Objective The clinical epidemiology of childhood encephalitis in Australia is inadequately understood. We aimed to describe recent trends in childhood encephalitis-related hospitalisation.
Study design We identified encephalitis-related hospital admissions (2000–2012) in national datasets among children ≤14 years using ICD encephalitis codes. We calculated hospitalisation rates and analysed trends by year, age, gender, location, indigenous status and aetiology.
Results Rates of childhood encephalitis hospitalisations significantly declined over an 11-year period (2000–2012; average hospitalisation rate 3.2/100 000). Varicella encephalitis hospitalisations decreased significantly, associated with high levels of varicella vaccine coverage since 2006. Acute disseminated encephalomyelitis (ADEM) was the most common ‘specified’ cause of encephalitis hospitalisation (15%–17%), and its rate has significantly increased. The highest hospitalisation rates occurred in the <1 year age group (5.8/100 000) and varied by location (highest in Northern Territory). The majority (58.9%) of hospitalised encephalitis had no cause identified; this proportion was highest in the <1 year age group (77%). The most common specified infectious causes included: herpes simplex virus, enterovirus, bacterial meningoencephalitis and varicella. When aggregated, the proportion of childhood encephalitis coded as viral was 21.2%.
Conclusion Hospitalisation of childhood encephalitis has slightly decreased in Australia. High rates of childhood immunisation have been associated with a reduction of varicella-associated encephalitis in Australian children. ADEM, an immune-mediated encephalitis, is the most common recognised cause of encephalitis in children. Young children (<1 year) have the highest admission rates. The high proportion of ‘unspecified’ encephalitis deaths and hospitalisations is an ongoing challenge.
- Infectious Diseases
Statistics from Altmetric.com
What is already known on this topic
Encephalitis occurs with highest incidence in young children.
Immunisation has been associated with reductions in the incidence of and mortality from encephalitis caused by measles (including subacute sclerosing panencephalitis).
The cause of encephalitis in children is frequently unknown.
What this study adds
We calculate age-disaggregated and location-specific incidence estimates for encephalitis in Australian children.
We show a significant reduction in varicella-associated encephalitis in the context of high varicella immunisation coverage.
We show that acute disseminated encephalomyelitis is the most common single specified cause of childhood encephalitis, and its rate is increasing.
Encephalitis is inflammation of the brain parenchyma, which results in neurological dysfunction.1 Encephalitis can result in mortality and severe neurological morbidity in children. Across all ages, the most commonly identified cause of encephalitis is infection, with viral pathogens most frequently isolated.2 ,3 However, for many patients presenting to hospital with encephalitis, a specific cause is never found.4–7
In Australia, mortality and hospital admission data for encephalitis in all age groups have been previously analysed,8 ,9 although in these studies, the data were categorised in 10-year age brackets. The highest rates of hospitalisation occurred in the 0–9 years age bracket, but this was not disaggregated. This is important to define because studies from Finland, Sweden and the USA have shown that the highest incidence of encephalitis occurs in the very young.5 ,6 ,10
The major causes of encephalitis in children differ between geographical locations; this limits extrapolation from existing studies of the Australian population. For example, in Thailand, dengue fever and Japanese encephalitis virus were most common,11 while in Finland, varicella zoster virus (VZV) and enterovirus were dominant.10 Australia, although a highly developed, industrialised society, akin to Europe and North America, has a unique environment and novel endemic encephalitides.3 Our understanding of the epidemiology of encephalitis in Australian children is currently inadequate.
We aimed to study encephalitis-related deaths and hospitalisations in Australian children using existing datasets. Specifically, we aimed to examine the rate and causes of encephalitis mortality, to establish incidence estimates for childhood encephalitis in Australia and to determine the contemporary causes of this important childhood condition.
We obtained hospitalisation data on encephalitis-associated conditions from the Australian Institute of Health and Welfare National Hospital Morbidity Database. This database receives administrative, demographic and clinical information about all patients admitted to public and private hospitals in Australia, which are coded by financial year. The full list of encephalitis-related ICD codes is contained in table 2. The available hospitalisation data were limited: the period 2000/2001–2008/2009 only included cases identified by the principal diagnostic field; the period 2009/2010–2011/2012 included cases identified using any diagnostic fields; codes of herpes simplex virus (HSV), enterovirus, mosquito-borne and listeria encephalitis were only available for a single financial year (2010/2011). All children aged ≤14 years (‘childhood’) were included, and we excluded those admissions associated with a ‘same-day’ flag or with ‘bed-days’ ≤1 day as being inconsistent with acute encephalitis (likely duplicates because of, eg, interhospital transfer or follow-on, day-only treatment).
