Video Abstract

Video Abstract

Close modal
OBJECTIVES

Vaccine hesitancy is a growing threat to health in the United States. Facing the fourth highest vaccine exemption rate in the United States in 2014, Michigan changed its state Administrative Rules, effective January 1, 2015, requiring parents to attend an in-person vaccine education session at their local health department before obtaining a nonmedical exemption (NME). In this article, we evaluate the longer-term impact of this policy change on the rate, spatial distribution, and sociodemographic predictors of NMEs in Michigan.

METHODS

Using school-level kindergarten vaccination data from Michigan from 2011 to 2018, we evaluated sociodemographic predictors of NMEs before and after this Administrative Rule change using Bayesian binomial regression. We measured the persistence and location of school district–level geographic clustering using local indicators of spatial association.

RESULTS

Immediately after the rule change, rates of NMEs fell by 32%. However, NME rates rebounded in subsequent years, increasing by 26% by 2018, although income disparities in NME rates decreased after the rule change. Philosophical, religious, and medical vaccine exemptions exhibited distinct geographic patterns across the state, which largely persisted after 2015, illustrating that NME clusters remain a concern despite this rule change.

CONCLUSIONS

Although Michigan’s Administrative Rule change caused a short-term decline in NME rates, NME rates have risen dramatically in the following 4 years since the policy was implemented. Michigan’s administrative effort to require parental education at the local health department before receiving an exemption did not cause a sustained reduction in the rate or spatial distribution of NMEs.

What’s Known on This Subject:

Michigan was the first state to mandate parental education at the local health department before obtaining a nonmedical exemption. This policy has only been evaluated in the single year after its implementation, which revealed that philosophical exemption rates decreased.

What This Study Adds:

No long-term evaluation of Michigan’s administrative policy has been conducted. In this study, we illustrate that while the rule change caused an initial drop in nonmedical exemption rates, there has since been a significant rebound, and spatial exemption clusters have persisted.

Vaccine hesitancy is a growing problem, with the World Health Organization declaring it 1 of the 10 leading threats to global health.1  Parental concerns around vaccine safety and religious and civil liberties are leading to increasing rates of nonmedical exemptions (NMEs) (philosophical or religious) of vaccines,2  leading to increasingly frequent and severe outbreaks of vaccine-preventable diseases (VPDs), such as measles. In 2019, the United States experienced the most measles cases in 27 years,3  nearly losing its measles elimination status granted in 2000.4 

In 2019, US measles outbreaks primarily occurred in areas with high rates of religious exemptions, including Orthodox Jewish communities. The 4 states most affected were New York, Washington, California, and Michigan.5  In New York, statewide measles vaccination coverage for children in prekindergarten through 12th grade was 98%,6  well above the threshold thought to be sufficient to confer herd immunity (∼95%); however, the outbreak occurred in schools with a measles vaccination rate of 77%, illustrating how heterogeneity can lead to outbreaks even when overall coverage reaches herd immunity thresholds.7  In March 2019, an infectious person traveled from New York to Michigan, initiating the largest measles outbreak in Michigan since 1991.8 

In response to accelerating VPD outbreaks, some states have sought to reduce the rate of NMEs, which are regulated at the state level for school entry.9  Although some state-specific exemption policies changed in 2019 because of measles concerns (NY, ME, and WA tightened exemption policies to reduce the number of NMEs),10  45 states still allow NMEs for religious and/or philosophical reasons (all but CA, ME, MS, NY, and WV). More restrictive NME policies have been revealed to decrease the number of NMEs, reducing outbreak risk.11  However, as seen in New York, schools with high NME rates, even in well-vaccinated communities, can create local regions of susceptibility to disease. There is increasing evidence that such geographic clustering of NMEs is a significant driver of outbreaks.12,13 

In 2014, facing the fourth highest exemption rate in the United States, Michigan changed Administrative Rule 325.176(12), effective January 1, 2015,14  to require parents to attend an in-person vaccine education session at the local health department before obtaining an NME. In one study, researchers found that philosophical exemptions decreased the year after the rule change, but they did not examine longer-term trends, and thus they were unable to evaluate whether NMEs rebounded.15  Michigan was the first state to require in-person waiver education at a local health department, although Washington (SB5005 in 201116 ) and California (AB2109 in 201417 ) both implemented legislation requiring parents to receive counseling from a health care provider before obtaining an NME. These policies had different results: SB5005 decreased rates and geographic clustering of exemptions,16  whereas AB2109 reduced NME rates for incoming kindergarteners but had no apparent effect on geographic clustering, which is just as important a driver of outbreaks.11  California passed SB277 in 2015, removing NMEs entirely, making it impossible to evaluate the longer-term impact of AB2109.18 

Michigan’s 2015 Administrative Rule change thus provides a unique opportunity to evaluate the lasting impact of in-person vaccine education sessions at a local health department on NME rates, implemented via administrative action. In this study, we examine the 4-year impact of the rule change on NME rates in Michigan, describe the geography and persistence of exemption clustering, and, finally, explore predictors of NMEs before and after the policy, accounting for spatial variation.

