Symptoms in blastomycosis, coccidioidomycosis, and histoplasmosis versus other respiratory illnesses in commercially insured adult outpatients, United States, 2016–2017

Abstract

Background:

Blastomycosis, coccidioidomycosis, and histoplasmosis cause various symptoms and syndromes, which may present similarly to other infections such as bacterial or viral community-acquired pneumonia, influenza, and tuberculosis.

Methods:

We used the IBM^® MarketScan^® Research Databases to identify adult outpatients with International Classification of Diseases, Tenth Revision, Clinical Modification (ICD-10-CM) diagnosis codes during 2016–2017 for blastomycosis; coccidioidomycosis; histoplasmosis; pneumonia (viral, bacterial, Streptococcus pneumoniae, and unspecified pneumonia); influenza; tuberculosis; and other lower and upper respiratory infections. We compared symptoms on and in the 90 days before diagnosis between patients with these diagnosis codes.

Results:

Fever was less common in blastomycosis (2.6%), histoplasmosis (5.3%) and coccidioidomycosis (9.4%) than in influenza (18.5%) or pneumonia (12.6%–16.3%) patients. Fungal diseases resembled bacterial, viral, and unspecified pneumonias for many pulmonary symptoms. However, cough was more common with coccidioidomycosis (31.4%) and less common with histoplasmosis (14.0%) and blastomycosis (13.1%) vs. influenza (20.2%). Although less frequent, solitary pulmonary nodule (5.2%–14.4%), enlarged lymph nodes (3.7%–9.0%), hyperhidrosis (<2%), and erythema nodosum (<2%) were particularly suggestive of fungal diseases.

Conclusions:

Despite limitations inherent in administrative coding, this analysis of symptom codes across disease types suggests that fungal diseases may be difficult to clinically distinguish from other causes of pneumonia except when certain uncommon symptoms are present. Healthcare providers caring for patients with pneumonia, especially if non-responsive to conventional treatment, should consider fungal diseases as possible etiologies.

40-word summary:

Health insurance claims data suggest that fungal diseases may be difficult to clinically distinguish from other causes of pneumonia except when certain uncommon symptoms are present. Healthcare providers caring for patients with pneumonia should consider fungal diseases as possible etiologies.

Introduction

Blastomycosis, coccidioidomycosis, and histoplasmosis cause wide spectra of illnesses ranging from asymptomatic to life-threatening disseminated infections. Severity and clinical presentation depend on exposure intensity and host immune status. These fungal infections can be difficult to identify because patients present with nonspecific symptoms and syndromes, frequently resulting in delayed diagnosis and misdiagnoses such as non-fungal community-acquired pneumonia (CAP), influenza, tuberculosis, and lung cancer.^1–4

Empiric treatment for bacterial CAP is common before a diagnosis of blastomycosis, coccidioidomycosis, or histoplasmosis.^5–8 Previous studies have found little evidence that specific symptoms or clinical findings can clearly distinguish fungal CAP from non-fungal CAP.^5,9 “Influenza-like” is a common description for blastomycosis, coccidioidomycosis, and histoplasmosis because they can cause cough and fever.^10–12 However, this term can have different meanings for the general public, clinicians, and public health professionals. Such nuances have implications for how these communities communicate about these under-recognized diseases. Accurate, focused messaging is critically important for raising awareness about these diseases and identifying them early in a patient’s illness course, which can reduce delays in diagnosis and appropriate treatment.¹³

Given limited information directly comparing symptoms and clinical findings associated with blastomycosis, coccidioidomycosis, and histoplasmosis to those of other diseases, we analyzed health insurance claims data to inform whether certain features could help increase clinicians’ suspicion for these fungal diseases.

Methods

We used data from the 2016–2017 IBM^® MarketScan^® Research Databases. The MarketScan Commercial Database and the Medicare Supplemental database contain health services insurance claims data including outpatient visits, outpatient prescriptions, and hospitalizations for >37 million employees, dependents, and retirees across the United States. These data are generally representative of the US insured population but are not necessarily representative of all US residents. The datasets are fully de-identified, so the analysis was not subject to review by the Centers for Disease Control and Prevention institutional review board.

