Table 4.

Results for death-censored ischemic stroke

ModelAnalytic ApproachImputation ModelaHazard RatioLower 95% Confidence IntervalUpper 95% Confidence IntervalP Value
1MI m=8b (sample size n=62,692)Variables in analytic modelc plus auxiliary variablesd1.361.101.68<0.01
2MI m=42b (sample size n=62,692)Variables in analytic modelc plus auxiliary variablesd1.361.101.68<0.01
3Complete casee (sample size n=36,107)1.260.951.700.10
4MI m=8b (sample size n=62,692)Variables in analytic modelc plus no auxiliary variablesd1.361.101.68<0.01
5MI m=8b (sample size n=62,692)Variables in analytic modelc plus BMI at listing and time since BMI was measured1.361.101.69<0.01
  • MI, multiple imputation; m, number of imputations.

  • a Imputation model also included censoring indicator and time to event.

  • b Cox proportional hazards model with multiple imputation to handle missing data (m imputed datasets).

  • c Analytic model included age, sex, race, BMI at transplant, cause of ESRD, dialysis vintage, dialysis modality, skilled nursing facility use indicator, number of hospital days, number of non-nephrology clinic visits, previous solid organ transplant, comorbidities, patient blood type, panel reactive antibody, donor age, donor sex, transplant type, number of HLA mismatches, and cold ischemia time.

  • d Auxiliary variables were BMI at listing, time since BMI was measured, weight at listing, and height at listing.

  • e Complete case analysis: Cox proportional hazards model on the dataset where observations with at least one covariate missing were removed.