{"columns":["round","reviewer","comment_id","severity","section","comment","response","action","protocol_version_after"],"rows":[["1","gpt-6-astra","C1","major","Transitivity; populations.csv P07","Unresolved intransitivity still allows indirect comparisons; heart failure phenotypes split only if connected.","Agreed. Indirect and network estimates are not calculated for comparisons with unresolved intransitivity. HFpEF/HFmrEF (P07A) and HFrEF (P07B) are fixed separate populations. LVEF, albuminuria, fibrosis stage, AHI with PAP use, amyloid status and antipsychotic drug added as condition specific modifiers.","revised","0.2.0"],["1","gpt-6-astra","C2","major","outcomes.csv O13 O24 O100","Incompatible scales share outcome rows (BMI SDS with BMI; CDR-SB with ADAS-Cog; HRQoL MD with SMD threshold).","Agreed. Split into O13A (BMI SDS) and O13B (percent BMI, only when SDS absent; never pooled); O100A (CDR-SB) and O100B (ADAS-Cog); O24A (SF-36 PCS, MD) and O24B (SMD across instruments only when O24A cannot be used).","revised","0.2.0"],["1","gpt-6-astra","C3","major","outcomes.csv O110","Abstinence coded lower_better.","Agreed. One row per substance and construct (O110A heavy drinking days; O110B alcohol abstinence; O111 tobacco abstinence; O112 return to opioid use; O113 stimulant negative urine; O114 cannabis use days), each with its own measure, direction and threshold; populations split P14A to P14E.","revised","0.2.0"],["1","gpt-6-astra","C4","major","outcomes.csv O22 O23 O36 O90","Clinically different definitions pooled.","Agreed. O22 is kidney failure, sustained eGFR decline of at least 50 percent or kidney death; broader composites (40 percent decline, cardiovascular death) are O22B and never pooled. O23 is hospitalization only and O23B adds urgent visits. O36A pulmonary aspiration and O36B retained gastric content; O90A ovulation, O90B menstrual regularity, O90C live birth.","revised","0.2.0"],["1","gpt-6-astra","C5","major","Bayesian analysis","Count based cloglog models substitute for survival estimates without assumptions; no overdispersion rule for recurrent events.","Agreed for time to event outcomes: reported or validly reconstructed log hazard ratios are primary and count based models are sensitivity analyses only. For binary adverse events (participants with at least one event), which trials report as counts over very different durations, the cloglog model with a log follow-up offset is primary with its constant hazard ratio assumption stated, following NICE DSU TSD2, and the odds ratio model and duration restriction are sensitivity analyses (this also answers fable-5.1 F12). Recurrent events use reported rate ratios or a negative binomial model.","revised","0.2.0"],["1","gpt-6-astra","C6","major","Frequentist analysis; zero events","Frequentist random effects cannot avoid continuity corrections; MH is common effect.","Agreed. Inverse variance random effects only when every arm has events (no correction); otherwise the frequentist estimate is Mantel-Haenszel, labeled as a common effect model, with the change in contributing trials reported. The Bayesian binomial model remains primary.","revised","0.2.0"],["1","gpt-6-astra","C7","major","Estimands","Trial product results substituted into treatment policy analyses.","Agreed. Each result is classified by intercurrent event strategy; primary networks use treatment policy results only; hypothetical results form separate networks; no substitution. The RoB 2 effect of interest follows the causal question of each network.","revised","0.2.0"],["1","gpt-6-astra","C8","major","Study designs","No executable rules for crossover and cluster trials.","Agreed. New section on cluster (design effect with trial ICC or 0.05, sensitivity 0.01 and 0.10) and crossover trials (paired contrast only with reported SE and addressed carryover; otherwise first period).","revised","0.2.0"],["1","gpt-6-astra","C9","major","Harms synthesis","No column marks nonrandomized outcomes; pooling rules undefined.","Agreed. Added nrs_synthesis to outcomes.csv and harms_nrs.csv specifying population, exposure, comparators, designs, time zero and latency, estimand, minimum confounders and effect measures per harm; pooling separately by measure, design and comparator type; critical risk of bias excluded.","revised","0.2.0"],["1","gpt-6-astra","C10","major","Risk of bias; meta-biases","ROB-ME is for pairwise meta-analysis; ROB-MEN is for NMA.","Agreed. ROB-MEN for network estimates and ROB-ME for pairwise syntheses; CINeMA reporting bias domain uses ROB-MEN.","revised","0.2.0"],["1","gpt-6-astra","C11","major","Time points","Windows mix 20 and 56 week results; O10 window excludes shorter obesity trials.","Agreed. New windows.csv with contiguous primary and secondary windows per outcome and population, nearest target rule within each, duration as covariate. Obesity primary window is now 40 to 104 weeks (target 68), which includes the 46 week survodutide trial; binary harms use the cloglog model with follow-up offset.","revised","0.2.0"],["1","gpt-6-astra","C12","major","Effect measures; thresholds","Absolute effects lack reference node, baseline risk source and horizon.","Agreed. Reference node rule declared before fitting; reference risk from a random effects meta-analysis of reference arms with low and high scenarios; outcome specific horizon (absolute_horizon_weeks); thresholds applied on the absolute scale at that horizon, with a relative threshold under the rare outcome rule.","revised","0.2.0"],["1","gpt-6-astra","C13","major","Priors","Priors incomplete (no HR or rate heterogeneity priors) and too diffuse.","Agreed. Complete prior table (priors.csv) for every likelihood and model; relative effects Normal(0, 2.5^2) on log scales (1.5^2 for rare outcomes) and Normal(0, (2.5 s_y)^2) for mean differences; Turner Table IV predictive distributions (turner_map.csv) applied to log hazard and log rate ratios; prior predictive checks required. We chose 2.5 rather than 1.5 because weight loss responder odds ratios can exceed 50.","revised","0.2.0"],["1","gpt-6-astra","C14","major","Nodes","No rules for flexible and maximum tolerated dose strategies; no formulation or ratio fields.","Agreed. nodes.csv now has formulation, dosing_strategy and component_ratio; flexible and maximum tolerated dose arms form their own nodes (for example TIRZ_SC_MTD_QW) with a highest dose sensitivity analysis; arms are never classified by achieved dose.","revised","0.2.0"],["1","gpt-6-astra","C15","major","populations.csv P02W","Definition required reaching a weight loss target, which excludes STEP 4.","Agreed. P02W is any randomization after a GLP-1 receptor agonist containing run in, with run in criteria recorded separately and the run in regimen as an effect modifier (EM22).","revised","0.2.0"],["1","gpt-6-astra","C16","major","populations.csv P01","Strata cannot classify basal bolus, premixed, mixed or replacement designs.","Agreed. Strata S0 to S3 (S3 prandial, premixed or basal bolus insulin) and S4 mixed background; the 80 percent rule governs assignment; replacement designs map each arm to its randomized regimen.","revised","0.2.0"],["1","gpt-6-astra","C17","major","Comparators","Comparator scope narrower than agreed; surgery excluded by design.","Agreed in part. Any comparator, including surgery, is eligible, and clinical compatibility decides network entry. Trials without a GLP-1 receptor agonist containing arm remain ineligible; the justification and its consequences are now stated in the protocol.","revised","0.2.0"],["1","gpt-6-astra","C18","major","Information sources; Appendix A","CENTRAL and Embase unresolved; Appendix A empty; benchmark not independent.","Agreed. CENTRAL is now searched; Embase depends on access confirmed before registration, with a stated justification if absent; Appendix A is generated from the concept file at build; the benchmark set adds regulatory pivotal trials and ClinicalTrials.gov linked publications independent of the harvest. PRESS review requires a human information specialist and is listed as a pre registration condition.","revised","0.2.0"],["1","gpt-6-astra","C19","major","concepts.yaml","Endogenous hormone exclusion untested; native GLP-1 infusion excluded; missing taspoglutide aliases.","Partly agreed. Tested: the receptor agonist heading adds 0 records and the endogenous hormone heading adds 6,988, of which 484 are human infusion experiments; the protocol now defines scope as therapeutic drugs and states this explicitly, so native GLP-1 infusion is a scope exclusion. Taspoglutide codes R1583 and BIM51077 added.","revised","0.2.0"],["1","gpt-6-astra","C20","minor","nodes.csv","Missing doses: retatrutide 1 mg, survodutide 0.6 mg, mazdutide 3 and 4.5 mg.","Agreed and extended: nodes are now defined by rule, the seed list includes these doses and others, and registry_dose_candidates.csv lists doses mined from registrations.","revised","0.2.0"],["1","gpt-6-astra","C21","minor","concepts.yaml hrs9531","HRS9531 and HRS-7535 conflated.","Agreed. Separate agents and node families (GIP_GLP1 injectable; GLP1_ORAL_SM oral).","revised","0.2.0"],["1","gpt-6-astra","C22","minor","Network meta-regression","Eligibility and modeling details missing.","Agreed. Priority order, centering and scaling, one covariate at a time, ten trials and overlap required, complete case with a 20 percent missingness limit, at most two covariates per multivariable model, and a stated prior.","revised","0.2.0"],["1","gpt-6-astra","C23","minor","Administrative information; reviews","PRISMA-P items 3 to 5 incomplete; publication methods treated as protocol.","Agreed. Items 3 to 5 are structured with placeholders the owner must complete before registration (administrative checklist). Reviews without a prospective protocol are recorded as such.","revised","0.2.0"],["1","fable-5.1","F1","major","outcomes.csv O10","O10 window empties P12 and P15 networks.","Agreed. O10B with a 12 to 52 week window (target 24) for P12 and P15 (windows.csv W09).","revised","0.2.0"],["1","fable-5.1","F2","major","outcomes.csv O110","O110 not executable.","Agreed; see gpt-6-astra C3.","revised","0.2.0"],["1","fable-5.1","F3","major","outcomes.csv O11 O27","Several syntheses packed under one outcome ID.","Agreed. O11A to O11E and O27A to O27E.","revised","0.2.0"],["1","fable-5.1","F4","major","outcomes.csv O32 O34 O36","Absolute thresholds above plausible risks of rare harms.","Agreed. Rare outcome rule: when the median reference risk at the horizon is below 2 per 1,000 the threshold is a hazard or odds ratio of 1.25 and absolute effects are given per 10,000.","revised","0.2.0"],["1","fable-5.1","F5","major","Harms synthesis","No nrs column; long latency harms missing.","Agreed. nrs_synthesis column; added pancreatic cancer (O38), any cancer (O39), gastroparesis (O37A), obstruction or ileus (O37B), completed suicide (O35B), major congenital malformations (O120) and acute kidney injury (O121); harms_nrs.csv specifies each.","revised","0.2.0"],["1","fable-5.1","F6","major","populations.csv P01; outcomes.csv O02","Weight management trials in diabetes lose their primary time point.","Agreed. P01 added to O10 and O11 with the 40 to 104 week primary window; such trials are analyzed in their background stratum.","revised","0.2.0"],["1","fable-5.1","F7","major","populations.csv P01 strata","Prandial, premixed and basal bolus insulin trials have no stratum.","Agreed in substance; we split S2 (basal insulin) and S3 (prandial, premixed or basal bolus) rather than using a covariate, and added S4 for mixed backgrounds, as grok-4.7 G3 also proposed.","revised","0.2.0"],["1","fable-5.1","F8","major","Comparators","Trials with GLP-1 as shared background produce self loops.","Agreed. A trial enters a network only when at least two arms map to different nodes; such trials are inventoried only.","revised","0.2.0"],["1","fable-5.1","F9","major","Nodes","No rules for flexible dose, free combinations, randomized comparator doses.","Agreed for flexible dose and free combinations. For randomized comparator doses we keep one molecule level node and combine arms within the trial, so that node sets stay identical across networks.","revised","0.2.0"],["1","fable-5.1","F10","major","populations.csv P07","HF phenotypes split only if connected.","Agreed; see gpt-6-astra C1. We use LVEF above 40 percent versus 40 percent or lower, matching the eligibility of the dedicated GLP-1 receptor agonist heart failure trials, with restriction to 50 percent or more as a sensitivity