Acute undifferentiated febrile illness (AUFI) is a common presentation of tropical infections. Dengue is hyperendemic in tropical and subtropical countries, predominantly in urban and semi-urban areas. Dengue infection is the most common AUFI etiology in Bangkok, Thailand, accounting for 39.6% of non-malarial febrile cases from 2013 to 2015. Despite the global malaria recession, it remains the most common AUFI diagnosis in various countries, particularly in Sub-Saharan Africa, but was only observed at 1% in Southeast Asia. Distinguishing between tropical infection, such as dengue or malaria, and bacterial infection at the early presentation of AUFI is difficult. The non-specificity of symptoms and signs and lack of availability of diagnostic tests often result in irrational antibiotic use. Following the overuse of antibiotics, antimicrobial resistance (AMR) has emerged as a serious global public health threat.
1. Acute Undifferentiated Febrile Illness
1.1. Causes
Acute undifferentiated febrile illness (AUFI) is a common presentation of tropical infectious diseases or bacterial infections without organ specific infection. This presentation is caused by a diverse range of pathogens and can vary between regions. A review in Southeast and South Asia found that viral etiologies were most common, followed by bacterial. Of the viral causes, dengue fever was the most common, accounting for 11.8% of cases, followed by leptospirosis, typhoid, scrub typhus and influenza
[1]. The multinational multicenter cross-sectional study of community-acquired sepsis and severe sepsis among children and adults in Southeast Asia showed that dengue infections, followed by leptospirosis and rickettsiosis, were the most common causative pathogens for sepsis in this area
[2]. As dengue infection is caused by a virus, antibiotics are not the key treatment.
Specific rapid diagnostic tests (RDTs) are required in malaria and dengue, both of which do not need empirical antibiotic therapy. Furthermore, this strategy may directly affect patients and indirectly introduce selective pressure, leading to antimicrobial resistance (AMR). The lack of a decision algorithm following negative RDTs in healthcare facilities can increase unnecessary antimicrobial use
[3].
1.2. Antibiotic Use in AUFI
Monitoring antibiotic use in AUFI is important to improve prescribing practices in the future. Previous literature on the types of antibiotics used in both malaria- and non-malaria- endemic regions was reviewed.
A study conducted in Pune, India explored the use of antibiotics in suspected and diagnosed mosquito-borne illnesses among children and adults with acute febrile illnesses. Robinson et al. demonstrated, that on admission, 82% of adults and 94% of children received empirical antibiotics
[4], i.e., the vast majority in both cases. Only 6% of patients were diagnosed with culture-positive bacterial infection, suggesting that a large proportion of antibiotic usage was unnecessary. The high rates of empirical antibiotic prescriptions were attributed to diagnostic uncertainty, even when suspicion of non-bacterial illness prevailed. Third generation cephalosporins were the most frequently prescribed antibiotics, administered to 52% of participants, followed by amoxicillin/clavulanate (28%) and macrolides (20%).
Anand Paramadhas et al. conducted a point prevalence study on antimicrobial use in public and private hospitals in Botswana, where human immunodeficiency virus (HIV) infections, tuberculosis (TB), and malaria were endemic. Cefotaxime followed by parenteral metronidazole were the most commonly prescribed antimicrobial agents in all public hospitals, whereas ceftriaxone was the second highest antimicrobial prescribed in specialist private hospitals
[5]. Of 711 patients, approximately 70% had a documented bacterial infection, most commonly pneumonia, skin and soft-tissue infections, intra-abdominal infections, and urinary tract infections. However, malaria was not a potential risk factor influencing antibiotic prescription. Okoth et al. found that antibiotics were frequently used in AUFI, at 17%. The most prescribed antibiotics are third-generation cephalosporins, imidazole derivatives and broad-spectrum penicillin
[6].
