Interactions in Epidemiology: Relevance, Identification, and Estimation
The DOI matches its Crossref record; 29 of the 30 references it cites were found in a registry or an index.
Abstract
Sufficient-cause models represent biologic interactions (including synergism) in the form of coparticipation of factors in a sufficient cause1–3 (causal coaction). In this issue of the journal, VanderWeele1 provides new results relating sufficient-cause models to regression models. These results reinforce the point that concepts of biologic interaction do not in general correspond to the concept of statistical interaction, because the latter is only the need for a product term in a statistical model.4,5 They further underscore that, without strict assumptions, sufficient-cause interactions do not correspond to interdependence of effects in potential-outcomes (counterfactual) models. I will examine the practical relevance and limitations of using epidemiologic regressions to study interactions, however the latter are defined. Relevance of Interactions A practical issue concerning interactions is their relevance to public health goals of finding cost-effective intervention strategies for disease reduction. Part of public health involves identifying population subgroups that would benefit most from a given intervention. These are not always high-risk subgroups—it is possible that a lower-risk group would obtain the greatest absolute risk reduction from the proposed intervention. In the absence of bias, departures from risk additivity imply that some subgroups would obtain a greater absolute risk reduction from the intervention than others would. Thus earlier authors noted that if costs were measured in terms of case-load per unit population (average risk), the relevant null model for detecting special groups would be additive on the risk scale.6,7 Risk nonadditivity (or “public health interaction”) implies the existence of such groups.5,6 It also implies existence of interdependent effects in individuals, also known as “interactions” in basic potential-outcome (counterfactual) causal models; that is, risk nonadditivity implies there are individuals such that the presence or direction of effect of one factor depends on the other factor.5,8 Risk nonadditivity is equivalent to heterogeneity or modification of risk differences for one variable across levels of another.5 For factors that can act only causally, excess risk above additivity (superadditivity) among those exposed to both factors signals the presence of individuals who get the disease only when exposed to both factors (“synergism” in the potential-outcomes sense).4,5,8 Identifying such individuals is valuable for preventive medicine and public health, because for them the outcome can be prevented by removing either of the factors. Nonidentifiability of Biologic Interactions Epidemiologic studies observe population distributions of risks, which submerge details of individual experiences. One consequence of this submersion is that many different biologic models will lead to the same population model.9,10 In particular, the implications of biologic models for interaction are unidirectional; for example, risk additivity could hold because no interdependence of effects exists, or instead may hold because of cancellations of the population effects of different individual interactions.5,8 Perfect cancellation of interactions is a higher-order analog of “unfaithfulness” in causal diagrams,11,12 where 2 variables may be unassociated despite having causal connections. For example, a treatment with effects could appear independent of the outcome if it causes and prevents equal numbers of cases. Although perfect cancellation is often implausible, any degree of cancellation will make it more difficult to detect effects or their interdependencies. Even if we could observe each individual's response under each exposure pattern, we still could not determine from those observations which biologic mechanisms or sufficient-cause types were operative8,13–15; hence, individual mechanisms are nonidentifiable even from a perfect factorial or crossover trial.13,15 For example, even with no sufficient-cause synergism, variation in background characteristics could nonetheless produce superadditivity.14 More generally, estimation of degrees of interaction will require further assumptions (eg, monotonicity) that cannot be tested with epidemiologic data, even if the latter are perfect, although valid tests for interactions can be constructed from weaker conditions.1,5,8,13–15 Statistical Modeling Issues Epidemiologic theories of interactions1,4,5,8,9,14–17 derive population causal models from models for individual etiology (eg, risk additivity is derived from the absence of interdependent effects). Study of these models and forms does not require that one use the derived population model as the statistical analysis model. One can estimate risks or rates by using any statistical model, then use these estimates to test or estimate the parameters in the causal model.1,18 For example, one can use risk estimates from a logistic statistical model for inference on parameters in an additiv
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
“Regarding studies that employ the product term specification, epidemiologists have called for an end to the practice of assessing interactions based purely on product terms in multiplicative models (Greenland, 2009; Rajaleid, Janszky, & Hallqvist, 2011).”