analysis.","revised","0.2.0"],["1","fable-5.1","F11","major","Priors","No heterogeneity priors for HR and rate outcomes; comparison type unfixed.","Agreed. Turner distributions applied to log hazard and log rate ratios. The comparison type follows a rule fixed from network geometry before outcome extraction (active versus active distribution when more than half of direct comparisons are between active drugs), which grok-4.7 G8 proposed; we judged it closer to the evidence than fixing one type for all networks.","revised","0.2.0"],["1","fable-5.1","F12","major","Effect measures O25 to O36","Odds ratios at longest follow-up pool very different durations.","Agreed. Binary harms use the cloglog model with a log follow-up offset as primary; the odds ratio model is a sensitivity analysis.","revised","0.2.0"],["1","fable-5.1","F13","major","Information sources; Appendix A","Appendix A empty; Embase and CENTRAL undecided; no Chinese databases.","Agreed. Appendix A generated at build; CENTRAL searched; Embase decided before registration with a fallback justification; CNKI and Wanfang Data searched with Chinese terms in concepts.yaml, verified by a Chinese-reading team member.","revised","0.2.0"],["1","fable-5.1","F14","major","nodes.csv","Node list incomplete and already behind the evidence; 9 mg retatrutide unverified.","Agreed. Nodes are now defined by rule so that new regimens do not need amendments; the seed list adds the named agents and doses (node families for agents whose doses registrations mask); retatrutide 9 mg is kept with a verification note because phase 3 programs use it; registry_dose_candidates.csv lists 861 dose candidates mined from registrations. A complete enumeration needs full text extraction, which cannot precede screening.","revised","0.2.0"],["1","fable-5.1","F15","major","Data collection","Single extraction plus verification for outcome data.","Agreed. Outcome data are extracted independently in duplicate.","revised","0.2.0"],["1","fable-5.1","F16","minor","Outcomes and time points","Critical outcomes wording mismatched to files.","Agreed; wording corrected.","revised","0.2.0"],["1","fable-5.1","F17","minor","outcomes.csv O22 O50","Mixed kidney composites.","Agreed; see gpt-6-astra C4 (O22 and O22B; O50 uses the O22 definition).","revised","0.2.0"],["1","fable-5.1","F18","minor","outcomes.csv O04","No mapping for older hypoglycemia categories.","Agreed. Mapping rule added in mapping_rules.","revised","0.2.0"],["1","fable-5.1","F19","minor","populations.csv P14; nodes.csv","P14 ambiguous; P99 note repetition; dose note on placebo rows.","Agreed. P14A to P14E; P99 note fixed; notes on placebo, standard care, lifestyle and surgery corrected.","revised","0.2.0"],["1","fable-5.1","F20","minor","effect_modifiers.csv","No condition specific modifiers; EM11 undefined.","Agreed. EM16 to EM22 added; EM11 categories defined.","revised","0.2.0"],["1","fable-5.1","F21","minor","Selection process","Agreement statistic unnamed.","Agreed. Cohen kappa of at least 0.8 with percent agreement reported.","revised","0.2.0"],["1","fable-5.1","F22","minor","Data collection hierarchy","Hierarchy could conflict.","Agreed. Ordered: prespecified analysis for the estimand with longest follow-up of that estimand, then regulatory review, registry, publication.","revised","0.2.0"],["1","fable-5.1","F23","minor","Frequentist analysis","Contribution algorithm unnamed.","Agreed. netcontrib shortest path method, as used by CINeMA (verified as the netmeta 3.7.0 default).","revised","0.2.0"],["1","fable-5.1","F24","minor","Network meta-regression","Subgroups cannot feed ML-NMR without IPD.","Agreed; wording corrected.","revised","0.2.0"],["1","fable-5.1","F25","minor","Confidence in evidence","Only critical outcomes rated.","Agreed. Critical and important outcomes are rated.","revised","0.2.0"],["1","fable-5.1","F26","minor","Rationale","Overstates approved uses.","Agreed. Rationale distinguishes approved from studied uses.","revised","0.2.0"],["1","fable-5.1","F27","minor","concepts.yaml","Untested MeSH claim; missing mechanism terms and codes.","Agreed. Claims tested and reported with counts; mechanism block added (190 additional PubMed records on 2026-10-12); codes AM833, NNC0174-0833, SHR20004 (noiiglutide), PB-119, HM15211 (efocipegtrutide), GZR18, CT-388 and HS-20094 added, plus UBT251, zenagamtide and DR10624 found by the mechanism block. NN9487 returned no PubMed record and was not added.","revised","0.2.0"],["1","grok-4.7","G1","major","nodes.csv","Oral semaglutide note wrong; missing doses and agents; dose response across routes.","Agreed. 3, 7 and 14 mg labeled original tablet; reformulated 1.5, 4 and 9 mg tablets are separate nodes with a bioequivalence pooling sensitivity; 25 and 50 mg labeled the higher dose program; missing doses and agent node families added (cotadutide once daily in micrograms; maridebart cafraglutide every 4 weeks); dose response restricted to one molecule, route and formulation.","revised","0.2.0"],["1","grok-4.7","G2","major","outcomes.csv O13 O22 O24 O50 O100 O110","Outcomes cannot be pooled as coded.","Agreed; implemented as proposed (O13A and O13B; O22 primary kidney composite and O22B; O24A with a 3 point threshold and O24B SMD; O100A and O100B; substance specific rows).","revised","0.2.0"],["1","grok-4.7","G3","major","Participants; populations.csv P01","80 percent rule contradicted; strata incomplete.","Agreed. Single assignment rule (80 percent at baseline or a baseline defined subgroup); strata S0 to S4 as proposed.","revised","0.2.0"],["1","grok-4.7","G4","major","outcomes.csv O01 O10","Windows drop primary outcomes for P04, P05, P12, P15.","Agreed. Population specific windows: P04 and P05 HbA1c 12 to 40 weeks (target 26) with a 40 to 64 secondary window; P12 and P15 weight 12 to 52 weeks (target 24); P02 and P03 40 to 104 weeks (target 68). We used nearest to target within the window for every outcome rather than the trial primary time point, as gpt-6-astra C11 asked for one consistent rule.","revised","0.2.0"],["1","grok-4.7","G5","major","Harms; thresholds","Odds ratios over unstated follow-up; recurrent events; impossible thresholds; sparse networks.","Agreed in part. Binary harms use the cloglog model with follow-up offset; recurrent hypoglycemia episodes have a rate ratio outcome (O04B); networks with fewer than 20 events are descriptive. For rare outcomes we adopted a relative threshold (hazard or odds ratio 1.25, as fable-5.1 F4 proposed and as CINeMA commonly uses) rather than 0.1 per 1,000, because an absolute threshold that small cannot be estimated with useful precision from trials. Gastrointestinal events and pancreatitis remain participant level outcomes because trials rarely report episode counts with exposure.","revised","0.2.0"],["1","grok-4.7","G6","major","R/search_strategy.R; concepts.yaml","PubMed does not truncate inside quotes; HRS agents merged; cagrilintide alone too broad.","Partly agreed. Tested on 2026-10-12: \"GLP-1 receptor agonist*\"[tiab] returns 8,109 records, identical to the singular and plural phrases combined, and \"incretin-based therap*\" returns 1,260 against 1,239 for the two explicit forms, so PubMed does truncate inside quoted phrases and the renderer is correct. HRS9531 and HRS-7535 separated. Cagrilintide alone is kept: CagriSema trials randomize cagrilintide monotherapy arms and registrations name it alone, and the term retrieves about 100 records.","partly revised","0.2.0"],["1","grok-4.7","G7","major","Effect measures; GRADE","Reference node and horizon undefined.","Agreed. Reference node rule (placebo, then standard care, then most frequent active node) and a 156 week horizon for O14, O20 to O23, O50; 52 weeks for harms and O40 (heart failure trials); stored in outcomes.csv.","revised","0.2.0"],["1","grok-4.7","G8","major","Priors","Baseline and relative effect priors mismatched; Turner comparison type.","Agreed in part. Logit and cloglog baselines Normal(0, 5^2) rather than Normal(0, 2^2): a prior centered on 50 percent with SD 2 shrinks rare event baselines (risk near 1 in 1,000 lies 3.5 SD away). Relative effects Normal(0, 2.5^2) and Normal(0, 1.5^2) for rare outcomes (between your proposal and gpt-6-astra C13). The comparison type rule is adopted as proposed. Nonrandomized models use a half normal prior with scale 1 on tau.","revised","0.2.0"],["1","grok-4.7","G9","major","Harms synthesis","No nrs column; overlap rule.","Agreed in part. nrs_synthesis and harms_nrs.csv added with the source overlap hierarchy. Mortality, MACE, heart failure, serious adverse events and discontinuation are not synthesized from nonrandomized studies because the review owner limited nonrandomized synthesis to harms; trials provide that evidence.","partly revised","0.2.0"],["1","grok-4.7","G10","major","populations.csv P07 P08","HF and CKD defined as both one and two networks.","Agreed in part. Fixed separate populations: P07A (LVEF above 40 percent) and P07B (40 percent or lower); P08A (with diabetes) and P08B (without). We did not split HFmrEF from HFpEF because the dedicated GLP-1 receptor agonist containing heart failure trials enrolled participants with LVEF of at least 45 or 50 percent, so a separate 41 to 49 percent network would have no dedicated trials; restriction to 50 percent or more is a sensitivity analysis.","partly revised","0.2.0"],["1","grok-4.7","G11","major","Confidence in evidence","CINeMA mixes frameworks; no numeric rules; thresholds not locked.","Agreed. CINeMA applied separately to Bayesian and frequentist estimates with the lower rating kept; numeric rules for every domain written into the protocol; thresholds locked before any outcome data are extracted.","revised","0.2.0"],["1","grok-4.7","G12","major","Designs; data collection","No cluster or crossover rules; source hierarchy conflict; completers.","Agreed; implemented as proposed (design effect with ICC 0.05 default; crossover paired contrast only with reported SE; ordered hierarchy; non treatment policy results flagged).","revised","0.2.0"],["1","grok-4.7","G13","major","Administrative; sources; Appendix A","Pre registration items open; no INN monitoring.","Agreed. Administrative checklist added; Appendix A generated; CENTRAL committed and Embase decision required before registration; quarterly check of WHO proposed INN lists added as a source.","revised","0.2.0"],["1","grok-4.7","G14","minor","node_classes.csv","Beinaglutide half life.","Agreed; corrected.","revised","0.2.0"],["1","grok-4.7","G15","minor","nodes.csv","Prandial insulin and glinides not molecule level; no OTHER_ACTIVE node.","Agreed. Aspart, lispro, glulisine and regular insulin; repaglinide and nateglinide; OTHER_ACTIVE rule node.","revised","0.2.0"],["1","grok-4.7","G16","minor","Frequentist analysis","netmeta adds 0.5 to zero cells by default.","Agreed; see gpt-6-astra C6 (no continuity correction; Mantel-Haenszel when zero cells occur).","revised","0.2.0"],["1","grok-4.7","G17","minor","Harms synthesis","HKSJ misnamed.","Agreed. REML (Sidik and Jonkman if REML fails) with the Hartung-Knapp adjustment.","revised","0.2.0"],["1","grok-4.7","G18","minor","effect_modifiers.csv EM02","BMI in children.","Agreed. BMI SDS for P03.","revised","0.2.0"],["1","grok-4.7","G19","minor","outcomes.csv O05 O10 O12","Thresholds too small.","Not adopted at this stage. A 5 percentage point difference is the regulatory efficacy benchmark against placebo, not a minimal important difference between active regimens, and a 2 mmHg systolic difference is associated with meaningful population level cardiovascular risk change. All thresholds are provisional and will be confirmed or changed by clinical advisors and patient partners and locked before any outcome data are extracted.","declined (provisional thresholds to be locked by clinical panel)","0.2.0"],["1","grok-4.7","G20","minor","outcomes.csv O51 O80","WOMAC scale and eGFR slope sign.","Agreed. WOMAC rescaled to 0 to 100; eGFR slope defined with decline negative.","revised","0.2.0"],["1","grok-4.7","G21","minor","populations.csv P04 P05","Missing hypoglycemia (P04) and insulin dose (P05).","Agreed. O04 for P04; O06 total daily insulin dose for P05.","revised","0.2.0"],["1","grok-4.7","G22","minor","outcomes.csv O21","No MACE mapping rule.","Agreed. O21 is three point MACE; expanded composites are O21B and never pooled.","revised","0.2.0"],["1","grok-4.7","G23","minor","Meta-regression; component models","Overfitting; component coding for fixed ratio products.","Agreed; implemented as proposed (one covariate