School-level vaccine exemption data on all (public, private, charter, and virtual) schools with ≥5 students from 2008 to 2018 were obtained from the Michigan Department of Health and Human Services (MDHHS). These data were aggregated and were previously publicly available on the MDHHS Web site, and thus no institutional board review approval was required. Data included school name, year, grade, number of children enrolled, number of students with up-to-date vaccinations, and number of exemption waivers issued by type. Michigan kindergarteners were required to receive a second dose of the varicella vaccine in 201019 ; thus, we selected an analytic period of 2011–2018 to maintain consistent vaccine requirements. Data cleaning and geocoding, described in Masters et al,20  resulted in a sample of 2769 schools that were spatially joined to 2010 school districts to link school district–level demographics from the 2018 American Community Survey 5-year estimate (data from 2013 to 2017).

For each year, the percentage of kindergarteners in each school with vaccine exemptions was calculated by dividing the number of students with an exemption by the number of enrolled students. Data were aggregated to the school district level by summing NME counts in schools on the basis of the school district boundary in which the school was located. Geographic trends in exemption rates were broken down by type and evaluated from 2011 to 2014 and from 2015 to 2018. Analyses were performed in R version 3.6.0. Maps were generated by using Environmental Systems Research Institute ArcMap version 10.7.1.

We used a hierarchical binomial regression model (using the R package lme421 ) to understand variation in school-level NME rates. This model included random intercepts at the school district level and a binary variable indicating whether the time period was before or after the rule change. Geographic clustering of exemptions before and after the change was assessed by using the local indicator of spatial association (LISA),22  which identifies spatial clusters of high and low exemption rates (Supplemental Equation 1). The LISA statistic was calculated for school districts (a meaningful administrative unit for policy action and parental decision-making) for each year, and clusters were evaluated for persistence over time in the pre- and postimplementation periods.

We used a hierarchical Bayesian binomial model with school district–level random intercepts (using the R package rstanarm23 ) to evaluate predictors of NMEs over the study period, accounting for variation at the school district level inherent in the data. We used zero-mean gaussian priors for the intercept (SD = 10) and coefficients (SD = 2.5). We used ArcGIS origin-destination cost matrix calculation service to regress school-level NME rates by school district–level percentages of adults with a college education; per-capita income; and percentages of white residents (all categorized into tertiles); school type; and a continuous variable, calculated as the travel time (in hours) to the local health department from each school. Health department travel time and per-capita income were interacted with the year (centered at 2014) to evaluate whether associations changed over time. We ran a counterfactual exercise using the model output to generate the posterior mean of the marginal probability of obtaining an NME in each year, fixing the distribution of school types, travel times, and demographics. This presents the predicted probability of an NME if every kindergartener in Michigan were in each category of each variable and allows us to make counterfactual comparisons of predicted NME probabilities.

From 2011 to 2014, overall exemption rates remained stable, averaging 5.6% per year. However, the proportion of total exemptions due to NMEs increased during this period, whereas medical exemptions decreased (Fig 1). After Michigan’s rule change was implemented in 2015, exemption rates decreased to 3.6%. The unadjusted binomial model (Supplemental Table 2) revealed that the odds of obtaining an NME were significantly lower in the 4 years after the policy compared with the previous 4 years. However, after 2015, exemption rates increased each year, rising to 3.7% in 2016, 4.2% in 2017, and 4.5% in 2018 (26% higher than 2015). Since 2015, medical exemption rates stayed stable, whereas philosophical and religious exemption rates increased by 18% and 70%, respectively.

FIGURE 1

Percentage of children with vaccine exemptions in the state of Michigan, broken out by exemption type (philosophical, medical, and religious) from 2011 to 2018, with Administrative Rules change going into effect on January 1, 2015, making NMEs harder to obtain.

FIGURE 1

Percentage of children with vaccine exemptions in the state of Michigan, broken out by exemption type (philosophical, medical, and religious) from 2011 to 2018, with Administrative Rules change going into effect on January 1, 2015, making NMEs harder to obtain.

