Placebo Arms in Phase 3 Trials: Therapeutic-Area Prevalence and Design Limits

Among Phase 3 records with described arms, placebo appears in 23.2 percent of 6,413 oncology records. The corresponding shares are 52.8 percent across 2,192 metabolic records, 55.1 percent across 1,123 psychiatry records, and 57.3 percent across 701 respiratory records. Restricting the cohort to four open statuses lowers the oncology share to 17.7 percent and raises the metabolic share to 68.5 percent. A placebo mention does not establish placebo-only control, add-on design, ethical rationale, or comparator quality. The registry supports a prevalence comparison of described arms, not an explanation for the design difference.

1. Cohort and counting rules

The cohort includes records whose phase field contains Phase 3 and whose arm array is not empty. Therapeutic areas use condition-text rules with oncology first, followed by metabolic, psychiatry, and respiratory. A record enters one area after precedence. Placebo presence means that the word appears anywhere in a described arm. The unit is one registered study record, not one participant, treatment comparison, or completed trial.

2. Therapeutic-area assignment is explicit and overlapping

Before precedence, oncology keywords appear on 6,413 records, metabolic keywords on 2,199, psychiatry keywords on 1,141, and respiratory keywords on 713. Oncology overlaps metabolic on seven records and psychiatry on fifteen. Metabolic overlaps psychiatry on four. The precedence rule assigns those overlaps once, which reduces later categories. Counts therefore depend on both the keyword set and its declared order.

3. Placebo-arm prevalence differs by assigned area

Placebo appears in 1,486 oncology records, 1,157 metabolic records, 619 psychiatry records, and 402 respiratory records. The denominators differ substantially, so percentages provide the area comparison. Oncology records show less than half the placebo prevalence of every other measured area. The registry does not identify the causal mechanism behind that difference.

4. Open-status records change the estimates

Restricting the cohort to recruiting, not-yet-recruiting, active-not-recruiting, and enrolling-by-invitation records changes every estimate. Placebo appears in 17.7 percent of 2,200 oncology records, 68.5 percent of 270 metabolic records, 58.3 percent of 139 psychiatry records, and 60.0 percent of 70 respiratory records. The sensitivity result shows that status composition is part of the cross-area comparison.

5. Arm text does not resolve comparator structure

Among oncology records with placebo text, 400 also contain an arm typed as an active comparator, while 39 mention standard care. The corresponding active-comparator counts are 334 in metabolic, 182 in psychiatry, and 166 in respiratory records. These markers overlap and remain incomplete. Absence of either marker does not prove placebo-only control because comparator information can appear under other labels or descriptions.

6. Supported interpretation

The registry supports three conclusions within the classification rules. Placebo text appears less often in oncology Phase 3 arm descriptions than in the other three assigned areas. The difference remains and widens under an open-status sensitivity analysis. Arm text and types do not reliably separate placebo-only, add-on, active-control, or standard-care designs. Ethical and scientific guidance governs comparator selection, but the registry does not encode the rationale used for each study.

7. Limitations

Area rules can misclassify studies or omit synonyms. Phase text can include combined phases. Records without described arms are excluded. Placebo text can appear in an intervention name, arm label, or description with different meanings. Active-comparator and standard-care markers are incomplete. Registered plans can change, and the analysis does not establish conduct, completion, ethics approval reasoning, or treatment effect.