Regression calibration method for correcting measurement-error bias in nutritional epidemiology
American Journal of Clinical NutritionPublished 1 April 1997Open access
Donna Spiegelman, Ann Yelmokas McDermott, Bernard Rosner
Citations255
SJR quartileQ1
SJR score2.11
SNIP1.75
Generate an AI Snapshot to get a quick, structured summary of this paper.
Study Snapshot
ObjectiveStudy objective
MethodsResearch methodology
PopulationPopulation studied
Sample sizeSample sizes
OutcomesStudy outcomes here
ResultsStudy results comes here
LimitationsResearch study limitations comes here
A concise AI-generated summary of the paper will appear here once you click Generate AI Snapshot.
TL;DR
Regression calibration is illustrated by correction of rate ratios describing the relations between the incidence of breast cancer and dietary intakes of vitamin A, alcohol, and total energy in the Nurses' Health Study.
Abstract
Regression calibration is a statistical method for adjusting point and interval estimates of effect obtained from regression models commonly used in epidemiology for bias due to measurement error in assessing nutrients or other variables. Previous work developed regression calibration for use in estimating odds ratios from logistic regression. We extend this here to estimating incidence rate ratios from Cox proportional hazards models and regression slopes from linear-regression models. Regression calibration is appropriate when a gold standard is available in a validation study and a linear measurement error with constant variance applies or when replicate measurements are available in a reliability study and linear random within-person error can be assumed. In this paper, the method is illustrated by correction of rate ratios describing the relations between the incidence of breast cancer and dietary intakes of vitamin A, alcohol, and total energy in the Nurses' Health Study. An example using linear regression is based on estimation of the relation between ultradistal radius bone density and dietary intakes of caffeine, calcium, and total energy in the Massachusetts Women's Health Study. Software implementing these methods uses SAS macros.
Keywords
MathematicsMedicineHealth Professions
The Annals of StatisticsCox's Regression Model for Counting Processes: A Large Sample Study
4,297 Citations1982P. Andersen, Richard D. Gill
Journal of the American Statistical AssociationEstimation of Regression Coefficients When Some Regressors are not Always Observed
2,647 Citations1994James M. Robins, Andrea Rotnitzky +1 more
American Journal of EpidemiologyTHE USE OF A SELF-ADMINISTERED QUESTIONNAIRE TO ASSESS DIET FOUR YEARS IN THE PAST
799 Citations1988Walter C. Willett, Laura Sampson +5 more
It is demonstrated that useful estimates of nutrient intake several years previously can be obtained by a relatively inexpensive, mailed, self-administered questionnaire.
BiometrikaCovariate measurement errors and parameter estimation in a failure time regression model
644 Citations1982Ross L. Prentice
Statistics in MedicineCorrection of logistic regression relative risk estimates and confidence intervals for systematic within‐person measurement error
638 Citations1989Bernard Rosner, Walter C. Willett +1 more
Two methods are provided to correct relative risk estimates obtained from logistic regression models for measurement errors in continuous exposures within cohort studies that may be due to either random (unbiased) within-person variation or to systematic errors for individual subjects.
American Journal of EpidemiologyINTERPRETATION AND CHOICE OF EFFECT MEASURES IN EPIDEMIOLOGIC ANALYSES1
518 Citations1987Sander Greenland
The concept of the odds ratio is now wellestablished in epidemiology, largely because it serves as a link between results obtained from follow-up studies and those obtainable from case-control studies.
American Journal of EpidemiologyCORRECTION OF LOGISTIC REGRESSION RELATIVE RISK ESTIMATES AND CONFIDENCE INTERVALS FOR MEASUREMENT ERROR: THE CASE OF MULTIPLE COVARIATES MEASURED WITH ERROR
460 Citations1990Bernard Rosner, Donna Spiegelman +1 more
The results indicate that the failure to find a substantial positive association between breast cancer risk and saturated fat intake cannot be explained by measurement error in fat, calories, or alcohol.
