Acellular Pertussis Vaccine Effectiveness Over Time

The study setting was Kaiser Permanente Northern California (KPNC), an integrated health care delivery organization that provides comprehensive care to ∼4 million members. Members receive almost all medical care at KPNC-owned facilities, including clinics, hospitals, pharmacies, and laboratories. KPNC databases are used to capture detailed information on all medical services, including vaccinations and laboratory tests, as well as on enrollment and demographics. KPNC members are similar to the broad catchment population in Northern California in terms of sociodemographic characteristics, except the extremes of income distribution are underrepresented.²⁸ Members receive all their routine vaccinations free of charge.

The current study included children who were born between 1999 and 2016 and were members of KPNC at 2 months of age. Children born before 1999 were excluded to restrict the study population to those receiving only acellular pertussis vaccines. We also excluded children with culture- or polymerase chain reaction (PCR)-confirmed pertussis before study entry. Children had to be continuously enrolled in KPNC starting at 2 months of age to ensure accurate ascertainment of DTaP vaccination status.

The study was a retrospective cohort study. The outcome was pertussis identified by a real-time PCR test for B pertussis. Since 2006, nearly all pertussis testing is conducted in a centralized laboratory by using PCR on nasopharyngeal swabs. Eligible children were followed starting on January 1, 2006, or when they were 3 months of age, whichever was later, and continued until they tested PCR-positive for pertussis, disenrolled from KPNC, received their Tdap vaccine, turned 11 years of age, or the end of the study period on June 30, 2017.

The main exposure of interest was DTaP vaccination status, a time-varying covariate with 4 levels (unvaccinated, undervaccinated, fully vaccinated, or fully vaccinated plus 1 dose). We categorized DTaP vaccination status on the basis of the number of DTaP doses received in relation to the number of DTaP doses expected according to the ACIP-recommended ages for the 5-dose series (2, 4, 6, and 12–18 months and 4–6 years). ACIP-recommended ages for DTaP did not change during our study period.

We classified children as fully vaccinated if they received the expected number of DTaP doses by 1 month after the ACIP-recommended age (ie, 1 dose by 3 months, 2 doses by 5 months, 3 doses by 7 months, 4 doses by 19 months, and 5 doses by 84 months of age). Children with fewer doses than expected for their age were undervaccinated, and children with 1 more dose than expected were fully vaccinated plus 1 dose. We created the fully vaccinated plus 1 dose category so that children who had all their recommended doses were not combined with those who had more than the recommended doses because their risk of pertussis may be different (Table 1). Children are considered age-appropriately vaccinated if they are in either the fully vaccinated or the fully vaccinated plus 1 dose category.

We counted DTaP doses starting 8 days after receipt to allow time for the vaccine to take effect.²⁹ We censored children who had 2 more doses than expected (eg, 4 doses by 5 months of age) or children ≥7 years of age with 6 doses.

We updated DTaP vaccination status during follow-up as children aged or received additional DTaP doses. For example, at ages 3 to <5 months, children with 1 DTaP dose were fully vaccinated, and those with 2 DTaP doses were fully vaccinated plus 1 dose; at ages 5 to <7 months, children with 1 dose were undervaccinated, those with 2 doses were fully vaccinated, and those with 3 doses were fully vaccinated plus 1 dose. Children with no DTaP doses were classified as unvaccinated at all ages (Table 1).

We examined the risk of pertussis in relation to DTaP vaccination status using a series of Cox regression models. The models included a 4-level vaccination indicator variable to estimate the pertussis hazard ratios (HRs) for children who were unvaccinated, undervaccinated, or fully vaccinated plus 1 dose compared with children who were fully vaccinated. The Cox models were specified with a calendar time line and stratified by the year and month of birth, which, together, adjust for the month of age. We included covariates to adjust for sex, race or ethnicity, type of insurance (Medicaid or other subsidized insurance versus commercial insurance), and the medical clinic where children received most of their care. The last 2 covariates were time-varying and updated monthly (insurance type) or quarterly (medical clinic). We estimated HRs for the entire study population over the entire follow-up period. We also estimated HRs for the following age groups: 3 to <5, 5 to <7, 7 to <19, 19 to <84, and 84 to 132 months. Each of these age groups has a different number of DTaP doses (ranging from 1 to 5) corresponding to fully vaccinated status. For the youngest and oldest age groups, the model included a 3-level rather than a 4-level vaccination indicator variable (Table 1). We estimated vaccine effectiveness (VE) as 1 minus the adjusted hazard ratios (aHRs) using these same models, except that unvaccinated children constituted the reference group.

