Source of Surveillance Data

The sentinel site for JE surveillance in Kushinagar District is the district hospital. Using a standard format, the hospital reports all AES/JE cases that meet the standard World Health Organization case definition (3) to the district malaria officer; this officer then forwards compiled data to the state program officer for transmission to National Vector Borne Disease Control Programme (http://nvbdcp.gov.in/Doc/AES%20guidelines.pdf). BRD Medical College in Gorakhpur, Uttar Pradesh, is a nearby regional field laboratory of the Indian Council of Medical Research, which receives clinical samples from patients admitted to the district hospital in Kushinagar and completes most laboratory testing for JEV infections. In addition, BRD Medical College serves as a tertiary care center for patients who directly seek medical care or who are referred from nearby districts, including Kushinagar. A line list with laboratory results of these patients is reported by BRD Medical College to the district malaria officer in Kushinagar for submission to state program officers.

In July 2012, we obtained the AES line lists data for January 2011–June 2012 that were submitted by BRD Medical College to Kushinagar District Headquarters in Padrauna. The list was originally to be used to identify blocks in Kushinagar District with high, medium, and low numbers of AES cases so that sites could be selected for a larger study of the drivers of JEV transmission. Cases recorded in the line lists represented case-patients from Kushinagar District who were 1) admitted directly to BRD Medical College, 2) referred to BRD Medical College by the district hospital in Padrauna, and 3) admitted to the district hospital in Padrauna but had serum samples referred to BRD Medical College because no diagnostic kits were available and laboratory testing was limited at the district hospital. An individual AES patient tracking system does not exist in the state of Uttar Pradesh, so patients treated in the private sector or at the district hospital in Gorakhpur were not included in the surveillance lists.

Preparation of Surveillance Database for Analysis

Individual identifiers for the case-patients were removed from the line lists to ensure confidentiality. Data were cleaned (i.e., extraneous data were removed) in multiple steps, resulting in 4 changes being made to the AES line lists. First, the spelling of residential localities (block, village, or police station) were matched and standardized to result in a list of 32 residential areas. Residential areas were then categorized and combined by block (14 blocks in Kushinagar). Modifications were validated by cross-checking with village population lists provided by Savera, a local nongovernmental organization. Second, all age values were standardized to a uniform decimal system (e.g., 1 year and 6 months was changed to 1.5 years, and 8 months was changed to 0.67 year). Third, the dates of symptom onset, hospital admission, sample collection, and outcome (i.e., discharged, left against medical advice, or death) were standardized to 1 format. In the original list, dates were variously coded by using different notations (e.g., 4 March 2011 was mentioned within the same row as both 04/03/11 and 03/04/11). Assuming that the case data were entered in chronological order, we standardized dates by using, as a cue, the dates of the preceding and following AES cases. Fourth, case-patients who died, were absent, or left against medical advice were considered to have been discharged.

Data Analysis

Cleaned data were imported into SAS version 9.2.2 (SAS Institute, Inc., Cary, NC, USA) for analysis. AES cases were plotted by week and month by using the recorded dates of symptom onset. We tabulated demographic characteristics of AES case-patients for 2011 and 2012 and stratified case-patients by block, vaccination status, laboratory test result for JEV infection, and clinical outcome. Incidence for 2011 was calculated by using population denominators from the 2011 Census of India (20). We calculated incidence overall, by sex, and for children 0–6 years of age by using the age stratification available in the district-level census data. We estimated the incidence for each block by using block-specific population denominators projected with the decadal growth rate from the 2001 Census of India (20). Crude incidence rate ratios were estimated with Wald-based CIs.

We also determined the median number of days between key points in AES disease progression and diagnosis: time between onset of symptoms and hospital admission, onset of symptoms and serum sample collection, hospital admission and serum sample collection, and hospital admission and discharge or death. Using previously described methods (21), we evaluated the quality of the surveillance data by assessing the amount of data cleaning required, the proportion of missing or incomplete values in line list fields, and inconsistencies in dates recorded for key points in AES disease progression and diagnosis.

