4 December 2011 in Statistics in Medicine
Multiple imputation fills in missing data with posterior predictive draws from imputation models. To assess the adequacy of imputation models, we can compare completed data with their replicates simulated under the imputation model. We apply analyses of substantive interest to both datasets and use posterior predictive checks of the differences of these estimates to quantify the evidence of model inadequacy. We can further integrate out the imputed missing data and their replicates over the completed‐data analyses to reduce variance in the comparison. In many cases, the checking procedure can be easily implemented using standard imputation software by treating re‐imputations under the model as posterior predictive replicates. Thus, it can be applied for non‐Bayesian imputation methods. We also sketch several strategies for applying the method in the context of practical imputation analyses. We illustrate the method using two real data applications and study its property using a simulation. Copyright © 2011 John Wiley & Sons, Ltd.
27 July 2020 in Statistics in Medicine
30 March 2023 in Statistics in Medicine
17 December 2019 in Statistics in Medicine
8 November 2024 in Cancer Epidemiology, Biomarkers & Prevention
Background: In 2012, the US Preventive Services Task Force recommended against prostate cancer screening using the PSA test for all age groups. In 2018, the US Preventive Services Task Force’s recommendation shifted from a “D” (not recommended) to a “C” (selectively offering PSA-based screening based on professional judgment and patient preferences) in men ages 55 to 69. Limited reliable county-level prostate cancer screening data are available for cancer surveillance purposes. Methods: Utilizing data from the National Health Interview Survey (NHIS) and Behavioral Risk Factor Surveillance System (BRFSS) collected in 2012 to 2019, state- and county-level small area models were developed for estimating PSA testing. Model diagnosis, internal validation, and external validation examining associations of PSA testing and prostate cancer incidence were conducted. Results: Model-based estimates of PSA testing rates were produced for all US states and 3,142 counties for two data periods: 2012 to 2016 and 2018 to 2019. Geographic variations across counties were demonstrated through maps. Moderate positive correlations between PSA-based screening and prostate cancer incidence were observed, e.g., the state-level weighted Pearson’s correlation coefficients were 0.5025 (P value = 0.0002) and 0.3691 (P value = 0.0077) for 2012 to 2016 and 2018 to 2019, respectively. Conclusions: These modeled estimates showed improved precision and adjusted for the differences between the BRFSS and NHIS. The approach of combining the NHIS and BRFSS utilized strengths of the larger sample size of the BRFSS and generally higher response rates and better household coverage from the NHIS. Impact: The resulting small area estimates offer a valuable resource for the cancer surveillance community, aiding in targeted interventions, decision making, and further research endeavors.