Mollie E Wood1,2, Stavroula Chrysanthopoulou3, Hedvig M E Nordeng1,2,4, Kate L Lapane3. 1. PharmacoEpidemiology & Drug Safety Research Group, Department of Pharmacy, School of Pharmacy. 2. PharmaTox Strategic Research Initiative, Faculty of Mathematics and Natural Sciences, University of Oslo, Oslo, Norway. 3. Department of Quantitative Health Sciences, University of Massachusetts Medical School, Worcester, MA. 4. Department of Child Health, Norwegian Institute of Public Health, Oslo, Norway.
Abstract
PURPOSE: To investigate the ability of the propensity score (PS) to reduce confounding bias in the presence of nondifferential misclassification of treatment, using simulations. METHODS: Using an example from the pregnancy medication safety literature, we carried out simulations to quantify the effect of nondifferential misclassification of treatment under varying scenarios of sensitivity and specificity, exposure prevalence (10%, 50%), outcome type (continuous and binary), true outcome (null and increased risk), confounding direction, and different PS applications (matching, stratification, weighting, regression), and obtained measures of bias and 95% confidence interval coverage. RESULTS: All methods were subject to substantial bias toward the null due to nondifferential exposure misclassification (range: 0%-47% for 50% exposure prevalence and 0%-80% for 10% exposure prevalence), particularly if specificity was low (<97%). PS stratification produced the least biased effect estimates. We observed that the impact of sensitivity and specificity on the bias and coverage for each adjustment method is strongly related to prevalence of exposure: as exposure prevalence decreases and/or outcomes are continuous rather than categorical, the effect of misclassification is magnified, producing larger biases and loss of coverage of 95% confidence intervals. PS matching resulted in unpredictably biased effect estimates. CONCLUSIONS: The results of this study underline the importance of assessing exposure misclassification in observational studies in the context of PS methods. Although PS methods reduce confounding bias, bias owing to nondifferential misclassification is of potentially greater concern.
PURPOSE: To investigate the ability of the propensity score (PS) to reduce confounding bias in the presence of nondifferential misclassification of treatment, using simulations. METHODS: Using an example from the pregnancy medication safety literature, we carried out simulations to quantify the effect of nondifferential misclassification of treatment under varying scenarios of sensitivity and specificity, exposure prevalence (10%, 50%), outcome type (continuous and binary), true outcome (null and increased risk), confounding direction, and different PS applications (matching, stratification, weighting, regression), and obtained measures of bias and 95% confidence interval coverage. RESULTS: All methods were subject to substantial bias toward the null due to nondifferential exposure misclassification (range: 0%-47% for 50% exposure prevalence and 0%-80% for 10% exposure prevalence), particularly if specificity was low (<97%). PS stratification produced the least biased effect estimates. We observed that the impact of sensitivity and specificity on the bias and coverage for each adjustment method is strongly related to prevalence of exposure: as exposure prevalence decreases and/or outcomes are continuous rather than categorical, the effect of misclassification is magnified, producing larger biases and loss of coverage of 95% confidence intervals. PS matching resulted in unpredictably biased effect estimates. CONCLUSIONS: The results of this study underline the importance of assessing exposure misclassification in observational studies in the context of PS methods. Although PS methods reduce confounding bias, bias owing to nondifferential misclassification is of potentially greater concern.
Authors: Timothy L Lash; Matthew P Fox; Richard F MacLehose; George Maldonado; Lawrence C McCandless; Sander Greenland Journal: Int J Epidemiol Date: 2014-07-30 Impact factor: 7.196
Authors: Kristin Palmsten; Krista F Huybrechts; Helen Mogun; Mary K Kowal; Paige L Williams; Karin B Michels; Soko Setoguchi; Sonia Hernández-Díaz Journal: PLoS One Date: 2013-06-26 Impact factor: 3.240
Authors: David Hajage; Florence Tubach; Philippe Gabriel Steg; Deepak L Bhatt; Yann De Rycke Journal: BMC Med Res Methodol Date: 2016-03-31 Impact factor: 4.615
Authors: Abu Mohd Naser; Thomas F Clasen; Stephen P Luby; Mahbubur Rahman; Leanne Unicomb; Kazi M Ahmed; Solaiman Doza; Shadassa Ourshalimian; Howard H Chang; Jennifer D Stowell; K M Venkat Narayan; Mohammad Shamsudduha; Shivani A Patel; Bethany O'Shea; Matthew O Gribble Journal: Int J Environ Res Public Health Date: 2019-06-28 Impact factor: 3.390
Authors: Mollie E Wood; Kate L Lapane; Marleen M H J van Gelder; Dheeraj Rai; Hedvig M E Nordeng Journal: Pharmacoepidemiol Drug Saf Date: 2017-10-17 Impact factor: 2.890