at a time, Normal(0,1) prior on the standardized covariate, bias covariates as sensitivity analyses, 20 percent missingness limit, presence or absence coding, fixed ratio products excluded from Emax).","revised","0.2.0"],["1","grok-4.7","G24","minor","R/search_strategy.R; selection","OpenAlex query stored unfielded; agreement statistic.","Agreed in part. The run log already stores the fielded OpenAlex query (search.title_and_abstract=...), and Appendix A now states the parameter. Calibration requires Cohen kappa of at least 0.8 (stricter than 0.70, as fable-5.1 F21 proposed), with percent agreement reported.","revised","0.2.0"],["2","gpt-6-astra","C2-1","major","Event outcomes and harms; O36A O36B","Count based cloglog model needs more than a constant hazard ratio; procedure outcomes are not hazards.","Agreed in part, with a clarification now in the protocol. Because every trial has its own baseline, the relative effect from the complementary log log model is a log hazard ratio under proportional hazards within each trial whatever the shape of the baseline hazard, provided follow-up is similar across arms (trials with markedly different exposure are flagged); the constant hazard assumption enters only the baseline model used for absolute effects, which is never extrapolated beyond observed follow-up. Reported hazard ratios are used when available, and odds ratio analyses within compatible follow-up are sensitivity analyses. O36A and O36B are now risks per procedure on the odds ratio scale with no time horizon.","partly revised","0.3.0"],["2","gpt-6-astra","C2-2","major","Frequentist analysis","Mantel-Haenszel cannot estimate hazard ratios; zero non-events ignored.","Agreed. Frequentist binary analyses are on the log odds ratio scale; inverse variance only when every arm has events and non-events, otherwise Mantel-Haenszel common effect; for harms with a hazard scale primary model the frequentist odds ratio analysis is a sensitivity analysis within compatible follow-up, not a replicate.","revised","0.3.0"],["2","gpt-6-astra","C2-3","major","outcomes.csv O21B O22B O24B O110B O111","New subdivisions still mix definitions.","Agreed. O21B is four point MACE with unstable angina only (other expanded composites described); kidney composites are O22 (50 percent or doubling), O22B (40 percent), O22C (with cardiovascular death) and O22D (with macroalbuminuria or declines below 40 percent), never pooled, with P08A and P08B added to O22B and O22C; O24B is physical functioning and O24C total HRQoL; O110B defines abstinence in the final 4 weeks; O111 is verified continuous abstinence and O111B point prevalence abstinence.","revised","0.3.0"],["2","gpt-6-astra","C2-4","major","windows.csv","Overlapping O05 windows; population matching; O04B; narrow windows; kg versus percent.","Agreed. O05 has P01 windows (W01 to W03) and P02 windows (W27, W28); a stated rule applies windows only where outcome populations match; a check confirmed no overlaps and exactly one primary window per outcome and population; O04B is in W00 (longest follow-up); narrow windows are target plus or minus 4 weeks; the text distinguishes O01 and O02 from O10.","revised","0.3.0"],["2","gpt-6-astra","C2-5","major","Absolute effects","Fixed horizons cannot be obtained from arbitrary follow-up.","Agreed. Fixed time outcomes use the analyzed window target and that window reference risk with no extrapolation of odds ratios; hazard outcomes use the stated horizon capped at the median follow-up of the reference arms, with no extrapolation beyond observed follow-up.","revised","0.3.0"],["2","gpt-6-astra","C2-6","major","harms_nrs.csv; nonrandomized synthesis","Estimands not separated.","Agreed. Pooling is now separate by population, exposure and comparator contrast, estimand (initial treatment versus on treatment), latency and risk window, effect measure and design; estimates without the minimum confounders are excluded from quantitative synthesis; every harms_nrs.csv row is now explicit.","revised","0.3.0"],["2","gpt-6-astra","C2-7","major","Nodes; nodes.csv","Flexible nodes underdefined; IcoSema and iGlarLixi ratios.","Agreed. Flexible and maximum tolerated dose nodes are defined by formulation, permitted dose set and algorithm; IcoSema 700 U and 2 mg per mL (350 U per 1 mg, verified from the manufacturer); iGlarLixi pens (100 U with 50 or 33 micrograms per mL) with pen switching per trial algorithm and a ratio specific sensitivity analysis; IDegLira 100 U and 3.6 mg per mL.","revised","0.3.0"],["2","gpt-6-astra","C2-8","major","Confidence in cumulative evidence","Incoherence rule leaves cases unclassified; inestimable incoherence.","Agreed. Incoherence is judged from the interval of the direct minus indirect difference against plus or minus the threshold (within, outside, otherwise); inestimable incoherence is recorded and judged from transitivity and network structure with reasons.","revised","0.3.0"],["2","gpt-6-astra","C2-9","minor","priors.csv","SMD s_y; O04B prior; dispersion prior.","Agreed. s_y equals 1 for SMD outcomes; O04B uses only reported log rate ratios with prior PR14, so no dispersion prior is needed (a count model would require an amendment).","revised","0.3.0"],["2","gpt-6-astra","C2-10","minor","agents.csv; concepts.yaml","Amycretin and zenagamtide are one molecule.","Agreed and verified with the manufacturer release: merged under amycretin with zenagamtide and NN9487 as search terms.","revised","0.3.0"],["2","gpt-6-astra","C2-11","minor","effect_modifiers.csv; Participants; meta-regression","EM02 lacks P03; heart failure pooling wording; categorical coding.","Agreed. EM02 includes P03 and P04 (BMI SDS); heart failure populations are never pooled in the text too; categorical covariates use prespecified indicator coding, and class coefficients need at least three trials with covariate variation.","revised","0.3.0"],["2","grok-4.7","G2-1","major","windows.csv O05","Overlapping O05 windows.","Agreed; see C2-4.","revised","0.3.0"],["2","grok-4.7","G2-2","major","nodes.csv SEMAR2","Bioequivalence notes read as pooling 1.5 mg with 14 mg.","Agreed. Each reformulated row names its single bioequivalent original dose and node, and the text states the three pairs.","revised","0.3.0"],["2","grok-4.7","G2-3","major","outcomes.csv horizons","156 week horizon in heart failure populations.","Agreed in substance. Rather than population specific rows, the horizon of every hazard based outcome is capped at the median follow-up of the network reference arms, which gives about 52 weeks in heart failure networks and also prevents extrapolation elsewhere (C2-5).","revised","0.3.0"],["2","grok-4.7","G2-4","minor","priors.csv; rare outcome rule","Rare outcome switch depends on extracted data; cutoffs disagree.","Agreed. A rare_outcome flag in outcomes.csv, fixed before extraction, drives both the tighter prior and the relative threshold; it is never recomputed from observed risks. O35A, O36B and O120 are not flagged rare and have absolute thresholds.","revised","0.3.0"],["2","grok-4.7","G2-5","minor","windows.csv W00 W25","Longest follow-up rows not joinable.","Agreed. selection_rule column; W00 lists outcome IDs explicitly; blank upper limits mean no limit; nearest target does not apply.","revised","0.3.0"],["2","grok-4.7","G2-6","minor","populations.csv P01","Stratum delimiter.","Agreed; fixed.","revised","0.3.0"],["2","grok-4.7","G2-7","minor","effect_modifiers.csv EM02","P03 missing.","Agreed; P03 and P04 added.","revised","0.3.0"],["2","grok-4.7","G2-8","minor","outcomes.csv O21B O22B","Expanded composites lack heart failure and kidney populations.","Agreed. O21B includes P07A and P07B; O22B and O22C include P08A and P08B.","revised","0.3.0"],["2","grok-4.7","G2-9","minor","harms_nrs.csv","As above cells; estimand pooling.","Agreed; see C2-6.","revised","0.3.0"],["2","grok-4.7","G2-10","minor","CINeMA","No range for relative thresholds.","Agreed. Range 1/1.25 to 1.25; discrepancies as differences of log effects against log 1.25.","revised","0.3.0"],["2","grok-4.7","G2-11","minor","outcomes.csv O120","Blank horizon.","Agreed. Per pregnancy (and per procedure for O36A and O36B) outcomes are exempt from horizons and stated as such.","revised","0.3.0"],["2","grok-4.7","G2-12","minor","Absolute effects","Most frequent active node undefined.","Agreed. The active node that appears in the most trials.","revised","0.3.0"],["2","grok-4.7","G2-13","minor","nodes.csv SEMA_PO_FLEX_QD","Formulation blank.","Agreed. Original tablet.","revised","0.3.0"],["2","grok-4.7","G2-14","minor","nodes.csv templates","Template rows could collapse doses.","Agreed. is_template column; templates never enter a network.","revised","0.3.0"],["2","grok-4.7","G2-15","minor","CINeMA","Bayesian contribution weights undefined.","Agreed. Contribution based domains are judged once from the frequentist shortest path matrix and apply to both frameworks; imprecision, heterogeneity and incoherence are judged per framework; the lower rating is kept.","revised","0.3.0"],["2","fable-5.1","F2-1","major","outcomes.csv O22","Registered composites of LEADER, REWIND and SELECT unmapped.","Agreed. O22D added for composites with macroalbuminuria or eGFR declines below 40 percent; doubling of serum creatinine maps to a 50 percent decline; a composite without albuminuria is used for O22 or O22B when also reported.","revised","0.3.0"],["2","fable-5.1","F2-2","major","windows.csv W12","KCCQ window empties the P07B network.","Agreed. P07B primary window 12 to 36 weeks (target 26) with 36 to 64 as secondary; P07A keeps 36 to 64 (target 52) as primary with 12 to 36 as secondary.","revised","0.3.0"],["2","fable-5.1","F2-3","major","windows.csv O05","O05 overlaps.","Agreed; see C2-4.","revised","0.3.0"],["2","fable-5.1","F2-4","minor","windows.csv W00 W25","Free text and longest.","Agreed; see G2-5.","revised","0.3.0"],["2","fable-5.1","F2-5","minor","windows.csv; time windows text","Window gaps (sleep apnea, MASH, pediatric) and text claims.","Agreed. W14 lower limit 24 weeks; MASH secondary window 12 to 48 (target 24); pediatric secondary window 20 to 40 (target 26); text corrected for O01, O02 and O10 and for windows without a secondary.","revised","0.3.0"],["2","fable-5.1","F2-6","minor","concepts.yaml pubmed_mesh","Drug combination lines parse as text.","Agreed and verified (84 and 81 records as text; 0 as quoted concepts). IDegLira[nm] (57 records) replaces the first; the iGlarLixi line is removed because no supplementary concept exists.","revised","0.3.0"],["2","fable-5.1","F2-7","minor","nodes.csv RETA_SC_9MG_QW","9 mg cannot be verified.","Agreed. Row and notes removed; the node rule recreates it from a report.","revised","0.3.0"],["2","fable-5.1","F2-8","minor","priors.csv; O36B; O120","Offset units; cutoffs; non rare outcomes with relative thresholds.","Agreed. Offset in years stated; one a priori rare flag drives prior and threshold; O36B (50 per 1,000 procedures) and O120 (5 per 1,000 pregnancies) have absolute thresholds.","revised","0.3.0"],["2","fable-5.1","F2-9","minor","concepts.yaml chinese_terms","No Chinese terms for Chinese origin investigational agents.","Agreed. Candidate names added under candidates for verification, and both databases are also searched with English names and development codes, as the protocol now states.","revised","0.3.0"],["2","fable-5.1","F2-10","minor","turner_map.csv O14","Wrong Turner category for incident diabetes.","Agreed. Remapped to infection or onset of new disease.","revised","0.3.0"],["2","fable-5.1","F2-11","minor","effect_modifiers.csv EM02","P03 missing.","Agreed; see G2-7.","revised","0.3.0"],["2","fable-5.1","F2-12","minor","populations.csv P01","Stratum delimiter.","Agreed; see G2-6.","revised","0.3.0"],["2","fable-5.1","F2-13","minor","search/runs; Timeline","No logged run of the current concepts; ClinicalTrials.gov counts unreconciled.","Agreed. The run log now records reconciliation (PubMed: 28,454 identifiers, 28,320 articles and 134 book chapters; ClinicalTrials.gov: 2,384 and 1,038 across two queries with 667 in both, union 2,755), the code logs this automatically from now on, and the Timeline no longer claims validation: a logged harvest of all four API sources with the registered concepts version precedes the baseline search. The OpenAlex harvest awaits an API key from the owner.","revised","0.3.0"],["2","fable-5.1","F2-14","minor","F11 follow-up","Define direct comparison for the Turner comparison type rule.","Agreed. A direct comparison is a distinct pair of nodes with