In each bootstrapped dataset we computed binomial log likelihood for each model. For each model summary statistics of the distribution of log likelihood was presented. Abbreviations: LASSO: Least Absolute Shrinkage and Selection Operator; PCR: Principal Component Regression; GBM: Gradient Boosting Machine; NN: Neural Network; MLR: Multiple Linear Regression.

The live attenuated Japanese encephalitis (JE) vaccine SA14-14-2 has been used in Nepal for catch-up campaigns and is now included in the routine immunisation schedule. Previous studies have shown good vaccine efficacy after one dose in districts with a high incidence of JE. The first well-documented dengue outbreak occurred in Nepal in 2006 with ongoing cases now thought to be secondary to migration from India. Previous infection with dengue virus (DENV) partially protects against JE and might also influence serum neutralising antibody titres against JEV. This study aimed to determine whether serum anti-JEV neutralisation titres are: 1. maintained over time since vaccination, 2. vary with historic local JE incidence, and 3. are associated with DENV neutralising antibody levels. We conducted a cross-sectional study in three districts of Nepal: Banke, Rupandehi and Udayapur. Udayapur district had been vaccinated against JE most recently (2009), but had been the focus of only one campaign, compared with two in Banke and three in Rupandehi. Participants answered a short questionnaire and serum was assayed for anti-JEV and anti-DENV IgM and IgG (by ELISA) and 50% plaque reduction neutralisation titres (PRNT50) against JEV and DENV serotypes 1–4. A titre of ≥1:10 was considered seropositive to the respective virus. JEV neutralising antibody seroprevalence (PRNT50 ≥ 1:10) was 81% in Banke and Rupandehi, but only 41% in Udayapur, despite this district being vaccinated more recently. Sensitivity of ELISA for both anti-JEV and anti-DENV antibodies was low compared with PRNT50. DENV neutralising antibody correlated with the JEV PRNT50 ≥1:10, though the effect was modest. IgM (indicating recent infection) against both viruses was detected in a small number of participants. We also show that DENV IgM is present in Nepali subjects who have not travelled to India, suggesting that DENV may have become established in Nepal. We therefore propose that further JE vaccine campaigns should be considered in Udayapur district, and similar areas that have had fewer vaccination campaigns.