Estimating risk difference from relative association measures in meta-analysis can infrequently pose interpretational challenges

Mohammad H Murad, Victor Manuel Montori, Stephen D. Walter, Gordon H. Guyatt

Research output: Contribution to journalArticle

13 Citations (Scopus)

Abstract

Objective: Risk difference (RD) is often estimated from relative association measures generated by meta-analysis and a particular group's baseline risk. We describe a problematic situation in using this approach. Study Design and Setting: We encountered a meta-analysis in which a confidence interval (CI) of relative risk (RR) overlapped 1.0; the point estimate favored treatment A, but when we used RR and median baseline risk to calculate a CI for RD, a greater portion of the CI favored treatment B (a result that some may find counterintuitive). We then calculated 10 different RDs from recently published meta-analyses in outcomes in which CIs of RR crossed 1.0 using three methods: estimation from RR, estimation from the odds ratio, and pooling RDs across trials. Results: When RD is estimated from relative measures, the counterintuitive result occurred in 2 of 10 instances. This discordance of interpretation is because of the logarithmic transformation that makes CIs of relative measures asymmetric around their point estimates. Conclusion: When RD is estimated from relative association measures that are nonsignificant and this counterintuitive situation occurs, it may be more appropriate to pool RD across studies. Pooling is particularly valid when baseline risks across studies are homogeneous.

Original languageEnglish (US)
Pages (from-to)865-867
Number of pages3
JournalJournal of Clinical Epidemiology
Volume62
Issue number8
DOIs
StatePublished - Aug 2009

Fingerprint

Meta-Analysis
Confidence Intervals
Odds Ratio

Keywords

  • Absolute measures of association
  • Meta-analysis
  • Odds ratio
  • Relative measures of association
  • Relative risk
  • Risk difference

ASJC Scopus subject areas

  • Epidemiology

Cite this

Estimating risk difference from relative association measures in meta-analysis can infrequently pose interpretational challenges. / Murad, Mohammad H; Montori, Victor Manuel; Walter, Stephen D.; Guyatt, Gordon H.

In: Journal of Clinical Epidemiology, Vol. 62, No. 8, 08.2009, p. 865-867.

Research output: Contribution to journalArticle

@article{f833a6379cb7424c894c8512441f45c3,
title = "Estimating risk difference from relative association measures in meta-analysis can infrequently pose interpretational challenges",
abstract = "Objective: Risk difference (RD) is often estimated from relative association measures generated by meta-analysis and a particular group's baseline risk. We describe a problematic situation in using this approach. Study Design and Setting: We encountered a meta-analysis in which a confidence interval (CI) of relative risk (RR) overlapped 1.0; the point estimate favored treatment A, but when we used RR and median baseline risk to calculate a CI for RD, a greater portion of the CI favored treatment B (a result that some may find counterintuitive). We then calculated 10 different RDs from recently published meta-analyses in outcomes in which CIs of RR crossed 1.0 using three methods: estimation from RR, estimation from the odds ratio, and pooling RDs across trials. Results: When RD is estimated from relative measures, the counterintuitive result occurred in 2 of 10 instances. This discordance of interpretation is because of the logarithmic transformation that makes CIs of relative measures asymmetric around their point estimates. Conclusion: When RD is estimated from relative association measures that are nonsignificant and this counterintuitive situation occurs, it may be more appropriate to pool RD across studies. Pooling is particularly valid when baseline risks across studies are homogeneous.",
keywords = "Absolute measures of association, Meta-analysis, Odds ratio, Relative measures of association, Relative risk, Risk difference",
author = "Murad, {Mohammad H} and Montori, {Victor Manuel} and Walter, {Stephen D.} and Guyatt, {Gordon H.}",
year = "2009",
month = "8",
doi = "10.1016/j.jclinepi.2008.11.005",
language = "English (US)",
volume = "62",
pages = "865--867",
journal = "Journal of Clinical Epidemiology",
issn = "0895-4356",
publisher = "Elsevier USA",
number = "8",

}

TY - JOUR

T1 - Estimating risk difference from relative association measures in meta-analysis can infrequently pose interpretational challenges

AU - Murad, Mohammad H

AU - Montori, Victor Manuel

AU - Walter, Stephen D.

AU - Guyatt, Gordon H.

PY - 2009/8

Y1 - 2009/8

N2 - Objective: Risk difference (RD) is often estimated from relative association measures generated by meta-analysis and a particular group's baseline risk. We describe a problematic situation in using this approach. Study Design and Setting: We encountered a meta-analysis in which a confidence interval (CI) of relative risk (RR) overlapped 1.0; the point estimate favored treatment A, but when we used RR and median baseline risk to calculate a CI for RD, a greater portion of the CI favored treatment B (a result that some may find counterintuitive). We then calculated 10 different RDs from recently published meta-analyses in outcomes in which CIs of RR crossed 1.0 using three methods: estimation from RR, estimation from the odds ratio, and pooling RDs across trials. Results: When RD is estimated from relative measures, the counterintuitive result occurred in 2 of 10 instances. This discordance of interpretation is because of the logarithmic transformation that makes CIs of relative measures asymmetric around their point estimates. Conclusion: When RD is estimated from relative association measures that are nonsignificant and this counterintuitive situation occurs, it may be more appropriate to pool RD across studies. Pooling is particularly valid when baseline risks across studies are homogeneous.

AB - Objective: Risk difference (RD) is often estimated from relative association measures generated by meta-analysis and a particular group's baseline risk. We describe a problematic situation in using this approach. Study Design and Setting: We encountered a meta-analysis in which a confidence interval (CI) of relative risk (RR) overlapped 1.0; the point estimate favored treatment A, but when we used RR and median baseline risk to calculate a CI for RD, a greater portion of the CI favored treatment B (a result that some may find counterintuitive). We then calculated 10 different RDs from recently published meta-analyses in outcomes in which CIs of RR crossed 1.0 using three methods: estimation from RR, estimation from the odds ratio, and pooling RDs across trials. Results: When RD is estimated from relative measures, the counterintuitive result occurred in 2 of 10 instances. This discordance of interpretation is because of the logarithmic transformation that makes CIs of relative measures asymmetric around their point estimates. Conclusion: When RD is estimated from relative association measures that are nonsignificant and this counterintuitive situation occurs, it may be more appropriate to pool RD across studies. Pooling is particularly valid when baseline risks across studies are homogeneous.

KW - Absolute measures of association

KW - Meta-analysis

KW - Odds ratio

KW - Relative measures of association

KW - Relative risk

KW - Risk difference

UR - http://www.scopus.com/inward/record.url?scp=67649373583&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=67649373583&partnerID=8YFLogxK

U2 - 10.1016/j.jclinepi.2008.11.005

DO - 10.1016/j.jclinepi.2008.11.005

M3 - Article

VL - 62

SP - 865

EP - 867

JO - Journal of Clinical Epidemiology

JF - Journal of Clinical Epidemiology

SN - 0895-4356

IS - 8

ER -