Current commercial automated or biochemical identification systems misidentify C. auris, often in a predictable manner. Yeasts identified as any of the organisms by the corresponding presumptive identification method (see Table 1) should be suspected to be C. auris, particularly if found to be fluconazole resistant, and tested further as per the recommended laboratory algorithm (see Figure 1).

Early identification of C. auris is important to guide appropriate antifungal treatment and to implement appropriate IPC measures. The laboratory should suspect C. auris when specimens are submitted from facilities or units known to be endemic for this pathogen. In a recent South African study, the risk of C. auris candidaemia (vs. fungaemia caused by any other Candida species) was threefold higher among patients admitted to private-sector hospitals. Other risk factors included prior antifungal treatment, older age, longer hospitalisation before first positive culture and a central venous catheter in situ.⁹ Current commercial identification systems often misidentify C. auris as the organisms listed in Table 1.^19,20 C. auris is almost uniformly resistant to fluconazole¹⁰; if a yeast is found to be resistant to fluconazole and the first-line automated or biochemical identification system also yields an unexpected identity (Table 1), consider C. auris and refer to a laboratory with Vitek 2 YST software version 8.01 or a matrix assisted laser desorption ionisation-time of flight (MALDI-TOF) mass spectrometry instrument or molecular testing platform.

Candida auris isolates are frequently misidentified in the clinical laboratory. They are germ tube-negative yeasts and are able to grow at relatively high temperatures (42°C).¹¹ They appear pink or purple on chromogenic Candida agar (CHROMagar, Paris, France). Confirmation of species-level identification can be performed using either a MALDI-TOF instrument (such as VITEK MS [Biomérieux, Marcy l’Étoile, France] or Bruker Biotyper [Bruker, Billerica, MA, USA] using the corresponding research use only or customised databases) or the Vitek 2 YST ID system (Biomérieux) updated with software version 8.01.^19,24 Molecular identification is the reference standard method.^25,26 Candida should be routinely identified to species level if isolated from a sterile site such as blood, cerebrospinal fluid, tissue, pus from deep abscesses, etc. Not all diagnostic laboratories routinely identify Candida species other than Candida albicans from non-sterile sites to species level. This may result in under-reporting during outbreaks. The guideline development group believe that species-level identification is particularly important to detect C. auris from all specimens for the following reasons: C. auris outbreaks may be prolonged and difficult to control; patients who are colonised represent an important reservoir for transmission. C. auris is potentially multidrug-resistant, with consistently high fluconazole MICs and occasionally high amphotericin B and echinocandin MICs. Reported cases of therapeutic failure have been documented with azoles and amphotericin B.^3,16,17

If carefully standardised and quality-controlled, antifungal susceptibility testing can yield reproducible MICs that facilitate selection of the optimal antifungal agent for use in a particular clinical scenario. Most laboratories perform routine testing on isolates from sterile sites. In certain circumstances, outlined in the recommendation above, antifungal susceptibility testing should be performed on non-sterile site isolates. Although very important, an MIC is not the only factor to be considered when selecting an antifungal agent. The ability of an antifungal agent to kill the pathogen may be important for early treatment success and to reduce the chance of persistent, recurrent or relapsed infection.²⁸ Some infected body compartments or sites (e.g. the central nervous system, urinary tract, eye and intra-abdominal abscesses) are not easily penetrated by echinocandins and the pharmacokinetics/pharmacodynamics of various agents should be compared.

A standardised reference BMD test is the recommended antifungal susceptibility testing method to resolve discrepancies and to confirm unusual phenotypes. A direct comparison of the European Committee on Antifungal Susceptibility Testing (EUCAST) and US Clinical and Laboratory Standards Institute (CLSI) BMD methods for a C. auris isolate collection yielded similar MICs for fluconazole, itraconazole, voriconazole, isavuconazole, posaconazole, anidulafungin, micafungin and amphotericin B.²² When CLSI-BMD and the commercial automated Vitek AST-YS07 were compared, there was 100% agreement of MIC₅₀ values for voriconazole, caspofungin and micafungin and agreement for fluconazole and flucytosine within two dilutions. Of concern is that Vitek AST-YS07 yielded falsely elevated MICs (MIC₅₀ of 8 µg/mL) for amphotericin B compared to the CLSI-BMD MIC₅₀ of 1 µg/mL and an Etest MIC₅₀ of 0.5 µg/mL.²⁹ The guideline development group therefore recommends that all amphotericin B MIC results obtained with Vitek 2 AST-YS07 system should be confirmed with another method. There are no data comparing Sensititre YeastOne or Etest MICs to reference BMD MICs for C. auris; however, these methods provide MICs with close approximation to the reference methods for other Candida species. Laboratories should avoid testing or reporting caspofungin MICs for detection of echinocandin resistance because this method is subject to error²¹; however, any echinocandin (including caspofungin) can be used for clinical treatment if the pathogen is shown to be echinocandin susceptible. Mutations in the hotspot regions of the FKS genes are usually associated with echinocandin resistance in C. auris, although very few laboratories currently perform FKS gene sequencing.

