Skip to main content
Download PDF

Application of causal inference methods in individual-participant data meta-analyses in medicine: addressing data handling and reporting gaps with new proposed reporting guidelines

BMC Medical Research Methodology volume 24, Article number: 91 (2024) Cite this article

Abstract

Observational data provide invaluable real-world information in medicine, but certain methodological considerations are required to derive causal estimates. In this systematic review, we evaluated the methodology and reporting quality of individual-level patient data meta-analyses (IPD-MAs) conducted with non-randomized exposures, published in 2009, 2014, and 2019 that sought to estimate a causal relationship in medicine. We screened over 16,000 titles and abstracts, reviewed 45 full-text articles out of the 167 deemed potentially eligible, and included 29 into the analysis. Unfortunately, we found that causal methodologies were rarely implemented, and reporting was generally poor across studies. Specifically, only three of the 29 articles used quasi-experimental methods, and no study used G-methods to adjust for time-varying confounding. To address these issues, we propose stronger collaborations between physicians and methodologists to ensure that causal methodologies are properly implemented in IPD-MAs. In addition, we put forward a suggested checklist of reporting guidelines for IPD-MAs that utilize causal methods. This checklist could improve reporting thereby potentially enhancing the quality and trustworthiness of IPD-MAs, which can be considered one of the most valuable sources of evidence for health policy.

Peer Review reports

Introduction

Randomized controlled trials (RCTs) are often considered the gold standard for establishing causal relationships. However, they may not always be feasible or ethical, particularly when dealing with exposures that cannot be randomized (e.g., cancer, obesity) or other exposures that would present ethical issues (e.g., Ebola Virus, smoking). Observational study designs are more often less resource-intensive and have the ability to evaluate the effects of a wider range of exposures than RCTs. This can allow for a larger number of individuals to be studied over a longer period of time.

In population health or global health science, where the goal is to make population-level inferences, meta-analyzing results from multiple studies can be an efficient and cost-effective way to increase statistical power and explore heterogeneity of single study findings across different sites, settings or populations [1]. There are two ways to conduct a meta-analysis (MA): pooling estimates (the traditional approach, also known as an aggregate data MA, which we do not review in this paper) and pooling individual-level patient data (IPD) to conduct a combined analysis. IPD-MAs are widely considered the gold standard in evidence-based medicine, as they often provide more precise and reliable estimates than MAs of aggregate data.

Aggregate data MAs may provide similar estimates to IPD-MAs in some settings [2, 3], but they are more prone to reporting bias [4], publication bias [5], or low statistical power [6]. IPD-MAs offer several other benefits, including the ability to adjust for confounders across studies, thereby minimizing the impact of between-study heterogeneity and reducing ecological bias [7, 8]. Moreover, data quality can be evaluated (e.g., study design features including randomization or follow-up) [9], IPD-MAs may have greater power to conduct subgroup analyses [9, 10], and they provide an opportunity to test assumptions of models and include unreported data [10]. However, IPD-MAs also present key challenges, such as accessing relevant data sources, data harmonization, and handling missing data in each study.

To address threats to internal validity present in observational studies when estimating causal effects in health science, the most commonly used approach is to include potential confounders as covariates in a standard regression-based adjustment (RBA) analysis, which we define as the investigation of a statistical relationship between a dependent and one (or more) explanatory variables. However, RBAs may not adequately control for measured confounding in the presence of time-varying confounders affected by prior treatment [11]. Analytical tools such as the G-methods (Marginal Structural Models [MSM], G-formula, and structural nested models) were developed to address these issues [12, 13]. Unmeasured confounding is another threat to internal validity in observational studies. In certain circumstances, data will allow for methods such as difference-in-differences [14], interrupted time series [15, 16], regression discontinuity design [17], and instrumental variables analysis [18, 19], methods which can circumvent unmeasured confounding. However, in other cases, including a sensitivity analysis for unmeasured confounders might be the only possible approach [20]. The strength of the inference relies not only on the method selected but also on the rigor with which the required assumptions are evaluated and tested.

Several reviews have shown that causal methods, which employ statistical techniques beyond above-mentioned standard RBA analyses, are implemented in single observational studies in medicine [21, 22]. However, a recent review [23] revealed that causal methods are rarely applied to IPD-MAs with infectious disease data. The objective of this systematic review is to expand on the previous review [23], and investigate the rigor in the implementation and reporting of causal methods in pooled longitudinal IPD studies in medicine.

Methods

Search strategy

The search strategy for this systematic review was developed by four researchers (HH, LM, EM, SR) and was reviewed and edited by information scientists from University Hospital Heidelberg (UKHD), University of California San Francisco (UCSF), and Harvard University. Similar to a previous review on infectious diseases [23], we chose not to include names of methods we considered “causal” but instead, allowed for methods not considered “causal”, such as standard RBAs, to be reviewed to prevent bias in the results. The search strategy was tailored to four large platforms so as to include non-medical disciplines (EBSCO [PsycINFO, Academic Search Complete, Business Source Premier, CINAHL, EconLit], EMBASE, PubMed and Web of Science). Details of the search strategy can be found in Supplementary Material 1.

Prior to initiating the systematic review, a protocol was registered with PROSPERO (CRD42020143148). Studies were included if they (1) posed a clear causal question related to the effect of an exposure on a health outcome, (2) estimated an effect size directly related to the causal question, and (3) pooled longitudinal individual-level data from more than one study or cohort. If a study pooled longitudinal data from RCTs, it was eligible for inclusion as long as not all of the exposure variables of interest were randomized (i.e., randomized exposures included in the pooled study must have been combined with non-randomized exposure variables). Furthermore, eligible studies had to be published (4) in the English language, (5) in peer-reviewed journals (accessible in full-text through open access, university licenses or project collaborators), and (6) in the years 2009, 2014, or 2019 (according to the electronic publication date). Due to resource constraints, the review was limited to publications at these three time-points, which were five years apart. Additional details about the study selection process are provided elsewhere [24].

Study selection process

Search results were deduplicated in Endnote [25], version X9. Titles, abstracts and full-texts articles were screened in COVIDENCE systematic review software [26] by two reviewers each (SC, HH, NM, EY) using the double-blind tool. Discrepancies were resolved by consensus. This review originally sought to investigate the rigor of causal methods implementation and reporting across academic disciplines. However, as there were too few studies meeting inclusion criteria in non-medical fields (5 of 210 articles after title-abstract screening), making rigorous comparisons of implementation and reporting across disciplines was not feasible. Therefore, studies from fields other than medicine were excluded, and the focus of this review therefore shifted from ‘across disciplines’ to ‘within medicine’. As the search returned more studies than could feasibly be reviewed, we selected a random sample of 20% (n = 24) of eligible records using a stratified random sampling approach based on the year of publication. Randomly selected articles that did not meet the inclusion criteria (n = 11) were replaced by another random sample (n = 21) taken from the remaining pool of eligible articles.

