The Geographic Distribution of Dimorphic Mycoses in the United States for the Modern Era

Abstract

Background

The dimorphic mycoses (DMs) of the United States—Histoplasma, Coccidioides, and Blastomyces—commonly known as endemic mycoses of North America (in addition to Paracoccidioides) are increasingly being diagnosed outside their historical areas of endemicity. Despite this trend, the maps outlining their geographic distributions have not been updated in more than half a century using a large, nationwide database containing individual-patient–level data.

Methods

This was a retrospective analysis of >45 million Medicare fee-for-service beneficiaries from 1 January 2007 through 31 December 2016. Diagnoses of histoplasmosis, coccidioidomycosis, and blastomycosis were defined by International Classification of Diseases, Ninth/10th Revision, codes. The primary outcome was the incidence of histoplasmosis, coccidioidomycosis, and blastomycosis for each US county. Clinically meaningful thresholds for incidence were defined as 100 cases/100 000 person-years for histoplasmosis and coccidioidomycosis and 50 cases/100 000 person-years for blastomycosis.

Results

There were 79 749 histoplasmosis, 37 726 coccidioidomycosis, and 6109 blastomycosis diagnoses in unique persons from 2007–2016 across 3143 US counties. Considering all US states plus Washington, DC, 94% (48/51) had ≥1 county above the clinically relevant threshold for histoplasmosis, 69% (35/51) for coccidioidomycosis, and 78% (40/51) for blastomycosis.

Conclusions

Persons with histoplasmosis, coccidioidomycosis, and blastomycosis are diagnosed in significant numbers outside their historical geographic distributions established >50 years ago. Clinicians should consider DM diagnoses based on compatible clinical syndromes with less emphasis placed on patients' geographic exposure. Increased clinical suspicion leading to a subsequent increase in DM diagnostic testing would likely result in fewer missed diagnoses, fewer diagnostic delays, and improved patient outcomes.

Graphical Abstract

Graphical Abstract.

This study found clinically significant incidence rates of histoplasmosis in 94% of US states, coccidioidomycosis in 69% of US states, and blastomycosis in 78% of US states. These data suggest expanded geographic distributions of these endemic mycoses.

(See the Editorial Commentary by Schwartz and Bahr on pages 1302–3.)

Members of the genera Histoplasma, Coccidioides, and Blastomyces are dimorphic fungal pathogens collectively known as endemic mycoses of the United States due to their habitation in specific ecological niches within defined geographic distributions [1]. The geographic distribution for each of these pathogens was first described in the 1950s (Figure 1A , Figure 2A , and Figure 3A ) [2]. The historical maps for Histoplasma and Coccidioides were based on antigen skin testing (histoplasmin and coccidioidin) of persons with a history of residence in a single county [2, 3]. Blastomyces antigen testing was unavailable during the mid-20th century; therefore, its historical maps were based on case and outbreak reports [2]. Over the last half century, these dimorphic mycoses (DMs) have been increasingly diagnosed outside the historical geographic distribution of these organisms. The variables affecting the epidemiologic and geographic changes are likely multifactorial and may include climate change and other anthropogenic activities. Efforts are currently underway to identify the causes of these epidemiological changes.

Figure 1.

Geographic distribution of histoplasmosis. A, Historical geographic distribution of histoplasmosis derived from histoplasmin skin antigen testing. Map is adapted from Schwartz and Furcolow, 1955 [2]. B, Histoplasmosis incidence from 2007 through 2016 in Medicare fee-for-service beneficiaries by US county. Reported as histoplasmosis cases/100 000 person-years.

Figure 2.

Geographic distribution of coccidioidomycosis. A, Historical geographic distribution of coccidioidomycosis derived from coccidioidin skin antigen testing. Map is adapted from Schwartz and Furcolow, 1955 [2]. B, Coccidioidomycosis incidence from 2007 through 2016 in Medicare fee-for-service beneficiaries by US county. Reported as coccidioidomycosis cases/100 000 person-years.

