Trends and Characteristics of Proposed and Enacted State Legislation on Childhood Vaccination Exemption, 2011–2017

Abstract

Objectives. To examine trends and characteristics of proposed and enacted state legislation that would directly affect states’ immunization exemption laws.

Methods. We performed content analysis of proposed bills in state legislatures from 2011 to 2017. We classified bills as provaccination or antivaccination.

Results. State legislators proposed 175 bills, with the volume increasing over time: 92 (53%) bills expanded access to exemptions, and 83 (47%) limited the ability to exempt. Of the 13 bills signed into law, 12 (92%) limited the ability to exempt. Bills that expanded access to exemptions were more likely to come from Republican legislators and Northeastern and Southern states.

Conclusions. Although most proposed legislation would have expanded access to exemptions, bills that limited exemptions were more likely to be enacted into law. Legal barriers to exempt one’s children from vaccination persist despite vaccine hesitancy, which is encouraging for public health.

Public Health Implications. Most vaccine exemption laws introduced in state legislatures would pose threats to the public’s health. There is a need for constituents to engage their elected legislators and advocate provaccination policies.

A primary goal of vaccination policy is to obtain and sustain a sufficient level of vaccinated individuals to establish community immunity against vaccine-preventable diseases. All states require vaccination of children enrolling in public schools and state-funded day care and often private schools as well.¹ Although all states allow medical exemptions to this vaccination requirement (e.g., for immunocompromised individuals), all but 3—West Virginia, Mississippi, and, most recently, California—allow exempting for religious and, possibly, ideological reasons. The use of nonmedical exemptions increased nationwide by 19% between 2009 and 2013, from 1.6% to 1.9% among children in kindergarten, with substantial state by state variation.² Private schools tend to have higher exemption rates than do public schools, although the exact reasons are unclear and likely multifactorial.³ Because some diseases require a very high threshold of immunized individuals for herd protection (e.g., > 90% vaccination coverage for pertussis⁴), vaccination exemptions may jeopardize community health if nonimmune individuals cluster together, even in states with overall high vaccination rates.⁵

Laws related to the procedures required to exempt one’s children vary by state, and state law pertaining to childhood vaccination exemptions is dynamic and evolving. Many changes have occurred in response to public health crises or judiciary involvement. In 2015, California passed Senate Bill 277,⁶ removing all nonmedical exemptions from its mandatory vaccine requirements following a measles outbreak at Disneyland. The Mississippi legislature removed its nonmedical exemptions in 1983 following a court battle.⁷ Unlike Mississippi and California, West Virginia has always permitted only medical exemptions.

Previous research has examined enacted legislation related to vaccine exemptions and identified specific aspects of state laws that increase immunization rates—such as inclusion of Advisory Committee on Immunization Practices recommendations, state health department–run parental education, state health department review of applications for exemption, and scalable exemption policies (i.e., exemption from specific vaccinations).^8,9 Studies have also identified features of current state laws that decrease immunization rates, such as inclusion of nonmedical exemptions,^8–10 and the ease with which exemptions are obtainable.²

Because a bill has to be proposed, discussed, and enacted through the legislative process before it becomes a law, activity in state legislatures can serve as an indicator of potential changes in exemption policy. Our literature review identified only 1 study that has systematically examined proposed legislation in the United States pertaining to exemptions, and this study was limited to personal belief exemption laws for the years 2009 to 2012.¹¹ We build on this earlier work by analyzing trends in more recent state legislative sessions and expanding the scope of analysis to all types of immunization exemption.

METHODS

Since 2011, the Association of State and Territorial Health Officials (ASTHO) has tracked proposed state-level immunization exemption legislation.¹² ASTHO searches state legislative databases for introduced bills that contain 1 or more of the words “vaccine,” “vaccination,” “immunization,” or “exemption,” which they subsequently review to determine if the bills were relevant to childhood immunization exemptions. ASTHO’s yearly reports include data on the legislative chamber in which bills are introduced (e.g., house or senate), bill number, a summary, and the status (e.g., voted and enacted, voted and rejected or vetoed, withdrawn or died in committee). Reports were available for the years 2011 through 2017. We obtained additional bill details, including the full text and primary sponsor, via LegiScan (LegiScan, LLC, Elkview, WV) and state legislature Web sites.

