In this study, an expert panel developed 44 Likert scale statements across 6 key domains concerning the diagnosis and management of IFI in the ICU which were then evaluated by an online survey with 335 respondents in order to identify unmet needs and from this derive recommendations to improve IFI care.
In terms of patients at risk, there was strong agreement that physicians should be aware of the local incidence of different IFIs in their ICU, which has been shown to vary widely [24]. There was also strong agreement that the risk of developing IFI is dependent on patient characteristics. In line with evidence from the literature [16, 24–26], there was agreement that risk groups differ between IA (patients with COPD, patients who are immunocompromised, patients with viral pneumonia, liver cirrhosis, or autoimmune diseases) and invasive Candida (IC) infections (any patient with disruption of the skin and gastro-intestinal barrier, and those receiving treatments that change the composition of the gut microbiome).
There was also agreement that early reliable diagnosis of IFI is of utmost importance, especially for IA. IA requires diagnosis and treatment initiation at early phases as mortality is very high, despite appropriate antifungal therapy, once angio-invasion occurs [27]. Therefore, 90% agreed that the ideal sample for diagnosing IA in the ICU is a bronchoalveolar lavage (BAL) for galactomannan (GM) and culture. Also, for IC infections, delays in starting treatment is a major factor predicting mortality [28, 29].
A major obstacle to early diagnosis, and identified as a challenge from the survey, is the long turnaround time to receive results from mycological tests; 68% (Statement 10) stated that receiving a beta-D-glucan (BDG) test in their institution would take more than 3 days, and for 61% (Statement 11) the same was true for a GM test. Analysis by respondent country highlighted that the UK, Italy, and France experience delays greater than 3 days for turnaround of BDG test and serum GM (BDG: 81%, 75% and 80% respectively. GM: 79%, 72% and 72% respectively). For the UK, turnaround time for BDG (> 48h in 87%) and GM (> 48h in 86%) has been reported as a problem in a previous study [30]. This was driven by the fact that few laboratories had access to local testing for BDG (3/63, 5%) or GM (13/63, 21%), and most send samples out to reference laboratories [31]. These rates may be even lower in some of the more resource-limited countries in Europe, as shown recently by the European Confederation of Medical Mycology (ECMM) laboratory capacities survey [1].
To improve diagnosis and outcomes of IFI, there were very strong agreements (Statements 12, 13, and 14, 96% respectively) that access to (i) serum BDG test results, (ii) serum or BAL GM test results, and (iii) PJP polymerase chain reaction results should be available in all ICU facilities within around 48 hours. Furthermore, there was also strong agreement that fungal culture, species identification, and susceptibility testing should all be available in-house at every hospital. These results indicate that where possible, all fungal testing should be conducted in-house to ensure more rapid turnaround time (< 48 hours) and to ensure patient care in this setting where every hour counts. These findings and recommendations provide a key area for improvement and can become a target for programmes such as the ECMM, International Society of Human and Animal Mycology (ISHAM), and the American Society for Microbiology (ASM) One World – One Guideline initiative [22, 32] and other initiatives. Another obstacle to early diagnosis is the lack of reliable diagnostic criteria for IA in patients who are non-neutropenic, as the established criteria for the neutropenic setting often cannot be applied to those patients who are non-neutropenic (Statement 9, 86%). Currently, efforts are underway to create improved classification criteria for the ICU setting [33].
Early initiation of appropriate antifungal therapy is essential for a successful outcome of IFI [34]. This was recognised in the current study, where there was strong agreement among HCPs that empirical treatment should start immediately and prior to receiving the results from any diagnostic test if there is high clinical suspicion and, especially, high clinical severity.
As part of this study, respondents were also asked what they understood the role of serum BDG testing in cases with clinical suspicion of an IFI in the ICU. Here, 77% of respondents stated that a serum BDG antigen test should be taken before initiation of any antifungal treatment for patients with suspected Candida infection (Statement 20). It is perhaps interesting to note that lower agreement to this statement was seen from Germany, Spain, and France (70%, 73%, and 69% respectively). Respondents from the UK displayed an agreement of 83%, whilst Italy passed the threshold for very high consensus at 90%. This is perhaps as a result of the varying availability, turnaround time, and reimbursement of serum BDG testing. For example, some guidance within the UK National Health Service recommends the use of serum BDG testing, but does not state it is mandatory [35].