Close modal

Public and charter schools had the lowest exemption rates, with ∼5% for the duration of the study period. (Table 1). On average, 91.3% of kindergarteners in this data set attended public school, 7.8% attended private school, 0.6% attended charter school, and 0.2% attended virtual school. Private schools had higher rates of exemptions from 2011 to 2018, with ∼10% before the 2015 policy, dropping to 7.3% in 2015 and increasing steadily each year, reaching 8.6% by 2018. Virtual schools had the highest exemption rates: >27% in 2012, 2013, and 2018.

TABLE 1

Kindergarten Student Enrollment and NME Data, Broken Out by Exemption and School Type in Michigan From 2011 to 2018

YearStudent EnrollmentNMEs, %Philosophical Exemptions,a %Religious Exemptions,a %
Charter schools     
 2011 278 5.04 71.43 14.29 
 2012 359 6.41 47.83 47.83 
 2013 362 4.42 43.75 50.00 
 2014 562 3.91 54.55 45.45 
 2015 713 3.09 59.09 36.36 
 2016 961 2.91 57.14 35.71 
 2017 1009 4.56 67.39 30.43 
 2018 1280 5.55 56.34 40.85 
Private schools     
 2011 9459 8.84 70.33 14.83 
 2012 9072 10.50 75.03 14.27 
 2013 9113 10.07 74.73 17.86 
 2014 9128 10.28 72.17 21.86 
 2015 8901 7.27 73.88 20.56 
 2016 9263 7.69 73.60 23.03 
 2017 9134 8.02 72.85 24.01 
 2018 9184 8.61 68.90 26.93 
Public schools     
 2011 111 279 5.26 76.21 13.25 
 2012 111 375 5.47 76.01 14.59 
 2013 107 363 5.45 74.24 17.46 
 2014 105 130 4.85 71.92 21.85 
 2015 103 217 3.25 77.26 16.82 
 2016 105 286 3.27 75.44 19.27 
 2017 105 650 3.80 73.41 21.36 
 2018 104 742 4.06 71.89 22.58 
Virtual schools     
 2011 57 21.05 91.67 8.33 
 2012 74 27.03 55.00 45.00 
 2013 200 27.00 64.81 35.19 
 2014 241 19.50 53.19 46.81 
 2015 205 13.66 67.86 28.57 
 2016 287 19.86 66.67 31.58 
 2017 411 24.57 71.29 23.76 
 2018 487 27.52 61.19 26.12 
YearStudent EnrollmentNMEs, %Philosophical Exemptions,a %Religious Exemptions,a %
Charter schools     
 2011 278 5.04 71.43 14.29 
 2012 359 6.41 47.83 47.83 
 2013 362 4.42 43.75 50.00 
 2014 562 3.91 54.55 45.45 
 2015 713 3.09 59.09 36.36 
 2016 961 2.91 57.14 35.71 
 2017 1009 4.56 67.39 30.43 
 2018 1280 5.55 56.34 40.85 
Private schools     
 2011 9459 8.84 70.33 14.83 
 2012 9072 10.50 75.03 14.27 
 2013 9113 10.07 74.73 17.86 
 2014 9128 10.28 72.17 21.86 
 2015 8901 7.27 73.88 20.56 
 2016 9263 7.69 73.60 23.03 
 2017 9134 8.02 72.85 24.01 
 2018 9184 8.61 68.90 26.93 
Public schools     
 2011 111 279 5.26 76.21 13.25 
 2012 111 375 5.47 76.01 14.59 
 2013 107 363 5.45 74.24 17.46 
 2014 105 130 4.85 71.92 21.85 
 2015 103 217 3.25 77.26 16.82 
 2016 105 286 3.27 75.44 19.27 
 2017 105 650 3.80 73.41 21.36 
 2018 104 742 4.06 71.89 22.58 
Virtual schools     
 2011 57 21.05 91.67 8.33 
 2012 74 27.03 55.00 45.00 
 2013 200 27.00 64.81 35.19 
 2014 241 19.50 53.19 46.81 
 2015 205 13.66 67.86 28.57 
 2016 287 19.86 66.67 31.58 
 2017 411 24.57 71.29 23.76 
 2018 487 27.52 61.19 26.12 
a

Represented as a percentage of total NMEs.

