Study Design
A cross-sectional, population-level study of kindergarten children with teacher-identified UDNs, as reported on the EDI, was carried out in Canada between 2010 and 2015.
Study Population
The study population consisted of 603904 kindergarten children attending publicly-funded schools between the 2009/2010 and 2014/2015 school years from most Canadian provinces and territories (see Table 1). Based on population estimates,37 approximately 90% of children living in Canada attend publicly-funded kindergarten.38 These children were part of a population-level study of developmental outcomes of kindergarten children with health disorders, referred to as the Canadian Children’s Health in Context Study (CCHICS).39 The aim of the CCHICS was to establish a pan-Canadian database for monitoring developmental health and well-being of children with health disorders. The CCHICS merged pan-Canadian EDI data with neighbourhood-level SES data (see below for description of the measures). All children who met the following criteria were included in the study: 1) were enrolled in kindergarten; 2) were in their current classroom for at least one month; and 3) had a questionnaire with no more than 25% of items missing.
Table 1. Implementation of the EDI in Canada between 2009/2010 and 2014/2015
|
Province
|
2009/2010
|
2010/2011
|
2011/2012
|
2012/2013
|
2013/2014
|
2014/2015
|
Alberta
|
21976
|
20881
|
14492
|
20734
|
-
|
-
|
British Columbia
|
25033
|
21911
|
12485
|
30034
|
1289
|
-
|
Manitoba
|
-
|
12437
|
-
|
13538
|
-
|
13776
|
New Brunswick
|
-
|
-
|
-
|
-
|
-
|
-
|
Newfoundland & Labrador
|
|
1106
|
2135
|
4942
|
5182
|
|
Northwest Territories
|
-
|
-
|
672
|
659
|
654
|
645
|
Nova Scotia
|
913
|
2402
|
2257
|
8592
|
1397
|
8677
|
Nunavut
|
-
|
-
|
-
|
-
|
-
|
-
|
Ontario
|
33305
|
38728
|
57038
|
|
|
135936
|
Prince Edward Island
|
-
|
-
|
-
|
-
|
-
|
-
|
Quebec
|
-
|
-
|
65498
|
-
|
-
|
-
|
Saskatchewan
|
8625
|
5501
|
552
|
8427
|
|
|
Yukon
|
362
|
344
|
368
|
401
|
-
|
-
|
Measures
The primary exposure variable was teacher-reported UDNs. This was measured by teachers’ answers to the following question: “Does the student have a problem that influences his/her ability to do schoolwork in a regular classroom? If yes, please mark all that apply.” There were 11 different options, one of which was “unaddressed dental needs” (see Table 2). This option was added to the questionnaire in 2010, as a response to a review of comments from teachers received in the previous 6 years of implementation of the EDI, which indicated that dental needs were considered as an impairment of children’s ability to participate. A small group of kindergarten teachers provided feedback on the usefulness and wording of the item, indicating that it was a feasible and useful addition. It should be noted that the UDNs reported by teachers most likely represent the most severe cases as it is improbable that minor or even moderate dental caries would affect children’s ability to function in a regular classroom. As such, the prevalence of UDNs reported in our study is presumably an underestimate of the true prevalence of UDNs among 5-year-olds.
Table 2. Functional impairments available on the EDI
Does this child have a problem that influences his/her ability to do schoolwork in a regular classroom?
|
a. Physical disability
|
b. Visual impairment
|
c. Hearing impairment
|
d. Speech impairment
|
e. Learning disability
|
f. Emotional problem
|
g. Behavioural problem
|
h. Home environment/problems at home
|
i. Chronic medical/health problems
|
j. Unaddressed dental needs
|
k. Other
|
The primary outcome was vulnerability in any of the developmental domains of the EDI (i.e., vulnerable on one or more domains). Secondary outcomes were vulnerability in each of the five developmental domains of the EDI: physical health and well-being, social competence, emotional maturity, language and cognitive development, and communication skills and general knowledge. Covariates included variables which are known or suspected to be potential confounders of the relationship between oral health and child development, identified through an extensive literature review. These included child’s age, sex, special needs designation, having English or French as a second language (E/FSL), and area-level SES.
Early Development Instrument
The EDI is a 103-item, teacher-completed questionnaire that measures children’s ability to meet age-appropriate developmental expectations prior to entering Grade 1.33 The EDI is completed in the second half of the school year by kindergarten teachers for each student in their class. The EDI measures five general domains of development: physical health and well-being, social competence, emotional maturity, language and cognitive development, and communication skills and general knowledge. In addition to measuring children’s development, the EDI collects demographic information for each child, such as their date of birth, sex, whether they have E/FSL (yes or no), and whether they have been identified as having a special need (yes or no; this is a school-based designation that identifies children with a chronic condition or those who require additional assistance in the classroom).33 Since 2010, the EDI has collected information on whether children have UDNs, either through information provided by the parents or through teacher observations. Teachers can select either of these options or both; the response options are combined into one dichotomous variable (UDNs, yes/no).
There are two main EDI outcomes: overall vulnerability and individual domain scores. Domain scores are an average of the items that contribute to each domain, which vary from 0 to 10, with a higher score indicating greater ability.33 The EDI scores for each domain were then divided into categories representing the highest to the lowest scores in the given population. The distribution of scores across each of the five domains are utilized to determine percentages of children at various levels of developmental health. Children scoring below the 10th percentile in one or more of the five domains are categorized as “vulnerable” in terms of their developmental health based on national standards.40 The EDI has been used extensively throughout Canada, Australia, and other parts of the world.41 Over the past decade, several studies have examined different psychometric properties of the EDI, including between-group reliability,34 construct validity,35 cross-cultural validity,36 and predictive validity.42 The results have consistently shown that the EDI is valid and reliable and can be reliably used as a measure of early child developmental health. The internal consistencies, using Cronbach’s alphas for each domain were .78 for physical health and well-being, .96 for social competence, .93 for emotional maturity, .91 for language and cognitive development, and .94 for communication skills and general knowledge.
Neighbourhood-level SES
Information on neighbourhood-level SES was retrieved from the 2010 Taxfiler databases and the 2011 National Household Survey which are national Canadian surveys collected through Statistics Canada.42 A SES index identifying 10 developmentally-relevant socioeconomic variables was created for 2,058 custom-defined neighbourhoods across the country. The index measures aspects related to household income, education, mobility, immigration, single parenthood, and first language.43 The SES index was transformed into Z-scores, with a mean of 0 and a standard deviation of 1. The neighbourhood SES index was merged with the EDI dataset using children’s postal codes with a 99.3% match rate.
Analytical strategy
UDNs were examined, comparing the percentages of children with UDNs, as reported by their teacher, in each of the provinces/territories included in the study. Descriptive statistics including means and proportions were examined for children with and without UDNs. Children’s age, sex, special needs designation, E/FSL, and area-level SES were compared between children with and without UDNs using contingency tables. For the primary analysis, a binary logistic regression (BLR) model was developed to determine the association between UDNs and overall vulnerability in any of the developmental domains (i.e., vulnerable on one or more domains), while controlling for the pre-specified potential confounding variables mentioned above. For the secondary analysis, if the association between UDNs and overall vulnerability was statistically significant (p<0.05) then an additional five BLR models were constructed to examine the association between each specific EDI domain, as mentioned above, with adjustment for the same potential confounding variables as the primary model. To account for multiple hypothesis testing, the level of statistical significance for each secondary analysis was set at p<0.01. All statistical analyses were conducted using the statistical software SPSS, version 25.44 CCHICS has been approved by the Hamilton Integrated Research Ethics Board and the University of Manitoba Health Research Ethics Board.