Propensity score analysis with partially observed covariates: How should multiple imputation be used?
Open accessThe DOI matches its Crossref record; 11 of the 11 references it cites were found in a registry or an index.
Abstract
Inverse probability of treatment weighting is a popular propensity score-based approach to estimate marginal treatment effects in observational studies at risk of confounding bias. A major issue when estimating the propensity score is the presence of partially observed covariates. Multiple imputation is a natural approach to handle missing data on covariates: covariates are imputed and a propensity score analysis is performed in each imputed dataset to estimate the treatment effect. The treatment effect estimates from each imputed dataset are then combined to obtain an overall estimate. We call this method MIte. However, an alternative approach has been proposed, in which the propensity scores are combined across the imputed datasets (MIps). Therefore, there are remaining uncertainties about how to implement multiple imputation for propensity score analysis: (a) should we apply Rubin’s rules to the inverse probability of treatment weighting treatment effect estimates or to the propensity score estimates themselves? (b) does the outcome have to be included in the imputation model? (c) how should we estimate the variance of the inverse probability of treatment weighting estimator after multiple imputation? We studied the consistency and balancing properties of the MIte and MIps estimators and performed a simulation study to empirically assess their performance for the analysis of a binary outcome. We also compared the performance of these methods to complete case analysis and the missingness pattern approach, which uses a different propensity score model for each pattern of missingness, and a third multiple imputation approach in which the propensity score parameters are combined rather than the propensity scores themselves (MIpar). Under a missing at random mechanism, complete case and missingness pattern analyses were biased in most cases for estimating the marginal treatment effect, whereas multiple imputation approaches were approximately unbiased as long as the outcome was included in the imputation model. Only MIte was unbiased in all the studied scenarios and Rubin’s rules provided good variance estimates for MIte. The propensity score estimated in the MIte approach showed good balancing properties. In conclusion, when using multiple imputation in the inverse probability of treatment weighting context, MIte with the outcome included in the imputation model is the preferred approach.
Overview
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
The self-check needs the paper’s full text.
Citations
Cites
Estimating causal effects of treatments in randomized and nonrandomized studies
Rubin, D. B. (1974) · Journal of Educational Psychology · doi:10.1037/h0037350 · Cited by 9,436 · cited 1×
“The PS approach is best understood using the potential outcomes framework [27], in which the causal effect of the treatment is defined as the difference between the two potential outcomes Y Z=0 and Y , which are the outcomes that would have been observed if an individual had been not treated and treated, respectively…”
BackgroundCausal Inference for Statistics, Social, and Biomedical Sciences: An Introduction
Imbens, G. W., & Rubin, D. B. (2015) · doi:10.1017/cbo9781139025751 · Cited by 4,210
Randomized, Controlled Trials, Observational Studies, and the Hierarchy of Research Designs
Concato, J., Shah, N., & Horwitz, R. I. (2000) · New England Journal of Medicine · doi:10.1056/nejm200006223422507 · Cited by 3,563
Multiple imputation by chained equations: what is it and how does it work?
Azur, M. J., Stuart, E. A., Frangakis, C., & Leaf, P. J. (2011) · International Journal of Methods in Psychiatric Research · doi:10.1002/mpr.329 · Cited by 3,542 · cited 2×
“[38] Azur MJ, Stuart EA, Frangakis C, Leaf PJ.”
BackgroundMethodReducing Bias in Observational Studies Using Subclassification on the Propensity Score
Rosenbaum, P. R., & Rubin, D. B. (1984) · Journal of the American Statistical Association · doi:10.1080/01621459.1984.10478078 · Cited by 3,082
Constructing Inverse Probability Weights for Marginal Structural Models
Cole, S. R., & Hernan, M. A. (2008) · American Journal of Epidemiology · doi:10.1093/aje/kwn164 · Cited by 2,796 · cited 1×
“Three assumptions are usually made to consistently estimate the causal effect of a treatment: (i) positivity [28], (ii) Stable unit treatment value assumption (SUTVA) [29] and”
BackgroundA comparison of 12 algorithms for matching on the propensity score
Austin, P. C. (2014) · Statistics in Medicine · doi:10.1002/sim.6004 · Cited by 1,085
Hirano, K., & Imbens, G. W. (2001) · Health Services and Outcomes Research Methodology · doi:10.1023/a:1020371312283 · Cited by 1,039 · cited 1×
“This approach, known as inverse probability of treatment weighting (IPTW) [5] aims to emulate the sample that would have been observed in a randomised trial.”
Multiple Imputation and its Application
Carpenter, J. R., & Kenward, M. G. (2013) · doi:10.1002/9781119942283 · Cited by 913
Self-citationFor objective causal inference, design trumps analysis
Rubin, D. B. (2008) · The Annals of Applied Statistics · doi:10.1214/08-aoas187 · Cited by 831 · cited 2×
“Furthermore, the inclusion of the outcome in the imputation model may go against the principle of objective design, as described by Rubin.(44) However, when using MI, this is necessary to obtain a consistent estimator of the treatment effect.”
