We identified 44,292 cases of coccidioidomycosis (43,210 from the Commercial and Medicare databases and 1,082 from the Medicaid database) during 2001–2022. Of those cases, 26,905 met the inclusion criteria of being continuously enrolled for >1 year before their index diagnosis (26,379 from Commercial and Medicare and 526 from Medicaid). The median age of patients in the cohort was 52.0 years; 51.5% of the cohort patients were female and 48.5% were male (Table 1). Most patients were identified in the Commercial claims database (80.2%), followed by the Medicare (17.9%) and Multi-State Medicaid (2.0%) databases.

In the year before the index coccidioidomycosis diagnosis, 26,398 (98.1%) patients had >1 healthcare visit for any reason, and 23,191 (86.2%) patients had >1 SSD visit. A total of 177,679 SSD-associated visit days were noted in the 365 days before the index diagnosis. The frequency of SSD-related visits increased substantially beginning ≈100 days before the index diagnosis (Figure 1, panel A).

During the 13-week diagnostic opportunity window, 88,130 total SSD visit days from 19,670 (73.1%) patients occurred, representing potential missed opportunities. We plotted the expected trend (in red) of SSD-related visits, estimated on the basis of visits during the crossover period from 92–365 days before diagnosis (Figure 1, panel B). We also plotted the distribution of the estimated number of missed diagnostic opportunities each day before the index diagnosis (Figure 2).

On the basis of the bootstrapping approach, we estimated that 43,338 (95% CI 40,566–46,125) of total SSD visit days during the diagnostic opportunity window represented actual missed diagnostic opportunities (Table 2). Approximately 73.2% (95% CI 72.4%–74.1%) of missed opportunities occurred in an outpatient setting, 18.6% (95% CI 17.8%–19.5%) in inpatient settings, 7.5% (95% CI 7.2%–7.7%) in emergency department settings, and 0.7% (95% CI 0.6%–0.7%) in observational stay settings.

Approximately 59.7% (95% CI 58.3%–61.1%) of patients experienced >1 missed diagnostic opportunity (Table 3). Of the patients who experienced >1 missed opportunity, they experienced on average 2.70 (95% CI 2.58–2.81) visits representing missed opportunities and had an average diagnostic delay duration of 29.69 days (95% CI 28.25–31.18). An estimated 24.7% (95% CI 22.9%–26.7%) of patients experienced a diagnostic delay duration that lasted >30 days.

Patients with Medicare (odds ratio [OR] 0.61 [95% CI 0.47–0.78]) or Medicaid (OR 0.61 [95% CI 0.48–0.77]) were less likely to experience a delay (i.e., experience >1 missed diagnostic opportunity) than were patients with commercial insurance (Table 4; Appendix Table 3). Patients >65 years of age were more likely to experience a delay than were patients <18 years of age (OR 1.48 [95% CI 1.15–1.98]). Patients in rural locations were slightly more likely to experience a delay (OR 1.27 [95% CI 1.04–1.60]). Patients with underlying history of pulmonary conditions or related procedures were more likely to experience a diagnostic delay (i.e., asthma [OR 1.34 (95% CI 1.18–1.49)], COPD [OR 1.56 (95% CI 1.36–1.80)], chest computed tomography [OR 1.42 (95% CI 1.26–1.61)], and chest radiography [OR 1.21 (95% CI 1.12–1.30)]). Finally, receipt of medications to treat potential symptoms of coccidioidomycosis during the diagnostic opportunity window was positively associated with the likelihood of experiencing a diagnostic delay (i.e., antibiotics [OR 3.42 (95% CI 3.18–3.69)] and inhalers [OR 2.64 (95% CI 2.34–2.99)]).

We observed similar patient-level delay metrics for patients residing in Arizona and patients not residing in Arizona (Appendix Table 4). However, a greater proportion of patients in the Arizona cohort experienced >1 missed diagnostic opportunity than did patients in the non-Arizona cohort (69.6% vs. 55.5%).

The number of outpatient antibiotic prescriptions increased substantially beginning nearly 90 days before diagnosis (Appendix Figure 2). The trend in antibiotic prescriptions mirrored the trend in SSD visits, suggesting that SSD visits attributable to diagnostic delays were associated with excess antibiotic prescriptions (Figure 3).

We found that many patients who ultimately received a diagnosis of coccidioidomycosis experienced multiple healthcare encounters before diagnosis, and many of those visits represented potential missed opportunities for earlier diagnosis. Specifically, 60% of patients experienced >1 missed opportunity; among those persons, we estimated an average of 2.70 visits per patient to be potential missed opportunities, and ≈73% occurred in the outpatient setting. We estimated that the mean diagnostic delay was ≈30 days (median 24 days). In addition, elderly patients, patients living in rural areas, and those with underlying respiratory conditions were more likely to experience a delay in diagnosis. Furthermore, prescription of antibiotics or inhalers during the delay window increased the odds of experiencing a missed opportunity to diagnose coccidioidomycosis.

Coccidioidomycosis, similar to other endemic fungal infections, is an infectious disease that is commonly associated with diagnostic delays (28). Overall, our results largely align with previous studies reporting diagnostic delays for coccidioidomycosis. For example, a study of health claims for 139 patients in a large health system based in Phoenix, Arizona, found that 46% of patients had a diagnostic delay >1 month in length (19). A study in 14 US states found that among 339 patients, a median of 38 days passed between first seeking care and diagnosis (29). However, those studies might have described longer delays because determining when symptoms began is often difficult. For example, Benedict et al. (29) reported 1 case in their series with an extremely long interval of 1,654 days (29). Moreover, symptoms caused by other conditions have often been attributed to coccidioidomycosis in previous study designs of diagnostic delays (29,30). We included a much larger sample size than previous studies and used a case-crossover and bootstrapping approach, enabling us to distinguish attributable visits from background healthcare use, providing more accurate estimates of missed diagnostic opportunities. Indeed, our finding that 59.7% of patients experienced a diagnostic delay directly contrasts with a recent study (30) that used the same data source and similar study population but reported that 71.3% of patients experienced a diagnostic delay. The difference in our findings is most likely attributable to the fact that the previous study did not account for an expected level of background healthcare utilization and, consequently, labeled every patient with potential signs or symptoms in the 90 days before diagnosis as a case of diagnostic delay.

