Greedy caliper propensity score matching can yield variable estimates of the treatment‐outcome association—A simulation study
Purpose Greedy caliper propensity score (PS) matching is dependent on randomness, which can ultimately affect causal estimates. We sought to investigate the variation introduced by this randomness. Methods Based on a literature search to define the simulation parameters, we simulated 36 cohorts of d...
Gespeichert in:
Veröffentlicht in: | Pharmacoepidemiology and drug safety 2021-07, Vol.30 (7), p.934-951 |
---|---|
Hauptverfasser: | , , , , , , , |
Format: | Artikel |
Sprache: | eng |
Schlagworte: | |
Online-Zugang: | Volltext |
Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
container_end_page | 951 |
---|---|
container_issue | 7 |
container_start_page | 934 |
container_title | Pharmacoepidemiology and drug safety |
container_volume | 30 |
creator | Komen, Joris J. Belitser, Svetlana V. Wyss, Richard Schneeweiss, Sebastian Taams, Anne C. Pajouheshnia, Romin Forslund, Tomas Klungel, Olaf H. |
description | Purpose
Greedy caliper propensity score (PS) matching is dependent on randomness, which can ultimately affect causal estimates. We sought to investigate the variation introduced by this randomness.
Methods
Based on a literature search to define the simulation parameters, we simulated 36 cohorts of different sizes, treatment prevalence, outcome prevalence, treatment‐outcome‐association. We performed 1:1 caliper and nearest neighbor (NN) caliper PS‐matching and repeated this 1000 times in the same cohort, before calculating the treatment‐outcome association.
Results
Repeating caliper and NN caliper matching in the same cohort yielded large variations in effect estimates, in all 36 scenarios, with both types of matching. The largest variation was found in smaller cohorts, where the odds ratio (OR) ranged from 0.53 to 10.00 (IQR of ORs: 1.11‐1.67). The 95% confidence interval was not consistently overlapping a neutral association after repeating the matching with both algorithms. We confirmed these findings in a noninterventional example study.
Conclusion
Caliper PS‐matching can yield highly variable estimates of the treatment‐outcome association if the analysis is repeated. |
doi_str_mv | 10.1002/pds.5232 |
format | Article |
fullrecord | <record><control><sourceid>proquest_cross</sourceid><recordid>TN_cdi_crossref_primary_10_1002_pds_5232</recordid><sourceformat>XML</sourceformat><sourcesystem>PC</sourcesystem><sourcerecordid>2502805167</sourcerecordid><originalsourceid>FETCH-LOGICAL-c5262-7980b463043bedfcad94da273c3814393ed32e2726c6440bdd7cf4bc9795f8a53</originalsourceid><addsrcrecordid>eNqNks1u1DAUhSMEoqUg8QTIEptKKMXxT342SNUABakSSMDacuybjksSB9vpKLvyBix4wj4Jzsx0hiJVYpUb3-9eHx2fJHme4ZMMY_J60P6EE0oeJIcZrqo047x4ONecpiXPq4PkifeXGMdexR4nB5QWlHJKD5OfZw5AT0jJ1gzg0ODsAL03YUJeWQeok0EtTX8RiR5NBlqNrqQzsm4BgQ8m9sEj26CwBBQcyNBBH26uf9kxKNsBkt5bZWQwtr-5_n2KvOnGdv2LfBj19DR51MjWw7Pt9yj59v7d18WH9PzT2cfF6XmqOMlJWlQlrllOMaM16EZJXTEtSUEVLTNGKwqaEiAFyVXOGK61LlTDalUVFW9KyelRkm72-hUMYy0GF7W7SVhpxPboe6xAxEtYlUW-upePLun90O1gxvKsyIuyiLNvNrMR6ECr6IiT7d0Vdzq9WYoLeyVKwrOSzWKPtwuc_TFGn0VnvIK2lT3Y0QvCMSkxz_L5rpf_oJd2dH20MlKU59G8stwvVM5676DZicmwmDMkYobEnKGIvvhb_A68DU0EXm2AFdS28cpAr2CHxZTlcQtneazw7GP5__TChHU0Fnbsw_7JVqaF6V7F4vPbL2vlfwAgEve2</addsrcrecordid><sourcetype>Open Access Repository</sourcetype><iscdi>true</iscdi><recordtype>article</recordtype><pqid>2535626288</pqid></control><display><type>article</type><title>Greedy caliper propensity score matching can yield variable estimates of the treatment‐outcome association—A simulation study</title><source>SWEPUB Freely available online</source><source>Access via Wiley Online Library</source><source>Web of Science - Science Citation Index Expanded - 2021<img src="https://exlibris-pub.s3.amazonaws.com/fromwos-v2.jpg" /></source><creator>Komen, Joris J. ; Belitser, Svetlana V. ; Wyss, Richard ; Schneeweiss, Sebastian ; Taams, Anne C. ; Pajouheshnia, Romin ; Forslund, Tomas ; Klungel, Olaf H.</creator><creatorcontrib>Komen, Joris J. ; Belitser, Svetlana V. ; Wyss, Richard ; Schneeweiss, Sebastian ; Taams, Anne C. ; Pajouheshnia, Romin ; Forslund, Tomas ; Klungel, Olaf H.</creatorcontrib><description>Purpose
Greedy caliper propensity score (PS) matching is dependent on randomness, which can ultimately affect causal estimates. We sought to investigate the variation introduced by this randomness.
