Michael Schomaker1, Sara Hogger, Leigh F Johnson, Christopher J Hoffmann, Till Bärnighausen, Christian Heumann. 1. From the aCentre for Infectious Disease Epidemiology and Research, University of Cape Town, Cape Town, South Africa; bInstitut für Statistik, Ludwig-Maximilans-Universität München, München, Germany; cAurum Institute, Johannesburg, South Africa; dSchool of Medicine, Johns Hopkins University, Baltimore, MD; eDepartment of Global Health and Population, Harvard T.H. Chan School of Public Health, Boston, MA; and fWellcome Trust Africa Centre for Health and Population Studies, University of KwaZulu-Natal, Mtubatuba, South Africa.
Abstract
BACKGROUND: Both CD4 count and viral load in HIV-infected persons are measured with error. There is no clear guidance on how to deal with this measurement error in the presence of missing data. METHODS: We used multiple overimputation, a method recently developed in the political sciences, to account for both measurement error and missing data in CD4 count and viral load measurements from four South African cohorts of a Southern African HIV cohort collaboration. Our knowledge about the measurement error of ln CD4 and log10 viral load is part of an imputation model that imputes both missing and mismeasured data. In an illustrative example, we estimate the association of CD4 count and viral load with the hazard of death among patients on highly active antiretroviral therapy by means of a Cox model. Simulation studies evaluate the extent to which multiple overimputation is able to reduce bias in survival analyses. RESULTS: Multiple overimputation emphasizes more strongly the influence of having high baseline CD4 counts compared to both a complete case analysis and multiple imputation (hazard ratio for >200 cells/mm vs. <25 cells/mm: 0.21 [95% confidence interval: 0.18, 0.24] vs. 0.38 [0.29, 0.48], and 0.29 [0.25, 0.34], respectively). Similar results are obtained when varying assumptions about measurement error, when using p-splines, and when evaluating time-updated CD4 count in a longitudinal analysis. The estimates of the association with viral load are slightly more attenuated when using multiple imputation instead of multiple overimputation. Our simulation studies suggest that multiple overimputation is able to reduce bias and mean squared error in survival analyses. CONCLUSIONS: Multiple overimputation, which can be used with existing software, offers a convenient approach to account for both missing and mismeasured data in HIV research.
BACKGROUND: Both CD4 count and viral load in HIV-infectedpersons are measured with error. There is no clear guidance on how to deal with this measurement error in the presence of missing data. METHODS: We used multiple overimputation, a method recently developed in the political sciences, to account for both measurement error and missing data in CD4 count and viral load measurements from four South African cohorts of a Southern African HIV cohort collaboration. Our knowledge about the measurement error of ln CD4 and log10 viral load is part of an imputation model that imputes both missing and mismeasured data. In an illustrative example, we estimate the association of CD4 count and viral load with the hazard of death among patients on highly active antiretroviral therapy by means of a Cox model. Simulation studies evaluate the extent to which multiple overimputation is able to reduce bias in survival analyses. RESULTS: Multiple overimputation emphasizes more strongly the influence of having high baseline CD4 counts compared to both a complete case analysis and multiple imputation (hazard ratio for >200 cells/mm vs. <25 cells/mm: 0.21 [95% confidence interval: 0.18, 0.24] vs. 0.38 [0.29, 0.48], and 0.29 [0.25, 0.34], respectively). Similar results are obtained when varying assumptions about measurement error, when using p-splines, and when evaluating time-updated CD4 count in a longitudinal analysis. The estimates of the association with viral load are slightly more attenuated when using multiple imputation instead of multiple overimputation. Our simulation studies suggest that multiple overimputation is able to reduce bias and mean squared error in survival analyses. CONCLUSIONS: Multiple overimputation, which can be used with existing software, offers a convenient approach to account for both missing and mismeasured data in HIV research.
Authors: J M Raboud; J S Montaner; B Conway; L Haley; C Sherlock; M V O'Shaughnessy; M T Schechter Journal: J Infect Dis Date: 1996-07 Impact factor: 5.226
Authors: Margaret May; Andrew Boulle; Sam Phiri; Eugene Messou; Landon Myer; Robin Wood; Olivia Keiser; Jonathan A C Sterne; Francois Dabis; Matthias Egger Journal: Lancet Date: 2010-07-15 Impact factor: 79.321
Authors: Angela M Bengtson; Daniel Westreich; Patrick Musonda; Audrey Pettifor; Carla Chibwesha; Benjamin H Chi; Bellington Vwalika; Brian W Pence; Jeffrey S A Stringer; William C Miller Journal: Epidemiology Date: 2016-09 Impact factor: 4.822
Authors: Viviane D Lima; Lu Wang; Chanson Brumme; Lang Wu; Julio S G Montaner; P Richard Harrigan Journal: PLoS One Date: 2017-02-02 Impact factor: 3.240