direct evidence.","revised","0.3.0"],["2","fable-5.1","F2-15","minor","F27 follow-up","NN9487 should be kept even with zero hits.","Agreed. NN9487 and NN-9487 added; the protocol states that development codes are kept regardless of current counts.","revised","0.3.0"],["3","gpt-6-astra","C3-1","major","Event outcomes and harms","Censored participants are treated as non-events in the count based harms model.","Agreed in part. The protocol now states that the count model analyzes a dichotomous outcome over each trial observation period (the usual analysis of adverse effects in Cochrane reviews), prefers reported time to first event estimates, extracts incomplete follow-up by arm, judges it in RoB 2 domain 3, and adds a sensitivity analysis excluding trials with more than 20 percent incomplete follow-up or a difference above 10 percentage points between arms. We did not adopt descriptive reporting whenever any participant is censored, because that would leave almost all trial evidence on harms unsynthesized.","partly revised","0.4.0"],["3","gpt-6-astra","C3-2","major","outcomes.csv kidney composites","Kidney composites mix component sets and thresholds; doubling of creatinine is not a 50 percent decline.","Agreed. A general rule now pools only identical component sets and thresholds within any outcome identifier, otherwise definition strata; doubling of serum creatinine (about 57 percent) is its own outcome O22C; O22D holds composites with cardiovascular death and O22E those with albuminuria or smaller declines, both in definition strata; O50 follows O22.","revised","0.4.0"],["3","gpt-6-astra","C3-3","major","Absolute effects","Median follow-up cap still extrapolates from shorter reference arms.","Agreed. The reference risk at the analyzed horizon comes only from reference arms whose follow-up covers it, using reported Kaplan-Meier or cumulative incidence where available; shorter arms are excluded from the baseline estimate.","revised","0.4.0"],["3","gpt-6-astra","C3-4","major","nodes.csv iGlarLixi","Single pen and switching regimens pooled.","Agreed. Separate nodes for the 2:1 pen, the 3:1 pen and each switching strategy (permitted pens, switching threshold and titration algorithm); combined only in a sensitivity analysis.","revised","0.4.0"],["3","gpt-6-astra","C3-5","major","Confidence in cumulative evidence","Contributions borrowed from a restricted analysis with different evidence.","Agreed. Contribution matrices represent the same trials and network as the estimate rated (for hazard scale estimates from counts, an inverse variance network of trial level log hazard ratios used only for contributions); restricted sensitivity analyses are rated separately and never substitute their rating; the lower of two framework ratings applies only to the same evidence.","revised","0.4.0"],["3","gpt-6-astra","C3-6","minor","priors.csv PR04 PR08","O24C lacks priors.","Agreed. PR04 and PR08 apply to O24B and O24C; the text names both.","revised","0.4.0"],["3","gpt-6-astra","C3-7","minor","Bayesian network meta-analysis","Negative binomial option contradicts O04B.","Agreed. Normal likelihood on reported log rate ratios; a count model requires a logged amendment.","revised","0.4.0"],["3","gpt-6-astra","C3-8","minor","outcomes.csv O21","Obsolete expanded MACE mapping.","Agreed; replaced with the proposed text.","revised","0.4.0"],["3","gpt-6-astra","C3-9","minor","nodes.csv OTHER_ACTIVE","Comparator template activation rule.","Agreed. Dose templates and the comparator template have separate activation rules in the table and the text.","revised","0.4.0"],["3","fable-5.1","F3-1","major","windows.csv W22","Short substance use trials (6 to 9 weeks) fall in no window.","Agreed. W33 (4 to 12 weeks, target 8) added as a secondary window for P14A to P14E; overlap check passed.","revised","0.4.0"],["3","fable-5.1","F3-2","minor","priors.csv PR04 PR08","O24C priors.","Agreed; see C3-6.","revised","0.4.0"],["3","fable-5.1","F3-3","minor","outcomes.csv O22C","O22C pooled 40 and 50 percent thresholds.","Agreed; see C3-2 (composites with cardiovascular death are now O22D in definition strata).","revised","0.4.0"],["3","fable-5.1","F3-4","minor","Time windows; O120","O120 has no window.","Agreed. Per procedure and per pregnancy outcomes are stated to have no week window; O36A and O36B were removed from W00.","revised","0.4.0"],["3","fable-5.1","F3-5","minor","O04B relative threshold","Unstated exception.","Agreed. O04B uses the rate ratio range 1/1.25 to 1.25 and PR14, stated in Absolute effects and in the imprecision rule.","revised","0.4.0"],["3","fable-5.1","F3-6","minor","Time windows text","O10 versus O10B.","Agreed; replaced with the proposed text.","revised","0.4.0"],["3","fable-5.1","F3-7","minor","nodes.csv OTHER_ACTIVE","Placeholder wording.","Agreed; see C3-9.","revised","0.4.0"],["3","fable-5.1","F2-13 follow-up","minor","Administrative checklist","Four source harvest only in the Timeline.","Agreed. The OpenAlex key and the reconciled four source harvest are in the administrative checklist.","revised","0.4.0"],["3","grok-4.7","G3-1","major","priors.csv; Effect measures","O24C priors.","Agreed; see C3-6.","revised","0.4.0"],["3","grok-4.7","G3-2","minor","nodes.csv OTHER_ACTIVE","Comparator template.","Agreed; see C3-9.","revised","0.4.0"],["3","grok-4.7","G3-3","minor","Time windows; W00","Procedure and pregnancy outcomes.","Agreed; see F3-4.","revised","0.4.0"],["3","grok-4.7","G3-4","minor","Bayesian likelihoods","Negative binomial.","Agreed; see C3-7.","revised","0.4.0"],["3","grok-4.7","G3-5","minor","O04B","Relative threshold.","Agreed; see F3-5.","revised","0.4.0"],["3","grok-4.7","G3-6","minor","outcomes.csv horizons","Stored horizon versus cap.","Agreed. The text states that absolute_horizon_weeks is the uncapped horizon and the analyzed horizon is the smaller of it and the median reference arm follow-up.","revised","0.4.0"],["3","grok-4.7","G3-7","minor","outcomes.csv O22C","Mixed thresholds.","Agreed; see C3-2.","revised","0.4.0"],["3","grok-4.7","G3-8","minor","Time windows text","O10B label.","Agreed; see F3-6.","revised","0.4.0"],["4","gpt-6-astra","C4-1","major","Event outcomes and harms","Counts with incomplete observation must not be interpreted as hazard ratios.","Agreed; your structure is adopted. In round 4 fable-5.1 and grok-4.7 judged the previous decline sound, and the change keeps what they required (no odds ratios pooled across very different durations): a hazard ratio analysis uses reported hazard ratios, reported cumulative incidence at a common time, and counts only when event status is known for at least 95 percent of randomized participants in every arm (100 percent in a sensitivity analysis), with the attenuation from incomplete observation quantified in the protocol (true hazard ratio 2.0 with risks 0.50 and 0.75 gives 1.93, 1.88 and 1.79 with 5, 10 and 20 percent lost; under 2 percent bias for risks below 5 percent); a separately rated recorded harms analysis of all trials has no hazard ratio interpretation and no fixed horizon absolute risks. Conclusions use the hazard ratio analysis when it meets the connectivity rule.","revised","0.5.0"],["4","gpt-6-astra","C4-2","major","Confidence in cumulative evidence","Count derived contribution networks fail with zero cells.","Agreed. The contribution network uses the same trial level log hazard ratios as the primary model and adds 0.5 to every cell of a trial with a zero event or all event arm, for contributions only (they depend on precision weights and network geometry, and the correction never enters an effect estimate); every primary trial is kept; when valid contributions still cannot be computed, within study bias and indirectness are judged qualitatively across the full evidence set with reasons recorded. This is the zero cell rule with justification you asked for, and the rule fable-5.1 and grok-4.7 proposed.","revised","0.5.0"],["4","gpt-6-astra","C4-3","major","Absolute effects","Competing risks unaddressed.","Agreed. For outcomes that death can preclude, cause specific and subdistribution hazard ratios are extracted and synthesized separately; absolute risks use cumulative incidence accounting for competing deaths, never one minus Kaplan-Meier; cause specific hazard ratios are converted only with the competing event hazard, otherwise relative effects only; trials with negligible mortality are exempt.","revised","0.5.0"],["4","gpt-6-astra","C4-4","minor","nodes.csv iGlarLixi switching","Switching row lacks template protection.","Agreed. IGLARLIXI_SWITCH_<strategy>_SC_TIT_QD with is_template = yes and the proposed note; the text adds this third template category.","revised","0.5.0"],["4","fable-5.1","F4-1","minor","Event outcomes and harms","No likelihood route for Kaplan-Meier input.","Agreed. Reported cumulative incidence enters on the log cumulative hazard scale with a delta method standard error and offset log t, as proposed; for outcomes with competing deaths the cumulative incidence must account for them (C4-3).","revised","0.5.0"],["4","fable-5.1","F4-2","minor","Composite definitions; O04","Hypoglycemia mapping versus the composite rule.","Agreed. Named as a prespecified exception, flagged and excluded in a sensitivity analysis.","revised","0.5.0"],["4","fable-5.1","F4-3","minor","Confidence in cumulative evidence","Zero event arms break the contribution network.","Agreed; see C4-2.","revised","0.5.0"],["4","fable-5.1","F4-4","minor","outcomes.csv O22E","Populations and overlap with O22D.","Agreed. P08A and P08B added; composites with cardiovascular death are O22D (precedence rule in O22D, O22E and O50).","revised","0.5.0"],["4","fable-5.1","F4-5","minor","Sensitivity analyses","Missing two analyses.","Agreed. The list now includes iGlarLixi node combination, the follow-up completeness exclusion, the 100 percent known status restriction and the O04 mapping exclusion.","revised","0.5.0"],["4","grok-4.7","G4-1","major","Confidence in cumulative evidence","Hazard scale contribution matrices do not run on the primary evidence.","Agreed; see C4-2 (your wording for the network).","revised","0.5.0"],["4","grok-4.7","G4-2","minor","outcomes.csv O22D O22E O50","Kidney composite precedence.","Agreed; your three mapping rules adopted verbatim and the Composite definitions text updated.","revised","0.5.0"],["4","grok-4.7","G4-3","minor","Networks and nodes; connectivity","Network key omits definition stratum.","Agreed; definition stratum added to the network key and the connectivity rule.","revised","0.5.0"],["4","grok-4.7","G4-4","minor","Event outcomes and harms","Kaplan-Meier input and denominators.","Agreed. See F4-1; denominators are the number randomized, with the safety population only when the randomized number is not reported, flagged.","revised","0.5.0"],["4","grok-4.7","G4-5","minor","nodes.csv iGlarLixi switching","Switching node.","Agreed; see C4-4.","revised","0.5.0"],["4","grok-4.7","G4-6","minor","Sensitivity analyses","Missing analyses.","Agreed; see F4-5.","revised","0.5.0"],["4","grok-4.7","G4-7","minor","Absolute effects","One covering reference arm.","Agreed; that arm risk is used and no between arm heterogeneity is estimated.","revised","0.5.0"],["5","gpt-6-astra","C5-1","major","Event outcomes and harms","The 95 percent rule lets unknown outcomes become non-events; the bias claim is wrong under unequal loss.","Agreed; verified your example (hazard ratio 1.20 becomes 1.26 with 5 percent lost from one arm). Counts enter the primary hazard ratio analysis only when status is known for every randomized participant at t (with t and known status defined as grok-4.7 proposed); the 95 percent rule is a sensitivity analysis; the protocol now states that it implies no general bound or direction of bias and shows both directions. Counts with unresolved missing status enter the recorded harms analysis.","revised","0.6.0"],["5","gpt-6-astra","C5-2","major","Competing risks","Cumulative incidence contrasts target subdistribution hazards; hazard type missing from the network key; exemption undefined.","Agreed. Estimator and hazard type are recorded for every input including derived ones; cumulative incidence contrasts and complete status counts are subdistribution inputs and never enter cause specific networks; hazard type is in the network key and connectivity rule; the exemption is removed, so the rule applies to every outcome with competing events, including harms; cumulative incidence enters only when 0 < p < 1 with finite