Delays in Coccidioidomycosis Diagnosis and Associated Healthcare Utilization, Tucson, Arizona, USA

Tucson, Arizona, USA, is a highly coccidioidomycosis-endemic area. We conducted a retrospective review of 815 patients in Tucson over 2.7 years. Of 276 patients with coccidioidomycosis, 246 had a delay in diagnosis; median delay was 23 days. Diagnosis delay was associated with coccidioidomycosis-related costs totaling $589,053 and included extensive antibacterial drug use.

with the last full year that Banner used Epic software and to impose a uniform price scheme. We manually reviewed each CPT code and tabulated costs related to coccidioidomycosis diagnosis; costs are presented in US dollars. The University of Arizona institutional review board approved this study.
We summarized categorical variables using counts and percentages and summarized time to events and costs using medians and interquartile ranges (Figures 1, 2). We showed total costs by disease category. We compared the distributions of time to diagnosis and costs across coccidioidomycosis presentation categories using nonparametric Kruskal-Wallis tests. We based pairwise comparisons of groups on nonparametric Wilcoxon-Mann-Whitney tests and CIs for differences in medians. We used a χ 2 test to compare percentages between groups. We conducted statistical analyses in R version 3.1 (https://www.r-statistics. com/2014/11/r-3-1) and defined statistical significance as p<0.05.
Of 276 patients, 30 (11%) received a coccidioidomycosis diagnosis at presentation (zero delay) and 12 of the 30 during hospital admission. Because hospitalization of these 12 either was not required for diagnosis or was required for disease complication management, we excluded coccidioidomycosis-associated costs from the analysis. Of the remaining 264 patients, 246 had delayed diagnosis of >1 day, with coccidioidomycosis-associated costs of $589,053 (82%) of the $718,401 total. Coccidioidomycosis-related costs for patients with diagnostic delay were significantly greater (p = 0.0004) than for the 18 patients without delay ( Figure 2; Appendix Table 2).
Not all CPT codes are associated with charges or reflected in patients' bills, so we assessed total effort by analyzing the frequency of CPT codes and determining whether codes were related to coccidioidomycosis infection (Appendix 2, https://wwwnc.cdc.gov/EID/article/25/9/19-0023-App2.xls). Coccidioidomycosis-related charges for CPT code-related effort was higher in patients with delays in diagnosis compared with charges for patients without delays (Appendix 1 Figure 1).
Outpatient and inpatient prescriptions for a total of 1,103 antibacterial medication orders submitted before coccidioidomycosis diagnosis are an approximate measure of antibacterial use. Vancomycin and daptomycin comprised 22% of antimicrobial drugs ordered (Appendix 1 Figure 2).

Conclusions
Our findings demonstrate significant delays in accurate coccidioidomycosis diagnosis and substantial costs associated with these delays. Median delays for different disease categories ranged from 17 to 54 days; 43% of patients with coccidioidomycosis had a diagnosis delay >1 month. The 23-day median delay we presented (interquartile range 7-74 days) corresponds with a recently reported 23-day delay in Arizona (8) and in a 2010 study (9). In our cohort, these delays were associated with $589,053 in coccidioidomycosis-related costs, as determined from Medicare fee schedules. Although median coccidioidomycosis-related costs were lowest for acute coccidioidal pneumonia, the total was greatest for this group because it accounted for >63% of patients. If our findings were extrapolated across institutions in coccidioidomycosis-endemic regions, diagnostic delays and excess healthcare utilization would probably represent millions of dollars.
Overall healthcare effort as indicated by CPT code frequencies, irrespective of whether or not they resulted in billed charges, showed similar delay-associated excesses. Of additional concern is unnecessary use of antibacterial drugs, such as broad-spectrum medications like vancomycin, in coccidioidomycosis patients before an accurate diagnosis (Appendix 1 Figure 2). If coccidioidomycosis diagnostic delays were shortened, unnecessary antibacterial treatments could be reduced greatly.
This study extends earlier reports of the economic burden associated with coccidioidomycosis (9). It is certainly an underestimate of costs, because we did not include in our cohort an unknown number of patients with coccidioidomycosis who were misdiagnosed. Our results suggest earlier diagnosis will lower costs and provide secondary benefits including patient reassurance, decreased antibacterial drug use, and improved antibiotic stewardship. This study reinforces the ongoing challenge to increase coccidioidomycosis awareness for healthcare providers and the urgent need to improve the ease, rapidity, and reliability of coccidioidomycosis testing.

About the Author
Dr. Donovan is an assistant professor with the Valley Fever Center for Excellence in the University of Arizona Department of Medicine. Her primary research interest is medical mycology, with an emphasis on fungus-host interactions.