Air pollution estimation error and what it does to epidemiological analysis

Pickles, J.H.

doi:10.1016/0004-6981(82)90294-3

Cited by 6 publications

(5 citation statements)

References 3 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…On the other hand, exposure misclassification probably occurs in nearly every epidemiologic study.” Reviews of the effects of exposure measurement error and the numerous methods that have been proposed to correct for biases that result when exposure measurement error is present have been published (1, 8, 10, 26, 46). There have been many theoretical investigations of the effects of measurement error on point and interval estimates of exposure-disease associations, and the widespread and profound presence of exposure measurement error in environmental health data has been documented (6, 12, 14, 32, 45). Nevertheless, there have been few original scientific publications that make use of existing methods for explicit exposure measurement error correction in environmental or occupational health fields, in contrast, for example, to nutritional epidemiology where such corrections are becoming more common (16, 31, 49).…”

Section: Introductionmentioning

confidence: 99%

Approaches to Uncertainty in Exposure Assessment in Environmental Epidemiology

Spiegelman

2010

Annu. Rev. Public Health

View full text Add to dashboard Cite

Uncertainty in assessment of individual exposure levels leads to bias, often, but not always, toward the null in estimates of health effects, and to underestimation of the variability of the estimates, leading to anticonservative p-values. In the absence of data on the uncertainty in individual exposure estimates, sensitivity analysis, also known as uncertainty analysis and bias analysis, is available. Hypothesized values of key parameters of the model relating the observed exposure to the true exposure are used to assess the resulting amount of bias in point and interval estimates. In general, the relative risk estimates can vary from zero to infinity as the hypothesized values of key parameters of the measurement error model vary. Thus, we recommend that exposure validation data be used to empirically adjust point and interval estimates of health effects for measurement error. The remainder of this review gives an overview of available methods for doing so. Just as we routinely adjust for confounding, we can and should routinely adjust for measurement error.

show abstract

Section: Introductionmentioning

confidence: 99%

Approaches to Uncertainty in Exposure Assessment in Environmental Epidemiology

Spiegelman

2010

Annu. Rev. Public Health

View full text Add to dashboard Cite

show abstract

“…( ,~) The mortality-pollution regression coefficient is vulnerable to errors of this kind because of the difficulty of establishing representative figures for pollution levels. Detailed analysis (24) shows that the error consists of two components: a systematic underestimate in the regression coefficient, as noted by Lave and Seskin,@) and further, a random fluctuation with zero mean. A further source of systematic uncertainty centers around the choice of the pollution variable.…”

Section: Pollution Measurement Errormentioning

confidence: 81%

Health Risks and Air Pollution — Error Analysis for a Cross‐Sectional Mortality Study

Pickles

1986

Risk Analysis

View full text Add to dashboard Cite

An attempt is made to analyze in quantitative terms the uncertainties in multiple regression estimates of the effects of air pollution on death rates. A range of factors--statistical fluctuations in numbers of deaths, differences in local age distribution, differences in smoking habits, errors in estimated pollution levels, migration, and variability of the characterization of socioeconomic effects--are assessed as potential sources of error. Both the precision and the robustness of the regression calculation are shown to be poor. Examples and illustrative calculations are given based on a study of U. K. death rates around the 1971 Census.

show abstract

“…This point has not always been properly understood in the literature. For example, Pickles (1982) claimed that, in a univariate linear regression analysis of the relationship between SO, concentration and mortality in 16 British towns, random errors in the SO, measurements implied that the P-value for the hypothesis of no association was biased toward 0. While the naive regression coefficient, B, was biased towards 0 in his example, the variance of the naive estimator was consistently estimated by standard techniques.…”

Section: The Effects Of Errors In Measurement Of Air-pollution Exposurementioning

confidence: 99%

Statistical issues in epidemiologic studies of the health effects of ambient acid aerosols

Ware

Stram

1988

Can J Statistics

View full text Add to dashboard Cite

Although the ecologic effects of acid rain have been widely reported, relatively little is known about the effects of acidic air pollution on human health, Some epidemiologic and animal studies suggest, however, that acidity is an important determinant of the respiratory health effects of aerosols. This paper reviews some of that evidence and discusses its implications for the design and analysis of epidemiologic studies. We contrast two types of exposure patterns: peak exposures associated with air pollution episodes, and chronic exposures resulting from persistently high levels of air pollutants. Recent work on the analysis of repeated categorical outcome variables provides new methods for the analysis of episode studies. Studies of long-term exposure require comparisons among population groups, and these comparisons can be subject to the design effects characteristic of multistage sample surveys. We examine the implications ofthese design effects for epidemiologic studies. Finally, the paper discusses the measurement errors induced by the use of outdoor measurements to quantify personal exposure to air pollutants. Recent work on methods for errors-in-variables problems may aid in assessing the effects of such errors on conventional analyses of airpollution studies.

show abstract

Air pollution estimation error and what it does to epidemiological analysis

Cited by 6 publications

References 3 publications

Approaches to Uncertainty in Exposure Assessment in Environmental Epidemiology

Approaches to Uncertainty in Exposure Assessment in Environmental Epidemiology

Health Risks and Air Pollution — Error Analysis for a Cross‐Sectional Mortality Study

Statistical issues in epidemiologic studies of the health effects of ambient acid aerosols

Contact Info

Product

Resources

About