To explore whether political climate and exogenous factors in the legislative process were associated with proposed legislation,¹³ we retrieved yearly data from the National Conference of State Legislatures about each legislature’s partisan composition (i.e., Republicans controlled both house and senate, Democrats controlled both house and senate, or house and senate control was split between the 2 parties) and the governor’s political party affiliation.¹⁴ Using data from the US Census Bureau, we also retrieved information on each state’s population size and Census region (i.e., Northeast, South, Midwest, or West).

Policy Mapping and Coding

We performed a content analysis of bills in accordance with best practices for legal mapping.¹⁵ First, J. S. S. (a public health attorney) read each bill’s full text to initially classify the legislation as provaccination (i.e., restricting exemption) or antivaccination (i.e., broadening exemption). When the bill’s intention was unclear or perceived as neutral, all 3 authors independently reviewed the text and the majority consensus was used to classify the bill. For these ambiguous bills, we assumed the perspective of the health department to guide the final decision: we deemed bills that took discretion away from the health department antivaccination and bills that empowered the health department provaccination.

Second, we undertook a thematic content analysis to further classify bills beyond being pro- or antivaccination according to their goals (e.g., eliminate religious exemption, add personal belief exemption, change reporting requirements). We generated themes inductively after we reviewed all of the bills. Although some bills had multiple goals, we used the primary intention for coding.

Statistical Analysis

Statistical analysis included both descriptive and predictive methodologies. In the descriptive analysis, we quantified the number of bills proposed in each state by whether they would expand or restrict vaccine exemptions (i.e., antivaccination or provaccination), by specific bills’ goals, and by the bills’ final disposition (i.e., enacted into law or not). We then assessed this longitudinally to examine trends in legislation over time. We created choropleth maps to depict the spatial distribution of proposed legislation across the United States. Using χ² and analysis of variance statistical techniques, we assessed associations between the dependent variable (i.e., antivaccination or provaccination bill) and the independent variables—features of the bill (i.e., whether it was signed into law, originating legislative chamber, sponsor’s political party), state political climate (i.e., political party holding most seats in the legislature, governor’s political party), and exogenous factors (i.e., state population size, US Census region).

In the predictive analysis, we fit Bayesian generalized linear mixed effect regression models to isolate individual characteristics of bills stratified by their classification as well as bill passage (i.e., signed into law). This corresponded to 3 separate and distinct regression models: (1) predictors of antivaccination bills being introduced, (2) predictors of provaccination bills being introduced, and (3) predictors of these bills being signed into law. The use of mixed effect models allowed us to account for the within-state correlations; that is, because multiple bills may have been proposed within a particular state, these bills may be more similar than are between-state comparisons. The use of a Bayesian framework had several analytic benefits, including being computationally flexible for fitting multilevel regression models and the ability to overcome small samples by incorporating external data. Models controlled for state political climate and exogenous factor variables. Outputs were expressed as odds ratios (ORs) with corresponding 95% credible intervals (CIs).

We conducted all statistical analyses with R version 3.4.2 (R Foundation for Statistical Computing, Vienna, Austria), and we performed policy mapping in Excel version 16.15 (Microsoft Corp., Redmond, WA). The analytic source code and data set are available for download from https://doi.org/10.5281/zenodo.1409660.

RESULTS

Between 2011 and 2017, 175 bills were introduced in state legislatures that would have modified childhood vaccination exemption laws. The total number of bills increased over time (Figure 1; P for trend = .04). Four states had more than 10 proposed bills in this period: New Jersey (n = 32; 18%), New York (n = 28; 16%), West Virginia (n = 15; 9%), and Mississippi (n = 12; 7%). Thirty-four states had between 1 and 10 bills, whereas 12 states did not have any relevant legislation. Among the 175 introduced bills, 13 (7%) were signed into law. We classified 92 (53%) bills as antivaccination and 83 (47%) as provaccination. Table 1 provides a state by state breakdown of the legislation, and examples of language included in the bill can be found in Table A (available as a supplement to the online version of this article at http://www.ajph.org).

FIGURE 1—

Trends in State Vaccination Exemption Bills: United States, 2011–2017

Note. Anti = antivaccination; Pro = provaccination. P for trend = .04.

