Assessing countermeasures during a hepatitis A virus outbreak among men who have sex with men

Abstract

Background: A hepatitis A epidemic occurred among men who have sex with men (MSM) in Japan in 2017-2018. In this study, we employ a parsimonious mathematical model to epidemiologically investigate the dynamics of infection, aiming to evaluate the effectiveness of campaign-based interventions among MSM to raise awareness of the situation.

Methods: A mathematical model describing a mixture of human-to-human transmission and environmental transmission was fitted to surveillance data. Taking seasonally varying environmental transmission into account, we estimated the reproduction number of hepatitis A virus during the course of epidemic, and, especially, the abrupt decline in this reproduction number following campaign-based interventions.

Results: The reproduction number prior to the countermeasures ranged from 2.6 to 3.1 and then began to decrease following campaign-based interventions. After the first countermeasure, the reproduction number decreased, but the epidemic remained supercritical (i.e., R_t > 1). The value of R_t dropped well below one following the second countermeasure, which used web articles to widely disseminate information about the epidemic risk.

Conclusions: Although the effective reproduction number, R_t, changes because of both intrinsic and extrinsic factors, the timing of the examined countermeasures against hepatitis A in the MSM population was consistent with the abrupt declines observed in R_t. Even without vaccination, the epidemic was brought under control, and risky behaviors may have been changed by the increase in situation awareness reached through web articles.

Keywords: Awareness; Epidemic; Intervention; Mathematical model; Sexually transmitted infection.

Fig. 1

Epidemic curve of hepatitis A in Japan in 2018 by week, age, and sex distribution. A Temporal distribution of the incidence (serologically diagnosed cases) from 2017 to 2018. B Age distribution of serologically diagnosed cases. The data from 2017 are based on whole-year notifications, but the 2018 data were gathered only up to July 26th. In 2017, the age distribution was fairly flat for women and peaked at 35–39 years for men, and the male-to-female ratio was not particularly high. In contrast, most cases in 2018 occurred in young men aged 20–55 years. C Spatial distribution of serologically diagnosed cases from 2017 to 2018. Many cases were diagnosed in the Kanto region, which includes the Tokyo metropolitan area, followed by Osaka Prefecture, the third largest prefecture in western Japan. D Sex distribution of confirmed cases from 2006-19. Proportion male is measured on right vertical axis

Fig. 2

Interpreting the observed patterns of the epidemic curve. We assumed that the observed epidemic curve could be decomposed into environmental transmission and human-to-human transmission. Environmental transmission, which varies seasonally, corresponds to a sine-like curve. The other curve represents human-to-human transmissions among men who have sex with men

Fig. 3

Decomposition of the epidemic curve. A Comparison of predicted and observed weekly incidence. The solid line is the expected weekly incidence of hepatitis A, and the dotted line represents the empirical data. The gray shaded area represents the 95% confidence interval calculated by the parametric bootstrap method. The three vertical lines represent the times at which (i) human-to-human transmission began to increase, (ii) the first countermeasure was implemented, and (iii) the second countermeasure was implemented. B Decomposed temporal distribution. The gray line shows the predicted incidence of environmental transmission, the orange line represents the predicted incidence of human-to-human transmission, and the blue line represents the mixture of these two mechanisms, obtained as a convolution of the mixed incidence and the incubation period

Fig. 4

Estimated effective reproduction numbers. The solid line shows the reproduction number, and the gray shaded area shows the 95% confidence interval. The reproduction number decreased following the first and second countermeasures and, notably, fell below one following the second countermeasure

Fig. 5

Sensitivity of the relative risk of transmission to serial interval variations. We fixed the median of the serial interval at 27.5 days [10]. In this figure, we varied the variance value, calculating the coefficient of variation to vary from zero to one. The vertical line shows the point that was used in the actual estimation