Data collection process

Data from each article were extracted using a predefined, peer-reviewed extraction form that consisted of over 70 points and was based on the PRISMA-IPD reporting guidelines [27] related to the pooling of studies (see Supplementary Material 2 Data Extraction Form). The extraction form also contained many reporting items related to causal methods implementation, as published in reporting guidelines for mediation analysis [28] and mendelian randomization [29]. Extracted data were cross-checked by at least two reviewers (SC, HH, NM, EY), and conflicts resolved by discussion or a tie-breaker (AD, VDJ). For each study, details such as (i) study design, (ii) statistical methods implemented, (iii) reporting of methods, and (iv) evaluation of assumptions were extracted.

Data analysis

To measure the quality of reporting across studies, we developed and applied a scoring system that consisted of the following domains: data harmonization, accounting for missing data, causal methods, data pooling, and confounder control. Each domain included specific criteria related to the quality of reporting within that domain and was weighted equally. If a specific item from the data extraction list was not mentioned in the study documentation, 0 points were awarded. If the item was alluded to but not clearly addressed, 0.5 points were awarded. If the item was clearly addressed, 1 point was awarded. Findings of this systematic review are reported following the 2020 PRISMA Statement [30].

Results

Study selection

The search strategy yielded 16,443 unique articles. Of the 210 articles which were eligible at the initial title-abstract phase, seven duplicates and 31 articles with e-publication dates other than eligible years were excluded, as well as the five non-medical articles (explained in sections 2.2 Study Selection process), resulting in 167 eligible medical articles for full-text review (2009, n = 23; 2014, n = 44; 2019, n = 100), general medicine (32), neoplasms (24), vascular disease (23), internal medicine (15), public health (15), nutritional sciences (9), neurology (7), endocrinology (7), surgery, other specialty (6), environmental health (5), psychiatry (5), communicable diseases (4), drugs therapy (2), genetics (2), geriatrics (2), pregnancy (2), therapeutics (2), allergy and immunology (1), complementary therapies (1), critical care (1), dentistry (1), and metabolism (1). See Supplementary Material 3 Full Article List for information on all 167 eligible articles as well as the 45 articles which were reviewed in both of the random samples. Of the 45 articles reviewed, 29 articles [31,32,33,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50,51,52,53,54,55,56,57,58] were included in the final analysis (see Fig. 1. PRISMA flow diagram).

Fig. 1
figure 1

PRISMA flow diagram. Note. ADMA = aggregate data meta-analyses, ePUB = electronic publication, RCTs = randomized controlled trials

Full size image

Study characteristics

Of the 29 IPD-MAs included in the final analysis, two were published in 2009, 10 in 2014, and 17 in 2019. The included IPD-MAs pooling data from cohort studies, RCTs, and case-control studies. The number of studies pooled in each IPD-MA ranged from 2 to 37, with an average of 12 studies. The pooled sample sizes ranged from 156 to 284,345 individuals. See the Supplementary Material 4. Details of Included Studies for more information.

Reporting results

Results of the data extraction are presented below. Please see Table 1 Results in tabular form and Fig. 2 Summary results for data items that received points for more details.

Table 1 Results in tabular form
Full size table

Data harmonization

IPD-MAs were awarded 17 points of 29 possible points for describing the definitions and measurements of the variables collected from the individual cohorts. For describing the differences in definitions and measurements among the variables pooled, IPD-MAs were awarded 15.5 of 29 possible points, with 11 IPD-MAs providing insufficient detail, earning them a half-point each. IPD-MAs received 24 of 29 possible points for describing their efforts to manage the differences in variable definitions and measurements from individual cohorts, such as through harmonization or standardization.

Accounting for missing data

18 of 29 possible points were awarded to IPD-MAs for describing the presence of missing data within and across studies (e.g. missing at random (MAR), missing not at random (MNAR), missing completely at random (MCAR)). Two IPD-MAs reported that the data was sporadically missing (i.e., data is partly missing in variables in one or more individual studies), and none of the IPD-MAs clearly labelled data as systematically missing (i.e., variables are entirely missing in one or more individual studies [59]). Twenty-one IPD-MAs were unclear about the type of missingness. One IPD-MA was awarded one point for a full description of why the data were missing, and 20.5 of 29 possible points were given to IPD-MAs for describing how the authors accounted for missing data.

Imputation model

Of the 10 IPD-MAs using imputation methods to account for missing data, five-and-a-half of 10 possible points were given for listing the variables included in the imputation model. None of the IPD-MAs were awarded points for justifying the choice of variables in the imputation model, and four of the 10 IPD-MAs received full points for describing efforts to account for potential heterogeneity between studies in the imputation models.

Data pooling

No points were awarded for discussing assumptions required for the methods they have used to pool the data, and no IPD-MA received points for describing the testing or evaluating of assumptions for the pooling method. Eight of the 29 IPD-MAs received full-points for clearly stating whether they implemented a one-step (n = 4) or two-step (n = 4) meta-analysis [60, 61].

Causal methods

Out of the 29 IPD-MAs, three used causal methods (one used mediation analysis; two used propensity scores), while the remaining 26 used standard regression-based analyses, including Cox proportional hazards regression, logistic regression, and linear regression. While most studies reported drawing data from longitudinal studies, exact time-points for variables included in the analyses was not clearly reported across IPD-MAs. Three-and-a-half of 29 possible points were awarded for justifying the choice of method used for causal inference or other statistical methods used. Four points out of 29 were awarded for explicitly stating assumptions required for the causal inference or statistical modelling approach selected. Three points were awarded for reporting the testing at least one of the testable assumptions, all of which were proportional hazards assumption. Two IPD-MAs discussed the evaluation of untestable assumptions (e.g., no unmeasured confounding) and thus received one point each. No IPD-MA implemented weighting in their causal analyses. 15.5 points were awarded to IPD-MAs for reporting that they investigated the potential for heterogeneity in the results. Eleven-and-a-half of 29 possible points were awarded to IPD-MAs for discussing the possible impact of any heterogeneity on the generalizability of the results. Twenty-three points were awarded for the reporting of sensitivity analyses.

