Peer review round 11: LivingGLP1 protocol 0.11.0 (reviewer grok-4.7)

1. Round 10 comments

G10-1. Resolved. Piecewise integration now inverts constant cause specific hazards on each arm's own reporting intervals and carries those hazards onto the common grid. For Aalen-Johansen curves, (S = 1 - F_1 - F_2), with (F_2 = 1 - S - F_1) when event free survival is published instead, (H = -\log{S(t_e)/S(t_s)}/\Delta), and (h_k = H,\Delta F_k/(\Delta F_1 + \Delta F_2)). For a Kaplan-Meier curve of event (k) with the other event censored, (h_k = -\log{S_k(t_e)/S_k(t_s)}/\Delta). Both allowed curve pairs are identified, and event free survival alone does not qualify. Hazards 0.4 and 0.2 per year give one year incidences 0.3008 and 0.1504, and these formulas return 0.4 and 0.2. Hazards 0.2 and 0.1 per year over five years return 0.2 and 0.1. The event free survival pairing returns the same hazards. Adopting this inversion was the right correction. The wording gaps below are new.

2. New issues

There are no major errors. Relative effects, network keys, the O20 assignment, and the order of absolute risk routes are unchanged. The new hazard point estimate is exact under the stated constancy. Its weight is not yet an executable rule.

Minor 1. The curve count behind the variance is not an equation, and one unavailability rule covers the wrong inputs. File: protocol/protocol.qmd, Competing risks, Piecewise integration, from "These forms are exact" through the zero count sentence. The random effects weight uses (d_k). "Reconstructed from the increment and the number at risk at the start" admits more than one count. With 500 people at risk, event free survival 0.8, and increments 0.05 and 0.02, the increment divided by survival gives 31 events, the number at risk times the increment gives 25, and the number at risk times one minus the survival ratio gives about 44. Those weights change the pooled reference hazard when arms disagree. The same sentences require (0 < S(t_e) < S(t_s)) for both forms, while (S) is event free survival and the Kaplan-Meier form uses (S_k). A qualifying Kaplan-Meier pair can fail that condition. Data items call every digitized curve a cumulative incidence, including a Kaplan-Meier estimator, whose input here is the decreasing survival. The unavailability rule also withholds the route from a count interval that has events and person time and lacks a published number at risk, although that interval's variance is (1/d_k) from the reported count. In the exponential bullet, (d_k) is the same symbol and is not defined as the total count for (\lambda_k).

Replace "These forms are exact under the stated constancy and require (0 < S(t_e) < S(t_s)). The variance of each log hazard is (1/d_k), where (d_k) is the reported event count or, for curves, the count reconstructed from the increment and the number at risk at the start of the interval. An arm interval with a zero count for either event, or without a published number at risk, makes this route unavailable" with: "The Aalen-Johansen form requires (0 < S(t_e) < S(t_s)). The Kaplan-Meier form requires (0 < S_k(t_e) < S_k(t_s)) for each event, where (S_k) is that curve's survival. The variance of each log hazard is (1/d_k). For counts, (d_k) is the reported event count. For Aalen-Johansen curves, (d_k = n(t_s),\Delta F_k/S(t_s)), where (n(t_s)) is the number at risk at the start of the interval. For a Kaplan-Meier curve, (d_k = n_k(t_s),(1 - S_k(t_e)/S_k(t_s))), where (n_k(t_s)) is that curve's number at risk. These curve counts use the risk set at the start of the interval and are used for the variance and the zero count rule. An arm interval with a zero count for either event makes this route unavailable. A curve interval without a published number at risk at its start also makes it unavailable. A count interval uses the reported counts. When one reported interval is carried onto several grid intervals, each copy uses that same variance, and the copies are one shared draw when uncertainty in the integrated risk is propagated. No constants are added and no interval or arm is omitted."

In the exponential bullet, replace "variance (1/d_k)" with "variance (1/d_k), where (d_k) is that arm's total count of event (k) over the person time used for (\lambda_k)".

In Data items, replace "digitized reference arm cumulative incidence curves for both the target event and the competing first event (or event free survival), with the estimator of each curve (Kaplan-Meier with the other event censored, or Aalen-Johansen)" with "digitized reference arm curves for both first events, or the target cumulative incidence with event free survival: Aalen-Johansen cumulative incidence, or the Kaplan-Meier survival of each first event with the other censored".

Minor 2. Low and high scenarios under a reported cumulative incidence have no hazard source. File: protocol/protocol.qmd, Absolute effects, the parenthetical in the reference risk sentence. For a subdistribution hazard ratio the first source of (F_{\mathrm{ref}}) is the reported Aalen-Johansen cumulative incidence. The parenthetical still takes each scenario risk from that arm's hazards. When the reported value is the source and piecewise hazards are unavailable, the scenario has no hazard to recompute. The scenario risks are the covering arms' own values of the source actually in use.

Replace "each covering arm's own cumulative incidence from its own hazards by the route in use" with "each covering arm's own cumulative incidence at the analyzed horizon by the route in use: the reported Aalen-Johansen value when that value is the source, otherwise the cumulative incidence computed from that arm's own hazards".

3. Checks and registration

The authoritative tables match the version 0.10.0 snapshot except protocol/amendments.csv row A010, which logs this revision. outcomes.csv has 77 outcomes. search/strategies/concepts.yaml version 1.2.0 is unchanged. R/search_strategy.R builds the API queries from the class terms, agent terms, mechanism block, and PubMed vocabulary. The Chinese terms remain a separate key for the manual CNKI and Wanfang searches. The iGlarLixi switching row remains a template (is_template yes) and is excluded from every network until a strategy is recorded. The renderer was read and was not executed.

Team names, affiliations, postal address, funding, competing interests, confirmation of Embase access, PRESS review by a human information specialist, verification of the Chinese terms by a Chinese-reading team member, and confirmation and locking of provisional thresholds by clinical advisors and patient partners remain conditions on the administrative checklist. This verdict is based on the methods. The hazard point estimate is acceptable. The variance equation and the scenario source should be written as above before registration. With those sentences corrected, the methods are acceptable for registration once the administrative conditions are completed.

VERDICT: MINOR REVISION