Invasive fungal infections (IFIs) are a major problem in tertiary care hospitals. They affect several types of patient and are cared for by several types of physician. The difficulties in confirming a diagnosis, the excellent tolerance of new drugs, and the impact of early therapy have led to extended use of empirical antifungal agents.

Previous studies, however, have shown that up to 67–74% of antifungal drugs are used inappropriately in tertiary care hospitals.1–3 A recent multicenter study performed in French intensive care units (ICU) revealed that 7% of all patients admitted on a single day were receiving antifungals, and 65% patients included had no proven IFI.4

International societies have recommended implementation of institutional antifungal stewardship programs.5 When antifungal costs surpassed €3 million and while our incidence of proven fungal infection remained stable, we decided to initiate an antifungal stewardship program that included training initiatives. We performed a knowledge survey in order to identify areas requiring specific attention.

The questionnaire was completed by 200 of the 329 physicians invited to participate (60.8%; 115 staff [57.5%] and 85 residents [42.5%]). The percentage of responders was higher than 50% for all the departments involved, except for general surgery (40.8%), oncology (20.0%), and pneumology (12.5%). The distribution by departments was as follows: medical (including oncology and hematology) (60%), pediatrics (19%), intensive care unit (15.5%), and surgery (5.5%). The mean age of the responding physicians was 35.0±9.7 years, and 57% were women. Median duration of postgraduate training was 6 years (3.0–18.0).

The overall mean score was 5.16±1.73 (ICUs, 5.40±1.50; medical wards, 5.37±1.76; pediatrics, 4.62±1.75; and surgery 4.09±1.30). The results of the survey by hospital department are shown in Table 1. No statistically significant differences were found between the different departments in the simple linear regression analysis; however, differences were found between the medical departments and the remaining departments after adjusting for sex, postgraduate training, and physician category (p=0.017) (Table 1).

When the percentage of correct answers for each question was compared by department, statistical differences were found in 3 questions (Table 1). Unlike physicians from medical departments, the ICU, and surgical departments, more than half of the pediatricians responded incorrectly that L-AmB was the treatment of choice for candidemia before the species of Candida is identified, instead of fluconazole or a candin. Pediatricians were also more commonly unaware of the particularities of non-albicans Candida species and the correct length of therapy for invasive aspergillosis (IA). Statistical differences were only detected between the pediatrics department and the medical departments.

Comparison of the mean scores between the different categories of physicians revealed that staff (5.59±1.67) had better results than residents (4.58±1.65) (p<0.001). The multiple linear regression analysis showed statistical differences between both categories (p=0.002). The staff performed better (higher percentage of correct answers) than the residents in 11 out of 20 questions (Table 1).

Finally, most participants (84%) reported that they would be willing to take an online course to update their knowledge of antifungal therapy. Only 12% were not interested, and 4% of the physicians did not answer this question.

Ours is the first study to evaluate knowledge of current recommendations on diagnosis and treatment of IFI in a tertiary hospital with a very high consumption of systemic antifungal drugs. We demonstrated that even frequent prescribers have a significant need for continuing medical education in this area. The most common mistakes led to overconsumption of antifungal drugs, since many physicians prescribe combinations and high doses of L-AmB, as well as inappropriate treatment for fungal colonization. The physicians questioned welcomed the possibility of a custom training program.

The results from the questionnaire we present reveal gaps in both diagnosis and management of IFI in our institution. The most critical aspects for intervention are discussed below.

First, physicians sometimes find it difficult to distinguish between colonization and infection when Candida species is isolated in urine or in a tracheal aspirate. Consequently, the potential for overprescription of antifungal therapy is considerable.

Second, poor knowledge of the indications for antifungal prophylaxis and empirical treatment means that almost 50% of physicians would initiate Candida prophylaxis in an ICU patient colonized by Candida species, an ICU patient with a urinary catheter, an ICU patient with a central venous catheter, and a patient who has recently undergone surgery, without taking into consideration other risk factors. Although at least 8 clinical trials involving several types of ICU and surgery patients have demonstrated that prophylaxis with fluconazole can reduce the frequency of invasive candidiasis by around 50% and improve mortality,8–12 this strategy remains inefficient unless it targets high-risk patients13 and could lead to an increase in the frequency of subsequent azole-resistance and non-albicans candidemia.14–16 According to the latest guidelines,6 antifungal prophylaxis is only warranted in ICU patients who are at the highest risk of invasive candidiasis (>10%).17

Third, almost 80% of physicians failed to identify the indication for antifungal therapy in a patients with sepsis possibly caused by a femoral catheter infection, as recommended in the latest guidelines for the management of catheter-related infections.18 Furthermore, 30% would have delayed antifungal treatment after being informed of positive blood cultures, even though the critical window of opportunity for initiation of therapy is 12–24h after the first positive blood culture.19,20

