Assessing the impact of different imputation methods on serial measures of renal function: The Strong Heart Study N.-M. Shara, J.-G. Umans, W. Wang, B.-V. Howard, H.-E. Resnick Kidney International Volume 71, Issue 7, Pages 701-705 (April 2007) DOI: 10.1038/sj.ki.5002105 Copyright © 2007 International Society of Nephrology Terms and Conditions
Figure 1 Comparison of mean Scr in the imputed sets versus the complete Strong Heart Study data set, at three levels of missing data. Kidney International 2007 71, 701-705DOI: (10.1038/sj.ki.5002105) Copyright © 2007 International Society of Nephrology Terms and Conditions
Figure 2 Comparison of mean eGFR in the imputed sets versus the complete Strong Heart Study data set, at three levels of missing data. Kidney International 2007 71, 701-705DOI: (10.1038/sj.ki.5002105) Copyright © 2007 International Society of Nephrology Terms and Conditions
Figure 3 HRs for the relationship among Scr, eGFR, eGFR calculated using imputed Scr and risk of CVD from separate Cox regression models, according to imputation method at 30% missing data. The HR estimate is shown as a dot and the extended horizontal line represents the 95% CIs. The complete case data set is plotted as the first group in all three methods of measuring kidney function. Kidney International 2007 71, 701-705DOI: (10.1038/sj.ki.5002105) Copyright © 2007 International Society of Nephrology Terms and Conditions