From Syndemics of psychosocial problems and HIV risk: A systematic review of empirical tests of the disease interaction concept (2015) · cites this paper as Greenland, 2009
In the citing paper
“The concept of disease interaction has had a long and controversial history in the epidemiologic literature (Blot & Day, 1979; Greenland, 2009; Kaufman, 2009; Kupper & Hogan, 1978; Rothman, 1974, 1976a; Rothman, Greenland, & Walker, 1980; Saracci, 1980; Siemiatycki & Thomas, 1981; VanderWeele, 2009a; Walter & Holford, 1978).”
From Syndemics of psychosocial problems and HIV risk: A systematic review of empirical tests of the disease interaction concept (2015) · cites this paper as Greenland, 2009
In the citing paper
“At the same time, from the public health perspective, additive effects are favourable over multiplicative effects to increase the net benefit by allocating the treatment to the proper subgroup.12,13”
From Understanding of interaction (subgroup) analysis in clinical trials (2019) · cites this paper as [12]
Citations
Cites
The Elements of Statistical Learning: Data Mining, Inference, and Prediction
John Lu, Z. Q. (2010) · Journal of the Royal Statistical Society Series A: Statistics in Society · doi:10.1111/j.1467-985x.2010.00646_6.x · Cited by 12,390
Bootstrap Methods and their Application
Davison, A. C., & Hinkley, D. V. (1997) · doi:10.1017/cbo9780511802843 · Cited by 7,350
Friedman, J., Hastie, T., & Tibshirani, R. (2000) · The Annals of Statistics · doi:10.1214/aos/1016218223 · Cited by 6,884 · cited 1×
“included, this goal leads to fitting methods that can use a very large set of terms, such as shrinkage(27) and machine-learning algorithms such as boosting.(28,29) The same methods can also”
MethodROTHMAN, K. J., GREENLAND, S., & WALKER, A. M. (1980) · American Journal of Epidemiology · doi:10.1093/oxfordjournals.aje.a113015 · Cited by 859 · cited 2×
“Thus earlier authors noted that if costs were measured in terms of case-load per unit population (average risk), the relevant null model for detecting special groups would be additive on the risk scale.(6,7) Risk nonadditivity (or “public health interaction”) implies the existence of such groups.”
BackgroundSelf-citationROTHMAN, K. J. (1976) · American Journal of Epidemiology · doi:10.1093/oxfordjournals.aje.a112335 · Cited by 776
Multiple-bias modelling for analysis of observational data
Greenland, S. (2005) · Journal of the Royal Statistical Society Series A (Statistics in Society) · doi:10.1111/j.1467-985x.2004.00349.x · Cited by 497
Self-citationTHE EFFECT OF MISCLASSIFICATION IN THE PRESENCE OF COVARIATES
GREENLAND, S. (1980) · American Journal of Epidemiology · doi:10.1093/oxfordjournals.aje.a113025 · Cited by 494 · cited 1×
“ity (such as nonadditivity), and measurement error can cause any of these problems even if it is nondifferential.(21,32)”
BackgroundSelf-citationAn overview of relations among causal modelling methods
Greenland, S., & Brumback, B. (2002) · International Journal of Epidemiology · doi:10.1093/ije/31.5.1030 · Cited by 406 · cited 1×
“Even if we could observe each individual’s response under each exposure pattern, we still could not determine from those observations which biologic mechanisms or sufficient-cause types were operative; hence, individual mechanisms are nonidentifiable even from a perfect factorial or crossover trial.(13,15) For example…”
BackgroundSelf-citationInvited Commentary: Variable Selection versus Shrinkage in the Control of Multiple Confounders
Greenland, S. (2007) · American Journal of Epidemiology · doi:10.1093/aje/kwm355 · Cited by 283
Self-citationTests for interaction in epidemiologic studies: A review and a study of power
Greenland, S. (1983) · Statistics in Medicine · doi:10.1002/sim.4780020219 · Cited by 280
Self-citation
Research works printed but not linked: 1
- Greenland, 2007, Encyclopedia of Epidemiology, Vol. 1., 149
Cited by
Detecting gene–gene interactions that underlie human diseases
Cordell, H. J. (2009) · Nature Reviews Genetics · doi:10.1038/nrg2579 · Cited by 1,350
VanderWeele, T. J., & Knol, M. J. (2014) · Epidemiologic Methods · doi:10.1515/em-2013-0005 · Cited by 1,127
“This will be relevant whenever resources are constrained and thus relevant also for cost-effectiveness (Greenland, 2009).”
