Differential Misclassification Arising from Nondifferential Errors in Exposure Measurement

Division of Health Examination Statistics, Centers for Disease Control, Hyattsville, MD.
American Journal of Epidemiology (Impact Factor: 5.23). 12/1991; 134(10):1233-44.
Source: PubMed


Misclassification into exposure categories formed from a continuous variable arises from measurement error in the continuous variable. Examples and mathematical results are presented to show that if the measurement error is nondifferential (independent of disease status), the resulting misclassification will often be differential, even in cohort studies. The degree and direction of differential misclassification vary with the exposure distribution, the category definitions, the measurement error distribution, and the exposure-disease relation. Failure to recognize the likelihood of differential misclassification may lead to incorrect conclusions about the effects of measurement error on estimates of relative risk when categories are formed from continuous variables, such as dietary intake. Simulations were used to examine some effects of nondifferential measurement error. Under the conditions used, nondifferential measurement error reduced relative risk estimates, but not to the degree predicted by the assumption of nondifferential misclassification. When relative risk estimates were corrected using methods appropriate for nondifferential misclassification, the "corrected" relative risks were almost always higher than the true relative risks, sometimes considerably higher. The greater the measurement error, the more inaccurate was the correction. The effects of exposure measurement errors need more critical evaluation.

Download full-text


Available from: Katherine Mayhew Flegal, Jan 03, 2014
  • Source
    • "Exposure measurement error can lead to exposure misclassification when exposure surrogates for individual participants are classified into categories for analysis. Differential misclassification can arise in categorical exposure metrics even when there is nondifferential error (i.e., independent of disease status) in an exposure variable that is measured on a continuous scale (Flegal et al. 1991). In order for epidemiologic studies to be evaluated and utilized appropriately in risk assessment, it is important that exposure measurement error is characterized and evaluated thoroughly with consideration of the magnitude and direction of any potential exposure misclassification bias (for example, see Bergen et al. 2013). "
    [Show abstract] [Hide abstract]
    ABSTRACT: Background: There is a recognized need to improve the application of epidemiologic data in human health risk assessment especially for understanding and characterizing risks from environmental and occupational exposures. Although there is uncertainty associated with the results of most epidemiologic studies, techniques exist to characterize uncertainty that can be applied to improve weight-of-evidence evaluations and risk characterization efforts. Methods: This report derives from a Health and Environmental Sciences Institute (HESI) workshop held in Research Triangle Park, North Carolina, to discuss the utility of using epidemiologic data in risk assessments, including the use of advanced analytic methods to address sources of uncertainty. Epidemiologists, toxicologists, and risk assessors from academia, government, and industry convened to discuss uncertainty, exposure assessment, and application of analytic methods to address these challenges. Synthesis: Several recommendations emerged to help improve the utility of epidemiologic data in risk assessment. For example, improved characterization of uncertainty is needed to allow risk assessors to quantitatively assess potential sources of bias. Data are needed to facilitate this quantitative analysis, and interdisciplinary approaches will help ensure that sufficient information is collected for a thorough uncertainty evaluation. Advanced analytic methods and tools such as directed acyclic graphs (DAGs) and Bayesian statistical techniques can provide important insights and support interpretation of epidemiologic data. Conclusions: The discussions and recommendations from this workshop demonstrate that there are practical steps that the scientific community can adopt to strengthen epidemiologic data for decision making.
    Environmental Health Perspectives 07/2014; 122(11). DOI:10.1289/ehp.1308062 · 7.98 Impact Factor
  • Source
    • "Unconditional logistic regression modeling adjusted for the frequency-matching variables (10-year age group and sex) and for age and education (a surrogate for socioeconomic status: less than high school, high school graduate, and college graduate (referent group)) was conducted for cumulative years and estimated lifetime cumulative exposure of farm pesticide use. Age was included as well as age group to adjust for residual confounding within age groups [25,26]. Sex was included as men and women differ in pesticide use [27,28]. "