AES Epidemiology

Figure 1

Figure 1. . . . Weekly number of acute encephalitis syndrome cases, by month, in Kushinagar District, Uttar Pradesh State, India, 2011–2012. Numbers are based on data obtained from Baba Raghav Das Medical...

In 2011, a total of 721 AES cases from Kushinagar District were identified through BRD Medical College; in 2012 (January–June), 91 cases were identified. Using the cleaned line lists, we determined the weekly number of AES cases reported during January 2011–June 2012 (Figure 1). Cases peaked during August–October 2011; >150 cases were identified in each of these 3 months. This seasonal trend corresponds with an expected increase in cases during the monsoon season, when transmission of both waterborne and vector-borne diseases increases (vector density is at its maximum).

In 2011 and 2012, most case-patients were male (57.4% and 59.3%, respectively) (Table 1). In 2011, almost half of the AES cases were in children <5 years of age (44.7%); the distribution of cases by age group was not substantially different in 2012. The case-fatality rate was 18.0% in 2011 and 19.8% in 2012.

Using 2011 population data (20), we estimated that there were 20.2 AES cases/100,000 population in Kushinagar District in 2011 (Table 2). The incidence was higher among male residents than female residents (incidence rate ratio 1.29, 95% CI 1.11–1.49), and it was highest among children 0–6 years of age. The crude incidence rate ratio, comparing case-patients 0–6 years of age with those >6 years of age, was 7.97 (95% CI 6.87–9.25). Boys 0–6 years of age were at highest risk for AES. The incidence among 0- to 6-year-old boys was almost 50% greater than that among girls of the same age.

Figure 2

Figure 2. . . . Weekly number of acute encephalitis syndrome cases, by month, in Kushinagar District, Uttar Pradesh State, India, 2011–2012. Numbers represent results of laboratory testing for Japanese encephalitis and are...

The weekly numbers of AES cases, classified by JE IgM laboratory result (positive, negative, awaiting determination), is shown in Figure 2. Only 3 (4.2%) cases of JEV infection were identified in 2011: two cases were in 55-year-old men, 1 of whom died, and 1 case was in a 14-year-old girl. Vaccination status was reported for 119 case-patients, of whom 3 (2.6%) had been vaccinated. The case-patients who had received vaccine were boys 6, 7, and 8 years of age; they began experiencing symptoms in July 2011 and were discharged within 3 weeks of hospital admission. None of the 3 JE case-patients had been vaccinated.

Quality of Surveillance Data

During data cleaning, we modified 25% of the 2011 and 5% of the 2012 line list values for residential locality, age, and date parameters (Table 3). Nearly one fifth of the age data and more than one fourth of the dates were edited. For 3.2% and 13.2% of cases in 2011 and 2012, respectively, the block name could not be determined from the residential locality provided and was marked as “unknown” because the village or police station name was not found or was present in multiple blocks. In addition, several fields in the database were incomplete. As of July 2012, laboratory results for JEV infection were still classified as “awaited” (i.e., awaiting determination) in the line lists for 82% (590) of the 721 cases in 2011 and for all 91 cases in 2012 (Table 1). The line lists indicated the date of sample collection, but the type of sample collected (CSF and/or serum) and the laboratory test used (IgM ELISA, PCR, and/or cell culture) were not recorded for any of the case-patients. Most samples submitted after July 2011 were still awaiting laboratory results at the time of our study (Figure 2). In addition, for 602 (83.5%) of the 721 case-patients in 2011 and all 91 case-patients in 2012, JEV vaccination status was marked as “unknown” in the line lists (Table 1).

Key epidemiologic and programmatic indicators, such as the time between key points in AES disease progression and diagnosis, varied widely among case-patients (Table 4). In some instances, the dates were illogical (e.g., dates of symptom onset and sample collection following and preceding the dates of hospital admission, respectively). It was not possible to rectify these inconsistencies on the basis of the available data. In addition, the range of values was often large. For example, median time from onset of symptoms to hospital admission was 4 days in 2012, but some patients were admitted >2 months after symptom onset. Because AES is an acute syndrome, a long interval is not expected between symptoms and may indicate reporting of unrelated symptoms or misclassification of AES.