The DOI matches its Crossref record.
A critical look at methods for handling missing covariates in epidemiologic regression analyses.
Abstract
Epidemiologic studies often encounter missing covariate values. While simple methods such as stratification on missing-data status, conditional-mean imputation, and complete-subject analysis are commonly employed for handling this problem, several studies have shown that these methods can be biased under reasonable circumstances. The authors review these results in the context of logistic regression and present simulation experiments showing the limitations of the methods. The method based on missing-data indicators can exhibit severe bias even when the data are missing completely at random, and regression (conditional-mean) imputation can be inordinately sensitive to model misspecification. Even complete-subject analysis can outperform these methods. More sophisticated methods, such as maximum likelihood, multiple imputation, and weighted estimating equations, have been given extensive attention in the statistics literature. While these methods are superior to simple methods, they are not commonly used in epidemiology, no doubt due to their complexity and the lack of packaged software to apply these methods. The authors contrast the results of multiple imputation to simple methods in the analysis of a case-control study of endometrial cancer, and they find a meaningful difference in results for age at menarche. In general, the authors recommend that epidemiologists avoid using the missing-indicator method and use more sophisticated methods whenever a large proportion of data are missing.
Overview
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
Claims other papers make about this one
Each is checked against this paper’s own text: does it say what the citing paper says it does?
In the citing paper
“Commonly missing values are related to observed participant or disease characteristics.”
From Reporting and Methods in Clinical Prediction Research: A Systematic Review (2012) · cites this paper as [24]
In the citing paper
“Missing values in clinical study data rarely occur completely at random.”
From Reporting and Methods in Clinical Prediction Research: A Systematic Review (2012) · cites this paper as [24]
In the citing paper
“Exclusion of participants with missing values will therefore not only lead to loss of statistical power, but often also to biased results 10,12,13,23,24,57,58.”
From Reporting and Methods in Clinical Prediction Research: A Systematic Review (2012) · cites this paper as [24]
Citations
Cited by
Sorted by citations
Items 801–892 of 892
Bias in estimates of seat belt effectiveness
Koepsell, T. D., Rivara, F. P., Grossman, D. C., et al. (2003) · Injury Prevention · doi:10.1136/ip.9.1.91-a
Byford, M. (2009) · UCL Discovery (University College London)
Bénet, T., D'Óliveira, A., Voirin, N., et al. (2007) · Journal of Viral Hepatitis · doi:10.1111/j.1365-2893.2007.00863.x
Diagnostic strategies in patients with lower abdominal complaints in primary care
Kok, L. (2014) · Utrecht University Repository (Utrecht University)
Depression and Diabetes - Methodological Issues in Etiologic Research
Knol, M. J. (2004) · Utrecht University Repository (Utrecht University)
Richardson et al. Respond to "Missing Doses in the Life Span Study"
Richardson, D. B., Wing, S., & Cole, S. R. (2013) · American Journal of Epidemiology · doi:10.1093/aje/kws475
Towards Genetic Prediction of Coronary Heart Disease in Familial Hypercholesterolemia
van der Net, J. (2009) · RePub (Erasmus University Rotterdam)
Impact of prenatal screening on future resource utilization in pregnancy
Metcalfe, A. L. (2012) · Libraries and Cultural Resources (University of Calgary) · doi:10.11575/prism/26463
