An Introductory Lecture to Environmental Epidemiology

Part 1. Introductory Examples.Mark S. Goldberg

INRS-Institut Armand-Frappier, University of Quebec, and McGill University

July 2000

The author

Dr. Mark Goldberg obtained his MSc in 1985 and his PhD in 1991 from McGill, both degrees in epidemiology and biostatistics.

Dr. Goldberg is an associate professor at the INRS-Institut Armand-Frappier, University of Québec, and is adjunct professor at McGill University. He currently holds a health research scientist award from Health Canada.

His main interests are in occupational and environmental epidemiology, including cancer in textile manufacturing workers, health effects of exposures from municipal solid waste landfill sites, the relationship between ionizing radiation and cancer and reproductive outcomes, and the connection between tobacco smoking and back pain.

Currently, he is conducting research into the short- and long-term effects of air pollution, environmental case-control studies of breast cancer, and a study of waiting times for treatment of breast cancer in Quebec. Dr. Goldberg has published about 40 papers in scientific peer-review journals, is the recipient of research funds from a number of organizations, and sits on a number of scientific review panels.

ObjectivesThis is the first in a five-part series of an

introductory lecture on environmental epidemiology. The goal of the lecture is to provide the student with a basic understanding of the elements of environmental epidemiology. Throughout the lecture, examples from the literature are used to illustrate the basic methods. It is assumed that the student is familiar with basic epidemiology and with regression techniques.

Environmental Epidemiology

The study of the determinants of the distributions of disease that are exogenous to and nonessential for the normal functioning of human beings

Adapted from Hertz-Piccioto (in Rothman and Greenland, 1998)

Types of Environmental ExposuresPoint sources

Pollution from factories, municipal solid waste sites

Line sourcesEMF exposures from high tension power linesPollutants from internal combustion engines

around motorways

Area sourcesLong-range transport of combustion

products from trafficVolatile organic compounds contaminating

underground water reservoirs

Example: Cancer Rates Near a Solid Waste Landfill SiteEcological Analysis (Goldberg et al., Arch

Environ Health 1995;50:417-24):Landfill site opened in 1968100,000 persons lived within 2 km of the

siteIn 1993, it contained about 36x106 Tons of

domestic, commercial, & industrial waste

Rates for men and women living in zones around site 1981-1988

Zones defined by 3-character postal codes (fairly large areas)

Putative “upwind” and “downwind” zones

Putative “unexposed” zone far from the site

The “High” zone surrounds the landfill site to about 1 mile.

The “High A” zone is downwind and the “High B” zone is upwind; because of the crude geographic identifiers, there is a region directly surrounding the site that is in both sub-zones.

The “Medium” zone is further away from the site and exposure was likely to be very limited.

Map of the site showing the different exposure zones

Poisson regression adjusted for age and year, by sex

Reference zones selected from the “unexposed” areas to ensure similarities for:average household incomeproportion of immigrantsproportion first language was Frenchunemployment and poverty rates

Matching was not entirely successful, as some key factors were dissimilar (e.g., percentage of persons with an Italian family background)

Analytic study(Goldberg et al., Arch Environ Health 1999;54:291-6)

Multi-site cancer case-control study of occupation, men, 1979-85

Distance from site and by geographic zones (at time of interview)

Logistic regression for each site of cancer, adjusted for occupational and nonoccupational risk factors

Age, family income, cigarette smoking, alcohol consumption,

ethnicity, place of birth, body mass index, consumption of vitamins, occupational “salubrity”

Ecologic analysis Case-controlCancer site Geographic

regionNo. ofcases

RR 95% CI OR 95% CI

1Stomach High 190 1.2 1.1-1.4 0.8 0.5-1.3

High-A 121 1.2 0.9-1.5 0.8 0.5-1.4High-B 94 1.1 0.8-1.4 0.6 0.3-1.2

1Liver and High 72 1.5 1.2-2.0Intrahepatic High 72 1.3 0.9-1.8 1.8 0.8-4.3bile ducts High-A 53 1.8 1.2-2.6 1.5 0.5-4.4

High-B 30 1.2 0.8-2.0 1.5 0.5-4.8

Prostate High 563 1.0 0.9-1.1 1.5 1.0-2.1High-A 259 0.8 0.7-1.0 1.2 0.7-1.9High-B 350 1.2 1.0-1.4 2.0 1.3-3.0

Relative Risks for Cancer

Exposuremetric Categories

Numberof cases

Adjustedoddsratio*

95% Confidenceinterval (p-value for

linear trend test)Geographic Unexposed 24 1zone Low 2 0.7 0.2-3.1

Medium 6 0.9 0.4-2.4High 9 1.8 0.8-4.3High-A 5 1.5 0.5-4.4High-B 4 1.5 0.5-4.8

Distance (m) > 3000 28 11500-2999 7 1.3 0.6-3.2<1500 6 2.1 0.8-5.3Continuous 1.05 0.97-1.13 (0.219)

Relative Risks for Liver Cancer from the Case-control Analysis

Conclusions

1) Slightly different results obtain using different methodologies. Populations were somewhat different, although there was an overlap.

2) The results are inconclusive, except perhaps for liver cancer. Vinyl chloride monomer is one of the constituents of the biogas, and this is an accepted liver carcinogen.

3) Further studies are needed at other landfill sites. Results from such studies may be difficult to generalize if the constituents of the biogas differs and if exposure patterns in populations vary considerably.

Another Ecological Example

The following example is a complex longitudinal cohort study undertaken for the purposes of determining whether air pollution affects pulmonary function. The analysis presented here is for mortality and the comparison is between six cities in the US. As in the preceding case-control study, this study can be viewed as an ecological study standardized for personal risk factors.

Example: Harvard Six-cities Study (Dockery et al, NEJM 1993;329:1753-9)

Prospective cohort study of about 8,000 subjects selected randomly from 6 US cities with different levels of air pollution

Subjects followed every two years and lung function and questionnaires administered periodically

Ambient air exposures assessed from special fixed-site monitoring stations (particles, sulfates, gaseous pollutants)

Mortality analyses, comparing mean annual levels in each city for years near start of followup

Assumed that subjects did not move during followup and that the rank ordering of cities for levels of air pollution was invariant of followup time

Stratified Cox proportional hazards models to estimate cause-specific relative risks

Mortality rates by level of pollution by city and by pollutant

Estimates of Relative Rates of Mortality, Comparing Most Exposed to Least Exposed CityThis analysis is a Cox regression analysis

comparing the most polluted city (Steubenville) to the least polluted city (Portage). The range of exposures for fine particles is about 18.6 µg/m3. A wide range of key risks factors were included in the statistical model.

Results from Harvard Six-Cities Study: All-Cause Mortality RatesComparing Most Exposed to Least Exposed City (Fine Particles)

Comparing Steubenville to PortageRR 95%CI

All causes of death 1.26 1.08-1.47Lung cancer 1.37 0.81-2.31Cardiopulmonary disease 1.37 1.11-1.68All other causes 1.01 0.79-1.30

Range of exposure: 11-29.6 ug/m3Adjusted for age, sex, smoking, alcohol, body-mass index, occupational exposures to dust, gases, and fumes.

Source: Dockery et al., N Engl J Med 1993; 329:1753-1759

References

Environmental EpidemiologyHertz-Piccioto, I. “Environmental Epidemiology”, in

Rothman and Greenland: Modern Epidemiology, Second edition, Lippincott-Raven Publishers, 1998, Philadelphia, Chapter 28, pages 555-583.