In addition to the initial 17 COD case-patients who were examined by ophthalmologists and reported to the Ministry of Health, a total of 5,084 children 5–15 years of age (corresponding to 83% of this age group in the population) were examined at 40 schools by health workers. During these active searches, of 235 students triaged and referred for evaluation of possible ocular abnormalities, 64 (27%) were categorized by ophthalmologists as having COD and 103 (44%) had sequelae. In addition to the total 81 COD case-patients identified above by November 26, 2005, COD was diagnosed for an additional 18 by January 26, 2006, identified initially by local clinicians or through patient self-referral.

Of the 99 COD case-patients identified, 91 (92%) were interviewed, of whom 70 (77%) were male and the mean age ( ±1 SD) was 11.0 ± 4.4 years. Ocular-related signs and symptoms were conjunctival hyperemia (89%), conjunctival nodule (70%), excessive tearing (63%), conjunctival pruritus (60%), photophobia (57%), ocular pain (42%), and blurred vision (40%); other reported symptoms included headache (37%) and generalized pruritus (16%). The number of COD case-patients peaked during the dry season (July–September) (Figure 1) when schools were closed and contact with the local river was most common. Of those interviewed, 88 (97%) resided in urban areas of Araguatins, and 3 (3%) resided in rural areas.

Ophthalmologists identified unilateral ocular lesions in 73 (80%) of interviewed COD case-patients; in 18 (20%) patients, the lesions were bilateral. Unilateral subconjunctival nodule(s) of sclera (Figure 2), some of which extended to the corneal limbus, were identified in 43 (47%) case-patients and were present bilaterally in 12 (13%). Unilateral corneal opacities (Figure 3) were observed in 32 (35%) case-patients, bilaterally in 18 (20%).

Neighborhoods most commonly affected in the city of Araguatins included Centro, where 61 (67%) of all case-patients resided (prevalence of 8 cases per 100,000 inhabitants), Vila Cidinha (6 per 100,000), and Nova Araguatins (3 per 100,000). Seventy-five (82%) case-patients attended an urban school in Araguatins, and 3 (3%) attended a rural school (Table 1); 13 (14%) of case-patients were not enrolled in a school.

Of the 62 COD case-patients randomly selected to participate in the case–control study, 48 (77%) were male. Case-patients were significantly younger (mean 11.4 ± >3.5 years) than household controls (mean 25 ± 17.8 years; p<0.001, Student t test), but age distribution was similar to that of community controls (13 ± 6.0 years; p = 0.4, Student t test). In univariate analysis, male sex was significantly associated with disease when case-patients were compared with 178 household controls (OR 4.7, 95% CI 2.3–9.8, p<0.001); and with 186 community controls (OR 4.5, 95% CI 2.2–9.4, p<0.001).

Environmental exposures most strongly associated with increased risk for disease, which was significant when compared with both household and community controls (Table 2), were swimming or diving in the Araguaia River and frequenting Cais beach on the bank of the Araguaia River. Fishing in the river was associated with disease but only when case-patients were compared with community controls. Factors not significantly associated with disease (using either control group) were drinking untreated river water, washing clothes in the river, contact with various types of domesticated animals, a history of exposure to ticks, or a history of allergies. Frequency of river contact was also significantly associated with disease (Table 3). According to multivariate analysis, factors most strongly associated with disease were being of male sex, frequenting the Cais beach area, and diving underwater in the Araguaia River (Table 4).

Among 32 case-patients treated with corticosteroid (oral and/or topical prednisone) by ophthalmologists, disease was resolved or cured in 25 (78%); 7 (22%) case-patients had more severe symptoms and were referred to the Sao Geraldo Hospital in Belo Horizonte, Minas Gerais State.

Among those with nodules, 14 had biopsy samples taken under sterile surgical conditions for diagnostic purposes. Microscopic examination of nodules identified microulcerations of corneal epithelium (Figure 4), and a mixed acute inflammatory response mainly consisting of leukocytes, with some eosinophils, and lymphohistocytic and diffuse neutrophilic infiltrates with edema. Twelve case-patients (13%) had a granuloma of the anterior chamber of the eye unilaterally; 1 case-patient had bilateral anterior chamber granulomas. In addition, in 2 biopsy samples, subconjunctival inflammation was present surrounding large, 200–600-micron, thick-walled, spherical foreign bodies (Figure 5) consistent with adiaconidia of Emmonsia sp. fungus, a cause of adiaspiromycosis.

