Predicting the long-term impact of rotavirus vaccination in 112 countries from 2006 to 2034: A transmission modeling analysis

Abstract

Rotavirus vaccination has been shown to reduce rotavirus burden in many countries, but the long-term magnitude of vaccine impacts is unclear, particularly in low-income countries. We use a transmission model to estimate the long-term impact of rotavirus vaccination on deaths and disability adjusted life years (DALYs) from 2006 to 2034 for 112 low- and middle-income countries. We also explore the predicted effectiveness of a one- vs two- dose series and the relative contribution of direct vs indirect effects to overall impacts. To validate the model, we compare predicted percent reductions in severe rotavirus cases with the percent reduction in rotavirus positivity among gastroenteritis hospital admissions for 10 countries with pre- and post-vaccine introduction data. We estimate that vaccination would reduce deaths from rotavirus by 49.1 % (95 % UI: 46.6-54.3 %) by 2034 under realistic coverage scenarios, compared to a scenario without vaccination. Most of this benefit is due to direct benefit to vaccinated individuals (explaining 69-97 % of the overall impact), but indirect protection also appears to enhance impacts. We find that a one-dose schedule would only be about 57 % as effective as a two-dose schedule 12 years after vaccine introduction. Our model closely reproduced observed reductions in rotavirus positivity in the first few years after vaccine introduction in select countries. Rotavirus vaccination is likely to have a substantial impact on rotavirus gastroenteritis and its mortality burden. To sustain this benefit, the complete series of doses is needed.

Keywords: Direct effects; Indirect effects; Partial dose; Rotarix; Rotateq; Rotavirus.

Figure 1.

Model structure. Vaccination is indicated by the red arrows. We also separately tracked vaccinated and unvaccinated populations. Children are born with maternal immunity (M class). After this immunity wanes, children are fully susceptible to rotavirus infection (entering the S₁ class). Upon infection, children become infectious (entering the I class), after which they may either develop long term immunity (entering the R class) or enter a lower susceptibility class (progressing to the next susceptible compartment, S). Secondary, tertiary, and quaternary infections are less likely to result in severe disease or death. After four infections, children enter the long-term immunity class (R). Long-term immunity wanes over time, after which children are fully susceptible to subsequent rotavirus infections (entering the S₁ class). Vaccination is assumed to act like one natural infection, and vaccinated individuals who respond to the vaccine move up one infection class after each dose. Model parameters and equations are shown in the appendix.

Figure 2:

Predicted annual deaths averted by region under the default vaccination scenario. For similar figures for DALYs averted, see Figures A4 and A5.

Figure 3:

Deaths averted by age group, region, and vaccine coverage scenario over time for A) 0-3 year olds and B) 3-5 year olds. Solid lines and circles show predicted impacts for the default scenario and dashed lines and open square shapes show impacts for the best-case scenario. See Figure A6 for plots of total deaths stratified by yearly age groups.

Figure 4.

Direct vaccine effects only (dashed line) and overall vaccines effects (solid line) up to 12 years post vaccine introduction in PINE countries. Panels show A) Pakistan, B) India, C) Nigeria, and D) Ethiopia.

Figure 5.

A comparison of the impact of a one- vs. two-dose vaccine schedule for the first 12 years after vaccine introduction. A) The percent of deaths averted, and B) the relative performance of a one vs. two dose schedule.

Figure 6:

Predicted vaccine impacts on annual severe rotavirus cases for children under 1 year of age (model) and percent reduction in rotavirus positivity (GRSN data) (y-axis) by number of years since vaccine introduction (x-axis) for 10 countries (shown in different panels). Modeled impacts are shown in solid lines, with the grey ribbon showing 95% simulation intervals. Impacts based on surveillance data are shown with points, with error bars to show uncertainty (error bars were calculated using standard methods, assuming that the percent reduction in positivity approximated a risk ratio).

Figure 7.

Predicted vaccine impacts on annual severe rotavirus cases for children under 5 years of age (model) and percent reduction in rotavirus positivity (GRSN data) (y-axis) by number of years since vaccine introduction (x-axis) for 10 countries (shown in different panels). Modeled impacts are shown in solid lines, with the grey ribbon showing 95% credible intervals. Impacts based on GRSN surveillance data are shown with points, with error bars to show uncertainty (error bars were calculated using standard methods, assuming that the percent reduction in positivity approximated a risk ratio).