There are currently no clinical breakpoints for C. auris and any antifungal agent. As limited clinical and pharmacokinetic/pharmacodynamic data currently preclude the development of such breakpoints, ECOFFs may be helpful. Epidemiologic cut-offs distinguish organisms with and without phenotypically expressed resistance mechanisms for a species and an antifungal agent in a defined test system; within a species, this is the highest MIC of organisms lacking phenotypically expressed resistance. Epidemiologic cut-offs may thus be used to identify isolates that are less likely to respond to antimicrobial therapy because of acquired resistance mechanisms (Table 2). Surveillance data from the NICD (N.P. Govender, pers. comm., unpublished) obtained from C. auris bloodstream isolates from South African public and private-sector hospitals roughly align with tentative ECOFFs determined for 123 C. auris isolates.²² The US CDC has applied tentative non-validated clinical breakpoints developed for other Candida species to C. auris for epidemiological purposes; however, these may not necessarily be clinically relevant at an individual patient level.²⁷ Susceptibility data for C. auris isolates published from multiple countries demonstrate uniformly high fluconazole MICs, with variable susceptibility to other azoles, echinocandins and amphotericin B.¹⁰ Some isolates may demonstrate high MICs to ≥ 2 antifungal classes (i.e. multidrug-resistant).

Candida auris is an emerging and multidrug-resistant pathogen that spreads rapidly in healthcare settings. The overarching goal of national surveillance is to provide information to prevent C. auris from becoming endemic in healthcare facilities and communities across South Africa and facilitate preparedness in laboratories for accurate detection and in IPC programmes for prevention and control.³⁰ The objectives of surveillance should be:

At a healthcare facility level, all public-sector hospitals and private hospital groups should passively monitor the number of cases of C. auris disease and colonisation by maintaining a line-list of culture-confirmed cases. The facility IPC practitioner(s) should be promptly notified of every C. auris case and should keep a record of the number of cases, by site of infection, wards where cases occurred and rates of infection, if possible, on a monthly basis. Facilities may be classified into three tiers (regular re-classification should be done by the facility IPC practitioners).

At a national level, NICD should conduct regular cross-sectional surveys as part of integrated AMR surveillance. These surveys could be scheduled at the same time every year and could be integrated with national point prevalence surveys for healthcare-associated infections (HAI) and AMR.²³ NICD should coordinate nested epidemiologic studies through its existing surveillance platforms. C. auris is included in a list of alert organisms that South African healthcare facilities are encouraged to compile.³¹ Guidance has been issued from several other public health agencies across the world. US facilities are currently requested to report all cases to the CDC by using a dedicated email address.³² Public Health England (PHE) currently requests facilities to report all new cases of colonisation or infection to their local PHE Centre Health Protection Team. The European Centre for Disease Prevention and Control (ECDC) recommends that member states should consider laboratory-based notification of C. auris invasive disease and prospective data collection at national level. Surveillance systems for HAIs should be updated to include C. auris in the list of reportable pathogens associated with HAIs.

