Tingyi’s Geob479 blog

Example of Conducting Analysis of Spatial Patterns of Nonmedical Exemptions

for Public Health Concerns
Student: Tingyi Liu

Original Paper Access : https://www.tandfonline.com/doi/abs/10.1080/00330124.2019.1612257

More parents of students in Illinois schools opt for nonmedical exemptions (NMEs) and
causing decreasing in vaccination rates which can affect herd immunity levels. There are just a
few previous studies had conducted or proposed methods to conduct local-scale spatial and
temporal cluster pattern analysis, and this study conducted by Elizabeth A. Dzwonczyk explored
further for the clusters.
Before Dzwonczyk’s study, there is several NME cluster detection methodological research
that mostly focused on NME choice, school policies, or type of school (Salmon and Siegel 2001;
Salmon, Omer, et al. 2005; Lai et al. 2014). However, if Illinois would need to identify areas
with higher percentages of unvaccinated children, such as an outbreak or epidemic, the current
data format is not easily accessible or useful. This study contributes to the literature by
introducing an approachable cluster detection method using a GIS. The result identified the
presence and changes of spatial clustering of NMEs in Illinois K–12 schools between the
2003–2004 and 2013–2014 academic years using the Getis–Ord Gi* statistical test.
The main method used in this study is the Getis–Ord Gi* statistical test, which had been
commonly used in public health associated studies. This test can identify hot and cold spots,
including patterns of infectious disease, and the leading cause of death, etc. The vaccination
exemption data used in this study was from the Illinois School Board of Education’s (ISBE)
annual Immunization School Survey Results for the 2003–2004 and 2013–2014 academic years.

The author explained that the reason she chose these two academic years is data availability. The
author chose a ten-year time interval to analyze a large turnover of students and parents
signifying a shift toward increased spatial clustering of NME children.
The Getis–Ord General G statistic was used to identify cluster detection by the individual
vaccine at the school level. The Getis–Ord Gi* statistic was used to identify hot and cold spots
with two different parameter models – eight nearest neighbours (8NN) and a hybrid model
consisting of a distance threshold of 2.3 miles or a minimum of eight nearest neighbours. The
author used the incremental spatial autocorrelation tool in ArcGIS 10.2.2 to select a distance
threshold for the hybrid model.
The z score results from the Getis–Ord General G tests are varied among different vaccines
over time but still indicated high levels of NMEs in Illinois increased between the 2003–2004
and 2013–2014 academic years. The Getis–Ord Gi* 8NN model identified NME hot spot
clusters (90, 95, and 99 percent statistical significance) for all vaccines and academic years. The
Getis–Ord Gi* hybrid model identified 90 percent (all vaccines) and 95 percent (rubella and
mumps only) confidence level cold spots for the 2003–2004 academic year and all confidence
levels for the 2013–2014 academic year. Some visible differences between the hot spot cluster
results from two models existed and were compared in maps in the paper. Besides the increase
NMEs level, the author mentioned some spatial differences between 2003-2004 and 2013-2014
results, for instance, the 2003–2004 hot spots were located mostly in urban areas while the hot
spots of the 2013–2014 academic year were identified in both urban areas and rural and lower
population density areas.

The author also conducted the hot spots and cold spots analysis based on schools for
Chicago-area measles exemptions and provided maps showing the details. The pattern of hot and
cold spot clusters in Chicago was compared and discussed with other cities such as Los Angeles.
The author discussed the results by relating the results and provided the reason for the details
of the decreasing change. For example, the author mentioned the polio NMEs increased a lot
more than the other vaccines because the global polio incidence rate is very low and the United
States has been polio-free since 1993 (CDC 2018). Thus, the increased polio NMEs’ clustering
in Illinois is because of the relatively low risk for polio. The author also gave potential
explanations for other details in the results and related them with what the government or the
people should consider public health after being aware of the NMEs rates change.
The author concluded that generally the total number of schools decreased, but there were
more hot and cold spots identified during the time. The hybrid model displayed the dichotomy of
vaccination exemptions with the change in the hotspots and cold spots’ numbers, whereas the
8NN model best avoided potential effects of school density as the spatial area analyzed is based
on the distance of the eight nearest neighbours, rather than using a fixed distance.
The limitations of this paper include that the spatial data did not completely include all the
schools, and that differences among schools such as the number of students and types (i.e.
whether it is a public, private, charter, or magnet school). This study did not account for
homeschooled children as there was no data available about this. Other limitations such as the
certain degree of the coarseness and simplicity of this study and the scale that this study only
focused on the clusters of schools but not parents’ location, and not includes schools other than
K-12 schools.

Overall, the author conducted suitable methods and provided a basic but clear explanation of
the methodology in this paper. I like that the author provided a lot of the explanation of the result
and discussion about it. The quality of the maps of clusters is also good (simple but clear). In this
paper, the author introduced the theoretical basic, methodology and practical significance usually
with some short details about some supporting literature or background facts. This way is easy
for the audience to understands the steps she went through and make this paper’s structure good.
I hope there would be more new studies about this issue (i.e. Delamater, Leslie, et al. 2016) and
provided more helpful results for understanding the public health situation and making it better.

Bibliographic Information

Delamater, P. L., Leslie, T. F., & Yang, Y. T. (2016). A spatiotemporal analysis of non-medical
exemptions from vaccination: California schools before and after SB277. Social
Science & Medicine, 168, 230 238.

Dzwonczyk, E. A. (2020). Clusters of Nonmedical Exemptions to Vaccination in Illinois K–12
Schools. The Professional Geographer, 72(1), 22-36.
Salmon, D. A., Moulton, L. H., Omer, S. B., DeHart, M. P., Stokley, S., & Halsey, N. A. (2005).
Factors associated with refusal of childhood vaccines among parents of school-aged
children: a case-control study. Archives of pediatrics & adolescent medicine, 159(5), 470
-476.

Salmon, D. A., and A. W. Siegel. 2001. Religious and philosophical exemptions from
vaccination requirements and lessons learned from conscientious objectors from
conscription. Public Health Reports 116 (4):289–95. DOI: 10.1093/phr/116.4.289.