PubMed vs Symmathy: searching the dengue literature

The problem in one sentence

Dengue is endemic to more than 100 countries, with the highest burden in Brazil, Southeast Asia, the Indian subcontinent, and parts of the Pacific. The clinical and epidemiological literature on dengue is written and indexed primarily where dengue is studied: in Portuguese on LILACS and SciELO, in Spanish on LILACS, in Thai on ThaiJO, and in Japanese on J-STAGE. PubMed indexes roughly 5% of the world's medical journals, overwhelmingly North American and Western European. The dengue evidence in MEDLINE is real, but it is a sample — not the population.

What we did

We ran the same dengue query through PubMed alone and through Symmathy's federated search. The Symmathy search runs in parallel across PubMed, Europe PMC, Cochrane, ClinicalTrials.gov, J-STAGE, LILACS, HAL, ThaiJO, DOAJ, OpenAlex, and CrossRef. Non-English titles and abstracts are translated inline; every record links back to the original source.

We did not grade the records. We did not exclude based on quality, study design, or recency. The point of the comparison is volume and provenance, not appraisal — appraisal stays with the reviewer.

What we found

• A large share of the records returned by the federated search are not in MEDLINE. Most are Portuguese- or Spanish-language clinical and epidemiological papers from Brazilian and Latin American journals indexed on LILACS and SciELO.
• Japanese case series and seroprevalence studies from J-STAGE — including imported-case surveillance from travel clinics — are routinely absent from PubMed.
• Thai clinical studies on dengue hemorrhagic fever and warning signs are heavily represented on ThaiJO and underrepresented in MEDLINE.
• French-language studies from Réunion, Martinique, Guadeloupe, French Guiana, and Polynesia — regions with first-hand dengue outbreak data — surface on HAL and are sparse in MEDLINE.

We are not claiming that the missing literature would change the conclusion of any specific systematic review. We are claiming that you cannot know whether it would until you've actually looked.

Why this matters for systematic reviews

Cochrane handbook §4.3 already says you should search beyond MEDLINE and Embase, including regional and non-English databases, to reduce language and geographic bias. In practice that step is the one most often skipped or reduced to a sentence in the methods section, because the manual work is real: different interfaces, different query syntaxes, no shared export format, and abstracts in languages the reviewer doesn't read.

Skipping it has a predictable effect. Reviews of tropical and neglected diseases tend to exclude the populations most affected by the disease being reviewed. Treatment recommendations get written on the slice of evidence that happens to be in English. That isn't a research-ethics edge case for dengue, Chagas, leishmaniasis, schistosomiasis, or Japanese encephalitis — it's the default.

What Symmathy does and doesn't do

Does: federated search across 10+ databases, inline translation of titles and abstracts, a reproducible audit trail of sources queried, query terms, and result counts for the PRISMA flow.

Doesn't: rate or grade individual papers, recommend a "best" paper, score quality, or make clinical decisions. Every record links back to the original source — interpretation, risk-of-bias, and GRADE stay with the reviewer.

Try it on your own question

If you're working on a review where regional evidence matters — dengue, Chagas, leishmaniasis, Japanese encephalitis, schistosomiasis, snakebite, or any disease whose burden sits outside the English-language publishing world — run your existing query through Symmathy and look at what comes back from outside MEDLINE. If you'd like institutional access for your library or review team, email connect@symmathy.org.