Concept: Early Development Instrument (EDI) Outcomes

Introduction

This concept describes the Early Development Instrument (EDI) and the outcomes associated with this instrument. It also provides information on the areas (domains) covered by the EDI and the outcomes measures, describes the statistical modeling techniques employed to predict EDI outcomes, and lists the key findings related to EDI outcomes from several MCHP deliverables along with providing links to the relevant sections in these reports.

NOTE: The terminology and outcome measures related to the EDI have changed over time. Please see the section titled Early Development Instrument (EDI) Outcome Measures for more specific information on these changes.

The Early Development Instrument (EDI) was originally implemented in order to provide a consistent measure of specific outcomes indicative of a child's "School Readiness" in their first year of education. In terms of the EDI, "School Readiness" refers to a child's ability to meet the demands and expectations of the kindergarten classroom, such as being comfortable exploring and asking questions, the ability to hold a pencil, listening to the teacher, playing and working with other children, and remembering and following rules (Janus, M. et al., 2007). In current EDI research, the focus has changed from a child's "school readiness" to a child's "developmental vulnerability".

These traits of developmental vulnerability/school readiness are measured in an assessment filled out by the child's teacher that divides child development into five general domains (detailed below). The EDI is typically administered to children aged 4 to 5 at either a junior or senior kindergarten level, and includes 104 fundamental items to measure vulnerability/school readiness (Janus, M. et al., 2007). NOTE: In 2019, we noticed the EDI now includes 103 fundamental items (Source: Offord Centre for Child Studies - Early Development Instrument Web Site - What is the EDI? web page - accessed June 13, 2019).

In 2012, the EDI was collected throughout Manitoba in all 37 public school divisions on behalf of Healthy Child Manitoba (HCM) - services now delivered by Manitoba Child and Youth Programs - by all Kindergarten teachers for each student.

The majority of information in this concept was taken from the deliverable The Early Development Instrument (EDI) In Manitoba by Santos et al. (2012) and through discussion with staff directly involved in the research project.

Early Development Instrument (EDI) Domains and Sub-Domains

The EDI has five domains. Four of the five domains have a series of sub-domains. The domains and sub-domains are listed below:


1. Physical Health and Well-Being with the sub-domains:
   
   • Physical readiness for school day
     
   • Physical independence
     
   • Gross and fine motor skills
   
   
2. Social Competence with the sub-domains:
   
   • Overall social competence
     
   • Responsibility and respect
     
   • Approaches to learning
     
   • Readiness to explore new things
   
   
3. Emotional Maturity with the sub-domains:
   
   • Pro-social and helping behaviour
     
   • Anxious and fearful behaviour
     
   • Hyperactivity and inattention
   
   
4. Language and Cognitive Development with the sub-domains:
   
   • Basic literacy
     
   • Interest in literacy/numeracy and uses memory
     
   • Advanced literacy
     
   • Basic numeracy
   
   
5. Communication Skills and General Knowledge Domain

For more information, see: Table 1.1 - The Early Development Instrument (EDI) in Manitoba in Santos et al. (2012), or read the information available in the Links Section below.

Early Development Instrument (EDI) Outcome Measures

The five domains of the EDI serve to isolate the different aspects of child development. Therefore, a child can show vulnerability in one domain and not in another. When evaluating a child's developmental vulnerability/school-readiness using the EDI, there are three or four possible outcomes for each domain.

Note: The terminology and measures for EDI outcomes has changed over time. The following lists indicate the previous terminology and measures published in several published MCHP research projects, and the current terminology and measures in use.

In previous research, the terminology and outcome measures related to a child's school readiness included:


• Not Ready (lowest 10%)
  
• Ready (11-69%)
  
• Very Ready (top 30th percentile)

In current research, the terminology and outcome measures related to a child's developmental vulnerability include:


• Vulnerable (lowest 10%)
  
• At Risk (10-25%)
  
• Middle (25-75%)
  
• Top (75-100%)

In MCHP research, usually a dichotomous (yes or no) Vulnerable variable is calculated for the outcome.