Metabolic syndrome and the development of vascular disease and type 2 diabetes in high-risk patients
Wassink, A. M. J. (2009) · Utrecht University Repository (Utrecht University)
Exhaled nitric oxide and asthma in childhood
van der Valk, R. J. P. (2013)
Abímbọ́lá, Ṣ. (2010) · UPT. Syiah Kuala University Library (Syiah Kuala University)
Treatment and outcome in basilar artery occlusion
Schonewille, W. J. (2014) · Utrecht University Repository (Utrecht University)
Recent Advances in Missing Data Methods: Imputation and Weighting
Stuart, E. A. (2012)
Impact of distance on mortality in trauma patients with an out-of-hospital intubation attempt
Cudnik, M. T. (2007) · OHSU Digital Commons · doi:10.6083/m4br8qf2
The use of multiple imputation in molecular epidemiologic studies assessing interaction effectsPreprint
Desai, M., Esserman, D. A., Gammon, M. D., et al. (2010) · Collection of Biostatistics Research Archive
Clinical aspects of Postpartum Hemorrhage
Prick, B. W. (2014) · RePub (Erasmus University Rotterdam)
Bradshaw, P. T. (2019) · Carolina Digital Repository (University of North Carolina at Chapel Hill) · doi:10.17615/gtg1-4626
Treatment strategies impacting ceftiofur resistance among enteric bacteria in cattle
Kanwar, N. (2013) · K-State Research Exchange (Kansas State University)
Typologies of alcohol consumption in adolescence: predictors and adult outcomes
Cable, N., & Sacker, A. (2006) · UCL Discovery (University College London)
Racial and Ethnic Disparities in Type-2 Diabetes: A Multilevel Perspective
Piccolo, R. S. (2016) · LSHTM Research Online (London School of Hygiene and Tropical Medicine) · doi:10.17037/pubs.02537689
Longitudinal Effects of Cognitive Reserve and Vascular Risks in Aging and Dementia
Lo, Y. Y. (2012) · eScholarship (California Digital Library)
Methodologie Van Wetenschappelijk Onderzoek
Schouten, H. J. A. (1999) · Bohn Stafleu van Loghum eBooks · doi:10.1007/978-90-313-9661-0_19
Adams, R. N. (2016) · Purdue e-Pubs (Purdue University System) · doi:10.7912/c2hg66
“Single imputation methods, such as the one used, are considered acceptable when few data are missing and the data are thought to be MCAR (Greenland & Finkle, 1995).”
MethodBengtson, A. M. (2019) · Carolina Digital Repository (University of North Carolina at Chapel Hill) · doi:10.17615/1qkp-9k14
Lagrutta, J., Quijano, F. J. R., de Fernandez, C. D. V., et al. (2015) · Revista Peruana de Ginecología y Obstetricia · doi:10.31403/rpgo.v12i1322
Phenotypic Characterization of Paediatric Inflammatory Bowel Disease
Sherlock, M. E. (2011)
Trotta, A. (2016) · Research Portal (King's College London)
The diagnosis of serious bacterial infections in the children’s emergency department
Irwin, A. (2016) · University of Liverpool · doi:10.17638/02005599
Wilson, J. R. (2013) · TSpace (University of Toronto)
SPECIAL SERIES: MISSING DATA Review: A gentle introduction to imputation of missing values
Rogier, A., Donders, T., & Stijnen, T. (2006)
Statistical Approaches to Revealing Structure in Complex Health Data
Tierney, N. J. (2018) · Queensland University of Technology · doi:10.5204/thesis.eprints.115536
Novel methods for distributed acoustic sensing data
Wilson, R. (2019) · Lancaster EPrints (Lancaster University) · doi:10.17635/lancaster/thesis/651
“, 2004; Plaia and Bondi, 2006), epidemiology (Greenland and Finkle, 1995; Clark and Altman, 2003) and signal processing (Gemmeke et al.”
BackgroundComparison of Methods for Processing Missing Values in Large Sample Survey Data
Wang, L., Zhang, D., Duan, J., et al. (2019) · Science Journal of Public Health · doi:10.11648/j.sjph.20190705.13
“In contrast, if the observed data are highly predictive of the missing values the imputations, it will be smaller bias in the results [28-29].”
