During January 1–December 31, 2016, routine surveillance conducted in 14 states (Louisiana, Michigan, Minnesota, Missouri, Montana, Nevada, New Mexico, North Dakota, Ohio, Oregon, Pennsylvania, Utah, Wisconsin, and Wyoming) identified coccidioidomycosis cases according to the CSTE case definition. State or local health department personnel contacted patients to participate in a voluntary telephone interview. A parent or guardian was interviewed for patients <18 years of age, and a relative or medical provider could complete the interview if the patient was incapacitated or deceased. Using a standardized questionnaire, health department personnel asked patients about symptoms, healthcare-seeking behaviors, diagnosis, treatment, outcomes, underlying medical conditions, and travel history. They also collected information about laboratory tests used to diagnose coccidioidomycosis from electronic surveillance databases. Some patients who met the CSTE case definition by laboratory criteria and symptoms but who clearly had a different diagnosis (such as histoplasmosis) or whose illnesses were not believed to be clinically consistent with coccidioidomycosis were not contacted for an interview. We further excluded from the analysis interviewed case-patients with compelling evidence that their illness was caused by something other than coccidioidomycosis.

We classified Nevada, New Mexico, and Utah as low-endemic because the risk for coccidioidomycosis is lower in those states than in Arizona and California. The 11 other states in this analysis were not known to be endemic for coccidioidomycosis and were considered nonendemic. We performed descriptive analysis and examined differences between cases in low- versus nonendemic states. We analyzed categorical variables by using χ² or Fisher exact tests and used Wilcoxon rank-sum tests to compare continuous variables. A human subjects review by CDC determined this project to be nonresearch.

These enhanced surveillance data provide much-needed insight into coccidioidomycosis-related illness among patients with cases reported from states where the disease is not highly endemic. Our findings are generally consistent with the similar surveillance conducted in Arizona during 2007–2008 in terms of effects on patients (7). Delays in diagnosis, unnecessary antibacterial use, and prolonged symptoms were common, and a high proportion of patients had culture-positive coccidioidomycosis, suggesting that less severe cases might go undiagnosed. We found several differences between patients in low- and nonendemic states, which appear to be related to the underlying populations and testing patterns.

Patients experienced considerable diagnostic delays. Although the median time from symptom onset to seeking healthcare (5.5 days) was shorter than in Arizona surveillance (11 days), the median time from seeking healthcare to diagnosis (38 days) was longer than in Arizona (23 days) and in a study of patients in Missouri (25 days from onset to diagnosis) (7,10). Compared with Arizona surveillance, the shorter time from onset to seeking care in this investigation could be related to an older patient population, whereas the longer time from seeking healthcare to diagnosis is likely related to lower coccidioidomycosis awareness in low-endemic and nonendemic states. Many patients also reported initial misdiagnosis before being tested for coccidioidomycosis; however, it is unclear whether those who said they were first diagnosed with pneumonia did, in fact, receive a correct initial coccidioidomycosis diagnosis but were either not informed of the specific etiology or did not remember being told, or whether they were truly misdiagnosed with bacterial pneumonia. Misdiagnosis seems likely because of the high proportion of patients who reported being diagnosed with another illness and receiving antibacterial medication, similar to other studies (7,11). We did not observe statistically significant differences in misdiagnosis or delays in diagnosis between low-endemic and nonendemic states, indicating a need for increased healthcare provider awareness about coccidioidomycosis in all areas.

The differences between patients from low-endemic and nonendemic states appear to reflect underlying population demographics and provider testing practices rather than differences in disease severity. Enzyme immunoassay is commonly used in highly endemic areas as an initial test for coccidioidomycosis because it is high-throughput, requires less expertise, and is more sensitive (though less specific) than other serologic methods (2). Providers and laboratories in low-endemic areas might be more familiar with this test than those in nonendemic areas, whereas patients in nonendemic areas were more likely to have positive coccidioidomycosis cultures. Providers in nonendemic areas might not be testing specifically for coccidioidomycosis, but rather diagnosing it incidentally, given that Coccidioides organisms can grow on various culture media. Nearly all cultures came from invasive procedures and <10% from sputum, suggesting that diagnosis by culture was uncommon among patients with uncomplicated primary pulmonary disease not warranting invasive procedures. Some patients from nonendemic states first sought care in endemic areas and might have had coccidioidomycosis diagnosed there; however, we did not collect data on diagnosis location, making geographic differences in testing patterns difficult to fully understand. In contrast to possible differences in providers’ knowledge of coccidioidomycosis tests, patients themselves were more likely to have known about coccidioidomycosis before being diagnosed with it and were more likely to ask for coccidioidomycosis testing if they were reported from nonendemic states. The modest awareness among these patients is probably related to the fact that a high proportion resided part-time in Arizona, where public outreach about coccidioidomycosis is frequent and awareness is likely to be greater than in other areas. Patients in nonendemic states were also more likely to know about their positive results before the interview. Possible reasons that patients did not know of their positive test results include that the patient misunderstood or did not remember their diagnosis or that the provider did not inform the patient because they believed the results were not clinically relevant. The second explanation would also support the finding that patients in low-endemic states were less likely to be prescribed antifungal medications.