Although NME rates fell in the immediate aftermath of the 2015 rule change, rates have increased steadily since and maintained relatively stable patterns of geographic clustering. School district–level clusters of persistent, high vaccine exemption rates from 2011 to 2014 and from 2015 to 2018 are shown in Fig 2. Different types of NMEs followed distinct clustering patterns. Figure 2A reveals that philosophical exemption clusters persisted in rural, remote regions of the Upper Peninsula, a cluster in northwestern Michigan disappeared after 2015, and a new cluster in mid-Michigan emerged after 2015. There were 56 school districts in a high philosophical exemption cluster from 2011 to 2014, decreasing to 26 for the period from 2015 to 2018, although the number of persistent clusters (>3 years) was unchanged (Supplemental Table 3). Religious exemption clusters revealed little change (Fig 2B): there were 32 religious exemption school district clusters in both time periods, although fewer persistent clusters afterward. Finally, for medical exemptions (Fig 2C), some persistent clusters in northeastern Michigan disappeared, yet a large, more persistent cluster appeared in southeast Michigan, overlapping with a philosophical exemption cluster. The number of medical exemption clusters dropped from 43 to 25. Overall, the 2015 rule change appeared to reduce the number of philosophical exemption clusters (although not persistent clusters) and diminished the spatial distribution of some medical exemption clusters. Religious exemption clusters were largely unchanged.

FIGURE 2

Persistence of LISA clusters of philosophical, religious, and medical exemptions at the school district level, represented as the number of years in which each school district was in a high-high LISA exemption cluster, before and after the 2015 rule change. High-high exemption clusters indicate that the highlighted school districts were identified by the LISA statistic as significant clusters of high exemption rates (indicating that both a given school district and the average of its neighboring school districts had significantly high exemption rates). (A) School districts in a high-high philosophical exemption cluster from 2011 to 2014 and from 2015 to 2018. (B) School districts in a high-high religious cluster from 2011 to 2014 and from 2015 to 2018. (C) School districts in a high-high medical exemption cluster from 2011 to 2014 and from 2015 to 2018.

FIGURE 2

Persistence of LISA clusters of philosophical, religious, and medical exemptions at the school district level, represented as the number of years in which each school district was in a high-high LISA exemption cluster, before and after the 2015 rule change. High-high exemption clusters indicate that the highlighted school districts were identified by the LISA statistic as significant clusters of high exemption rates (indicating that both a given school district and the average of its neighboring school districts had significantly high exemption rates). (A) School districts in a high-high philosophical exemption cluster from 2011 to 2014 and from 2015 to 2018. (B) School districts in a high-high religious cluster from 2011 to 2014 and from 2015 to 2018. (C) School districts in a high-high medical exemption cluster from 2011 to 2014 and from 2015 to 2018.

Close modal

Our counterfactual analysis of the predicted probability of an NME if every kindergartener in Michigan were in each category of select demographic variables revealed that the average marginal probability of obtaining an NME was similar for kindergarteners whose school district was in the 2 lowest tertiles of per-capita income but was higher for school districts in the highest tertile of per-capita income (Fig 3, Supplemental Table 4). After the rule change, the discrepancy in the posterior mean average probability of an NME diminished between those in the wealthiest tertile and those in the poorer 2 tertiles. School district–level percentage of white residents was monotonically associated with NME rates: kindergarteners in the lowest tertile had the lowest NME probability, and those in the highest tertile had the highest (Supplemental Table 4). For percentage of adults with a college education, a different association was observed; kindergarteners in the middle tertile had the highest probability of an NME versus those in the other tertiles (Supplemental Table 4). Increased travel time to the local health department did not have a predictive effect on the probability of an NME (Supplemental Tables 5 and 6; Supplemental Figs 410).

FIGURE 3

Bayesian binomial logistic hierarchical model output revealing posterior mean average marginal effects of the probability of getting an NME waiver for selected demographic predictors at the school district level: tertiles of the percentage of white residents, tertiles of the percentage of adults aged >25 years with a college education, and tertiles of per-capita income, with demarcation for prevaccination and postvaccination rule change before the 2015 academic year.

FIGURE 3

Bayesian binomial logistic hierarchical model output revealing posterior mean average marginal effects of the probability of getting an NME waiver for selected demographic predictors at the school district level: tertiles of the percentage of white residents, tertiles of the percentage of adults aged >25 years with a college education, and tertiles of per-capita income, with demarcation for prevaccination and postvaccination rule change before the 2015 academic year.

