Random measurement error: Why worry? An example of cardiovascular risk factors

With the increased use of data not originally recorded for research, such as routine care data (or ‘big data’), measurement error is bound to become an increasingly relevant problem in medical research. A common view among medical researchers on the influence of random measurement error (i.e. classical measurement error) is that its presence leads to some degree of systematic underestimation of studied exposure-outcome relations (i.e. attenuation of the effect estimate). For the common situation where the analysis involves at least one exposure and one confounder, we demonstrate that the direction of effect of random measurement error on the estimated exposure-outcome relations can be difficult to anticipate. Using three example studies on cardiovascular risk factors, we illustrate that random measurement error in the exposure and/or confounder can lead to underestimation as well as overestimation of exposure-outcome relations. We therefore advise medical researchers to refrain from making claims about the direction of effect of measurement error in their manuscripts, unless the appropriate inferential tools are used to study or alleviate the impact of measurement error from the analysis.

[1]  Raymond J. Carroll,et al.  Measurement error in nonlinear models: a modern perspective , 2006 .

[2]  A. Algra,et al.  Second Manifestations of ARTerial disease (SMART) study: Rationale and design , 1999, European Journal of Epidemiology.

[3]  Mustafa Dosemeci,et al.  RE: “DOES NONDIFFERENTIAL MISCLASSIFICATION OF EXPOSURE ALWAYS BIAS A TRUE EFFECT TOWARD THE NULL VALUE?” , 1991 .

[4]  James A Hanley,et al.  Random measurement error and regression dilution bias , 2010, BMJ : British Medical Journal.

[5]  M. Nitzan,et al.  More accurate systolic blood pressure measurement is required for improved hypertension management: a perspective , 2017, Medical devices.

[6]  D. Hall.  Measurement Error in Nonlinear Models: A Modern Perspective , 2008 .

[7]  Z. Obermeyer,et al.  Predicting the Future - Big Data, Machine Learning, and Clinical Medicine. , 2016, The New England journal of medicine.

[8]  R Core Team,et al.  R: A language and environment for statistical computing. , 2014 .

[9]  N. Campbell,et al.  Misclassification of blood pressure by usual measurement in ambulatory physician practices. , 2005, American journal of hypertension.

[10]  M. Wong,et al.  Correlated measurement error--implications for nutritional epidemiology. , 2004, International journal of epidemiology.

[11]  Michael Böhm,et al.  2013 ESH/ESC Guidelines for the management of arterial hypertension: the Task Force for the management of arterial hypertension of the European Society of Hypertension (ESH) and of the European Society of Cardiology (ESC). , 2007, Journal of hypertension.

[12]  M Dosemeci,et al.  Does nondifferential misclassification of exposure always bias a true effect toward the null value? , 1990, American journal of epidemiology.

[13]  T. Lang,et al.  Poor blood pressure control in general practice: in search of explanations. , 2009, Archives of cardiovascular diseases.

[14]  Rafael Perera,et al.  Hypertension guideline recommendations in general practice: awareness, agreement, adoption, and adherence. , 2007, The British journal of general practice : the journal of the Royal College of General Practitioners.

[15]  B Rosner,et al.  Regression calibration method for correcting measurement-error bias in nutritional epidemiology. , 1997, The American journal of clinical nutrition.

[16]  I. White,et al.  A toolkit for measurement error correction, with a focus on nutritional epidemiology , 2014, Statistics in medicine.

[17]  John M. McAdams.  Alternativesfor Dealing with Errors in the Variables: An Example Using Panel Data , 1986 .

[18]  S. Greenland,et al.  How far from non-differential does exposure or disease misclassification have to be to bias measures of association away from the null? , 2008, International journal of epidemiology.

[19]  Robert P. Freckleton,et al.  Dealing with collinearity in behavioural and ecological data: model averaging and the problems of measurement error , 2010, Behavioral Ecology and Sociobiology.

[20]  B Rosner,et al.  Correction of logistic regression relative risk estimates and confidence intervals for systematic within-person measurement error. , 2006, Statistics in medicine.

[21]  Iris Pigeot,et al.  Handbook of epidemiology , 2005 .

[22]  S. Greenland,et al.  Proper interpretation of non-differential misclassification effects: expectations vs observations. , 2005, International journal of epidemiology.

[23]  P. Levy.  Measurement Error and Misclassification in Statistics and Epidemiology: Impacts and Bayesian Adjustments , 2004 .

[24]  S. Greenland,et al.  Exposure-measurement error is frequently ignored when interpreting epidemiologic study results , 2007, European Journal of Epidemiology.

[25]  M Y Wong,et al.  Measurement error in epidemiology: the design of validation studies II: bivariate situation. , 1999, Statistics in medicine.

[26]  James V. Zidek,et al.  CAUSALITY, MEASUREMENT ERROR AND MULTICOLLINEARITY IN EPIDEMIOLOGY , 1996 .

[27]  Andrew Gelman,et al.  Measurement error and the replication crisis , 2017, Science.

[28]  S. Sheps,et al.  Does evidence-based medicine suggest that physicians should not be measuring blood pressure in the hypertensive patient? , 2004, American journal of hypertension.

[29]  C. van Weel,et al.  A Novel Approach to Office Blood Pressure Measurement: 30-Minute Office Blood Pressure vs Daytime Ambulatory Blood Pressure , 2011, The Annals of Family Medicine.

[30]  S. Muff,et al.  Reverse attenuation in interaction terms due to covariate measurement error , 2015, Biometrical journal. Biometrische Zeitschrift.

[31]  J. R. Cook,et al.  Simulation-Extrapolation Estimation in Parametric Measurement Error Models , 1994 .