Fungal diagnostics in Latin America and the Caribbean: what 619 institutions say they have

Asked, not audited

Most fungal-disease papers are about an organism. This one, published online on 28 July 2026 in Nature Communications, is about equipment. Working through EFISG, the fungal infection study group of the European Society of Clinical Microbiology and Infectious Diseases, a team with Alessandro C. Pasqualotto as first author and Jon Salmanton-García as senior author sent a questionnaire to hospitals, clinical laboratories and reference centres that diagnose or treat invasive fungal disease across Latin America and the Caribbean. A total of 619 institutions in 23 countries answered. The paper records that the survey ran in English, French, Portuguese and Spanish, and was distributed between April 2023 and May 2025. Every figure below belongs to that window, not to 2026.

What the study is shapes what each figure means. Participation was voluntary, the unit of analysis was the institution rather than the patient, and a respondent ticked what their centre had access to. Nobody walked into these laboratories with a checklist. The authors say as much in their limitations: reported availability may run ahead of real routine access, and there is a potential response bias towards better-resourced centres. Their discussion goes further, observing that many centres hold beta-D-glucan and galactomannan assays and still use them rarely. Institutions were grouped by their country's nominal GDP per capita — 101 from countries below US$5,000, 318 from the US$5,000 to 10,000 band, and 200 from countries above US$10,000.

Classical methods in place, newer tools thin

The basics are there. Microscopy was reported by 475 of the 497 centres that answered that item (95.6%), culture by 428 of 479 (89.4%), and China/India ink — still the classical laboratory route into suspected cryptococcosis — by 408 of 493 (82.8%). Tests allowing identification to species level came in at 353 of 463 (76.2%) and antifungal susceptibility testing at 323 of 477 (67.7%).

Above that line, provision thins out fast. MALDI-TOF mass spectrometry was reported by 107 of 449 (23.8%) and DNA sequencing by 80 of 445 (18.0%). Among antigen tests, Aspergillus galactomannan reached 253 of 404 (62.6%), Cryptococcus antigen 286 of 406 (70.4%), Histoplasma antigen 206 of 403 (51.1%) and Histoplasma serology 188 of 304 (61.8%); the authors flag access to tools for identifying endemic mycoses as a particular gap, and one that persists above all in low-income countries. Molecular testing as a category stood at 166 of 388 (42.8%), most often for Pneumocystis jirovecii (135 of 327; 41.3%) and Aspergillus (118 of 316; 37.3%). Beta-D-glucan, which in wealthier systems is often the earliest signal of invasive candidiasis, was reported by 62 centres out of 391 (15.9%).

The denominators change from row to row, and that is not sloppiness. Many centres skipped individual questions, so the base ranges from a few hundred to just under five hundred. A percentage quoted here without its denominator would be close to meaningless.

The drug cupboard

Fluconazole was reported by 421 of 436 centres (96.6%), and it is the only item the authors describe as nearly universally available. Amphotericin B in any formulation reached 378 of 429 (88.1%), but the split matters clinically: deoxycholate 317 of 418 (75.8%), the liposomal formulation 251 of 419 (59.9%). Echinocandins as a class stood at 295 of 429 (68.8%) — caspofungin 202 of 415 (48.7%), anidulafungin 175 of 413 (42.4%), micafungin 91 of 413 (22.0%). Among the other azoles, itraconazole 284 of 425 (66.8%), voriconazole 274 of 422 (64.9%), isavuconazole 159 of 409 (38.9%), posaconazole 119 of 408 (29.2%). Terbinafine 129 of 401 (32.2%).

The sharpest number is flucytosine: 88 centres of 409 (21.5%) for a drug the discussion calls essential in cryptococcal meningitis. Therapeutic drug monitoring was reported by 121 of 363 (33.3%), mostly for voriconazole (94 of 338; 27.8%) and itraconazole (74 of 325; 22.8%), less often for posaconazole (36 of 318; 11.3%). Surgical procedures for diagnostic or therapeutic purposes were reported by 228 of 313 (72.8%). Access to liposomal amphotericin B, echinocandins, voriconazole and posaconazole was significantly higher in countries above US$10,000 GDP per capita, in transplant centres, and in institutions caring for people living with HIV — an association, not a demonstration that money is the only cause.

A paper that disagrees with itself, and why that is worth saying

Every figure above is taken from Tables 2 and 3 of the open-access full text, and that choice needs stating out loud. The same paper's abstract gives several items differently: galactomannan 41% rather than 62.6%, beta-D-glucan 29% rather than 15.9%, molecular testing 23% rather than 42.8%, liposomal amphotericin B 38%, echinocandins 51%, voriconazole 57%, posaconazole 33%, drug monitoring 36%. The running Results text adds a second set — monitoring at 35.7% against the table's 33.3%, posaconazole monitoring at 12.4% against 11.3% — and the limitations paragraph supplies a third value again for molecular diagnostics, 42.6%. None of it can be reconciled from outside the dataset, so this article uses one set throughout, the tables, because only the tables carry denominators, and names the others rather than quietly picking a favourite. Anyone who cites 29% for beta-D-glucan has read the abstract, not invented a number.

One more detail that travels badly. Table 2 heads the 286-of-406 row "Cryptococcus spp. GM", and the same table's key expands GM as galactomannan. Galactomannan is an Aspergillus cell-wall antigen and not a cryptococcal one, and the sub-rows beneath that heading are latex agglutination and lateral flow assays, which are the standard ways of detecting cryptococcal antigen. The Results text and the discussion both call the item Cryptococcus antigen, and that is the only correct way to quote it.

What the survey cannot support

It counts no patients and no deaths. It does not follow outcomes, and it cannot say how many people with cryptococcal meningitis went without flucytosine. It records what institutions said they could reach for. Nor is it a representative sample of the region: participation was voluntary, the replies fell very unevenly — the largest single contributor was Colombia, which the Results text puts at 205 centres — and the authors list uneven geographical representation, naming Brazil, Colombia, Honduras and Peru, alongside limited participation from the smaller Caribbean island states. For the same reason no country should be credited with or denied a given test on the strength of these data; the survey pooled the answers of centres, it did not audit national supply.

The map at the top of this entry is the paper's own Figure 1 and is a data figure, not a photograph of anyone's bench. Country colour there is nominal GDP per capita, a white dot is a location, and several centres sharing one location are drawn as a single dot, so the dots are fewer than the institutions. Grey means a country that sent no respondents, not a country without laboratories.

Why it belongs beside the big numbers

The catalogue already carries the estimate of 3.8 million deaths a year linked to serious fungal disease and the WHO priority list of fungal pathogens. Those describe scale and ranking. This survey describes the shelf someone has to reach for before any of it becomes a diagnosis, and the distance between a dangerous pathogen and a recognised one lives in that gap. The figures are all checkable in an afternoon: the paper is open, Tables 2 and 3 print n and denominator on every row, and the disagreement with the abstract is visible within two minutes. That habit — read the table, not the summary — is worth keeping for any paper, including the ones about mushrooms that grow in woods rather than on plates.