# Retatrutide Phase 3 Results Ledger: data dictionary

Version 1.0. Checked October 14, 2026. Compiled by Peptide Care Index.

This release contains four UTF-8 CSVs. They record published group estimates and selected safety reports, not patient-level data. Every result has an evidence URL and a check date. Blank means unavailable or not asserted; it never means zero. A printed `0.0` is a reported zero.

## Results file

**File:** `retatrutide-phase-3-results.csv`. **Rows:** 22: 18 main-trial arms across five studies, plus four original-assignment groups in the selected TRIUMPH-1 extension.

| Column | Format | Meaning |
| --- | --- | --- |
| `record_id` | Text; unique | Stable identifier for the trial, cohort, assignment and time point. |
| `trial` | Text | Study name. |
| `registry_id` | Text | ClinicalTrials.gov NCT identifier; see its separate evidence URL. |
| `population` | Text | Studied population, including material eligibility or extension selection rules. |
| `cohort` | Text | Main randomized trial or selected extension. |
| `randomized_trial_n` | Integer or blank | Whole main-trial randomized sample. Repeated on each arm row: count each trial once, never sum this column across rows. Blank for extension rows. |
| `randomized_arm_n` | Integer or blank | Participants originally randomized to that arm, when verified. Not a safety or completer denominator. Blank when not available; also blank for extension subgroups. |
| `arm_n_source_url` | URL | Evidence for an arm count when available; a link on a blank count does not assert that count. |
| `endpoint` | Text | Mean percentage change from original baseline in body weight. These are statistical estimates, not responder percentages. |
| `timepoint_weeks` | Integer | Nominal endpoint time from original baseline: 40, 68, 80 or 104 weeks. |
| `original_assignment` | Text | Original randomized research group, such as 12 mg or Placebo. Research labels are not dosing advice. |
| `extension_treatment` | Text or blank | Actual extension treatment. Original placebo assignment does not mean continuing placebo in the extension. |
| `efficacy_change_pct` | Decimal or blank | Published efficacy-estimand weight change, with its negative sign retained. `-28.3` means a 28.3% reduction from baseline. |
| `treatment_regimen_change_pct` | Decimal or blank | Published treatment-regimen-estimand weight change, also signed. |
| `treatment_regimen_status` | Text | Reported or not reported in the retrieved primary material. |
| `efficacy_source_url` | URL | Exact evidence source for the efficacy estimate. |
| `efficacy_source_date` | YYYY-MM-DD | Publication date of that evidence source. |
| `treatment_regimen_source_url` | URL or blank | Exact evidence source for the regimen estimate. |
| `treatment_regimen_source_date` | YYYY-MM-DD or blank | Publication date of that evidence source. |
| `publication_status` | Text | Evidence maturity at the check date; publication existence and extent of paper reading are separate facts. |
| `last_checked` | YYYY-MM-DD | Actual verification date, not the trial observation date. |
| `notes` | Text | Selection, treatment, missingness and interpretation limits. |
| `registry_id_source_url` | URL | Primary evidence for the NCT identifier. |
| `trial_n_source_url` | URL or blank | Primary evidence for the main-trial randomized total. |

The total of **6,422** is 2,339 + 1,152 + 1,949 + 445 + 537, counting each main trial once. It is the sum of reported randomized samples, not an independently deduplicated count of people. Neither the 532-person selected extension nor nested groups are added again.

TRIUMPH-3's three treatment-regimen cells and exact arm counts remain blank. Do not infer arm counts from an allocation ratio. The extension's original-assignment subgroup counts were not verified and remain blank.

## Analysis file

**File:** `retatrutide-estimand-analysis.csv`. **Rows:** four main trials with verified 12 mg and placebo estimates under both analysis methods. TRIUMPH-3 is excluded because a paired regimen estimate was not found.