Evidence suggest that serum BDG has a use in guiding the cessation of empirical antifungal therapy via a negative result [36] because of the high negative predictive value (NPV) the test presents (of around 90–95%) [21]. Given that serum BDG tests become negative under appropriate antifungal therapy (and are therefore also used for treatment monitoring) [37] it is therefore important to obtain a sample for serum BDG testing before initiation of antifungal therapy, otherwise the NPV would decrease [38]. In contrast, serum BDG may be less reliable as a marker driving initiation of antifungal therapy in suspected cases of Candida infection with positive test results. Data from a recent randomised controlled trial examining the use of antifungal therapy in ICU patients with sepsis at a high risk of Candida infection determined that, in this select group of at-risk patients, that serum BDG guided initiation of antifungal therapy did not improve 28-day mortality and may be associated with overtreatment (i.e., treating patients with antifungals without IC infection) [39].
Therefore, with the support shown by the respondents to the statements (particularly Statement 31, 80%), serum BDG should be taken at initiation of treatment and results used thereafter to help guide the discontinuation of empirical antifungal therapy for invasive candidiasis when there is no further evidence of an infection present in the patient. While being supported by a strong recommendation (low quality evidence) by the European Society of Intensive Care Medicine-European Society of Clinical Microbiology and Infectious Diseases Candida infection task-force [40], there was a lack of consensus observed for Statement 32, “In treatment of suspected (and not proven) Candida infection without a clear clinical focus, if no improvement in clinical situation is seen within 4–5 days, then treatment can be stopped” (69%). This suggests a view amongst respondents that clinical findings alone are not sufficient for the discontinuation of treatment and further outlines the need for more BDG testing in order to support decisions on IC treatment discontinuation. The use of biomarkers to support the decision to stop antifungal treatment can increase the likelihood of appropriateness of this decision.
In terms of selecting initial treatment, there was strong agreement among HCPs that the choice of treatment for an IFI in the ICU should be based on the fungal pathogen, susceptibility testing, pharmacokinetics, co-morbidities, and the site of infection. This highlights the importance of knowing the local epidemiology in terms of Candida species and susceptibility in order to increase the likelihood of appropriate empiric antifungal treatment, while waiting for the microbiological results of susceptibility testing. Another important consideration is drug-drug interactions, and there was an 82% agreement (Statement 34) that hospitals should select antifungal treatment based on their ability to perform therapeutic drug monitoring (TDM) as low drug levels are one of the potential effects of drug interactions. This seems of particular relevance for voriconazole treatment where TDM is strongly recommended in guidelines [41–43], but not always available in institutions across Europe [1]. Furthermore, a recent ECMM survey of IFI diagnostic capacity in Europe showed that availability of TDM was closely related to the gross domestic product of the country in which the institution was located [1].
In cases where culture/specific diagnostics are not possible, a broad-spectrum antifungal treatment should be selected. In cases of suspected mould infections, these broad-spectrum agents would include liposomal amphotericin B, isavuconazole, and posaconazole with activity not only against Aspergillus (for which voriconazole would be a viable option [8]) but a wide range of non-Aspergillus moulds [22]. For Candida infections, echinocandins would be the initial treatment of choice [44]. It was recognised that azole resistance in Aspergillus [45] and other fungal pathogens [18] is an emerging problem that requires future actions. HCPs agreed that in case of local azole resistance rates of > 10%, alternative choice of antifungal treatment should be considered empirically.
In terms of special populations in the ICU, HCPs strongly agreed that patients on extracorporeal membrane oxygenation (ECMO) are a population with great unmet needs, having generally lower levels of antifungals [46]. Therefore, either higher dosages or TDM with rapid dosage adjustments is needed. However, it was also acknowledged that data on antifungal treatment in patients on ECMO are limited, and further studies are needed. There was also agreement that in patients with renal impairment, renal replacement therapy options should be taken into account when selecting antifungal treatment.
HCPs agreed with the statement that the fungicidal mode of action of antifungal treatment in abdominal infections is an important treatment attribute. There was also overwhelming agreement that drug-drug interactions are an important consideration when selecting treatments for patients. This may have contributed to the 42% of HCPs agreeing with the statement that voriconazole is not a suitable treatment option in ICU patients (Statement 21). However, if voriconazole is used for treatment, there was a 91% agreement that TDM of voriconazole is essential, and results should be available within 48 hours after achieving steady state to optimise the treatment regimen (Statement 22). Similarly, 92% agreed that if voriconazole fails to reach the required therapeutic levels after reaching steady state, either a dose escalation or a change to another antifungal would be the most appropriate strategy (Statement 24).
