Protocol Deviations and Selective Outcome Reporting in Longevity Trials

Who This Is Useful For

This page is useful for readers assessing trials of interventions intended to affect ageing biology, healthspan, function, age-related disease, or biological-age measures. It is especially relevant when a publication highlights a different outcome, time point, subgroup, or analysis from the one emphasized in the trial's earlier records. [1] [3] [7]

A clinical trial begins with a planned question: who will be enrolled, what will be compared, which outcome is primary, when it will be measured, and how it will be analysed. The protocol and statistical analysis plan preserve those decisions so that readers can distinguish analyses specified before the results were known from analyses developed after examining the data. [1] [2]

Trials do not always proceed exactly as planned. Recruitment may be slower than expected, an assay may fail, safety information may require a dose change, or follow-up procedures may become impractical. Transparent amendment and reporting allow such changes to be evaluated. The central concern is not the mere existence of a change, but its timing, rationale, effect on validity, and completeness of disclosure. [1] [2]

Protocol Changes and Selective Reporting at a Glance

These situations are not interchangeable. A documented prospective amendment can be compatible with a credible trial, while an undeclared, result-driven outcome change prevents readers from reconstructing the original confirmatory question. [1] [2] [3]

Why Outcome Prespecification Matters

A trial may collect many plausible outcomes. If investigators can choose which outcomes to emphasize after seeing the data, chance variation is more likely to be mistaken for a consistent treatment effect. Selective reporting also removes unfavourable or inconclusive results from the visible evidence, which can distort both individual articles and later systematic reviews. [4] [8]

Direct comparisons of trial protocols with publications have found that statistically significant outcomes are more likely to be fully reported than non-significant outcomes. A broader systematic review likewise found empirical evidence that outcome reporting is associated with the direction and statistical significance of results. [3] [4]

Why Longevity Trials Need Particular Care

Longevity trials can measure molecular markers, physiological measures, functional outcomes, diseases, adverse events, quality of life, and mortality at several time points. Geroscience has not established a single outcome that captures every relevant dimension, and proposed biological-age measures are not automatically validated surrogates for clinical benefit. This breadth creates legitimate scientific uncertainty, but it also creates many opportunities to select a favourable result after analysis. [7] [9]

• Multiple biomarkers: a report may foreground the clock, inflammatory marker, metabolic measure, or composite that changed most, although another measure was designated primary. [7] [9]
• Multiple time points: a short-term difference may disappear later, or a late difference may not have been the prespecified assessment time. Complete reporting is needed to distinguish persistence from a selected snapshot. [1] [3]
• Composite and clinical outcomes: results can depend on which age-related diseases, functional events, or deaths are included and how they are combined. The outcome definition should therefore be fixed precisely enough to reproduce. [1] [7]
• Long follow-up: attrition, intervention discontinuation, missed assessments, and changes in usual care can accumulate over time; each can produce deviations or missing data that affect interpretation. [1] [2]

Changes That May Be Reasonable

Some changes protect participants or preserve the scientific value of a trial. Examples include modifying an intervention after new safety evidence, replacing a discontinued assay, extending recruitment after slower enrolment, or revising procedures during an external disruption. SPIRIT guidance treats the protocol as a living, version-controlled document and calls for important amendments to be communicated to oversight bodies and registries and described in completed trial reports. [2]

A reasonable change can still alter the question the trial answers. Readers need to know whether the decision occurred before or after outcome data were available, who made it, whether group assignments or interim results were visible, and whether the original and revised analyses are both presented. Those details make it possible to judge the change without assuming either that every deviation invalidates a trial or that every stated rationale removes the risk of bias. [1] [2]

Common Forms of Selective Outcome Reporting

• Omitting a prespecified outcome: an outcome listed in the protocol or registry does not appear in the article. [3] [5]
• Changing outcome status: a secondary outcome becomes primary, or the original primary outcome is presented as secondary, without a clear declaration. [3] [5]
• Adding an undeclared outcome: a favourable new outcome is presented without identifying it as post hoc or exploratory. [5]
• Selecting a measurement: one scale, threshold, biomarker definition, or follow-up time is reported from several possibilities. [1] [3]
• Selective analysis reporting: one covariate adjustment, population, subgroup, or missing-data method is emphasized while other prespecified analyses are absent. [1] [6]

How to Compare the Trial Record With the Article

• Start with dates: check whether registration, protocol versions, amendments, and the statistical analysis plan predate enrolment, unblinding, or access to outcome data. [1] [2]
• Match the primary outcome exactly: compare the variable, participant-level metric, method of aggregation, and assessment time rather than relying only on a similar outcome name. [1]
• Account for every prespecified outcome: note outcomes that disappeared, changed status, or appeared only in supplements, registry results, or later reports. [3] [5]
• Read the amendment history: look for dates, reasons, oversight, and whether the change was made without knowledge of comparative results. [1] [2]
• Separate confirmation from exploration: additional analyses can generate useful hypotheses when their post hoc status and multiplicity are clear. [1] [6]