Methods
Based on a literature search to define the simulation parameters, we simulated 36 cohorts of different sizes, treatment prevalence, outcome prevalence, treatment‐outcome‐association. We performed 1:1 caliper and nearest neighbor (NN) caliper PS‐matching and repeated this 1000 times in the same cohort, before calculating the treatment‐outcome association.
Results
Repeating caliper and NN caliper matching in the same cohort yielded large variations in effect estimates, in all 36 scenarios, with both types of matching. The largest variation was found in smaller cohorts, where the odds ratio (OR) ranged from 0.53 to 10.00 (IQR of ORs: 1.11‐1.67). The 95% confidence interval was not consistently overlapping a neutral association after repeating the matching with both algorithms. We confirmed these findings in a noninterventional example study.
Conclusion
Caliper PS‐matching can yield highly variable estimates of the treatment‐outcome association if the analysis is repeated.</description><identifier>ISSN: 1053-8569</identifier><identifier>ISSN: 1099-1557</identifier><identifier>EISSN: 1099-1557</identifier><identifier>DOI: 10.1002/pds.5232</identifier><identifier>PMID: 33733533</identifier><language>eng</language><publisher>Chichester, UK: John Wiley & Sons, Inc</publisher><subject>caliper matching ; greedy ; Life Sciences & Biomedicine ; matching ; Medicin och hälsovetenskap ; nearest neighbor matching ; Original ; pharmacoepidemiology ; Pharmacology & Pharmacy ; propensity score ; Public, Environmental & Occupational Health ; Science & Technology ; simulation ; Variation</subject><ispartof>Pharmacoepidemiology and drug safety, 2021-07, Vol.30 (7), p.934-951</ispartof><rights>2021 The Authors. published by John Wiley & Sons Ltd.</rights><rights>2021 The Authors. Pharmacoepidemiology and Drug Safety published by John Wiley & Sons Ltd.</rights><rights>2021. This article is published under http://creativecommons.org/licenses/by/4.0/ (the “License”). Notwithstanding the ProQuest Terms and Conditions, you may use this content in accordance with the terms of the License.</rights><lds50>peer_reviewed</lds50><oa>free_for_read</oa><woscitedreferencessubscribed>true</woscitedreferencessubscribed><woscitedreferencescount>9</woscitedreferencescount><woscitedreferencesoriginalsourcerecordid>wos000632354600001</woscitedreferencesoriginalsourcerecordid><citedby>FETCH-LOGICAL-c5262-7980b463043bedfcad94da273c3814393ed32e2726c6440bdd7cf4bc9795f8a53</citedby><cites>FETCH-LOGICAL-c5262-7980b463043bedfcad94da273c3814393ed32e2726c6440bdd7cf4bc9795f8a53</cites><orcidid>0000-0002-4208-3583 ; 0000-0002-5471-3132 ; 0000-0003-2575-467X</orcidid></display><links><openurl>$$Topenurl_article</openurl><openurlfulltext>$$Topenurlfull_article</openurlfulltext><thumbnail>$$Tsyndetics_thumb_exl</thumbnail><linktopdf>$$Uhttps://onlinelibrary.wiley.com/doi/pdf/10.1002%2Fpds.5232$$EPDF$$P50$$Gwiley$$Hfree_for_read</linktopdf><linktohtml>$$Uhttps://onlinelibrary.wiley.com/doi/full/10.1002%2Fpds.5232$$EHTML$$P50$$Gwiley$$Hfree_for_read</linktohtml><link.rule.ids>230,315,554,782,786,887,1419,27931,27932,39265,45581,45582</link.rule.ids><backlink>$$Uhttps://www.ncbi.nlm.nih.gov/pubmed/33733533$$D View this record in MEDLINE/PubMed$$Hfree_for_read</backlink><backlink>$$Uhttp://kipublications.ki.se/Default.aspx?queryparsed=id:146176787$$DView record from Swedish Publication Index$$Hfree_for_read</backlink></links><search><creatorcontrib>Komen, Joris J.</creatorcontrib><creatorcontrib>Belitser, Svetlana V.</creatorcontrib><creatorcontrib>Wyss, Richard</creatorcontrib><creatorcontrib>Schneeweiss, Sebastian</creatorcontrib><creatorcontrib>Taams, Anne C.</creatorcontrib><creatorcontrib>Pajouheshnia, Romin</creatorcontrib><creatorcontrib>Forslund, Tomas</creatorcontrib><creatorcontrib>Klungel, Olaf H.</creatorcontrib><title>Greedy caliper propensity score matching can yield variable estimates of the treatment‐outcome association—A simulation study</title><title>Pharmacoepidemiology and drug safety</title><addtitle>PHARMACOEPIDEM DR S</addtitle><addtitle>Pharmacoepidemiol Drug Saf</addtitle><description>Purpose