To identify patients with diseases of interest and apply enrollment and exclusion criteria, we accessed data through Treatment Pathways, a web-based platform with data from people whose health insurance plans contribute prescription drug data to MarketScan (^~95% of enrollees). We identified 11 independent cohorts of adult (age ≥18 years) patients based on International Classification of Diseases, Tenth Revision, Clinical Modification (ICD-10-CM) diagnosis codes at an outpatient visit for blastomycosis, coccidioidomycosis, histoplasmosis, and eight other respiratory conditions: unspecified viral pneumonia, Streptococcus pneumoniae pneumonia, any bacterial pneumonia, pneumonia caused by an unspecified organism, influenza, respiratory tuberculosis, bronchitis or unspecified lower acute respiratory infection, and acute nasopharyngitis or acute upper respiratory infections (URIs) of multiple and unspecified sites (Supplemental Table). The index date was the first date the diagnosis code was used during April 1, 2016–December 1, 2017. We required continuous enrollment in the 90 days before and the 30 days after the index date. We excluded patients with a diagnosis code for the condition of interest in the 90 days before the index date to attempt to identify incident cases. Because symptoms are likely coded differently for hospitalized patients than for outpatients, we also excluded patients hospitalized in the 90 days before the index date. Patients could have been included if they were hospitalized on the index date in addition to having an outpatient visit, but symptoms coded on inpatient claims were not included. We excluded patients from the histoplasmosis cohort if they received a histoplasmosis diagnosis code (ICD-10-CM B39) at an eye care provider and excluded patients from the tuberculosis cohort if they had a diagnosis code for inactive tuberculosis (ICD-10-CM R7611, R7612).¹⁴ We did not stratify blastomycosis, coccidioidomycosis, or histoplasmosis diagnoses by subcode because most (~75%) were unspecified forms.

For each cohort, we analyzed symptoms and clinical findings documented in ICD-10-CM codes on and in the 90 days before the index date (Supplemental Table). We refer to these collectively as symptoms for simplicity. To better analyze patterns, we grouped symptoms into categories: systemic, pulmonary, upper respiratory, and “other.” We explored frequencies of chronic obstructive pulmonary disease (COPD) or other chronic lower respiratory diseases, asthma, diabetes, and immunocompromised status (HIV/AIDS, immune-mediated inflammatory disease, transplant, or malignancy). We also examined pre-index date outpatient visit frequency for each of the 11 disease cohorts and pre-index date diagnoses with other respiratory infections among the blastomycosis, coccidioidomycosis, and histoplasmosis cohorts.

We used Poisson regression, adjusting for age and sex, to compare proportions of patients with symptom codes among those with fungal diseases and 1) those with influenza, given the “influenza-like” descriptor, and 2) patients with pneumonia caused by S. pneumoniae, as it is the leading cause of bacterial pneumonia. We chose specific viral and bacterial etiologies for these comparisons rather than unspecified pneumonia or other infections because we were evaluating specified fungal etiologies.

Results

Demographic features and underlying conditions

Disease cohort size ranged from 268 patients for blastomycosis to 1,934,543 for acute URI (Table 1). Median age ranged from 42.8 years for influenza to 60.7 years for S. pneumoniae pneumonia. Most (54%–63%) patients were female except for the blastomycosis (24%) and coccidioidomycosis (48%) cohorts. Census Bureau region was most commonly the West for coccidioidomycosis patients (81%) and the Midwest for histoplasmosis (43%) and blastomycosis (39%) patients. Immunocompromised status was more frequent in histoplasmosis (14%) than any other cohort besides respiratory tuberculosis (21%). The blastomycosis cohort had the highest proportion of patients with diabetes (23%).

