Elsevier

Annals of Epidemiology

Volume 7, Issue 2, February 1997, Pages 154-164
Annals of Epidemiology

Review
The correction of risk estimates for measurement error

https://doi.org/10.1016/S1047-2797(96)00149-4Get rights and content

Abstract

PURPOSE: The methods available for the correction of risk estimates for measurement errors are reviewed. The assumptions and design implications of each of the following six methods are noted: linear imputation, absolute limits, maximum likelihood, latent class, discriminant analysis and Gibbs sampling.

METHODS: All methods, with the exception of the absolute limits approach, require either repeated determinations on the same subjects with use of the methods that are prone to error or a validation study, in which the measurement is performed for a number of persons with use of both the error-prone method and a more accurate method regarded as a “gold standard.”

RESULTS: The maximum likelihood, latent class and absolute limits methods are most suitable for purely discrete risk factors. The linear imputation methods and the closely related discrimination analysis method are suitable for continuous risk factors which, together with the errors of measurement, are usually assumed to be normally distributed.

CONCLUSIONS: The Gibbs sampling approach is, in principle, useful for both discrete and continuous risk factors and measurement errors, although its use does mandate that the user specify models and dependencies that may be very complex. Also, the Bayesian approach implicit in the use of Gibbs sampling is difficult to apply to the design of the case-control study.

References (34)

  • S MacMahon et al.

    Blood pressure, stroke, and coronary heart disease: Part 1

    Lancet

    (1990)
  • J Berkson

    Are there two regressions?

    J Am Stat Assoc

    (1950)
  • AM Walker et al.

    Misclassification of covariates

    Stat Med

    (1991)
  • B Cox et al.

    The effect on stratum specific odds ratios of nondifferential misclassification of a dichotomous covariate

    Am J Epidemiol

    (1991)
  • M Blettner et al.

    What does an observed relative risk convey about possible misclassification?

    Methods Inf. Med

    (1984)
  • AT Marinos et al.

    Experimental quantiles of epidemiological indices in case-control studies with non-differential misclassification

    Stat Med

    (1995)
  • N Qizilbash et al.

    Repeat measurement of case-control data: Correcting risk estimates for misclassification due to regression dilution of lipids in transient ischaemic attacks and minor ischaemic strokes

    Am J Epidemiol

    (1991)
  • B Rosner et al.

    Correction of logistic regression relative risk estimates and confidence intervals for systematic within-person measurement error

    Stat Med.

    (1989)
  • DV Lindley

    Introduction to Probability and Statistics from a Bayesian viewpoint (Part 1)

    (1969)
  • SW Duffy et al.

    Misclassification in more than one factor in a case-control study: A combination of Mantel-Haenszel and maximum likelihood approaches

    Stat Med

    (1989)
  • SW Duffy et al.

    External validation, repeat determination, and precision of risk estimation in misclassified exposure data in epidemiology

    J Epidemiol Community Health

    (1992)
  • RA Elton et al.

    Correcting for the effect of misclassification bias in a case-control study using data from two different questionnaires

    Biometrics

    (1983)
  • S Greenland

    On correcting for misclassification in twin studies and other matched-pair studies

    Stat Med

    (1989)
  • S Greenland

    The effect of misclassification in matched-pair case-control studies

    Am J Epidemiol

    (1982)
  • S Greenland

    The effect of misclassification in the presence of covariates

    Am J Epidemiol

    (1980)
  • S Greenland et al.

    Correcting for misclassification in two-way tables and matched studies

    Int J Epidemiol

    (1983)
  • J Palmgren et al.

    Exponential family nonlinear models for categorical data with errors of observation

    Appl Stochastic Models Data Analysis

    (1987)
  • Cited by (36)

    • Breast cancer incidence using administrative data: correction with sensitivity and specificity

      2009, Journal of Clinical Epidemiology
      Citation Excerpt :

      However, the main challenge has been to differentiate incident from prevalent cases [8–12,15–17,25,26,33]. Because of all the known imperfections of claims data to identify incident cancer cases, a true or exact measurement of the number of incident cancer cases cannot be made [34,35]. To correct this measurement, we used a statistical method based on a two-phase study design that we had previously developed [36,37].

    • Simulations showed that validation of database-derived diagnostic criteria based on a small subsample reduced bias

      2007, Journal of Clinical Epidemiology
      Citation Excerpt :

      Some limitations of our simulations have to be recognized. First, we considered only those methods that can be easily implemented by a relatively unsophisticated user of standard commercial statistical software, and we did not investigate some more refined approaches [31]. Furthermore, as in any simulation study, the underlying assumptions were somewhat arbitrary [19,23,28,32], even if such crucial parameters as the PPV were clinically plausible [5].

    View all citing articles on Scopus
    1

    For the duration of this work, S. A. Bashir was supported by a studentship from the Medical Research Council.

    View full text