uncertainty.","revised","0.6.0"],["5","gpt-6-astra","C5-3","major","Recorded harms analysis","Pooled absolute difference at the median duration not operationally defined.","Agreed. No pooled absolute effect is derived from the recorded harms analysis; each trial recorded proportions and within trial risk difference are displayed with observation period and completeness; magnitude dependent certainty domains use the relative range 1/1.25 to 1.25 on the recorded event rate ratio, with the scale stated. fable-5.1 and grok-4.7 had proposed defining a pooled absolute difference at the median; we followed your position because the observation processes differ between trials, and gave the analysis a fully specified relative rule so that the certainty rules remain executable, which is what their comments required.","revised","0.6.0"],["5","gpt-6-astra","C5-4","minor","Confidence in cumulative evidence","Contribution rule omits cumulative incidence inputs.","Agreed. The contribution network uses, per trial, the input that entered the primary model (reported log hazard ratio, cumulative incidence contrast with its delta method variance, or count derived estimate with the 0.5 rule), preserving shared arm covariance.","revised","0.6.0"],["5","gpt-6-astra","C5-5","minor","Absolute effects; certainty; ranking","Relative only exception incomplete.","Agreed. When valid absolute conversion is unavailable, absolute effects and categories are marked unavailable, magnitude dependent domains are judged against the relative range 1/1.25 to 1.25 with the reason stated (never no concerns merely because conversion is unavailable), and rankings needing absolute effects are omitted; this overrides the general requirement. We used the relative range rather than a purely qualitative judgment so that the rule stays numeric and reproducible.","revised","0.6.0"],["5","gpt-6-astra","C5-6","minor","Composite definitions","Duplicated O22C.","Agreed; corrected.","revised","0.6.0"],["5","fable-5.1","F5-1","minor","Recorded harms absolute effects","Absolute effects and imprecision undefined for the recorded harms analysis.","Resolved differently from your proposal: gpt-6-astra (C5-3) argued that a pooled absolute difference at the median observation period is not defined when observation processes differ, so no pooled absolute effect is derived; per trial proportions are displayed and magnitude dependent domains use the relative range 1/1.25 to 1.25 on the recorded event rate ratio, which makes the certainty rules executable.","revised (alternative)","0.6.0"],["5","fable-5.1","F5-2","minor","Frequentist analysis; certainty; sensitivity list","Compatible follow-up undefined.","Agreed. 20 to 104 weeks restored in both places and added to the sensitivity list.","revised","0.6.0"],["5","fable-5.1","F5-3","minor","Competing risks","No primary hazard type.","Agreed. Cause specific is primary for time to event outcomes (outcome trials report Cox models) and subdistribution for binary harms (counts and cumulative incidence estimate it); hazard type is part of the network key.","revised","0.6.0"],["5","fable-5.1","F5-4","minor","Competing risks","Negligible mortality undefined.","Resolved by removal rather than a cutoff, as gpt-6-astra proposed: because every input now carries its hazard type and counts with complete status are subdistribution inputs, no exemption is needed.","revised (alternative)","0.6.0"],["5","fable-5.1","F5-5","minor","Composite definitions","Duplicated O22C.","Agreed; corrected.","revised","0.6.0"],["5","fable-5.1","F5-6","minor","Outcomes and time points","Stale version number.","Agreed; corrected.","revised","0.6.0"],["5","fable-5.1","F5-note","note","Implementation","Arm level cumulative incidence entry is not a multinma likelihood.","Adopted: cumulative incidence enters as a contrast level normal likelihood on the difference of log(-log(1 - p)) between arms.","revised","0.6.0"],["5","grok-4.7","G5-1","major","Competing risks; network key","Competing risk synthesis is not yet a network; conversions and exemption.","Agreed. Hazard type in the network key and connectivity rule; your conversion formulas adopted (subdistribution: 1 - (1 - F_ref)^HR; cause specific with both hazards and the stated assumption for the competing hazard); applied to every outcome death can preclude. The exemption is removed rather than set at 1 percent, because counts with complete status are now subdistribution inputs and need none.","revised","0.6.0"],["5","grok-4.7","G5-2","major","Hazard ratio analysis","Time and known status undefined for the count rule.","Agreed; your definitions of t and known status adopted. Following gpt-6-astra (C5-1), the primary analysis requires known status for every randomized participant and the 95 percent rule is a sensitivity analysis; your statement that coding unknowns as event free pulls toward the null holds for equal loss, and the protocol now also shows that unequal loss can push away from the null.","revised","0.6.0"],["5","grok-4.7","G5-3","minor","Composite definitions","Duplicate O22C.","Agreed; your sentence adopted.","revised","0.6.0"],["5","grok-4.7","G5-4","minor","Contributions","Cumulative incidence inputs.","Agreed; see C5-4. Cumulative incidence of 0 or 1 never enters (C5-2), so no correction is needed for it.","revised","0.6.0"],["5","grok-4.7","G5-5","minor","Frequentist analysis","Compatible follow-up bounds.","Agreed; see F5-2.","revised","0.6.0"],["5","grok-4.7","G5-6","minor","Recorded harms analysis","Absolute differences unnamed.","Resolved as in C5-3 and F5-1: the coefficient is named the log relative rate of recorded first events, not a hazard ratio, and certainty uses the relative range on that scale; no pooled absolute difference is derived.","revised (alternative)","0.6.0"],["6","gpt-6-astra","C6-1","minor","Competing risks","Cause specific conversion omits its constant hazard assumption.","Agreed. The conversion is stated to assume constant cause specific hazards over the analyzed horizon and is labeled an exponential competing risks approximation; leaving the competing hazard unchanged is stated as assuming no effect on it; when time varying baseline hazards are available cumulative incidence is computed by integration; otherwise, if the approximation cannot be supported, only relative effects are reported.","revised","0.7.0"],["6","gpt-6-astra","C6-2","minor","Competing risks","Proportional subdistribution hazards unstated.","Agreed. Subdistribution networks assume proportional subdistribution hazards; proportionality assessments are extracted and curves inspected; with material nonproportionality, horizon specific contrasts are reported separately and no constant subdistribution hazard ratio converts effects across horizons.","revised","0.7.0"],["6","gpt-6-astra","C6-3","minor","Recorded harms analysis","The rate ratio label does not match the fitted quantity.","Agreed. The reported quantity is Q, the exponentiated complementary log log contrast with nominal observation periods, given as a formula and interpreted descriptively; certainty uses 0.8 to 1.25 on Q.","revised","0.7.0"],["6","gpt-6-astra","C6-4","minor","Known status","Competing deaths and administrative censoring.","Agreed; your definition adopted (fable-5.1 F6-2 proposed the same).","revised","0.7.0"],["6","gpt-6-astra","C6-5","minor","Hazard ratio analysis","Equal loss given a general direction of bias.","Agreed. The text states there is no general bound or direction even with equal missing fractions and labels the numbers as an illustration with missingness independent of the outcome at baseline.","revised","0.7.0"],["6","fable-5.1","F6-1","minor","Conclusions for harms","Order of networks when the primary hazard type network fails.","Agreed; your ordering adopted (primary hazard type network, then the other labeled with its hazard type, then the recorded harms analysis), and a trial reporting both types contributes each to its own network. grok-4.7 (G6-2) proposed excluding the cause specific network from conclusions; we kept it as the second step because a connected network of reported Cox hazard ratios is a valid relative effect on the harm, more interpretable than the recorded harms analysis.","revised","0.7.0"],["6","fable-5.1","F6-2","minor","Known status","Competing deaths omitted.","Agreed; see C6-4.","revised","0.7.0"],["6","fable-5.1","F6-3","minor","Competing risks","Not applicable not anchored to the outcome.","Agreed; not applicable only for outcomes with no competing event (such as O20); otherwise the hazard type follows the estimator.","revised","0.7.0"],["6","fable-5.1","F6-4","minor","Competing risks; absolute effects","Source of the hazards; Kaplan-Meier wording.","Agreed. Both rates are events per person year from the covering reference arms, pooled on the log scale, under constant hazards (stated); the reference risk bullet names Aalen-Johansen for outcomes with competing events and Kaplan-Meier otherwise.","revised","0.7.0"],["6","fable-5.1","F6-5","minor","Competing risks","Kaplan-Meier contrasts are valid cause specific inputs.","Agreed. A contrast between arms of one minus Kaplan-Meier with competing deaths censored is a cause specific input; it is never an absolute or reference risk.","revised","0.7.0"],["6","fable-5.1","F6-6","minor","Certainty; data items","Cross references and data items.","Agreed. The imprecision rule names the relative only rule and the recorded harms range; data items add estimator, hazard type, known status counts and observation period.","revised","0.7.0"],["6","grok-4.7","G6-1","minor","Hazard ratio analysis","The 95 percent sensitivity does not choose t.","Agreed; your sentence adopted.","revised","0.7.0"],["6","grok-4.7","G6-2","minor","Conclusions; time to event counts","Conclusions across hazard type networks; counts for time to event outcomes.","Agreed for time to event counts: they enter only the subdistribution network and are never pooled with cause specific estimates. For conclusions we adopted fable-5.1 ordering (primary hazard type network, then the other hazard type network labeled, then the recorded harms analysis) rather than excluding the cause specific network, because a connected network of reported Cox hazard ratios gives a valid relative effect on the harm and is more interpretable than the recorded harms quantity Q.","partly revised","0.7.0"],["6","grok-4.7","G6-3","minor","Competing risks","Conversion sentences ambiguous.","Agreed; your wording adopted (constant rates for the reference node, pooled on the log scale; replacement of lambda1 everywhere; definition of F_ref; relative only fallback).","revised","0.7.0"],["7","gpt-6-astra","C7-1","minor","Competing risks","Absolute effect fallback overrides valid conversions.","Agreed; your replacement adopted (cause specific: relative only when neither integration nor a supported exponential approximation is possible; subdistribution: reference cumulative incidence and proportional subdistribution hazards, without constant cause specific hazards).","revised","0.8.0"],["7","gpt-6-astra","C7-2","minor","Event outcomes and harms; O20","Count routing conflicts with all cause mortality.","Agreed (grok-4.7 G7-M1 and fable-5.1 F7-2 raised the same point). Counts for outcomes with a competing event enter only the subdistribution network; for outcomes without one (at present only O20), hazard ratios and complete status counts share the not applicable network, which is primary.","revised","0.8.0"],["7","gpt-6-astra","C7-3","minor","Recorded harms analysis","Offset written as t instead of log t.","Agreed; corrected.","revised","0.8.0"],["7","fable-5.1","F7-1","minor","Absolute effects","Constant hazard reference risk is a net risk for outcomes with competing events.","Agreed. For outcomes with competing events a single event rate is never used: the reference risk is the reported Aalen-Johansen cumulative incidence, otherwise the exponential competing risks expression from the pooled rates (which then serves as F_ref), otherwise the relative only rule. This also answers grok-4.7 G7-4.","revised","0.8.0"],["7","fable-5.1","F7-2","minor","Time to event counts","Downstream sentences for time to event counts.","Agreed; your three replacements adopted (conclusions for time to event outcomes with descriptive reporting as the last step, likelihood list, sensitivity list).","revised","0.8.0"],["7","fable-5.1","F7-3","minor","Data items","Items needed for the competing hazard.","Agreed; competing deaths without a preceding target event, person years by arm (with the stated