Bmc Infectious Diseases Identification and Validation of Clinical Predictors for the Risk of Neurological Involvement in Children with Hand, Foot, and Mouth Disease in Sarawak

Background: Human enterovirus 71 (HEV71) can cause Hand, foot, and mouth disease (HFMD) with neurological complications, which may rapidly progress to fulminant cardiorespiratory failure, and death. Early recognition of children at risk is the key to reduce acute mortality and morbidity.


Background
Hand, foot, and mouth disease (HFMD) is a common childhood exanthema caused by species A human enteroviruses (HEVA), particularly Coxsackievirus A16 (CVA16) [1]. In most instances, this is a mild self-limiting illness. The affected children are often given out-patient care with symptomatic treatment. However over the last decade HFMD has emerged as a growing public health problem in Asia following frequent outbreaks of deathassociated HFMD caused by a more virulence member of HEVA, human enterovirus 71 (HEV71), in a number of countries in the region [2][3][4][5]. This was first recognized with large outbreaks of HFMD associated with neurological disease and alarming fatalities in Sarawak, Malaysia in 1997 and in Taiwan in 1998 [2,3]. Fatal cases typically presented with a brief duration of febrile illness, subtle neurological signs and died dramatically of acute refractory cardiac dysfunction and fulminant pulmonary oedema within hours of developing signs of tachycardia, poor peripheral perfusion and tachypnea. Indeed, most of them died shortly after hospital admission, and some even before or on arrival at hospital [2,[6][7][8]. Although severe neurological complications and death only occur in a small minority of children with HFMD, the fulminant disease course of the fatal cases has caused great public alarm in Asia. Experience from recent outbreaks of HEV71 associated HFMD (HEV71-HFMD) in Asia showed that primary care doctors are often overwhelmed with large number of children with HFMD seeking medical attention for the fear of neurological complications and death. Because of the risk of sudden death, coupled with tremendous parental pressure to admit children with HFMD into hospital for observation, children with HFMD are often routinely admitted into hospital for observation in Sarawak, which has imposed a huge burden on the healthcare system. Cerebrospinal fluid (CSF) pleocytosis has so far been the universal finding in fatal cases even though many have no obvious neurological signs prior to sudden onset of cardiorespiratory failure and death [2,6,8]. In the absence of clear neurological sign, CSF pleocytosis (indicative of neurological involvement) has thus been considered an objective marker of complicated disease, allowing clinicians to focus their attention and provide timely intervention in these patients before they develop fatal cardiorespiratory failure. We therefore examined data collected through a prospective study of HFMD to identify and validate risk factors associated with neurological involvement in children with HFMD that may be used by clinicians managing children with HFMD.

Setting and study period
A prospective clinical study was conducted from January 2000 through December 2006, which included 3 distinct outbreaks that occurred in 2000 /1, 2003 and 2006, at the paediatric wards and intensive care unit at Sibu Hospital (Sarawak, Malaysia). The study was approved by the Director of Health for Sarawak and the Ethics Committee of the Liverpool School of Tropical Medicine (UK). Informed consent was obtained verbally from each child's accompanying parent or guardian. Figure 1 shows the algorithm of the investigation and the classification of the disease severity of children with HFMD in the study [9]. A child was defined as having HFMD if they had new onset of at least one (≥ 1) of the following: maculopapular or vesicular rash on the palms and/or soles; vesicles or ulcers in the mouth or herpangina (defined as multiple oral ulcers predominantly affecting the posterior parts of the oral cavity). Children with HFMD were considered to have more serious illness if they have the following features: a history of fever, or fever on examination (≥ 38°C), and ≥ 1 of the following features indicative of more serious illness: toxic and ill in appearance, recurrent vomiting (at least twice), tachycardia (heart rate ≥ 150/min) breathlessness, poor perfusion (cold clammy skin), reduced consciousness (irritability, lethargy, drowsiness, coma), limb weakness, meningism (neck stiffness or positive Kernig's sign), seizures. They were subjected to CSF examination after written consent to exclude central nervous system (CNS) involvement. Children with > 5 cells/μL (i.e. CSF pleocytosis) and negative microscopy and culture for bacteria were classified as "HFMD with CNS complications" (HFMD-CNS), while those with normal CSF examination were considered to have "severe HFMD without CNS involvement" (HFMD-Non-CNS). Children with HFMD-CNS were diagnosed to have aseptic meningitis (ASM) if they were fully conscious, had headache, meningism, and no focal neurological signs. Encephalitis was defined by the presence of impaired consciousness including lethargy, drowsiness or coma, seizures or myoclonus. Acute flaccid paralysis (AFP) was characterized by the acute onset of areflexic limb weakness. Cardiorespiratory failure was defined by the presence of tachycardia, respiratory distress, pulmonary oedema, poor peripheral perfusion requiring inotropes, pulmonary congestion on chest radiography and reduced cardiac contractility on echocardiography. Children without features of more serious illness were classified as "mild HFMD", and were observed in hospital until they became afebrile for at least 12-24 hours. A child was considered to be positive for HEV71 if HEV71 was isolated by tissue culture or HEV71 RNA was detected by HEV71 specific RT-PCR from ≥ 1 clinical sample.

