Association Between the Decline in Pneumococcal Disease in Unimmunized Adults and Vaccine-Derived Protection Against Colonization in Toddlers and Preschool-Aged Children

Abstract

Vaccinating children with pneumococcal conjugate vaccine (PCV) disrupts transmission, reducing disease rates in unvaccinated adults. When considering changes in vaccine dosing strategies (e.g., removing doses), it is critical to understand which groups of children contribute most to transmission to adults. We used data from Israel (2009-2016) to evaluate how the buildup of vaccine-associated immunity in children was associated with declines in invasive pneumococcal disease (IPD) due to vaccine-targeted serotypes in unimmunized adults. Data on vaccine uptake and prevalence of colonization with PCV-targeted serotypes were obtained from children visiting an emergency department in southern Israel and from surveys of colonization from central Israel. Data on IPD in adults were obtained from a nationwide surveillance study carried out in Israel. We compared the trajectory of decline of IPD due to PCV-targeted serotypes in adults with the decline of colonization prevalence and increase in vaccine-derived protection against pneumococcal carriage among different age groupings of children. The declines in IPD in adults were most closely associated with the declines in colonization and increased vaccination coverage among children in the age range of 36-59 months. This suggests that preschool-aged children, rather than infants, are responsible for maintaining the indirect benefits of PCVs.

Figure 1.

Changes in the prevalence of carriage of vaccine-targeted pneumococcal serotypes following the introduction of vaccine, Israel, 2009–2016. A) Observed proportion of nasopharyngeal swabs that were positive for 7-valent pneumococcal conjugate vaccine (PCV7) serotypes among Jewish children under 5 years of age (in 1-year age groups), by epidemiologic year (July to June). B) Smoothed proportion of children carrying PCV7 serotypes in each month during the period 2009–2016.

Figure 2.

Numbers of cases of invasive pneumococcal disease (IPD) due to 7-valent pneumococcal conjugate vaccine (PCV7) serotypes among Jewish adults, Israel, July 2009–June 2016. A) Adults aged 18–39 years; B) adults aged 40–64 years; C) adults aged ≥65 years. The observed number of cases in each age group is indicated by dots. The black line shows the smoothed trend (fitted with a generalized additive model with a spline for time and monthly dummy variables).

Figure 3.

Increase in the population direct effect of 7-valent pneumococcal conjugate vaccine against pneumococcal carriage for Jewish children under 5 years of age (in 1-year age groups), Israel, 2009–2016. The population direct effect represents the vaccine effectiveness that would be expected if there were no effect of the vaccine on transmission. Estimates were smoothed with splines prior to plotting.

Figure 4.

Relative goodness of fit of regression models evaluating the association between the population direct effect (PDE) of 7-valent pneumococcal conjugate vaccine in children of different age ranges (indicated by the horizontal bars) and invasive pneumococcal disease (IPD) in adults, Israel, 2009–2016. A) Adults aged 18–39 years; B) adults aged 40–64 years; C) adults aged ≥65 years. Each horizontal bar indicates a specific age range in which the PDE was calculated on the basis of uptake of the vaccine in that age group and the expected efficacy against colonization. The vertical position of the bar along the y-axis indicates the goodness of fit, as measured by the likelihood of the model given the data in comparison with the best-fitting model (which had a relative likelihood of 1). These values were calculated from the Akaike Information Criterion (AIC) score. The PDE in age ranges that are placed higher on the y-axis fit the adult IPD data better. Black bars above the dotted line were not meaningfully different from the best-fit model (AIC score within 2 points).

Figure 5.

Relative goodness of fit of regression models evaluating the association between the population direct effect of 7-valent pneumococcal conjugate vaccine in children of different age ranges and invasive pneumococcal disease in adults, Israel, 2009–2016. A) Adults aged 18–39 years; B) adults aged 40–64 years; C) adults aged ≥65 years. The goodness-of-fit (model likelihood) values were averaged across all of the models in which the population direct effect included the indicated pediatric age.

Figure 6.

Relative goodness of fit of regression models evaluating the association between the prevalence of 7-valent pneumococcal conjugate vaccine serotypes among healthy children in different age ranges (indicated by the horizontal bars) and invasive pneumococcal disease (IPD) in adults, Israel, 2009–2016. A) Adults aged 18–39 years; B) adults aged 40–64 years; C) adults aged ≥65 years. Each horizontal bar indicates an age range in which (smoothed) carriage prevalence was calculated. The vertical position of the bar along the y-axis indicates the goodness of fit, as measured by the likelihood of the model given the data in comparison with the best-fitting model (which had a relative likelihood of 1). These values were calculated from the Akaike Information Criterion (AIC) scores. Carriage prevalence in age ranges that are placed higher on the y-axis fit the adult IPD data better. Black bars above the dotted line were not meaningfully different from the best-fit model (AIC score within 2 points).

Figure 7.

Relative goodness of fit of regression models evaluating the association between the prevalence of 7-valent pneumococcal conjugate vaccine serotypes among healthy children in different age ranges and invasive pneumococcal disease in adults, Israel, 2009–2016. A) Adults aged 18–39 years; B) adults aged 40–64 years; C) adults aged ≥65 years. The goodness-of-fit (model likelihood) values were averaged across all of the models in which the population direct effect included the indicated pediatric age.