1.3. Challenges in Diagnosis of AUFI Aetiology
Distinguishing AUFI etiologies on presentation has been difficult, particularly when the mainstay of diagnosis is clinical history and examinations. The non-specificity of symptoms and signs and the lack of availability of diagnostic tests often lead to empirical and irrational use of antibiotics
[7][8][9]. Nevertheless, some high-income countries (HICs) with advanced and rapid diagnostic investigations did not limit antibiotics prescriptions. High and inappropriate use of antibiotics was found in both HICs and low- and middle-income countries (LMICs)
[10][11][12][13], mostly driven by high rates of HIV, TB and malaria among patients
[5]. Rickettsiosis is the best example for this issue because of its diagnostic difficulty; thus, empirical treatment of doxycycline should be considered to prevent complications, particularly in some resource-limited hospitals
[14]. Collateral damage of doxycycline, as well as cephalosporins or fluoroquinolones, in AMR is not well recognised. However, tetracyclines have been reported to cause vaginal flora suppression by an unknown mechanism and
Clostridioides difficile diarrhea
[15]. COVID-19 also needs to be included in AUFI etiologies, since respiratory symptoms may be absent in the early phase
[16]. However, symptoms probably vary in different variants.
In addition, several limitations have been reported in various diagnostic tests; for example, blood culture sensitivity for
Salmonella spp. ranged between 40% and 80%, and in dengue or rickettsiosis it is impossible to obtain both acute and convalescent sera, which results in delayed diagnosis
[17]. Improved diagnosis of AUFI etiologies could help reduce inappropriate antibiotic prescriptions and improve patient outcomes
[18].
The aforementioned limitation of diagnostic tests remains problematic. To help clinicians in their decision, biomarkers such as C-reactive protein (CRP) and procalcitonin play some role in distinguishing bacterial and viral infections
[19]. In a study conducted in Northern Thailand, Wangrangsimakul et al. demonstrated that low CRP and white blood cell count were significant predictors of viral infection and that CRP was highly sensitive and specific for bacterial infections when comparing bacterial and viral groups. Moreover, a high procalcitonin level was sensitive in detecting bacterial infection, whereas low levels were not specific to viral infections
[20]. If laboratory results are effective in ruling out a bacterial diagnosis, this can reduce antibiotic prescription and help combat the issue of AMR.
1.4. Antimicrobial Use in Malaria and Dengue
The treatment for malaria involves the use of antimalarials rather than antibiotics. Like antibiotics, inappropriate use of antimalarials can lead to resistance. The barriers to definite diagnosis can result in misdiagnosis and subsequently overtreatment, as displayed in Leslie et al.’s observational study in Afghanistan, where 413 of 414 patients had a negative malaria smear but 412 (99%) were prescribed an antimalarial drug
[21]. The WHO 2010 guidelines implemented malaria testing at pre-treatment, which has reduced antimalarial use, but at the expense of increasing antibiotic use
[18]. D’Acremont et al. conducted a study in Tanzania, which found that the introduction of RDTs reduced antimalaria prescriptions by 23% but increased antibiotic use by 23%
[22]. Although the WHO guidelines were implemented, antimalarials are still administered without diagnosis confirmation, resulting in unnecessary antimalarial selective pressure
[23]. Presumptive treatment is especially prevalent in high malaria transmission rates, with lack of laboratory expertise and without availability of RDTs
[24].
In a different study in rural Tanzania, Njozi et al. found that predictors associated with the risk of antibiotic co-prescription with antimalarials were age groups and types of diagnosis
[24]. Children aged < 5 years were more likely to be co-prescribed antibiotics, probably because they are a higher-risk group. As bacterial co-infection cannot be confidently diagnosed with bedside examination, antibiotics are currently prescribed in addition to antimalarials in children with severe malaria
[25]. The type of diagnosis herein
[24] referred to RDT, whether it was positive, negative or not tested. Co-prescription was observed more frequently in patients with a negative malaria test or in those who were not tested.
Another factor to consider in malaria resistance is the implementation of over-the-counter antimalarial prescriptions in some countries, such as Kenya, which eventually result in overuse, as highlighted by its popularity even in areas with low and seasonal malaria transmission
[26]. Abuya et al. suggested an impact of underdosing on the development of malarial resistance when over-the-counter drugs are prescribed.