Table 1:

Demographic features and underlying conditions on or in the 90 days before index date among patients with blastomycosis, coccidioidomycosis, histoplasmosis, or other respiratory infections, 2016–2017

	Coccidioidomycosis	Histoplasmosis	Blastomycosis	S. pneumoniae pneumonia	Bacterial pneumonia	Pneumonia, organism unspecified	Viral pneumonia	Influenza	Bronchitis or unspecified lower acute respiratory infection	Acute nasopharyngitis or upper acute respiratory infections of multiple and unspecified sites	Respiratory tuberculosis
	n=1,929	n=2,205	n=268	n=3,272	n=42,602	n=302,675	n=5,233	n=307,648	n=1,120,957	n=1,934,543	n=3,659
Demographic features
Median age, years	50.9	55.8	56.4	60.7	58.3	56.3	58.7	42.8	48.3	43.8	49.3
Age group, years
18–34	14.9%	11.9%	12.3%	8.4%	13.2%	14.0%	12.4%	32.5%	21.4%	31.5%	23.1%
35–44	16.8%	12.2%	8.6%	10.4%	10.9%	12.8%	11.3%	21.4%	18.0%	19.6%	16.6%
45–54	24.7%	18.1%	20.2%	15.2%	15.6%	17.6%	16.1%	21.9%	23.1%	21.1%	21.0%
55–64	30.0%	29.0%	28.0%	27.9%	24.9%	25.9%	24.6%	18.8%	26.0%	20.4%	23.0%
65 and older	13.6%	28.8%	31.0%	38.1%	35.3%	29.7%	35.6%	5.4%	11.6%	7.4%	16.4%
Male	52.1%	43.9%	75.8%	45.3%	46.3%	45.5%	43.8%	41.9%	39.0%	37.4%	40.2%
Census region*
Northeast	4.3%	7.2%	16.8%	21.9%	21.0%	19.2%	23.4%	14.4%	15.8%	17.6%	23.7%
Midwest	6.0%	43.0%	39.2%	29.7%	25.1%	25.9%	24.4%	13.8%	20.9%	17.0%	16.1%
South	8.8%	42.6%	37.3%	37.6%	42.7%	41.1%	39.0%	63.0%	52.6%	52.8%	38.4%
West	80.6%	7.0%	6.7%	10.9%	11.0%	13.7%	12.9%	8.5%	10.5%	12.3%	21.7%
Unknown	0.3%	0.1%	0.0%	0.1%	0.2%	0.2%	0.2%	0.3%	0.2%	0.3%	0.2%
Underlying conditions
COPD or other chronic lower respiratory disease	10.2%	12.2%	10.1%	23.3%	21.1%	16.2%	20.2%	3.5%	6.8%	3.3%	11.0%
Asthma	10.0%	6.9%	4.9%	10.1%	10.1%	9.0%	11.9%	4.3%	8.1%	4.9%	7.0%
Diabetes	13.8%	15.8%	23.1%	18.8%	18.6%	16.1%	18.9%	6.7%	9.7%	7.2%	11.8%
Immunocompromised**	12.2%	14.0%	11.6%	9.6%	7.8%	6.3%	11.6%	2.8%	3.4%	2.9%	20.5%

^*primary beneficiary’s residence

^**HIV, hematologic malignancy, immune-mediated inflammatory disease, or solid organ or stem cell transplant

Systemic symptoms

Systemic symptoms other than fever were coded for <10% of patients for all conditions (Figure 1). Fever was most common for bacterial, viral, and unspecified pneumonia (12.6–16.3%) and influenza (18.5%), whereas malaise/fatigue was most common for fungal (15.3% coccidioidomycosis, 7.9% histoplasmosis, 6.0% blastomycosis) and other diseases, ranging from 4.4% for URI to 9.7% for tuberculosis. On statistical testing, fever was significantly less common for fungal diseases than for influenza and S. pneumoniae pneumonia, and coccidioidomycosis patients were more likely to have malaise/fatigue compared with influenza (adjusted relative risk [aRR] 2.9, CI 2.6–3.3) and S. pneumoniae pneumonia (aRR 2.4, CI 2.0–2.9) (Figure 2). Generalized hyperhidrosis (ICD terminology that includes night sweats) was significantly more common in coccidioidomycosis and histoplasmosis (aRR 12.1, CI 8.5–17.1 for coccidioidomycosis vs. influenza).