BackgroundGene–environment-wide association studies: emerging approaches
Thomas, D. (2010) · Nature Reviews Genetics · doi:10.1038/nrg2764 · Cited by 692
Bally, M., Dendukuri, N., Rich, B., Nadeau, L., Helin-Salmivaara, A., Garbe, E., & Brophy, J. M. (2017) · BMJ · doi:10.1136/bmj.j1909 · Cited by 528
“73 Since methods exist for pooling IPD and aggregate data 51 we considered performing a combined IPD aggregate data meta-analysis.”
MethodHart, C. L., Morrison, D. S., Batty, G. D., Mitchell, R. J., & Davey Smith, G. (2010) · BMJ · doi:10.1136/bmj.c1240 · Cited by 422
“Our findings have important clinical and public health implications. New perspectives on the risk of liver disease may need to be considered for people who are overweight and consume alcohol and lower, BMI specific “safe” limits of alcohol consumption may need to be defined. Liver disease is often advanced when it is…”
Metadata
- Authors
- Greenland, S.
- Affiliations
- University of California, Los Angeles (US)
- Corresponding author
- Greenland, Sander
- ORCID
- Greenland, Sander
- Venue
- Epidemiology
- Published
- 2009-01
- Volume
- 20
- Issue
- 1
- Pages
- 14-17
- ISSN
- 1044-3983 (ISSN-L)
- OpenAlex
- W1989872151
- Cited by
- 204
- References
- 30
- Citations by year
- 2026: 9 · 2025: 15 · 2024: 15 · 2023: 10 · 2022: 11 · 2021: 13 · 2020: 9 · 2019: 10 · 2018: 6 · 2017: 6 · 2016: 16 · 2015: 14 · 2014: 17 · 2013: 14 · 2012: 15
- FWCI
- 7.57
- Type
- journal article
- Language
- en
- Open access
- closed
- Locations
- Epidemiology (publishedVersion)
- PubMed (publishedVersion)
- Topics
- Physical Sciences › Mathematics › Statistics and Probability › Advanced Causal Inference Techniques
- Physical Sciences › Mathematics › Statistics and Probability › Statistical Methods in Epidemiology
- Social Sciences › Arts and Humanities › History and Philosophy of Science › Philosophy of Biology and Scientific Explanation
- Keywords
- statistical interaction, synergistic effect, superadditivity, non-identifiability, intervention targeting
- MeSH terms
- Models, Theoretical, Terminology as Topic, Population Surveillance, Regression Analysis, Reproducibility of Results, Causality, Epidemiologic Studies, Risk Assessment
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
How to cite
Cite CoChat for the checks on this page, and the paper for its own findings.
- This page
- CoChat. (2026). Citation check: Interactions in Epidemiology: Relevance, Identification, and Estimation. https://app.cochat.ai/papers/10.1097/ede.0b013e318193e7b5
- The paper
- Greenland, S. (2009). Interactions in Epidemiology: Relevance, Identification, and Estimation. Epidemiology. https://doi.org/10.1097/ede.0b013e318193e7b5
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.