1.5. Bacterial Co-Infection in Tropical Infectious Diseases
Antibiotic use in dengue or malaria would be appropriate for bacterial co-infection. The incidence of bacterial co-infection in dengue appears to be low, although the exact numbers have not been extensively elucidated. Studies report the incidence to be as low as 7% and as high as 25%
[27][28]. Concurrent infections among dengue cases in Bangkok were found to be 7.8%; nevertheless, bacterial co-infection requiring antibiotic therapy was only 5.3%
[29]. In a study conducted by Sunil et al. in North India, among the 124 pediatric patients with dengue fever, 10.4% had concurrent bacterial sepsis
[30]. Adrizain et al. conducted a retrospective study in Indonesia and found that 17.8% of patients who tested positive for the dengue virus received antibiotics for presumed concurrent upper respiratory tract infection, typhoid fever or urinary tract infection
[31]. After diagnostic testing for these presumed concurrent bacterial infections, they concluded that both the indication and choice of antibiotic in hospitalised patients with dengue were inappropriate in most patients. Kay C et al. constructed a diagnostic model (Dengue Dual Infection Score (DDISI)) for bacterial co-infection in dengue patients. The DDIS was created by five parameters for bacterial coinfection: pulse rate ≥ 90 beats/minute, total white cell count ≥ 6 × 10
9/L, hematocrit < 40%, serum sodium < 135 mmol/L, and serum urea ≥ 5 mmol/L. The DDIS showed the satisfied validation set area and good specificity to differentiate dengue patients for empirical antibiotics
[32].
Bacterial co-infection in malaria can result in higher morbidity and mortality and most commonly occur in falciparum malaria. Although both antibiotics and antimalarials are recommended in children with severe malaria, the aforementioned guidelines suggest that adults with severe malaria should not receive empirical antibiotics
[25]. Bacterial and malarial co-infection rates varied among the literature in Figure 1, from 1.07% in a study from Vietnam by Phu et al.
[33] to 14.9% in a study from Myanmar by Aung et al.
[34] Antibiotic overuse in malaria-endemic countries should be cautiously considered, as Nyein et al. showed that nearly half of the bacterial isolates in malaria cases were resistant to empirical antibiotic treatment
[35].
2. Overview of Antimicrobial Stewardship (AMS)
Although RDTs were generally used for the early diagnosis of dengue and malaria, several limitations have been recognised. For example, the capability of malarial RDTs for parasite detection and the intervention of temperature and humidity to the test accuracy had been observed
[36][37]. Dengue RDTs provided high sensitivity and specificity, whereas the cross-reactivity and past infection detection are still controversial
[38]. AMS is an additive strategy to enhance appropriate antimicrobial use in tropical diseases.
AMS refers to a coordinated programme that promotes the appropriate use of antimicrobials to improve patient outcomes, reduce AMR and decrease the spread of infections caused by multidrug-resistant organisms
[39]. This term has many current definitions, but they broadly employ a coherent set of actions that promote the responsible use of antimicrobials
[40]. An example of activities to describe AMS is the selection of the most appropriate antibiotic, duration, dose and route of administration for a patient with confirmed or suspected infection
[41].
AMS can vary in low- and middle-income countries (LMICs) and HICs. Initially, most research and evidence of AMS was restricted mostly to HICs but has now been redefined on a global scale
[42]. LMICs, where high rates of antibiotic resistance are reported, are facing several challenges and barriers
[43]. Limited resources create diagnostic challenges where correct identification of pathogens and susceptibility testing are not performed before initiating antibiotic therapy. LMICs may also face the challenge of limited access to quality-assured antibiotics or, conversely, the widespread use of non-prescribed over-the-counter antibiotics. In contrast, a study by Ashiru-Oredope et al. in England demonstrated the rapid implementation of AMS programmes in both primary and secondary care settings. They experienced high response rates from trusts across all geographical regions
[44], whereas Kpokiri et al. found that strategies to combat AMR in low-resource settings are yet to be successfully implemented
[45].
AMS programmes require support and collaboration from all healthcare workers. A knowledge and perception study by Adegbite et al. in Gabon found that 64% of prescribers reported awareness that AMR is an issue in their country; however, only 30% considered it an issue in their health facilities
[46]. This could be explained by the lack of regular training on antibiotic prescription, which in turn can lead to overuse and eventually resistance. Rational antibiotic prescription can be achieved by altering the prescriber’s attitude and perception.