Overrepresentation of South Asian ethnic groups among cases of influenza A(H1N1)pdm09 during the first phase of the 2009 pandemic in England

Abstract Background During the first wave of the influenza A(H1N1)pdm09 pandemic in England in 2009, morbidity and mortality were higher in patients of South Asian (Indian, Pakistani or Bangladeshi) ethnic minority groups. Objectives This study aims to provide insights in the representation of this group among reported cases, indicating susceptibility and exposure. Methods All laboratory‐confirmed cases including basic demographic and limited clinical information that were reported to the FluZone surveillance system between April and October 2009 were retrieved. Missing ethnicity data were imputed using the previously developed and validated South Asian Names and Group Recognition Algorithm (SANGRA). Differences between ethnic groups were calculated using chi‐square, log‐rank and t tests and rate ratios. Geographic clustering was compared using Ripley's K functions. Results SANGRA identified 2447 (28%) of the total of 8748 reported cases as South Asian. South Asian cases were younger (P < .001), more often male (P = .002) and more often from deprived areas (P < .001) than cases of other ethnic groups. Time between onset of symptoms and laboratory sampling was longer in this group (P < .001), and they were less often advised antiviral treatment (P < .001), however, declined treatment less. The highest cumulative incidence was seen in the West Midlands region (32.7/10 000), London (7.0/10 000) and East of England region (5.7/10 000). Conclusions People of South Asian ethnic groups were disproportionally affected by the first wave of the influenza pandemic in England in 2009. The findings presented contribute to further understanding of demographic, socioeconomic and ethnic factors of the outbreak and inform future influenza preparedness to ensure appropriate prevention and care.


| INTRODUC TI ON
Greater pandemic-related morbidity and mortality were reported for indigenous and minority ethnic groups in several countries including the UK. [4][5][6][7][8][9][10] Sociodemographic variation in the impact of infectious diseases can arise due to differences in exposure, susceptibility and/or treatment; there may be language and cultural barriers to seeking and receiving appropriate care. 11 The South Asian ethnic group (UK residents of Indian, Pakistani or Bangladeshi birth or heritage) is one of the largest ethnic minority groups in England, accounting for 5.6% of the total population in the 2011 national census. 12 During the UK pandemic, there was an overrepresentation of South Asian ethnic groups among hospital admissions and deaths (including paediatric deaths) associated with pandemic influenza. 9,13,14 This study aims to ascertain if this overrepresentation also existed in the overall number of laboratory-confirmed pandemic influenza cases in England suggesting greater exposure or susceptibility in this population rather than greater severity of illness (eg due to co-morbidities).
The study aimed to estimate the occurrence of influenza

