Measurement error is common in epidemiologic research and may affect the validity of research results. It is therefore important to scrutinise the effects of measurement error in epidemiologic... Show moreMeasurement error is common in epidemiologic research and may affect the validity of research results. It is therefore important to scrutinise the effects of measurement error in epidemiologic research. Even simple forms of measurement error, for instance random measurement error in an exposure, can introduce bias in exposure-outcome associations. And even though there are situations in which measurement error does not introduce bias in the exposure-outcome association, for instance in case of random measurement error in a continuous outcome, it nearly always affects the precision and power of a study. In addition, other forms of measurement error, for example systematic measurement error or differential measurement error in an exposure, covariate or outcome, can affect exposure-outcome associations in complex ways that may not easily be anticipated. Adjusting for measurement error using measurement error correction methods may thus be necessary to obtain reliable estimates of exposure-outcome associations.To facilitate measurement error correction, information about the underlying measurement error mechanism (i.e., model) and its parameters is needed. The measurement error model can sometimes be estimated from internal or external validation data, replicates data or calibration data. Collection and the use of such measurement error mechanism data will likely improve the quality of epidemiologic analyses in the presence of measurement error. This can be done through the application of measurement error correction methods, which adjust the analyses taking into account the information from the measurement error mechanism. Alternatively, in the absence of concrete data about the mechanisms or the parameters of measurement error, sensitivity analysis for measurement error can be used, in which the impact on the epidemiologic analyses of one or a range of hypothesised measurement error mechanisms or their parameters can be investigated. The studies described in the thesis were set out to improve the understanding of the impact of measurement error, to facilitate the application of measurement error correction methods, to improve the design of epidemiologic studies when measurement error in a variable is suspected and, to develop tools to quantitatively assess the impact of measurement error in epidemiologic research. Show less
Nab, L.; Groenwold, R.H.H.; Smeden, M. van; Keogh, R.H. 2020
Observational data are increasingly used with the aim of estimating causal effects of treatments, through careful control for confounding. Marginal structural models estimated using inverse... Show moreObservational data are increasingly used with the aim of estimating causal effects of treatments, through careful control for confounding. Marginal structural models estimated using inverse probability weighting (MSMs-IPW), like other methods to control for confounding, assume that confounding variables are measured without error. The average treatment effect in an MSM-IPW may however be biased when a confounding variable is error prone. Using the potential outcome framework, we derive expressions for the bias due to confounder misclassification in analyses that aim to estimate the average treatment effect using an marginal structural model estimated using inverse probability weighting (MSM-IPW). We compare this bias with the bias due to confounder misclassification in analyses based on a conditional regression model. Focus is on a point-treatment study with a continuous outcome. Compared with bias in the average treatment effect in a conditional model, the bias in an MSM-IPW can be different in magnitude but is equal in sign. Also, we use a simulation study to investigate the finite sample performance of MSM-IPW and conditional models when a confounding variable is misclassified. Simulation results indicate that confidence intervals of the treatment effect obtained from MSM-IPW are generally wider, and coverage of the true treatment effect is higher compared with a conditional model, ranging from overcoverage if there is no confounder misclassification to undercoverage when there is confounder misclassification. Further, we illustrate in a study of blood pressure-lowering therapy, how the bias expressions can be used to inform a quantitative bias analysis to study the impact of confounder misclassification, supported by an online tool. Show less