Clinical methods
All children with HFMD admitted into the hospital were assessed by pediatricians of the study team. A detailed history and clinical examination was performed with special

NO YES
attention to mucocutaneous lesions, cardiovascular and neurological signs. All details were recorded on standardized forms. Swabs were taken from the throat and rectum of every patient, as well as ≥ 1 swab from vesicles on the skin and oral ulcers (if present). The clinical samples were stored immediately in a -70°C freezer until further testing. Blood was taken for flavivirus serology, and in patients with suspected CNS involvement for full blood count, urea, electrolytes, and glucose. Electrocardiogram and echocardiogram was also performed on children with suspected CNS involvement. CSF was examined for cell count and differential, protein, glucose, Gram stain, bacterial culture and processed for viral studies. If there was a strong clinical suspicion of viral CNS infection, but the initial CSF examination was acellular, a second lumbar puncture was performed. Lumbar punctures were delayed in those with unstable vital signs. Patients were examined daily or more frequently as indicated, by a member of the study team. Children with HFMD-CNS complications (particularly those with encephalitis and acute flaccid paralysis) were treated with intravenous immunoglobulin (IVIG) at the discretion of the treating physician [10].

Virological methods
Virus isolation was attempted on all swab specimens, CSF specimens, and any serum samples remaining after other investigations had been completed through the inoculation of human rhabdomyosarcoma and human embryonic kidney cells. Isolated enteroviruses were typed by nucleotide sequencing of VP1 and VP4 genes and genogrouped by phylogenetic analysis [11,12]. During the 2006 outbreak, in addition to virus isolation, all swab specimens were also tested for presence of HEV71 RNA using a HEV71 specific RT-PCR [13]. Paired serum samples (obtained on the day of admission and on day 7, or on the day of discharge or after death) and CSF specimens were also tested for IgM against dengue and Japanese encephalitis virus (JEV) in parallel, using an IgM-capture ELISA that distinguishes responses to these two viruses [14].

Statistical analysis
Data from HFMD patients recruited in the first phase of the study (mostly during 2 outbreaks that occurred between January 2000 and July 2003) were used to identify risk factors for neurological involvement (evident by CSF pleocytosis). The primary analysis was for variables associated with neurological involvement by comparing children with HFMD-CNS (i.e. with CSF pleocytosis) to those with HFMD-Non-CNS (i.e. no CSF pleocytosis). Variables that were considered potentially useful to primary care doctors in identifying children with neurological involvement were included in a multiple logistic regression analysis to look for independent risk factors for neurological involvement (i.e. CSF pleocytosis). Variables were selected backward and remained in the model only if they were statistically associated with neurological involvement (p < 0.05). (SPSS software, Version 13.0; SPSS). The association between the independent risk factors identified and neurological involvement were validated in the second phase of the study, where most patients were admitted during the 2006 outbreak. The utility of the identified risk factors as clinical predictors for neurological involvement at the point of first contact for care was also examined. Normally distributed data were compared using Student's t test; data that were not normally distributed were compared by the Mann-Whitney U test (Statview 4.02; Abacus Concepts). Differences between proportions were tested using the Chi-square test with Yates's correction or Fisher's exact test as appropriate (Epi Info, version 6; Centers for Disease Control and Prevention). A p value < 0.05 was considered statistically significant. Detailed results of the epidemiology, diagnostic virology and molecular epidemiology of this phase of the study have been reported previously [9,15].