Figure 1:

Symptoms and clinical findings on or in the 90 days before index date among patients with blastomycosis, coccidioidomycosis, histoplasmosis, or other respiratory infections, 2016–2017

Figure 2:

Comparison of symptoms and clinical findings on or in the 90 days before index date between patients with blastomycosis, coccidioidomycosis, histoplasmosis and those with influenza or S. pneumoniae pneumonia, 2016–2017

*Erythema nodosum not shown for coccidioidomycosis vs. S. pneumoniae pneumonia (aRR 4.03 x 10¹²). No patients with blastomycosis or S. pneumonia had erythema nodosum.

Pulmonary symptoms

For fungal disease patients, overall patterns of many pulmonary symptoms resembled those of various pneumonias more than those of influenza or other acute respiratory infections (Figure 1). Cough was the most common symptom for coccidioidomycosis patients (31.4%), similar to S. pneumoniae pneumonia (33.2%). “Nonspecific abnormal finding of lung field” was the most common symptom among blastomycosis (15%) and histoplasmosis (18%) patients and was the second most common symptom among coccidioidomycosis patients (21%), similar to S. pneumoniae pneumonia (15.9%). Compared with influenza on statistical testing, cough was significantly more common in coccidioidomycosis (aRR 1.4, CI 1.3–1.5) but less common in histoplasmosis (aRR 0.6, CI 0.5–0.7) and blastomycosis (aRR 0.6, CI 0.4–0.8) (Figure 2). Solitary pulmonary nodule was more common with fungal diseases, with aRR ranging from 2.4 (CI 1.3–4.2) for blastomycosis vs. S. pneumoniae pneumonia to 36.4 (CI 31.5–41.9) for coccidioidomycosis vs. influenza. “Diseases of mediastinum” diagnoses were infrequent (maximum 1.3%, in histoplasmosis patients) but were also significantly more common in fungal disease patients. Dyspnea, wheezing, hypoxemia, and pleural effusion were all significantly less common with fungal diseases than with S. pneumoniae pneumonia.

Upper respiratory symptoms

Pharyngitis was infrequent among fungal disease patients (<4%) compared with influenza (12.6%), bronchitis (8.9%), or acute URIs (13.3%). Nasal congestion was also infrequent in fungal diseases (<2.1%).

Other symptoms

Erythema nodosum was more frequent in coccidioidomycosis (1.5%) than in any other disease. Enlarged lymph nodes were more frequent in all 3 fungal diseases (3.7%–9.0%) compared with other diseases, including tuberculosis (2.7%). Compared with influenza, erythema nodosum (aRR 320.1, CI 176.9–579.0) was significantly more common in coccidioidomycosis, and enlarged lymph nodes (aRR 22.2, CI 19.0–26.0) were more common in histoplasmosis. “Pain in joint” was significantly more common in coccidioidomycosis than in S. pneumoniae pneumonia (aRR 1.5, CI 1.2–1.8) and influenza (aRR 1.8, CI 1.6–2.0).

Pre-index date visits and post-index date diagnoses

Blastomycosis, coccidioidomycosis, and histoplasmosis patents each had a median of 4 (interquartile range 2–8) outpatient visits in the 90 days before the index date, similar to viral pneumonia, bacterial pneumonia, and S. pneumoniae pneumonia (Table 2). In contrast, patients with influenza, bronchitis or unspecified lower acute respiratory infection, and acute nasopharyngitis or acute URIs had a median of 1 pre-index date visit. “Pneumonia, organism unspecified” was a common pre-index date diagnosis among blastomycosis (10.0%), coccidioidomycosis (16.7%), and histoplasmosis patients (4.9%) (Table 3). In the 30 days post-index date, the proportion diagnosed with any of these 3 fungal diseases was 0.4% in the tuberculosis cohort, 0.2% in each of the viral pneumonia, bacterial pneumonia, and unspecified pneumonia cohorts, and <0.2% in other cohorts.