An outbreak is defined as a sudden temporal increase in the number of cases of C. auris colonisation or infection within a unit or facility compared to a baseline, with epidemiological links which suggest clustering. The definition of an outbreak will not necessarily be the same for all units or facilities; therefore, each facility should be aware of their own tier status and distribution of prior cases within the facility. All suspected clusters or outbreaks should be reported by the facility IPC practitioner or laboratory to the relevant district CDC team and to the NICD in the following high-priority scenarios (Table 3). Not all outbreaks will require the same type of response. As resources for outbreak detection and response are limited, particularly in the public sector, urgent outbreak response efforts should be focused on:

Outbreak response activities may include, but are not limited to:

Outbreak investigations reported from other countries describe response activities that have been effective. Following a large outbreak in a cardiothoracic facility in the United Kingdom, screening of all direct contacts was recommended. Screening of hospital personnel had a very low yield and was not recommended.³³ In the United States, screening of close contacts of 77 case patients resulted in identification of an additional 45 patients with C. auris colonisation. Public health surveillance and ongoing investigations were recommended.²³

Two sets of precautions are recommended (Table 4):

Standard precautions apply to all patients and in all situations, regardless of diagnosis or presumed infection/colonisation status. Standard precautions apply to blood, all other body fluids, secretions and excretions except sweat (regardless of whether they contain visible blood or not), non-intact skin and mucous membranes. As part of standard precautions, 70% alcohol-based hand rub is recommended for hand hygiene; a combination of chlorhexidine and alcohol may provide additional benefit.³⁴ Personnel should perform hand hygiene before touching a patient, before a clean/aseptic procedure (e.g. inserting a peripheral line), after body fluid exposure, after touching a patient and after touching patient surroundings. Hand hygiene adherence should be measured with a standardised checklist and adherence should be monitored on a regular basis in all wards of a facility on a rotating basis. Routine hand sampling of staff to monitor adherence to hand hygiene is not recommended.

Contact transmission-based precautions (including isolation, cohorting and use of personal protective equipment such as disposable aprons and gloves) are not specific to C. auris and are recommended for several other multidrug-resistant organisms.³⁵ Adherence to contact precautions should be monitored on a regular basis in all wards with patients who have contact precautions implemented because of C. auris infection and/or colonisation. If this level of monitoring is not possible, consider monitoring adherence primarily in the isolation unit where patients with C. auris are cohorted.

The duration of colonisation is not clearly defined; in some cases, colonisation with C. auris may persist for many months, perhaps indefinitely.^3,36 The optimal approach to reduce the skin or mucosal surface microbial load (i.e. decolonisation) of infected or colonised patients with C. auris has not been determined.³⁷ While daily topical application of chlorhexidine gluconate 0.5% (including body washes and mouth gargles) has been recommended by at least one public health agency, patients have been documented to remain colonised with C. auris in prolonged outbreak settings despite this intervention.³³ Similarly, the use of chlorhexidine-impregnated central vascular catheter dressings or topical nystatin has not been evaluated and these interventions are not recommended. Therefore, the most conservative approach for patients who are known to be infected or colonised with C. auris is to maintain contact precautions for the duration of admission. Patients known to be colonised or infected with C. auris should also be isolated when re-admitted to a healthcare facility; we have not specified a recommended time limit since the last admission because colonisation may be prolonged.

Routine screening of all newly admitted patients is not feasible or recommended in a resource-constrained setting. However, screening may be considered in an outbreak situation to establish colonisation of epidemiologically linked patients. Epidemiologically linked contacts are defined as patients who are currently sharing a cubicle with a confirmed case. In areas that do not have cubicles, but are shared rooms with or without semi-permanent barriers, epidemiologically linked contacts include all patients in a shared physical area. Given the likely rapid colonisation potential of C. auris, the IPC practitioner could also consider screening any roommates the case patient may have had during the last month. Screening of healthcare personnel during an outbreak is not routinely recommended owing to the difficulty of evaluating the role of healthcare workers in the transmission of pathogens between patients and because the reported prevalence of carriage is relatively low.³³

In an outbreak situation to establish colonisation of epidemiologically linked patients, specimens that could be submitted include the following: axillary skin swabs, groin skin swabs, nose/throat swabs, rectal swabs or stool samples, urine, wound fluid and respiratory tract specimens. The axillae and groin areas appear to be the most common and consistent sites of colonisation. We recommend that IPC practitioners should wait at least 48 h after administration of topical antiseptics, for example chlorhexidine, before collecting specimens for C. auris colonisation. An enrichment protocol has been described to optimise laboratory isolation of C. auris from colonisation samples.¹⁴ If a patient screens positive for C. auris, no further sampling is indicated. A negative colonisation screen should not be used as evidence to discontinue contact transmission-based precautions in a person with prior culture-confirmed invasive disease or colonisation; in such patients, it may be prudent to isolate but not cohort with other infected or colonised patients.

