Geographic Distribution of Endemic Fungal Infections among Older Persons, United States

To the Editor: We read with interest the article by Baddley et al. (1) and appreciate their efforts to characterize incidence rates of mycoses. We agree that histoplasmosis, blastomycosis, and coccidioidomycosis are differential diagnoses for patients with consistent symptoms but who reside outside mycosis-endemic areas. 
 
However, we believe that the methods of Baddley et al. probably do not determine the true incidence of these mycoses in sparsely populated states such as Arkansas. Their estimates contrast markedly with surveillance data from the Arkansas Department of Health (Table) and with our clinical experience as infectious disease physicians. We characterize Arkansas as a state in which histoplasmosis and blastomycosis incidence is high and coccidioidomycosis incidence is low; however, Baddley et al. indicate that in Arkansas, incidence of blastomycosis is relatively low and incidence of coccidioidomycosis is high. 
 
 
 
Table 
 
Reported cases of fungal diseases in Arkansas, by year* 
 
 
 
To investigate whether this finding might be associated with their small 5% sample of Medicare beneficiaries, we used data from the Arkansas census to determine that in 2008 the population of adults >65 years of age was ≈407,014, and during 1999–2008, there were ≈3,840,896 person-years for persons in this age group. A 5% sample would account for ≈192,045 person-years. Using their rate ranges (7.84–12.3 cases/100,000 person-years for histoplasmosis, 3.97–6.71 for coccidioidomycosis, and 0.39–0.86 for blastomycosis), we calculated the approximate numbers of cases in their sample: 15–23 histoplasmosis cases, 7–12 coccidioidomycosis cases, and only 1 blastomycosis case. Compared with rates from surveillance averaged over the 10 years, the midpoints of the Baddley et al. estimates are ≈6-fold higher for histoplasmosis, ≈60-fold higher for coccidioidomycosis, and ≈0.4-fold lower for blastomycosis. Only their estimate for blastomycosis incidence falls within the 10-year 95% CIs from surveillance data. We believe that the small cell sizes require that the rate estimates of Baddley et al. be interpreted with care, especially with respect to less populous states.

The case we describe is consistent with the cutaneous variant of melioidosis. However, the patient's initial general symptoms (probably attenuated by early treatment with antimicrobial drugs) could have indicated a transitory, disseminated phase of disease such as that experienced by 4 (all adults) of the 58 cases of primary cutaneous melioidosis in the Australian study (9). It is not known whether B. pseudomallei was transmitted to the patient by an airborne route or percutaneously as in most cases (i.e., wounds infected by contaminated water or mud); other transmission modes are anecdotal (1)(2)(3)(4)(5). Moreover, our patient had none of the classic risk factors, although dengue fever as an underlying coinfection has been described (10).
The patient was treated with intravenous ceftazidime and oral cotrimoxazole at the minimum treatment duration recommended for melioidosis (1)(2)(3)(4)(5). Purely cutaneous variants of melioidosis may be treated exclusively by oral cotrimoxazole over 12 weeks (9), but we opted to prescribe initial intravenous treatment because of her general symptoms. We stopped follow-up 11 weeks after the treatment period ended because of persisting illness remission, but lifelong monitoring is recommended for adult patients (1,4) because relapses occur in ≈10% of adult patients despite well-conducted antimicrobial drug treatment (3,4).
In conclusion, melioidosis as a potential emerging infectious disease should be considered in cases of isolated skin lesions as well as in cases of unexplained fever with nonspecifi c symptoms. Furthermore, the disease should be considered not only among travelers returning from known disease-endemic regions but also in those coming from the Caribbean.

Geographic Distribution of Endemic Fungal Infections among Older Persons, United States
To the Editor: We read with interest the article by Baddley et al. (1) and appreciate their efforts to characterize incidence rates of mycoses. We agree that histoplasmosis, blastomycosis, and coccidioidomycosis are differential diagnoses for patients with consistent symptoms but who reside outside mycosis-endemic areas.
However, we believe that the methods of Baddley et al. probably do not determine the true incidence of these mycoses in sparsely populated states such as Arkansas. Their estimates contrast markedly with surveillance data from the Arkansas Department of Health (Table) and with our clinical experience as infectious disease physicians. We characterize Arkansas as a state in which histoplasmosis and blastomycosis incidence is high and coccidioidomycosis incidence is low; however, Baddley et al. indicate that in Arkansas, incidence of blastomycosis is relatively low and incidence of coccidioidomycosis is high.
To investigate whether this fi nding might be associated with their small 5% sample of Medicare benefi ciaries, we used data from the Arkansas census to determine that in 2008 the population of adults  (2). They raise the relevant concern about potential instability of incidence rates from our data because of small cell sizes. We agree that use of administrative data has major limitations. As such, our intent was not to compare infection incidences of individual states; but rather, our intent was to focus on geographic distribution of endemic mycoses and whether infections occurred in nonmycosis-endemic areas.
Specifi cally, for blastomycosis, our study showed incidence in Arkansas to be 0.8 (0.12-5.8) cases per 100,000 person-years, comparable to the rate provided by Haselow et al. of 1.1 case per 100,000 personyears (1). For coccidioidomycosis, our study found the rate to be much higher than that calculated from the Arkansas surveillance data. Potential reasons for this discrepancy might be lack of case capture with surveillance data, because mandatory reporting for coccidioidomycosis is not required in Arkansas, or misclassifi cation of incident cases in the administrative data. Finally, for histoplasmosis, the incidence rate calculated from administrative data was much higher than that reported by Haselow et al. By using administrative data, we identifi ed a large number (15) of cases and doubt that rate instability is present. We agree that surveillance that uses administrative data has inherent limitations, which require that care be taken when interpreting epidemiologic measures, especially when sample sizes are small.