Aetiology and clinical spectrum of acute undifferentiated febrile illness in hospitalized children

Introduction:Acute fever of 2 weeks duration with non-specific signs and symptoms is known as acute undifferentiated febrile illness (AUFI). Owing to non-specific presentation, it remains a diagnostic challenge. Hence the present study focuses on etiology and clinical profile of undifferentiated febrile illness. Methodology: All children aged 2-12 years admitted with fever of 5-15 days duration, for which no cause was found after a thorough history and clinical examination were included from July 2015 to June 2016 prospectively. History, examination findings, investigations and the treatment details were recorded. Data analyzed using SSPS software. Results: Total sample size was 263. The mean age was 6.7 ± 3.4 years. Most common symptoms were cough and vomiting. The most common diagnosis arrived at was Scrub typhus (22.4%) followed by Dengue (11%), Enteric fever (11%), Coinfections (6.1%), Urinary tract infections (3%) and Lower respiratory tract infections (2.3%). Fever was still undiagnosed in 116 children (44.1%). No malarial infection was noted. Conclusion: Non-malarial infections are common in this part of the country in children with AUFI.


Introduction
Fever whether low or high grade is a distressing symptom for parents. It is the most common presentation in children attending the pediatric outpatient service and in children with infectious disease. Acute fever of 2 weeks duration with nonspecific signs and symptoms is known as acute undifferentiated febrile illness (AUFI) [1]. This is distinguished from pyrexia of unknown origin which refers to fever of at least 3 weeks for which no cause is identified even after investigation [2]. Since the presentation is non-specific, AUFIs remain a diagnostic challenge unless information of the regional etiological pattern of the AUFIs is known. It can then serve as a clinical and therapeutic guide. Otherwise, treatment may get delayed and contribute to mortality in the child. In developed countries, viral fever is reportedly the most common cause for AUFI in children [3]. In South Asian Inclusion criteria: All children aged 2-12 years admitted with fever 38 0 Celsius or more, of 5-15 days duration with no cause found after a thorough history and clinical examination.
Exclusion criteria:Children with fever less than 5 days, those with fever more than 15 days, children with clear localizing signs and critically ill children and infants.
Sampling Methods: All children satisfying inclusion criteria were included.

Methods
Detailed history was recorded concerning fever pattern, duration and other presenting complaints after informed consent from parents.
History pertaining to respiratory, cardio-vascular, gastro-intestinal and central nervous system, development and immunization was recorded. Past history of infections, contact with tuberculosis, family history and history of travel was noted. Detailed general and systemic examination findings were recorded. Initial investigations recorded were complete blood count, peripheral smear, urine analysis, smear and rapid diagnostic test (RDT) for malarial parasite, urine culture and blood culture. Further investigations included chest X-ray, Ultra sonogram (USG) of abdomen, liver function test (LFT), renal function test (RFT) and serological analysis including blood Widal, Dengue Ns1Ag detection/IgM Elisa and Scrub typhus IgM Elisa which were done as per clinical discretion. Other serological tests were not done in the hospital.
Treatment details pertaining to antimicrobial use was recorded. The peak temperature, time for defervescence and duration of hospital stay were noted.
Statistical analyses-SSPS software was used for analyzing data collected. Mean and standard deviation was used for continuous variables which were compared using t test. Proportions were used for categorical variables which were compared using Chi test.
Univariate and multivariate logistic regression was done to find significant association of variables in common etiological patterns identified. p value less than 0.05 was considered significant.
Clinical features-Mean duration of fever for all the children was 8.3 ± 3.6 days. Mean peak temperature was 39.2 ± 0.9 0 C and the maximum recorded was 44 0 C. Fever was intermittent in 211 children (80.2%) and continuous in 52 (19.8%). Fever was associated with chills in 122 children (46.4%) and rigors in 107 (40.7%).

Diagnosis-
The pattern of acute undifferentiated fever is shown in Figure 1.

Gender, age group and regional differences
There was no gender difference noted for any of the etiological patterns (p>0.05). More number of female children presented with chills (χ²3.944, p=0.047).
The clinico-lab profile in different etiologies of AUFI is shown in    After univariate and multivariate logistic regression, local residence, abdominal pain and conjunctival congestion were significantly higher in scrub typhus (Table 4). Leucopenia and thrombocytopenia were significantly higher in Dengue (Table 4).