Environmental surfaces are a reservoir for C. auris.³⁸ Like C. parapsilosis, C. auris has been documented to persist on plastic surfaces for up to 28 days in a controlled environment mimicking a healthcare setting.¹⁴ C. auris forms biofilms which may enhance its persistence in the environment.^11,12,13 Guidance for environmental cleaning is not consistent, with variability across the recommendations from several public health agencies.³⁷

Daily cleaning: All surfaces and equipment should be cleaned daily with a neutral detergent and water. Standard cleaning should be followed by wiping surfaces with an appropriate disinfectant. Chlorine-based disinfectants effectively kill C. auris in suspension and inoculated on surfaces.^16,17,34,39 Chlorine disinfectants also kill other multidrug-resistant pathogens such as methicillin-resistant Staphylococcus aureus and carbapenem-resistant Enterobacterales. A sodium-hypochlorite solution (1000 parts per million) is recommended for daily cleaning. While some public health agencies recommend higher concentrations of sodium hypochlorite, there is limited evidence to support this and the guideline development group had concerns about corrosive damage to re-useable equipment and adverse (noxious) effects on personnel working with a concentrated solution.³⁷ New chlorine-based solution should be prepared daily at a minimum and stored away from sunlight and heat to preserve potency. Cleaners should be given clear instructions how to prepare the chlorine solutions, including pictorial depictions of the dilution process. Cleaning should proceed from cleanest to dirtiest areas, for example cleaning patient’s bedside table prior to cleaning the commode. Cleaning supplies, for example mop heads and buckets, should be decontaminated regularly. Adequate contact time should be allowed with the disinfectant (at least 3 min).¹⁶ Frequently touched areas should be cleaned and disinfected more often (at least twice a day). Quaternary ammonium compounds and ethyl alcohol appear to be less effective for environmental disinfection of C. auris and should not be used.^17,37,39 Routine environmental sampling to culture C. auris from patient care areas as a proxy for efficacy of terminal cleaning is not recommended.

Equipment: Single-use equipment is preferred, but if it is not available, dedicated equipment should be used for the duration of the patient’s stay. Equipment should be cleaned thoroughly and disinfected according to the manufacturer’s recommendations. Surfaces of equipment should be cleaned adequately to remove dirt and organic material prior to disinfection; sodium hypochlorite is less effective in the presence of organic material.

Terminal cleaning: Terminal cleaning protocols must be strictly adhered to using checklists which are completed by the IPC team. Terminal cleaning should involve cleaning and disinfection of all items and surfaces in the patient care area or room as well as laundering or changing any difficult-to-clean items, for example curtains and movable partitions. Terminal cleaning or disinfection should begin with removing all disposable items (e.g. suction canisters, glove boxes, tubing and waste) and items intended to be removed and cleaned outside patient care area (e.g. laundry items). All surfaces and equipment should be cleaned with a neutral detergent and water and then wiped with a sodium-hypochlorite solution. Although higher concentrations of this solution have been used for terminal disinfection in outbreaks,³³ we recommend 1000 parts per million. Hydrogen peroxide vapour or wipes appear to be effective against C. auris and may be added as an additional measure after cleaning and disinfection.^16,17,39 There is limited evidence for the use of UV light disinfection for C. auris. A recent study examining the efficacy of UV-C light (254 nm) showed that an exposure time of 20 min was required to destroy C. auris; this was substantially longer than the time required to kill MRSA.⁴⁰ It is important to note that ‘non-touch’ environmental disinfection methods, such as hydrogen peroxide vapour and UV light, cannot replace traditional methods and may only be considered an adjunct to traditional cleaning and contact disinfection of the environment.

