Evaluation of the Propensity score methods for estimating marginal odds ratios in case of small sample size View Full Text


Ontology type: schema:ScholarlyArticle      Open Access: True


Article Info

DATE

2012-05-30

AUTHORS

Romain Pirracchio, Matthieu Resche-Rigon, Sylvie Chevret

ABSTRACT

BackgroundPropensity score (PS) methods are increasingly used, even when sample sizes are small or treatments are seldom used. However, the relative performance of the two mainly recommended PS methods, namely PS-matching or inverse probability of treatment weighting (IPTW), have not been studied in the context of small sample sizes.MethodsWe conducted a series of Monte Carlo simulations to evaluate the influence of sample size, prevalence of treatment exposure, and strength of the association between the variables and the outcome and/or the treatment exposure, on the performance of these two methods.ResultsDecreasing the sample size from 1,000 to 40 subjects did not substantially alter the Type I error rate, and led to relative biases below 10%. The IPTW method performed better than the PS-matching down to 60 subjects. When N was set at 40, the PS matching estimators were either similarly or even less biased than the IPTW estimators. Including variables unrelated to the exposure but related to the outcome in the PS model decreased the bias and the variance as compared to models omitting such variables. Excluding the true confounder from the PS model resulted, whatever the method used, in a significantly biased estimation of treatment effect. These results were illustrated in a real dataset.ConclusionEven in case of small study samples or low prevalence of treatment, PS-matching and IPTW can yield correct estimations of treatment effect unless the true confounders and the variables related only to the outcome are not included in the PS model. More... »

PAGES

70

Identifiers

URI

http://scigraph.springernature.com/pub.10.1186/1471-2288-12-70

DOI

http://dx.doi.org/10.1186/1471-2288-12-70

DIMENSIONS

https://app.dimensions.ai/details/publication/pub.1005384649

PUBMED

https://www.ncbi.nlm.nih.gov/pubmed/22646911


Indexing Status Check whether this publication has been indexed by Scopus and Web Of Science using the SN Indexing Status Tool
Incoming Citations Browse incoming citations for this publication using opencitations.net

JSON-LD is the canonical representation for SciGraph data.

TIP: You can open this SciGraph record using an external JSON-LD service: JSON-LD Playground Google SDTT