Fourth, since most physicians were not aware of the relatively low incidence of azole resistance in Candida in our institution, they tended to prescribe more broad-spectrum antifungals. Although non-albicans strains are clearly more frequent, the rate of fluconazole resistance in most European centers is still less than 5%.21–23 It is also noteworthy that 26.5% of physicians would choose L-AmB and 27.5% voriconazole instead of a candin to treat fluconazole-resistant Candida infections. In our opinion, L-AmB should be restricted to cases of intolerance to other antifungal agents owing to its potential for toxicity and higher cost, and voriconazole should be limited to oral step-down therapy for selected cases of candidiasis due to voriconazole-susceptible C. krusei or C. glabrata.6

Finally, physicians often did not know the appropriate fluconazole dosage for treatment of Candida infections. Failure to achieve pharmacodynamic targets for fluconazole has been associated with worse outcomes.24–27 Dosing should aim to achieve target AUC/MIC ratios of at least 25 (based on Clinical and Laboratory Standards Institute MIC methodology)28 or 100 (based on European Committee on Antimicrobial Susceptibility Testing MIC methodology) to maximize the probability of clinical success; these targets usually require 6mg/kg/d (following a 12mg/kg loading dose) for susceptible isolates or 12mg/kg/d for C. glabrata or other isolates with MICs of 16–32mg/L, as recommended in the latest guidelines.6

Problems differentiating colonization from infection were also evident for invasive aspergillosis. In a previous study, only 22.3% of Aspergillus isolates in our institution corresponded to cases of probable/proven invasive aspergillosis. We proposed a prediction score that took into account the procedure used to obtain the sample, the presence of leukemia or neutropenia, and the use of corticosteroids to help clinicians in the interpretation of Aspergillus cultures.29 The use of this score could help physicians in cases where colonization has to be excluded.

Furthermore, a high percentage of the prescribing physicians still considered L-AmB as first-line therapy, whereas the largest prospective, randomized trial for the treatment of invasive aspergillosis demonstrated that voriconazole was superior to amphotericin B deoxycholate and is also considered the first-choice antifungal in international guidelines.7,30–33 The role of combination therapy as primary or salvage therapy is uncertain and warrants evaluation in a prospective, controlled clinical trial. However, many physicians consider combination therapy to be standard of care. Guidelines clearly recommend it as an approach in patients who do not respond to voriconazole or L-AmB.7

Most physicians did not know that current guidelines recommend that treatment of invasive aspergillosis be continued for a minimum of 6–12 weeks and that voriconazole plasma levels be monitored during treatment to avoid toxicity and therapeutic failure.34

Finally, we found differences in knowledge between experienced physicians and residents. As expected, hematologists and infectious diseases specialists were more knowledgeable about prescription of antifungal therapy in our institution, since they treat the largest number of patients with IFIs.

Our study is subject to 2 main limitations. First, we only assessed the knowledge of those physicians who most frequently prescribe antifungal agents: the results could have been worse if all the physicians in the institution had participated. Second, the questionnaire addressed only invasive Candida and Aspergillus infections, and knowledge of other non-invasive fungal infections and of fungal infections caused by other agents was not assessed.

Our questionnaire can be used to assess basic knowledge of antifungal therapy among antifungal prescribers in a single institution. Baseline assessment of knowledge should be part of antifungal stewardship bundles and enables us to evaluate the impact of training initiatives. The results obtained would highlight which areas, professionals, and units are amenable to intervention. However, training without an active intervention is only marginally effective in changing antimicrobial prescribing practices. Other strategies that can be implemented include the following: (i) prospective audit of antifungal use involving direct interaction with and feedback to the prescriber, with emphasis on improving the use of broad-spectrum and costly antifungals; (ii) definition and tracking of process and outcome measures in order to determine the impact of interventions, for example, a point score that evaluates the appropriateness of each antifungal prescription, the percentage of recommendations accepted, mortality attributable to IFI, incidence of adverse drug events [grade III–IV], antifungal consumption [defined daily dose], and the incidence of azole-resistant Candida infections); (iii) implementation of computerized physician order entry with clinical decision support that improves decisions on antifungal therapy through the incorporation of local guidelines and data on patient-specific microbiology cultures, liver and kidney function, drug–drug interactions, allergies, and cost. In our opinion, infectious diseases specialists and clinical pharmacists should work together to control hospital infection. The pharmacy and therapeutics committee must also play a key role, with the support of the hospital administration and medical staff management.

A simple test enabled us to assess the knowledge of prescribing physicians about important aspects of the diagnosis, prophylaxis, and treatment of IFIs. Our results revealed serious gaps that require a custom training program that could act as the first step in the implementation of an antifungal stewardship program.

This study was partially financed by PROMULGA Project (Proyecto Multidisciplinar para Optimizar la Gestión del Uso de Antifúngicos). Instituto de Salud Carlos III. PI1002868.