Close modal

High rates of NMEs are a critical public health challenge that has begun to reverse decades of public health success in the control of VPDs. Although New York, Maine, and Washington tightened NME regulations in the wake of the 2019 measles outbreaks, Michigan passed no comparable legislation. In fact, even minor policy remedies presented in the state have been stymied: Michigan introduced HB4610 in May 2019 to make schools with >5% exemption rates publicly post such information, but there has been no movement on this bill.24  As such, Michigan’s 2015 Administrative Rule change, which did not require legislative action, was a strategic avenue for MDHHS to attempt to reduce NMEs.11 

Although the rule change induced a sharp decline in the number of NMEs in the year after it went into effect, NME rates in Michigan have since rebounded nearly to pre-2015 levels, suggesting that this change has not had an enduring impact. Interestingly, medical exemption rates in Michigan decreased after the rule change and have remained at these reduced levels since 2015. This is in contrast to what was observed in California after SB277, in which medical exemption rates increased 250%, indicative of a replacement effect in which parents were obtaining medical exemptions in place of nonmedical ones.25  The rate of religious exemptions in Michigan has rebounded faster than philosophical exemptions, which is concerning given that the 2019 outbreaks were primarily driven by religious exemption clusters.

Michigan’s increasing exemption rates mirror national trends: from 1991 to 2004, the mean state-level NME rate increased from 0.98% to 1.48%,2  and from 2011 to 2016, the national rate of NMEs increased from 1.75% to 2.25%.26  Michigan’s high NME rate has put the state at risk: Olive et al27  identified Michigan’s Oakland, Macomb, and Wayne counties to be among the 10 counties in the United States with the highest numbers of NMEs, foreshadowing the 2019 measles outbreak in Oakland County. Our analysis confirmed high NME rates in these counties, and identified school district–level clustering of religious exemptions in Macomb County, indicating regions in which VPD risk may also be high at a finer scale. Overall, the fact that in 4 years, exemption rates have already rebounded to nearly pre–rule change levels indicates that stronger action, combined with multifaceted approaches that do not rely exclusively on legislative and administrative changes, is needed to curb increasing outbreak risk in Michigan.

Our findings that private schools had approximately twice the rate of exemptions as public and charter school align with research from California.11  Although the vast majority of kindergarteners in this data set attended public school, ∼8% of the students attended private schools, and thus the increase in private school exemption rates is concerning. Virtual schools had extremely high exemption rates, providing a glimpse of nonvaccination among homeschooled children. The number of kindergarteners enrolled in virtual schools increased nearly 10-fold over the study period, from 57 students in 2011 to 487 in 2018; thus, the high rates of exemptions in virtual schools is particularly worrying if these numbers continue to rise, which may occur especially in the aftermath of the coronavirus disease 2019 pandemic.

We also identified spatial clusters with persistently high exemption rates both before and after the rule change. This is important because spatial clustering of nonvaccination may dramatically increase the risk of outbreaks at the local and population level.28  Given that students who obtain kindergarten vaccine exemptions age through the educational system, a region with persistently high kindergarten exemption rates is likely to have markedly reduced vaccination levels among students, accumulating children susceptible to VPDs.18  Our analysis revealed that many exemption clusters remained persistent after the rule change, which more strongly aligns with the aftermath of California’s AB2109 (which decreased exemption rates but not clustering)11,17  than with Washington’s SB5005. Additionally, the geography of philosophical and religious NMEs were distinct. These findings generally concur with Mashinini et al.29  However, our analysis used school districts as the clustering unit and evaluated clusters for their persistence over time, providing results at an actionable geographic scale. The distinct patterns of spatial clustering observed for different exemption types indicate that the downstream impacts of policy changes to further restrict access to NMEs would likely play out heterogeneously across a state characterized by a diverse socioeconomic, racial, and religious landscape.

Results from our counterfactual exercise by using Bayesian regression output underscored that school type was a predictor of NME rates, with virtual and private schoolers having higher probabilities of exemptions than their public school counterparts. Distance to the local health department was not a strong predictor of school-level exemption rates, highlighting that this policy likely only reduced exemptions because of convenience rather than conviction and potentially indicating that those who pursued an NME after 2015 were sufficiently motivated that the opportunity cost associated with traveling to the health department was not a high barrier. We found nonlinear patterns across school district–level percentage of white residents, percentage of college-educated adults, and per-capita income. This generally concurs with previous research in which undervaccinated children are often minorities of lower socioeconomic status and educational attainment, whereas completely nonvaccinated children are often white, wealthy, educated, and privately insured.3032 

A notable strength of this study is the use of school-level data to identify potential geographic clustering and regions in which herd immunity might be broken because of high exemption rates. This data source represents all schools with at least 5 students in Michigan, providing a near-complete assessment of kindergarten vaccination. Additionally, school-level data are appropriate here given that school is the unit of aggregation at which much of transmission occurs. Geocoding these schools allowed for linkage of sociodemographic variables from the American Community Survey, permitting measures of community-level demographics while using transmission-level vaccination data.