Encephalitis-associated hospitalisations were analysed according to age, gender, indigenous status (Aboriginal and Torres Strait Islander origin people), year, geographical location and cause. National, state-specific and Aboriginal and Torres Strait Islander people population statistics were obtained from the Australian Bureau of Statistics and the mid-year population data for each year was used to calculate crude childhood (≤14 years), age-specific and location-specific hospitalisation and mortality rates.
For those hospitalisations coded as otherwise-specified encephalitis in the principal diagnostic field (see table 2), we reviewed the additional ICD-10 diagnostic codes assigned for the admission with particular reference to those coded A00 to B99 (certain infectious and parasitic diseases) and G00 to G99 (diseases of the nervous system). Where a known pathogen associated with encephalitis or a non-infectious aetiology was specified among the first six diagnostic codes assigned, we determined this to be the ‘otherwise specified’ aetiology of the encephalitis-associated admission.
We assessed secular trends using a χ2 for trend analysis; however, this could not be validly performed for individual causes of mortality because of their low frequency. The analysis was performed using Microsoft Excel 2010, SPSS V.22.0 and Epi Info V.7.
From 2000 to 2012, there were 1847 childhood encephalitis hospitalisations in Australia coded in the principal diagnostic field; an average of 154 per year with an average annual hospitalisation rate of 3.8 per 100 000 population ≤14 years (95% CI 3.6 to 4.0) (figure 1). When all diagnostic fields were considered (2009–2012), we identified an average of 213 admissions per year (56–66 additional cases per year) with an average annual hospitalisation rate of 5.0 per 100 000 population ≤14 years (95% CI 4.6 to 5.4). The total and cause-specific annual hospitalisation rates are shown in figure 2. Overall childhood encephalitis hospitalisations in Australia have decreased significantly across the study period (p=0.003). Between 2000 and 2012, there has been a significant decrease in encephalitis hospitalisations coded as varicella-associated and ‘unspecified’ (p<0.001); there has been a significant increase in hospitalisations coded as ‘otherwise specified’ and acute disseminated encephalomyelitis (ADEM) (p<0.001 and 0.002, respectively).
For a subperiod of this hospitalisation dataset (2000–2004), we have analysed national childhood encephalitis mortality data (unpublished results). There were 21 childhood (≤14 years) deaths in this period; assuming that all deaths in this period were admitted to hospital prior to death; the case fatality ratio was 2.6%.
The hospitalisation rate was higher in males than females (table 1). The highest hospitalisation rates were in the <1 year age group, and rates reduced with increasing age (table 1 and figure 2). Hospitalisation rates varied by location; they were highest in the Northern territory (tropical and subtropical), though these estimates are based on a small number of cases (range 0–10 cases per annum) and lowest in South Australia and Tasmania (temperate). Children of indigenous background had higher encephalitis hospitalisation rates from 2009 to 2012, when all diagnostic coding fields are included, but this was not evident across the full study period when considering only the principal diagnostic field.
From 2009 to 2012, when all coding fields are considered, 50% of childhood encephalitis in Australia was coded as unspecified; 27.8% was coded as ‘otherwise specified’ (table 2). ADEM (an immune-mediated cause) was the most frequently coded specified cause of childhood encephalitis (17.1%). The specified infectious causes included: HSV, enterovirus, bacterial meningoencephalitis, varicella and mosquito-borne encephalitis. When aggregated, for the single year for which most complete data were available (2010–2011), the proportion of childhood encephalitis coded as viral was 21.2%, although one-fifth of these cases are coded as otherwise-specified viral encephalitis (G05.1, A85.8). HSV and enterovirus were the most frequent, specified viral causes in this year.
There were age-related differences in aetiology. VZV was most frequent in the 1–9 years groups (8%–9%) compared with <2% in the <1 year group where otherwise-specified viral encephalitis was the most common coded aetiology. ADEM accounted for 17%–19% of encephalitis in the 1–9 years group and 5% and 13% in the <1 year and 10–14 years groups, respectively. The highest proportion of unspecified encephalitis was in the <1 year group (77%) and then in the 10–14 years group (66%). In the single year for which data were available, 30% and 45% of HSV encephalitis occurred in the <1 year group and 1–4 years groups, respectively; 12 of 13 enteroviral encephalitis cases occurred in the 0–4 years groups and all cases of mosquito-borne encephalitis occurred in the 1–4 years group.