TABLE 1—

State by State Breakdown of Proposed State Legislation on Childhood Vaccination Exemptions: United States, 2011–2017

No. of Bills	States
Total
0	AK, AL, FL, GA, ID, IN, KY, ND, NE, SC, TN, WY
1–4	AR, AZ, CA, CO, CT, DE, HI, IA, IL, KS, LA, MA, MD, ME, MI, MO, MT, NC, NH, NM, NV, OH, OK, OR, PA, SD, UT, VA, WA, WI
5–9	MN, RI, TX, VT
≥ 10	MS, NJ, NY, WV
Antivaccination
0	AK, AL, AR, CA, DE, FL, GA, HI, ID, IL, IN, KY, LA, MD, ME, MI, MO, NC, ND, NE, NM, NV, OH, OR, PA, SC, TN, UT, VA, VT, WI, WY
1–4	AZ, CO, CT, IA, KS, MA, MN, MT, NH, OK, SD, TX, WA
5–9	RI
≥ 10	MS, NJ, NY, WV
Provaccination
0	AK, AL, FL, GA, ID, IN, KS, KY, MA, MS, ND, NE, NH, RI, SC, SD, TN, WY
1–4	AR, AZ, CA, CO, CT, DE, HI, IA, IL, LA, MD, ME, MI, MO, MT, NC, NM, NV, OH, OK, OR, PA, UT, VA, WA, WI, WV
5–9	MN, NJ, TX, VT
≥ 10	NY

Thematic analysis revealed 5 primary goals of exemption bills. Most bills (n = 78; 45%) would have added exemptions or removed requirements to make existing exemptions easier to obtain. Fifty-nine (34%) bills would have removed exemptions or added requirements to make existing exemptions more difficult to obtain. Seventeen (10%) bills affected data collection and reporting requirements or clarified language in existing legislation, whereas 10 bills (6%) included language to notify parents about the vaccination status of children in the school. Eight (5%) bills shifted the governing authority for exemptions either away from or toward the health department.

Table 2 presents characteristics of the bills overall and compared by their classification (i.e., antivaccination or provaccination). Most bills were introduced in the house or assembly legislative chambers (n = 112; 64%) and from states in the Northeast (n = 86; 48%). Of the 13 bills signed into law, 12 (92%) were provaccination and 1 (8%) was antivaccination (P < .01). Antivaccination bills were more likely to be sponsored by a Republican legislator (P < .01) and in states in the Northeastern and Southern Census regions (P < .01). Choropleth maps depict the spatial distribution of all exemption-related bills (Figure A, available as a supplement to the online version of this article at http://www.ajph.org), those classified as antivaccination (Figure B, available as a supplement to the online version of this article at http://www.ajph.org), and those classified as provaccination (Figure C, available as a supplement to the online version of this article at http://www.ajph.org). Provaccination bills appeared more evenly distributed throughout the United States, whereas antivaccination bills tended to be concentrated in fewer states. Mississippi and West Virginia, states that currently do not have nonmedical exemptions, were particularly active in introducing legislation to add these exemptions. Bill classification did not statistically differ with respect to the partisan composition of the legislature or the governor.

TABLE 2—

Characteristics of State Vaccination Exemption Bills Overall and by Classification (Antivaccination or Provaccination): United States, 2011–2017

Characteristic	Overall (n = 175 Bills), No. (%) or Median (IQR)	Bill Classification		P
		Antivaccination (n = 92 Bills; 53%), No. (%) or Median (IQR)	Provaccination (n = 83 Bills; 47%) No. (%) or Median (IQR)
Signed into law				< .01
No	162 (93)	91 (99)	71 (86)
Yes	13 (7)	1 (1)	12 (14)
Legislative chamber				.97
House/Assembly	112 (64)	59 (64)	53 (64)
Senate	63 (36)	33 (36)	30 (36)
Legislative majority				.39
Democratic	88 (50)	49 (53)	39 (47)
Republican	60 (34)	32 (35)	28 (34)
Split	27 (16)	11 (12)	16 (19)
Bill sponsor’s party				< .01
Democratic	93 (53)	30 (33)	63 (77)
Republican	81 (47)	62 (67)	19 (23)
Governor’s party				.08
Democratic	89 (51)	41 (45)	48 (58)
Republican	86 (49)	51 (55)	35 (42)
State population size	6 113 532 (7 049 188)	6 233 486 (7 189 787)	6 113 532 (13 225 635)	.21
Census region				< .01
Northeast	86 (48)	48 (51)	38 (46)
Midwest	24 (14)	8 (9)	16 (19)
South	43 (25)	30 (33)	13 (16)
West	22 (13)	6 (7)	16 (19)

Note. IQR = interquartile range.