fallback) and any reported competing event hazard ratio are now data items.","revised","0.8.0"],["7","fable-5.1","F7-4","minor","Competing risks","Two judgments without criterion or timing.","Agreed; your replacement adopted (two reviewers, before relative effects are modeled, logged with reasons, with the stated criteria and the alternative as a sensitivity analysis).","revised","0.8.0"],["7","fable-5.1","F7-5","minor","Competing risks","Not applicable set open.","Agreed; at present only O20.","revised","0.8.0"],["7","grok-4.7","G7-M1","major","Event outcomes and harms; O20","Count sentence contradicts the not applicable rule for all cause mortality.","Agreed; your replacement adopted (see C7-2).","revised","0.8.0"],["7","grok-4.7","G7-1","minor","Competing risks","Time varying integral omits where the hazard ratio enters.","Agreed; your sentence adopted (the hazard ratio enters inside the integral and inside all cause survival).","revised","0.8.0"],["7","grok-4.7","G7-2","minor","Competing risks","Kaplan-Meier contrast has no stated estimator.","Agreed; it enters as the contrast of log(-log S) at a common t with a delta method standard error, only when 0 < 1 - S < 1, without continuity correction in the contribution network.","revised","0.8.0"],["7","grok-4.7","G7-3","minor","Conclusions; 95 percent sensitivity","Conclusion order for time to event outcomes; hazard type of sensitivity counts.","Agreed; conclusions now cover time to event outcomes (cause specific network primary, then subdistribution labeled, then descriptive reporting) and the 95 percent counts are a sensitivity analysis of the subdistribution network (or the not applicable network for O20), never of a cause specific network.","revised","0.8.0"],["7","grok-4.7","G7-4","minor","Absolute effects","Single rate fallback for outcomes with competing events.","Agreed; see F7-1. Reported Aalen-Johansen incidence first, then the exponential competing risks expression from both pooled rates, otherwise the relative only rule.","revised","0.8.0"],["8","gpt-6-astra","C8-1","major","Competing risks","Integration used all cause survival instead of survival free of both events.","Agreed; verified (hazards 0.2 and 0.1 per year give 0.518 at five years with event free survival and 0.787 with death only survival). This error entered in version 0.8.0 from wording we adopted. The integration is now written with survival free of both the target and the competing event, using your formula, and grok-4.7 G8-5 flagged the same gloss.","revised","0.9.0"],["8","gpt-6-astra","C8-2","major","Data items; competing risks","Person time fallback permits invalid denominators.","Agreed; your rule adopted: rates use time at risk until the first target event, competing event, censoring or cutoff; the number randomized times mean follow-up only when it is exactly that; median follow-up, treatment exposure and general follow-up are not substitutes; without valid person time no rate is computed by division. grok-4.7 (G8-6) proposed allowing a flagged median; we chose your stricter rule because a median times the number randomized is not time at risk and flagging does not correct the rate.","revised","0.9.0"],["8","gpt-6-astra","C8-3","minor","Absolute effects","Reference risk hierarchy omits integration.","Agreed. Subdistribution conversions take F_ref from the reported Aalen-Johansen incidence, then piecewise integration of reference hazards, then the exponential approximation under supported constant hazards, otherwise relative only; cause specific conversions require hazards, not a reference cumulative incidence alone (see F8-4 for the reference risk shown).","revised","0.9.0"],["8","gpt-6-astra","C8-4","minor","Competing risks; conclusions","O20 exception missing from the primary network rule and conclusions.","Agreed; the primary network sentence now names the O20 exception, and the conclusions rule states that O20 has only its not applicable network, else descriptive reporting. In round 7 our response recorded C7-2 as adopted although the primary network sentence and the one input per trial rule had not been carried (fable-5.1 noted this); both are now in the text.","revised","0.9.0"],["8","fable-5.1","F8-1","minor","Competing risks","Primary network for O20.","Agreed; see C8-4.","revised","0.9.0"],["8","fable-5.1","F8-2","minor","Event outcomes and harms","One input per trial.","Agreed; your order adopted (reported hazard ratio, then cumulative incidence or Kaplan-Meier contrast, then arm counts; alternatives logged).","revised","0.9.0"],["8","fable-5.1","F8-3","minor","Competing risks; absolute effects","F_ref and the constant hazard requirement.","Agreed. F_ref follows the stated order, and constant cause specific hazards are required only when F_ref comes from the exponential approximation, which then carries its label.","revised","0.9.0"],["8","fable-5.1","F8-4","minor","Absolute effects; competing risks","Two baselines on the cause specific route; pooling of time varying hazards.","Agreed. For cause specific hazard ratios the reference risk shown is the reference node cumulative incidence computed by the same route as the treated risk; time varying hazards are piecewise constant on the intervals of digitized reference arm curves, pooled per interval on the log scale. In round 7 our response to grok-4.7 G7-4 did not disclose that the integral route had not been carried into the absolute effects bullet; it now is.","revised","0.9.0"],["8","fable-5.1","F8-5","minor","Contributions; Kaplan-Meier input","Kaplan-Meier contrast missing from the contribution rule; source of its standard error.","Agreed; the contribution rule lists the Kaplan-Meier contrast and the standard error comes from the reported confidence intervals or the Greenwood variance.","revised","0.9.0"],["8","grok-4.7","G8-1","minor","Known status","Parenthesis sends O20 counts to a subdistribution network.","Agreed; complete status counts are subdistribution inputs only for outcomes with a competing event and not applicable inputs for O20.","revised","0.9.0"],["8","grok-4.7","G8-2","minor","Competing risks","Primary network sentence without exception.","Agreed; see C8-4.","revised","0.9.0"],["8","grok-4.7","G8-3","minor","Competing risks","F_ref definition against absolute effects.","Agreed; see F8-3 (one definition with the stated order).","revised","0.9.0"],["8","grok-4.7","G8-4","minor","Kaplan-Meier contrast; contributions","Likelihood and contribution entry.","Agreed; it enters as a normal likelihood on the difference in log cumulative cause specific hazards and is listed in the contribution rule.","revised","0.9.0"],["8","grok-4.7","G8-5","minor","Competing risks","All cause survival gloss is wrong for nonfatal outcomes.","Agreed; see C8-1.","revised","0.9.0"],["8","grok-4.7","G8-6","minor","Data items","Person years fallback with a median.","Resolved by the stricter rule of gpt-6-astra (C8-2): a median follow-up times the number randomized is not time at risk, so no rate is computed from it; another supported route or the relative only rule applies.","revised (alternative)","0.9.0"],["9","gpt-6-astra","C9-1","minor","Competing risks; data items","Piecewise route lacks eligibility and failure rules.","Agreed. The piecewise route now requires both cause specific hazards with valid joint uncertainty from every covering reference arm (events and person time for both events, or the target curve with the competing curve or event free survival from the same arm and risk set); a target curve alone does not qualify; intervals use one grid fixed before pooling; zero event intervals are neither omitted nor patched with constants; otherwise the route is unavailable and the exponential route or the relative only rule applies. Data items list both curves, their uncertainty and risk set information.","revised","0.10.0"],["9","fable-5.1","F9-1","minor","Competing risks","Exponential route has no support source without curves.","Agreed; support comes from curves where published, otherwise from cumulative incidences or event rates at two or more time points, otherwise the exponential approximation is not used.","revised","0.10.0"],["9","fable-5.1","F9-2","minor","Competing risks","Mixed curve availability among covering arms.","Agreed; the piecewise route is used only when every covering reference arm qualifies.","revised","0.10.0"],["9","fable-5.1","F9-3","minor","Data items; competing risks","Which curves; dependence assumption.","Agreed; both curves are required and the assumption for a competing hazard from a competing event curve is stated.","revised","0.10.0"],["9","fable-5.1","F9-4","minor","Competing risks","Common grid for per interval pooling.","Agreed; one grid fixed before pooling (union of reported time points), with hazards derived per interval from counts and person time or from curve increments over event free survival.","revised","0.10.0"],["9","fable-5.1","F9-5","minor","Absolute effects","Leftover Aalen-Johansen parenthetical.","Agreed; the reference risk text now separates outcomes without a competing event from the two hazard types, each with its own order (grok-4.7 G9-1 proposed the same).","revised","0.10.0"],["9","grok-4.7","G9-1","minor","Absolute effects","Subdistribution order applied to cause specific ratios.","Agreed; your sentence adopted.","revised","0.10.0"],["9","grok-4.7","G9-2","minor","Competing risks; data items","A target curve treated as if it supplied both hazards; variances.","Agreed; see C9-1 (requirements, per interval variances and pooling with the predictive median).","revised","0.10.0"],["9","grok-4.7","G9-3","minor","Event outcomes; O20","Kaplan-Meier contrast for O20 has no stated network.","Agreed; Kaplan-Meier contrasts for O20 are not applicable inputs to its single network, and for outcomes with a competing event they remain cause specific.","revised","0.10.0"],["10","gpt-6-astra","C10-1","major","Competing risks","Curve increments were converted to hazards incorrectly.","Agreed; verified (hazards 0.4 and 0.2 per year give one year incidences 0.3008 and 0.1504; the previous rule reintegrates to 0.2421 and the corrected inversion recovers 0.4 and 0.2). The error was ours: in round 9 we replaced the derivation fable-5.1 proposed with an incorrect one. The text now uses the exact inversion under constant hazards, H = -log{S(t_e)/S(t_s)}/Delta and h_k = H Delta F_k/(Delta F_1 + Delta F_2), plus the Kaplan-Meier form, as all three reviewers proposed.","revised","0.11.0"],["10","gpt-6-astra","C10-2","minor","Competing risks","Competing mortality curve exception conflicts with first event risk sets.","Agreed; your stricter rule adopted. Curves qualify only as cumulative incidences of mutually exclusive first events, or as Kaplan-Meier curves of each first event with the other censored; a mortality curve that includes deaths after a nonfatal target event cannot substitute. This replaces the dependence assumption fable-5.1 proposed in round 9, which is no longer needed because such curves are not used.","revised","0.11.0"],["10","gpt-6-astra","C10-3","minor","Competing risks","Failure wording overrode a valid subdistribution conversion.","Agreed; when no support source exists, only the exponential approximation is unavailable and every other eligible route, including a reported Aalen-Johansen incidence for subdistribution conversion, is kept.","revised","0.11.0"],["10","fable-5.1","F10-1","minor","Competing risks; data items","Biased per interval estimator; curve estimator unrecorded.","Agreed; your replacement paragraph adopted, with the estimator of each curve recorded in the data items. Our round 9 response recorded F9-4 as agreed although the derivation was replaced by an incorrect one; this is corrected here.","revised","0.11.0"],["10","fable-5.1","F10-2","minor","Competing risks","Reported intervals undefined for curves.","Agreed; intervals are those of reported events and person time or, for curves, between published numbers at risk.","revised","0.11.0"],["10","fable-5.1","F10-3","minor","Competing risks","No variance source; exponential route pooling and zero counts.","Agreed; variance 1/d_k from reported or reconstructed counts, unavailability with a zero count or without numbers at risk, and the same pooling and zero count rule for the exponential route.","revised","0.11.0"],["10","fable-5.1","F10-4","minor","Absolute effects","Low and high risk scenarios under competing risks.","Agreed; the scenarios use each covering arm own cumulative incidence by the route in