherto investigated. From a public health point of view, identification of new agents whose exposures are amenable to modification will allow new public health initiatives to reduce risk.

The present study is meant as a first step to investigating occupation and, to a lesser extent, the general environment. We hypothesize that there are chemical and physical agents present in the workplace that may induce breast cancer. In particular, we have developed an hypothesis that exposure to organic solvents is associated with the incidence of breast cancer. The present investigation is the first major study of invasive breast cancer and its relationship with occupation. The objective of the study is to determine whether exposure to organic solvents and exposure to selected chemical and physical agents are associated with the incidence of post-menopausal malignant breast cancer in women. Subjects will be incident, histologically-confirmed cases of breast cancer aged 51-70 years diagnosed in Montreal hospitals. Controls will be drawn from other histologically-confirmed sites of cancer. The study uses a powerful methodology for assessing occupational exposures: Probing interviews of subjects provide a detailed picture regarding the subject's occupational history. With this information, an experienced team of occupational hygienists then attribute exposure to a list of agents. This and other design features of the study will allow the assessment of a wide range of occupational risk factors that have not been hitherto investigated.

Publications:

Goldberg MS, Labrèche F. Occupational risk factors for female breast cancer: A review. Occupational and Environmental Medicine, 1996; 53:145-56.

Labrèche F, Goldberg MS. Exposure to organic solvents cause breast cancer in women: A hypothesis. Am J Indust Med, 1997; 32:1-14.

Lenz SK, Goldberg MS, Labrèche F, Parent M-É, Valois M-F. The association between alcohol consumption and postmenopausal breast cancer: Results of A case-control study in Montreal, Quebec, Canada. Cancer Causes and Control, 2002;13(8):701-710.

Labrèche F, Goldberg MS, Valois M-F, Nadon L, Richardson L, Lakhani R, Latreille B. Occupational Exposures to Extremely Low Frequency Magnetic Fields and Postmenopausal Breast Cancer. Am J Indust Med, 2003, 44:643-52.

Waiting time for breast cancer surgery: An analysis of Quebec, 1979-99

Granting agency: Canadian Breast Cancer Research Initiative

PI: Nancy E Mayo (McGill University)

Collaborators: Mark Goldberg, Richard Margolese, John Sampalis, Jeannie Haggerty, Jim Hanley

Abstract: The purpose of this study is to determine the amount of time that women in Quebec wait for treatment for breast cancer. This study will: 1) determine the time from screening mammography to diagnostic mammography, diagnostic mammography to surgical consultation, consultation to biopsy, biopsy to surgery and surgery to adjuvant treatment (if received); 2) determine whether waiting times are influenced by age, socioeconomic and geographic region, characteristics of the referring physician, characteristics of the surgeon, and type of surgery or treatment received; 3) determine whether waiting time has changed over recent times.

Publications:

Mayo N, Scott S, Shen NY, Hanley JA, Goldberg MS, MacDonald N. Waiting time for breast cancer surgery in Quebec. CMAJ 2001;164(8):1133-8.

Shen N, Mayo NE, Scott SC, Hanley JA, Goldberg MS, Abrahamowicz M, Tamblyn R. Factors associated with pattern of care before surgery for breast cancer in Quebec, 1992-1997. Medical Care, 2003;41(12):1353-66.

Fortin Bernard, Goldberg MS, Mayo NE, Valois M-F, Scott SC, Hanley J. Waiting Time for Radiation Therapy in Breast Cancer Patients in Quebec from 1992 to 1998. A study of surgically treated breast cancer patients in Quebec helps to explain increased waiting times between surgery and post-operative radiation therapy. Healthcare Policy 2006; 1(2): 142-157.

The Occurrence of Arm Dysfunction and Lymphedema After Treatment for Stage I or II Breast Cancer

Granting agency: CIHR, CBCRI

PI: M S Goldberg

Collaborators: Nancy Mayo, Sarkis Meterissian

Abstract: Breast carcinoma is the most frequently occurring cancer in women. Survival rates after breast cancer have been increasing and this is due partly to improved diagnostic and therapeutic techniques. One frequent source of morbidity post-surgery is lymphedema and arm dysfunction, which may entail weakness, stiffness, pain, edema, loss or change in sensation, decreased fine motor coordination, and restricted range of motion. These arm problems may be associated with axillary dissection, a procedure performed on the majority of women undergoing breast cancer surgery. Research on arm morbidity has focused on the problem of lymphedema, which is the result of a functional overload of the lymphatic system in which lymphatic volume exceeds transport capabilities. Chronic lymphedema (present for at least 3 months) has noticeable changes in skin colour and texture and can cause chronic arm dysfunction. Treatment includes elevation, massage and exercise, and application of external pressure with compression garments or compression pumps. These methods give a wide range of results, depending on the subject and timing of treatment, but may only lead to partial improvement. Lymphedema persisting over a long period results in hypertrophy of the subcutaneous fat tissue and development of fibrosis, both of which make compression therapy less effective. Thus, it is thought that early identification of patients with lymphedema is important since compression therapy seems to be more effective the shorter the time the edema has persisted.