Figure 3.

Geographic distribution of blastomycosis. A, Historical geographic distribution of blastomycosis derived from case reports. Map is adapted from Schwartz and Furcolow, 1955 [2]. B, Blastomycosis incidence from 2007 through 2016 in Medicare fee-for-service beneficiaries by US county. Reported as blastomycosis cases/100 000 person-years.

Due to the evidence supporting evolving areas of endemicity for Histoplasma, Coccidioides, and Blastomyces, we will refer to these fungi as “dimorphic mycoses” (DMs) rather than endemic mycoses for the remainder of the article. Dimorphic mycoses, as general terminology, would include other pathogens such as Sporothrix spp.; DMs in the context of this manuscript will specifically refer to Histoplasma, Coccidioides, and Blastomyces. Several studies based on literature reviews, mathematical models, and patient data limited to US state mandatory pathogen reporting have been conducted to update the geographic distribution of the DMs [3–9]. However, since 1969, there have been no large, nationwide studies utilizing patient-level data to update DM geographic distributions. This study aims to update the geographic distributions of histoplasmosis, coccidioidomycosis, and blastomycosis using geographically granular data from a large, nationwide population of Medicare recipients.

METHODS

Medicare fee-for-service (FFS) claims data were obtained from the Centers for Medicare & Medicaid Services Chronic Condition Warehouse. The cohort includes all persons aged 65 years and older from 2007 through 2016; persons less than 65 years were excluded. DM diagnoses were identified using International Classification of Diseases, Ninth Revision (ICD-9) (histoplasmosis 115.x, coccidioidomycosis 114.x, blastomycosis 116.x), and International Classification of Diseases, 10th Revision (ICD-10), diagnostic codes (histoplasmosis B39*, coccidioidomycosis B38*, blastomycosis B40*). To avoid misclassification of cases during diagnostic workup, patients having solely diagnostic (eg, radiography) and laboratory claims were not considered DM cases and excluded from analysis. The county where the patient lives was considered the geographic location of their DM diagnosis. Data from the 50 US states and Washington, DC, were included.

Incidence for each US county was the primary outcome and reported as number of cases per 100 000 person-years (PYs). Incidence was defined for each DM as the number of persons diagnosed divided by the available PYs of Medicare FFS beneficiaries for each year and summated as a 10-year cumulative total. The first nondiagnostic claim coded with a DM diagnosis was counted as the patient's incident diagnosis. To avoid skewing the data resultant to small numbers of DM diagnoses, we considered counties with 5 or fewer histoplasmosis, 3 or fewer coccidioidomycosis, or 1 or fewer blastomycosis diagnoses as zero cases in that county. Investigators defined clinically meaningful thresholds as 100 cases per 100 000 PYs for histoplasmosis and coccidioidomycosis and 50 cases per 100 000 PYs for blastomycosis. The DM thresholds were assigned to be between estimated US rates of Legionnaire's disease from Legionella spp. infection (∼2 cases/100 000 PYs) and pulmonary embolism (PE) (∼112 cases/100 000 PYs) [10, 11]. As with DM infections, Legionnaire's disease and PE have variable clinical presentations, have high mortality if untreated, and require specific laboratory/radiographic testing for diagnosis. The choice to use different clinically relevant incidence thresholds and different adjustment cutoffs for counties with very few cases was made to reflect the underlying epidemiology of each pathogen.

Incidence rates were mapped for each US county using R (version 4.1.2; R Foundation for Statistical Computing, Vienna, Austria). Colorization thresholds for incidence rates for each DM map were optimized for visual discrimination.

RESULTS

Among the Medicare FFS beneficiaries, there were 79 749 histoplasmosis, 37 726 coccidioidomycosis, and 6109 blastomycosis diagnoses from 2007 through 2016. The geographic distribution of each DM by incidence for each county is presented in Figure 1B , Figure 2B , and Figure 3B . Incidence maps for Alaska and Hawaii are presented in Supplementary Figures 1 and 2. A comprehensive table of DM incidence rates for each US county and Washington, DC, is available in Supplementary Table 1.