use.","revised","0.11.0"],["10","grok-4.7","G10-1","minor","Competing risks","Curve formula is not a piecewise constant hazard; grid and curve pairs.","Agreed; see C10-1 and F10-1. Both allowed curve pairs are handled (event free survival gives F_2 = 1 - S - F_1), and hazards are estimated on each arm own intervals and carried onto the grid.","revised","0.11.0"],["11","gpt-6-astra","C11-1","major","Competing risks","Curve derived counts and variances are not identified.","Agreed; your rule adopted. Counts are not inferred from curve increments and numbers at risk; curve inversion requires a reported or author supplied joint covariance of the curve ordinates, propagated through the Jacobian of the inversion and preserved in pooling; otherwise the route is unavailable and another eligible route is kept. Reported counts with valid person time keep the 1/d_k approximation. fable-5.1 (F11-2a) and grok-4.7 (G11-1) proposed explicit reconstruction formulas; we did not adopt them because, as your counterexample shows, within interval censoring leaves the counts unidentified, and even known counts do not give the variance of a curve inverted log hazard.","revised","0.12.0"],["11","gpt-6-astra","C11-2","minor","Absolute effects","Scenario wording required hazards for the reported incidence route.","Agreed; scenarios use the reported Aalen-Johansen value when that is the source, otherwise the cumulative incidence from the arm own qualifying hazards.","revised","0.12.0"],["11","fable-5.1","F11-1","minor","Competing risks","Shared uncertainty across subdivided grid intervals.","Agreed; hazards are carried onto grid intervals with their uncertainty and as one draw.","revised","0.12.0"],["11","fable-5.1","F11-2","minor","Competing risks","Count reconstruction; Kaplan-Meier condition; final partial interval; number at risk rule.","Partly adopted. (a) Count reconstruction was replaced by the joint covariance requirement of C11-1, so no counts are reconstructed from curves; (b) the Kaplan-Meier condition 0 < S_k(t_e) < S_k(t_s) is stated; (c) the interval from the last published time to the analyzed horizon is a reported interval when the curve extends to it; (d) count intervals do not need a published number at risk (the number at risk rule no longer exists).","partly revised","0.12.0"],["11","fable-5.1","F11-3","minor","Absolute effects","Scenario wording.","Agreed; see C11-2.","revised","0.12.0"],["11","fable-5.1","F11-4","minor","Front matter","Date.","Agreed; dated 2026-10-13.","revised","0.12.0"],["11","grok-4.7","G11-1","minor","Competing risks; data items","Curve count equation; Kaplan-Meier condition; number at risk rule; exponential count; data items wording.","Partly adopted. The reconstructed count equations were not adopted (see C11-1: curve inversion now requires reported or author supplied joint covariance); the separate Kaplan-Meier condition, count intervals without a number at risk, the definition of d_k for the exponential route, the data items wording and one shared draw for subdivided intervals are adopted.","partly revised","0.12.0"],["11","grok-4.7","G11-2","minor","Absolute effects","Scenarios under a reported cumulative incidence.","Agreed; your wording adopted (see C11-2).","revised","0.12.0"],["12","grok-4.7","G12-M1","major","Competing risks","The Jacobian was of the hazards, not the log hazards; per interval pooling dropped the shared draw.","Agreed; your replacement passage adopted. J is the Jacobian of the map from ordinates to log hazards; reported counts give independent log hazards with variance 1/d_k; copies of one reported interval are one coordinate; the arms are pooled in one multivariate random effects model, the reference hazard per grid interval is the median of its marginal predictive distribution, and integrated risks and scenarios use joint predictive draws. fable-5.1 (F12-1) raised the same dependence point.","revised","0.13.0"],["12","fable-5.1","F12-1","minor","Competing risks","How per interval pooling preserves within arm dependence.","Agreed; see G12-M1 (one multivariate model with joint predictive draws).","revised","0.13.0"],["12","fable-5.1","F12-2","minor","Data items","Self contradictory where published wording.","Agreed; where published now qualifies the curves, and covariances are requested from authors with their ordinate estimates when not published.","revised","0.13.0"],["12","fable-5.1","F12-3","minor","Competing risks","Ordinates paired with an author supplied covariance.","Agreed; author supplied covariances are paired with the authors own ordinate estimates, and digitized values are inverted only with a published covariance at the same times.","revised","0.13.0"],["12","fable-5.1","F12-note","note","Absolute effects; priors","Reported Aalen-Johansen values rarely sit at the analyzed horizon; no heterogeneity prior row for the reference arm meta-analysis.","Noted as pre-existing matters to settle by amendment before the first analysis, as the reviewer suggests; they are recorded here so that they are not lost.","deferred to amendment","0.13.0"],["13","gpt-6-astra","C13-1","major","Competing risks; priors","Multivariate pooling lacked common coordinates and a between arm covariance.","Agreed; your conservative restriction adopted: piecewise integration requires identical reported interval boundaries across all covering arms through the horizon, otherwise the route is unavailable. The model is now written: y_a ~ N(mu + Z u_a, V_a) with one arm level random effect per event shared across intervals (the structure fable-5.1 proposed), u_ak ~ N(0, tau_k^2), priors PR15 (mu) and PR16 (tau_k) with sensitivity priors, hazards per year, and a one arm rule. We used this structure rather than an unstructured covariance with an LKJ prior because with two to five covering arms a free between arm covariance over intervals and events is weakly identified, and a proportional deviation of each arm hazard function is the usual form of baseline risk heterogeneity.","revised","0.14.0"],["13","gpt-6-astra","C13-2","minor","Competing risks; absolute effects","Scenario uncertainty targeted the predictive distribution.","Agreed; your text adopted: scenario values remain empirical percentiles of arms own cumulative incidences, with uncertainty from arm specific draws N(y_a, V_a) recomputed per draw.","revised","0.14.0"],["13","gpt-6-astra","C13-3","minor","Competing risks","Count variance stated as exact.","Agreed; the working normal approximation with variance approximately 1/d_k and zero covariance under its standard assumptions.","revised","0.14.0"],["13","fable-5.1","F13-1","minor","Competing risks; priors","Multivariate model structure and prior.","Agreed; your structure (one shared random effect per event per arm) adopted with priors PR15 and PR16. Because identical interval boundaries are now required (C13-1), no observation equation across different partitions is needed. PR17 also settles the reference arm heterogeneity prior deferred in round 12.","revised","0.14.0"],["13","fable-5.1","F13-2","minor","Competing risks","Scenario values tied to the wrong distribution.","Agreed; see C13-2.","revised","0.14.0"],["13","fable-5.1","F13-note","note","Absolute effects","Point estimate versus median of draws.","Adopted: the displayed reference cumulative incidence is the median of the integrated draws.","revised","0.14.0"],["13","grok-4.7","G13-M1","major","Competing risks","No mean for a reported interval covering several grid intervals; displayed incidence unspecified.","Resolved by requiring identical reported interval boundaries across covering arms (C13-1), so every coordinate maps to one common interval and no observation equation across partitions is needed; the displayed reference cumulative incidence is the median of the joint predictive draws of the integrated incidence, as you proposed. Your length weighted observation equation was not needed under this restriction; it remains the natural amendment if unequal partitions must be accommodated.","revised (alternative)","0.14.0"],["13","grok-4.7","G13-1","minor","Competing risks; absolute effects","Scenario uncertainty.","Agreed; see C13-2.","revised","0.14.0"],["14","grok-4.7","G14-1","minor","Competing risks","The one arm rule contradicted the predictive draw, and exponentiation was not stated.","Agreed; your text adopted: with one covering arm u_new is not drawn and each draw is from N(y_a, V_a); with two or more, mu + Z u_new at that draw tau_k; hazards are the exponentials of the log hazards, integrated jointly; scenario draws are exponentiated before integration. fable-5.1 (F14-3) raised the one arm point too.","revised","0.15.0"],["14","grok-4.7","G14-2","minor","Competing risks","y_a and V_a not assigned for counts.","Agreed; y_a is the vector of estimated log hazards, one coordinate per event and common interval, in that order, and for counts V_a is diagonal with variances approximately 1/d_k.","revised","0.15.0"],["14","grok-4.7","G14-3","minor","Competing risks","A single shared interval skipped the constant hazard gate; d_k ambiguous.","Agreed; your text adopted: when the identical boundaries are the single interval from 0 to the horizon, the piecewise fit is the exponential approximation, with its support judgment and label; the separate exponential route applies only when the piecewise route is unavailable and every arm supplies counts and person time from 0 to the horizon; d_k is the count from 0 to the horizon. fable-5.1 (F14-1) raised the same overlap.","revised","0.15.0"],["14","fable-5.1","F14-1","minor","Competing risks","One interval per arm belonged to two routes.","Agreed; resolved with the wording of G14-3, which also settles arms whose totals cover different follow-up: only counts from 0 to the horizon qualify for the single interval.","revised","0.15.0"],["14","fable-5.1","F14-2","minor","Absolute effects","PR17 cited where PR16 applies.","Agreed; now \"heterogeneity prior PR17; PR16 for the competing risk routes\".","revised","0.15.0"],["14","fable-5.1","F14-3","minor","Competing risks","One arm distribution implicit.","Agreed; see G14-1.","revised","0.15.0"],["14","fable-5.1","F14-4","minor","Absolute effects","Pooling scale for mixed Kaplan-Meier and rate inputs.","Agreed; your text adopted: both enter one model on the log cumulative hazard at the horizon (log{-log(1-F)} with delta method variance; log(lambda t) with variance approximately 1/d) under PR17.","revised","0.15.0"],["14","fable-5.1","F14-note","note","Priors; absolute effects","PR16 scale 1 versus 0.5 without stated reason; reported Aalen-Johansen values that miss the horizon remain deferred.","PR16 uses Half-normal(0; 1^2) to match PR17 and the weakly informative log scale used for baseline risks, with 0.5 and 2 as sensitivity priors; the horizon matter stays deferred to the pre-analysis amendment, as agreed in round 12.","noted; deferred to amendment","0.15.0"],["15","gpt-6-astra","C15-1","minor","Absolute effects","Undefined inputs in reference risk pooling: F = 0 or 1 and zero event rates make the normal model undefined.","Agreed that the inputs needed a rule; resolved by an alternative that keeps the arms. Under your replacement a single zero event reference arm would disable absolute effects for the whole network, and zero event reference arms are routine for the many complementary log log harms outcomes (for example diabetic ketoacidosis, retinopathy complications, acute kidney injury). The reference arm meta-analysis now uses the likelihood of each arm own data: binomial with logit link for fixed time outcomes; for hazard based outcomes a Poisson likelihood for counts with person time and a binomial likelihood with complementary log log link and log follow-up offset for participants with an event, both on the log cumulative hazard at the horizon; a reported Kaplan-Meier value enters by the normal approximation only when 0 < F < 1 with a reported confidence interval or Greenwood variance. Zero counts need no correction and no arm is dropped for having none, which keeps the two requirements of your text (no omission because of the event count, no continuity correction). The mean prior is PR02 and the one arm case uses that arm likelihood under PR02. The competing risk routes keep the working normal likelihood, because curve inputs have no count likelihood; the protocol now states this.","revised (alternative)","0.16.0"],["15","grok-4.7","G15-1","minor","Competing risks","The separate exponential route contradicted its own availability rule; gate and label only in the first