Using Big Data Wisely

In recent months, the editorial team has seen a rapid rise in submissions that use large-scale electronic health records (EHRs), claims data, and administrative databases. These data sources offer significant advantages, including large sample sizes and the ability to investigate rare events across multiple institutions. However, their increasing prevalence has underscored a critical reality: not all clinical questions are equally suited to these data sources. Although database studies can effectively examine population-level trends and hard clinical events, they lack the granularity needed to address many questions central to foot and ankle practice. The data do not capture the nuances of surgeon expertise, surgical indications, or patient compliance, nor do they consistently capture variables such as radiographic alignment, reduction quality, or bone healing. When these limitations are overlooked, large data sets can yield results that are statistically precise yet clinically misleading.

The value of database research depends not on sample size alone, but on alignment between the research question, the outcomes assessed, and what conclusions the data can support. For this reason, Foot & Ankle International (FAI) and Foot & Ankle Orthopaedics (FAO) have adopted structured guidelines for evaluating database studies, with distinct standards reflecting each journal’s mission. FAI publishes studies that support decisive clinical inference—meaning we have a reasonable degree of confidence that observed outcomes are accurately attributed to the index procedure and not driven by misclassification or uncontrolled confounding. This requires procedure-confirmed endpoints (eg, revision, reoperation, implant removal) and rigorous attribution to the correct side and anatomic site. FAO publishes methodologically rigorous studies that provide population-level insights and hypothesis-generating findings, including those that rely on diagnosis codes, health care utilization measures, or risk stratification, with the expectation that limited inference is explicitly acknowledged. Studies based primarily on diagnosis codes or indirect outcome measures are generally better suited for FAO; studies with procedure-confirmed endpoints and rigorous laterality attribution are better suited for FAI.

Methodologic Standards

Both journals now require that exposure groups be clinically homogeneous. Combining distinct conditions or procedures with different severities or prognoses, for example, grouping all ankle fractures together, creates case-mix confounding that statistical adjustment cannot address. Studies that group heterogeneous entities without clinical justification and without supporting subgroup analysis will be rejected by either journal.

For comparative effectiveness studies, matching or adjustment limited to demographics (age, sex, race, body mass index) without validated comorbidity measures (eg, the Charlson or Elixhauser indices) is insufficient for health status–dependent outcomes and will not be considered for publication. When propensity score matching is used, authors must report the matching algorithm, caliper width, matching ratio, and balance assessment using standardized mean differences (SMDs). Both journals require SMD <0.1. FAO may accept SMD between 0.1 and 0.2 when accompanied by sensitivity analyses demonstrating robust conclusions; SMD >0.2 will result in rejection.

Anatomic Specificity and Laterality

Decisive inference depends not only on procedure-confirmed outcomes but also on confidence that those outcomes occurred on the same side and at the same anatomic site as the index procedure. Current Procedural Terminology (CPT) codes in most large EHR databases do not capture laterality modifiers, requiring investigators to infer laterality through linkage to same-encounter diagnosis codes—a methodology with inherent limitations. Authors must report coding completeness for anatomic site and laterality, describe their inference methodology, and acknowledge potential misclassification.

We have established clear thresholds for laterality completeness. FAI requires <20% missing laterality data; studies exceeding this threshold will be rejected without review. For FAI, bilateral foot or ankle surgery cases must be excluded because of the high risk of outcome misattribution. FAO requires <30% missing laterality data for studies making side-specific claims; studies exceeding this threshold will be rejected without review. FAO permits bilateral cases when outcomes are systemic (eg, mortality, venous thromboembolism) or when the focus is epidemiologic, provided authors report the proportion of bilateral cases and acknowledge the limitation. In all cases, sensitivity analyses are expected, and laterality limitations must be prominently discussed. Studies that examine only systemic outcomes, where laterality is irrelevant, are exempt from these requirements.

Medication Exposure and Outcomes Beyond the Data

Studies are generally unsuitable for either journal when primary conclusions depend on variables that administrative databases cannot reliably capture, such as radiographic outcomes, operative technique, implant-specific performance, or patient-reported outcomes. Separately, medication and substance exposure studies face a distinct challenge: Prescription records cannot confirm actual use, dose, adherence, or temporal relationship to the outcome. FAI generally considers such studies unsuitable unless inpatient medication administration records confirm exposure. FAO can accept them if they are hypothesis-generating, but limitations must be highlighted.

Looking Ahead

Even sophisticated statistical methods cannot overcome fundamental data limitations. We recognize that some database studies published in recent issues of FAI were accepted prior to the development of these guidelines and may not fully meet the standards outlined here. The criteria described in this editorial reflect an evolving understanding of what these data sources can and cannot support, and they will be applied prospectively to all new submissions. This transition is important, and we are committed to applying these standards consistently going forward. Our responsibility as editors and peer reviewers is to ensure that conclusions are fully commensurate with the limitations of the data, including the ability, or inability, to reliably attribute outcomes to a specific treatment. Aligning the research question, the outcome, and the level of anatomic specificity to what the data can reliably support is not a barrier to innovation; it is essential for scientific integrity and meaningful clinical progress. These guidelines are intended to help authors direct their work to the journal where it will have the greatest impact, whether through decisive surgical inference in FAI or through carefully framed, hypothesis-generating insights in FAO.

Please refer to the Submission Guidelines for more detailed instructions: https://journals.sagepub.com/author-instructions/FAO. We encourage all authors planning clinical database studies to consult these guidelines before submission.

Charles L. Saltzman, MD Robert B. Anderson, MD Brad D. Blankenhorn, MD Timothy R. Daniels, MD, FRCSC James R. Holmes, MD John Y. Kwon, MD Ellie Pinsker, PhD Stefan Rammelt, MD, PhD John T. Campbell, MD Robert A. Vander Griend, MD