Confounder control

Twenty-four points out of 29 were awarded for reporting the method(s) used to account for clustering/heterogeneity at the cohort level. These methods included stratification (n = 12), random effects (n = 9), interaction terms (n = 2), confounder adjustment (n = 2), and fixed-effects (n = 1). Seventeen-and-one-half points were awarded to IPD-MAs for indicating how the covariates were conceptualized (e.g., confounders, mediators), and 15.5 points were awarded for describing how they selected their covariates— e.g., two IPD-MAs [43, 55] reviewed the literature; one [36] used statistical testing procedures.

Effect estimates

Most IPD-MAs did not clearly report how they accounted for (potential) heterogeneity. Due to these ambiguities, we used a rough categorization based on our assumptions about the authors’ intentions: five IPD-MAs took strata-specific estimates; three IPD-MAs excluded specific patients or data points responsible for baseline heterogeneity; two IPD-MAs suggest the use of data standardization or harmonization was used to account for this; one IPD-MA reported that they found no baseline heterogeneity; and the remaining IPD-MAs (n = 17) only reported adjusting for variables. It was unclear for any of the studies which effect—marginal or conditional— was estimated. However, we inferred that 10 IPD-MAs estimated a conditional effect, one IPD-MA estimated a marginal effect, and four IPD-MAs estimated both. For the remaining IPD-MAs, no precise statement can be made.

Fig. 2
figure 2

Bowman [62]. Summary results for data items that received points

Full size image

Discussion

This systematic review evaluates the use of causal methods in IPD-MAs in medicine. Specifically, we investigated the implementation and reporting of methods to address causal questions and the critical role of data handling and reporting in this context. Overall, we found that the use of standard regression methods was the most common approach, and note the lack of utilization of other causal methods tailored to address time-varying and unmeasured confounding. While sensitivity analyses can be used to alleviate concerns with unmeasured confounding [20], these were not reported to have been used. We also observed major gaps in the reporting of the methodology used in pooled longitudinal, observational studies, including issues related to harmonization, missing data, and data standardization—all crucial to the implementation of causal methods. Table 2 provides guidance as to the various technical aspects that researchers engaging in IPD-MAs should consider for robust and reliable results.

Pooled studies

Our results suggest an increase in the number of pooled longitudinal observational studies within the medical field between 2009 and 2019. This upward trend mirrors the findings of an earlier review [7], and may reflects a greater understanding of the importance of IPD-MAs [63], improved digitization of records and data sharing efforts, and/or improved statistical software (both for conducting IPD-MAs as well as causal methodologies).

Causal methods

Although all included IPD-MAs were considered to have causal intent, all but three used standard RBA analyses. This finding is consistent with other reviews suggesting underutilization of causal methods in medicine or medical subfields [21, 23, 64]. This may be due to the high variability in data elements and study designs, which can pose challenges in applying certain methods for pooled data as compared to analyzing a single dataset. Alternatively, it could reflect a general lack of knowledge about or understanding of how to apply these methods to pooled studies. To address these issues, investigators may want to review (introductory) articles on causal methods [65,66,67]. We want to point out that no one method is universally better suited for one scientific field over another. Rather, each of these causal methods entail tradeoffs [68], and the choice of method should be determined by the research question at hand and the data available. In the context of causal analysis, researchers often rely on tools like Directed Acyclic Graphs (DAGs) to map the (assumed) relationships between factors, allowing them to identify potential confounders and mediators, e.g., and understand what variables need to be controlled or adjusted for [69]. This approach can provide valuable insights into the causal relationships under investigation and enhance the validity of results.

Table 2 Technical considerations required by IPD-MAs
Full size table

Several previous reviews have identified discrepancies in causal effects between RCTs and observational studies with the same exposures and outcomes [73]. As researchers in academia and industry are increasingly interested in the use of real-world evidence for regulatory decision making, IPD-MAs and other approaches to the pooled analyses of multiple longitudinal studies should consider applying causal methods (e.g., G methods with IV approaches) to account for time-varying confounding and unmeasured confounding.

Reporting guidelines

While many IPD-MAs included in this systematic review reported variable definitions, measurement methods, and efforts to harmonize data, they would benefit from reporting additional details like, e.g., between-study differences in measurement methods and variable definitions, which were rarely discussed and may affect the validity and interpretation of causal effect estimates.

When following best practices for harmonization [74], authors should consider reporting their detailed efforts in appendices, where there is ample space. However, we found that descriptions of data standardization were low, potentially due to a lack of lack of understanding of data standard availability, usability, or lack of feasibility of implementation based on specific needs of epidemiological studies.

Our review also revealed that few of the included IPD-MAs described the type of missing data present; either sporadically missing values, when these data are missing on observations within a particular study, or systematically missing, when variables are not defined consistently across studies and therefore missing entirely from specific studies [59]. Many studies simply omitted participants with missing data, using complete cases, a common practice in medicine. Although multiple imputation is generally recommended to account for missing data, the implementation becomes problematic when variable definitions or measurement methods differ across studies. For this reason, several multi-level imputation methods have been proposed that are better capable of preserving between-study heterogeneity and uncertainty when imputing missing data in IPD-MA and other types of pooled cohort studies [75].

The strength of causal inferences that can be made from any approach rely on the rigor with which assumptions were tested (if testable) or evaluated (if untestable). Although there are many forms of bias that can adversely influence the results of a study (e.g., reverse causation, measurement error), authors reported almost entirely on confounding bias. We would recommend that the authors explicitly report other forms of bias that they considered in their analysis, both to make the interpretability of their results transparent and to inform future studies on similar research questions.

There are reporting guidelines which exist in JAMA for mediation analyses [28] and mendelian randomization analyses [29], as well as reviews of and suggested reporting checklists for Instrumental Variable analyses [76, 77] but all of these publications appear to be intended for use in single studies. In addition, despite reporting guidelines existing for IPD-MAs [27], researchers have previously reported lower-than-desired reporting patterns from authors of IPD-MAs with regards to their statistical methods [23, 72]. There are currently no reporting guidelines for IPD-MAs which employ causal methods. We, therefore, propose that reporting guidelines for pooled studies employing causal methodologies be developed (see Supplementary Material 5 Proposed Reporting Guidelines Checklist for IPD-MAs implementing Causal Methods), based on the aforementioned published reporting guidelines (see Supplementary Material 6 Reporting Guidelines Comparison). We would appreciate any feedback to the proposed checklist.