Our study also outlines important areas that should be targeted by educational efforts and evidence building. For example, there was a clear divide between the different respondent specialities, as well as country of practice, around the role of voriconazole treatment and antifungal prophylaxis (Fig. S2, Statements 21 and 41). This suggests that there is an area for improvement to raise the awareness of all HCPs around what constitutes optimal treatment modalities so there is there is greater consistency. Statement 44 discusses the lack of confidence HCPs may have in regard to using antifungal agents and interpreting the diagnostic results. It is concerning that 79% agreement was displayed to this statement. This highlights a major area for improvement and educational efforts focussing on these often rare and orphan disease to be undertaken. Proactive formal training, with a rigorous education package around IFI, could help improve clinical confidence of HCPs in using antifungal treatments and interpreting the diagnostic results. This should include the available treatment methods to combat infections and provide the necessary support to implement antifungal use so that patient outcomes can be improved.
Identified areas of IFI management in need of improvement:
Based on the results obtained over the course of this study and the discussion held by the expert panel to review the findings, the authors have identified the following areas requiring attention and the below measures as a way forward as to how these remaining IFI challenges in the ICU can be addressed (Table 2).
Table 2
Expert panel proposed measures to address challenges identified from the ICU consensus survey
No. | Measure |
1 | Education around appropriate diagnostic tests and interpretation of the results for different types of IFI (and particularly that of invasive aspergillosis) should be increased as well as education on the appropriate use of different antifungal treatments to improve patient outcomes. |
2 | Physicians should be aware of the incidence of different invasive fungal infections and their associated rate of azole resistance, especially for Aspergillus and non-albicans Candida species in their unit, through utilisation of centralised databases to therefore take steps to improve the standards of care within their institution for patients with high risk factors for IFI. |
3 | All testing should be done in house where available to reduce turnaround times and increase efficiency of care. |
4 | Access to testing for invasive fungal infections should be available to all ICU centres, with provisions made to supply point of care and single sample tests to smaller centres or where labs are centralised and off-site. |
5 | Mycological test results should be available within 48 hours to aid diagnosis, appropriate treatment initiation and discontinuation. |
6 | Serum beta-D-glucan tests should be used prior to the initiation of treatment with antifungal agents in patients with suspected Candida infections to guide the early discontinuation of treatment were appropriate. |
7 | A greater body of evidence around the benefit of prophylaxis for invasive fungal infections in the ICU should be developed. |
8 | To improve the standards of care for patients affected by IFI, and support clinicians in the management of IFIs, a rigorous education package should be provided around the utilisation of antifungal agents and provided the necessary support to implement their use as swiftly as possible when an IFI case is suspected. |
This study has a number of strengths. The main strength of this study compared to expert statement publications is the validation of the consensus statements by a multidisciplinary cohort of healthcare professionals involved in the management of IFIs in the ICU in five different European countries. There was good representation across all countries included within the study with over 300 respondents, lending greater weight to what ICU healthcare professionals actually view as current status of care, optimal care and subsequent challenges faced in their clinical practice. The large study population provides further weight to the findings, especially given the distribution of results across the ICU roles tested.
The study had some limitations. Firstly, the wording of some statements may have been ambiguous and therefore have impacted the agreement displayed. Statements were developed by the expert panel and then ratified by each member individually before being used to develop the survey to reduce any potential bias that may have emerged.
Secondly, responses were sought using an incentivised methodology through a standard market research approach, which may have introduced participant bias. Respondents may have felt pressured to align their answers with the perceived objectives of the research group.
Thirdly, respondents’ expertise in IFI management was likely heterogeneous. However, all respondents were screened to confirm involvement in IFI diagnosis and management, and had to have a predefined number of years of experience in their roles, so responses were valid in reflecting the current situation in centres across Europe.
Considering these factors, all results were analysed by the project facilitators for potential problems. The full data set was also presented to the expert panel for ratification and to ensure that no bias was present of inadvertently introduced.
Finally, although externally validated, the proposed statements and measures have been developed through modified Delphi consensus approach and as such are expert opinions of the panel and respondents and as such are not evidence-based recommendations.