    [Show abstract] [Hide abstract]
    ABSTRACT: An excess incidence of brain cancer in farmers has been noted in several studies. The National Institute for Occupational Safety and Health developed the Upper Midwest Health Study (UMHS) as a case-control study of intracranial gliomas and pesticide uses among rural residents. Previous studies of UMHS participants, using "ever-never" exposure to farm pesticides and analyzing men and women separately, found no positive association of farm pesticide exposure and glioma risks. The primary objective was to determine if quantitatively estimated exposure of pesticide applicators was associated with an increased risk of glioma in male and female participants. The study included 798 histologically confirmed primary intracranial glioma cases (45 % with proxy respondents) and 1,175 population-based controls, all adult (age 18-80) non-metropolitan residents of Iowa, Michigan, Minnesota, and Wisconsin. The analyses used quantitatively estimated exposure from questionnaire responses evaluated by an experienced industrial hygienist with 25 years of work on farm pesticide analyses. Odds ratios (ORs) and 95 % confidence intervals (CIs) using unconditional logistic regression modeling were calculated adjusting for frequency-matching variables (10-year age group and sex), and for age and education (a surrogate for socioeconomic status). Analyses were separately conducted with or without proxy respondents. No significant positive associations with glioma were observed with cumulative years or estimated lifetime cumulative exposure of farm pesticide use. There was, a significant inverse association for phenoxy pesticide used on the farm (OR 0.96 per 10 g-years of cumulative exposure, CI 0.93-0.99). No significant findings were observed when proxy respondents were excluded. Non-farm occupational applicators of any pesticide had decreased glioma risk: OR 0.72, CI 0.52-0.99. Similarly, house and garden pesticide applicators had a decreased risk of glioma: OR 0.79, CI 0.66-0.93, with statistically significant inverse associations for use of 2,4-D, arsenates, organophosphates, and phenoxys. These results are consistent with our previous findings for UMHS of reported farm pesticide exposure and support a lack of positive association between pesticides and glioma.
    Environmental Health 06/2012; 11(1):39. DOI:10.1186/1476-069X-11-39 · 3.37 Impact Factor
  • Source
    • "Another limitation is that our findings were likely to be affected by imprecise measurement of red and processed meat consumption and potential confounders. Categorisation of main exposures and confounders that are measured with error may induce misclassification and may bias the expected RR toward or away from the null value (Flegal et al, 1991; Wacholder et al, 1991; Wacholder, 1995). Thus, misclassification of red and processed meat consumption and of potential confounders might have resulted in an over- or underestimation of the association between red and processed meat consumption and risk of pancreatic cancer. "
    [Show abstract] [Hide abstract]
    ABSTRACT: Whether red and processed meat consumption is a risk factor for pancreatic cancer remains unclear. We conducted a meta-analysis to summarise the evidence from prospective studies of red and processed meat consumption and pancreatic cancer risk. Relevant studies were identified by searching PubMed and EMBASE databases through November 2011. Study-specific results were pooled using a random-effects model. Eleven prospective studies, with 6643 pancreatic cancer cases, were included in the meta-analysis. An increase in red meat consumption of 120 g per day was associated with an overall relative risk (RR) of 1.13 (95% confidence interval (CI)=0.93-1.39; P(heterogeneity)<0.001). Red meat consumption was positively associated with pancreatic cancer risk in men (RR=1.29; 95% CI=1.08-1.53; P(heterogeneity)=0.28; five studies), but not in women (RR=0.93; 95% CI=0.74-1.16; P(heterogeneity)=0.21; six studies). The RR of pancreatic cancer for a 50 g per day increase in processed meat consumption was 1.19 (95% CI=1.04-1.36; P(heterogeneity)=0.46). Findings from this meta-analysis indicate that processed meat consumption is positively associated with pancreatic cancer risk. Red meat consumption was associated with an increased risk of pancreatic cancer in men. Further prospective studies are needed to confirm these findings.
    British Journal of Cancer 01/2012; 106(3):603-7. DOI:10.1038/bjc.2011.585 · 4.84 Impact Factor
Show more