Early aggressive treatment of invasive Candida disease is vital for improved outcomes in critically ill adults.⁴¹ In the vast majority of adults with invasive Candida disease (including C. auris), an echinocandin is recommended as first-line treatment.⁴² Amphotericin B deoxycholate is an alternative agent in settings where echinocandins are unavailable. Amphotericin B is also preferred in invasive infections of the central nervous system, eye and urinary tract.⁴³ Although amphotericin B deoxycholate is known to exhibit concentration-dependent killing activity, continuous infusion may be associated with better tolerability and less renal toxicity and may therefore be desirable in those settings where this is possible.⁴⁴ Azole antifungal agents such as fluconazole and voriconazole are not recommended as initial treatment for suspected or confirmed C. auris invasive disease. In many centres, reduced susceptibility or high-level resistance has been demonstrated to these agents.¹⁰ While posaconazole MICs for South African C. auris strains are relatively low (MIC₅₀ of 0.12 mg/L), the first-line use of this agent should only be considered in consultation with an ID specialist or a specialist with a particular interest in this field. Posaconazole is currently only available as an oral formulation in South Africa. Clinicians are advised to check for potential drug–drug interactions and adverse effects when prescribing antifungals. A useful antifungal interactions smartphone application can be accessed at https://www.aspergillus.org.uk/content/antifungal-drug-interactions. Currently available antifungal agents with efficacy against C. auris are shown in Table 5.

Neonates or infants aged < 2 months: For neonates or infants less than 2 months old, amphotericin B deoxycholate should be used as a first-line treatment of invasive infections (Table 6).⁴⁵ Amphotericin B is efficacious and well tolerated in neonates. Fluconazole should not be used for treatment of C. auris; fluconazole also has no activity against azole-resistant strains of C. parapsilosis which are endemic in some South African neonatal units.¹⁸ Echinocandin use should be limited and reserved for cases of salvage therapy or where severe toxicity precludes the use of amphotericin B. There is no evidence for combination antifungal therapy in this age group for the treatment of C. auris.

Children aged > 2 months: Echinocandins are the preferred agents for most cases of candidaemia and invasive candidiasis (Table 7). Exceptions include infections of the central nervous system, eye and urinary tract where amphotericin B deoxycholate should be used. Patients should be closely monitored for treatment failure, as indicated by persistently positive clinical cultures. Switching to amphotericin B should be considered if the patient has persistent fungaemia for > 5 days or is unresponsive to echinocandin treatment. Fluconazole should not be used for treatment of C. auris. No supporting evidence exists for combination antifungal therapy in children.

Candida auris bloodstream infections are usually associated with healthcare settings and occur among patients with intravascular catheters and prosthetic devices. While many of these bloodstream infections represent candidaemia alone, attempts to exclude deep-seated infections, such as infective endocarditis, osteomyelitis, meningitis, pyelonephritis and endophthalmitis (by dilated retinal examination), should be undertaken.^23,46 This will influence treatment duration and penetration of antifungal agents into the source area will need to be considered. In such cases, consultation with an ID specialist (or a specialist with a particular interest in this condition) is recommended. C. auris fungaemia may be difficult to control. Without adequate and appropriate source control, antifungal treatment alone may be futile. All attempts should be made to remove or replace indwelling central venous and arterial devices, as well as urinary catheters. Infected prosthetic material such as heart valves, shunts and bone fixation devices should be surgically removed, where feasible. Any collections should be drained. In addition, risk factors for candidaemia should be modified where possible. A summary of recommended source control and risk factor modification measures is presented in Table 8. In neonates with blood and/or urine cultures positive for C. auris, a lumbar puncture and a dilated retinal examination are recommended. If cultures are persistently positive, imaging of the genitourinary tract, heart, liver and spleen should be performed. Central venous catheter removal is strongly recommended. Surgical intervention should be considered for fungal balls in the kidneys and for endocarditis.⁴²

Blood cultures and laboratory or biochemical markers (including peripheral white cell count [WCC], platelet count and C-reactive protein [CRP]) should be repeated at least three times a week to monitor clearance after candidaemia is confirmed by blood culture.

Blood cultures for initial diagnosis of candidaemia or monitoring clearance of bloodstream infection should be collected using strict aseptic technique. Among adults, each blood culture bottle should be inoculated with at least 10 mL of blood from a peripheral venepuncture site (total volume of a blood culture set: up to 40 mL–60 mL).⁴⁷ Follow-up blood cultures can help to determine the appropriate duration of antifungal therapy. Blood cultures should be repeated at least three times a week in order to document clearance of candidaemia.⁴² Many laboratories routinely perform MIC testing on all invasive Candida strains: MICs of subsequently cultured strains should be closely monitored to identify antifungal resistance which may require treatment modification.²³ In addition, we suggest that markers such as a peripheral WCC, platelet count and CRP should be measured regularly to assist with treatment monitoring and clinical response. Kidney function and electrolytes (especially potassium and magnesium) should be monitored closely, particularly if the patient is being treated with amphotericin B deoxycholate.⁴⁸ Serum procalcitonin levels usually remain between 2.0 ng/mL and 2.5 ng/mL among patients with invasive Candida infections; thus, procalcitonin is not a useful marker for monitoring response to treatment.⁴⁹ A negative serum (1,3) beta-D-glucan (BDG) level may be a useful adjunct to exclude a diagnosis of candidaemia in critically ill adults.^42,50,51 There are no published data on the utility of serum BDG for initial diagnosis of invasive C. auris infection. A decrease in serially collected serum BDG levels during treatment for candidaemia is associated with clinical/microbiological resolution.^52,53 However, no recommendation can be made on the use of serum BDG for monitoring response to C. auris infection because no data are currently available.