Clinical features Comparison of patients with HFMD with CNS complications with those had more serious HFMD without CNS involvement (January 2000 to July 2003)
The clinical features of the children with HFMD-CNS (i.e. with CSF pleocytosis) are compared to those with HFMD-Non-CNS (i.e. no CSF pleocytosis) (see Additional file 1). Children with HFMD-CNS were more likely to be male and of Chinese ethnic group. Children of Iban ethnic group, however, were less likely to have HFMD with CNS complications. Children with HFMD-CNS complications were more likely to have higher mean peak temperature, peak temperature ≥ 38.5°C, longer mean total duration of fever and total duration of fever ≥ 3 days. Findings of leth-argy (from the parent's history or physical examination), faster mean heart rate, mean heart rate ≥ 150/min and limb weakness on examination were more frequently present in children with HFMD-CNS. There was no difference in the proportion of children with positive HEV71 isolation between children with HFMD-CNS and those with HFMD-Non-CNS.
To look for independent risk factors that could be used to predict neurological involvement evident by CSF pleocytosis, total duration of fever ≥ 3 days, peak temperature ≥ 38.5°C, being lethargic (from the parent's history or physical findings), history of breathlessness, history of vomiting, history of or witnessed myoclonus, neck stiffness were included in a multiple logistic regression analysis. Total duration of fever ≥ 3 days, peak temperature ≥ 38.5°C and history of lethargy were found to be independent risk factors of neurological involvement after multivariate analysis (Table 1).  (Table 2), and HEV71-postive mild HFMD (11/187), respectively.

Validation of the association between the risk factors and neurological involvement in children with HFMD in 2006 outbreak
The association between the identified risk factors (total duration of fever

The usefulness of the risk factors in predicting neurological involvement in children with HFMD in the 2006 outbreak
We were particularly interested to assess the utility of the three clinical risk factors in predicting neurological involvement in children with HFMD at the point of first contact for care. While a febrile illness ≥ 3 day was an important risk factor for severity, primary care physicians often see many children on the first 2 days of HFMD illness. To determine if peak temperature ≥ 38.5°C and history of lethargy are useful in identifying children who sought treatment within the first 2 days of the febrile illness we performed a separate analysis for children who presented within the first 2 days of the illness during the 2006 outbreak. Figure 2 shows the distribution and classification of disease severity of 730 children with HFMD in the 2006 outbreak according to the duration of febrile illness and the risk factors that were present when they first presented to hospital. Five hundred and seventy nine (79%) of 730 children were admitted within the first 2 days of febrile illness. Sixty five (11%) of the 579 children had history of lethargy plus peak temperature ≥ 38.5C. All but two (97%) of the 65 children had features of more serious illness and warranted CSF examination. About three quarter of them had CSF pleocytosis and was classified as HFMD-CNS. Only 2 (3%) of the 65 children were labeled as mild HFMD. Two hundred and twenty (38%) children had only either history of lethargy or peak temperature ≥ 38.5C. Of the 167 (76%) children who warranted a CSF examination, 102 (61%) of them had CSF pleocytosis, and were classified as HFMD-CNS. The remaining 53 (24%) children without feature of more serious illness were considered as mild HFMD. Two hundred and ninety four (51%) children had neither of the two risk factors. Eighty four (29%) of the 294 children, however, had other features of more serious illness, and hence underwent CSF examination. CSF pleocytosis was   One hundred and fifty one (21%) of the 730 children were seen on the 3 rd or later days of their febrile illness.  Table 3.