In each of the 2 oldest age groups of children, we examined waning of vaccine immunity over time. We restricted the analyses to children who had received at least 1 dose of DTaP and estimated the risk of pertussis in relation to the number of years since the last DTaP dose, adjusting for DTaP vaccination status. Years since last DTaP dose was a time-dependent covariate that was divided into categories and updated as children got further from their last DTaP dose or received a new DTaP dose. For children ages 19 to <84 months, there were 4 categories: <1 year (reference), 1 to <2 years, 2 to <3 years, and ≥3 years. For children ages 84 to 132 months, there were 5 categories: <3 years (reference), 3 to <4 years, 4 to <5 years, 5 to <6 years, and ≥6 years. We estimated the pertussis HR for each category for years since vaccination in relation to the reference category with a Cox model like those described above. We used a calendar time line stratified by the year and month of birth and included the same covariates. We also conducted analyses in which we did not control for the year and month of birth because years since last DTaP dose and age are closely correlated.

We conducted all analyses with SAS version 9.3 (SAS Institute, Inc, Cary, NC). We used the Lexis macro to partition person-time (http://bendixcarstensen.com/Lexis/Lexis.sas).

The study was approved by the KPNC Institutional Review Board with a waiver of written informed consent because the study had no direct contact with study participants.

The study included 469 982 children born between 1999 and 2016 and followed them from 2006 to 2017. The average duration of follow-up was 4.6 years per child. Most children were born at a KPNC hospital (89%) and entered the study at 3 months of age (73%) (Table 2). During follow-up, the study population contributed 2 138 835 person-years (P-Ys) distributed as follows by DTaP vaccination status: 1% unvaccinated, 3% undervaccinated, 70% fully vaccinated, and 26% fully vaccinated plus 1 dose (Table 3). We identified 738 PCR-confirmed pertussis cases classified into the following categories according to their vaccination status: 99 (13%) unvaccinated, 36 (5%) undervaccinated, 515 (70%) fully vaccinated, and 88 (12%) fully vaccinated plus 1 dose.

Overall incidence rates of pertussis among children in the study population varied from 389 out of 100 000 P-Ys for unvaccinated children to 16 out of 100 000 P-Ys for fully vaccinated plus 1 dose children. Incidence rates of pertussis also varied by vaccination status within each age category (Table 3).

After adjustment for covariates, pertussis risk was 13 times higher among unvaccinated children (aHR = 13.53; 95% confidence interval [CI] 10.64–17.21) and 1.9 times higher among undervaccinated children (aHR = 1.86; 95% CI 1.32–2.63) compared with fully vaccinated children across all follow-up in all age groups. Children who were fully vaccinated plus 1 dose had a lower risk of pertussis compared with fully vaccinated children (aHR = 0.48; 95% CI 0.34–0.68; Table 4).

Risk of pertussis according to DTaP vaccination status varied somewhat by age. In each age group, unvaccinated children were at a significant increased risk for pertussis compared with fully vaccinated children, ranging from 4 times higher (aHR = 4.54; 95% CI 1.62–12.69) for children ages 3 to <5 months to 23 times higher (aHR = 23.62; 95% CI 16.32–34.17) for children ages 19 to <84 months. Undervaccinated children ages 5 to <7 and 19 to <84 months were also at significantly increased risk for pertussis compared with fully vaccinated children, and children who were fully vaccinated plus 1 dose at ages 7 to <84 months were at significantly reduced risk compared with the fully vaccinated reference group (Table 4).

Across all follow-up and all age groups, VE was 86% (95% CI 80%–91%) for undervaccinated children compared with unvaccinated children. VE was even higher for fully vaccinated children and for those who were fully vaccinated plus 1 dose. VE = 93% (95% CI: 91%–94%) for fully vaccinated children and VE = 96% (95% CI: 95%–97%) for children who were fully vaccinated plus 1 dose (Supplemental Table 6).

Vaccinated children who were further away from their last DTaP dose were at increased risk of pertussis. Among children ages 19 to <84 months, the crude incidence rate of pertussis rose from 11 to 32 out of 100 000 P-Ys when time since last DTaP dose increased from <1 to ≥3 years. Adjusting for vaccination status and other covariates, including the child’s age, risk of pertussis was 5 times higher for children whose last DTaP dose was ≥3 years before compared with those whose last dose was <1 year before (aHR = 5.04; 95% CI 1.84–13.80; Table 5). Adjusting for the same covariates except the child’s age, risk of pertussis was >2 times higher for children whose last DTaP dose was ≥3 vs <1 year before (aHR = 2.58; 95% CI 1.49–4.46; Table 5).

Similarly, among fully vaccinated children ages 84 to 132 months, crude incidence rates of pertussis rose from 24 to 133 out of 100 000 P-Ys when time since last DTaP dose increased from <3 to ≥6 years. After adjusting for covariates, including age, the risk of pertussis was >2 times higher ≥6 years after the last DTaP vaccination compared with <3 years (aHR = 2.32; 95% CI 0.97–5.59; Table 5). After adjusting for all covariates except age, the risk of pertussis was >4 times higher ≥6 years after the last DTaP vaccination compared with <3 years (aHR = 4.66; 95% CI 2.81–7.71; Table 5).

