A causal framework for understanding the effect of losses to follow-up on epidemiologic analyses in clinic-based cohorts: The case of HIV-infected patients on antiretroviral therapy in Africa

Elvin H. Geng, David V. Glidden, David Bangsberg, Mwebesa Bosco Bwana, Nicholas Musinguzi, Denis Nash, John Z. Metcalfe, Constantin T. Yiannoutsos, Jeffrey N. Martin, Maya L. Petersen

Research output: Contribution to journalArticle

28 Citations (Scopus)

Abstract

Although clinic-based cohorts are most representative of the "real world," they are susceptible to loss to follow-up. Strategies for managing the impact of loss to follow-up are therefore needed to maximize the value of studies conducted in these cohorts. The authors evaluated adult patients starting antiretroviral therapy at an HIV/AIDS clinic in Uganda, where 29% of patients were lost to follow-up after 2 years (January 1, 2004-September 30, 2007). Unweighted, inverse probability of censoring weighted (IPCW), and sampling-based approaches (using supplemental data from a sample of lost patients subsequently tracked in the community) were used to identify the predictive value of sex on mortality. Directed acyclic graphs (DAGs) were used to explore the structural basis for bias in each approach. Among 3,628 patients, unweighted and IPCW analyses found men to have higher mortality than women, whereas the sampling-based approach did not. DAGs encoding knowledge about the data-generating process, including the fact that death is a cause of being classified as lost to follow-up in this setting, revealed "collider" bias in the unweighted and IPCW approaches. In a clinic-based cohort in Africa, unweighted and IPCW approaches-which rely on the "missing at random" assumption-yielded biased estimates. A sampling-based approach can in general strengthen epidemiologic analyses conducted in many clinic-based cohorts, including those examining other diseases. American Journal of Epidemiology

Original languageEnglish (US)
Pages (from-to)1080-1087
Number of pages8
JournalAmerican Journal of Epidemiology
Volume175
Issue number10
DOIs
StatePublished - May 15 2012
Externally publishedYes

Fingerprint

HIV
Lost to Follow-Up
Uganda
Mortality
Therapeutics
Acquired Immunodeficiency Syndrome
Epidemiology

Keywords

  • Africa
  • antiretroviral therapy
  • clinic-based cohorts
  • directed acyclic graphs
  • informative censoring
  • inverse probability of censoring weights
  • loss to follow-up
  • missing at random

ASJC Scopus subject areas

  • Epidemiology

Cite this

A causal framework for understanding the effect of losses to follow-up on epidemiologic analyses in clinic-based cohorts : The case of HIV-infected patients on antiretroviral therapy in Africa. / Geng, Elvin H.; Glidden, David V.; Bangsberg, David; Bwana, Mwebesa Bosco; Musinguzi, Nicholas; Nash, Denis; Metcalfe, John Z.; Yiannoutsos, Constantin T.; Martin, Jeffrey N.; Petersen, Maya L.

In: American Journal of Epidemiology, Vol. 175, No. 10, 15.05.2012, p. 1080-1087.

Research output: Contribution to journalArticle

Geng, Elvin H. ; Glidden, David V. ; Bangsberg, David ; Bwana, Mwebesa Bosco ; Musinguzi, Nicholas ; Nash, Denis ; Metcalfe, John Z. ; Yiannoutsos, Constantin T. ; Martin, Jeffrey N. ; Petersen, Maya L. / A causal framework for understanding the effect of losses to follow-up on epidemiologic analyses in clinic-based cohorts : The case of HIV-infected patients on antiretroviral therapy in Africa. In: American Journal of Epidemiology. 2012 ; Vol. 175, No. 10. pp. 1080-1087.
@article{97a6f708fece4a58bb4a18af929b2a1a,
title = "A causal framework for understanding the effect of losses to follow-up on epidemiologic analyses in clinic-based cohorts: The case of HIV-infected patients on antiretroviral therapy in Africa",
abstract = "Although clinic-based cohorts are most representative of the {"}real world,{"} they are susceptible to loss to follow-up. Strategies for managing the impact of loss to follow-up are therefore needed to maximize the value of studies conducted in these cohorts. The authors evaluated adult patients starting antiretroviral therapy at an HIV/AIDS clinic in Uganda, where 29{\%} of patients were lost to follow-up after 2 years (January 1, 2004-September 30, 2007). Unweighted, inverse probability of censoring weighted (IPCW), and sampling-based approaches (using supplemental data from a sample of lost patients subsequently tracked in the community) were used to identify the predictive value of sex on mortality. Directed acyclic graphs (DAGs) were used to explore the structural basis for bias in each approach. Among 3,628 patients, unweighted and IPCW analyses found men to have higher mortality than women, whereas the sampling-based approach did not. DAGs encoding knowledge about the data-generating process, including the fact that death is a cause of being classified as lost to follow-up in this setting, revealed {"}collider{"} bias in the unweighted and IPCW approaches. In a clinic-based cohort in Africa, unweighted and IPCW approaches-which rely on the {"}missing at random{"} assumption-yielded biased estimates. A sampling-based approach can in general strengthen epidemiologic analyses conducted in many clinic-based cohorts, including those examining other diseases. American Journal of Epidemiology",
keywords = "Africa, antiretroviral therapy, clinic-based cohorts, directed acyclic graphs, informative censoring, inverse probability of censoring weights, loss to follow-up, missing at random",
author = "Geng, {Elvin H.} and Glidden, {David V.} and David Bangsberg and Bwana, {Mwebesa Bosco} and Nicholas Musinguzi and Denis Nash and Metcalfe, {John Z.} and Yiannoutsos, {Constantin T.} and Martin, {Jeffrey N.} and Petersen, {Maya L.}",
year = "2012",
month = "5",
day = "15",
doi = "10.1093/aje/kwr444",
language = "English (US)",
volume = "175",
pages = "1080--1087",
journal = "American Journal of Epidemiology",
issn = "0002-9262",
publisher = "Oxford University Press",
number = "10",