In the subanalysis of otherwise-specified encephalitis (ICD-10 codes: G04.8, A85.8, G05.1), a specific infectious or immune-mediated aetiology was not able to be determined for 64% of these admissions (n=185; see figure 3). Of those admissions where a specific aetiology could be determined, the most frequent causes identified were Mycoplasma pneumoniae (n=22; B96.0, A49.3, J15.7), enterovirus (n=11; B97.1, B34.1), rotavirus (n=8; A08.0), Epstein–Barr virus (EBV: n=6; B27.0) and respiratory syncytial virus (RSV: n=5; B97.4, J12.1). A large number of other aetiologies (most infectious; see figure 3) were identified, although for many causes, there was only a single case across the entire study period. The rate of the ‘aetiology undetermined’ subgroup increased significantly across the study period, whereas the rate of M. pneumoniae-associated encephalitis did not increase (respective p values <0.001 and 0.32).
Encephalitis is a critical disease that requires hospital care, such that hospitalisation should be considered a good proxy for incidence. A range of childhood encephalitis incidence rates have been reported in international studies, from 2.8 to 10.5 per 100 000 in England, Sweden and the USA, with highest rates in the under 1 year age group (13.7 to 18.4 per 100 000).6 ,7 ,10 The Australian incidence estimates reported here (3.8 per 100 000 (principal diagnostic field) and 5.0 per 100 000 (all fields)) are within the range of these international estimates and show a similar age-related variability. However, they fall below that expected according to a systematic review of the incidence of childhood encephalitis in industrialised countries where a minimum of 10.5 per 100 000 was calculated as a guide for encephalitis surveillance programmes.12 This suggests that hospitalisation coding, even when all admission diagnostic coding fields are included, may be relatively insensitive for childhood encephalitis. Coding of hospitalisation data in Australia is considered to be of high quality, although coding errors may occur more frequently for uncommon conditions, such as encephalitis.13 Furthermore, we would emphasise that because some important codes were not included in the data across the full time period (especially HSV and enteroviral encephalitis) these rates must be interpreted as minimum estimates. These limitations are unlikely to have caused significant year-to-year variation affecting the validity of the secular trend analysis.
VZV has previously been shown to be one of the most, if not the most, common causes of childhood encephalitis.5 ,10 Varicella vaccine (Varilix) has been recommended for use (2004) and subsequently funded as part of the national immunisation programme (from 2006) in Australia. High rates of vaccine coverage have been associated with a significant decrease in varicella-associated hospitalisation.14 A recent study from Canada showed a decreased frequency of varicella-associated neurological disease at a single hospital associated with childhood varicella immunisation in the province.15 We have added to these studies by showing, for the first time at a population level, a significant reduction in the hospitalisation rate of varicella encephalitis in Australia associated with high levels of vaccine coverage.14
ADEM is an immune-mediated encephalitis that occurs most frequently in children. It is effectively treated with corticosteroids, most often with good prognosis. It has been previously estimated to occur in 0.4 and 0.64 per 100 000 person-years in people aged <20 and 15 years, respectively.16 ,17 In the three most recent years of our analysis, we identified a higher incidence in the range of 0.68–0.79 per 100 000 person-years in people aged ≤14 years. We have also, for the first time at a population level, shown a rising rate of the diagnosis of ADEM; our findings emphasise its importance as a cause of childhood encephalitis. A possible explanation for this increasing rate is that MRI is central to the diagnosis of ADEM and there has been an increasing availability of MRI. ADEM is frequently denoted as being ‘post-infectious’ and a many pathogens have been associated with ADEM. Unfortunately, these administrative data are not able to provide further insights into the infectious triggers of ADEM; however, our study highlights the need for further research in this area. ADEM can, rarely, be vaccine-associated18 and our study emphasises the need for further population-level studies of the association of ADEM and vaccine receipt.19
The case fatality ratio of 2.6% reported here is relatively low compared with other published studies. In a US study of hospital discharge data, the fatality ratio was 7.4% across all age groups,6 and in three recent large prospective studies including all age groups it ranged from 10% to 12%.4 ,20 ,21 Possible reasons for this difference are that children are at lower risk of death from encephalitis than adults, and regional differences in aetiology.
We have identified a high proportion of ‘unspecified’ encephalitis in this study. Most studies of encephalitis have noted this high proportion of encephalitis with unknown cause; in childhood cohorts, the range of ‘unspecified’ or ‘unknown’ encephalitis varies between 37% and 67%.4–7 ,9 ,10 ,22 ,23 Reasons for this include inadequate testing of cases for known causes, limited sensitivity of current diagnostic testing for known causes or truly novel infectious or immune-mediated causes of encephalitis.20 The testing of adult encephalitis cases in Australia has been shown to be not standardised and a call has been made for guidelines to assist in optimising diagnostic approach.24 Furthermore, even among the group of admissions coded as otherwise-specified encephalitis, in two-thirds, we were unable to determine an aetiology using this available coding data. Prospective research is required involving careful aetiological categorisation, and collection and banking of specimens from children with ‘unknown’ encephalitis with the application of advanced molecular techniques for pathogen discovery and screening for newer immune-mediated causes, for which, there are no specific ICD codes; we began such nationally in 2014. Recent prospective aetiological studies have shown a high prevalence of antibody-mediated encephalitides (including: anti-N-methyl D-aspartate reception and anti-voltage-gated potassium complex encephalitis).20 ,25 There are no specific ICD-10 codes for these conditions, so, it is a limitation of this study that we cannot determine their incidence from this dataset. It is likely that where they have been diagnosed, the admission would be coded as otherwise-specified encephalitis.