Greedy caliper propensity score (PS) matching is dependent on randomness, which can ultimately affect causal estimates. We sought to investigate the variation introduced by this randomness.
Methods
Based on a literature search to define the simulation parameters, we simulated 36 cohorts of different sizes, treatment prevalence, outcome prevalence, treatment‐outcome‐association. We performed 1:1 caliper and nearest neighbor (NN) caliper PS‐matching and repeated this 1000 times in the same cohort, before calculating the treatment‐outcome association.
Results
Repeating caliper and NN caliper matching in the same cohort yielded large variations in effect estimates, in all 36 scenarios, with both types of matching. The largest variation was found in smaller cohorts, where the odds ratio (OR) ranged from 0.53 to 10.00 (IQR of ORs: 1.11‐1.67). The 95% confidence interval was not consistently overlapping a neutral association after repeating the matching with both algorithms. We confirmed these findings in a noninterventional example study.
Conclusion
Caliper PS‐matching can yield highly variable estimates of the treatment‐outcome association if the analysis is repeated.</description><subject>caliper matching</subject><subject>greedy</subject><subject>Life Sciences & Biomedicine</subject><subject>matching</subject><subject>Medicin och hälsovetenskap</subject><subject>nearest neighbor matching</subject><subject>Original</subject><subject>pharmacoepidemiology</subject><subject>Pharmacology & Pharmacy</subject><subject>propensity score</subject><subject>Public, Environmental & Occupational Health</subject><subject>Science & Technology</subject><subject>simulation</subject><subject>Variation</subject><issn>1053-8569</issn><issn>1099-1557</issn><issn>1099-1557</issn><fulltext>true</fulltext><rsrctype>article</rsrctype><creationdate>2021</creationdate><recordtype>article</recordtype><sourceid>24P</sourceid><sourceid>WIN</sourceid><sourceid>HGBXW</sourceid><sourceid>D8T</sourceid><recordid>eNqNks1u1DAUhSMEoqUg8QTIEptKKMXxT342SNUABakSSMDacuybjksSB9vpKLvyBix4wj4Jzsx0hiJVYpUb3-9eHx2fJHme4ZMMY_J60P6EE0oeJIcZrqo047x4ONecpiXPq4PkifeXGMdexR4nB5QWlHJKD5OfZw5AT0jJ1gzg0ODsAL03YUJeWQeok0EtTX8RiR5NBlqNrqQzsm4BgQ8m9sEj26CwBBQcyNBBH26uf9kxKNsBkt5bZWQwtr-5_n2KvOnGdv2LfBj19DR51MjWw7Pt9yj59v7d18WH9PzT2cfF6XmqOMlJWlQlrllOMaM16EZJXTEtSUEVLTNGKwqaEiAFyVXOGK61LlTDalUVFW9KyelRkm72-hUMYy0GF7W7SVhpxPboe6xAxEtYlUW-upePLun90O1gxvKsyIuyiLNvNrMR6ECr6IiT7d0Vdzq9WYoLeyVKwrOSzWKPtwuc_TFGn0VnvIK2lT3Y0QvCMSkxz_L5rpf_oJd2dH20MlKU59G8stwvVM5676DZicmwmDMkYobEnKGIvvhb_A68DU0EXm2AFdS28cpAr2CHxZTlcQtneazw7GP5__TChHU0Fnbsw_7JVqaF6V7F4vPbL2vlfwAgEve2</recordid><startdate>202107</startdate><enddate>202107</enddate><creator>Komen, Joris J.</creator><creator>Belitser, Svetlana V.</creator><creator>Wyss, Richard</creator><creator>Schneeweiss, Sebastian</creator><creator>Taams, Anne C.</creator><creator>Pajouheshnia, Romin</creator><creator>Forslund, Tomas</creator><creator>Klungel, Olaf H.</creator><general>John Wiley & Sons, Inc</general><general>Wiley</general><general>Wiley Subscription Services, Inc</general><scope>24P</scope><scope>WIN</scope><scope>BLEPL</scope><scope>DTL</scope><scope>HGBXW</scope><scope>NPM</scope><scope>AAYXX</scope><scope>CITATION</scope><scope>7TK</scope><scope>K9.