}

TY - JOUR

T1 - A causal framework for understanding the effect of losses to follow-up on epidemiologic analyses in clinic-based cohorts

T2 - The case of HIV-infected patients on antiretroviral therapy in Africa

AU - Geng, Elvin H.

AU - Glidden, David V.

AU - Bangsberg, David

AU - Bwana, Mwebesa Bosco

AU - Musinguzi, Nicholas

AU - Nash, Denis

AU - Metcalfe, John Z.

AU - Yiannoutsos, Constantin T.

AU - Martin, Jeffrey N.

AU - Petersen, Maya L.

PY - 2012/5/15

Y1 - 2012/5/15

N2 - Although clinic-based cohorts are most representative of the "real world," they are susceptible to loss to follow-up. Strategies for managing the impact of loss to follow-up are therefore needed to maximize the value of studies conducted in these cohorts. The authors evaluated adult patients starting antiretroviral therapy at an HIV/AIDS clinic in Uganda, where 29% of patients were lost to follow-up after 2 years (January 1, 2004-September 30, 2007). Unweighted, inverse probability of censoring weighted (IPCW), and sampling-based approaches (using supplemental data from a sample of lost patients subsequently tracked in the community) were used to identify the predictive value of sex on mortality. Directed acyclic graphs (DAGs) were used to explore the structural basis for bias in each approach. Among 3,628 patients, unweighted and IPCW analyses found men to have higher mortality than women, whereas the sampling-based approach did not. DAGs encoding knowledge about the data-generating process, including the fact that death is a cause of being classified as lost to follow-up in this setting, revealed "collider" bias in the unweighted and IPCW approaches. In a clinic-based cohort in Africa, unweighted and IPCW approaches-which rely on the "missing at random" assumption-yielded biased estimates. A sampling-based approach can in general strengthen epidemiologic analyses conducted in many clinic-based cohorts, including those examining other diseases. American Journal of Epidemiology

AB - Although clinic-based cohorts are most representative of the "real world," they are susceptible to loss to follow-up. Strategies for managing the impact of loss to follow-up are therefore needed to maximize the value of studies conducted in these cohorts. The authors evaluated adult patients starting antiretroviral therapy at an HIV/AIDS clinic in Uganda, where 29% of patients were lost to follow-up after 2 years (January 1, 2004-September 30, 2007). Unweighted, inverse probability of censoring weighted (IPCW), and sampling-based approaches (using supplemental data from a sample of lost patients subsequently tracked in the community) were used to identify the predictive value of sex on mortality. Directed acyclic graphs (DAGs) were used to explore the structural basis for bias in each approach. Among 3,628 patients, unweighted and IPCW analyses found men to have higher mortality than women, whereas the sampling-based approach did not. DAGs encoding knowledge about the data-generating process, including the fact that death is a cause of being classified as lost to follow-up in this setting, revealed "collider" bias in the unweighted and IPCW approaches. In a clinic-based cohort in Africa, unweighted and IPCW approaches-which rely on the "missing at random" assumption-yielded biased estimates. A sampling-based approach can in general strengthen epidemiologic analyses conducted in many clinic-based cohorts, including those examining other diseases. American Journal of Epidemiology

KW - Africa

KW - antiretroviral therapy

KW - clinic-based cohorts

KW - directed acyclic graphs

KW - informative censoring

KW - inverse probability of censoring weights

KW - loss to follow-up

KW - missing at random

UR - http://www.scopus.com/inward/record.url?scp=84861388568&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84861388568&partnerID=8YFLogxK

U2 - 10.1093/aje/kwr444

DO - 10.1093/aje/kwr444

M3 - Article

C2 - 22306557

AN - SCOPUS:84861388568

VL - 175

SP - 1080

EP - 1087

JO - American Journal of Epidemiology

JF - American Journal of Epidemiology

SN - 0002-9262

IS - 10

ER -