The most common symptoms (cough, fatigue, and shortness of breath) and prolonged symptom duration (median 38.5 days among patients who had recovered at the time of the interview) were similar to those in Arizona patients (42 days) (7). Comparable to our findings, other studies show that coccidioidomycosis symptoms, particularly fatigue, can take months to resolve and profoundly impair physical activities, resulting in missed workdays and inability to perform usual daily activities (7,12,13). The reasons that patients in low-endemic states were more likely to report chest pain, headache, joint pain, and muscle pain than patients in nonendemic states are unclear but could further reflect geographic differences in testing practices if physicians in low-endemic areas are more likely to suspect and test for coccidioidomycosis based on those symptoms. In previous studies of coccidioidomycosis test positivity among community-acquired pneumonia (CAP) patients in highly endemic areas, myalgia (11) and rash (14) were the only clinical features that differentiated coccidioidal CAP from noncoccidioidal CAP. Another study found that chest pain was a significant predictor of being tested for coccidioidomycosis among CAP patients (15). In addition, approximately half of patients in our surveillance reported fever, similar to findings in Arizona surveillance (7).

Potential recall bias is our investigation’s main limitation. However, patient interviews can yield insightful data about effects on patients’ daily activities and other information that might not be routinely available from medical records, such as detailed travel histories.

Travel to or part-time residence in Arizona was frequent among patients reported from nonendemic states. Approximately 40 million persons visited Arizona in 2015 (16), but the number of seasonal residents is more challenging to measure; the most recent figures estimated Arizona’s winter-only resident population to be ≈300,000 during 2000–2001 (17). These seasonal residents, also known as snowbirds, are typically retired, older adults who have higher socioeconomic statuses than others in their age group and who usually cite a more enjoyable climate as the reason for their part-time residence outside their home state (18). Overall, the risk for acquiring coccidioidomycosis during travel to Arizona is likely small. One expert estimated that only 1 in 17,000 visitors would experience an infection serious enough to seek medical care (19). However, the total number of cases estimated to occur in Arizona visitors is estimated to be ≈1,300 per year (19), suggesting a public health problem far larger than surveillance detects. The high proportion of cases from nonendemic states whose coccidioidomycosis was diagnosed by culture (36%, compared with <10% in Arizona surveillance) also indicates that less severe cases go undiagnosed or unreported. Obtaining a patient’s history of travel to or residence in coccidioidomycosis-endemic areas is essential for early diagnosis (2).

Some patients (59% from low-endemic states and 17% from nonendemic states) did not report travel to endemic areas in the 4 months before symptom onset. A study of cases in Missouri residents also found that approximately one quarter of patients did not report travel to endemic areas during the 3 weeks before symptom onset (10). Incomplete travel histories or travel that occurred >4 months before symptom onset likely explain the lack of recent travel to endemic areas among patients from nonendemic states in our surveillance. In low-endemic states, most cases in patients without recent travel to other endemic areas could presumably be locally acquired. A deeper understanding of the highest-risk geographic areas in those states is needed.

We classified Oregon as nonendemic; although Coccidioides spp. DNA was identified from several soil samples in central Oregon in 2016, the fungus has not been cultured from environmental samples (20). Our surveillance identified culture-confirmed coccidioidomycosis in 1 Oregon patient who did not recently travel to known endemic areas, and the acute nature of his illness did not suggest reactivation of infection acquired during his earlier travel to Mexico. The patient reported extensive exposure to alfalfa hay (of unknown source), and rare coccidioidomycosis cases have been transmitted by similar fomites (21), suggesting that acquisition from the local environment and remote travel are not the only possible sources of infection. Unfortunately, a suitable clinical isolate was not available for whole-genome sequencing and comparison to isolates from nearby states. Such testing, in combination with environmental isolates, has enabled identification of cases acquired from the natural environment in south-central Washington (4,22) and could allow for discovery of similar transmission in Oregon, if present.

Our results could be used to inform minor revisions to the CSTE case definition. In nonendemic areas or areas with unknown endemicity, interpretation of positive coccidioidomycosis serologic test results can be challenging if the patient is asymptomatic, has no relevant travel or an unknown travel history, or has laboratory evidence of a different disease. In this surveillance, >16% (54/339) of patients whose illness met the CSTE definition likely did not have coccidioidomycosis; all but 1 were from nonendemic states, and many had histoplasmosis, which is known to cause cross-reactions with coccidioidomycosis serologic tests (23). Some states are already excluding such cases from their case-counts even though the CSTE definition does not specify exclusion criteria. Similarly, the CSTE definition does not state whether cases counted in a previous year should be counted again if subsequent positive laboratory tests are reported. Most states, including Arizona, only count cases once because infection is thought to confer lifelong immunity. Seven percent of patients in our analysis self-reported a history of coccidioidomycosis, but we were not able to determine if their cases had been previously reported in other states. Last, 9% of patients we interviewed did not report symptoms, although the actual proportion could be higher because some patients were not contacted for an interview because their illnesses were thought to be clinically incompatible with coccidioidomycosis. In Arizona’s enhanced surveillance, 5% of patients had no symptoms or symptoms inconsistent with coccidioidomycosis according to the CSTE definition, suggesting that the CSTE definition’s laboratory component alone is sufficiently specific for public health surveillance (7). Further characterization of clinical scenarios involving asymptomatic patients with positive coccidioidomycosis tests could inform clinical practice and disease surveillance. Overall, the contributions of false-positive laboratory tests, previously reported cases, and asymptomatic cases to national-level case-counts are undoubtedly small, but they serve as examples of ways that coccidioidomycosis surveillance could be improved.

Although Coccidioides is most common in Arizona and California, coccidioidomycosis is a disease of national importance. Our investigation revealed many cases associated with travel to or part-time residence in highly endemic areas, as well as cases presumably acquired in Nevada, New Mexico, and Utah. Patients experienced substantial delays in diagnosis and prolonged symptoms, leading to lost productivity. The high proportion of culture-positive cases suggests that less severe cases go undiagnosed, resulting in underestimates of the actual number of cases, which is typical for public health surveillance. Greater awareness nationwide among clinicians and the public about coccidioidomycosis is needed to minimize delays in diagnosis and appropriate treatment.