ResultBerman, L. (2019) · Routledge eBooks · doi:10.4324/9780203753910-4
Natural History Study of Prognosis
Cnaan, A. (2014) · Wiley StatsRef: Statistics Reference Online · doi:10.1002/9781118445112.stat05247
Ogollah, R. O. (2010) · Bournemouth University Research Online (Bournemouth University)
Tobacco Smoke Exposure and Pediatric Multiple Sclerosis
Lavery, A. M. (2016) · TUScholarShare (Temple University)
Modelos de predição de risco de morte para pacientes com carcinoma epidermoide de cabeça e pescoço
Rios, L. F. B. (2020) · Digital Library of Theses and Dissertations (Universidade de São Paulo) · doi:10.11606/t.5.2020.tde-29012021-143358
DETERMINANTS OF MATERNAL ANTENATAL STATE-ANXIETY: ROLE OF MATERNAL FEELINGS ABOUT THE PREGNANCY
Akiki, S. (2011) · Scholarship@Western (Western University)
Profiles of Parental Mental Health and Children’S Academic Coping
Fiat, A. E. (2020) · University of Minnesota Digital Conservancy (University of Minnesota)
Learning from the Data Heterogeneity for Data Imputation
liu, X. (2021)
Analysis of Clinical Trials with Missing Data by Modeling Causes of Drop-Outs
Kada, A., Matsuyama, Y., & 俊哉, 佐. (2002) · Japanese Journal of Biometrics · doi:10.5691/jjb.23.81
Dealing with missing data in randomized and cluster randomized trials
Kayembe, M. T. (2021) · doi:10.26481/dis.20211216mk
van Buuren, S. (2012) · Chapman & Hall/CRC machine learning & pattern recognition/Chapman & Hall/CRC machine learning & pattern recognition series · doi:10.1201/b11826-16
Shintani, A., & Eden, S. (2012) · Faculty Opinions – Post-Publication Peer Review of the Biomedical Literature · doi:10.3410/f.13972971.15429075
Direct Bias Modeling and Missing Data Methods for Bias Analysis
Fox, M. P., MacLehose, R. F., & Lash, T. L. (2021) · Statistics for biology and health · doi:10.1007/978-3-030-82673-4_10
REMOVED DUE TO POLICY VIOLATIONSPreprint
Krämer, M. D., van Scheppingen, M. A., Chopik, W. J., et al. (2022) · doi:10.31234/osf.io/87emk
Carpenito, T. (2022) · doi:10.17760/d20449633
Hendry, G. M., Zewotir, T., Naidoo, R. N., & North, D. (2022) · South African Statistical Journal · doi:10.37920/sasj.2017.51.1.10
Zuk, A. M., Tsuji, L. J. S., Nieboer, E., et al. (2023) · doi:10.32920/22782689.v1
Imputation Method for Multidimensional Data
Naik, J. S., & Jadhav, A. S. (2023) · International Conference on Intelligent Technologies (CONIT) · doi:10.1109/conit59222.2023.10205777
“In the article [7], authors review the results of studies on biased performances due to simple imputation methods in the context of logistic regression and simulation experiments showing the limitations of the methods.”
BackgroundOptimizing Geriatric Traumacare
Kusen, J. Q. (2023) · Utrecht University Repository (Utrecht University) · doi:10.33540/2049
Underwood, L., Kool, B., Pillai, A., et al. (2023) · Injury Prevention · doi:10.1136/ip-2023-044869
Schrager, S. M., Mamey, M. R., Rusow, J. A., & Goldbach, J. T. (2024) · Psychology of Sexual Orientation and Gender Diversity · doi:10.1037/sgd0000706
Holmberg, S., Lendahls, L., & Alle, K.-Å. (2024) · PLoS ONE · doi:10.1371/journal.pone.0303021
Harrison, J., Sullivan, F., Keenan, K. L., & Kulu, H. (2024) · BMC Public Health · doi:10.1186/s12889-024-20813-1
“Further since the use of ‘missing’ as a category generates scepticism in health research [84], we repeated the analysis using only complete cases, the results hold with only small changes to the magnitude, see ‘Additional Table 3, Additional Results’.”