There were 1971 out of 3143 US counties with more than 5 incident histoplasmosis diagnoses, and 92% (1806/1971) had an incidence of more than 100 cases per 100 000 PYs. In the 839 counties with more than 3 coccidioidomycosis diagnoses, 40% (339/839) had an incidence of more than 100 cases per 100 000 PYs. In the 1602 counties with more than 1 blastomycosis diagnosis, 34% (547/1602) had an incidence of more than 50 cases per 100 000 PYs (Supplementary Table 1). At least 1 county in 94% (48/51), 69% (35/51), and 78% (40/51) of states, including Washington, DC, was above the clinically relevant threshold for histoplasmosis, coccidioidomycosis, and blastomycosis incidence, respectively. These distributions of clinically significant incidence rates extend beyond the historical boundaries of each DM last described in 1969 (Figure 1B , Figure 2B , and Figure 3B ). Uncensored incidence maps and simplified maps of counties exceeding the designated clinical threshold are available as Supplementary Figures 3 and 4.

DISCUSSION

This study represents the largest systematic update of the geographic distribution of DMs in a half century. Most states had at least 1 county meeting the clinically relevant thresholds for each DM, with many outside the historical DM geographic distributions. Our results support expanded geographic distributions proposed by others [4, 5, 12]. There are ongoing efforts to explain why DMs are being diagnosed outside of their historical distributions, but it is time to recognize the widespread geography of DM diagnoses and implement this knowledge into clinical practice.

Increasing clinician awareness of more widespread distributions of DMs is vitally important to prevent delayed or missed diagnoses. Diagnosis is delayed by more than 1 month in 40% of blastomycosis cases and 46% of coccidioidomycosis cases [13, 14]. Patients with histoplasmosis experience an average of a 40-day delay in diagnosis, with over 80% having a healthcare visit, representing a missed opportunity for diagnosis [15]. The most common cause of diagnostic delay is attributed to failure to consider DM etiologies, resulting in a moderate to major impact in clinical care in 66% of cases [16]. A recent survey found that more than 75% of infectious disease physicians in areas that our study found to have clinically significant incident rates reported not seeing patients with histoplasmosis [17]. Continued use of outdated DM distribution maps will result in further missed or delayed diagnoses.

We believe the continued use of “endemic” in the context of Histoplasma, Coccidioides, and Blastomyces contributes to the false heuristic that these DMs are not frequently encountered and diagnosed across the United States. The data presented in our study are specific to the United States, although members of the DM genera infect patients across the globe. Histoplasma is present on every continent except for Antarctica, Coccidioides is present in North and South America, and Blastomyces has been reported in North America, Africa, and Asia [4]. Other fungal infections with environmental reservoirs, including mucormycosis and aspergillosis, have incidence rates that vary across their geographic distributions [18, 19], although the term “endemic” is not regularly applied to these pathogens. Rather, for mucormycosis and aspergillosis, there is focus on host risk factors and compatible clinical presentation when approaching potential diagnoses. We believe a similar diagnostic approach should be applied to DMs. As more evidence confirming the expansion of DM distributions accumulates [20–22], phasing out the “endemic” designation for Histoplasma, Coccidioides, and Blastomyces may aid clinicians and avoid misattributing patients' risk based on outdated “endemic regions” but instead focus on a patient's specific risk factors and clinical syndrome.