[
  {
    "@context": "https://springernature.github.io/scigraph/jsonld/sgcontext.json", 
    "about": [
      {
        "id": "http://purl.org/au-research/vocabulary/anzsrc-for/2008/11", 
        "inDefinedTermSet": "http://purl.org/au-research/vocabulary/anzsrc-for/2008/", 
        "name": "Medical and Health Sciences", 
        "type": "DefinedTerm"
      }, 
      {
        "id": "http://purl.org/au-research/vocabulary/anzsrc-for/2008/1117", 
        "inDefinedTermSet": "http://purl.org/au-research/vocabulary/anzsrc-for/2008/", 
        "name": "Public Health and Health Services", 
        "type": "DefinedTerm"
      }, 
      {
        "inDefinedTermSet": "https://www.nlm.nih.gov/mesh/", 
        "name": "Humans", 
        "type": "DefinedTerm"
      }, 
      {
        "inDefinedTermSet": "https://www.nlm.nih.gov/mesh/", 
        "name": "Monte Carlo Method", 
        "type": "DefinedTerm"
      }, 
      {
        "inDefinedTermSet": "https://www.nlm.nih.gov/mesh/", 
        "name": "Odds Ratio", 
        "type": "DefinedTerm"
      }, 
      {
        "inDefinedTermSet": "https://www.nlm.nih.gov/mesh/", 
        "name": "Propensity Score", 
        "type": "DefinedTerm"
      }, 
      {
        "inDefinedTermSet": "https://www.nlm.nih.gov/mesh/", 
        "name": "Sample Size", 
        "type": "DefinedTerm"
      }
    ], 
    "author": [
      {
        "affiliation": {
          "alternateName": "3- Service d'Anesth\u00e9sie-R\u00e9animation, H\u00f4pital Europ\u00e9en Georges Pompidou, Universit\u00e9 Paris 5 Descartes, Sorbonne Paris Cit\u00e9, Paris, France", 
          "id": "http://www.grid.ac/institutes/grid.508487.6", 
          "name": [
            "Service de Biostatistique et Information M\u00e9dicale, H\u00f4pital Saint-Louis, UMR-S717 Inserm; Sorbonne Paris Cit\u00e9, Universit\u00e9 Paris Diderot, 1 avenue Claude Vellefaux, 75010, Paris, France", 
            "3- Service d'Anesth\u00e9sie-R\u00e9animation, H\u00f4pital Europ\u00e9en Georges Pompidou, Universit\u00e9 Paris 5 Descartes, Sorbonne Paris Cit\u00e9, Paris, France"
          ], 
          "type": "Organization"
        }, 
        "familyName": "Pirracchio", 
        "givenName": "Romain", 
        "id": "sg:person.01047623565.27", 
        "sameAs": [
          "https://app.dimensions.ai/discover/publication?and_facet_researcher=ur.01047623565.27"
        ], 
        "type": "Person"
      }, 
      {
        "affiliation": {
          "alternateName": "Service de Biostatistique et Information M\u00e9dicale, H\u00f4pital Saint-Louis, UMR-S717 Inserm; Sorbonne Paris Cit\u00e9, Universit\u00e9 Paris Diderot, 1 avenue Claude Vellefaux, 75010, Paris, France", 
          "id": "http://www.grid.ac/institutes/None", 
          "name": [
            "Service de Biostatistique et Information M\u00e9dicale, H\u00f4pital Saint-Louis, UMR-S717 Inserm; Sorbonne Paris Cit\u00e9, Universit\u00e9 Paris Diderot, 1 avenue Claude Vellefaux, 75010, Paris, France"
          ], 
          "type": "Organization"
        }, 
        "familyName": "Resche-Rigon", 
        "givenName": "Matthieu", 
        "id": "sg:person.01036155166.50", 
        "sameAs": [
          "https://app.dimensions.ai/discover/publication?and_facet_researcher=ur.01036155166.50"
        ], 
        "type": "Person"
      }, 
      {
        "affiliation": {
          "alternateName": "Service de Biostatistique et Information M\u00e9dicale, H\u00f4pital Saint-Louis, UMR-S717 Inserm; Sorbonne Paris Cit\u00e9, Universit\u00e9 Paris Diderot, 1 avenue Claude Vellefaux, 75010, Paris, France", 
          "id": "http://www.grid.ac/institutes/None", 
          "name": [
            "Service de Biostatistique et Information M\u00e9dicale, H\u00f4pital Saint-Louis, UMR-S717 Inserm; Sorbonne Paris Cit\u00e9, Universit\u00e9 Paris Diderot, 1 avenue Claude Vellefaux, 75010, Paris, France"
          ], 
          "type": "Organization"
        }, 
        "familyName": "Chevret", 
        "givenName": "Sylvie", 
        "id": "sg:person.0625147223.89", 
        "sameAs": [
          "https://app.dimensions.ai/discover/publication?and_facet_researcher=ur.0625147223.89"
        ], 
        "type": "Person"
      }
    ], 
    "citation": [
      {
        "id": "sg:pub.10.1007/s00134-010-1933-2", 
        "sameAs": [
          "https://app.dimensions.ai/details/publication/pub.1037710572", 
          "https://doi.org/10.1007/s00134-010-1933-2"
        ], 
        "type": "CreativeWork"
      }, 
      {
        "id": "sg:pub.10.1038/bmt.2010.90", 
        "sameAs": [
          "https://app.dimensions.ai/details/publication/pub.1042034040", 
          "https://doi.org/10.1038/bmt.2010.90"
        ], 
        "type": "CreativeWork"
      }, 
      {
        "id": "sg:pub.10.1007/978-1-4757-3692-2", 
        "sameAs": [
          "https://app.dimensions.ai/details/publication/pub.1013392911", 
          "https://doi.org/10.1007/978-1-4757-3692-2"
        ], 
        "type": "CreativeWork"
      }, 
      {
        "id": "sg:pub.10.1007/s00134-010-1991-5", 
        "sameAs": [