Table 2:

Number of outpatient visits in the 90 days before index date among patients with blastomycosis, coccidioidomycosis, histoplasmosis or other respiratory infections, 2016–2017

	Mean	Median	Interquartile range	Range
Coccidioidomycosis	5.9	4	2-8	0-89
Blastomycosis	6.0	4	2-8	0-37
Histoplasmosis	5.8	4	2-8	0-89
Streptococcus pneumoniae pneumonia	6.5	4	2-8	0-89
Bacterial pneumonia	6.8	4	2-9	0-89
Pneumonia, organism unspecified	5.4	3	1-7	0-89
Viral pneumonia	7.4	4	2-9	0-89
Influenza	2.7	1	0-4	0-89
Bronchitis or unspecified lower acute respiratory infection	2.9	1	0-4	0-89
Acute nasopharyngitis or acute upper respiratory infections of multiple and unspecified sites	2.9	1	0-4	0-89
Respiratory tuberculosis	5.1	3	1-7	0-89

Table 3:

Other respiratory infection diagnoses in the 90 days before index date among blastomycosis, coccidioidomycosis, and histoplasmosis patients, 2016–2017

	Coccidioidomycosis N=1,929	Blastomycosis N=268	Histoplasmosis N=2,205
Streptococcus pneumoniae pneumonia	0.06%	0.00%	0.00%
Bacterial pneumonia	2.44%	0.42%	0.41%
Pneumonia, organism unspecified	16.72%	10.04%	4.87%
Viral pneumonia	0.41%	0.00%	0.21%
Influenza	0.93%	0.00%	0.88%
Bronchitis or unspecified lower acute respiratory infection	7.25%	3.77%	4.97%
Acute nasopharyngitis or acute upper respiratory infections of multiple and unspecified sites	6.85%	2.93%	3.89%
Respiratory tuberculosis	0.17%	0.42%	0.26%

Discussion

In this study of US health insurance claims data, many signs or symptoms associated with blastomycosis, coccidioidomycosis, and histoplasmosis more closely resembled those of various pneumonias than influenza. Solitary pulmonary nodule and certain less common symptoms such as hyperhidrosis, erythema nodosum, and enlarged lymph nodes were especially suggestive of fungal diseases. Blastomycosis, coccidioidomycosis, and histoplasmosis also appeared to be under detected based on prior diagnoses with unspecified pneumonia, indicating a need for improved healthcare provider awareness and better diagnostic options.

Systemic and pulmonary symptoms were generally similar between patients with fungal diseases and other pneumonia, with a few notable exceptions. Namely, fever and cough were less common (except cough in coccidioidomycosis). These differences may reflect these fungal diseases’ wide illness spectra, which include asymptomatic nodules and chronic infections; symptom prevalence would likely be higher if data permitted limiting analysis to acute pulmonary infections. Our results support the notion that clinical signs and symptoms alone are typically insufficient to distinguish blastomycosis, coccidioidomycosis, and histoplasmosis from other pneumonias. Therefore, it is important for healthcare providers to consider these diseases in patients with non-specific systemic and pulmonary symptoms and be aware that targeted fungal laboratory testing may be warranted, particularly when no other etiology is found.