Observability, Honesty, and the Social Image Costs of Lying *Preprint
Bosch-Rosa, C., Nosenzo, D., & Neyse, L. (2025) · SSRN Electronic Journal · doi:10.2139/ssrn.5080148
An investigation on synthetic data generation from original incomplete data
Seo, Y.-B., & Kim, Y. M. (2025) · Communications in Statistics - Simulation and Computation · doi:10.1080/03610918.2025.2496777
“This assumption is based on the MAR framework, which posits that appropriate imputation methods can yield unbiased estimates when replacing missing values (Greenland and Finkle 1995) and effectively mitigate bias even in the presence of substantial missingness (Madley-Dowd et al. 2019).”
Blevins, E. J., Chen, Y., Sánchez, S. E., Rondón, M. B., Sánchez, S. E., Gelaye, B., & Basu, A. (2025) · Child Psychiatry & Human Development · doi:10.1007/s10578-025-01872-w
The Impact of Imputation Timing on Model Performance Estimation
Mitchell, B., & Shrestha, S. (2025) · IEEE International Conference on Artificial Intelligence Testing (AITest) · doi:10.1109/aitest66680.2025.00032
“This problem extends beyond traditional machine learning to a range of other data-related fields, and leads to issues including performance degradation, data analysis challenges, and biased outcomes due to disparities between missing and complete data [4], [17], [25], [26].”
Creating Data Systems to Promote Health Equity
Tjoeng, Y. L., Mazwi, M., Goodwin, A., McCradden, M. D., & Chan, T. (2025) · PEDIATRICS · doi:10.1542/peds.2025-070739k
Setting Up a Multivariable Analysis
Katz, M. H. (2025) · Cambridge University Press eBooks · doi:10.1017/9781009558488.007
Fischer, M., Little, R. J. A., & West, B. T. (2025) · Journal of Statistical Computation and Simulation · doi:10.1080/00949655.2025.2558859
“Earlier literature does not recommend including RIs in an analysis model targeted at the whole population [7,8], because this approach can lead to biased estimates [9–11].”
How to manage missing covariates in randomized controlled trials: a comparison of strategies
Zhang, S., Si, Y., & Dziak, J. J. (2025) · BMC Medical Research Methodology · doi:10.1186/s12874-025-02708-w
“In the literature, strategies for handling incomplete covariates were originally developed based on observational studies [1, 7, 8], but RCTs represent a special case where strategies and assumptions from observational study data are not always directly applicable in the same way.”
Time to First Injury After Knee Surgery in U.S. Army Soldiers: A Survival Analysis
Adams, B., Westrick, R. B., Gochyyev, P., & Taylor, K. M. (2025) · Sports Medicine - Open · doi:10.1186/s40798-025-00955-1 · Cited by 1
“To account for the relatively small amount of missing covariate data ( 4.3%), we used multiple imputation with 5 imputations [30, 31], and results from each imputed regression were pooled.”
InfluentialMigration, Health and Help-seeking in Childhood
Flink, I. J. E. (2013) · RePub (Erasmus University Rotterdam)
Bellocco, R. (2004) · BOA (University of Milano-Bicocca)
Pan, Y., Bobak, M., Pikhart, H., et al. (2026) · Population Health Metrics · doi:10.1186/s12963-026-00459-4
“Since the proportion of missing data in baseline characteristics was about 10%, multiple imputation (for covariates) had little advantage over complete case analysis (CCA) [19, 20].”
Dağ, A., Dolatsara, H. A., Asilkalkan, A., et al. (2026) · Decision Analytics Journal · doi:10.1016/j.dajour.2026.100690
Characteristics and outcomes of men presenting with complications of metastatic prostate cancer
Nathan, A., Parry, M. G., Cook, A. D., et al. (2026) · British Journal of Urology · doi:10.1111/bju.70179
“Missing patient characteristics were handled using the missing indicator method, where an extra ‘ missing ’ category was created for each patient characteristic [20].”