          "https://app.dimensions.ai/details/publication/pub.1005546328", 
          "https://doi.org/10.1007/s00134-010-1991-5"
        ], 
        "type": "CreativeWork"
      }
    ], 
    "datePublished": "2012-05-30", 
    "datePublishedReg": "2012-05-30", 
    "description": "BackgroundPropensity score (PS) methods are increasingly used, even when sample sizes are small or treatments are seldom used. However, the relative performance of the two mainly recommended PS methods, namely PS-matching or inverse probability of treatment weighting (IPTW), have not been studied in the context of small sample sizes.MethodsWe conducted a series of Monte Carlo simulations to evaluate the influence of sample size, prevalence of treatment exposure, and strength of the association between the variables and the outcome and/or the treatment exposure, on the performance of these two methods.ResultsDecreasing the sample size from 1,000 to 40 subjects did not substantially alter the Type I error rate, and led to relative biases below 10%. The IPTW method performed better than the PS-matching down to 60 subjects. When N was set at 40, the PS matching estimators were either similarly or even less biased than the IPTW estimators. Including variables unrelated to the exposure but related to the outcome in the PS model decreased the bias and the variance as compared to models omitting such variables. Excluding the true confounder from the PS model resulted, whatever the method used, in a significantly biased estimation of treatment effect. These results were illustrated in a real dataset.ConclusionEven in case of small study samples or low prevalence of treatment, PS-matching and IPTW can yield correct estimations of treatment effect unless the true confounders and the variables related only to the outcome are not included in the PS model.", 
    "genre": "article", 
    "id": "sg:pub.10.1186/1471-2288-12-70", 
    "isAccessibleForFree": true, 
    "isPartOf": [
      {
        "id": "sg:journal.1024940", 
        "issn": [
          "1471-2288"
        ], 
        "name": "BMC Medical Research Methodology", 
        "publisher": "Springer Nature", 
        "type": "Periodical"
      }, 
      {
        "issueNumber": "1", 
        "type": "PublicationIssue"
      }, 
      {
        "type": "PublicationVolume", 
        "volumeNumber": "12"
      }
    ], 
    "keywords": [
      "treatment exposure", 
      "small sample size", 
      "treatment effects", 
      "small study sample", 
      "sample size", 
      "treatment weighting", 
      "PS matching", 
      "odds ratio", 
      "IPTW method", 
      "low prevalence", 
      "propensity score methods", 
      "score method", 
      "inverse probability", 
      "study sample", 
      "true confounders", 
      "confounders", 
      "outcomes", 
      "prevalence", 
      "exposure", 
      "treatment", 
      "subjects", 
      "marginal odds ratio", 
      "ConclusionEven", 
      "IPTW", 
      "MethodsWe", 
      "PS model", 
      "cases", 
      "association", 
      "effect", 
      "variables", 
      "evaluation", 
      "rate", 
      "IPTW estimators", 
      "method", 
      "size", 
      "such variables", 
      "samples", 
      "model", 
      "bias", 
      "ratio", 
      "series", 
      "variance", 
      "results", 
      "probability", 
      "PS", 
      "influence", 
      "weighting", 
      "PS method", 
      "relative biases", 
      "Type I error rates", 
      "matching estimators", 
      "I error rate", 
      "correct estimation", 
      "biased estimation", 
      "context", 
      "relative performance", 
      "strength", 
      "biases", 
      "performance", 
      "estimation", 
      "estimator", 
      "dataset", 
      "Monte Carlo simulations", 
      "error rate", 
      "Carlo simulations", 
      "real datasets", 
      "simulations"
    ], 
    "name": "Evaluation of the Propensity score methods for estimating marginal odds ratios in case of small sample size", 
    "pagination": "70", 
    "productId": [
      {
        "name": "dimensions_id", 
        "type": "PropertyValue", 
        "value": [
          "pub.1005384649"
        ]
      }, 
      {
        "name": "doi", 
        "type": "PropertyValue", 
        "value": [
          "10.1186/1471-2288-12-70"
        ]
      }, 
      {
        "name": "pubmed_id", 
        "type": "PropertyValue", 
        "value": [
          "22646911"
        ]
      }
    ], 
    "sameAs": [
      "https://doi.org/10.1186/1471-2288-12-70", 
      "https://app.dimensions.ai/details/publication/pub.1005384649"
    ], 
    "sdDataset": "articles", 
    "sdDatePublished": "2022-12-01T06:29", 
    "sdLicense": "https://scigraph.springernature.com/explorer/license/", 
    "sdPublisher": {
      "name": "Springer Nature - SN SciGraph project", 
      "type": "Organization"
    }, 
    "sdSource": "s3://com-springernature-scigraph/baseset/20221201/entities/gbq_results/article/article_558.jsonl", 
    "type": "ScholarlyArticle", 
    "url": "https://doi.org/10.1186/1471-2288-12-70"
  }
]
 