PubMedValidation of a dietary questionnaire used in a large-scale prospective cohort study on diet and cancer.
455 Citations1994R. Alexandra Goldbohm, Piet A. van den Brandt +5 more
It is concluded that the FFQ is able to rank subjects according to intake of food groups and nutrients, and this conclusion also applies to the cohort at large.
JAMADietary Fat and Fiber in Relation to Risk of Breast Cancer
392 Citations1992W. C. Willett
Evidence is provided against both an adverse influence of fat intake and a protective effect of fiber consumption by middle-aged women on breast cancer incidence over 8 years, and the positive association between intake of animal fat and risk of colon cancer observed in many studies provides ample reason to limit this source of energy.
American Journal of EpidemiologyDietary Assessment of Older Iowa Women with a Food Frequency Questionnaire: Nutrient Intake, Reproducibility, and Comparison with 24-Hour Dietary Recall Interviews
372 Citations1992Ronald G. Munger, Aaron R. Folsom +3 more
This food frequency questionnaire appears to be reasonably reproducible and accurate, so that its use may be extended to epidemiologic studies of older women with a broad range of socioeconomic backgrounds.
New England Journal of MedicineA Prospective Study of the Intake of Vitamins C, E, and A and the Risk of Breast Cancer
314 Citations1993David J. Hunter, JoAnn E. Manson +6 more
Large intakes of vitamin C or E did not protect women in this study from breast cancer; any benefit of vitamin A supplements may be limited to women with diets low in vitamin A.
Annual Review of Public HealthExposure Measurement Error: Influence on Exposure-Disease Relationships and Methods of Correction
310 Citations1993Duncan Thomas, Daniel O. Stram +1 more
The results of the former line of research have often been used qualitatively in the interpretation of epidemiologic findings, but the quantitative predictions have seldom been applied, and the latter line ofResearch appears to be particularly promising in terms of practical applications.
Journal of the American Statistical AssociationApproximate Quasi-likelihood Estimation in Models with Surrogate Predictors
281 Citations1990Raymond J. Carroll, Leonard A. Stefanski
American Journal of EpidemiologyCorrection of Logistic Regression Relative Risk Estimates and Confidence Intervals for Random Within-Person Measurement Error
265 Citations1992Bernard Rosner, Donna Spiegelman +1 more
The authors present measurement error methods for logistic regression when there is error (possibly correlated) in one or more covariates and one has data from both a main study and a reproducibility substudy, and applied to the Framingham Heart Study.
JAMAThe Framingham Study: The Epidemiology of Atherosclerotic Disease
226 Citations1981Robert I. Levy
Dr Dawber has produced an informative book whose purpose, he states, is to tell the "entire story of the Framingham Study—a narrative that without being too detailed would give the reader a reasonably complete knowledge of the overall project."
American Journal of EpidemiologyTHE EFFECTS OF MEASUREMENT ERRORS ON RELATWE RISK REGRESSIONS
212 Citations1990Ben Armstrong
The effects of random error in numerical measurements of risk factors (covariates) in relative risk regressions in simple models in which distributions are assumed normal are concerns.
Journal of the American Statistical AssociationStatistical Models for Longitudinal Studies of Health.
191 Citations1993Paul S. Albert, James H. Dweyer +3 more
Introduction to statistical models for longitudinal observation, James Dwyer and Manning Feinleib considerations in the design of longitudinal surveys of health, and logistic and related models in longitudinal studies of chronic disease risk.
American Journal of EpidemiologyValidation Studies using an Alloyed Gold Standard
161 Citations1993Sholom Wacholder, Ben Armstrong +1 more
A key assumption made when using a validation study to correct an estimate of relative risk for bias due to misclassification or measurement error is that the available measure is compared to a gold standard measured without error.
American Journal of EpidemiologyASSESSING NUTRIENT INTAKE WITH AN ABBREVIATED DIETARY INTERVIEW1
157 Citations1985Tim Byers, James R. Marshall +3 more
An abbreviated dietary history ascertaining the intake of as few as 15-20 food items may be all that is required in order to test the association between disease outcome and the Intake of a single nutrient.