| Column | Format | Meaning |
| --- | --- | --- |
| `trial` | Text | Main study name. |
| `timepoint_weeks` | Integer | Endpoint follow-up. |
| `arm` | Text | Research arm, 12 mg in this file. |
| `efficacy_loss_pct` | Decimal | Positive loss magnitude, E. |
| `efficacy_placebo_loss_pct` | Decimal | Positive placebo loss magnitude under efficacy, EP. |
| `treatment_regimen_loss_pct` | Decimal | Positive loss magnitude under regimen, R. |
| `treatment_regimen_placebo_loss_pct` | Decimal | Positive placebo loss magnitude under regimen, RP. |
| `active_arm_gap_pp` | Decimal | E − R, in percentage points. |
| `efficacy_minus_placebo_pp` | Decimal | E − EP, in percentage points. |
| `treatment_regimen_minus_placebo_pp` | Decimal | R − RP, in percentage points. |
| `placebo_adjusted_gap_pp` | Decimal | (E − EP) − (R − RP), in percentage points. |
| `formula` | Text | Open arithmetic used to generate the derived values. |
| `efficacy_source_url` | URL | Evidence for efficacy active and placebo values. |
| `treatment_regimen_source_url` | URL | Evidence for regimen active and placebo values. |
| `last_checked` | YYYY-MM-DD | Verification date. |
| `notes` | Text | Rounding and interpretation limits. |

For TRIUMPH-1, E = 28.3, EP = 2.2, R = 25.0 and RP = 3.9. Thus the active gap is 3.3 points, and the gap after placebo subtraction is (28.3 − 2.2) − (25.0 − 3.9) = 5.0 points.

These are descriptive calculations from rounded published arm estimates. They are **not** the modeled treatment contrasts. TRIUMPH-1's paper reports a 21.0-point modeled regimen contrast; rounded subtraction gives 21.1. TRIUMPH-2's paper reports 13.8; rounded subtraction gives 13.7. Keep both labels when citing them. The differences cannot be assigned solely to treatment discontinuation.

## Safety file

**File:** `retatrutide-safety.csv`. **Rows:** 25 selected endpoint records, each pairing a 12 mg arm with placebo. Twenty are selected adverse-event reports; five are definition-sensitive discontinuation reports. This is not an exhaustive safety dataset.

| Column | Format | Meaning |
| --- | --- | --- |
| `record_id` | Text; unique | Trial and endpoint identifier. |
| `trial` | Text | Main study name. |
| `timepoint_weeks` | Integer | Trial endpoint follow-up; percentages are not annualized. |
| `endpoint` | Text | Event or discontinuation definition as reported. Preserve this exact definition. |
| `active_arm` | Text | 12 mg research group. |
| `active_n` | Integer or blank | Verified active denominator where appropriate; not asserted for the unresolved TRIUMPH-1 discontinuation row. |
| `placebo_n` | Integer or blank | Verified placebo denominator where appropriate. |
| `active_reported_pct` | Decimal or blank | Printed active percentage, not a rate reconstructed from a count. |
| `placebo_reported_pct` | Decimal or blank | Printed placebo percentage. |
| `active_reported_count` | Integer or blank | Published active event count when used. |
| `placebo_reported_count` | Integer or blank | Published placebo event count when used. |
| `denominator_description` | Text | Population or denominator qualification. |
| `source_url` | URL | Original evidence for the event value and definition. |
| `source_date` | YYYY-MM-DD | Evidence publication date. |
| `last_checked` | YYYY-MM-DD | Verification date. |
| `notes` | Text | Source conflicts, missing denominators and endpoint qualifications. |
| `denominator_source_url` | URL | Separate primary evidence for a denominator, where known. Consult notes before using it. |

TRIUMPH-1's printed discontinuation percentages, 11.2% and 4.6%, are retained. The source's denominator footnote and printed count/percentage pairs do not fully reconcile. Both generic denominator cells stay blank; do not reverse-engineer a rate or silently replace the paper's values with the earlier sponsor release.

TRIUMPH-2's discontinuation endpoint includes **adverse events or death**. Its original abstract supplies counts of 22 and 14. Percentage cells stay blank so this measure is not presented as a pure adverse-event percentage. For TRIUMPH-3 and TRIUMPH-4, exact safety denominators were not verified and stay blank. TRIUMPH-4's sponsor discontinuation measure includes perceived excessive weight loss.

Reported adverse events are not proof of causation. Trials differ in population, follow-up and reporting: do not pool these records or rank drugs by these percentages.