If no evidence of a deep-seated fungal infection is found (e.g. infective endocarditis, meningitis, osteomyelitis, pyelonephritis, endophthalmitis or prosthetic infection) and disease is thus considered uncomplicated, antifungals are recommended to be continued for a minimum period of 2 weeks from the date of clearance of the candidaemia, as documented by negative blood cultures, in conjunction with clinical resolution.⁴² Treatment of deep-seated or complicated infections is usually prolonged and should be in consultation with an ID specialist.

Although there is currently no evidence for combination therapy in any patient population with invasive C. auris disease, crude (unadjusted) mortality is unacceptably high,⁵⁴ especially among critically ill and immunosuppressed patients. We therefore recommend initial combination therapy in the sub-groups mentioned above, along with prompt source control. Where initial combination antifungal therapy is commenced among patients in septic shock (defined as a mean arterial blood pressure [MABP] ≤ 65 mmHg or requiring vasopressor support and lactate > 2 mmol/L⁵⁵), daily evaluation for the ongoing requirement of combination therapy should be reviewed while awaiting antifungal susceptibility results and/or clinical stabilisation of the patient. Following susceptibility testing results, de-escalation to a single antifungal agent to which the pathogen is susceptible should be considered, provided that the patient has clinical and laboratory improvement and has undergone adequate, appropriate source control measures. This should happen within a 72-h time frame. Combination therapy may be considered among patients who remain blood culture positive after 5–7 days (defined as persistent fungaemia) despite attempts at suitable source control, appropriate antifungal dosing and optimised antifungal penetration to the site of infection; isolate MICs should be reviewed by a clinical microbiologist. Patients who become culture positive following completion of initial antifungal treatment and presumed clearance of infection (defined as recurrent fungaemia), as well as patients who become culture positive after a period of negative cultures while still receiving appropriate treatment (defined as relapsing fungaemia), may also be considered for combination therapy, as well as detailed further investigations. In all patients, appropriate antifungal dosing and source control are of paramount importance. Treatment of these complex patients is recommended to be continued in consultation with an ID specialist and clinical microbiologist.

Isolation of C. auris from a non-sterile body site (such as skin, rectum, upper or lower respiratory tract or urinary tract) in the absence of markers of inflammation or organ dysfunction and clinical signs of infection is usually an indication of colonisation and not disease. In this setting, antifungal treatment should be avoided; however, colonisation may prompt removal of indwelling devices (such as urinary catheters) and institution of appropriate IPC measures (refer to Section 3). In the presence of clinical signs of infection, attempts to isolate C. auris from a sterile site (such as blood, CSF, tissue, central venous catheters, etc.) should be made. Ancillary markers of fungaemia such as a serum BDG assay may be useful to exclude cases of candidaemia (this assay has excellent negative predictive value [NPV] among critically ill adults).^42,50