Children ages 3 months to 11 years who did not receive any DTaP vaccine or who were undervaccinated according to the ACIP-recommended schedule were at increased risk of pertussis compared with children who were fully vaccinated. Although noncompliance with the vaccination schedule is an important public health problem that leads to increased pertussis risk, most children in our study received all their recommended DTaP doses. Unvaccinated and undervaccinated children accounted for only 4% of P-Ys of follow-up in our study population. Furthermore, most undervaccinated children had only 1 fewer DTaP dose than recommended for their age so that <2% of follow-up was spent missing ≥2 doses. Thus, despite the pertussis risk being 13 times higher among unvaccinated children and the risk 1.9 times higher among undervaccinated children, these 2 groups together comprised <20% of pertussis cases.

The majority of pertussis cases (>80%) in our study population occurred among children who had received all their recommended DTaP doses. Children who were further away from their last DTaP dose were at increased risk of pertussis. Our results reveal that waning of DTaP immunity was an important cause of pertussis in children >18 months of age who have a longer interval between recommended doses. By age 19 months, nearly all children had received at least 1 DTaP dose, and children who were age-appropriately vaccinated accounted for >95% of follow-up.

Although undervaccinated children were at a much lower risk than unvaccinated children, their risk was twice that of fully vaccinated children. These results are broadly consistent with 2 previous case-control studies. Authors of 1 study, which included 72 children PCR-positive for pertussis and 288 matched controls, found a significant increased risk for pertussis in undervaccinated children aged 3 to 36 months,¹⁵ whereas authors of another study evaluated 145 cases and 2900 controls and found undervaccination associated with a more than twofold increased risk of pertussis among children 3 to 35 months old.³⁰ In both studies, undervaccination was defined on the basis of the number of missing DTaP doses in relation to the number of recommended doses. Children who were age-appropriately vaccinated were the comparison group. The results for waning of DTaP immunity in this study are consistent with our 2 previous studies, although we use different methods and measures of waning.^22,23 Our results are also consistent with those of a meta-analysis of 11 studies in which authors found the odds of pertussis increased by 33% for every additional year after the third or the fifth DTaP dose.²⁴ 

The study strengths include its large, racially diverse population and precise data on the number and timing of DTaP doses. The large population allowed us to analyze pertussis risk relative to vaccination status separately by age group for each level of recommended DTaP doses. We also carefully adjusted for calendar date and age as we examined pertussis risk in relation to vaccination status. Because we used Cox models on a calendar time line, pertussis cases were only compared with other children at risk on the same day who have nearly the same age.

Our study has some limitations. We could not fully disentangle the effects on risk of pertussis of age and time since last DTaP dose because these 2 factors are highly correlated. In our primary analyses, we closely adjusted for months of age by comparing children who were the same age in months, which resulted in minimal variability in time since last DTaP doses and reduced precision of our estimates. Although we likely captured most pertussis cases in our study, we may have missed mild cases that did not come to medical attention or cases that presented >2 weeks after cough onset.³¹ In addition, PCR testing may have misclassified pertussis status for a few individuals; however, such misclassification was unlikely to be related to time since vaccination or to vaccination status.

Although waning immunity is clearly an important factor driving pertussis epidemics in recent years, other factors that we did not evaluate in this study might also contribute to pertussis epidemics individually or in synergy. Results from studies in baboons suggest that the acellular pertussis vaccines are unable to prevent colonization, carriage, and transmission.^32,–34 If this is also true for humans, this could contribute to pertussis epidemics. The causes of recent pertussis epidemics are complex, and we were only able to address some aspects in our study.

Compared with fully vaccinated children, pertussis risk was 13 times higher among unvaccinated children and 1.9 times higher among undervaccinated children. Although undervaccinated and especially unvaccinated children were at greater risk of pertussis, they represented a small fraction of all pertussis cases. Within our study population, >80% of pertussis cases occurred among age-appropriately vaccinated children. Children who were further away from their last DTaP dose were at increased risk of pertussis, even after controlling for undervaccination. Our results suggest that in this population, possibly in conjunction with other factors not addressed in this study, suboptimal VE and waning played a major role in recent pertussis epidemics.

ACIP

Advisory Committee on Immunization Practice

aHR

adjusted hazard ratio

CI

confidence interval

DTaP

diphtheria-tetanus-acellular pertussis

HR

hazard ratio

KPNC

Kaiser Permanente Northern California

PCR

polymerase chain reaction

P-Y

person-year

Tdap

tetanus toxoid, reduced diphtheria toxoid, and acellular pertussis, adsorbed

VE

vaccine effectiveness