The literature on lymphedema is rather sketchy and its interpretation is plagued by difficulties in its definition. There are very little data on arm dysfunction. The objective of the present study is to increase our limited understanding regarding the relationships between lymphedema, arm dysfunction, and quality of life among women treated for stage I or II breast cancer and to identify important risk factors for lymphedema and arm dysfunction. Specifically, we will 1) estimate the prevalence, extent, and impact on quality of life of lymphedema and arm dysfunction after treatment for stage I or II breast cancer: 2) define the association between the presence and degree of lymphedema and arm dysfunction; and 3) identify risk factors for the development of lymphedema and arm dysfunction post-treatment.

The study is designed as a follow-up of patients treated at the McGill University Hospital Centre (MUHC) between 1992 and 2002. We will make use of the MUHC Cancer Registry to identify patients, obtain relevant clinical data, and address information. We will trace subjects and they will fill out a “screening” questionnaire about signs and symptoms of lymphedema and a set of questionnaires on arm dysfunction and quality of life. Women who are positive on the screening questionnaire will be invited for a clinical evaluation as well as a random sample of persons who are negative (to estimate sensitivity and specificity). These data will allow us to answer the three objectives of the study. In identifying those patients at greatest risk of developing these complications, we may be able to determine which ones may benefit from sentinel lymph node mapping. This study may eventually make the long-term follow-up of breast cancer survivors routine and may lead eventually to randomized studies to determine optimal treatment regimens.

Publications:

Diana J. Dawes, Sarkis Meterissian, Mark Goldberg, Nancy E. Mayo. Impact of Lymphoedema on Arm Function and Health-Related Quality of Life in Women following Breast Cancer Surgery. J Rehabil Med, 2008; 40: 651–658.

Prostate Cancer Research

Occupational and lifestyle factors in the etiology of prostate cancer, and establishing a platform for studying susceptibility biomarkers

Granting agency: National Cancer Institute of Canada

PI: M-E Parent

Collaborators: J Siemiatycki, K Aronson, Mark Goldberg

Abstract: The present study will explore, using expert-based exposure assessment, the role of some 100 occupational substances which appear the most promising based on the evidence from studies of occupational groups or based on their suspected biological role. Moreover, we will assess the role of selected lifestyle factors which have received relatively little attention to date, i.e., physical activity (occupational, household, recreational), sexual habits, lifetime body size and abdominal obesity, smoking, and alcohol. Finally, we will collect biological specimen in order to be able to assess the role of relevant genetic polymorphisms (carcinogen metabolizing and hormone-related genes) which may influence the degree of susceptibility of developing prostate cancer either alone, or in combination with occupational or lifestyle risk factors. We propose to carry out a population-based case-control study of prostate cancer. Eligible subjects will be males, aged under 76, resident in the Montreal area, and Canadian citizens. Incident cases (n=1,500) will be actively ascertained from the 18 largest hospitals in the area. Controls (n=1,500) will be age and sex stratified and selected from electoral lists. Data will be collected as part of interviews. The data collection methods for the study of occupational agents will be based on an approach that was developed and extensively used by our research group. It is widely recognized as the reference method for such study design. Buccal cells will be collected under the direction of trained interviewers. The statistical analysis of this data set will describe the association between prostate cancer on the one hand and the various occupational exposures and lifestyle factors. The possible confounding effect of numerous other variables will be explored and, whenever appropriate, accounted for in the analytical models.

Epidemiological Methods and Biostatistics

Regression modelling of continuous covariates in epidemiology: A simulation study of nonparametric smoothers

Granting agency: Canadian Institutes of Health Research

PI: Michal Abrahamowicz

Collaborators: Mark Goldberg, Karen Leffondre

Abstract: Modern statistical analysis of data arising from investigations in epidemiology, clinical medicine, and the social sciences make extensive use of multiple regression techniques to estimate associations between outcome and explanatory variables. For continuously measured covariates, parametric representations of the response function are generally used. Current methods for estimating the shape of the response function are inadequate, possibly resulting in incorrect specification of the model and possible residual confounding effects. Nonparametric regression methods relax the strict assumptions imposed by the parametric representations thereby allowing the data to dictate the appropriate functional form. Although these flexible methods are well-developed and are available in some software packages, there are a number of outstanding problems that need to be tackled. Thus, the objectives of this simulation study are to evaluate different nonparametric smoothers to determine which methods and what amount of smoothing more accurately represent the true response functions, and to identify the most accurate model selection criteria. The simulations studies will cover a wide class of GLMs, the Cox model, different response functions for the main exposure variable, and different response functions for the covariates. in addition, we will develop some new computational routines to facilitate inference about nonparametric estimates.

Publications:

Jiguo Cao, Marie-France Valois, Mark S. Goldberg. An S-Plus Function to Calculate Relative Risks for Regression Models Using Natural Splines. Computer Methods and Programs in Biomedicine. 2006;84(1):58-62.

A Benedetti, M Abrahamowicz, Goldberg, MS. Accounting for data-dependent degrees of freedom selection when testing the effect of a continuous covariate in generalized additive models, In press, Communications in Statistics - Simulation and Computation.