</scope><scope>7X8</scope><scope>5PM</scope><scope>ADTPV</scope><scope>AOWAS</scope><scope>D8T</scope><scope>ZZAVC</scope><orcidid>https://orcid.org/0000-0002-4208-3583</orcidid><orcidid>https://orcid.org/0000-0002-5471-3132</orcidid><orcidid>https://orcid.org/0000-0003-2575-467X</orcidid></search><sort><creationdate>202107</creationdate><title>Greedy caliper propensity score matching can yield variable estimates of the treatment‐outcome association—A simulation study</title><author>Komen, Joris J. ; Belitser, Svetlana V. ; Wyss, Richard ; Schneeweiss, Sebastian ; Taams, Anne C. ; Pajouheshnia, Romin ; Forslund, Tomas ; Klungel, Olaf H.</author></sort><facets><frbrtype>5</frbrtype><frbrgroupid>cdi_FETCH-LOGICAL-c5262-7980b463043bedfcad94da273c3814393ed32e2726c6440bdd7cf4bc9795f8a53</frbrgroupid><rsrctype>articles</rsrctype><prefilter>articles</prefilter><language>eng</language><creationdate>2021</creationdate><topic>caliper matching</topic><topic>greedy</topic><topic>Life Sciences & Biomedicine</topic><topic>matching</topic><topic>Medicin och hälsovetenskap</topic><topic>nearest neighbor matching</topic><topic>Original</topic><topic>pharmacoepidemiology</topic><topic>Pharmacology & Pharmacy</topic><topic>propensity score</topic><topic>Public, Environmental & Occupational Health</topic><topic>Science & Technology</topic><topic>simulation</topic><topic>Variation</topic><toplevel>peer_reviewed</toplevel><toplevel>online_resources</toplevel><creatorcontrib>Komen, Joris J.</creatorcontrib><creatorcontrib>Belitser, Svetlana V.</creatorcontrib><creatorcontrib>Wyss, Richard</creatorcontrib><creatorcontrib>Schneeweiss, Sebastian</creatorcontrib><creatorcontrib>Taams, Anne C.</creatorcontrib><creatorcontrib>Pajouheshnia, Romin</creatorcontrib><creatorcontrib>Forslund, Tomas</creatorcontrib><creatorcontrib>Klungel, Olaf H.</creatorcontrib><collection>Wiley Online Library (Open Access Collection)</collection><collection>Wiley Online Library (Open Access Collection)</collection><collection>Web of Science Core Collection</collection><collection>Science Citation Index Expanded</collection><collection>Web of Science - Science Citation Index Expanded - 2021</collection><collection>PubMed</collection><collection>CrossRef</collection><collection>Neurosciences Abstracts</collection><collection>ProQuest Health & Medical Complete (Alumni)</collection><collection>MEDLINE - Academic</collection><collection>PubMed Central (Full Participant titles)</collection><collection>SwePub</collection><collection>SwePub Articles</collection><collection>SWEPUB Freely available online</collection><collection>SwePub Articles full text</collection><jtitle>Pharmacoepidemiology and drug safety</jtitle></facets><delivery><delcategory>Remote Search Resource</delcategory><fulltext>fulltext</fulltext></delivery><addata><au>Komen, Joris J.</au><au>Belitser, Svetlana V.</au><au>Wyss, Richard</au><au>Schneeweiss, Sebastian</au><au>Taams, Anne C.</au><au>Pajouheshnia, Romin</au><au>Forslund, Tomas</au><au>Klungel, Olaf H.</au><format>journal</format><genre>article</genre><ristype>JOUR</ristype><atitle>Greedy caliper propensity score matching can yield variable estimates of the treatment‐outcome association—A simulation study</atitle><jtitle>Pharmacoepidemiology and drug safety</jtitle><stitle>PHARMACOEPIDEM DR S</stitle><addtitle>Pharmacoepidemiol Drug Saf</addtitle><date>2021-07</date><risdate>2021</risdate><volume>30</volume><issue>7</issue><spage>934</spage><epage>951</epage><pages>934-951</pages><issn>1053-8569</issn><issn>1099-1557</issn><eissn>1099-1557</eissn><abstract>Purpose
Greedy caliper propensity score (PS) matching is dependent on randomness, which can ultimately affect causal estimates. We sought to investigate the variation introduced by this randomness.
