Adjusted analyses in studies addressing therapy and harm: Users' guides to the medical literature

Thomas Agoritsas, Arnaud Merglen, Nilay D Shah, Martin O'Donnell, Gordon H. Guyatt

Research output: Contribution to journalReview article

35 Citations (Scopus)

Abstract

Observational studies almost always have bias because prognostic factors are unequally distributed between patients exposed or not exposed to an intervention. The standard approach to dealing with this problem is adjusted or stratified analysis. Its principle is to use measurement of risk factors to create prognostically homogeneous groups and to combine effect estimates across groups. The purpose of this Users' Guide is to introduce readers to fundamental concepts underlying adjustment as a way of dealing with prognostic imbalance and to the basic principles and relative trustworthiness of various adjustment strategies. One alternative to the standard approach is propensity analysis, in which groups are matched according to the likelihood of membership in exposed or unexposed groups. Propensity methods can deal with multiple prognostic factors, even if there are relatively few patients having outcome events. However, propensitymethods do not address other limitations of traditional adjustment: investigators may not have measured all relevant prognostic factors (or not accurately), and unknown factors may bias the results. A second approach, instrumental variable analysis, relies on identifying a variable associated with the likelihood of receiving the intervention but not associated with any prognostic factor or with the outcome (other than through the intervention); this could mimic randomization. However, as with assumptions of other adjustment approaches, it is never certain if an instrumental variable analysis eliminates bias. Although all these approaches can reduce the risk of bias in observational studies, none replace the balance of both known and unknown prognostic factors offered by randomization.

Original languageEnglish (US)
Pages (from-to)748-759
Number of pages12
JournalJAMA - Journal of the American Medical Association
Volume317
Issue number7
DOIs
StatePublished - Feb 21 2017

Fingerprint

Random Allocation
Observational Studies
Therapeutics
Research Design
Research Personnel

ASJC Scopus subject areas

  • Medicine(all)

Cite this

Adjusted analyses in studies addressing therapy and harm : Users' guides to the medical literature. / Agoritsas, Thomas; Merglen, Arnaud; Shah, Nilay D; O'Donnell, Martin; Guyatt, Gordon H.

In: JAMA - Journal of the American Medical Association, Vol. 317, No. 7, 21.02.2017, p. 748-759.

Research output: Contribution to journalReview article

Agoritsas, Thomas ; Merglen, Arnaud ; Shah, Nilay D ; O'Donnell, Martin ; Guyatt, Gordon H. / Adjusted analyses in studies addressing therapy and harm : Users' guides to the medical literature. In: JAMA - Journal of the American Medical Association. 2017 ; Vol. 317, No. 7. pp. 748-759.
@article{f339bab9fff44eeab0476e0a39cecb64,
title = "Adjusted analyses in studies addressing therapy and harm: Users' guides to the medical literature",
abstract = "Observational studies almost always have bias because prognostic factors are unequally distributed between patients exposed or not exposed to an intervention. The standard approach to dealing with this problem is adjusted or stratified analysis. Its principle is to use measurement of risk factors to create prognostically homogeneous groups and to combine effect estimates across groups. The purpose of this Users' Guide is to introduce readers to fundamental concepts underlying adjustment as a way of dealing with prognostic imbalance and to the basic principles and relative trustworthiness of various adjustment strategies. One alternative to the standard approach is propensity analysis, in which groups are matched according to the likelihood of membership in exposed or unexposed groups. Propensity methods can deal with multiple prognostic factors, even if there are relatively few patients having outcome events. However, propensitymethods do not address other limitations of traditional adjustment: investigators may not have measured all relevant prognostic factors (or not accurately), and unknown factors may bias the results. A second approach, instrumental variable analysis, relies on identifying a variable associated with the likelihood of receiving the intervention but not associated with any prognostic factor or with the outcome (other than through the intervention); this could mimic randomization. However, as with assumptions of other adjustment approaches, it is never certain if an instrumental variable analysis eliminates bias. Although all these approaches can reduce the risk of bias in observational studies, none replace the balance of both known and unknown prognostic factors offered by randomization.",
author = "Thomas Agoritsas and Arnaud Merglen and Shah, {Nilay D} and Martin O'Donnell and Guyatt, {Gordon H.}",
year = "2017",
month = "2",
day = "21",
doi = "10.1001/jama.2016.20029",
language = "English (US)",
volume = "317",
pages = "748--759",
journal = "JAMA - Journal of the American Medical Association",
issn = "0002-9955",
publisher = "American Medical Association",
number = "7",