In messaging to patients and the public, clinicians and public health professionals sometimes refer to blastomycosis, coccidioidomycosis, and histoplasmosis as “influenza-like illnesses” (ILI). However, based on our findings, this term could be misleading. In general, symptoms most predictive of uncomplicated influenza in adult outpatients are abrupt-onset fever and cough,¹⁵ whereas in this analysis, fever was substantially less common among fungal disease patients, and cough was less common for histoplasmosis and blastomycosis. The Centers for Disease Control and Prevention’s ILI criteria are fever and cough and/or sore throat,¹⁶ and although this definition is for public health surveillance rather than clinical purposes, it further supports the notion that blastomycosis, coccidioidomycosis, and histoplasmosis are not always similar to influenza, as sore throat (“acute pharyngitis, unspecified”) was several-fold less common.

Presence of pulmonary nodules and hyperhidrosis, erythema nodosum, and enlarged lymph nodes were more frequently associated with blastomycosis (except erythema nodosum), coccidioidomycosis, and histoplasmosis than with other respiratory infections. Though most of these symptoms were uncommonly coded for, they might help increase healthcare providers’ suspicion for these fungal diseases. All of these symptoms have been classically associated with tuberculosis,^17,18 including hyperhidrosis out of proportion to fever; notably, frequencies among fungal disease patients were even higher than in tuberculosis patients. Similar to findings that erythema nodosum and fatigue were substantially more common in coccidioidomycosis patients, two small studies found that rash, higher fatigue scores, and myalgia were the only clinical features that differentiated coccidioidal CAP from CAP caused by other pathogens.^5,9 The number of blastomycosis patients with cutaneous symptoms in our analysis was small, but a prior study showed that presence of skin involvement increased the percent who received a correct initial blastomycosis diagnosis from 18% to 64%.⁸ Although fungal reviews frequently note the presence of rash, erythematous conditions other than erythema nodosum were not more common in fungal diseases than in other respiratory infections. Arthralgia is a known complication of coccidioidomycosis (termed “desert rheumatism”),¹ and notably, “pain in joint” was more frequent in this disease than in others, including blastomycosis and histoplasmosis. The relatively high proportions of joint and low back pain likely reflect how common they are in primary care; these and other symptoms could be unrelated to fungal disease diagnoses. Studies to identify symptoms predictive of histoplasmosis have been performed for HIV-infected patients in Latin America, but not in the United States.^19,20 Our results support existing literature that enlarged lymph nodes might be particularly indicative of histoplasmosis.²¹

Our findings suggest that blastomycosis, coccidioidomycosis, and histoplasmosis are likely under-recognized causes of lower respiratory infections. Unspecified pneumonia was a common pre-index date diagnosis, consistent with previous studies describing diagnostic delays, often involving inappropriate antibacterial prescriptions for suspected bacterial pneumonia.^6,7,13 Together, the three fungal diseases in this study outnumbered S. pneumoniae pneumonia, classically considered to be the leading cause of bacterial pneumonia. The contribution of fungal diseases to the overall CAP burden in the United States is poorly understood. In the largest US study to examine causes of inpatient CAP, fungal testing was not performed; notably, no pathogen was identified in nearly two-thirds of cases, raising the possibility that some cases could have been caused by fungi.²² Specific causes of CAP are probably under-represented by ICD-10-CM codes because of delays between laboratory testing and receipt of results or because of lack of laboratory testing altogether. The latest American Thoracic Society (ATS) and Infectious Diseases Society of America (IDSA) guidelines do not recommend obtaining respiratory cultures for adults with CAP in the outpatient setting.²³ Targeted fungal testing is also uncommon, and incidental diagnoses of fungal diseases (e.g., on biopsy), particularly of histoplasmosis, occur regularly.⁷ Although ATS has guidelines for diagnosis of fungal infections,²⁴ ATS/IDSA guidelines for CAP do not include testing for endemic mycoses, stating that they are less common causative pathogens.²³ Further research is needed to better quantify the burden of fungal diseases as a cause of CAP given overlapping symptoms and the possibility that they are more common than currently appreciated, particularly in certain regions.