Strength and limitations

Strengths of our systematic review include the search strategy, which was built on other systematic reviews of non-randomized exposures and reviews of similar methods and was also built in consultation with three experienced librarian scientists, and with input from colleagues in other fields regarding synonyms that could be used in other disciplines. The search strategy was also implemented in non-medical platforms to ensure potential identification of non-medical articles. The strategy also did not employ specific names, or variations of or acronyms of the names of the methods, so as not to bias our results by including only methods which we considered “causal”. Another strength is the sheer number of articles that we screened and data points that we extracted—nearly four and six times the number of titles and abstracts screened by similar reviews [21, 22], and far exceeded the extraction numbers of those same reviews (five and 24 items to our 70 + items).

Weaknesses of our systematic review include the scarcity of articles found from disciplines other than medicine. This low number could be because health outcomes from pooled data sets are not often being investigated in disciplines other than medicine, but we must also consider the possibility that this small number we found is related to the search strategy, despite the rigor with which it was built. We also recognize that the global generalizability of our results is limited due to language and year restrictions. We must also consider that, although we appeared to reach theoretical saturation, our findings may be limited by the fact we were only able to review roughly 27% of potentially-eligible full-text articles. Another limitation is that the review was limited to the reporting of the use of causal methods in IPD-MAs with in medicine which may not reflect what was actually done. Further, it is possible that we included some studies that did not (primarily) aim to infer a causal relationship as the study aims were not always entirely clear. We attempted to counter subjectivity with blind assessment by at least two reviewers per study, rounds of discussion between reviewers in case of disagreement, and consultation with additional scientists from four universities across four countries.

Conclusions

To encourage better reporting and implementation of causal methods in future pooled longitudinal IPD studies, we propose the following approaches. First, we suggest that authors always clearly describe their methods. The domain criteria evaluated in this study can serve as a basis for developing or building on existing reporting standards. Although most medical journals set a predefined word limit for publications, the appendix, which usually has no word limit, is a simple way to include an in-depth description and justification of each aspect of the methodological approach. Second, the research community could publish accessible “how-to” documents that apply causal inference methods to pooled IPD studies and are accompanied by open-source data and code to ensure that investigators can better apply these methods to studies that pool longitudinal, observational data. This will lower the barrier to engaging with appropriate and potentially unfamiliar methods and could ultimately increase application of these methods in the broader research contexts, as well as inform health policy and decision making.

Availability of data and materials

Data and materials can be found in the Supplementary Materials 1, 2, 3, 4, 5 and 6.

Abbreviations

RCTs:

Randomized controlled trials

IPD-MA:

Individual-level patient data meta-analysis

RBA:

Regression-based adjustment

References

  1. van der Steen JT, Kruse RL, Szafara KL, van der Mehr DR, Ribbe MW, et al. Benefits and pitfalls of pooling datasets from comparable observational studies: combining US and Dutch nursing home studies. Palliat Med. 2008;22(6):750–9.

    Article  PubMed  Google Scholar 

  2. Tudor Smith C, Marcucci M, Nolan S, Iorio A, Sudell M, Riley R, et al. Individual participant data meta-analyses compared with meta-analyses based on aggregate data (Review). 2016.

  3. Tierney JF, Fisher DJ, Burdett S, Stewart LA, Parmar MKB. Comparison of aggregate and individual participant data approaches to meta-analysis of randomised trials: an observational study. PLoS Med. 2020;17(1):1–22.

    Article  Google Scholar 

  4. McCormack K, Grant A, Scott N. Value of updating a systematic review in surgery using individual patient data. Br J Surg. 2004;91(4):495–9.

    Article  CAS  PubMed  Google Scholar 

  5. Jeng GT, Scott JR, Burmeister LF. A comparison of meta-analytic results using literature vs individual patient data: paternal cell immunization for recurrent miscarriage. JAMA. 1995;274(10):830–6. https://doi.org/10.1001/jama.1995.03530100070037.

    Article  CAS  PubMed  Google Scholar 

  6. Berlin JA, Santanna J, Schmid CH, Szczech LA, Feldman HI. Individual patient- versus group-level data meta-regressions for the investigation of treatment effect modifiers: ecological bias rears its ugly head. Stat Med. 2002;21(3):371–87.

    Article  PubMed  Google Scholar 

  7. Riley RD, Lambert PC, Abo-Zaid G. Meta-analysis of individual participant data: Rationale, conduct, and reporting. BMJ. 2010;340(7745):521–5.

    Google Scholar 

  8. Riley RD. Commentary: like it and lump it? Meta-analysis using individual participant data. Int J Epidemiol. 2010;39(5):1359–61.

    Article  PubMed  Google Scholar 

  9. Stewart LA, Michael JC. Practical methodology of meta-analyses (overviews) using updated individual patient data. Stat Med. 1995;14(19):2057–79.

    Article  CAS  PubMed  Google Scholar 

  10. Tierney JF, Vale C, Riley R, Smith CT, Stewart L, Clarke M, et al. Individual participant data (IPD) metaanalyses of randomised controlled trials: Uidance on their use. PLoS Med. 2015;12(7):1–16.

    Article  Google Scholar 

  11. Robins JM, Hernán MÁ, Brumback B. Marginal structural models and causal inference in epidemiology. Epidemiology. 2000;11(5):550–60.

    Article  CAS  PubMed  Google Scholar 

  12. Mansournia MA, Etminan M, Danaei G, Kaufman JS, Collins G. Handling time varying confounding in observational research. BMJ. 2017;359:1–6.

    Google Scholar 

  13. Doosti-Irani A, Mansournia MA, Collins G. Use of G-methods for handling time-varying confounding in observational research. Lancet Glob Heal. 2019;7(1):e35. https://doi.org/10.1016/S2214-109X(18)30471-6.

    Article  Google Scholar 

  14. Streeter AJ, Lin NX, Crathorne L, Haasova M, Hyde C, Melzer D, et al. Adjusting for unmeasured confounding in nonrandomized longitudinal studies: a methodological review. J Clin Epidemiol. 2017;87:23–34.

    Article  PubMed  PubMed Central  Google Scholar 

  15. Kontopantelis E, Doran T, Springate DA, Buchan I, Reeves D. Regression based quasi-experimental approach when randomisation is not an option: interrupted time series analysis. BMJ. 2015;350:1–4.

    Article  Google Scholar 

  16. Hudson J, Fielding S, Ramsay CR. Methodology and reporting characteristics of studies using interrupted time series design in healthcare. BMC Med Res Methodol. 2019;19(1):1–7.