Our results should be interpreted with the following limitations: (1) we used administrative claims data from the Medicare population and only included patients 65 years or older and (2) our methodology could not account for travel-related exposure. These limitations likely affect the specific incidence rate for a given county, but the incidence rate relative to other counties should not be affected as all counties were treated uniformly. The ability to compare the results of each county relative to the more than 3000 other counties and utilizing a dataset of more than 45.4 million patients decreases the potential for the limitations in our methodology to disguise general geographic trends in DM diagnoses. To further mitigate these limitations, we chose conservative cutoffs for clinically relevant incidence and considered counties with extremely low numbers of diagnoses as zero diagnoses. Most of the Medicare population is older than 65 years, although younger patients can qualify for Medicare benefits if they have certain diagnoses (eg, end-stage renal disease, amyotrophic lateral sclerosis). We excluded this younger population to avoid overrepresentation of specific Medicare-eligibility diagnoses in our study relative to the general population. Due to this age limitation, our cohort likely overrepresents symptomatic presentation with severity sufficient to seek medical care. This limitation would result in an underestimation of the true incidence that would include younger, more active individuals who may manifest asymptomatic, mild, or self-limited presentations.

For this study, a patient's county of residence was used to calculate DM incidence rather than the location where the patient sought care and received diagnostic testing. This decision was made to minimize the effect of patients being sent to referral centers. However, relative proximity to a referral center with additional resources (eg, mycology experts, increased availability of diagnostic testing) may skew our results. However, this type of bias would suggest that our methods result in an underestimation of true DM incidence.

Travel-related exposure remains an important contributor to DM incidence; however, we believe additional factors (eg, anthropogenic climate change, land development practices) may contribute to the general trend of expanded geographic distributions of these fungi. For example, we observed areas of clinically relevant coccidioidomycosis incidence in the upper Midwest, a pattern also observed in recent Centers for Disease Control and Prevention data [6]. The high incidence of coccidioidomycosis in the upper Midwest may be attributable to winter travel to popular warm weather destinations in the southwestern United States, although our study is unable to account for travel exposure. Alternatively, it is possible that suitable habitats for Coccidioides exist in the northern Midwest, and these diagnosed infections could potentially be acquired locally. Outside its historical area of endemicity, Coccidioides has been found in Washington state, suggesting the potential for suitable habitats at similar northern latitudes [21, 22]. Further, climate change has been shown to increase DM infection rates [23] and climate modeling predicts northern expansion of suitable Coccidioides habitats over the next several decades [24]. Additional research is needed to (1) identify new areas with DM reservoirs by detecting these pathogens in local soil samples and (2) identify the extent that specific climate factors influence environmental reservoirs and the epidemiology of each DM. Targeting potential environmental reservoirs in areas of high incidence identified in our study, climatologists, microbiologists, and other environmental researchers can better explore the interactions between local and climate factors with these fungal pathogens.

CONCLUSIONS

Histoplasmosis, coccidioidomycosis, and blastomycosis diagnoses are often delayed or missed. Since the last systematic update using nationwide patient-level data in 1969, the geographic distributions of DM diagnoses have expanded to regions beyond their historical borders. Acknowledging expanded geographic distributions for DM diagnoses is important to maintain high clinical suspicion of these pathogens. Dimorphic mycoses should be considered when there is concern for a fungal infection anywhere in the United States in patients with a compatible clinical syndrome, particularly in those who are immunocompromised.

Supplementary Data

Supplementary materials are available at Clinical Infectious Diseases online. Consisting of data provided by the authors to benefit the reader, the posted materials are not copyedited and are the sole responsibility of the authors, so questions or comments should be addressed to the corresponding author.

Notes

Financial support. This work was supported by a Washington University Institute of Clinical and Translational Sciences grant from the National Center for Advancing Translational Sciences of the National Institutes of Health (grant number UL1 TR002345), reported by P. B. M. and A. S.; a Metabolic Skeletal Disorders Training grant through the National Institute of Arthritis and Musculoskeletal and Skin Diseases (grant number T32AR060719); and by Mayne Pharma. M. A. O. reports support from the National Institutes of Health/National Center for Advancing Translational Sciences (paid to their institution). A. S. reports support from an Astellas Pharma US ISR grant (grant number MYCA-15L03).