This study also has some limitations, most importantly that we only analyzed exemption data for kindergarteners, thus creating an incomplete picture of the vaccination status of the full student population in these schools. Additionally, it is possible that there are missing data if not all students were present when schools were surveyed for vaccination and enrollment records. Assigning school districts to schools on the basis of spatial location alone may not perfectly represent the body of students in virtual and private schools, which may draw from a broader geographic area. Finally, using school district–level demographics may not be a perfect match to the student body from each school, as school catchment areas may extend beyond the boundaries of the units chosen or be specific subsegments of a geographic unit, introducing the possibility of ecological bias.

Michigan’s Administrative Rule change reduced the number of NMEs immediately after its implementation, yet NMEs have since rebounded, and many school district–level NME clusters have persisted, illustrating that this change did not have a lasting impact on NMEs in Michigan. Navin et al33  found that Michigan’s vaccine waiver educators rarely convinced parents to vaccinate their children, underscoring that such a policy is effectively imposing a cost to reduce convenience exemptions, yet unlikely to change perception. They also found there may be a threshold of burden beyond which increasing inconvenience does not further reduce exemptions.34  As a result, it is important to balance the implementation of stronger policies to curb NME rates to reduce the frequency of outbreaks with the possibility of backlash against restrictions of individual liberty.35,36 

Michigan’s administrative policy should be viewed within the larger context of the interventions available to reduce incentives to obtain exemptions. In addition to state regulation of vaccine exemptions, interventions should be aimed to counter growing levels of vaccine hesitancy through education, building confidence in vaccines and government, curbing misinformation, educating doctors about the importance of vaccination, minimizing missed opportunities, and increasing affordability.37  These recommendations are particularly important against a backdrop of the coronavirus disease 2019 pandemic, which has led to reduced ambulatory care visits, causing a precipitous drop in pediatric vaccination rates.38  At this critical juncture, we must use new tactics (more than the usual slate of interventions) to fight vaccine misinformation, build back trust, increase vaccine uptake, and minimize the risk of additional outbreaks.

Dr Masters wrote the first draft of the research, did the data cleaning, helped formulate the research plan, conducted the analysis, and created maps and figures; Dr Zelner assisted with formulating the research plan, provided assistance in the Bayesian modeling component, and edited the manuscript; Dr Delamater assisted with mapping and performing the local indicators of spatial association analysis, as well as mapping the results of such analysis in ArcGIS, and edited the final manuscript; Dr Hutton helped to determine the most effective way to analyze the results of the Administrative Rules change, helped plan the analysis, and edited the manuscript; Dr Kay helped to conduct the Bayesian modeling approach and wrote the R package TidyBayes, which allows for visualization of such complex models in an interpretable way, and reviewed the manuscript; Dr Eisenberg assisted with programming and modeling for the Bayesian analysis and with writing the manuscript; Dr Boulton helped conceive of the data analysis plan, evaluate the policy given his experience as a state epidemiologist, and assisted with writing the manuscript; and all authors approved the final manuscript as submitted and agree to be accountable for all aspects of the work.

FUNDING: Dr Zelner was supported by funding from the Michigan Institute for Clinical and Health Research (Pathway award) and the Michigan Institute for Computational Discovery and Engineering (Catalyst grant). The funders had no role in the design or conceptualization of the study. Dr Masters’ research was funded by Dr Boulton.

COMPANION PAPER: A companion to this article can be found online at www.pediatrics.org/cgi/doi/10.1542/peds.2021-050445.

