Search

Background: Encephalitis, brain inflammation and swelling, most often caused by an infection or the body’s immune defences, can have devastating consequences, especially if diagnosed late. We looked for clinical predictors of different types of encephalitis to help clinicians consider earlier treatment. Methods: We conducted a multicentre prospective observational cohort study (ENCEPH-UK) of adults (> 16 years) with suspected encephalitis at 31 UK hospitals. We evaluated clinical features and investigated for infectious and autoimmune causes. Findings: 341 patients were enrolled between December 2012 and December 2015 and followed up for 12 months. 233 had encephalitis, of whom 65 (28%) had herpes simplex virus (HSV), 38 (16%) had confirmed or probable autoimmune encephalitis, and 87 (37%) had no cause found. The median time from admission to 1st dose of aciclovir for those with HSV was 14 hours (IQR 5-50); time to 1st dose of immunosuppressant for the autoimmune group was 125 hours (IQR 45-250). Compared to non-HSV encephalitis, patients with HSV more often had fever, lower serum sodium and lacked a rash. Those with probable or confirmed autoimmune encephalitis were more likely to be female, have abnormal movements, normal serum sodium levels and a cerebrospinal fluid white cell count < 20 cells x106/L, but they were less likely to have a febrile illness. Interpretation: Initiation of treatment for autoimmune encephalitis is delayed considerably compared with HSV encephalitis. Clinical features can help identify patients with autoimmune disease and could be used to initiate earlier presumptive therapy. Clinical Trial Registration Details: Funding Information: National Institute for Health Research, UK. Declaration of Interests: None to declare. Ethics Approval Statement: The study protocols were approved by participating sites and the National Research Ethics Service (now part of the Health Research Authority) East Midlands Nottingham 1 committee (reference 11/EM/0442). Written consent for entry into the study was obtained from patients or an accompanying relative.

Variable selection in the presence of outliers may be performed by using a robust version of Akaike's information criterion (AIC). In this paper, explicit expressions are obtained for such criteria when S- and MM-estimators are used. The performance of these criteria is compared with the existing AIC based on M-estimators and with the classical non-robust AIC. In a simulation study and in data examples, we observe that the proposed AIC with S and MM-estimators selects more appropriate models in case outliers are present.