Administrative data are not ideally suited to provide detailed clinical information, but they offer unique opportunities to study large sample sizes, which can be difficult to achieve with other sources of fungal disease data. The primary limitations of this analysis involve inherent under-coding and misclassification. In general, symptom data are likely vastly under-represented on insurance claims. Precedence for similar descriptions of pre-diagnosis symptoms using claims data include neuroendocrine tumors and ovarian cancer,^25,26 and a recent study used ICD codes for symptoms to estimate diagnostic delays in coccidioidomycosis.²⁷ Official guidelines for ICD-10-CM coding and reporting from the Centers for Medicare and Medicaid Services (CMS) and National Center for Health Statistics (NCHS) state that codes describing signs and symptoms are adequate for reporting if a specific diagnosis has not been established, but that symptoms regularly associated with a known disease process generally should not be assigned in addition to the disease diagnosis code(s).²⁸ For this reason, influenza coding, which most frequently occurred on the first visit, may have fewer associated symptom codes than the other diseases we examined. The extent to which healthcare providers and coders follow this guidance, however, is unclear, particularly for pneumonia diagnoses in which the specific pathogen is unknown; “pneumonia, organism unspecified” appears to represent a final diagnosis for many patients. A further ICD-10-CM coding complication is that some signs and symptoms can be considered definitive diagnoses, for example, “acute pharyngitis, unspecified” (J02.9).²⁹ However, pre-index date visit frequencies suggest that we identified patients with fungal diseases, bacterial pneumonia, and viral pneumonia at similar points in their care-seeking and diagnosis processes, so symptom coding may be comparable.

Disease severity and manifestation (e.g., pulmonary vs. disseminated) could also affect the presence and type of symptom codes. Because claims data provide limited information about severity, we focused our study on outpatients to attempt to minimize effects from hospital-acquired pneumonia, which often represents different pathogenic processes and patient populations than CAP. This could affect the overall numbers of patients with codes for specific diseases because cultures are more frequently ordered for inpatients, but we do not expect that it would alter symptom distributions on outpatient claims across diseases.

In addition to limitations related to symptom coding, potential disease misclassification is also a concern. Pneumonia has been studied extensively using administrative data, and ICD codes appear to be moderately accurate for identifying pneumonia patients, though the sensitivity varies widely depending on the code(s) used, setting, and patient population.³⁰ For influenza, the small number of pre-index date visits in this analysis suggests more acute illness, point-of-care testing, or empiric coding compared with fungal diseases and pneumonia. Although inpatient coding for influenza appears to have moderate to high sensitivity and moderate positive predictive value,^31,32 few available data describe influenza coding accuracy among outpatients. Similarly, few data exist regarding ICD code accuracy for blastomycosis, coccidioidomycosis, and histoplasmosis, though under-coding likely exists. Given these limitations, we elected to perform descriptive analyses rather than complex modeling to examine symptom profiles. Future work could also better assess certain underlying conditions; in particular, the high diabetes prevalence among blastomycosis patients is a notable association for additional study.

This study of symptoms among blastomycosis, coccidioidomycosis, and histoplasmosis outpatients provides evidence that these diseases can appear clinically similar to other pneumonia etiologies, but their symptom profile is typically inconsistent with influenza or other acute respiratory infections. Our results underscore the importance of considering these fungal diseases in patients with pulmonary or systemic symptoms and obtaining fungal-specific testing, especially if illness does not resolve with empiric CAP-directed antibacterials. Public and healthcare provider awareness is critical to reducing diagnostic delays, which can be lengthy and associated with unnecessary diagnostic tests, inappropriate treatment, patient anxiety, and excess costs.^6,7,27 Although these fungal diseases have regional patterns and classic exposure scenarios, healthcare providers should be aware that they can occur outside their traditionally-recognized geographic ranges and do not always involve a clear environmental source.³³ That these diseases are of national rather than regional significance is underscored by the finding that they occurred in each Census region. Future studies to better characterize their epidemiology and various clinical presentations will help further guide awareness efforts and facilitate earlier diagnosis.