     
  • LISA

    local indicator of spatial association

  •  
  • MDHHS

    Michigan Department of Health and Human Services

  •  
  • NME

    nonmedical exemption

  •  
  • VPD

    vaccine-preventable disease

1
World Health Organization
.
Ten threats to global health in 2019
.
2
Omer
SB
,
Salmon
DA
,
Orenstein
WA
,
deHart
MP
,
Halsey
N
.
Vaccine refusal, mandatory immunization, and the risks of vaccine-preventable diseases
.
N Engl J Med
.
2009
;
360
(
19
):
1981
1988
3
Centers for Disease Control and Prevention
.
Measles cases and outbreaks
.
2019
.
Available at: https://www.cdc.gov/measles/cases-outbreaks.html. Accessed June 28, 2019
4
Silverberg
R
,
Caceres
J
,
Greene
S
,
Hart
M
,
Hennekens
CH
.
Lack of measles vaccination of a few portends future epidemics and vaccination of many
.
Am J Med
.
2019
;
132
(
9
):
1005
1006
5
Patel
M
,
Lee
AD
,
Clemmons
NS
, et al
.
National update on measles cases and outbreaks - United States, January 1-October 1, 2019
.
MMWR Morb Mortal Wkly Rep
.
2019
;
68
(
40
):
893
896
6
Centers for Disease Control and Prevention
.
Measles
. In:
Hamborsky
J
,
Kroger
A
,
Wolfe
SE
, eds.
Epidemiology and Prevention of Vaccine-Preventable Diseases
. 13th ed.
Washington, DC
:
Public Health Foundation
;
2015
:
209
229
7
McDonald
R
,
Ruppert
PS
,
Souto
M
, et al
.
Notes from the field: measles outbreaks from imported cases in orthodox Jewish Communities - New York and New Jersey, 2018-2019
.
MMWR Morb Mortal Wkly Rep
.
2019
;
68
(
19
):
444
445
8
Michigan Department of Health and Human Services
.
2019 Michigan measles outbreak information
.
9
Gastañaduy
PA
,
Paul
P
,
Fiebelkorn
AP
, et al
.
Assessment of the status of measles elimination in the United States, 2001-2014
.
Am J Epidemiol
.
2017
;
185
(
7
):
562
569
10
McKinley
J
.
Measles outbreak: N.Y. eliminates religious exemptions for vaccinations
.
New York Times. June 13, 2019. Available at: https://www.nytimes.com/2019/06/13/nyregion/measles-vaccinations-new-york.html. Accessed July 30, 2019
11
Bednarczyk
RA
,
King
AR
,
Lahijani
A
,
Omer
SB
.
Current landscape of nonmedical vaccination exemptions in the United States: impact of policy changes
.
Expert Rev Vaccines
.
2019
;
18
(
2
):
175
190
12
Aloe
C
,
Kulldorff
M
,
Bloom
BR
.
Geospatial analysis of nonmedical vaccine exemptions and pertussis outbreaks in the United States
.
Proc Natl Acad Sci USA
.
2017
;
114
(
27
):
7101
7105
13
Omer
SB
,
Enger
KS
,
Moulton
LH
,
Halsey
NA
,
Stokley
S
,
Salmon
DA
.
Geographic clustering of nonmedical exemptions to school immunization requirements and associations with geographic clustering of pertussis
.
Am J Epidemiol
.
2008
;
168
(
12
):
1389
1396
14
Michigan Department of Health and Human Services Department of Community Health Bureau of Epidemiology Communicable and Related Diseases
.
MDHHS communicable and related diseases administrative rules
.
2015
.
15
Mashinini
DP
,
Fogarty
KJ
,
Potter
RC
,
Lagerwey
MD
.
The impact of Michigan’s nonmedical vaccine exemption rule change on philosophical exemption rates
.
J Community Health
.
2020
;
45
(
1
):
148
153
16
Omer
SB
,
Allen
K
,
Chang
DH
, et al
.
Exemptions from mandatory immunization after legally mandated parental counseling
.
Pediatrics
.
2018
;
141
(
1
):
e20172364
17
Delamater
PL
,
Leslie
TF
,
Yang
YT
.
Examining the spatiotemporal evolution of vaccine refusal: nonmedical exemptions from vaccination in California, 2000-2013
.
BMC Public Health
.
2018
;
18
(
1
):
458
18
Delamater
PL
,
Leslie
TF
,
Yang
YT
.
A spatiotemporal analysis of non-medical exemptions from vaccination: California schools before and after SB277
.
Soc Sci Med
.
2016
;
168
(
13
):
230
238
19
Michigan Department of Health and Human Services
.
Immunization program
.
2020
.
20
Masters
NB
,
Delamater
PL
,
Boulton
ML
,
Zelner
J
.