Antifungal prophylaxis among non-neutropenic critically ill patients remains controversial, including among surgical patients with severe acute pancreatitis.^56,57 While fluconazole prophylaxis may reduce the incidence of invasive candidiasis in critically ill adults and neonates, emergence of resistance in Candida species other than C. albicans is a concern with universal prophylaxis in this high-risk population. Previous exposure to antifungals is associated with a shift in Candida species distribution and an upward antifungal MIC ‘creep’.⁵⁸ In addition, the threat of emergence of cross-resistance to both triazoles and echinocandins exists, as described in Candida glabrata, a species which notoriously sequentially acquires and expresses multiple resistance genes.⁵⁹ The dominance of triazole-resistant C. parapsilosis causing bloodstream infections in South Africa was recently confirmed, particularly in ICU patients in the private sector.¹⁸ Overuse of triazoles for prophylaxis and treatment of candidaemia and other fungal infections may have led to the emergence and subsequent nosocomial transmission of these triazole-resistant strains. Similar factors may apply to C. auris in South Africa.⁹ The epidemiology of candidaemia in South Africa is unusual: C. albicans and C. parapsilosis dominate in the public and private sectors, respectively.¹⁸ Multidisciplinary AFS teams should choose prophylactic agents based on local surveillance data. The recommended antifungal options and doses for prophylaxis in adults and children are summarised in Table 9.^42,60 However, the optimal duration of prophylactic treatment is not known.⁶¹

There is insufficient evidence to make a firm recommendation on the optimal strategy to identify patients who may benefit from early antifungal treatment.

From a clinical point of view, early diagnosis and treatment of invasive candidiasis is the key to reduction in mortality. To minimise the negative impact of this infection, several management strategies had previously been described: antifungal prophylaxis, empirical therapy, pre-emptive therapy and directed culture-based treatment. However, both universal antifungal prophylaxis and empirical therapy (based on the persistence of fever non-responsive to antibacterial agents and a combination of risk factors) may overexpose the patients to antifungal treatment, potentially increasing antifungal resistance.⁶² Notably, up to 70% of critically ill patients receive systemic antifungal therapy although they have no documented invasive fungal infection,⁶³ suggesting a need for alternative strategies. With the use of biomarkers such as the serum BDG assay and to simplify auditing of AFS process measures, the concepts of pre-emptive or empirical therapy should be substituted by ‘early’ antifungal treatment. Identifying patients at risk for invasive candidiasis includes recognition of a combination of risk factors. The Candida score was developed for critically ill non-neutropenic adults in Spanish intensive care units (ICUs) and is calculated by adding the following scores for each risk factor that is present: 1 (total parenteral nutrition), 1 (surgery), 1 (multifocal Candida species colonisation) and 2 (severe sepsis).⁶⁴ Such predictive scores can help distinguish Candida colonisation and invasive candidiasis in ICUs, permit selection of high-risk patients who may benefit from early antifungal therapy and can also be used by AFS teams.⁶⁵ However, given the low positive predictive values (PPVs) of such scores, many prescribed antifungal regimens have been shown to be unnecessary.⁶⁶ In contrast, predictive scores have far better NPVs.⁶⁷

Studies using non-culture-based assays, particularly serum BDG, together with a Candida score, have aided in establishing whether initiation of antifungal therapy in at-risk patients followed by close follow-up and discontinuation of antifungal therapy when invasive candidiasis is excluded has an impact on the outcomes of ICU patients. Combining BDG and the Candida score improves the sensitivity and NPV compared with either serum BDG or the Candida score alone.⁶³ Using this approach, antifungal therapy was safely avoided in 73% of treatment-eligible ICU patients and treatment duration was shortened in another 20% of patients.⁶⁸ In another cohort, early discontinuation of antifungal therapy (initiated in high-risk ICU patients following a positive Candida score ≥ 3) based on two consecutive negative serum BDG tests appeared to be a reasonable AFS strategy such that the combined assay is potentially usable and safe for the therapeutic decision-making process and discontinuing of early antifungal therapy.⁶⁹ Similar outcomes were observed in a biomarker-based strategy using an algorithm involving serum BDG, mannan and anti-mannan assays.⁷⁰ A recent study also aimed to assess the combined performance of serum BDG and procalcitonin to differentiate between invasive candidiasis and bacteraemia.⁷¹ When both markers indicated invasive candidiasis (BDG ≥ 80 pg/mL and procalcitonin < 2 ng/mL), they had a higher PPV (96%) compared to 79% and 66% for BDG or procalcitonin alone, respectively. When both markers indicated bacteraemia (BDG < 80 pg/mL and procalcitonin ≥ 2 ng/mL), the NPV for invasive candidiasis was similar to that of BDG used alone (95% vs. 93%). The combined use of procalcitonin (PCT) and β-D-glucan (BDG) could therefore be helpful in the diagnostic workflow for critically ill patients with suspected candidaemia. The data suggest that the concurrent use of the Candida score, BDG and other biomarkers may improve diagnostic stewardship in ICU patients at risk for Candida sepsis, but additional investigations are needed and their use as AFS tools remains to be established. In addition, the negative BDG cut-off < 80 pg/mL for C. auris and Candida species other than C. albicans in South Africa needs to be confirmed.

