Transmission dynamics of the 2009 influenza A (H1N1) pandemic in India: the impact of holiday-related school closure

Abstract

The role of social-distancing measures, such as school closures, is a controversial aspect of pandemic mitigation planning. However, the timing of 2009 pandemic provides a natural experiment for evaluating the impact of school closure during holidays on influenza transmission. To quantify the transmission intensity of the influenza A (H1N1) pdm'09 in India, by estimating the time varying reproduction number (Rt) and correlating the temporal changes in the estimates of Rt for different regions of India with the timing of school holidays. We used daily lab-confirmed case reports of influenza A (H1N1) pdm'09 in India (during 17 May'09 to 17 May'10), stratified by regions. We estimated the transmissibility of the pandemic for different regions from these time-series, using Bayesian methods applied to a branching process model of disease spread and correlated the resulting estimates with the timing of school holidays in each region. The North-west region experienced two notable waves, with the peak of the first wave coinciding with the start of a 4 week school holiday (September-October'09). In the southern region the two waves were less clear cut, though again the first peak of the first wave coincided with the start of school holidays--albeit of less than 2 weeks duration (August'09). Our analysis suggests that the school holidays had a significant influence on the epidemiology of the 2009 pandemic in India. We estimate that school holidays reduced the reproduction number by 14-27% in different regions of India, relative to levels seen outside holiday periods. The estimates of the reproduction number obtained (with peak R values below 1.5) are compatible with those reported from other regions of the world. This work reinforces past studies showing the significant impact of school holidays on spread of 2009 pandemic virus, and by inference the role of contact patterns in children on transmission.

Keywords: India; Influenza; Mathematical models; Reproduction number; School holidays.

Figure 1

Boundaries of regions used in this paper.

Figure 2

a. Assumed discretised serial interval distribution (mean=2.6 days, SD=1.5 days); b, c. Assumed prior distributions for the serial interval distribution mean and standard deviation, respectively.

Figure 3

Daily reported confirmed daily cases of H1N1 pandemic influenza in different regions of India in 2009–10 by region. Shaded areas indicate school holidays. a. India as a whole; b. North-west region; c. South region; d. Mid-East region.

Figure 4

Cumulative absolute incidence of confirmed cases of influenza A (H1N1)pdm’09 virus infection, by month and different states & union territories of India during June’09 to May’10 (till 17^th May).

Figure 5

Estimated reproduction number R_t (posterior median in black line) with 95% CrI (grey zone) for different regions of India: a) India as a whole, b) North-west region and c) South region. A smoothing window of τ=14 days was used.