Measuring multiple dimensions and indices of non-vaccination clustering in Michigan: 2008-2018 [published online ahead of print December 11, 2020]
.
Am J Epidemiol
.
doi:10.1093/aje/kwaa264
21
Bates
D
,
Mächler
M
,
Bolker
B
,
Walker
S
.
lme4: linear mixed-effects models using Eigen and S4
.
2015
.
22
Anselin
L
.
Local indicators of spatial association—LISA
.
Geogr Anal
.
1995
;
27
(
2
):
93
115
23
Goodrich
B
,
Gabry
J
,
Ali
I
,
Brilleman
S
.
rstanarm: Bayesian applied regression modeling via Stan
.
2020
.
Available at: https://mc-stan.org/rstanarm/. Accessed June 1, 2020
24
Elder
B
.
HB4610. Michigan Legislature
.
2019
.
Available at: http://legislature.mi.gov/doc.aspx?2019-HB-4610. Accessed February 1, 2020
25
Mohanty
S
,
Buttenheim
AM
,
Joyce
CM
,
Howa
AC
,
Salmon
D
,
Omer
SB
.
Experiences with medical exemptions after a change in vaccine exemption policy in California
.
Pediatrics
.
2018
;
142
(
5
):
e20181051
26
Omer
SB
,
Porter
RM
,
Allen
K
,
Salmon
DA
,
Bednarczyk
RA
.
Trends in kindergarten rates of vaccine exemption and state-level policy, 2011-2016
.
Open Forum Infect Dis
.
2017
;
5
(
2
):
ofx244
27
Olive
JK
,
Hotez
PJ
,
Damania
A
,
Nolan
MS
.
The state of the antivaccine movement in the United States: a focused examination of nonmedical exemptions in states and counties
.
PLoS Med
.
2018
;
15
(
6
):
e1002578
28
Masters
NB
,
Eisenberg
MC
,
Delamater
PL
,
Kay
M
,
Boulton
ML
,
Zelner
J
.
Fine-scale spatial clustering of measles nonvaccination that increases outbreak potential is obscured by aggregated reporting data
.
Proc Natl Acad Sci USA
.
2020
;
117
(
45
):
28506
28514
29
Mashinini
DP
,
Fogarty
KJ
,
Potter
RC
,
Berles
JD
.
Geographic hot spot analysis of vaccine exemption clustering patterns in Michigan from 2008 to 2017
.
Vaccine
.
2020
;
38
(
51
):
8116
8120
30
Olive
JK
,
Matthews
KRW
.
How Too Much Freedom of Choice Endangers Public Health: The Effect of Nonmedical Exemptions From School-Entry Vaccinations in Texas. Policy Brief 10.13.16
.
Houston, TX
:
James A. Baker III Institute for Public Policy of Rice University
;
2016
31
Yang
YT
,
Delamater
PL
,
Leslie
TF
,
Mello
MM
.
Sociodemographic predictors of vaccination exemptions on the basis of personal belief in California
.
Am J Public Health
.
2016
;
106
(
1
):
172
177
32
Wang
E
,
Clymer
J
,
Davis-Hayes
C
,
Buttenheim
A
.
Nonmedical exemptions from school immunization requirements: a systematic review
.
Am J Public Health
.
2014
;
104
(
11
):
e62
e84
33
Navin
MC
,
Kozak
AT
,
Clark
EC
.
The evolution of immunization waiver education in Michigan: a qualitative study of vaccine educators
.
Vaccine
.
2018
;
36
(
13
):
1751
1756
34
Navin
MC
,
Largent
MA
,
McCright
AM
.
Efficient burdens decrease nonmedical exemption rates: a cross-county comparison of Michigan’s vaccination waiver education efforts
.
Prev Med Rep
.
2020
;
17
:
101049
35
Salmon
DA
,
MacIntyre
CR
,
Omer
SB
.
Making mandatory vaccination truly compulsory: well intentioned but ill conceived
.
Lancet Infect Dis
.
2015
;
15
(
8
):
872
873
36
Mello
MM
,
Studdert
DM
,
Parmet
WE
.
Shifting vaccination politics–the end of personal-belief exemptions in California
.
N Engl J Med
.
2015
;
373
(
9
):
785
787
37
Gostin
LO
,
Hodge
JG
 Jr
,
Bloom
BR
, et al
.
The public health crisis of underimmunisation: a global plan of action
.
Lancet Infect Dis
.
2020
;
20
(
1
):
e11
e16
38
Santoli
JM
,
Lindley
MC
,
DeSilva
MB
, et al
.
Effects of the COVID-19 pandemic on routine pediatric vaccine ordering and administration - United States, 2020
.
MMWR Morb Mortal Wkly Rep
.
2020
;
69
(
19
):
591
593

Competing Interests

POTENTIAL CONFLICT OF INTEREST: The authors have indicated they have no potential conflicts of interest to disclose.

FINANCIAL DISCLOSURE: The authors have indicated they have no financial relationships relevant to this article to disclose.

Supplementary data