Classification of 730 Children with HFMD
Between 2000 and 2006, a total of 352 children with CNS involvement were admitted into the study. One hundred and twenty eight (36%) children had ASM (a mild and benign CNS involvement) and 224 (64%) had severe and potentially fatal CNS complications (205 had encephalitis, 14 had encephalitis associated with cardiorespiratory failure, 2 had encephalitis associated with AFP, 3 had AFP). Among the 224 children that had severe CNS complications, 204 (95%) of 215 children that survived had timely hospital admission and IVIG treatment compared to one (11%) of 9 children that died (OR 148.36, 95%CI 16.34-6609.04, p < 0.0001). Table 4 shows the clinical details and the risk factors that were present in the 9 fatal case children on hospital admission. Two or more risk factors associated with neurological involvement were present in all the 9 fatal children, and were noted for ≥ 24-48 hours before hospital admission.

Discussion
Early recognition of children at risk of neurological involvement and death (particularly those with encephalitis and encephalomyelitis) is critical as the disease progression from the onset of neurological involvement to fulminant cardiorespiratory failure may be remarkably rapid [8]. However the clinical manifestations of neurological involvement may be very subtle, particularly in young children with early CNS disease [8,16]. While the signs of cardiorespiratory distress such as breathlessness, tachypnea, tachycardia, poor perfusion are easy to recognize, they invariably appear very late shortly before most fatal case collapsed. Our results and other published studies showed that timely diagnosis and intervention, including the use of IVIG infusion, may reduce acute mortality [10,[17][18][19]. Hence the primary care doctors are confronted with a clinical challenge of identifying a small fraction of children who are at risk of neurological complication from an overwhelmingly large number of chil-  For this reason it is important to find clinical predictors for neurological involvement that can guide primary care doctors perform a proper patient triage, which should be aimed to admit high risk children into hospital early for close observation and further management, while those at low risk of neurological complication may be given outpatient care after parental education and advice. Few studies have systemically examined how to identify children at risk early before they develop cardiorespiratory failure, particularly at the primary care setting where the majority of children with HFMD would first seek treatment during a community outbreak of HFMD.
In this study we identified and validated that history of lethargy, mean peak temperature ≥ 38.5°C and total duration of fever ≥ 3 days were important risk factors for neu-rological involvement. Our study also shows that neurological involvement occurs at early course of complicated HFMD, and may be detectable within the first 2 days of the febrile illness because CSF pleocytosis was present in 174 (30%) of 579 children seen within the first 2 days of febrile illness, where they also formed 70% (174/250) of children with HFMD-CNS in the 2006 outbreak ( Figure 2). Since CSF pleocytosis may be detectable within the first 2 days of the febrile illness and fulminant cardiorespiratory failure seen in the fatal case children typically occurred on the 3 rd or later day of febrile illness, it is imperative to attempt to identify children at risk of neurological involvement before the 3 rd day of febrile illness so that they can be admitted into hospital early for close monitoring and investigation, and intervention may be instituted when necessary. Examination of body temperature and careful enquiry into history of lethargy, duration of fever and home record of body temperature should form an integral part of HFMD patient triage at the primary care level. The three risk factors are readily elicited, and can also be used after minimal training by paramedics, who are the key primary care providers in many developing countries including in Sarawak (Malaysia) in Asia. The parents of children with HFMD can also play an important role in early diagnosis of neurological complication in children with HFMD. They should be educated about the 3 risk factors, and be encouraged to monitor the children's body temperature regularly and observe the children's physical activity closely. Body temperature ≥ 38.5°C and history of lethargy may be particularly useful clinical clues for neurological involvement during the first 2 days of febrile illness since at this time the presentation of complicated HFMD is typically undifferentiated and subtle, even to the experienced clinicians [8]. Indeed both history of lethargy and temperature ≥ 38.5°C were observed for 24-48 hours in all the 9 fatal case children before they succumbed