# Screening codebook: which adverse-event groups compare a GLP-1 receptor agonist with placebo

Each record is a randomized, parallel-group or factorial trial on ClinicalTrials.gov with posted results, found by searching interventions for a GLP-1 receptor agonist. You see its titles, conditions, phases, masking, arms with their descriptions and interventions, and the groups its adverse-event tables are reported by (`ae_groups`), with no counts. Decide, for each trial, which adverse-event groups form clean comparisons of a GLP-1 receptor agonist against placebo.

## What counts

**A GLP-1 group** is an adverse-event group whose participants were randomized to receive a drug whose action includes agonism of the GLP-1 receptor: exenatide (including extended-release forms and implanted pumps such as ITCA 650), liraglutide, lixisenatide, albiglutide, dulaglutide, semaglutide (injected or oral), efpeglenatide, taspoglutide, beinaglutide, polyethylene glycol loxenatide, tirzepatide, retatrutide, survodutide, cotadutide, mazdutide, efinopegdutide, pemvidutide, orforglipron, danuglipron, maridebart cafraglutide, ecnoglutide, or another molecule the record describes as a GLP-1 receptor agonist (name it in `note`). Any dose, schedule, or route counts, and several doses may be pooled.

**A placebo group** is an adverse-event group whose participants were randomized to placebo for that drug (or no added drug in a double-blind design, e.g. a placebo injection or tablet), with everything else the same as the GLP-1 group: the same background therapy, the same co-interventions, the same period.

**A comparison** is a set of GLP-1 groups and a set of placebo groups that differ only in the GLP-1 receptor agonist. Most trials have one comparison. A factorial trial, or one with several background regimens reported in separate groups, may have several, one per background (for example liraglutide + exercise against placebo + exercise, and liraglutide alone against placebo alone). Each adverse-event group belongs to at most one comparison.

## What doesn't count

Leave a group out of every comparison when:

- **It combines the GLP-1 drug with another active drug** that its comparator doesn't get (cagrilintide with semaglutide against placebo, a fixed-ratio insulin combination against placebo, liraglutide with phentermine against placebo).
- **It gets an active comparator**, even under a double dummy (sitagliptin with placebo-liraglutide is an active-comparator group, not a placebo group).
- **It is open-label "no treatment" or usual care** with no placebo. Only placebo controls count.
- **It isn't a randomized arm over the randomized period**: a run-in, an open-label extension, a group of everyone in the trial, a period after placebo participants switched to the drug, or a group that mixes exposures.
- **The GLP-1 drug is background therapy in every arm** and the trial tests something else.

When no clean comparison remains, the trial is ineligible: set `eligible` to false and give the reason.

## Population

Give the population randomized, from the conditions, titles, and descriptions:

- `type2_diabetes`: type 2 diabetes, with or without obesity or other conditions.
- `type1_diabetes`.
- `overweight_obesity`: overweight or obesity without diabetes (prediabetes counts here).
- `other`: anything else (fatty liver disease without diabetes required, Parkinson's disease, polycystic ovary syndrome, alcohol use disorder, heart failure, healthy volunteers, ...): say what in `population_note`.
- `mixed`: a population that mixes these by design (e.g. people with and without diabetes): say what in `population_note`.

## Output

One JSON object per trial, in the order given, as a JSON array written to the file named in your task:

```json
{
  "nct": "NCT01234567",
  "eligible": true,
  "exclusion_reason": null,
  "comparisons": [
    {"glp1_groups": ["EG000", "EG001"], "placebo_groups": ["EG002"], "molecules": ["semaglutide"], "background": "metformin"}
  ],
  "population": "type2_diabetes",
  "population_note": "",
  "note": ""
}
```

- `exclusion_reason` (when `eligible` is false), one of: `no_glp1_group`, `no_placebo_group`, `glp1_combined_with_other_active`, `active_comparator_only`, `background_differs`, `groups_not_randomized_arms`, `glp1_is_background`, `other` (explain in `note`).
- `molecules`: the GLP-1 receptor agonists in that comparison's GLP-1 groups, lower case, as listed above.
- `background`: a few words on what both sides of the comparison also received, or `"none"`.
- `note`: anything a second reader would need to follow your decision, especially when a group's identity or the background is unclear. Keep it short.

Decide from the record alone. Don't look up results, publications, or other records, and don't open any other files.