Outcome-
The mean duration of fever after hospital admission was 3.1±2.3 days. The mean hospital stay was 6.7±3.1days and was longest in LRTI (9.7±5.5 days) and shortest in undifferentiated illness (5.2±2.5 days). No mortality was recorded in present study.

Discussion
Final diagnosis was possible in 147 children (55.9%). The most common diagnosis was scrub typhus (22.4%). This is similar to finding reported by Kashinkunti in Karnataka [1]. However, regional patterns vary. In Chennai, enteric fever was identified as most common specific etiology in children [7]. All these children diagnosed with Scrub typhus in this study did not have eschar.
Various studies have reported presence of eschar as ranging from 11-44% only [8,9]. Therefore, high index of suspicion is necessary to make a prompt diagnosis of Scrub typhus in the absence of eschar. Serology is useful in this regard to confirm the clinical suspicion. This is vital as early treatment reduces morbidity and mortality associated with Scrub typhus.
After Scrub typhus, Enteric fever (11%) and Dengue (11%) were the common specific diagnosis reported in our study. Prevalence in other studies range from 5-26% [5,7]. None ofthe children in our study had received typhoid vaccination. However, the low prevalence may be related to absence of specific reliable serological test and low yield of blood culture. Though Widal is only supportive investigation, eosinopenia was a valuable guide in diagnosis. Typhidot which is considered superior to Widal was not available in the hospital.
Prevalence of Dengue in AUFI ranged from 10-25% in adults [1,10]. A lower proportion was reported in children in Chennai [7]. Thus, there is wide regional variation in the prevalence. The common infections noted in AUFI are malaria and non-malarial infections such as Scrub typhus, Enteric fever, Dengue and Leptospirosis in South India [1,7,10], North India [11] and even other Asian countries like Cambodia and Vietnam [5,12].
It is noteworthy that no case of malaria was reported in our study during this time period though mosquito-borne Dengue was noted in 11% of the children. It may be because of different mosquito species and different breeding habits of the mosquitoes involved.Respiratory tract infections and urinary tract infections were other identifiable causes in our study, similar to that reported by Shamikumar [7]. Co-infectionswere observed in 6.1% comparable to 7.6% reported by Abhilashet al [6]. It is highly probable that these might not be true co-infections but only serological cross-reactivity. However, true co-infection have been reported [13,14].
Cough and vomiting were the two principal symptoms with which children with AUFI presented, similar to that reported by Shamikumar [7]. In their study, vomiting and cough were predominant in that order. Abdominal pain and conjunctival suffusion were significantly associated with scrub typhus in our study while Abhilash observed breathlessness as significant in adults [6]. Lab finding of leukopenia and thrombocytopenia was significantly associated with Dengue similar to that reported by Gopalakrishnan in adults [10].
Undiagnosed fever was seen in 44.1% in our study. Similar proportion was reported by other studies ranging from 50-60% [7,15]. Higher proportion of undiagnosed cases can be explained by the limited diagnostic facilities available.
Most of the undiagnosed fever in our study were treated as viral fever and was indeed self-limiting. Shamikumar also reported a similar observation where non-specific viral illness and URI formed the major category of children with AUFI [7]. In our study, a small number of them, received empirical antibiotics (16.4%) and empirical Chloroquine (2.6%). Overall, empirical antibiotics were given in 26 children (9.9%) and empirical chloroquine in 5 (1.9%). Observation from our study emphasizes that empirical antimicrobials can well be avoided in children presenting with AUFI as most of them are self-limiting.

Conclusion
Non-malarial infections are common causes of AUFI in children in this part of the country. The most common specific infection was Scrub typhus, followed by Dengue and Enteric fever. Proportion of undiagnosed fever, most of which were presumed non-specific viral fever, still remain high owing to limited diagnostic facilities. Due to high prevalence of Scrub typhus in this area, a thorough search for eschar should be made in all patients of AUFI. Empirical Chloroquine can be safely avoided unless tested positive for malaria. Similarly, empirical antibiotics can be avoided in children presenting with AUFI.
Recommendations-All children should receive Typhoid vaccination. Dengue control and mite control measures should be undertaken to curb the same.
What this study adds-Non-malarial infections are presently the leading causes of AUFI in children in this region.Empirical chloroquine can be avoided. As most of the AUFI are self-limiting, even empirical antimicrobials can be avoided.