The STROBE Extensions

To the Editor:

A decade after the publication of the STROBE (STrengthening the Reporting of Observational studies in Epidemiology) Statement, we use this anniversary as a time to reflect on STROBE’s impact and future avenues for addressing the incomplete reporting of observational studies.^1,2 As an aid to authors, the STROBE Statement and an explanation and elaboration article were published in 2007 with generic guidance for reporting cohort, case–control, or cross-sectional studies. Subsequently, several extensions to STROBE were published, some including authors involved in the original Statement, to provide more nuanced and tailored guidance.^3–15 In principal, these efforts are valuable, but inconsistencies may arise because extension production is not coordinated, and there is no clear guidance on their creation.

We qualitatively assessed the published STROBE extensions to identity perceived gaps and deficiencies in the current STROBE checklist and to detect nonspecific or redundant guidance. As detailed in the protocol,¹⁶ as of 1 March2017, we found 13 STROBE extensions.^3–15 Collectively, there were 298 additions to the STROBE checklist (Table 1). Most additions were directly related to the field on which the extension was focused but, based on independent coding by two reviewers, over one third were not specific to the extension’s field. Rather, they were general epidemiologic or methodologic tenets applicable to most observational research (e.g., details about potential confounders, biases, etc.). The Methods section contained the most changed or added items, one third of which were nonspecific changes (Table 1).

TABLE 1.

Qualitative Assessment of Extensions to STROBE Checklist

Nonspecific additions were mainly in the following areas (Table 2):

TABLE 2.

Examples of Nonspecific Additions Added in STROBE Extensions

1. Participants, including sample size rationale, changes in exposure status, time points of assessment, and recruitment details;

2. Potential confounders and biases;

3. Subgroup and sensitivity analyses;

4. Generalizability;

5. Ethics disclosure/approval; and

6. Access to supplemental information (e.g., open source data, code, or protocols).

These results, highlighting nonspecific recommendations, complement previous research demonstrating particular problems with the reporting of bias, study size calculations, and subgroup and sensitivity analyses.^17,18 Nonspecific additions were of particular concern when they were found to be nearly identical to original STROBE checklist items (Table 3).

TABLE 3.

Examples of Redundant Suggestions

While the focused nature of the extensions varies widely, nonspecific additions could represent perceived gaps in content or indicate that information in the explanation and elaboration should be included in the checklist. Checklists provide valuable structure to research articles and serve as a reminder of what should be considered while writing. One cannot expect that all relevant epidemiologic or statistical information will be included; however, the trend of extensions adding general epidemiologic tenets points to a different reality.

The majority of additions made across the extensions were valuable, field-specific recommendations that experts in their respective disciplines determined necessary to report. However, nonspecific and redundant suggestions should not be ignored. EQUATOR (the Enhancing the QUAlity and Transparency Of health Research) Network guidance for guideline developers is a useful starting point to the process of how to develop an extension,¹⁹ but more direction is needed in terms of what to report about the process. For example, why it was deemed necessary to duplicate existing items in different words or to add nonspecific information.

Erik von Elm conceived of reporting guidelines as life jackets, not strait jackets.²⁰ STROBE is not meant to be a strict and rigid list, hence why many authors have used it as a base for their own more focused extensions. However, redundant or nonspecific content additions may create confusion rather than help. STROBE is an “evolving document that requires continual assessment, refinement, and if necessary change.”² The adaptable nature of STROBE is indispensable to its successful implementation. Updating STROBE was discussed at a 2010 meeting,²¹ but only minor revisions were identified, thus not justifying a new version of the guidelines; perhaps, this should now be reconsidered.

Melissa K. Sharp
Department of Psychology
Faculty of Humanities and Social Sciences
University of Split
Split, Croatia
INSERM
U1153 Epidemiology and Biostatistics
Sorbonne Paris Cité Research Center (CRESS)
Methods of therapeutic evaluation of chronic diseases Team (METHODS)
Paris France
Paris Descartes University
Sorbonne Paris Cité
France
msharp@unist.hr; melissa.sharp@etu.parisdescartes.fr

Darko Hren
Department of Psychology
Faculty of Humanities and Social Sciences
University of Split
Split, Croatia

Douglas G. Altman†
Centre for Statistics in Medicine
University of Oxford, Oxford
United Kingdom