## Other outcomes file

**File:** `retatrutide-other-outcomes.csv`. **Rows:** three selected 12 mg/placebo pairs: TRIUMPH-2 HbA1c, TRANSCEND-T2D-1 HbA1c and TRIUMPH-4 WOMAC knee pain. This file does not enumerate every trial endpoint or nested subgroup.

| Column | Format | Meaning |
| --- | --- | --- |
| `record_id` | Text; unique | Trial, endpoint and time-point identifier. |
| `trial` | Text | Main study name. |
| `population` | Text | Studied group, matching the main results ledger. |
| `timepoint_weeks` | Integer | Endpoint follow-up. |
| `endpoint` | Text | Exact selected outcome; primary/co-primary status retained where specified. |
| `unit` | Text | HbA1c percentage points or WOMAC points on a 0–10 scale. |
| `analysis_method` | Text | Efficacy, treatment-regimen or pain-specific treatment-regimen estimand. |
| `active_arm` | Text | 12 mg research group. |
| `randomized_active_arm_n` | Integer | Original randomized active arm size, not a denominator for a responder percentage or a count of observed endpoint measurements. |
| `randomized_placebo_arm_n` | Integer | Original randomized placebo arm size with the same qualification. |
| `active_change` | Decimal | Signed source estimate in the stated unit, not a relative percentage change. |
| `placebo_change` | Decimal | Signed source estimate in the stated unit. |
| `source_url` | URL | Exact primary source for both estimates and their analysis definition. |
| `source_date` | YYYY-MM-DD | Evidence publication date. |
| `arm_n_source_url` | URL | Evidence for the randomized arm sizes. |
| `last_checked` | YYYY-MM-DD | Actual verification date. |
| `notes` | Text | Endpoint-specific handling of events during follow-up and interpretation limits. |

TRIUMPH-2's HbA1c values are rounded efficacy estimates; TRANSCEND's are the original paper's treatment-regimen estimates. Do not infer that their difference is a drug comparison. TRIUMPH-4's pain-specific regimen analysis assumes no benefit after significant actions associated with insufficient efficacy; it is not identical to its weight regimen analysis. Lower WOMAC scores indicate less pain. Do not subtract these rounded values and present that arithmetic as a paper's modeled treatment contrast.

<a id="source-retrieval"></a>
## Source retrieval

We cite the original publishers and source producers. A publicly hosted copy was used to inspect two original papers when full-text publisher access failed; no copyrighted paper is redistributed in this package.

| Evidence | What was inspected | Canonical source and retrieval detail |
| --- | --- | --- |
| TRIUMPH-1 | NEJM publisher abstract directly; original paper including Table 3 through a PDF copy | [NEJM DOI 10.1056/NEJMoa2604169](https://www.nejm.org/doi/10.1056/NEJMoa2604169); [copy inspected](https://waltersport.com/wp-content/uploads/2026/09/NEJM-Retatrutide-a-Triple-Hormone-Receptor-Agonist-for-Treatment-of-Obesity-Jastreboff-et-al.-2026.pdf). |
| TRIUMPH-2 | Complete original abstract plus Lilly's detailed announcement | [Original abstract, PMID 42810372](https://pubmed.ncbi.nlm.nih.gov/42810372/). No independent full-paper reading is claimed. |
| TRANSCEND-T2D-1 | Original results paper through a PDF copy; sponsor release | [Lancet DOI 10.1016/S0140-6736(26)00967-0](https://www.thelancet.com/journals/lancet/article/PIIS0140-6736%2826%2900967-0/abstract); [copy inspected](https://static.poder360.com.br/2026/06/Lancet-1.pdf). |
| TRIUMPH-3 | Lilly's July 23 numerical topline release | [Original sponsor PDF](https://investor.lilly.com/node/54736/pdf). No full results paper located by the cutoff. |
| TRIUMPH-4 | Original syndicated sponsor release, Lilly trial identity and public ACR abstract | [Trial identity](https://trials.lilly.com/en-US/trial/408355); [ACR abstract](https://acrabstracts.org/abstract/retatrutide-once-weekly-for-the-treatment-of-knee-osteoarthritis-pain-and-obesity-results-from-a-68-week-phase-3-randomized-clinical-trial/). The correct NCT number is NCT05931367; the legacy syndicated announcement displays a different identifier. |

All checks were performed October 14, 2026. Source publication dates belong to the reports; trial endpoint durations belong to the observations. The webpage's numbered source list provides the complete evidence references and current regulatory statement.

## Reuse and versioning

Peptide Care Index's original calculations and chart designs may be reused with credit by name. Underlying reports retain their own rights and terms. Retain the trial, endpoint, time point, analysis method, evidence URL and check date when quoting a record. Archive each released CSV version before changing it so a citation to version 1.0 can still be reconstructed.