    Article  CAS  Google Scholar 

  17. Thistlewaite DL, Campbell DT. Regression-discontinuity analysis: an alternative to the ex-post facto experiment. J Educ Psychol. 1960;51:309–17. (https://obsstudies.org/reprint-regression-discontinuity-analysis-an-alternative-to-the-ex-post-facto-experiment/).

    Article  Google Scholar 

  18. Baiocchi M, Cheng J, Small DS. Instrumental variable methods for causal inference. Stat Med. 2014;33(13):2297–340.

    Article  PubMed  PubMed Central  Google Scholar 

  19. Boef AGC, Dekkers OM, Vandenbroucke JP, Le Cessie S. Sample size importantly limits the usefulness of instrumental variable methods, depending on instrument strength and level of confounding. J Clin Epidemiol. 2014;67(11):1258–64. https://doi.org/10.1016/j.jclinepi.2014.05.019.

    Article  PubMed  Google Scholar 

  20. D’Agostino ML. Sensitivity analyses for unmeasured confounders. Curr Epidemiol Reports. 2022;9(4):361–75. https://doi.org/10.1007/s40471-022-00308-6.

    Article  Google Scholar 

  21. Clare PJ, Dobbins TA, Mattick RP. Causal models adjusting for time-varying confounding-a systematic review of the literature. Int J Epidemiol. 2019;48(1):254–65. https://doi.org/10.1093/ije/dyy218.

    Article  PubMed  Google Scholar 

  22. Farmer RE, Kounali D, Walker AS, Savović J, Richards A, May MT, et al. Application of causal inference methods in the analyses of randomised controlled trials: a systematic review. Trials. 2018;19(1):1–14.

    Article  Google Scholar 

  23. Hufstedler H, Rahman S, Danzer AM, Goymann H, de Jong VMT, Campbell H, et al. Systematic review reveals lack of causal methodology applied to pooled longitudinal observational infectious disease studies. J Clin Epidemiol. 2022;145:29–38. https://doi.org/10.1016/j.jclinepi.2022.01.008.

    Article  PubMed  Google Scholar 

  24. Yeboah E, Mauer NS, Hufstedler H, Carr S, Matthay EC, Maxwell L, et al. Current trends in the application of causal inference methods to pooled longitudinal non-randomised data: a protocol for a methodological systematic review. BMJ Open. 2021;11(11):1–5.

    Article  Google Scholar 

  25. Team TE. EndNote. Philadelphia: Clarivate; 2013.

    Google Scholar 

  26. Covidence systematic review software. Melbourne: Veritas Health Innovation. Available from: http://www.covidence.org.

  27. Stewart LA, Clarke M, Rovers M, Riley RD, Simmonds M, Stewart G, et al. Preferred reporting items for a systematic review and meta-analysis of individual participant data: the PRISMA-IPD statement. JAMA. 2015;313(16):1657–65.

    Article  PubMed  Google Scholar 

  28. Lee H, Cashin AG, Lamb SE, Hopewell S, Vansteelandt S, Vanderweele TJ, et al. A Guideline for reporting mediation analyses of randomized trials and observational studies: the AGReMA Statement. JAMA - J Am Med Assoc. 2021;326(11):1045–56.

    Article  Google Scholar 

  29. Skrivankova VW, Richmond RC, Woolf BAR, Yarmolinsky J, Davies NM, Swanson SA, et al. Strengthening the reporting of observational studies in epidemiology using mendelian randomization: the STROBE-MR Statement. JAMA - J Am Med Assoc. 2021;326(16):1614–21.

    Article  Google Scholar 

  30. Page MJ, McKenzie JE, Bossuyt PM, et al. The PRISMA 2020 statement: an updated guideline for reporting systematic reviews. BMJ. 2021;372:n71. https://doi.org/10.1136/bmj.n71.

  31. Gibson TM, Park Y, Robien K, Shiels MS, Black A, Sampson JN, et al. Body mass index and risk of second obesity-associated cancers after colorectal cancer: A pooled analysis of prospective cohort studies. J Clin Oncol. 2014;32(35):4004–11. Available from: http://www.embase.com/search/results?subaction=viewrecord&from=export&id=L600782409).

    Article  PubMed  PubMed Central  Google Scholar 

  32. Martin ET, Krantz E, Gottlieb SL, Magaret AS, Langenberg A, Stanberry L, et al. A pooled analysis of the effect of condoms in preventing HSV-2 acquisition. Arch Intern Med. 2009;169(13 PG-1233–40):1233–40.

    Article  PubMed  PubMed Central  Google Scholar 

  33. Keller A, O’Reilly EJ, Malik V, Buring JE, Andersen I, Steffen L, et al. Substitution of sugar-sweetened beverages for other beverages and the risk of developing coronary heart disease: Results from the Harvard Pooling Project of Diet and Coronary Disease. Prev Med (Baltim). 2020;131(December 2019):105970. https://doi.org/10.1016/j.ypmed.2019.105970.

    Article  Google Scholar 

  34. Mondul AM, Shui IM, Yu K, et al. Vitamin-D associated genetic variation and risk of breast cancer in the breast and prostate cancer cohort consortium (BPC3). Cancer Epidemiol Biomarkers Prev. 2016;24(3):627–30.

    Article  Google Scholar 

  35. Crowe FL, Appleby PN, Travis RC, Barnett M, Brasky TM, Bueno-De-mesquita HB, et al. Circulating fatty acids and prostate cancer risk: Individual participant meta-analysis of prospective studies. J Natl Cancer Inst. 2014;106(9). https://doi.org/10.1093/jnci/dju240, https://www.embase.com/search/results?subaction=viewrecordandid=L604449288andfrom=export.

  36. Dilworth TJ, Casapao AM, Ibrahim OM, Mercierg RC. Adjuvant ␤-Lactam therapy combined with vancomycin for methicillin-resistant staphylococcus aureus bacteremia: does ␤-Lactam class matter? Antimicrob Agents Chemother. 2019;63(3):1–4.