American Journal of EpidemiologyMeasurement Error Correction for Logistic Regression Models with an "Alloyed Gold Standard"
151 Citations1997D Spiegelman, Sebastian Schneeweiß +1 more
It is proven that if the errors between Z and X are uncorrelated, the regression calibration method has no bias even when the gold standard is "alloyed," and when a third method of exposure assessment is available, point and interval estimates of the correlation between the errors in X and Z are derived.
American Journal of Clinical NutritionEvaluation of a self-administered dietary questionnaire for use in a cohort study
126 Citations1982MG Jain, Laurie Harrison +2 more
It is concluded that when applied to large numbers of women in a cohort study, the self-administered questionnaire is feasible to administer and will be sufficiently accurate in estimating nutrient intake as compared to estimates from a detailed, interview-type diet history.
EpidemiologyWhen Measurement Errors Correlate with Truth
107 Citations1995Sholom Wacholder
Epidemiologists need to be wary of deviations from the classical error model, since poor measurement might occasionally explain a positive finding even when the error does not differ by disease status.
Statistics in MedicineAnalysis of case‐control data with covariate measurement error: Application to diet and colon cancer
106 Citations1989Ben Armstrong, Alice S. Whittemore +1 more
A simple correction to naive estimates of odds ratios from logistic regression of disease on fallible measurements of covariates removes bias and yields confidence intervals and significance tests.
EpidemiologyCaffeine and Other Predictors of Bone Density among Pre- and Perimenopausal Women
98 Citations1993Mauricio Hernández‐Ávila, Meir J. Stampfer +6 more
Caffeine was inversely associated with bone density independent of dietary, anthropometric, and hormonal factors, indicating that caffeine may be an important determinant of bone density.
Journal of the Royal Statistical Society Series B (Statistical Methodology)Semiparametric Efficient Estimation of a Conditional Density with Missing or Mismeasured Covariates
97 Citations1995James M. Robins, Fushing Hsieh +1 more
Statistics in MedicineEstimating the accuracy of dietary questionnaire assessments: Validation in terms of structural equation models
95 Citations1994Rudolf Kaaks, Elio Ríboli +3 more
It is described how, by additional comparison with a third type of measurement such as a biochemical marker, the validity of dietary questionnaire measurements can be evaluated even in situations where the random errors of repeat reference measurements can no longer be assumed to be independent.
BiometricsCost-Efficient Study Designs for Binary Response Data with Gaussian Covariate Measurement Error
82 Citations1991Donna Spiegelman, R Gray
When mismeasurement of the exposure variable is anticipated, epidemiologic cohort studies may be augmented to include a validation study, where a small sample of data relating the imperfect exposure measurement method to the better method is collected.
Journal of Clinical EpidemiologyStatistical uncertainty due to misclassification: Implications for validation substudies
80 Citations1988Sander Greenland
The present paper shows that the alternative of a fully-validated design may provide more information per unit cost than a larger study coupled with a validation substudy.
Statistics in MedicineMeasurement error in dietary assessment: An investigation using covariance structure models. Part I
80 Citations1993Martyn Plummer, David Clayton
This work applies the covariance matrix of the repeated measures of diet, widely used in social science, to assessments of nitrogen intake, which is essentially equivalent to protein intake.
Journal of the American Statistical AssociationCase-Control Studies With Errors in Covariates
77 Citations1993Raymond J. Carroll, Mitchell H. Gail +1 more
Statistics in MedicineMeasurement error in dietary assessment: An investigation using covariance structure models. Part II
43 Citations1993Martyn Plummer, David Clayton
Inclusion of urine nitrogen data produces more pessimistic estimates of the quality of dietary intake measurements and shows that previous assumptions about independence of measurement error were wrong, highlighting the need for well founded assumptions in the analysis of measurementerror.