Methods
Based on a literature search to define the simulation parameters, we simulated 36 cohorts of different sizes, treatment prevalence, outcome prevalence, treatment‐outcome‐association. We performed 1:1 caliper and nearest neighbor (NN) caliper PS‐matching and repeated this 1000 times in the same cohort, before calculating the treatment‐outcome association.
Results
Repeating caliper and NN caliper matching in the same cohort yielded large variations in effect estimates, in all 36 scenarios, with both types of matching. The largest variation was found in smaller cohorts, where the odds ratio (OR) ranged from 0.53 to 10.00 (IQR of ORs: 1.11‐1.67). The 95% confidence interval was not consistently overlapping a neutral association after repeating the matching with both algorithms. We confirmed these findings in a noninterventional example study.
Conclusion
Caliper PS‐matching can yield highly variable estimates of the treatment‐outcome association if the analysis is repeated.</abstract><cop>Chichester, UK</cop><pub>John Wiley & Sons, Inc</pub><pmid>33733533</pmid><doi>10.1002/pds.5232</doi><tpages>18</tpages><orcidid>https://orcid.org/0000-0002-4208-3583</orcidid><orcidid>https://orcid.org/0000-0002-5471-3132</orcidid><orcidid>https://orcid.org/0000-0003-2575-467X</orcidid><oa>free_for_read</oa></addata></record> |
fulltext | fulltext |
identifier | ISSN: 1053-8569 |
ispartof | Pharmacoepidemiology and drug safety, 2021-07, Vol.30 (7), p.934-951 |
issn | 1053-8569 1099-1557 1099-1557 |
language | eng |
recordid | cdi_crossref_primary_10_1002_pds_5232 |
source | SWEPUB Freely available online; Access via Wiley Online Library; Web of Science - Science Citation Index Expanded - 2021<img src="https://exlibris-pub.s3.amazonaws.com/fromwos-v2.jpg" /> |
subjects | caliper matching greedy Life Sciences & Biomedicine matching Medicin och hälsovetenskap nearest neighbor matching Original pharmacoepidemiology Pharmacology & Pharmacy propensity score Public, Environmental & Occupational Health Science & Technology simulation Variation |
title | Greedy caliper propensity score matching can yield variable estimates of the treatment‐outcome association—A simulation study |
url | https://sfx.bib-bvb.de/sfx_tum?ctx_ver=Z39.88-2004&ctx_enc=info:ofi/enc:UTF-8&ctx_tim=2024-12-05T19%3A32%3A35IST&url_ver=Z39.88-2004&url_ctx_fmt=infofi/fmt:kev:mtx:ctx&rfr_id=info:sid/primo.exlibrisgroup.com:primo3-Article-proquest_cross&rft_val_fmt=info:ofi/fmt:kev:mtx:journal&rft.genre=article&rft.atitle=Greedy%20caliper%20propensity%20score%20matching%20can%20yield%20variable%20estimates%20of%20the%20treatment%E2%80%90outcome%20association%E2%80%94A%20simulation%20study&rft.jtitle=Pharmacoepidemiology%20and%20drug%20safety&rft.au=Komen,%20Joris%20J.&rft.date=2021-07&rft.volume=30&rft.issue=7&rft.spage=934&rft.epage=951&rft.pages=934-951&rft.issn=1053-8569&rft.eissn=1099-1557&rft_id=info:doi/10.1002/pds.5232&rft_dat=%3Cproquest_cross%3E2502805167%3C/proquest_cross%3E%3Curl%3E%3C/url%3E&disable_directlink=true&sfx.directlink=off&sfx.report_link=0&rft_id=info:oai/&rft_pqid=2535626288&rft_id=info:pmid/33733533&rfr_iscdi=true |