Table 3 presents results from the 3 regression models to identify independent predictors of bill classification (i.e., antivaccination or provaccination) and final disposition (signed into law or not). In model 1, year of bill proposal, not signed into law, and sponsor’s political party were associated with a bill being classified as antivaccination. Bills were less likely to be labeled as antivaccination for the more recent years, and for bills signed into law, they had 91% lower odds of being classified as antivaccination. A bill had a 12-fold increase in odds of being classified as antivaccination if the sponsor was Republican. Model 2 revealed similar, yet inverse trends, as expected. Bills were more likely to be labeled as provaccination for the more recent years, being executed into law, and being sponsored by a Democrat. In model 3, which assessed factors related to passage, only provaccination bill classification was independently associated with a bill being signed into law (OR = 18.78; 95% CI = 1.88, 187.42).

TABLE 3—

Bayesian Generalized Linear Mixed Effect Regression Models Stratified by Bill Classification (Antivaccination or Provaccination) and Passage (Signed Into Law) on the Basis of Political Climate and Exogenous Factors in the Legislative Process: United States, 2011–2017

Characteristic	Regression Modela
	Model 1: Antivaccination, OR (95% CI)	Model 2: Provaccination, OR (95% CI)	Model 3: Passage, OR (95% CI)
Signed into law
No (Ref)	1	1	. . .
Yes	0.09 (0.01, 0.93)	11.11 (1.08, 114.90)	. . .
Bill classification
Antivaccination (Ref)	. . .	. . .	1
Provaccination	. . .	. . .	18.78 (1.88, 187.42)
Year of bill proposalb	0.44 (0.26, 0.74)	2.26 (1.35, 3.78)	0.64 (0.26, 1.60)
Legislative chamber
House/Assembly (Ref)	1	1	1
Senate	0.95 (0.38, 2.36)	1.05 (0.42, 2.61)	2.10 (0.43, 10.24)
Legislative majority
Democratic (Ref)	1	1	1
Republican	1.03 (0.16, 6.68)	0.98 (0.15, 6.36)	0.57 (0.08, 3.91)
Split	0.94 (0.25, 3.48)	1.07 (0.29, 3.97)	0.07 (0.00, 1.33)
Bill sponsor’s party
Democratic (Ref)	1	1	1
Republican	12.01 (3.83, 37.61)	0.08 (0.03, 0.26)	0.83 (0.15, 4.51)
Governor’s party
Democratic (Ref)	1	1	1
Republican	0.70 (0.12, 4.16)	1.43 (0.24, 8.54)	0.18 (0.03, 1.21)
State population sizeb	0.75 (0.26, 2.16)	1.33 (0.46, 3.82)	1.02 (0.45, 2.30)
Census region
South (Ref)	1	1	1
Midwest	0.44 (0.04, 4.48)	2.30 (0.22, 23.66)	0.51 (0.04, 6.27)
Northeast	1.91 (0.17, 21.80)	0.52 (0.05, 5.95)	0.31 (0.03, 3.08)
West	0.45 (0.04, 4.68)	2.24 (0.21, 23.52)	4.35 (0.54, 34.95)

Note. CI = credible interval; OR = odds ratio.

^aEstimates correspond to mean OR of the posterior distribution with corresponding 95% CIs. A weak Gaussian prior was incorporated into coefficient estimation.

^bVariable mean centered and scaled for modeling. Effect can be interpreted per SD change relative to the average value.

DISCUSSION

Most bills proposed in state legislatures between 2011 and 2017 sought to expand, rather than restrict, nonmedical exemptions. However, and encouraging for public health, the bills ultimately enacted into law were overwhelmingly provaccination in that they eliminated or made it more difficult for parents to opt their children out of mandatory school vaccination. There have been calls in the scientific literature (and popular press) to increase vaccination rates by making exemptions more difficult to obtain.^16,17 Following suit, major national medical organizations, such as the American Medical Association and the American College of Physicians, have endorsed the removal of existing nonmedical exemptions from all states’ immunization laws.^18,19 Although these calls have merit and are appealing from a public health perspective, such regulations face substantial challenges in state legislatures. Although vaccine hesitancy is not the only reason for the increased incidence of vaccine-preventable diseases (e.g., a waning host immunity has been reported for the acellular pertussis vaccine²⁰), from a public health perspective, vaccine hesitance is perhaps the most important reason because outbreaks are often started and sustained by persons choosing not to vaccinate.

