This study demonstrates that patients with AES of suspected viral aetiology, either where JE was confirmed or where viral aetiology remained unknown, were significantly more likely to have a bad outcome compared to AES patients with bacterial or malaria infection.
Appropriately, JE surveillance is a health priority in Nepal. However, public health and clinical teams should be aware that patients with AES of unknown viral aetiology also have a high risk of morbidity and mortality. Furthermore, since there are up to 3 times more Nepali children with AES of unknown viral aetiology than proven JE, a bad outcome among the former group impacts on a larger number of children Therefore, identification and optimizing management of patients with AES should also be a priority.
The lower frequency of bad outcome among the AES patients with bacterial or malaria infection is likely to reflect the availability and effective use of antibiotics and anti-malaria treatment to reduce morbidity among these patients.
AES patients without LP results exhibited a significantly higher rate of death. The finding is likely to reflect lumbar punctures being undertaken less frequently on children who were critically ill. The finding highlights that restricting analyses to patients where LP results are available can lead to an underestimation of the frequency of death and bad outcome linked to AES.
Interestingly, the study has identified that the number of days of fever (reflecting number of days of illness) the patient experienced prior to hospital admission is a prognostic indicator of bad outcome in both patients with AES of unknown viral aetiology and JE. When analysis was applied to AES patients of bacterial aetiology no significant association was identified. Antibiotic use in the community may reduce fever duration prior to hospital admission in patients with bacterial infection and may confound the association between fever duration and bad outcome. In contrast, antibiotics would have limited impact on fever duration during viral infection.
Shorter duration of illness (or fever) prior to admission has previously been associated with good outcome among children for a range of diseases . What is striking about our finding is that there is no specific treatment for JE or AES of suspected viral aetiology, yet attending hospital earlier in the illness course appears to be of benefit. The findings would suggest that hospital admission and the supportive management received there improves outcome.
Patients with impaired consciousness are often unable to drink themselves. Consequently, dehydration and metabolic acidosis may complicate AES . Dehydration and acidosis would fit the significantly lower body weight and higher respiratory rate observed among JE compared to Non-JE patients. A previous prospective survey of JE management identified that fluid supplementation was associated with a positive influence on outcome . Although appropriate fluid provision is a delicate balance when managing brain injury [16, 17], one possible explanation of the positive influence of hospital admission may be that patients receive fluid support in hospital during the illness.
Given the commonest reported mode of presentation was self-referral, the focus on hastening patient attendance would lie with improving patient awareness of the features of AES and encouraging families to attend hospital promptly. Further research is needed to understand the factors that underlie families' health-seeking behaviours when a child is ill.
As shown previously, a low Glasgow coma scale (GCS) and/or a focal neurological deficit at hospital admission were independent clinical markers of bad outcome among children with AES . JE patients frequently exhibit raised intra-cranial pressure, brain herniation syndromes  and focal brain lesions on neuro-imaging studies , explaining the preponderance of focal neurolgical deficits in this group.
In line with previous reports, mannitol was prescribed significantly more frequently among AES patients who exhibited a bad outcome . Interestingly, a randomized clinical trial of mannitol among children with raised raised ICP secondary to cerebral malaria (another cause of non-traumatic brain injury) did not identify any beneficial effect . A similar study would help clarify whether mannitol is a useful supportive treatment in AES .
JE patients exhibited an increased respiratory rate compared to Non-JE patients. Furthermore, JE patients who exhibited a high respiratory rate were associated with a good outcome, while those with a lower respiratory rate were associated with a bad outcome. This pattern can be observed during evolution of many complications, including metabolic acidosis, pneumonia, acute flaccid paralysis (involving the inter-costal muscles) or brain damage, where there may be an intial compensatory rise in respiratory rate followed by a fall when the body decompensates.
Tachypnea is also a feature of the severe brain injury syndrome - paroxysmal autonomic instability with dystonia (PAID). Patients with this brain injury syndrome exhibit intermittent agitation, diaphoresis, hyperthermia, hypertension, tachycardia, tachypnea, and extensor posturing. All of these signs overlap with features reported in both JE and AES. PAID may exist among AES patients.
As a descriptive analysis this study cannot distinguish cause from consequence. Consequently although several clinical features and interventions appear linked with outcome, this study is unable to determine whether these parameters are causal. Similarly, as a retrospective study it is limited by the breadth and quality of information available in the hospital notes. More detailed information on acid-base status would help discriminate between respiratory and metabolic causes of tachypnea. Similarly, more systematic measurement of urea and electrolytes and additional indicators of fluid balance would help assess the influence of fluid support on outcome. Based on their discharge diagnosis, the AES patients without LP contributed to all AES groups. However, a prospective study with a more systematic investigation of pathogen aetiology is warranted to confirm these findings.