We identified multiple factors associated with the odds of diagnostic delays for coccidioidomycosis. First, delays were more common in older patients (i.e., >65 years of age compared with persons <18 years of age), an unsurprising finding, given that older adults frequently demonstrate more vague or subtle signs and symptoms for infectious diseases (31,32). For example, older adults are substantially less likely to experience a fever than are younger patients (33). Second, patients living in rural areas were more likely to experience a missed opportunity to diagnose coccidioidomycosis. Diagnostic delays among rural populations might be partially attributed to lack of access to medical care. Rural patients are substantially less likely to have access to healthcare, especially specialty services, and might lack access to some diagnostic tests and services (34,35). Similarly, we found that most missed diagnostic opportunities occurred in outpatient settings where less diagnostic testing is available than in hospital settings. Previous reports have also indicated that a large proportion of cases are diagnosed during hospital stays rather than during outpatient visits (36).

Patients with previous lung disease (e.g., asthma, COPD) or pulmonary imaging were also more likely to experience a missed opportunity, which is likely secondary to attributing the signs and symptoms associated with coccidioidomycosis to the patient’s previous lung disease. Indeed, missed opportunities were associated with prescriptions for inhalers, suggesting that some missed opportunities were caused by clinicians attributing symptoms related to coccidioidomycosis to reactive or obstructive airway diseases. Comparable results have been reported for other lung infections, including tuberculosis (25) and histoplasmosis (23).

Surprisingly, patients living in Arizona were at higher risk of experiencing a missed diagnostic opportunity. We propose 2 possible reasons for this finding. First, in Arizona, the harder-to-diagnose cases (e.g., mild cases with subtle symptoms) might be diagnosed, but only after longer-than-average periods of time. However, in non–coccidioidomycosis-endemic regions, harder-to-diagnose cases might go undiagnosed altogether, leaving only the cases with classic symptoms, which are easier to diagnose. As a result, the time to diagnosis might appear shorter than it would otherwise in nonendemic regions. Second, although coccidioidomycosis is much more common in Arizona than in nonendemic regions, physicians in Arizona might not prioritize coccidioidomycosis as highly in their differential diagnosis because of a form of base-rate neglect. Indeed, physicians in nonendemic regions might consider coccidioidomycosis in their differential diagnosis as soon as history of travel to an endemic region is discovered. Rapid consideration of coccidioidomycosis is based on activation of a classical illness script rather than a consideration of the actual probability of disease risk. That overconsideration of coccidioidomycosis might help to reduce diagnosed delays for this disease outside of endemic regions.

Several studies have focused on the inappropriate use of antibiotics before the diagnosis of coccidioidomycosis (13–16,37,38); most patients receive an antibiotic before diagnosis, and a substantial proportion of patients receive multiple rounds of antibiotics (38). In a California study (37), 70% of patients testing positive for coccidioidomycosis received an antibiotic before diagnosis; the median number of antibiotic prescriptions for that population was 3. Our modeling framework enabled us to explore the potential association of antibiotic prescriptions and the likelihood of experiencing a missed opportunity before the coccidioidomycosis diagnosis. We found that persons who had been prescribed an antibiotic during the delay window had >3 times the odds of experiencing a missed opportunity. That finding might suggest that diagnostic delays are a potentially unappreciated harm with respect to inappropriate antibiotic use. Similar findings have been reported for tuberculosis (25), histoplasmosis (23), pertussis (21), herpes encephalitis (24), and sepsis (20). Current Infectious Diseases Society of America community-acquired pneumonia guidelines are silent on testing for endemic mycoses (39). Therefore, it might be useful for patients not improving on an initial course of antibiotics to be reevaluated for alternative diagnoses, including endemic mycoses, given their frequency in certain regions or after travel to high-risk areas.

The first limitation of this study is that we relied on administrative claims data that are routinely collected for billing purposes but might not capture key factors related to diagnostic delays. For example, we used diagnostic codes to identify cases of coccidioidomycosis and previous symptomatic visits. We did not have access to laboratory results or radiology reports to confirm cases of disease. We also did not have access to clinical notes that could have captured signs and symptoms not assigned to a diagnosis code. Thus, some patients with coccidioidomycosis might be misclassified, and symptoms that occurred before diagnosis might have been underidentified. In addition, although we have limited geographic data (i.e., enrollee state, metropolitan statistical area for Commercial and Medicaid populations), details of travel history or localized exposures, which might be confounders in the diagnostic delay process, were unknown. Finally, the study population was restricted to commercially insured persons who were enrolled sometime during 2001–2022 and maintained continuous enrollment coverage for >1 year before diagnosis. That factor might limit generalizability to uninsured populations, persons who have frequent gaps in insurance coverage, or current and future patient populations outside the study period.

In conclusion, our study suggests that diagnostic delays in coccidioidomycosis are common and that multiple missed opportunities occur frequently. Raising clinician awareness and expanding access to timely diagnostic testing, particularly in outpatient settings, are critical steps toward improving outcomes for patients with coccidioidomycosis.