| ME THODS
The study is based on laboratory data obtained during the 2009 pandemic. In the "containment" phase of the public health response (from April to July 2009), suspected influenza H1N1pdm09 cases (both ambulatory and hospitalised) were routinely tested for infection, confirmed cases were treated with antivirals, and contacts were traced and offered post-exposure antiviral chemoprophylaxis. A "treatment" phase (from July 2009) started once sustained community transmission occurred, where all suspected cases were treated with antivirals without laboratory confirmation and contact tracing was discontinued. 15 Data on laboratory-confirmed cases of influenza A(H1N1)pdm09 in England were collected by the Health Protection Agency using "FluZone", a centralised online case management system used for capturing demographic and clinical information on cases, between April and October 2009. Ethnic group data in FluZone were largely incomplete but, as full names of all cases were available, the computerised South Asian Names and Group Recognition Algorithm (SANGRA) was used to impute the missing data. The SANGRA algorithm can identify if a name is likely to be of South Asian origin, that is from the Indian subcontinent. The development and validation of the method are described elsewhere 15; it has been used in several studies lacking complete ethnic group data. [16][17][18] The limited data on ethnic group recorded in FluZone were compared to the imputed data to estimate the sensitivity and specificity of the SANGRA algorithm. A slight male predominance was seen overall (51%) which was more pronounced among cases of South Asian ethnic groups (54%) than among cases of other ethnic groups combined (50%, P = .002, Table 1). Overall, age ranged from 0 to 90 years with a median age of 14 years. Cases of South Asian ethnic group were younger compared to all other groups; their median age was 12 years, and more than half of the cases were in the 5 to 14 years age group. For other ethnic groups, the median age of cases was significantly higher at 16 years (t test; P < .001) and the highest percentage of cases was in the 15 to 49 years age group.
More than three-quarters of the cases of South Asian ethnic group were from the most deprived national IMD quintile areas (78%), significantly more than for cases of other ethnicities (46%; P < .001). ONS data are not readily available to undertake combined age and IMD stratification of cases.
Limited clinical information was available. An underlying medical condition was reported for only 49 cases, mainly chronic respiratory disease (n = 26), immunosuppression (n = 9) and diabetes (n = 7). There were no significant differences in underlying medical  The South Asian proportion of the total population is highest in London region (11.1%) followed by the West Midlands region (7.8%) ( Table 2). The highest cumulative incidence of confirmed influenza A(H1N1) pdm09 infection for South Asian cases was seen in the West Midlands region (32.7/10 000) followed by London (7.0/10 000) and East of England (5.7/10 000). For England, the relative risk of laboratory-diagnosed infection among South Asian cases compared to cases of other ethnic group was 6.8, with the highest relative risks in the West Midlands (10.4), Yorkshire & the Humber (8.8), East of England (6.1) and the East Midlands (5.5).
The time from onset of symptoms to laboratory sampling was significantly longer for South Asian cases than for other cases (logrank test chi-squared value 30.6; P=<0.001). showed that mortality rates were higher for Indian cases (aIRR 1.87) and Pakistani cases (aIRR 3.37) in England than those of the White British ethnic group. 9 A study describing laboratory-confirmed cases in the West Midlands identified 57.9% as South Asian cases by manual classification, 25 a striking proportion that is higher than the amount estimated by this study.

| D ISCUSS I ON
In the early pandemic, following multiple importations local transmission was mainly found in the urban areas of London and the West Midlands. These areas accounted for two-thirds of the overall number of cases and over three-quarters of South Asian cases reported. Urban residents are initially possibly more exposed to the pandemic influenza than rural residents as there is higher mobility, higher population density and a more diverse population. Much of the transmission was driven by school-aged children with outbreaks reported in school settings. 26 Urban areas in the United States were also more affected by pandemic influenza than rural areas. 27 In a previous UK study, crude analysis showed that the risk of mortality was higher for urban cases; however, this association was absent after adjustment for deprivation. 9 Relatively, high proportions of the London and West Midlands populations comprise of South Asians (11.1% and 7.8% respectively) compared to the national average (5.4%). According to our findings, South Asian cases were also significantly more spatially clustered than other cases in urban centres. Compared to the average house-

TA B L E 1 Characteristics of cases by ethnic group, England, 2009
All cases n (%)

P-value (test performed)
Gender (n = 8287) The percentage of cases classified as being in the most deprived quintile areas was high at 56%, providing further evidence that transmission is associated with deprivation as previously found in a study describing early cases in London. 29 Reflecting observations in the general population, cases of South Asian ethnic group resided significantly more often in the most deprived quintile areas (79%) than other cases (46%); however, this percentage was lower in South Asian cases attending hospitals in England (26%) 20 possibly as a result of different referral practices or differences in severity of disease. Deprivation has also been associated with higher mortality among influenza cases previously. 9,30 Just under half of all cases were younger than 15 years (46%). In addition, the difference in antiviral treatment advice between South Asian and other cases is unexplained and its implications for impact on South Asian groups are unclear. Antiviral treatment was advised significantly less often for South Asian cases (P = <0.001).
There were no apparent differences in pregnancy or underlying health conditions between the groups; however, limited incomplete data were available. Differences in antiviral treatment could arise from delayed clinical presentation or laboratory sampling in South Asian groups (our analysis provides some evidence for the latter); language barriers; and/or poorer access to healthcare services in areas with high South Asian populations.

ACK N OWLED G EM ENTS
The authors thank Professor Peter Diggle and Dr Barry Rowlingson from the University of Lancaster for statistical advice in the early stages of analysis. Dr Ioannis Karagiannis is gratefully acknowledged for comments on the draft paper.

CO N FLI C T O F I NTE R E S T S
The authors declare that they have no competing interests. 6.8 6.7-6.8