Loading…

Sensitivity analysis for unobserved confounding of direct and indirect effects using uncertainty intervals

To estimate direct and indirect effects of an exposure on an outcome from observed data, strong assumptions about unconfoundedness are required. Since these assumptions cannot be tested using the observed data, a mediation analysis should always be accompanied by a sensitivity analysis of the result...

Full description

Saved in:
Bibliographic Details
Published in:Statistics in medicine 2018-05, Vol.37 (10), p.1744-1762
Main Authors: Lindmark, Anita, Luna, Xavier, Eriksson, Marie
Format: Article
Language:English
Subjects:
Citations: Items that this one cites
Items that cite this one
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
cited_by cdi_FETCH-LOGICAL-c3870-7910b408a42c2b21da6e6c2c85c1a2bd2f4bea56ddc271eb8fefdf4655bdb5d3
cites cdi_FETCH-LOGICAL-c3870-7910b408a42c2b21da6e6c2c85c1a2bd2f4bea56ddc271eb8fefdf4655bdb5d3
container_end_page 1762
container_issue 10
container_start_page 1744
container_title Statistics in medicine
container_volume 37
creator Lindmark, Anita
Luna, Xavier
Eriksson, Marie
description To estimate direct and indirect effects of an exposure on an outcome from observed data, strong assumptions about unconfoundedness are required. Since these assumptions cannot be tested using the observed data, a mediation analysis should always be accompanied by a sensitivity analysis of the resulting estimates. In this article, we propose a sensitivity analysis method for parametric estimation of direct and indirect effects when the exposure, mediator, and outcome are all binary. The sensitivity parameters consist of the correlations between the error terms of the exposure, mediator, and outcome models. These correlations are incorporated into the estimation of the model parameters and identification sets are then obtained for the direct and indirect effects for a range of plausible correlation values. We take the sampling variability into account through the construction of uncertainty intervals. The proposed method is able to assess sensitivity to both mediator‐outcome confounding and confounding involving the exposure. To illustrate the method, we apply it to a mediation study based on the data from the Swedish Stroke Register (Riksstroke). An R package that implements the proposed method is available.
doi_str_mv 10.1002/sim.7620
format article
fullrecord <record><control><sourceid>proquest_swepu</sourceid><recordid>TN_cdi_swepub_primary_oai_DiVA_org_umu_125929</recordid><sourceformat>XML</sourceformat><sourcesystem>PC</sourcesystem><sourcerecordid>2007121177</sourcerecordid><originalsourceid>FETCH-LOGICAL-c3870-7910b408a42c2b21da6e6c2c85c1a2bd2f4bea56ddc271eb8fefdf4655bdb5d3</originalsourceid><addsrcrecordid>eNp1kV1rFDEUhoNY7FoFf4EMeNObaZOzM8nMZalVCxUvWrwN-TgpWWaSNZm07L83a9cWBG_yEnh4OOe8hHxg9IxRCufZz2eCA31FVoyOoqXQD6_JioIQLResPyZvc95QylgP4g05hrHjMKzHFdncYsh-8Q9-2TUqqGmXfW5cTE0JUWdMD2gbE4OLJVgf7pvoGusTmqXStvHh8EHnauSm5D1UgsG0KB-qtD7Voqb8jhy5Gvj-kCfk7svV3eW39ubH1-vLi5vWrAdBWzEyqjs6qA4MaGBWceQGzNAbpkBbcJ1G1XNrDQiGenDorOt432ure7s-Ie2TNj_itmi5TX5WaSej8vKz_3khY7qXZS6SQT_CWPnTJ36b4q-CeZGzzwanSQWMJUugVDBgTIiKfvoH3cSS6tH2FKwHxnlHX4QmxZwTuucRGJX7umStS-7rqujHg7DoGe0z-Lefl2Ue_YS7_4rk7fX3P8LfyZihRg</addsrcrecordid><sourcetype>Open Access Repository</sourcetype><iscdi>true</iscdi><recordtype>article</recordtype><pqid>2023816640</pqid></control><display><type>article</type><title>Sensitivity analysis for unobserved confounding of direct and indirect effects using uncertainty intervals</title><source>Wiley-Blackwell Read &amp; Publish Collection</source><creator>Lindmark, Anita ; Luna, Xavier ; Eriksson, Marie</creator><creatorcontrib>Lindmark, Anita ; Luna, Xavier ; Eriksson, Marie</creatorcontrib><description>To estimate direct and indirect effects of an exposure on an outcome from observed data, strong assumptions about unconfoundedness are required. Since these assumptions cannot be tested using the observed data, a mediation analysis should always be accompanied by a sensitivity analysis of the resulting estimates. In this article, we propose a sensitivity analysis method for parametric estimation of direct and indirect effects when the exposure, mediator, and outcome are all binary. The sensitivity parameters consist of the correlations between the error terms of the exposure, mediator, and outcome models. These correlations are incorporated into the estimation of the model parameters and identification sets are then obtained for the direct and indirect effects for a range of plausible correlation values. We take the sampling variability into account through the construction of uncertainty intervals. The proposed method is able to assess sensitivity to both mediator‐outcome confounding and confounding involving the exposure. To illustrate the method, we apply it to a mediation study based on the data from the Swedish Stroke Register (Riksstroke). An R package that implements the proposed method is available.</description><identifier>ISSN: 0277-6715</identifier><identifier>ISSN: 1097-0258</identifier><identifier>EISSN: 1097-0258</identifier><identifier>DOI: 10.1002/sim.7620</identifier><identifier>PMID: 29462839</identifier><language>eng</language><publisher>England: Wiley Subscription Services, Inc</publisher><subject>direct effects ; indirect effects ; mediation ; Sensitivity analysis ; sequential ignorability ; Statistics ; statistik ; unmeasured confounding</subject><ispartof>Statistics in medicine, 2018-05, Vol.37 (10), p.1744-1762</ispartof><rights>Copyright © 2018 John Wiley &amp; Sons, Ltd.</rights><lds50>peer_reviewed</lds50><oa>free_for_read</oa><woscitedreferencessubscribed>false</woscitedreferencessubscribed><citedby>FETCH-LOGICAL-c3870-7910b408a42c2b21da6e6c2c85c1a2bd2f4bea56ddc271eb8fefdf4655bdb5d3</citedby><cites>FETCH-LOGICAL-c3870-7910b408a42c2b21da6e6c2c85c1a2bd2f4bea56ddc271eb8fefdf4655bdb5d3</cites><orcidid>0000-0003-3298-1555 ; 0000-0003-3187-1987 ; 0000-0002-4600-0060</orcidid></display><links><openurl>$$Topenurl_article</openurl><openurlfulltext>$$Topenurlfull_article</openurlfulltext><thumbnail>$$Tsyndetics_thumb_exl</thumbnail><link.rule.ids>230,314,780,784,885,27924,27925</link.rule.ids><backlink>$$Uhttps://www.ncbi.nlm.nih.gov/pubmed/29462839$$D View this record in MEDLINE/PubMed$$Hfree_for_read</backlink><backlink>$$Uhttps://urn.kb.se/resolve?urn=urn:nbn:se:umu:diva-125929$$DView record from Swedish Publication Index$$Hfree_for_read</backlink></links><search><creatorcontrib>Lindmark, Anita</creatorcontrib><creatorcontrib>Luna, Xavier</creatorcontrib><creatorcontrib>Eriksson, Marie</creatorcontrib><title>Sensitivity analysis for unobserved confounding of direct and indirect effects using uncertainty intervals</title><title>Statistics in medicine</title><addtitle>Stat Med</addtitle><description>To estimate direct and indirect effects of an exposure on an outcome from observed data, strong assumptions about unconfoundedness are required. Since these assumptions cannot be tested using the observed data, a mediation analysis should always be accompanied by a sensitivity analysis of the resulting estimates. In this article, we propose a sensitivity analysis method for parametric estimation of direct and indirect effects when the exposure, mediator, and outcome are all binary. The sensitivity parameters consist of the correlations between the error