Download the RDF metadata as:  json-ld nt turtle xml License info

HOW TO GET THIS DATA PROGRAMMATICALLY:

JSON-LD is a popular format for linked data which is fully compatible with JSON.

curl -H 'Accept: application/ld+json' 'https://scigraph.springernature.com/pub.10.1186/1471-2288-12-70'

N-Triples is a line-based linked data format ideal for batch operations.

curl -H 'Accept: application/n-triples' 'https://scigraph.springernature.com/pub.10.1186/1471-2288-12-70'

Turtle is a human-readable linked data format.

curl -H 'Accept: text/turtle' 'https://scigraph.springernature.com/pub.10.1186/1471-2288-12-70'

RDF/XML is a standard XML format for linked data.

curl -H 'Accept: application/rdf+xml' 'https://scigraph.springernature.com/pub.10.1186/1471-2288-12-70'


 

This table displays all metadata directly associated to this object as RDF triples.

181 TRIPLES      21 PREDICATES      101 URIs      89 LITERALS      12 BLANK NODES

Subject Predicate Object
1 sg:pub.10.1186/1471-2288-12-70 schema:about N490819bd54264e69bc42df33c99a6b4f
2 N595bf4a392384102a5fd89a03bb6a104
3 Nb0aeca64b65d42119cfb0ec4a8a0d62a
4 Nb4cd181848344b3db3ed418889e1d8b9
5 Nc0262253cab144db9f8afb937a377e22
6 anzsrc-for:11
7 anzsrc-for:1117
8 schema:author N98c00cd8acfe44dfb339d401a51c9d36
9 schema:citation sg:pub.10.1007/978-1-4757-3692-2
10 sg:pub.10.1007/s00134-010-1933-2
11 sg:pub.10.1007/s00134-010-1991-5
12 sg:pub.10.1038/bmt.2010.90
13 schema:datePublished 2012-05-30
14 schema:datePublishedReg 2012-05-30
15 schema:description BackgroundPropensity score (PS) methods are increasingly used, even when sample sizes are small or treatments are seldom used. However, the relative performance of the two mainly recommended PS methods, namely PS-matching or inverse probability of treatment weighting (IPTW), have not been studied in the context of small sample sizes.MethodsWe conducted a series of Monte Carlo simulations to evaluate the influence of sample size, prevalence of treatment exposure, and strength of the association between the variables and the outcome and/or the treatment exposure, on the performance of these two methods.ResultsDecreasing the sample size from 1,000 to 40 subjects did not substantially alter the Type I error rate, and led to relative biases below 10%. The IPTW method performed better than the PS-matching down to 60 subjects. When N was set at 40, the PS matching estimators were either similarly or even less biased than the IPTW estimators. Including variables unrelated to the exposure but related to the outcome in the PS model decreased the bias and the variance as compared to models omitting such variables. Excluding the true confounder from the PS model resulted, whatever the method used, in a significantly biased estimation of treatment effect. These results were illustrated in a real dataset.ConclusionEven in case of small study samples or low prevalence of treatment, PS-matching and IPTW can yield correct estimations of treatment effect unless the true confounders and the variables related only to the outcome are not included in the PS model.
16 schema:genre article
17 schema:isAccessibleForFree true
18 schema:isPartOf N52b341373c114937bc9bbb9051d9dc88
19 N8c1908ca16e64d61bec0af3efdf96966
20 sg:journal.1024940
21 schema:keywords Carlo simulations