}

TY - JOUR

T1 - Adjusted analyses in studies addressing therapy and harm

T2 - Users' guides to the medical literature

AU - Agoritsas, Thomas

AU - Merglen, Arnaud

AU - Shah, Nilay D

AU - O'Donnell, Martin

AU - Guyatt, Gordon H.

PY - 2017/2/21

Y1 - 2017/2/21

N2 - Observational studies almost always have bias because prognostic factors are unequally distributed between patients exposed or not exposed to an intervention. The standard approach to dealing with this problem is adjusted or stratified analysis. Its principle is to use measurement of risk factors to create prognostically homogeneous groups and to combine effect estimates across groups. The purpose of this Users' Guide is to introduce readers to fundamental concepts underlying adjustment as a way of dealing with prognostic imbalance and to the basic principles and relative trustworthiness of various adjustment strategies. One alternative to the standard approach is propensity analysis, in which groups are matched according to the likelihood of membership in exposed or unexposed groups. Propensity methods can deal with multiple prognostic factors, even if there are relatively few patients having outcome events. However, propensitymethods do not address other limitations of traditional adjustment: investigators may not have measured all relevant prognostic factors (or not accurately), and unknown factors may bias the results. A second approach, instrumental variable analysis, relies on identifying a variable associated with the likelihood of receiving the intervention but not associated with any prognostic factor or with the outcome (other than through the intervention); this could mimic randomization. However, as with assumptions of other adjustment approaches, it is never certain if an instrumental variable analysis eliminates bias. Although all these approaches can reduce the risk of bias in observational studies, none replace the balance of both known and unknown prognostic factors offered by randomization.

AB - Observational studies almost always have bias because prognostic factors are unequally distributed between patients exposed or not exposed to an intervention. The standard approach to dealing with this problem is adjusted or stratified analysis. Its principle is to use measurement of risk factors to create prognostically homogeneous groups and to combine effect estimates across groups. The purpose of this Users' Guide is to introduce readers to fundamental concepts underlying adjustment as a way of dealing with prognostic imbalance and to the basic principles and relative trustworthiness of various adjustment strategies. One alternative to the standard approach is propensity analysis, in which groups are matched according to the likelihood of membership in exposed or unexposed groups. Propensity methods can deal with multiple prognostic factors, even if there are relatively few patients having outcome events. However, propensitymethods do not address other limitations of traditional adjustment: investigators may not have measured all relevant prognostic factors (or not accurately), and unknown factors may bias the results. A second approach, instrumental variable analysis, relies on identifying a variable associated with the likelihood of receiving the intervention but not associated with any prognostic factor or with the outcome (other than through the intervention); this could mimic randomization. However, as with assumptions of other adjustment approaches, it is never certain if an instrumental variable analysis eliminates bias. Although all these approaches can reduce the risk of bias in observational studies, none replace the balance of both known and unknown prognostic factors offered by randomization.

UR - http://www.scopus.com/inward/record.url?scp=85014638634&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85014638634&partnerID=8YFLogxK

U2 - 10.1001/jama.2016.20029

DO - 10.1001/jama.2016.20029

M3 - Review article

C2 - 28241362

AN - SCOPUS:85014638634

VL - 317

SP - 748

EP - 759

JO - JAMA - Journal of the American Medical Association

JF - JAMA - Journal of the American Medical Association

SN - 0002-9955

IS - 7

ER -