terms of the exposure, mediator, and outcome models. These correlations are incorporated into the estimation of the model parameters and identification sets are then obtained for the direct and indirect effects for a range of plausible correlation values. We take the sampling variability into account through the construction of uncertainty intervals. The proposed method is able to assess sensitivity to both mediator‐outcome confounding and confounding involving the exposure. To illustrate the method, we apply it to a mediation study based on the data from the Swedish Stroke Register (Riksstroke). An R package that implements the proposed method is available.</description><subject>direct effects</subject><subject>indirect effects</subject><subject>mediation</subject><subject>Sensitivity analysis</subject><subject>sequential ignorability</subject><subject>Statistics</subject><subject>statistik</subject><subject>unmeasured confounding</subject><issn>0277-6715</issn><issn>1097-0258</issn><issn>1097-0258</issn><fulltext>true</fulltext><rsrctype>article</rsrctype><creationdate>2018</creationdate><recordtype>article</recordtype><recordid>eNp1kV1rFDEUhoNY7FoFf4EMeNObaZOzM8nMZalVCxUvWrwN-TgpWWaSNZm07L83a9cWBG_yEnh4OOe8hHxg9IxRCufZz2eCA31FVoyOoqXQD6_JioIQLResPyZvc95QylgP4g05hrHjMKzHFdncYsh-8Q9-2TUqqGmXfW5cTE0JUWdMD2gbE4OLJVgf7pvoGusTmqXStvHh8EHnauSm5D1UgsG0KB-qtD7Voqb8jhy5Gvj-kCfk7svV3eW39ubH1-vLi5vWrAdBWzEyqjs6qA4MaGBWceQGzNAbpkBbcJ1G1XNrDQiGenDorOt432ure7s-Ie2TNj_itmi5TX5WaSej8vKz_3khY7qXZS6SQT_CWPnTJ36b4q-CeZGzzwanSQWMJUugVDBgTIiKfvoH3cSS6tH2FKwHxnlHX4QmxZwTuucRGJX7umStS-7rqujHg7DoGe0z-Lefl2Ue_YS7_4rk7fX3P8LfyZihRg</recordid><startdate>20180510</startdate><enddate>20180510</enddate><creator>Lindmark, Anita</creator><creator>Luna, Xavier</creator><creator>Eriksson, Marie</creator><general>Wiley Subscription Services, Inc</general><scope>NPM</scope><scope>AAYXX</scope><scope>CITATION</scope><scope>K9.</scope><scope>7X8</scope><scope>ADHXS</scope><scope>ADTPV</scope><scope>AOWAS</scope><scope>D8T</scope><scope>D93</scope><scope>ZZAVC</scope><orcidid>https://orcid.org/0000-0003-3298-1555</orcidid><orcidid>https://orcid.org/0000-0003-3187-1987</orcidid><orcidid>https://orcid.org/0000-0002-4600-0060</orcidid></search><sort><creationdate>20180510</creationdate><title>Sensitivity analysis for unobserved confounding of direct and indirect effects using uncertainty intervals</title><author>Lindmark, Anita ; Luna, Xavier ; Eriksson, Marie</author></sort><facets><frbrtype>5</frbrtype><frbrgroupid>cdi_FETCH-LOGICAL-c3870-7910b408a42c2b21da6e6c2c85c1a2bd2f4bea56ddc271eb8fefdf4655bdb5d3</frbrgroupid><rsrctype>articles</rsrctype><prefilter>articles</prefilter><language>eng</language><creationdate>2018</creationdate><topic>direct effects</topic><topic>indirect effects</topic><topic>mediation</topic><topic>Sensitivity analysis</topic><topic>sequential ignorability</topic><topic>Statistics</topic><topic>statistik</topic><topic>unmeasured confounding</topic><toplevel>peer_reviewed</toplevel><toplevel>online_resources</toplevel><creatorcontrib>Lindmark, Anita</creatorcontrib><creatorcontrib>Luna, Xavier</creatorcontrib><creatorcontrib>Eriksson, Marie</creatorcontrib><collection>PubMed</collection><collection>CrossRef</collection><collection>ProQuest Health &amp; Medical Complete (Alumni)</collection><collection>MEDLINE - Academic</collection><collection>SWEPUB Umeå universitet full text</collection><collection>SwePub</collection><collection>SwePub Articles</collection><collection>SWEPUB Freely available online</collection><collection>SWEPUB Umeå universitet</collection><collection>SwePub Articles full text</collection><jtitle>Statistics in medicine</jtitle></facets><delivery><delcategory>Remote