Many of the vaccine exemption bills in our analysis contained text that was not supported by empirical evidence. For example, a 2015 New Jersey assembly bill (A497) linked hepatitis B immunization with autism spectrum disorder (Table A presents an example of this text). Such instances demonstrate the ongoing damage done by the now infamous study implicating receipt of the measles, mumps, and rubella immunization with risk for developmental delays in childhood,²¹ which has been repeatedly disproven.²² Nevertheless, public opinion has substantial influence on legislative policymaking processes.²³ Although most Americans vaccinate their children, a vocal minority has permeated social media with antivaccination rhetoric and has eroded trust in the vaccine industry.^24,25

The results of our study highlight areas for future research about the sociopolitical processes through which pro- and antivaccination bills are developed, introduced, and voted on in state legislatures. Specifically, there would be benefit from studies that shed light on how research evidence is and is not used in these policymaking processes. Previous studies have used qualitative and document analysis methods to demonstrate how research evidence is used in obesity policymaking,²⁶ and these methods could be applied to immunization exemption legislation. Such studies could inform how provaccine advocates might more effectively infuse research evidence about vaccine exemptions into legislative processes.

Perhaps unsurprisingly, the sponsor’s political party was associated with proposed exemption legislation, in that Republican legislators were more likely to introduce a bill that would expand access to exemptions. Political party has also been shown elsewhere to correlate with general support for propublic health legislation (Democratic representatives were more likely to vote in favor of public health legislation¹³) and use of nonmedical exemptions (higher rates of vaccine refusal in areas with a greater proportion of registered Republicans²⁷). These results highlight the importance of learning how to effectively communicate evidence about the risks of immunization exemption to Republican legislators. Emphasizing information about the financial costs of vaccine-preventable disease outbreaks (e.g., a total of $394 448 and 10 054 personnel hours in the case of a 2013 measles outbreak in New York City²⁸) could be an effective approach, as previous research has shown that Republican and ideologically conservative state legislators place more value on economic evaluation data in health research than do Democratic and liberal state legislators.^29,30

Given the procedural hurdles to altering vaccination exemption laws, as acknowledged by others,³¹ short-term solutions are needed to voluntarily increase vaccination rates. To avoid long, drawn-out legislative and judicial action, public health practitioners and clinicians should consider improving grassroots advocacy and communication with antivaccination and vaccine-hesitant communities. Further, it is not inconceivable that widespread elimination of nonmedical exemptions could have an opposite effect and give rise to medical exemptions granted by fringe practitioners or an increase in homeschooling,³² a potentially self-damaging scenario to the public’s health. In a 2018 article, Weithorn and Reiss describe tools that states can use to increase vaccination rates.³³ These tools range from the use of force (compulsory vaccination) to education and incentives, and some techniques have yet to be tried, such as tort liability for failure to vaccinate. Administrative fees for processing nonmedical exemptions have also been proposed.³⁴

Limitations and Strengths

There are several limitations as well as notable strengths to this analysis. First, vaccination exemption bills may have been underestimated or misclassified. Budget reconciliation processes may include language affecting health law, and these were not included in the ASTHO data set. In most cases the intention of the bill was apparent, such as the 2015 Vermont provaccination House Bill H98, which removed the language “or philosophical convictions” from existing vaccination law. Nevertheless, some bills were somewhat nebulous in nature and the original intent of the bill’s sponsor was unclear. Second, an examination of past legislation does not necessarily indicate future trends. Our analysis should be viewed in the context of the years under study and not extrapolated beyond this period. Third, as mentioned, most bills introduced do not become law. Even though antivaccination bills rarely became law, provaccine groups should remain vigilant because of the increased legislative activity in the area of immunization exemptions. Fourth, the relatively small bill sample size and limited number of observations for some outcomes (e.g., number of bills passed into law) may have resulted in type II error in the regression models.

Our analysis has several strengths, including applying systematic and replicable policy-mapping methods and a robust statistical analysis. The policy-mapping approach allowed us to assess proposed legislation to identify important themes and subsequently inform the statistical analysis. Additionally, the Bayesian mixed effects regression approach provided a flexible framework for fitting hierarchical models that accounted for state-level effects in the data.

Public Health Implications

This analysis demonstrated that whereas most bills brought before state legislatures between 2011 and 2017 would have relaxed the ability to exempt one’s child from mandatory school entry vaccination, the bills ultimately executed into law restricted the ability to exempt. This is encouraging for public health, as fewer exemptions correlate to increased immunization with subsequently lower rates of community vaccine-preventable disease. Unfortunately, several bills were clearly not evidence based; this serves as an opportunity for provaccination constituents to become involved in the legislative process to ensure that proposed legislation reflects the state of the science.