to unexpected fulminant cardiorespiratory failure (Table 4). Primary care doctors should have high index of suspicions of neurological complication when they are presented with children with HFMD who have been febrile ≥ 3 days. The children should be admitted into hospital for close observation and investigated for CNS involvement, if necessary. Our study showed that 92 (31%) of 293 children with total duration of fever ≥ 3 days in the 2000/3 outbreak, and 183 (61%) of 300 children in the 2006 outbreak had neurological involvement (Table 2). CSF pleocytosis was present in 25% (19/74) of children with a single risk factor of being febrile ≥ 3 days on hospital admission ( Figure 2). The risk of CNS complication is increased significantly when there are added risk factors of having history of lethargy and temperature ≥ 38.5°C. In contrast children who have a brief duration of low grade fever (≥ 38.5°C) and no history of lethargy are of low risk of neurological disease, and may be provided with outpatient care and parental reassurance.
Our results are in keeping with findings reported by Chang and co-authors where fever ≥ 39°C, fever duration ≥ 3 days and lethargy were more frequently observed in children with CNS involvement and in children with HEV71-HFMD than in those with CVA16-HFMD [8,20]. Although several other clinical features and laboratory abnormalities have been associated with fatal HEV71-HFMD, they have yet been validated, and been shown useful in detecting neurological disease or disease progression [8,21,22]. For example, Chong and co-authors reported that absence of mouth ulcers predicted a more complicated or fatal HFMD, and have recommended that children without mouth ulcers should be monitored closely [22]. However, in our study we did not find that children without mouth ulcers were more likely than children with mouth ulcers to have features of more severe HFMD or develop neurological complication. Not all the risk factors identified in these published studies can readily be translated into clinical practice, particularly at primary care settings. Hyperglycemia and leucocytosis have been shown as risk factors for fatal HEV71 disease [8]. However, in our experience hyperglycemia and leucocytosis are late laboratory changes in children with fulminant cardiorespiratory failure (unpublished observations), and thus are not helpful clinically in identifying children at high risk of complication and death. Elevated cardiac Troponin I, a sensitive cardiac-specific biomarker for myocardial injury, has been noted in children who subsequently developed left ventricular failure, and may be useful in identifying patients at risk of left ventricular failure and pulmonary oedema [21]. Although cardiac Troponin I has been used widely in developed countries for early diagnosis of acute coronary syndrome, it is expensive and not widely available in many developing countries, including in Sarawak. Screening for heart rate variability abnormalities, an index of autonomic nervous system, through non-invasive continuous ECG monitoring may provide early warning of impending cardiorespiratory failure about 7 hours before its onset [16]. The labour-intensive approach is most suited in a critical care setting for children already diagnosed of CNS involvement because it requires the use of expensive and sophisticated device, and close monitoring for the heart rate abnormalities.
A limitation of our study is that the clinical predictors developed for use at primary care setting were identified and tested using data collected through a hospital-based study. Clinical characteristics of children treated at primary care settings may differ from hospitalized children. However, as a large number of children with mild HFMD were admitted into our study, we had the opportunity to systemically examine the clinical feature of HFMD of varying severity, including children with mild disease that would normally be treated at primary care clinics, where we have shown history of lethargy, peak temperature 38.5C and total duration of fever 3 days were reported infrequently in children with mild HFMD.

Conclusion
Currently there is no vaccine against HEV71 infection. Early recognition of children at risk of fulminant pulmonary oedema and cardiac dysfunction is the key to reduce acute mortality and morbidity. We have identified three clinical risk factors that may help detect children at risk of neurological involvement and death at primary care settings, which can guide primary care doctors decide if hospital admission is warranted when they see with children with HFMD. These risk factors are readily elicited through history taking and measurement of body temperature.
They may also provide useful guide to help clinicians to decide the need for CSF examination, as well as to monitor disease progression in children with HFMD.