Investigation of Pertussis Resurgence in King County, Washington

Introduction: Pertussis or “whooping cough” is an infectious disease caused by bacterium Bordetella pertussis. Despite widespread vaccination since the 1950s, it continues to cause epidemics in several countries. In fact, in the last three decades, many countries with high vaccination coverage have reported a resurgence in pertussis activity. Understanding the factors that led to pertussis resurgence and its persistence is important because pertussis remains a leading cause of infant mortality worldwide with close to 100,000 deaths annually (WHO, 2018). This study examines the role of the Diphtheria-Tetanus-acellular-Pertussis (DTaP) vaccination coverage and vaccine failure in pertussis resurgence and persistence. Specifically, we described long-term, fine-scale spatio-temporal trends of pertussis incidence and characterized spatial dependence in pertussis cases within King County (Aim 1); estimated the association between pertussis epidemics and area-level vaccination coverage and socio-demographic factors (Aim 2); assessed the role of vaccine schedules and vaccine timeliness on pertussis incidence (Aim 3); estimated the population-level effects of acellular pertussis vaccines (Aim 4); and estimated waning effects of the acellular pertussis vaccine using Schoenfeld residuals (Aim 5) Methods: We obtained pertussis incidence data for all ages for the period between 1999 and 2017 from the Communicable Disease and Immunizations Department within Public Health Seattle and King County. Diphtheria-Tetanus-acellular-Pertussis (DTaP) vaccination records, as well as records for all pediatric vaccinations, for all children born or living in King County, WA, between 2008 and 2017 were obtained from the Washington State Immunization Information System (WA-IIS) maintained by the Washington State Department of Health. All five aims use information from either one or both datasets. These datasets were linked using probabilistic linkage methods to obtain DTaP vaccination and pertussis status for the study participants. Information on census-tract level and school-district level socio-demographic factors was obtained from the US Census and the National Historical Geographic Information System (NHGIS) databases. Aim 1 included pertussis incidence data for all ages reported between 1999 and 2017 and used Bayesian hierarchical disease mapping models and the tau statistic to characterize spatio-temporal dependence between pertussis cases. The Kulldorff spatial scan statistic was used to examine location of pertussis clusters and their overlap with clusters of non-medical vaccine exemptions. For aim 2, we used pertussis incidence data for all ages reported between 2010 to 2017 and estimated annual school-district level vaccination coverage as proportion of 19-35 month old children who received ≥4 DTaP doses using immunization data from the WA-IIS. Association between pertussis epidemics and vaccination coverage and other socio-demographic factors was estimated using epidemic-endemic models and the ecological vaccine model. For aims 3, 4, and 5, we used the linked dataset with individual level vaccination and pertussis status for all children born or living in King County between 2008 and 2017. Log binomial models were used to estimate the association between DTaP vaccination schedules and age-specific pertussis incidence. Cox proportional hazards models were used to estimate vaccine direct effects among children older than 3 months and population-level effects among children older than 7 months. Schoenfeld residuals obtained from fitting Cox proportional hazards models were used to estimate waning of vaccine effectiveness among 5-9 year old children. Results: There was no overall increase in pertussis incidence between 1999 and 2017, but we found spatial dependence between pertussis cases at very small spatial scales. Pertussis clusters overlapped with clusters of vaccine refusal suggesting an association between the two. We estimated the vaccine effectiveness of DTaP vaccine to be 83% (95% credible intervals: 63%, 95%) using the ecological vaccine model but found no correlation between the effective reproduction number of pertussis and area-level vaccine coverage. The association between area-level under-vaccination and pertussis epidemics was statistically significant as estimated using the epidemic-endemic models (adjusted Relative Risk, aRR: 2.76; 95% confidence interval: 1.44, 16.6), suggesting areas with low vaccination coverage had higher risk of experiencing pertussis outbreaks. We found significant association between under-vaccination and age-specific pertussis risk, but a short delay of a few weeks in receiving DTaP doses did not significantly alter pertussis risk. Using the Cox proportional hazards models and DTaP vaccine series as a time-dependent exposure, direct vaccine effects were estimated to be 72% (95% CI: 65%, 77%) comparing vaccinated time at risk to under-vaccinated time at risk for the entire cohort. The estimated indirect protection for the 3-dose primary series was 45% (95% CI: 1%, 70%), total protection was 94% (95% CI: 91%, 96%), and overall protection was 42.2% (95% CI: 19%, 60%). We found no evidence of waning of vaccine effectiveness after 5 doses of DTaP among 5-9 year old children. Vaccine effectiveness remained high at 83% (95% CI: 39%, 95%) four years after vaccination with 5th DTaP dose. Conclusion: Our findings show that although pertussis transmission is ongoing in King County, there is no clear evidence for resurgence between 1999 and 2017 as seen in the rest of the country. We found that the current schedule for the 5-dose childhood DTaP vaccine series effectively reduces pertussis risk and adding or delaying booster doses may not be required. We estimated direct vaccine effectiveness (VE) of the acellular pertussis vaccine using different models in this dissertation and found that the estimates were high and consistent across analyses. However, direct VE estimates from the statistical models used in this dissertation do not provide information about mechanism of vaccine failure (i.e.\ leaky vs.\ all-or-none). We found significant vaccine indirect effects for the acellular vaccine suggesting that vaccination with DTaP may contribute to herd effects, although we cannot deduce from these data if the indirect effects are due to decrease in susceptibility to infection or decrease in infectiousness after exposure among vaccinated individuals. We found no evidence of rapid waning of vaccine effects among children who were fully vaccinated with 5 doses of DTaP, suggesting that immediate waning of vaccine effects is likely not the mechanism of vaccine failure for the DTaP vaccine. Understanding the nature of the acellular pertussis vaccine failure in shaping the epidemiology of pertussis is challenging and deserves continued research.