What Registration Can and Cannot Show

Prospective registration creates a dated public record and can make discrepancies easier to detect. Access to the full protocol and statistical analysis plan adds detail that a short registry entry may not contain. CONSORT 2025 therefore asks trial reports to identify where these documents can be accessed and to report changes to outcomes after the trial began, with reasons. [1]

Registration is not self-validating. Entries can be late, vague, incomplete, or updated without enough explanation, and publications can still differ from registered plans. A prospective study comparing registered outcomes with reports in five major medical journals found substantial undeclared outcome discrepancies even among journals endorsing reporting guidance. [5] [6]

What This Does Not Mean

• It does not mean every protocol deviation invalidates a trial; the type, cause, timing, frequency, and analytic consequence matter. [1] [2]
• It does not mean an unchanged protocol guarantees an unbiased result; randomization, blinding, missing data, measurement quality, and statistical precision still require appraisal. [1]
• It does not mean exploratory outcomes are unscientific; they become misleading when presented as if they were prespecified confirmatory tests. [1] [6]
• It does not mean a non-significant primary outcome makes all secondary findings irrelevant; it means their multiplicity, status, and uncertainty should remain visible. [4] [8]

Practical Interpretation Examples

• If a biological-age clock becomes primary after enrolment: check whether the change was made before outcome data were examined, why the original endpoint changed, and whether both results are reported. [1] [7]
• If one of many biomarkers improves: determine how many were measured, which were prespecified, and whether the report accounts for multiple testing. [7] [9]
• If missed visits are called protocol deviations: look for counts and reasons by trial group and for an explanation of how missing observations affected the analysis. [1]
• If a new subgroup is highlighted: treat it as exploratory unless the subgroup and interaction analysis were specified before the results were known. [1] [6]

Related Reading

• Preregistration and Registered Reports in Ageing Research
• Intervention Adherence in Longevity Trials
• Multiple Comparisons and False-Positive Findings in Longevity Studies
• How to Read the Methods and Limitations of a Longevity Study

Summary

Protocol deviations describe departures from a trial's planned conduct or analysis; selective outcome reporting describes a biased choice about which results become visible. The two can overlap, but they are not synonymous. In longevity trials, numerous biomarkers, time points, and candidate health outcomes make a dated and specific analysis plan especially important. Credible interpretation depends on transparent comparison of the registry, protocol, amendments, statistical analysis plan, and complete results. [1] [2] [3] [7]

References

1. Hopewell, S., et al. (2025). CONSORT 2025 explanation and elaboration: updated guideline for reporting randomised trials. BMJ. https://www.bmj.com/content/389/bmj-2024-081124
2. Chan, A. W., et al. (2025). SPIRIT 2025 explanation and elaboration: updated guideline for protocols of randomised trials. BMJ. https://www.bmj.com/content/389/bmj-2024-081660
3. Chan, A. W., Hrobjartsson, A., Haahr, M. T., Gotzsche, P. C., and Altman, D. G. (2004). Empirical evidence for selective reporting of outcomes in randomized trials: comparison of protocols to published articles. JAMA. https://jamanetwork.com/journals/jama/fullarticle/198809
4. Dwan, K., Gamble, C., Williamson, P. R., and Kirkham, J. J. (2013). Systematic review of the empirical evidence of study publication bias and outcome reporting bias: an updated review. PLoS ONE. https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0066844
5. Goldacre, B., et al. (2019). COMPare: a prospective cohort study correcting and monitoring 58 misreported trials in real time. Trials. https://link.springer.com/article/10.1186/s13063-019-3173-2
6. Kirkham, J. J., et al. (2018). Outcome reporting bias in trials: a methodological approach for assessment and adjustment in systematic reviews. BMJ. https://www.bmj.com/content/362/bmj.k3802
7. Cummings, S. R., and Kritchevsky, S. B. (2022). Endpoints for geroscience clinical trials: health outcomes, biomarkers, and biologic age. GeroScience. https://pmc.ncbi.nlm.nih.gov/articles/PMC9768060/
8. Kirkham, J. J., et al. (2010). The impact of outcome reporting bias in randomised controlled trials on a cohort of systematic reviews. BMJ. https://pubmed.ncbi.nlm.nih.gov/20156912/
9. Justice, J. N., et al. (2018). A framework for selection of blood-based biomarkers for geroscience-guided clinical trials: report from the TAME Biomarkers Workgroup. GeroScience. https://pmc.ncbi.nlm.nih.gov/articles/PMC6294728/