    Article  Google Scholar 

  37. Bethea TN, Kitahara CM, Sonderman J, Patel AV, Harvey C. APooled analysis of body mass index and pancreatic cancer mortality in African Americans. Cancer Epidemiol Biomarkers Prev. 2014;23(10):2119–25. http://www.embase.com/search/results?subaction=viewrecordandfrom=exportandid=L600120623.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  38. Bastos ML, Hussain H, Weyer K, Garcia-Garcia L, Leimane V, Leung CC, et al. Treatment outcomes of patients with multidrug-resistant and extensively drug-resistant tuberculosis according to drug susceptibility testing to first- and second-line drugs: an individual patient data meta-analysis. Clin Infect Dis. 2014;59(10):1364–74.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  39. Grams ME, Surapaneni A, Ballew SH, Appel LJ, Boerwinkle E, Boulware LE, et al. APOL1 kidney risk variants and cardiovascular disease: An individual participant data meta-analysis. J Am Soc Nephrol. 2019;30(10):2027–36. https://doi.org/10.1681/ASN.2019030240.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  40. Voerman E, Santos S, Golab BP, Amiano P, Ballester F, Barros H, et al. Maternal body mass index, gestational weight gain, and the risk of overweight and obesity across childhood: An individual participant data meta-analysis. PLoS Med. 2019;16(2):e1002744. https://doi.org/10.1371/journal.pmed.1002744.

    Article  PubMed  PubMed Central  Google Scholar 

  41. Jokela M, Airaksinen J, Virtanen M, Batty GD, Kivimäki M, Hakulinen C. Personality, disability-free life years, and life expectancy: individual participant meta-analysis of 131,195 individuals from 10 cohort studies. J Pers. 2020;88(3):596–605.

    Article  PubMed  Google Scholar 

  42. Yamamoto K, Shiomi H, Morimoto T, Natsuaki M, Takeji Y, Watanabe H, et al. Effect of renal dysfunction on the risks for ischemic and bleeding events in patients with atrial fibrillation receiving percutaneous coronary intervention. Am J Cardiol. 2020;125(3):399–408. https://doi.org/10.1016/j.amjcard.2019.10.049.

    Article  PubMed  Google Scholar 

  43. Rogozińska E, Zamora J, Marlin N, Betrán AP, Astrup A, Bogaerts A, et al. Gestational weight gain outside the Institute of Medicine recommendations and adverse pregnancy outcomes: Analysis using individual participant data from randomised trials. BMC Pregnancy Childbirth. 2019;19(1). Available from: https://www.embase.com/search/results?subaction=viewrecord&id=L629151253&from=export. https://doi.org/10.1186/s12884-019-2472-7.

  44. Elke G, Wang M, Weiler N, Day AG, Heyland DK. Close to recommended caloric and protein intake by enteral nutrition is associated with better clinical outcome of critically ill septic patients: secondary analysis of a large international nutrition database. Crit Care. 2014;18(1):1–8 (Available from: Critical Care).

    Article  Google Scholar 

  45. Bosetti C, Rosato V, Li D, Silverman D, Petersen GM, Bracci PM, et al. Diabetes, antidiabetic medications, and pancreatic cancer risk: an analysis from the International pancreatic cancer case-control consortium. Ann Oncol. 2014;25(10):2065–72.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  46. Gall S, Huynh QL, Magnussen CG, Juonala M, Viikari JSA, Kähönen M, et al. Exposure to parental smoking in childhood or adolescence is associated with increased carotid intima-media thickness in young adults: evidence from the cardiovascular risk in young finns study and the childhood determinants of adult health study. Eur Heart J. 2014;35(36):2484–91.

    Article  PubMed  Google Scholar 

  47. Hansson J, Galanti MR, Hergens MP, Fredlund P, Ahlbom A, Alfredsson L, et al. Snus (Swedish smokeless tobacco) use and risk of stroke: pooled analyses of incidence and survival. J Intern Med. 2014;276(1):87–95.

    Article  CAS  PubMed  Google Scholar 

  48. Kuramatsu JB, Biffi A, Gerner ST, Sembill JA, Sprügel MI, Leasure A, et al. Association of surgical hematoma evacuation vs conservative treatment with functional outcome in patients with cerebellar intracerebral hemorrhage. JAMA - J Am Med Assoc. 2019;322(14):1392–403.

    Article  Google Scholar 

  49. Fretts AM, Imamura F, Marklund M, Micha R. Associations of circulating very-long-chain saturated fatty acids and incident type 2 diabetes: A pooled analysis of prospective cohort studies. Am J Clin Nutr. 2019;109(4):1216–23. (http://www.embase.com/search/results?subaction=viewrecordandfrom=exportandid=L628286599).

    Article  PubMed  PubMed Central  Google Scholar 

  50. Marklund M, Wu JHY, Imamura F, Del Gobbo LC, Fretts A. Biomarkers of dietary omega-6 fatty acids and incident cardiovascular disease and mortality: an individual-level pooled analysis of 30 cohort Studies. Circulation. 2019;139(21):2422–36. Available from: http://www.embase.com/search/results?subaction=viewrecord&from=export&id=L629616822.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  51. Danforth K, Townsend MK, Curhan GC, Resnick N, Grodstein F. Type 2 diabetes mellitus and risk of stress, urge, and mixedurinary incontinence. J Urol. 2009;181(1):193–7.

    Article  PubMed  Google Scholar 

  52. Peres LC, Mallen AR, Townsend MK, Poole EM, Trabert B, Allen NE, et al. High levels of C-reactive protein are associated with an increased risk of ovarian cancer: results from the ovarian cancer cohort consortium. Cancer Res. 2019;79(20):5442–51.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  53. Sun L, Yim WS, Fahey P, Wang S. Investigation on advanced non-small-cell lung cancer among elderly patients treated with Chinese herbal medicine versus chemotherapy: a pooled analysis of individual data. Evid Based Complement Altern Med. 2019;2019. Available from: http://www.embase.com/search/results?subaction=viewrecord&from=export&id=L625956756.

  54. Leon ME, Schinasi LH, Lebailly P, Beane Freeman LE. Pesticide use and risk of non-Hodgkin lymphoid malignancies in agricultural cohorts from France, Norway and the USA: A pooled analysis from the AGRICOH consortium. Int J Epidemiol. 2019;48(5):1519–35. Available from: http://www.embase.com/search/results?subaction=viewrecord&from=export&id=L630105407).

    Article  PubMed  PubMed Central  Google Scholar 

  55. Hach M, Christensen LB, Lange T, Hvidtfeldt UA, Danielsen B, Diderichsen F, et al. Social inequality in tooth loss, the mediating role of smoking and alcohol consumption. Community Dent Oral Epidemiol. 2019;47(5):416–23.

