Heterogeneity in norovirus shedding duration affects community risk

Abstract

Norovirus is a common cause of gastroenteritis in all ages. Typical infections cause viral shedding periods of days to weeks, but some individuals can shed for months or years. Most norovirus risk models do not include these long-shedding individuals, and may therefore underestimate risk. We reviewed the literature for norovirus-shedding duration data and stratified these data into two distributions: regular shedding (mean 14-16 days) and long shedding (mean 105-136 days). These distributions were used to inform a norovirus transmission model that predicts the impact of long shedders. Our transmission model predicts that this subpopulation increases the outbreak potential (measured by the reproductive number) by 50-80%, the probability of an outbreak by 33%, the severity of transmission (measured by the attack rate) by 20%, and transmission duration by 100%. Characterizing and understanding shedding duration heterogeneity can provide insights into community transmission that can be useful in mitigating norovirus risk.

Fig. 1.

Model schematic. S, Susceptible; E, exposed; I, infected; R, recovered; subscript (_R), regular shedding; subscript (_L), long shedding. Parameters: β, transmission probability; ε, 1/incubation period; γ_R, 1/regular-shedder infectious period; γ_L, 1/long-shedder infectious period; ρ, fraction that is long-shedding.

Fig. 2.

Shedding lengths and fitted gamma distributions of individual shedding duration data from empirical review. Data are truncated at 200 days; the figure does not include seven individuals who shed for >200 days (up to 898 days). Grey bars indicate operational long shedders (infants and immunocompromised individuals) and white bars indicate operational regular shedders (immunocompetent non-infants). The dotted verticalline at 34 days delineates the cut-off point for the functional definition. Dashed lines represent fitted gamma distributions for the operational definition (regular-shedder parameters = 2·2, 7·4; long-shedder parameters = 0·8, 129·1) and solid lines are gamma distributions fitted to the functionally defined populations, where regular shedders <34 days, long-shedders ⩾34 days (regular-shedder parameters = 2·7, 5·3; long-shedder parameters = 0·7, 199·4).

Fig. 3.

Outbreak duration and severity using shedding length parameter values from our empirical review (Table 2). Only model runs resulting in high transmission events (>200 cases) are shown. Runs with both long-shedding and regular-shedding groups are represented by squares, runs with only regular shedders are represented by circles. Filled shapes indicate use of the operational (a priori) definition and open shapes indicate use of the functional definition (>34 days) for long shedders. Black lines show the mean and standard distribution for each scenario.

Fig. 4.

Sensitivity of outbreak probability to variable long-shedding duration (1/γ_L). Outbreaks are defined by having >200 cases. Dashed lines show the probability for each definition in the absence of long shedders, and dotted lines represent the shedding duration of the regular-shedding groups (γ = 16·4 days, γ = 14·5 days). Open circles represent use of the operational definition, and filled grey circles represent use of the functional definition for stratification.

Fig. 5.

Sensitivity of outbreak severity to variable long-shedding duration (1/γ_L). Outbreaks are defined by having >200 cases. Dashed line shows average and standard definition for 1000 runs with the functional definition (solid line) shows average and standard deviation for operational definition. Dotted lines show average number of cases in the absence of long shedders.

Fig. 6.

Sensitivity of outbreak duration to variable long-shedding duration (1/γ_L). Outbreaks are defined by having >200 cases. End of outbreak occurs when there are no exposed or infected individuals remaining. Dotted lines show average duration for each definition in the absence of long shedders.