Supplemental for the Research letter: Do the Data Support ESVS Advising Against Cryopreserved Allografts for Vascular Graft Infection?

Scope This supplement accompanies the Research Letter: Do Data Support ESVS Advising Against Cryopreserved Allografts for Vascular Graft Infection?. No individual-patient data were used. The primary analysis is based on graft-related outcomes reported in ESVS Tables 13 and 15 as published. The source-verification audit is a separate reproducibility assessment. Primary dataset For the primary analysis, ESVS-reported denominators and percentages were transcribed from Tables 13 and 15. Where required for analysis, an integer event count was reconstructed from the reported denominator and rounded percentage when uniquely compatible with that percentage. These reconstructed counts are analytical reconstructions of the ESVS table and should not automatically be interpreted as raw event counts reported in the original publication. In particular, a published percentage may represent a time-to-event estimate rather than events divided by the full study denominator. NR and NA were treated as missing and were not imputed as zero. Source-verification audit We assessed each outcome separately to determine whether the ESVS-reported population and outcome value could be independently reconstructed from source information accessible to us. Full-text publications were used where available; otherwise abstracts or indexed publication records were used. Positive verification required sufficient source information to support the relevant population and outcome value, allowing for rounding. Failure to verify an ESVS value was classified as unverifiable from the accessible source information, not as erroneous. Absence of an outcome from an abstract or other limited source was not interpreted as a zero event count or as evidence that the outcome was absent from the full publication. Additional subgroup data, supplementary information, or direct author correspondence may have been available to the guideline authors. Whole-cohort observations were not substituted for ESVS subgroup values when populations or denominators differed. Such observations are described only as audit context. The source-verification audit was not used to generate an alternative comparative meta-analysis. Verification categories Status Definition Verified Accessible source information positively supports the ESVS outcome value for the relevant population, allowing for rounding. A time-to-event percentage may be verified as a reported rate without being converted to a raw event count. Discordant Accessible source information provides a materially different value or mapping for the same identifiable population/outcome. This does not exclude additional data available to the guideline authors. Unverifiable Accessible source information is insufficient to reconstruct the ESVS value. This is not evidence that the ESVS value is incorrect. NR/NA The ESVS table does not provide an analysable value for that outcome. Statistical analysis The prespecified graft-related outcomes were rupture, reinfection, and occlusion/thrombosis. Random-effects pooled proportions were estimated using binomial-normal generalised linear mixed models with a logit link. Zero-event studies were retained without continuity correction. Graft type was examined as a study-level moderator. The resulting odds ratios are ecological comparisons across separate case series and are not causal head-to-head treatment effects. Non-significant moderator tests do not establish equivalence or non-inferiority. Follow-up and time at risk Reported follow-up ranged from 14-70 months in the autologous-vein series and 12-96 months in the cryopreserved-allograft series. Formal incidence-rate analyses were not performed because individual patient-time was unavailable and reported follow-up summaries were not consistently defined as means or medians. Multiplying study size by reported follow-up would therefore not provide a consistently defined estimate of person-time. The analyses use cumulative event proportions; differences in observation time remain an important limitation, particularly for late structural events. Versioning Version 1.3 supersedes the earlier