MethodLewis, J. D., Schinnar, R., Bilker, W. B., Wang, X., & Strom, B. L. (2007) · Pharmacoepidemiology and Drug Safety · doi:10.1002/pds.1335 · Cited by 602
Research works printed but not linked: 35
- Cochran WG, 1973, Sankhyā: Indian J Stat Ser A (1961-2002), 35, 417
- 10.1093/biomet/70.1.41 · DOI · Paper page
- Guo S, 2009, Propensity score analysis: statistical methods and applications
- 10.1002/sim.3944 · DOI · Paper page
- Mitra R, 2012, Stat Methods Med Res, 0, 1
- 10.1080/03610926.2012.700371 · DOI
- 10.1016/j.jclinepi.2006.01.009 · DOI · Paper page
- 10.1002/sim.3549 · DOI
- 10.1016/j.jpedsurg.2014.10.035 · DOI
- 10.1016/j.urology.2016.04.042 · DOI
- 10.7326/M13-1505 · DOI
- 10.1007/s40264-015-0292-x · doi:10.1007/s40264-015-0292-x
- 10.1186/s12889-016-3437-7 · DOI
- 10.1371/journal.pone.0166444 · DOI
- 10.1093/rheumatology/ket390 · DOI
- 10.1016/j.jpedsurg.2007.12.043 · DOI
- 10.1136/bmj.d1642 · DOI
- Blak BT, 2011, Inform Prim Care, 19, 251
- 10.1002/(SICI)1097-0258(19981015)17:19<2265::AID-SIM918>3.0.CO;2-B · doi:10.1002/(sici)1097-0258(19981015)17:19<2265::aid-sim918>3.0.co;2-b
- Rubin DB, 1986, J Am Stat Assoc, 81, 961
- 10.1111/j.1467-9531.2010.01226.x · DOI
- 10.1002/sim.5884 · DOI
- 10.1080/01621459.1987.10478441 · DOI · Paper page
- 10.1002/sim.5991 · DOI · Paper page
- 10.1002/sim.1903 · DOI · Paper page
- Hill J. Reducing bias in treatment effect estimation in observational studies suffering from missing data. Columbia University Commons , http://hdl.handle.net/10022/AC:P:9697 (2004, accessed 20 May 2017).
- 10.1080/01621459.2000.10474263 · DOI · Paper page
- 10.1093/biomet/87.3.706 · DOI · Paper page
- 10.1002/sim.2781 · DOI · Paper page
- Gelman A, 2011, J Stat Softw, 40, 1
- 10.1201/b16654 · doi:10.1201/b16654 · Paper page
- Tsiatis A, 2007, Semiparametric theory and missing data
- 10.1177/0962280216674296 · DOI
- 10.1177/0962280214521348 · DOI · Paper page
- 10.1002/sim.6532 · DOI
Cited by
Lee, K. J., Tilling, K. M., Cornish, R. P., Little, R. J. A., Bell, M. L., Goetghebeur, E., Hogan, J. W., & Carpenter, J. R. (2021) · Journal of Clinical Epidemiology · doi:10.1016/j.jclinepi.2021.01.008 · Cited by 374
Self-citationPropensity score matching with R: conventional methods and new features
Zhao, Q.-Y., Luo, J.-C., Su, Y., Zhang, Y.-J., Tu, G.-W., & Luo, Z. (2021) · Annals of Translational Medicine · doi:10.21037/atm-20-3998 · Cited by 291
Wong, C. K. H., Au, I. C. H., Lau, K. T. K., Lau, E. H. Y., Cowling, B. J., & Leung, G. M. (2022) · The Lancet Infectious Diseases · doi:10.1016/s1473-3099(22)00507-2 · Cited by 269
Diabetes medications and risk of Parkinson’s disease: a cohort study of patients with diabetes
Brauer, R., Wei, L., Ma, T., Athauda, D., Girges, C., Vijiaratnam, N., Auld, G., Whittlesea, C., Wong, I., & Foltynie, T. (2020) · Brain · doi:10.1093/brain/awaa262 · Cited by 233
Ali, M. S., Prieto-Alhambra, D., Lopes, L. C., Ramos, D., Bispo, N., Ichihara, M. Y., Pescarini, J. M., Williamson, E., Fiaccone, R. L., Barreto, M. L., et al. (2019) · Frontiers in Pharmacology · doi:10.3389/fphar.2019.00973 · Cited by 200
Self-citation
Metadata
- Authors
- Leyrat, C., Seaman, S. R., White, I. R., Douglas, I., Smeeth, L., Kim, J., Resche-Rigon, M., Carpenter, J. R., & Williamson, E. J.
- Venue
- Statistical Methods in Medical Research
- Publisher
- SAGE Publications
- Published
- 2017-06-02
- Volume
- 28
- Issue
- 1
- Pages
- 3-19
- ISSN
- 0962-2802 (ISSN-L)
- 1477-0334
- OpenAlex
- W2509191638
- Cited by
- 249
- References
- 11
- FWCI
- 10.55
- Type
- journal article
- Language
- en
- Open access
- hybrid
- Licence
- journals.sagepub.com/page/policies/text-and-data-mining-license (tdm)
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.