This study revealed the median duration of antifungal treatment to be 123.5 days (IQR, 235.0), which appeared shorter considering the recommendations of the guidelines. In addition, 29.7% of the patients had recurrence and 78.7% were hospitalized. In patients with recurrence, the initial treatment was continued for a median of 90 days, indicating an insufficient treatment duration. Moreover, only 46.8% of the patients continued treatment for 6 months, highlighting the potential challenges in sustaining treatment management and adherence.

Among the azoles, echinocandins, and polyene groups, echinocandin agent were predominantly utilized in elderly patients with lower body mass index (BMI) and more compromised baseline conditions compared to azoles (Table 1), since they are administered intravenously. This characteristic may be associated with a less favorable prognosis.

The European Society of Clinical Microbiology and Infectious Diseases/European Respiratory Society guidelines recommend a minimum of 6 months of oral triazole therapy [9]. Notably, the risk of recurrence is higher in patients with bilateral lung involvement, who may require an even longer treatment [15]. A randomized controlled trial that compared the outcomes of 6 and 12 months of treatment with orally administered itraconazole in patients with CPA reported a lower relapse rate at 2 years in the 12-month group [8]. This may be attributed to the limited evidence on the treatment duration or challenges associated with low treatment tolerability.

The percentage of medication switches was 37.9% and 29.4% for voriconazole and itraconazole, respectively. Switching may occur because of drug resistance or adverse drug reactions [16]. However, considering that Aspergillus species in Japan have a lower frequency of azole resistance than that in Europe and America [17,18,19], we believe that the primary reason for medication switches was adverse events.

The overall mortality rate at the 1-year mark was 24.2%, excluding the censored cases. This mortality rate in Japanese patients appeared higher than the treated mortality estimates for CPA reported by Denning [1]. The advanced age of our patients (Table 1), with a mean age of 72.73 years, may have contributed to this, as they lived through the 1950s—a period associated with a high risk of tuberculosis. However, our study lacks data on latent tuberculosis infection. The wide variation in background lung disease in patients with CPA across countries may also explain the differences in mortality rates [20]. Adverse events were observed in both groups, although liver function disturbances appeared to be more prominent with voriconazole, which is consistent with previous reports [21]. Further studies are required to determine the accumulation and availability of data on CPA treatment using isavuconazole.

Recently, Kimura et al. reported a database study about the epidemiology of CPA using a national administrative database in Japan [22]. The aforementioned study and ours examined newly diagnosed CPA cases and evaluated actual treatment practices, with nearly identical initial treatment outcomes. However, distinctions exist in the databases utilized and the observation duration, which was 3 years in our study, exceeding theirs by 1 year. Their research concentrated on CPA prevalence, age distribution, geographic features, expenses, and survival and prognostic factors. In contrast, our investigation emphasized actual treatment practices, detailed treatment duration, and treatment transitions using Sankey diagrams, as well as exploring the connection between treatment periods and recurrence. We also examined the rate of switching between azoles, the frequency of TDM, and adverse events potentially linked to short treatment durations or medication changes. The differing databases employed in both studies, encompassing varied healthcare settings and patient populations, may account for some result discrepancies. A significant limitation of Kimura et al.’s paper is their failure to acknowledge that follow-up likely ceased at age 75 because of presumed insurance changes. This oversight might underestimate treatment duration, potentially explaining the longer treatment periods observed in our study.

This study has some limitations. First, the database from which the data was obtained in this study is based on DPC data voluntarily provided by acute-care hospitals, which raises concerns regarding the representability of the study’s findings, and as of 2022, 1764 hospitals were participating in the DPC program in Japan [12, 13]. However, because the majority of patients with CPA are treated at these hospitals, concerns regarding representability are minimal.

Second, all adverse events may not be observed because they were determined by only ICD codes. In addition, when patients are transferred to another hospital, their data becomes inaccessible, resulting in a loss of continuity in tracking. Mortality data was limited to in-hospital deaths. However, CPA-related deaths and end-of-life care, except for terminal patients with cancer, generally occur in hospital settings. Therefore, we believe that the effect of this limitation was relatively minor.

Third, we defined relapse as the resumption of antifungal medication after more than 30 days. However, this definition could potentially include a discontinuation due to an adverse event or other circumstance.

Fourth, chest X-ray was utilized as one of the inclusion criteria instead of chest CT. Ideally, chest CT should have been performed; however, as it may have been conducted prior to the inclusion criterion of 30 days, chest X-rays were also permitted in this study.

Another limitation stems from the potential biases associated with the selected definition. For example, this study did not trace cases transferred to other hospitals or account for out-of-hospital deaths, all of which were censored. As a result of the limitations of our data in tracking hospital transfers, we lacked data on the characteristics of the censored patients. In addition, the definition of adverse events may be limited because it does not clarify the duration between the occurrence of liver function disturbances and the use of antifungal agents before onset. Moreover, other significant adverse events may have been omitted from this study. Nevertheless, despite these limitations, our study offers valuable insights into the treatment practices, trajectories, and subsequent prognoses of patients with CPA.