    Article  PubMed  Google Scholar 

  56. Yang JJ, Yu D, Wen W, Shu XO, Saito E, Rahman S, et al. Tobacco smoking and mortality in Asia: A pooled meta-analysis. JAMA Netw open. 2019;2(3):e191474. Available from: http://www.embase.com/search/results?subaction=viewrecord&from=export&id=L627148837.

    Article  PubMed  PubMed Central  Google Scholar 

  57. Ding J, Davis-Plourde KL, Sedaghat S, Tully PJ, Wang W, Phillips C, et al. Antihypertensive medications and risk for incident dementia and Alzheimer’s disease: a meta-analysis of individual participant data from prospective cohort studies. Lancet Neurol. 2020;19(1):61–70.

    Article  CAS  PubMed  Google Scholar 

  58. Rota M, Alicandro G, Pelucchi C, Bonzi R, Bertuccio P, Hu J, et al. Education and gastric cancer risk—An individual participant data meta-analysis in the StoP project consortium. Int J Cancer. 2020;146(3):671–81.

    Article  CAS  PubMed  Google Scholar 

  59. Resche-Rigon M, White IR, Bartlett JW, Peters SAE, Thompson SG. Multiple imputation for handling systematically missing confounders in meta-analysis of individual participant data. Stat Med. 2013;32(28):4890–905.

    Article  PubMed  Google Scholar 

  60. Fisher DJ, Copas AJ, Tierney JF, Parmar MKB. A critical review of methods for the assessment of patient-level interactions in individual participant data meta-analysis of randomized trials, and guidance for practitioners. J Clin Epidemiol. 2011;64(9):949–67. https://doi.org/10.1016/j.jclinepi.2010.11.016.

    Article  CAS  PubMed  Google Scholar 

  61. Debray TPA, Moons KGM, Abo-Zaid GMA, Koffijberg H, Da Riley R. Individual participant data meta-analysis for a binary outcome: one-stage or two-stage? PLoS One. 2013;8(4):e60650.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  62. Bowman KC. GitHub. Radial bar chart part 3. 2016. Available from: https://plnkr.co/edit/vIwzn3piaBSqTfmVxrKt?p=preview&preview. Cited 2021 Nov 30.

  63. Katkade VB, Sanders KN, Zou KH. Real world data: an opportunity to supplement existing evidence for the use of long-established medicines in health care decision making. J Multidiscip Healthc. 2018;11:295–304.

    Article  PubMed  PubMed Central  Google Scholar 

  64. Moscoe E, Bor J, Bärnighausen T. Regression discontinuity designs are underutilized in medicine, epidemiology, and public health: a review of current and best practice. J Clin Epidemiol. 2015;68(2):122–33. https://doi.org/10.1016/j.jclinepi.2014.06.021.

    Article  PubMed  Google Scholar 

  65. Hernán MA, Brumback B, Robins JM. Marginal structural models to estimate the joint causal effect of nonrandomized treatments. J Am Stat Assoc. 2001;96(454):440–8.

    Article  Google Scholar 

  66. Jacob RT, Zhu P, Somers MA, Bloom HS. A practical guide to regression discontinuity. 2012. Available from: http://www.mdrc.org/publications/644/full.pdf.

  67. Lousdal ML. An introduction to instrumental variable assumptions, validation and estimation. Emerg Themes Epidemiol. 2018;15(1):1–7. https://doi.org/10.1186/s12982-018-0069-7.

    Article  PubMed  PubMed Central  Google Scholar 

  68. Matthay EC, Hagan E, Gottlieb LM, Tan ML, Vlahov D, Adler NE, et al. Alternative causal inference methods in population health research: Evaluating tradeoffs and triangulating evidence. SSM - Popul Heal. 2020;10:100526. https://doi.org/10.1016/j.ssmph.2019.100526.

    Article  Google Scholar 

  69. Greenland S, Pearl J, Robins JM. Causal diagrams for epidemiologic research. Epidemiology. 1999;10(1):37–48. Available from: http://www.jstor.org/stable/3702180).

    Article  CAS  PubMed  Google Scholar 

  70. Trivella M, Pezzella F, Pastorino U, Harris AL, Altman DG. Microvessel density as a prognostic factor in non-small-cell lung carcinoma: a meta-analysis of individual patient data. Lancet Oncol. 2007;8(6):488–99.

    Article  PubMed  Google Scholar 

  71. Riley RD, Lambert PC, Staessen JA, Wang J, Gueyffier F, Thijs L, et al. Meta-analysis of continuous outcomes combining individual patient data and aggregate data. Stat Med. 2008;27:1870–93.

    Article  PubMed  Google Scholar 

  72. Simmonds M, Stewart G, Stewart L. A decade of individual participant data meta-analyses: A review of current practice. Contemp Clin Trials. 2015;45:76–83. https://doi.org/10.1016/j.cct.2015.06.012.

    Article  PubMed  Google Scholar 

  73. Ramagopalan SV, Simpson A, Sammon C. Can real-world data really replace randomised clinical trials? BMC Med. 2020;18(1):13. https://doi.org/10.1186/s12916-019-1481-8.

    Article  PubMed  PubMed Central  Google Scholar 

  74. Fortier I, Raina P, Heuvel ER, Van Den, Griffith LE, Craig C, Saliba M, et al. Maelstrom research guidelines for rigorous retrospective data harmonization. 2017;(June 2016):103–15.

  75. Audigier V, White I, Jolani S, Debray T, Carpenter J, Van Buuren S, et al. Multiple imputation for multilevel data with continuous and binary variables. Stat Sci. 2018;3(2):160–83.

    Google Scholar 

  76. Davies NM, Smith GD, Windmeijer F, Martin RM. Issues in the reporting and conduct of instrumental variable studies: a systematic review. Epidemiology. 2013;24(3):363–9.

    Article  PubMed  Google Scholar 

  77. Swanson SA, Hernán MA. Commentary: How to report instrumental variable analyses (suggestions welcome). Epidemiology. 2013;24(3):370–4.

    Article  PubMed  Google Scholar 

Download references

Acknowledgements

We would like to sincerely thank Leonard Levin, Deputy Director of the Francis A. Countway Library of Medicine at Harvard Medical School, for his insight, patience and dedication to this project. For the publication fee we acknowledge financial support by Heidelberg University. We would also like to thank the editors and reviewers of the BMC Medical Research Methodology journal for their valuable time.

Registration and protocol (registration)

The protocol for the systematic review and the systematic review were registered with PROSPERO (registration number: CRD42020143148) and published in BMJ Open [24].