22 ConclusionEven
23 I error rate
24 IPTW
25 IPTW estimators
26 IPTW method
27 MethodsWe
28 Monte Carlo simulations
29 PS
30 PS matching
31 PS method
32 PS model
33 Type I error rates
34 association
35 bias
36 biased estimation
37 biases
38 cases
39 confounders
40 context
41 correct estimation
42 dataset
43 effect
44 error rate
45 estimation
46 estimator
47 evaluation
48 exposure
49 influence
50 inverse probability
51 low prevalence
52 marginal odds ratio
53 matching estimators
54 method
55 model
56 odds ratio
57 outcomes
58 performance
59 prevalence
60 probability
61 propensity score methods
62 rate
63 ratio
64 real datasets
65 relative biases
66 relative performance
67 results
68 sample size
69 samples
70 score method
71 series
72 simulations
73 size
74 small sample size
75 small study sample
76 strength
77 study sample
78 subjects
79 such variables
80 treatment
81 treatment effects
82 treatment exposure
83 treatment weighting
84 true confounders
85 variables
86 variance
87 weighting
88 schema:name Evaluation of the Propensity score methods for estimating marginal odds ratios in case of small sample size
89 schema:pagination 70
90 schema:productId N126436ba2f464e1f8eb14e701b486cb1
91 N1e78b9bd7ff5455e881148526dc3947d
92 Nb45081d56e6a4e32be84322e6a92f788
93 schema:sameAs https://app.dimensions.ai/details/publication/pub.1005384649
94 https://doi.org/10.1186/1471-2288-12-70
95 schema:sdDatePublished 2022-12-01T06:29
96 schema:sdLicense https://scigraph.springernature.com/explorer/license/
97 schema:sdPublisher N54e3923e8c594a7cbb6c84601323e209
98 schema:url https://doi.org/10.1186/1471-2288-12-70
99 sgo:license sg:explorer/license/
100 sgo:sdDataset articles
101 rdf:type schema:ScholarlyArticle
102 N126436ba2f464e1f8eb14e701b486cb1 schema:name doi
103 schema:value 10.1186/1471-2288-12-70
104 rdf:type schema:PropertyValue
105 N1e78b9bd7ff5455e881148526dc3947d schema:name pubmed_id
106 schema:value 22646911
107 rdf:type schema:PropertyValue
108 N2304130da8f24e96a18b8af04ec9d984 rdf:first sg:person.0625147223.89
109 rdf:rest rdf:nil
110 N490819bd54264e69bc42df33c99a6b4f schema:inDefinedTermSet https://www.nlm.nih.gov/mesh/
111 schema:name Propensity Score
112 rdf:type schema:DefinedTerm
113 N4bb71e5bbaa94f369cc8ca28476ed2e2 rdf:first sg:person.01036155166.50
114 rdf:rest N2304130da8f24e96a18b8af04ec9d984
115 N52b341373c114937bc9bbb9051d9dc88 schema:issueNumber 1
116 rdf:type schema:PublicationIssue
117 N54e3923e8c594a7cbb6c84601323e209 schema:name Springer Nature - SN SciGraph project
118 rdf:type schema:Organization
119 N595bf4a392384102a5fd89a03bb6a104 schema:inDefinedTermSet https://www.nlm.nih.gov/mesh/
120 schema:name Sample Size
121 rdf:type schema:DefinedTerm
122 N8c1908ca16e64d61bec0af3efdf96966 schema:volumeNumber 12
123 rdf:type schema:PublicationVolume
124 N98c00cd8acfe44dfb339d401a51c9d36 rdf:first sg:person.01047623565.27
125 rdf:rest N4bb71e5bbaa94f369cc8ca28476ed2e2
126 Nb0aeca64b65d42119cfb0ec4a8a0d62a schema:inDefinedTermSet https://www.nlm.nih.gov/mesh/
127 schema:name Monte Carlo Method
128 rdf:type schema:DefinedTerm
129 Nb45081d56e6a4e32be84322e6a92f788 schema:name dimensions_id
130 schema:value pub.1005384649
131 rdf:type schema:PropertyValue
132 Nb4cd181848344b3db3ed418889e1d8b9 schema:inDefinedTermSet https://www.nlm.nih.gov/mesh/