Onchocerciasis, toxoplasmosis, toxocariasis, and microfilaria were discarded as possible etiologies for COD. All 17 samples tested for onchocerciasis were nonreactive for IgG, and no evidence of microfilaria was found in blood smears (n = 17), aqueous humor (n = 1), or biopsy samples of cutaneous nodules (n = 6) examined microscopically. All samples tested for toxoplasmosis (n = 46) were nonreactive for IgM antibodies. Serologic tests for detection of IgG for toxocariasis were reactive for 59 (88%) of 67 COD case-patients, 14 (74%) of 19 household controls, and 53 (64%) of 82 community controls; helminth eggs and larvae, including Toxocara spp., were found in 7 of 10 soil samples.

These results confirm the existence of an outbreak of conjunctivitis and severe ocular disease probably caused by adiaspiromycosis, mainly among school-aged children. Risk factors for COD identified in this investigation included diving underwater and frequenting a specific beach (Cais) on the Araguia River in the Amazon region of Brazil. The precise reasons eye contact with river water increased risk remain unclear. Perhaps exposure to freshwater sponge spicules caused an initial conjunctival irritation, as suggested in previous publications (6–8). However, the microscopic identification of probable adiaconidia of Emmonsia sp. fungus in the scleral biopsy samples from children with severe disease in this outbreak suggests that conjunctival irritation was most likely followed by conjunctival exposure to conidia of this fungus, perhaps in dust caused in part by dry environmental conditions, similar to the exposure of respiratory mucosa described in case reports of pulmonary adiaspiromycosis (1–5,9). Adiaspiromycosis causes an inflammatory and often granulomatous response in tissue because of the presence of nonbudding, thick-walled adiaconidia of Emmonsia sp. fungus. Disease is thought to result from exposure to conidia (through inhalation or mucosal contact with dust); these conidia subsequently cause a marked inflammatory response and enlarge to become adiaconidia ranging in diameter from 300 to 600 microns (1–5,9,10).

Boys were at higher risk than girls most likely because boys had more facial and eye contact with the river water while swimming and diving. To minimize bias, we randomly selected asymptomatic controls among persons 5–25 years of age in the community, but some selection bias may have resulted because boys and adolescents were absent at the time of interview (only 42% of community-based controls were boys). The clinical characteristics of conjunctivitis in this outbreak were unusual for several reasons. First, unlike conjunctivitis caused by common bacterial or viral pathogens, neither purulent conjunctival discharge nor hemorrhage was reported, and family members of case-patient households were not commonly affected. In addition, disease was characterized by unusual, single or multiple, white, opaque scleral nodules, often with hyperemia or local edema, and in some cases with opacification (changes in the normally transparent characteristics of the cornea or superficially on scleral tissue) extending to the limbus, or angular corneal opacities and anterior uveitis with granulomas in the anterior chamber. We believe that the clinical improvement of nearly all patients treated with corticosteroids also argues strongly against a bacterial cause or fungal species other than Emmonsia because conidia of Emmonsia sp. enlarge and cause a localized inflammatory response but do not commonly have the potential to disseminate.

Characteristics of this outbreak are similar to those of an outbreak of anterior uveitis and granuloma previously reported in India, where the etiology was traced to trematodes (10). Although the thick-walled foreign body observed microscopically on slides from 2 case-patients was initially suspected to be trematodes, the round, apparently spherical shape, thick walls, and vacuous central area with lack of organized, internal structures is most consistent with the adiaconidia of the Emmonsia sp. Morphologic appearance differs from that of the fungus Coccidiodes immitis, in which spores contain internal microsporules (11).

The natural history of this disease is unknown. However, we identified COD case-patients in several stages of disease, including patients with sequalae. Moreover, after obtaining school surveys, we identified ≈5% of children with ocular abnormalities; COD was diagnosed in one third of children after an ophthalmologic exam. We educated the population about risks for eye contact with river water; active searches were conducted to identify all ill persons in the population and in neighboring cities, and health officials limited recreational access to the Araguaia River. These findings suggest that the extent of this problem may be more widespread in the Amazon region than is currently recognized.