How to search ClinicalTrials.gov for a systematic review

Why the registry search is not optional

A systematic review that only searches published literature has a structural blind spot: trials that finished but were never written up. Published trials are not a random sample of conducted trials — studies with null or unfavorable results are published less often and more slowly. Registry searching is how you detect that gap, count it in your PRISMA flow, and compare registered outcomes against what eventually appeared in print.

ClinicalTrials.gov holds over half a million registrations, including thousands of completed studies with posted results and no corresponding publication. For many interventions, it is the only public record that a trial ever happened.

Classic vs modern interface

The modern interface (clinicaltrials.gov) is the default and supports fielded search, facets, and JSON/CSV export. The legacy "classic" expert-search syntax still works in the modern search box and is worth knowing because it appears in older published review protocols:

• AREA[ConditionSearch] dengue — restrict to the condition field.
• AREA[InterventionSearch] praziquantel — intervention field only.
• AREA[OverallStatus] completed — recruitment status.
• dengue AND vaccine AND AREA[Phase] PHASE3 — combine free text with field restrictions.

In the modern interface you can build the same queries visually with the Advanced Search form, which is less error-prone and produces a shareable URL you can paste straight into your review protocol.

A search strategy that holds up to peer review

1. Search the condition with synonyms and spelling variants, not just the index term. "Dengue" also appears as "breakbone fever" and "dengue hemorrhagic fever."
2. Search the intervention separately with generic names, brand names, and investigational codes (e.g. a drug's development code). Registry entries are less standardized than MEDLINE indexing.
3. Record status filters explicitly. "Completed" and "Terminated" both matter; terminated trials still produce usable data and are disproportionately unpublished.
4. Filter "With results" as a separate pass, not a replacement. You need the full denominator of conducted trials, then the subset with posted results.
5. Export on the day you search and record the date, the exact query URL, and the hit count. PRISMA requires all three, and registry contents change daily as sponsors update records.

Mistakes that quietly shrink your yield

• Assuming "completed" means "results posted." It doesn't. A large share of completed trials never submit results; check the results-posted fields separately.
• Searching only the condition field. Interventions often appear in titles and outcome measures but not in the structured intervention field. Pair fielded terms with an "other terms" free-text pass.
• Treating ClinicalTrials.gov as global coverage. It skews toward US-registered and industry trials. The WHO ICTRP portal aggregates national registries — jRCT (Japan), CTRI (India), ChiCTR (China), Thai Clinical Trials Registry — that contain trials absent from ClinicalTrials.gov.
• Not matching registrations to publications. Cross-reference NCT numbers against your PubMed results; many published papers cite their registration number in the abstract, which is also how you spot outcome switching between registration and publication.

Searching the registry alongside the literature

The tedious part of registry searching isn't the registry — it's reconciling it with everything else. NCT numbers in one tab, PubMed results in another, CENTRAL in a third, deduplicated by hand in a spreadsheet.

Symmathy queries ClinicalTrials.gov in parallel with PubMed, Europe PMC, Cochrane, LILACS, J-STAGE, DOAJ, HAL, OpenAlex, CrossRef, ThaiJO, and KoreaMed from one search box, deduplicates across sources, and shows which databases each record came from — so a registered trial and its eventual publication land next to each other instead of in separate exports. Every record links back to its original registry or database entry, and per-source hit counts are kept for your PRISMA flow diagram.

Symmathy doesn't screen, grade, or recommend studies. That judgment stays with you.