Development and validation of risk scores in vascular diseases
Lapébie, F.-X. (2021) · theses.fr (ABES) · doi:10.70675/4893255az220cz4e47z8833ze45283a937a4
Evaluating Sex‐Based Disparities in STEMI Care: A Call for Methodological Rigor
Haider, N. (2026) · Clinical Cardiology · doi:10.1002/clc.70298
Mehta, P., Yang, M., Kharouf, F., Abarza, V. C., Gao, S., Cook, R. J., Gladman, D. D., Chandran, V., & Poddubnyy, D. (2026) · Lara D. Veeken · doi:10.1093/rheumatology/keag303
“Missing data were addressed by multiple imputation with chained equations (MICE) in R, and twenty ‘complete’ datasets were generated [37].”
Kalthoum, D. E., Dakin, H., Bennett, D., & Bolbocean, C. (2026) · doi:10.2196/preprints.105748
Lomenzo, K., & Lopes, V. D. F. (2026) · Journal of Interpersonal Violence · doi:10.1177/08862605261463274
“Little’s MCAR test indicated data were not missing completely at random, χ 2 (21) = 480, p .001 (Little, 1988), supporting the MAR assumption (Greenland & Finkle, 1995).”
Exploratory visual analysis of missing data for multiple imputation
Matute, J., Beule, A. G., Huebner, M., Schmidt, C. O., Völzke, H., & Linsen, L. (2026) · Information Visualization · doi:10.1177/14738716261465405
Kalthoum, D. E., Dakin, H., Bennett, D., & Bolbocean, C. (2026) · JMIR Formative Research · doi:10.2196/105748
Studies of the Epidemiology and Prognosis of Patients with Heart Failure in Leicestershire
Newton, J. D. (2009) · Leicester Research Archive (University of Leicester)
Denholm, R. (2013) · UCL Discovery (University College London)
McDowell, S. E. (2011) · University of Birmingham Institutional Research Archive (University of Birmingham)
Multiple imputation of missing blood pressure covariates in survival analysis
van Buuren, S., Boshuizen, H. C., & Knook, D. L. (1999) · Statistics in Medicine · doi:10.1002/(sici)1097-0258(19990330)18:6
Metadata
- Authors
- Greenland, S., & Finkle, W. D.
- Affiliations
- University of California, Los Angeles (US)
- Corresponding author
- Greenland, Sander
- ORCID
- Greenland, Sander
- Venue
- American Journal of Epidemiology
- Publisher
- Oxford University Press (OUP)
- Published
- 1995-12-15
- Volume
- 142
- Issue
- 12
- Pages
- 1255-1264
- ISSN
- 0002-9262 (ISSN-L)
- 1476-6256
- OpenAlex
- W2166561686
- Cited by
- 891
- References
- 0
- Citations by year
- 2026: 13 · 2025: 18 · 2024: 13 · 2023: 24 · 2022: 26 · 2021: 21 · 2020: 33 · 2019: 29 · 2018: 34 · 2017: 32 · 2016: 27 · 2015: 40 · 2014: 57 · 2013: 53 · 2012: 67
- FWCI
- 6.57
- Type
- journal article
- Language
- en
- Open access
- closed
- Locations
- American Journal of Epidemiology (publishedVersion)
- PubMed (publishedVersion)
- American Journal of Epidemiology
- Funders
- National Institutes of Health (n01cp095680-001)
- National Cancer Institute (N01 CP 95680)
- Topics
- Physical Sciences › Mathematics › Statistics and Probability › Statistical Methods and Bayesian Inference
- Physical Sciences › Mathematics › Statistics and Probability › Statistical Methods in Epidemiology
- Physical Sciences › Mathematics › Statistics and Probability › Advanced Causal Inference Techniques
- Keywords
- missing covariates, missing data, logistic regression, bias, model misspecification, multiple imputation, maximum likelihood estimation, weighted estimating equations, endometrial cancer, age at menarche, case-control study
- MeSH terms
- Epidemiologic Methods, Female, Humans, Menarche, Research Design, Logistic Models, Odds Ratio, Case-Control Studies, Endometrial Neoplasms
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
