Despite being uncommon (5.2/100,000 population annually), encephalitis is of public health importance because rapid response to cases may prevent additional transmitted cases of the underlying etiologies. Currently, nearly 70% of encephalitis cases have no identified etiology, so encephalitis is presently not amenable to public health prevention and control measures. This trend of encephalitis with unidentified etiology is seen elsewhere in developed countries and seems to occur even in the presence of extensive laboratory testing (22). Because of the number of emerging infectious diseases associated with encephalitis in Australia and elsewhere in recent times, surveillance should be enhanced.

The encephalitis hospitalization rate and case-fatality rate in NSW are lower than estimates from the United States (rate of 5.2 vs. 7.3/100,000/year, respectively; case-fatality rate of 4.6% vs. 7.4%, respectively); however, these differences may result in part from differences in study methods. The US study included hospitalizations for which encephalitis was the primary, secondary, or other discharge diagnosis, whereas our study included only those cases for which encephalitis was the primary discharge diagnosis. Similarly, a study in the United Kingdom found a hospitalization rate of 1.5 per 100,000 per year. The UK study included only cases of viral encephalitis and thus used a smaller number of ICD codes in the data collection than our study. This selection could account for the lower number of cases, although a true difference cannot be ruled out.

Our findings demonstrated a higher rate of encephalitis admissions that had no identified etiology (69.6%) in NSW than has been found in either the United States (59.5%) or the United Kingdom (60.0%). Although different methods of data extraction and different sample sizes may have contributed to these differences, a true difference in pathogens causing encephalitis could exist among these 3 countries. The high proportion of encephalitis cases with unknown etiology across all 3 countries emphasizes the limits of current diagnostic tools and/or the lack of systematic investigation of encephalitis cases. In Australia, the proportion of encephalitis deaths with no identified etiology increased from 47.0% during 1979–1992 to 57.2% during 1993–2006 (C. Huppatz, unpub. data). Although the datasets for hospital inpatients and for death registry data are not directly comparable, this relative increase in deaths from encephalitis with unknown etiology further emphasizes the need for enhanced routine surveillance.

Our data showed a higher rate of total encephalitis admissions for men (54.3%) compared with women (45.7%); this difference is largely due to a higher rate of toxoplasmosis encephalitis hospitalizations for men, a finding similarly observed in the US hospitalization data (11). Toxoplasma encephalitis is likely associated with HIV infection in the early 1990s.

The proportion of patients with encephalitis with unidentified etiology did not change during the study period, despite innovations in laboratory testing, particularly increased use of PCR testing after its described use for herpes encephalitis in 1990 (23). Interestingly, encephalitis with unknown etiology was not higher in rural areas, where lack of laboratory facilities and access to specialty units may constrain diagnosis. We found a similar rate of known to unknown etiologies for urban and rural hospitals, with the exception of 1 HIV referral center that had a high number of toxoplasmosis encephalitis admissions.

Our study showed that the most common known etiology was herpes virus (12.9%); encephalitis hospitalization rates changed little for patients with this virus during the study period, a finding consistent with the US and UK studies (11,17).

A decrease in encephalitis admissions due to Toxoplasma sp. was observed after a peak occurred for this pathogen in 1993. This decline in toxoplasmosis can be explained by the increasing use of prophylactic treatment for Toxoplasma gondii infection in people with HIV infection, a recommendation first widely published in Australia in 1994 (24).

The decrease in SSPE, a neurodegenerative disorder caused by the persistence of measles virus in the central nervous system (25), reflects the success of immunization in Australia. The median incubation period for the development of SSPE after measles infection is 6–8 years (25). Our data show a decline in SSPE admissions after this illness peaked in 1991. This decline most likely resulted from a national measles vaccination campaign in Australia in 1987 (26). The surprising increase of SSPE in 2005 can be partly explained by 3 admissions of 1 patient with SSPE. This patient’s recurring hospitalizations were identified during a detailed review of encephalitis patients at the John Hunter Hospital, NSW, Australia (C. Huppatz, unpub. data). The decline in SSPE diagnoses builds on the growing evidence of the success of vaccination in decreasing illness and death from measles and its complications (26,27).

Several limitations are associated with use of hospital coding data to estimate the impact of disease. During the study period, the ICD coding system changed, and ICD-10 (1998–2007) contained several new codes not present in ICD-9 (1990–1997), including codes for zoster encephalitis, listeria meningoencephalitis, and enteroviral encephalitis. Before 1998, ICD-9 codes must have been designated to conditions associated with the new ICD-10 codes; however, which codes were used cannot definitely be determined. Fortunately, hospitalizations coded with the unknown ICD-9 codes account for a small proportion (<6%) of all encephalitis cases reported (Table).

In addition to limitations due to the ICD coding changes, differences may exist in diagnostic criteria for encephalitis used by clinicians and in inconsistencies among hospitals regarding coding practices and hospital admission criteria. Our data represent hospital admissions, rather than cases of encephalitis. Although this study criterion is useful for estimating the impact of disease as it relates to hospital service provision, case-fatality rates will be underestimated by the use of hospital admission data, rather than true cases, as some patients had multiple hospital admissions. The proportion of encephalitis cases with known and unknown etiologies remained constant, despite the ICD coding system change.

A previous study undertaken in the Northern Territory of Australia also found a high proportion (53%) of encephalitis cases with unknown etiologies (21). The Northern Territory has a distinctly different climate from that of NSW, and geographic variation exists in cases of encephalitis with known etiologies. For example, Murray Valley encephalitis and Kunjin encephalitis occur predominantly in the north. The relatively high proportion of encephalitis infections with unknown etiology appears consistent in these 2 different Australian regions.

Acute encephalitis has heralded the emergence of multiple virulent pathogens (28), including West Nile virus, Hendra virus, Nipah virus, Murray Valley encephalitis virus, and Japanese encephalitis virus, all of which can cause severe illness and death (2,4,29–31). Although emerging infectious diseases may become apparent due to large outbreaks in humans, such as the Nipah virus outbreak in Malaysia (4), the diseases that emerged recently in Australia (e.g., Hendra virus infection and Australian bat lyssavirus infection ) have been found in only a few persons (2,32). Consistent with US and UK studies (11,17), we found a high proportion (70%) of encephalitis cases with an unknown etiology in NSW. This finding highlights the need to monitor trends in encephalitis illness in Australia and to improve identification of the etiologies of encephalitis for detection of emerging infectious diseases. Based on our findings, we recommend that in Australia, surveillance be considered for encephalitis to assist in pathogen identification, alert authorities to outbreaks, and allow timely public health action.