Date of Award

Fall 2020

Document Type

Open Access Dissertation

Department

Health Services and Policy Management

First Advisor

Nicole L. Hair

Abstract

Introduction

2009 H1N1 Pandemic: The historical 2009 H1N1 Influenza pandemic, which had a CDC estimated accrued disease burden of 100.5 million illnesses, 936,000 hospitalizations, and 75,000 deaths from 2009 to 2018, resulted in a declared state of emergency nationally, with ensuing diminished vaccine confidence and amplified fears of infection, prompting some to pursue flu vaccination, and others to forego. Although the Centers for Disease Control and Prevention (CDC), and its Advisory Committee on Immunization Practices (ACIP) recommend an annual flu vaccine for individuals 6 months of age and older as the “first and best” defense against influenza, a low percentage of children are vaccinated, and parental decisions are not fully understood. Examining previous literature, a void exists in relation to parental perceptions and decisions for child immunizations, particularly concerning the U.S. nationally, with most studies being international. Furthermore, there is evidence of varied results with inadequate and conflicting conclusions, specifically for children. 2015 LAIV Policy Shift: The 2015 Centers for Disease Control and Prevention’s (CDC) Advisory Committee on Immunization Practices (ACIP) retraction of its original preferential recommendations for usage of the live attenuated influenza vaccine (LAIV), which is the intra-nasal version of the vaccine, has resulted in varied responses, with fluctuations in ensuing CDC vaccine advisements affecting its implementation and uptake among children. Although the CDC’s ACIP recommend an annual flu vaccine for individuals 6 months of age and older as the “first and best” defense against influenza, a low percentage of children are vaccinated, and parental decisions are not completely comprehended, particularly in regards to the LAIV formulation. Reviewing the literature, in certain studies a decline in flu vaccine uptake was concluded, whereas in other instances, it conversely increased, or remained static, yielding inconsistent outcomes. Furthermore, there exists a great void in the number, scale, and scope of studies published, with none being nationally representative, and examining parental perspectives, decisions, and responses in regard to child flu vaccine uptake following the 2015 ACIP LAIV policy shift.

Methods

2009 H1N1 Pandemic: To assess impacts of the 2009 H1N1 pandemic on decisions to uptake influenza vaccines for children age 6 months to 17 years of age, data from NIS was used as a series of weighted consecutive annual surveys in order to synthesize a longitudinal panel dataset spanning from 2003 to 2018. Population adjusted measures of influenza like illness (ILI) by state and season procured from CDC’s FluView application and ILI Net from 2008 to 2018 was used in order to supplement the primary NIS dataset. Quasi-experimental (QE) approaches in the form of segmented interrupted time series (ITS), and fixed effects model (FEM) logistic estimations were executed on the integrated dataset yielding logistic regression coefficients and post-estimation marginal effects signifying the impact of the pandemic on child influenza vaccine uptake (CIVU). ITS regressions examined both level and trend changes due to pandemic occurrence via binary and continuous pandemic incidence variables respectively. FEM regressions examined fluctuations in CIVU as a function of influenza disease progression across seasons and geographic jurisdictions. 2015 LAIV Policy Shift: To assess impacts of the 2015 ACIP LAIV preferential recommendation revocation on decisions to uptake influenza vaccines for children age 6 months to 17 years of age, data from NIS was used as a series of weighted consecutive annual surveys in order to synthesize a longitudinal panel dataset spanning from 2003 to 2018. Quasi-experimental (QE) approaches in the form of segmented interrupted time series (ITS), and difference in differences (DID) logistic estimations were executed on the integrated dataset yielding logistic regression coefficients and post-estimation marginal effects signifying the impact of the 2015 ACIP LAIV policy shift on child influenza vaccine uptake (CIVU). ITS regressions examined both level and trend changes due to policy shift occurrence via binary and continuous policy shift incidence variables respectively. DID regressions incorporated LAIV eligibility indicators to ascertain the level and trend differences in CIVU between LAIV eligible (age 2 years and greater), and LAIV ineligible (age 6 to 23 months) individuals, pre and post policy shift. This additionally allowed for ascertainment of spillover effects and impacts of the policy shift on individuals who were only eligible for the injected influenza vaccine (IIV) formulation. Vaccine specific ITS estimations for individual formulations were executed applying previous procedures, in addition to regressions assessing heterogeneity effects.