sentence.","Agreed; the bullet is rewritten in one passage: a piecewise fit whose common interval is a single interval from 0 is the exponential approximation; when the piecewise route is unavailable, arms with counts and person time over one period from 0 that reaches or passes the horizon (finer intervals summed) give one coordinate per event, pooled with the same model; in both cases the constant hazard gate and the label apply. The clause that the constructed interval met the identical boundary requirement is removed. Periods that pass the horizon are admitted (fable-5.1 F15-1), matching the rule for outcomes without a competing event, since under the constant hazard gate a rate over a longer period estimates the same hazard.","revised","0.16.0"],["15","grok-4.7","G15-2","minor","Absolute effects","Units of t and the variance input of the log cumulative hazard model unspecified.","Agreed; t is the analyzed horizon in years (weeks divided by 52.18), rates are per person year, and the Kaplan-Meier variance comes from the reported confidence interval or the Greenwood variance, requiring 0 < F < 1. Counts now enter by their exact likelihoods (C15-1).","revised","0.16.0"],["15","fable-5.1","F15-1","minor","Competing risks","Single reported intervals that pass the horizon.","Agreed; your second closure adopted in substance (see G15-1): single periods from 0 that reach or pass the horizon qualify, finer count intervals are summed, and the gate and label apply in every case; the piecewise bullet now states that a last count interval may end after the horizon, with its rates applied up to the horizon.","revised","0.16.0"],["15","fable-5.1","F15-2","minor","Absolute effects","\"Reported interval\" ambiguous; Kaplan-Meier without uncertainty.","Agreed; \"reported confidence interval or Greenwood variance\". A Kaplan-Meier value without either enters through the arm counts; an arm with no qualifying input is listed with the reason. Because zero counts always qualify under the count likelihoods, such exclusion depends on reporting, not on the event count.","revised","0.16.0"],["15","fable-5.1","F15-3","minor","Priors; absolute effects","No prior on the mean of the reference arm meta-analysis.","Agreed; PR02 now applies to the mean of the reference arm meta-analysis on the same scales, and the text cites it.","revised","0.16.0"],["15","fable-5.1","F15-note","note","Absolute effects","Reported Aalen-Johansen values that miss the analyzed horizon.","Remains deferred to the pre-analysis amendment, as agreed in round 12.","deferred to amendment","0.16.0"],["16","gpt-6-astra","C16-0","note","Response record","The round 15 response to C15-1 cited ketoacidosis, retinopathy and acute kidney injury, but the new count likelihoods apply only to hazard based outcomes without a competing event (at present O20) and to fixed time binomial outcomes.","Agreed; that justification was wrong and is corrected here. The alternative retains zero event reference arms for O20 and for fixed time binomial outcomes. It does not change the zero count restriction of the competing risk integration routes, which still applies to the harms outcomes named. Extending count likelihoods to those routes (fable-5.1, round 16) is listed for the pre-analysis amendment together with the Aalen-Johansen horizon matter of F12-note.","corrected","0.17.0"],["16","gpt-6-astra","C16-1","minor","Absolute effects","Reporting exclusions bypassed the one arm safeguard; contributing set for scenarios undefined; binomial input needed the known status rule.","Agreed; a usable reference arm covers the horizon and supplies a qualifying input, arms without one are listed and excluded from the reference risk and the scenarios, two or more usable arms fit the random effects model, one uses its likelihood with PR02 and no heterogeneity, none triggers the relative only rule, and one usable arm gives both scenarios. The binomial input uses the longest follow-up at which status is known for every randomized participant, reaching the horizon, with the 95 percent rule only in its sensitivity analysis.","revised","0.17.0"],["16","gpt-6-astra","C16-2","minor","Priors","PR02 rationale did not distinguish the annual hazard from the cumulative hazard at the horizon.","Agreed; the rationale now states both parameterizations and the logit scale of fixed time reference arm models.","revised","0.17.0"],["16","grok-4.7","G16-1","minor","Absolute effects","The one arm sentence conflicted with the competing risk one arm draw.","Agreed; the one arm rule for outcomes without a competing event uses the arm likelihood with PR02; for outcomes with competing events the text now restates the rule of the competing risks section (tau_k not estimated, draws from N(y_a, V_a)).","revised","0.17.0"],["16","grok-4.7","G16-2","minor","Absolute effects","An arm with person time and a follow-up proportion had two count likelihoods.","Agreed; Poisson when the count is reported with valid person time over one period, otherwise binomial, and an arm that reports both enters only the Poisson likelihood.","revised","0.17.0"],["16","grok-4.7","G16-3","minor","Priors","PR02 rationale describes a per year baseline for every complementary log log use.","Agreed; see C16-2.","revised","0.17.0"],["16","fable-5.1","F16-1","minor","Absolute effects","Precedence between count likelihoods and definition of f_a.","Agreed; see G16-2 and C16-1 (known status rule for f_a).","revised","0.17.0"],["16","fable-5.1","F16-2","minor","Absolute effects","Scenario risks for arms that enter by counts.","Agreed; your text adopted: each usable arm risk is 1 - exp(-exp(theta_a)) from its own likelihood under PR02 alone, with uncertainty from independent draws and percentiles recomputed per draw.","revised","0.17.0"],["16","fable-5.1","F16-3","minor","Absolute effects","Scale of the delta method from a reported confidence interval.","Agreed; standard error from both limits transformed by log{-log(1 - .)}, half width over 1.96, or the Greenwood variance by the delta method.","revised","0.17.0"],["16","fable-5.1","F16-4","minor","Competing risks","Boundary matching when a last interval passes the horizon.","Agreed; boundaries at or before the horizon must be identical, a boundary after the horizon counting as the horizon.","revised","0.17.0"],["16","fable-5.1","F16-note","note","Competing risks; priors","Reason for keeping the working normal likelihood in competing risk routes is thin; PR02 rationale clause.","The reason now states only the actual choice (one multivariate normal likelihood for count and curve arms); count likelihoods for those routes are deferred to the pre-analysis amendment (see C16-0). The PR02 rationale is revised (C16-2).","revised; part deferred to amendment","0.17.0"],["17","gpt-6-astra","C17-1","minor","Absolute effects","Transformed confidence limits can be infinite, and 1.96 assumes 95 percent coverage.","Agreed; a Kaplan-Meier value needs 0 < F < 1 and a finite positive standard error on g(F) = log{-log(1 - F)}: from a two sided interval (L, U) of known level with 0 < L < U < 1, {g(U) - g(L)}/(2 z_{1 - alpha/2}), otherwise from the Greenwood variance; when no Kaplan-Meier value qualifies, the Poisson and then binomial inputs apply. fable-5.1 (F17-6b) raised the coverage point too.","revised","0.18.0"],["17","gpt-6-astra","C17-2","minor","Absolute effects","The one arm restatement extended log hazard sampling to a reported Aalen-Johansen value used directly.","Agreed; the restatement is limited to the piecewise integration and exponential approximation routes, and the text says it does not apply to a reported Aalen-Johansen cumulative incidence used directly as F_ref.","revised","0.18.0"],["17","fable-5.1","F17-0","note","Response record","The round 16 response to F16-2 said \"your text adopted\" although the Kaplan-Meier branch was simplified.","Agreed; the record should have said partly adopted: every usable arm, including a Kaplan-Meier arm, takes its scenario risk from its own likelihood under PR02 alone, which you judged harmless.","corrected","0.18.0"],["17","fable-5.1","F17-1","minor","Absolute effects","Poisson input \"over one period\" excluded summed finer intervals.","Agreed; \"over one period, or summed over reported finer intervals\".","revised","0.18.0"],["17","fable-5.1","F17-2","minor","Absolute effects; priors","Reference risk with one usable arm unnamed; PR02 scope.","Agreed; with one usable arm the reference risk is the posterior median of 1 - exp(-exp(theta_a)), and PR02 applies_to names the single arm and scenario uses.","revised","0.18.0"],["17","fable-5.1","F17-3","minor","Absolute effects","Known status for f_a read trial wide.","Agreed; status known for every participant randomized to that arm.","revised","0.18.0"],["17","fable-5.1","F17-4","minor","Absolute effects","Scenario inputs for fixed time outcomes unstated.","Deferred to the pre-analysis amendment, as you allowed; the matter predates 0.17.0 and adding a new rule now would extend the revision beyond the comments of this round.","deferred to amendment","0.18.0"],["17","fable-5.1","F17-5","minor","Absolute effects","Scenario values on zero count arms are prior dependent.","Deferred to the pre-analysis amendment with F17-4 (labeling and display under the sensitivity prior of PR02).","deferred to amendment","0.18.0"],["17","fable-5.1","F17-6","minor","Editorial","Name the relative only rule at its definition; level of the confidence interval; pointer to deferred matters.","Agreed; the defining passage now names the relative only rule, the confidence level is general (C17-1), and the Amendments paragraph points to the comments marked deferred to amendment.","revised","0.18.0"],["18","grok-4.7","G18-1","minor","Absolute effects","The arm level f_a clause cited the trial wide complete status rule of the hazard ratio analysis.","Agreed; known status is the definition of the outcomes section applied to that arm alone rather than trial wide, and the sensitivity analysis uses the longest follow-up at which status is known for at least 95 percent of that arm, with unknown participants kept in the denominator as event free. fable-5.1 (F18-1) raised the same point.","revised","0.19.0"],["18","grok-4.7","G18-2","minor","Amendments","The pointer matched only the exact action \"deferred to amendment\".","Agreed; the pointer now covers every row whose action field contains \"deferred to amendment\" (F12-note, F14-note, F15-note, F16-note, F17-4, F17-5). fable-5.1 (F18-2) raised the same point.","revised","0.19.0"],["18","fable-5.1","F18-0","note","Response record","The stated reason for deferring F17-4 did not hold on its own terms.","Agreed; the reason should have been that fixed time scenario inputs belong with the deferred reference arm likelihood work (F12-note, F16-note) in the same pre-analysis amendment. The record is corrected here; the deferral itself is unchanged.","corrected","0.19.0"],["18","fable-5.1","F18-1","minor","Absolute effects","Cross reference after the arm level rule.","Agreed; see G18-1.","revised","0.19.0"],["18","fable-5.1","F18-2","minor","Amendments","Scope of the Amendments pointer.","Agreed; see G18-2.","revised","0.19.0"],["18","fable-5.1","F18-3","note","Absolute effects","A reported Aalen-Johansen value used directly has no stated one arm or pooling rule.","Agreed that it belongs to the deferred F12-note matter and stays there.","deferred to amendment","0.19.0"],["19","fable-5.1","F19-1","minor","Response record","The G18-2 response listed six deferred rows; F18-3, deferred in the same round, makes seven.","Agreed; the set covered by the pointer at version 0.19.0 is F12-note, F14-note, F15-note, F16-note, F17-4, F17-5 and F18-3. The protocol rule needed no change.","corrected","0.19.0"],["19","fable-5.1","F19-2","minor","Absolute effects","Scope of \"and f_a reaches the analyzed horizon\" after the sensitivity parenthetical.","The reading that the condition governs f_a in both analyses is the intended one, and the text forbids extrapolation in any case. Version 0.19.0 is the version all three reviewers accepted, so the wording \"in either analysis\" is deferred to the pre-analysis amendment rather than made now.","deferred to amendment","0.19.0"],["19","fable-5.1","F19-3","note","Sensitivity analyses","The reference arm 95 percent analysis is defined inline but not listed in the Sensitivity analyses section.","Deferred to the pre-analysis amendment for the same reason as F19-2; the inline definition is executable.","deferred to amendment","0.19.0"]]}