133 schema:name Odds Ratio
134 rdf:type schema:DefinedTerm
135 Nc0262253cab144db9f8afb937a377e22 schema:inDefinedTermSet https://www.nlm.nih.gov/mesh/
136 schema:name Humans
137 rdf:type schema:DefinedTerm
138 anzsrc-for:11 schema:inDefinedTermSet anzsrc-for:
139 schema:name Medical and Health Sciences
140 rdf:type schema:DefinedTerm
141 anzsrc-for:1117 schema:inDefinedTermSet anzsrc-for:
142 schema:name Public Health and Health Services
143 rdf:type schema:DefinedTerm
144 sg:journal.1024940 schema:issn 1471-2288
145 schema:name BMC Medical Research Methodology
146 schema:publisher Springer Nature
147 rdf:type schema:Periodical
148 sg:person.01036155166.50 schema:affiliation grid-institutes:None
149 schema:familyName Resche-Rigon
150 schema:givenName Matthieu
151 schema:sameAs https://app.dimensions.ai/discover/publication?and_facet_researcher=ur.01036155166.50
152 rdf:type schema:Person
153 sg:person.01047623565.27 schema:affiliation grid-institutes:grid.508487.6
154 schema:familyName Pirracchio
155 schema:givenName Romain
156 schema:sameAs https://app.dimensions.ai/discover/publication?and_facet_researcher=ur.01047623565.27
157 rdf:type schema:Person
158 sg:person.0625147223.89 schema:affiliation grid-institutes:None
159 schema:familyName Chevret
160 schema:givenName Sylvie
161 schema:sameAs https://app.dimensions.ai/discover/publication?and_facet_researcher=ur.0625147223.89
162 rdf:type schema:Person
163 sg:pub.10.1007/978-1-4757-3692-2 schema:sameAs https://app.dimensions.ai/details/publication/pub.1013392911
164 https://doi.org/10.1007/978-1-4757-3692-2
165 rdf:type schema:CreativeWork
166 sg:pub.10.1007/s00134-010-1933-2 schema:sameAs https://app.dimensions.ai/details/publication/pub.1037710572
167 https://doi.org/10.1007/s00134-010-1933-2
168 rdf:type schema:CreativeWork
169 sg:pub.10.1007/s00134-010-1991-5 schema:sameAs https://app.dimensions.ai/details/publication/pub.1005546328
170 https://doi.org/10.1007/s00134-010-1991-5
171 rdf:type schema:CreativeWork
172 sg:pub.10.1038/bmt.2010.90 schema:sameAs https://app.dimensions.ai/details/publication/pub.1042034040
173 https://doi.org/10.1038/bmt.2010.90
174 rdf:type schema:CreativeWork
175 grid-institutes:None schema:alternateName Service de Biostatistique et Information Médicale, Hôpital Saint-Louis, UMR-S717 Inserm; Sorbonne Paris Cité, Université Paris Diderot, 1 avenue Claude Vellefaux, 75010, Paris, France
176 schema:name Service de Biostatistique et Information Médicale, Hôpital Saint-Louis, UMR-S717 Inserm; Sorbonne Paris Cité, Université Paris Diderot, 1 avenue Claude Vellefaux, 75010, Paris, France
177 rdf:type schema:Organization
178 grid-institutes:grid.508487.6 schema:alternateName 3- Service d'Anesthésie-Réanimation, Hôpital Européen Georges Pompidou, Université Paris 5 Descartes, Sorbonne Paris Cité, Paris, France
179 schema:name 3- Service d'Anesthésie-Réanimation, Hôpital Européen Georges Pompidou, Université Paris 5 Descartes, Sorbonne Paris Cité, Paris, France
180 Service de Biostatistique et Information Médicale, Hôpital Saint-Louis, UMR-S717 Inserm; Sorbonne Paris Cité, Université Paris Diderot, 1 avenue Claude Vellefaux, 75010, Paris, France
181 rdf:type schema:Organization
 




Preview window. Press ESC to close (or click here)


...