Results

2009 H1N1 Pandemic: The interrupted time series (ITS) regression for the NIS-Child sample yielded statistically significant coefficients. Post-estimation average marginal effects (AMEs) were as follows. The H1N1 pandemics occurrence yielded a 12.57 percentage point (pp), 95% CI [10.28, 14.32], immediate level change increase in the probability of a child being immunized, on average. It also yielded a 3.77 pp, 95% CI [-4.32, -2.55], sustained slope change decrease in the probability of a child being immunized annually, on average. Pre-pandemic, a 1.64 pp, 95% CI [1.47, 1.81], sustained increase in the probability of a child being immunized annually, on average, was evident. Restricted scale epidemic (RSE) occurrences of the influenza virus yielded post-estimation AMEs that were statistically significant for RSEs on 2012, 2013, and 2014. These coefficients were a 1.79 pp, 95% CI [-2.22, 0.38], 5.23 pp, 95% CI [-6.27, -4.77], and 1.92 pp, 95% CI [2.74 1.10], decrease in the probability of a child being immunized, on average, respectively. The respective trend change increases post RSE occurrences were 0.85 pp, 95% CI [0.74, 0.96], 0.34 pp, 95% CI [0.28, 0.40], and 1.24 pp, 95% CI [1.12 1.35], on average, in the probability of the same outcome. Sensitivity analysis fixed effects model (FEM) regressions yielded logit and AME coefficients that were statistically insignificant with the exception of a single variable in subgroup 5, which indicated a decrease of 2.29 pp, on average, in immunization rates during peak season weeks registering at a ILI intensity magnitude of 9 or greater. FEM regressions for the NIS-Teen sample yielded logit and AME coefficients that were statistically insignificant with the exception of three variables in subgroup 5. The initial variable indicated a 1.31 pp increase, and the subsequent variables indicated a 0.135 pp, and a 0.212 pp decrease, on average, in immunization rates respectively. 2015 LAIV Policy Shift: The interrupted time series (ITS) regression for the NIS-Child sample yielded statistically significant coefficients. Post-estimation average marginal effects (AMEs) were as follows. The LAIV preferential recommendation revocation yielded a 3.01 percentage point (pp), 95% CI [2.54, 4.74], immediate level change increase in the probability of a child being immunized, on average. It also yielded a 2.41 pp, 95% CI [-2.62, -2.11], sustained slope change decrease in the probability of a child being immunized annually, on average. Pre-policy shifts, a 2.06 pp, 95% CI [1.91, 2.22], sustained increase in the probability of a child being immunized annually, on average, was evident. The LAIV preferential recommendation of 2014, and the subsequent LAIV recommendation rescindment of 2016, respectively yielded a 5.25 pp decrease, 95% CI [-7.05, -3.25], and a 1.02 pp increase, 95% CI [0.55, 1.12], in the probability of a child being immunized, on average. The respective trend changes post-policy shifts were a 1.21 pp increase, 95% CI [1.11, 1.31], and a 5.30 pp decrease, 95% CI [-6.22, -4.38], on average. The sensitivity analysis difference in differences (DID) estimation yielded statistically significant coefficients. Comparing the differences between LAIV-eligible and LAIV-ineligible individuals, pre and post 2015 policy shift, yielded a DID of 20.70 pp, 95% CI [19.52, 21.88], indicating an increase occurred in the probability of a LAIV-eligible child being immunized as compared to an LAIV-ineligible child, on average, following the 2015 policy shift. Examining the LAIV-eligibility indicator’s AME, it is evident that an LAIV-eligible child experiences a 1.34 pp, 95% CI [0.64, 2.03], increase in the probability of being immunized, on average, as compared to an LAIV-ineligible child. The interrupted time series (ITS) regression for the NIS-Teen sample yielded statistically significant coefficients. Post-estimation AMEs are as follows. The LAIV preferential recommendation revocation yielded a 4.25 pp, 95% CI [2.31, 6.22], immediate level change increase in the probability of a teen being immunized, on average. It also yielded a 3.02 pp, 95% CI [-4.77, -2.33], sustained slope change decrease in the probability of a teen being immunized annually, on average. Pre-policy shifts, a 2.70 pp, 95% CI [2.12, 3.16], sustained increase in the probability of a teen being immunized annually, on average, was evident. The LAIV preferential recommendation of 2014, and the subsequent LAIV recommendation rescindment of 2016, respectively yielded a 8.41 pp decrease, 95% CI [-10.35, -6.41], and a 6.52 pp decrease, 95% CI [-8.21, -4.42], in the probability of a teen being immunized, on average. The respective trend changes post-policy shifts were a 7.17 pp increase, 95% CI [6.11, 8.58], and a 2.84 pp increase, 95% CI [1.96, 3.71], on average.

Conclusion

2009 H1N1 Pandemic: Preliminary escalations in the probability of child immunization uptake are evident following the pandemic. This is possibly linked to immediate vaccination promoting factors connected to the pandemics occurrence, but cannot be ascertained. These factors are possibly paramount in the initial post-pandemic phase, and gradually diminish with the progression of time, theoretically yielding reductions in uptake rates in the long term. Public health immunization professionals should expect preliminary increases in uptake behavior, followed by gradual decreases in the same outcome for influenza pandemics such as H1N1. They should anticipate decreases in uptake behavior following smaller scale epidemics. For pandemic intensity ILI seasons, uptake behavior is not sensitive to weekly fluctuations in ILI severity for children, but slightly sensitive for teens during peak and late phases of the influenza season, with fluctuating uptake behavior associated with peak season phases, and consistent increases for late season phases. This study contributes to the existing literature by enhancing the understanding of how vaccine uptake rates change following pandemic and epidemic events. However it is limited in determining why these changes occur, and due to what factors and mechanisms specifically, which future studies should attempt to discern and ascertain. 2015 LAIV Policy Shift: The 2015 policy shift was associated with preliminary increases in vaccine uptake, followed by annual declines, for both children and teens. Reductions in overall immunization uptake following the preceding 2014 policy shift, and subsequent 2016 policy shift were evident, for both samples for 2014, and teens for 2016. Public health policies concerning influenza immunization for children and adolescents should concentrate on refraining from issuing preferential advisements for either vaccine formulation if possible. Immunization policies should focus on consistent and stable annual advisements, which may promote greater trust in immunization policies. This study contributes to the existing literature by enhancing the understanding of how vaccine uptake rates change following policy shifts. However it is limited in determining why these changes occur, and due to what factors and mechanisms specifically, which future studies should attempt to discern and ascertain.

Available for download on Saturday, December 31, 2022

Share

COinS