Peer review round 8: LivingGLP1 protocol 0.8.0 (reviewer grok-4.7)
Round 7 comments
G7-M1. Resolved. Complete status counts for an outcome with a competing event enter only its subdistribution network. For an outcome with no competing event, hazard ratios and complete status counts share one not applicable network. The text limits that set, at present, to all cause mortality (O20). That matches outcomes.csv: O20 is the only all cause mortality row. O21, O22D, O35B, and O40 still have a competing event. Calling the shared O20 network primary is the right label.
G7-1. Resolved. The time varying sentence places HR times λ1(u) inside the integral and inside survival, and replaces λ2(u) in those same places only when a cause specific hazard ratio for the competing event is reported. Otherwise λ2(u) is unchanged, under the no effect assumption already stated for the exponential formula. That is the treated cumulative incidence. The name given to the survival function is Minor 5.
G7-2. Partly resolved. A contrast of one minus Kaplan-Meier, with competing deaths censored, now enters as log(-log S) at a common t, with a delta method standard error, only when 0 < 1 - S < 1. It is classified as cause specific, and it remains barred as an absolute or reference risk. The likelihood list points to that description. The cumulative incidence contrast is explicitly a normal likelihood; this one is not. "Without continuity correction in the contribution network" does not say that the contribution network uses this contrast. The contribution list still names only the log hazard ratio, the cumulative incidence contrast, and the complementary log log count estimate. See Minor 4.
G7-3. Resolved. Conclusions for a time to event outcome use the primary hazard type network when it meets the connectivity rule, otherwise the other hazard type network labeled with its hazard type, otherwise descriptive reporting. Where a competing event exists, cause specific is primary and subdistribution is the other network, which is the requested order. Descriptive reporting as the last step is a sound addition when neither network can be synthesized. Recorded harms remains limited to binary harms. Counts from the 95 percent sensitivity are a sensitivity of the subdistribution network, or of the not applicable network for O20, and the sensitivity list assigns them the same way.
G7-4. Resolved. The response adopts the other reviewer's order rather than limiting subdistribution conversions to a reported Aalen-Johansen incidence. That reason is sound. For an outcome with competing events a single event rate is forbidden. The reference risk is the reported Aalen-Johansen cumulative incidence, otherwise the two rate exponential cumulative incidence under a supported constant hazard, otherwise the relative only rule. That exponential value is a cumulative incidence, and using it as F in 1 - (1 - F)^HR does not revive the net risk 1 - exp(-λ1 t). The earlier "otherwise their event rate under a constant hazard" is overridden by the next sentence. The symbol F_ref in Competing risks was not updated. See Minor 3.
New issues
Minor 1. Known status, Event outcomes and harms. Competing deaths are "counted as event free, which is why complete status counts are subdistribution inputs." That reason holds only when a competing event exists. Applied to O20, it sends all cause mortality counts to a subdistribution network, which the opening paragraph and the not applicable definition forbid. Replace the parenthesis with: "counted as event free; complete status counts are subdistribution inputs only for an outcome with a competing event, and for O20 they are not applicable inputs."
Minor 2. Competing risks, primary network sentence. "For time to event outcomes the cause specific network is primary" has no exception, while the opening paragraph says the O20 network is primary. O20 is a time to event outcome, so the conclusions rule has two primaries. Replace with: "For a time to event outcome with a competing event, the cause specific network is primary, because outcome trials report Cox models. For an outcome with no competing event, the not applicable network is the only network and is primary. For binary harms the subdistribution network is primary, because counts and cumulative incidence estimate it."
Minor 3. Competing risks, subdistribution formula, against Absolute effects. The formula says F_ref is the reference node's Aalen-Johansen cumulative incidence. Absolute effects says that when that incidence is unreported, the exponential expression serves as F_ref. Replace the definition with: "where F_ref is the reference risk at that horizon from the reference risk procedure: the reported Aalen-Johansen cumulative incidence, otherwise the exponential expression from the pooled λ1 and λ2 when the constant hazard assumption is supported. That value is the reference risk shown with the estimate. If neither is available, apply the relative only rule."
Minor 4. Kaplan-Meier contrast and Confidence in cumulative evidence. In Competing risks, add: "That contrast enters the cause specific network as a normal likelihood and is the difference in log cumulative cause specific hazards." In the contribution sentence, add: "or the Kaplan-Meier contrast of log(-log S) with its delta method variance and no continuity correction."
Minor 5. Competing risks, time varying integral. Placing λ1(u) and λ2(u) inside survival is the correct event-free survival. The gloss "all cause survival (any death an event)" is wrong for a nonfatal outcome such as hospitalization for heart failure (O23): all cause survival keeps people who already had the event in the risk set. Replace that gloss with: "event-free survival, in which both the outcome and competing death are events."
Minor 6. Data items, person years. The fallback is the number randomized times mean follow-up, "with median follow-up flagged," and it does not state the product when only a median is reported. λ1 and λ2 are events per person year on that denominator. Replace with: "person years by arm as reported; if unreported, the number randomized times the arm's mean follow-up; if only a median follow-up is reported, use the median, flag the arm, and do not treat the resulting rates as exact person time."
Checks that held
Q still matches the complementary log log contrast with offset log t, and the recorded harms offset is now written log t. The exponential replacement of λ1 everywhere, and 1 - (1 - F_ref)^HR, remain the right expressions. The iGlarLixi switching row remains a template and stays out of networks until the strategy identifier, permitted pens, switching threshold, and titration algorithm are recorded. Nonproportionality is material when a reported test rejects proportionality at the 5 percent level or the arms' cumulative incidence curves cross after the first quarter of follow-up. Both that judgment and support for constant hazards are made by two reviewers before relative effects are modeled and logged with reasons.
Pre-registration conditions
These remain for the review owner and do not by themselves withhold acceptance: team names, affiliations, and postal address; funding and competing interests; PRESS 2015 review by a human information specialist; confirmation of Embase access, with omission stated as a limitation if access is absent; verification of the Chinese terms by a Chinese-reading team member; confirmation and locking of provisional thresholds by clinical advisors and patient partners before any outcome data are extracted; an OpenAlex API key; and a logged harvest of all four API sources with the registered concepts version, with every count reconciled, before the baseline search.
Judgment
G7-M1, G7-1, G7-3, and G7-4 are resolved. G7-2 is resolved for the primary estimator and incomplete for the contribution network. The known status parenthesis and the unqualified primary network sentence still contradict the O20 rule. The other new items are sentence fixes. With those edits, and the administrative conditions completed, the methods would be acceptable for registration.
VERDICT: MINOR REVISION