Search Resource</delcategory><fulltext>fulltext</fulltext></delivery><addata><au>Lindmark, Anita</au><au>Luna, Xavier</au><au>Eriksson, Marie</au><format>journal</format><genre>article</genre><ristype>JOUR</ristype><atitle>Sensitivity analysis for unobserved confounding of direct and indirect effects using uncertainty intervals</atitle><jtitle>Statistics in medicine</jtitle><addtitle>Stat Med</addtitle><date>2018-05-10</date><risdate>2018</risdate><volume>37</volume><issue>10</issue><spage>1744</spage><epage>1762</epage><pages>1744-1762</pages><issn>0277-6715</issn><issn>1097-0258</issn><eissn>1097-0258</eissn><abstract>To estimate direct and indirect effects of an exposure on an outcome from observed data, strong assumptions about unconfoundedness are required. Since these assumptions cannot be tested using the observed data, a mediation analysis should always be accompanied by a sensitivity analysis of the resulting estimates. In this article, we propose a sensitivity analysis method for parametric estimation of direct and indirect effects when the exposure, mediator, and outcome are all binary. The sensitivity parameters consist of the correlations between the error terms of the exposure, mediator, and outcome models. These correlations are incorporated into the estimation of the model parameters and identification sets are then obtained for the direct and indirect effects for a range of plausible correlation values. We take the sampling variability into account through the construction of uncertainty intervals. The proposed method is able to assess sensitivity to both mediator‐outcome confounding and confounding involving the exposure. To illustrate the method, we apply it to a mediation study based on the data from the Swedish Stroke Register (Riksstroke). An R package that implements the proposed method is available.</abstract><cop>England</cop><pub>Wiley Subscription Services, Inc</pub><pmid>29462839</pmid><doi>10.1002/sim.7620</doi><tpages>19</tpages><orcidid>https://orcid.org/0000-0003-3298-1555</orcidid><orcidid>https://orcid.org/0000-0003-3187-1987</orcidid><orcidid>https://orcid.org/0000-0002-4600-0060</orcidid><oa>free_for_read</oa></addata></record>
fulltext fulltext
identifier ISSN: 0277-6715
ispartof Statistics in medicine, 2018-05, Vol.37 (10), p.1744-1762
issn 0277-6715
1097-0258
1097-0258
language eng
recordid cdi_swepub_primary_oai_DiVA_org_umu_125929
source Wiley-Blackwell Read & Publish Collection
subjects direct effects
indirect effects
mediation
Sensitivity analysis
sequential ignorability
Statistics
statistik
unmeasured confounding
title Sensitivity analysis for unobserved confounding of direct and indirect effects using uncertainty intervals
url http://sfxeu10.hosted.exlibrisgroup.com/loughborough?ctx_ver=Z39.88-2004&ctx_enc=info:ofi/enc:UTF-8&ctx_tim=2024-12-24T02%3A36%3A28IST&url_ver=Z39.88-2004&url_ctx_fmt=infofi/fmt:kev:mtx:ctx&rfr_id=info:sid/primo.exlibrisgroup.com:primo3-Article-proquest_swepu&rft_val_fmt=info:ofi/fmt:kev:mtx:journal&rft.genre=article&rft.atitle=Sensitivity%20analysis%20for%20unobserved%20confounding%20of%20direct%20and%20indirect%20effects%20using%20uncertainty%20intervals&rft.jtitle=Statistics%20in%20medicine&rft.au=Lindmark,%20Anita&rft.date=2018-05-10&rft.volume=37&rft.issue=10&rft.spage=1744&rft.epage=1762&rft.pages=1744-1762&rft.issn=0277-6715&rft.eissn=1097-0258&rft_id=info:doi/10.1002/sim.7620&rft_dat=%3Cproquest_swepu%3E2007121177%3C/proquest_swepu%3E%3Cgrp_id%3Ecdi_FETCH-LOGICAL-c3870-7910b408a42c2b21da6e6c2c85c1a2bd2f4bea56ddc271eb8fefdf4655bdb5d3%3C/grp_id%3E%3Coa%3E%3C/oa%3E%3Curl%3E%3C/url%3E&rft_id=info:oai/&rft_pqid=2023816640&rft_id=info:pmid/29462839&rfr_iscdi=true