The H1N1 influenza pandemic occurred within the study period, and there are data suggesting its association with an increased rate of neurological disease.26 We have not identified this in our study; however, we did not specifically search for the most relevant ICD-10 code (J10.8—influenza with other manifestations: including encephalopathy). We identified only two cases of influenza encephalitis/encephalopathy in our subanalysis of otherwise-specified encephalopathy, one from 2007 and one from 2011.
The infectious causes of encephalitis identified by this study are instructive. In the most recent years, no cases of measles (and subacute sclerosing panencephalitis), mumps and rubella encephalitis were identified, which emphasises the ongoing effectiveness of the national immunisation programme in Australia against these pathogens. In their absence, the most common causes are HSV, enterovirus, VZV (although decreasing), bacterial meningoencephalitis and the mosquito-borne viruses (includes the endemic Flaviviruses Murray Valley encephalitis and Kunjin); all of which have established neurotropism. M. pneumoniae, rotavirus and RSV were identified as prominent infectious causes of encephalitis among the otherwise-specified group. All three occur almost exclusively in childhood, and are considered controversial being rarely isolated from the central nervous system.27–29 There is a need to further clarify the aetiopathogenesis of these important encephalitis-associated infections.
As expected, childhood encephalitis hospitalisations showed a male predominance and occurred more commonly in younger age groups. There was some geographical variation in childhood encephalitis hospitalisations. We have hinted at possible climatic drivers of the higher rates of encephalitis mortality and hospitalisation in the Northern Territory compared with Southern states (Tasmania, South Australia, Victoria), though this requires further investigation. Another factor potentially underlying these geographical differences is the high indigenous population proportion in the Northern Territory. The high rates of encephalitis hospitalisation in indigenous (ie, Aboriginal and Torres Strait Islander origin) children is further evidence of health-related disadvantage in this population.
This study is limited by the available datasets including: the identification of hospitalised cases using only the principal diagnostic for the full hospitalisation study period and the availability of HSV, enterovirus and mosquito-borne codes only for a single year.
We have shown a significant decreasing trend in childhood encephalitis hospitalisations over a 12-year period; in particular, a recent decrease in varicella-associated encephalitis associated with high levels of varicella immunisation. ADEM is the most common contemporary ‘specified’ cause of encephalitis hospitalisation, and its rate is increasing. Younger children, especially infants aged <1 year, have the highest incidence of hospitalisation. At present, the majority of paediatric encephalitis hospitalisations do not have an identified cause. Further prospective research is required, in the context of enhanced surveillance, to better define the aetiology of childhood encephalitis especially given the high proportion of ‘unknown’ encephalitis and emergence and re-emergence of infectious encephalitides in our region.
The authors thank Dr Gulam Khandaker and Dr Camille Raynes-Greenow for their assistance with the analysis of the National Hospital Morbidity data. They also thank Dr Han Wang and Dr Aditi Dey for their assistance with extraction of the data.
Contributors PNB: performed the analysis of national hospitalisation data, drafted the initial manuscript and approved the final manuscript as submitted. LK: undertook preliminary analysis of NSW hospitalisation as a project as part of clinical training, and contributed to the manuscript and approved the final manuscript as submitted. RB: reviewed and revised the manuscript and approved the final manuscript as submitted. NW: jointly conceptualised the study with CAJ, reviewed and revised the manuscript and approved the final manuscript as submitted. CAJ: conceptualised and designed the study with NW, reviewed and revised the manuscript and approved the final manuscript as submitted.
Funding This study was supported by an National Health and Medical Research (NHMRC) postgraduate research scholarship (APP1074547), Norah Therese-Hayes scholarship and Royal Australasian College of Physicians award for excellence to PNB; NHMRC centre for research excellence in critical infections (APP1001021) to CAJ, RB and the Marie Bashir Institute (MBI) for infectious diseases and biosecurity.
Competing interests None declared.
Provenance and peer review Not commissioned; externally peer reviewed.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.