No specific AFS programmes focusing on C. auris have yet been designed, but it is likely that an environment with high and inappropriate antifungal utilisation will favour the emergence of multidrug-resistant fungi. Changes in the distribution of Candida species may have impact on treatment recommendations because of differences in susceptibility to antifungal agents among species, but previous exposure to antifungal agents has likely contributed to this shift in species distribution.⁶² Inappropriate use, as opposed to overuse, also needs to be considered. This was highlighted in a bedside audit of antifungal use in patients admitted to a general hospital, where 57% of the prescriptions were found to be sub-optimal.⁷² Reasons for inappropriate use included inappropriate choice, dosing, de-escalation and duration of treatment. While an overall reduction in antifungal consumption is necessary, using the correct agent at the correct dose for the correct duration is also important. In support of this, a 3-year comprehensive AFS programme not only resulted in improved overall utilisation but also a significant decrease in fluconazole consumption (from 242 to 117 defined daily doses [DDDs] per 1000 patient-days) which was associated with a significant reduction in the incidence of C. glabrata and C. krusei.^61,73 Therefore, to reduce overall consumption, enhance appropriate use of antifungal therapy and improve patient outcomes while minimising the risk of emergence of resistance, the implementation of an AFS programme is recommended in all South African hospitals.

Multidisciplinary teams encompassing the necessary expertise (in pharmacy, clinical microbiology, infectious diseases, internal medicine, surgery, paediatrics and anaesthetics) are an international recommendation for AFS.^74,75 Given the lack of ID human resources in most South African hospitals, utilising existing multidisciplinary resources in a collaborative manner may enable an AFS programme to be embedded in routine practice.

Depending on resources, circumstances and the health sector in South Africa, restrictive stewardship interventions (such as formulary restriction, prior authorisation, therapeutic substitutions and automatic stop orders), structural interventions (such as changing from paper to computerised records, rapid laboratory testing, therapeutic drug monitoring, computerised decision support systems and the introduction of quality monitoring mechanisms), persuasive strategies (such as distribution of educational materials, educational meetings and outreach visits, local consensus processes, reminders provided verbally, on paper or on computer) and prospective audit, intervention and feedback should be considered.⁷⁶ However, prospective audit, intervention and feedback has been shown to be a very effective and safe antibiotic stewardship strategy in South African hospitals, particularly in settings without ID specialists.⁷⁴ Potential multi-component AFS process and outcome measures for clinician, pharmacist and/or ICU nurse audits are proposed in Table 10.

Antifungal stewardship process measures (Table 10) should preferably be audited as an ‘AFS bundle’, which is defined as a small set of evidence-based interventions for a defined patient population and care setting. In contrast to check lists, adherence to bundle components is measured using an all-or-nothing measurement, with a goal of ≥ 95%. As mentioned, the first step in the development and implementation of AFS is to build a multidisciplinary team.^74,75 Using AFS bundles and all-or-nothing measurement may change the way care is provided for at-risk patients in important ways because bundles not only facilitate but also promote awareness that the entire care team must work together in a system designed for reliability.

The beneficial impact of ‘bundles’ on clinical outcomes in patients with invasive candidiasis was confirmed for the first time recently.⁷⁷ The composite adherence to nine measures (all-or-nothing) was only 6.9% in a Japanese study, but there was a significant difference in clinical success between patients with and without adherence (92.9% vs. 75.8%). When step-down oral therapy was excluded from the measures, adherence to the bundles was shown to be an independent predictor of clinical success (odds ratio [OR] 4.42, 95% confidence interval [CI] 2.05–9.52) and mortality (OR 0.27, 95% CI 0.13–0.57). Notably in none of the studies in Appendix 1, Table 1-A1, where the impact of various AFS interventions for invasive candidiasis in a variety of settings including non-academic hospitals has been summarised, were patient outcome measures negatively affected. This included length of stay, re-admissions, length of hospitalisation, time until clearance of candidaemia, persistent candidaemia, recurrent candidaemia, triazole-resistant Candida species other than C. albicans and mortality compared to the pre-implementation phase.