Funding

Open Access funding enabled and organized by Projekt DEAL. This review is part of the ReCoDID study funded by the European Union’s Horizon 2020 Research and Innovation Programme [grant number 825746] and the Canadian Institutes of Health Research, Institute of Genetics (CIHR-IG) [grant number N.01886-000]. Dr. Matthay is funded by National Institute on Alcohol Abuse and Alcoholism [grant number 1K99AA028256-01]. The authors declare that funders had no role in the review and have no competing interests.

Author information

Author notes

  1. Heather Hufstedler, Nicole Mauer, Edmund Yeboah and Sinclair Carr contributed equally as co-first authors.

  2. Ellicott C. Matthay and Till Bärnighausen contributed equally as co-last authors.

Authors and Affiliations

  1. Heidelberg Institute of Global Health, Faculty of Medicine and University Hospital, Heidelberg University, Heidelberg, Germany

    Heather Hufstedler, Nicole Mauer, Edmund Yeboah, Sinclair Carr, Lauren Maxwell, Thomas Jaenisch & Till Bärnighausen

  2. Center for Interdisciplinary Addiction Research, University Medical Center Hamburg-Eppendorf, Hamburg, Germany

    Sinclair Carr

  3. University of Massachusetts Medical School, University of Massachusetts, Worcester, USA, MA

    Sabahat Rahman

  4. KU Eichstätt-Ingolstadt, Ingolstadt School of Management and Economics (WFI), Ingolstadt, Germany

    Alexander M. Danzer

  5. IZA, Bonn, Germany

    Alexander M. Danzer

  6. CESifo, Munich, Germany

    Alexander M. Danzer

  7. Julius Center for Health Sciences and Primary Care, University Medical Center Utrecht, Utrecht University, Utrecht, Netherlands

    Thomas P. A. Debray & Valentijn M.T. de Jong

  8. Cochrane Netherlands, University Medical Center Utrecht, Utrecht University, Utrecht, Netherlands

    Valentijn M.T. de Jong

  9. Smart Data Analysis and Statistics B.V, Utrecht, The Netherlands

    Thomas P. A. Debray

  10. Department of Statistics, University of British Columbia, Vancouver, Canada, BC

    Harlan Campbell & Paul Gustafson

  11. Center for Global Health, Colorado School of Public Health, Aurora, USA, CO

    Thomas Jaenisch

  12. Department of Epidemiology, Colorado School of Public Health, Aurora, USA

    Thomas Jaenisch

  13. Department of Population Health, New York University Grossman School of Medicine, New York City, USA, NY

    Ellicott C. Matthay

  14. Harvard T H Chan School of Public Health, Harvard University, Boston, USA, MA

    Till Bärnighausen

  15. Department of Epidemiology, Harvard T.H. Chan School of Public Health, Harvard University, Boston, MA, USA

    Sinclair Carr

Authors
  1. Heather Hufstedler
    View author publications

    You can also search for this author in PubMed Google Scholar

  2. Nicole Mauer
    View author publications

    You can also search for this author in PubMed Google Scholar

  3. Edmund Yeboah
    View author publications

    You can also search for this author in PubMed Google Scholar

  4. Sinclair Carr
    View author publications

    You can also search for this author in PubMed Google Scholar

  5. Sabahat Rahman
    View author publications

    You can also search for this author in PubMed Google Scholar

  6. Alexander M. Danzer
    View author publications

    You can also search for this author in PubMed Google Scholar

  7. Thomas P. A. Debray
    View author publications

    You can also search for this author in PubMed Google Scholar

  8. Valentijn M.T. de Jong
    View author publications

    You can also search for this author in PubMed Google Scholar

  9. Harlan Campbell
    View author publications

    You can also search for this author in PubMed Google Scholar

  10. Paul Gustafson
    View author publications

    You can also search for this author in PubMed Google Scholar

  11. Lauren Maxwell
    View author publications

    You can also search for this author in PubMed Google Scholar

  12. Thomas Jaenisch
    View author publications

    You can also search for this author in PubMed Google Scholar

  13. Ellicott C. Matthay
    View author publications

    You can also search for this author in PubMed Google Scholar

  14. Till Bärnighausen
    View author publications

    You can also search for this author in PubMed Google Scholar

Contributions

Heather Hufstedler: Conceptualization; Data curation; Formal analysis; Investigation; Methodology; Project administration; Supervision; Validation; Visualization; Writing - original draft; Writing - review and editing. Nicole Mauer: Data curation; Formal analysis; Investigation; Methodology; Validation; Visualization; Writing - original draft; Writing - review and editing. Edmund Yeboah: Data curation; Formal analysis; Investigation; Methodology; Validation; Visualization; Writing - original draft; Writing - review and editing Sinclair Carr: Data curation; Formal analysis; Investigation; Methodology; Validation; Visualization; Writing - original draft; Writing - review and editing. Sabahat Rahman: Data curation; Formal analysis; Investigation; Methodology. Alexander M. Danzer: Data curation; Formal analysis; Investigation; Methodology; Supervision; Validation; Visualization; Writing - review and editing. Thomas P.A. Debray: Methodology; Supervision; Validation; Visualization; Writing - review and editing. Valentijn de Jong: Methodology; Validation; Visualization; Writing - review and editing. Harlan Campbell: Methodology; Visualization; Writing - review and editing. Paul Gustafson: Methodology; Visualization; Writing - review and editing. Lauren Maxwell: Conceptualization; Funding acquisition; Methodology; Project administration; Supervision; Writing - review and editing. Thomas Jaenisch: Funding acquisition. Ellicott Matthay: Conceptualization; Funding acquisition; Methodology; Project administration; Supervision; Validation; Visualization; Writing - original draft; Writing - review and editing. Till Bärnighausen: Funding acquisition; Conceptualization; Supervision; Writing - review and editing.

Corresponding author

Correspondence to Heather Hufstedler.

Ethics declarations

Ethics approval and consent to participate

Not applicable.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.

Additional information

Publisher’s Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary Information

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated in a credit line to the data.

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Hufstedler, H., Mauer, N., Yeboah, E. et al. Application of causal inference methods in individual-participant data meta-analyses in medicine: addressing data handling and reporting gaps with new proposed reporting guidelines. BMC Med Res Methodol 24, 91 (2024). https://doi.org/10.1186/s12874-024-02210-9

Download citation

  • Received:

  • Accepted:

  • Published:

  • DOI: https://doi.org/10.1186/s12874-024-02210-9

Keywords

BMC Medical Research Methodology

ISSN: 1471-2288

Contact us