For more information on current outcome measures, please read the following: How to interpret EDI results from the Offord Centre website (accessed January 25, 2022). According to this information, individuals who are identified as "vulnerable" or "at risk" can be classified as "not on track", and those individuals who are identified as "middle" or "top" can be classified as "on track".

In addition to vulnerability, the EDI also has a Multiple Challenge Index (MCI), an indicator of a child experiencing challenges in at least three EDI domains. The MCI is scored based on challenges in nine or more sub-domains.

Predicting EDI Outcomes using Statistical Modeling Techniques

In order to predict EDI outcomes, three statistical modeling techniques can be used:


1. Structural Equational Modeling (SEM)
   
2. Logistic Regression, and
   
3. Multi-level Logistic Regression

Structural Equational Modeling (SEM) is used to model children's average EDI scores for each of the five EDI domains. Both logistic and multilevel logistic regression are used to model the following outcomes: Very Ready in one or more domains (VR1+), Not Ready in one or more domains (NR1+), Multiple Challenge Index (MCI), Very Ready in each domain (VR), and Not Ready in each domain (NR).

The EDI data for the SEMs are continuous (e.g., scores ranging from 0 to 10), whereas the EDI data for the logistic and multilevel logistic regressions are dichotomous (i.e., present or absent, wherein we used "dummy variables" of 1 or 0 to denote the presence or absence of the outcome, respectively).

For more information about the statistical modeling techniques used to predict EDI outcomes, see the Modeling section in Chapter 1: Introduction and Methods in Santos et al. (2012).

Measuring Vulnerability

The EDI is administered at the Kindergarten level and is designed to measure population-level development during early childhood. Therefore, when measuring vulnerability, the EDI outcomes primarily indicate early development vulnerability at age five. Children's vulnerability at age five was analyzed in depth in Santos et al. (2012) as a function of:


1. socioeconomic status (SES);
   
2. early life health;
   
3. factors that put children at increased risk; and
   
4. the relationship of epigenetics and EDI outcome.

1. Socio-economic Status (SES) and EDI Outcomes

Across income quintiles, the prevalence of children's vulnerability at age five is considerable, ranging from 21% to nearly 40% of children in a given income quintile and representing thousands of Manitoban children at the crucial developmental transition from the early years into school at Kindergarten.

EDI outcomes as a function of SES can be further broken down into the percent of children not ready (vulnerable) at age five by urban or rural income quintile, age, gender, RHA of residence or Winnipeg community of residence.

For more information see:

• Chapter 2: Socioeconomic Adversity and Children's Vulnerability at Age Five in Santos et al. (2012).
  
• Table 2.2: Number of Children Not Ready at Age 5 by Urban and Rural Income Quintiles, Manitoba in Santos et al. (2012).

2. Early Life Health and EDI Outcomes

Early life health includes children's health at birth (gestational age, birth weight, preterm birth, long intensive care unit (ICU) stay, long hospital stay) up until their EDI outcomes at age five. This also includes children's major and minor illness in early childhood to age four.

In summary, in the context of socioeconomic adversity, the five EDI outcomes all share a common pathway originating from biological vulnerability at birth.

For more information see:

• Chapter 3: Biological Vulnerability at Birth and Children's Vulnerability at Age Five in Santos et al. (2012).
  
• Figure 3.5: Birth Measures by Urban and Rural Income Quintiles, Manitoba in Santos et al. (2012).
  
  • This figure details the measures of biological vulnerability at birth that also exhibit SES gradients, including, long ICU stay, preterm birth, and long hospital stay.

3. Children "At-Risk"

Researchers have identified three "at-risk" groups of children when implementing the EDI. That is, three groups which are more likely to have a higher prevalence of children that fall into the "Not Ready" category or the MCI. These groups are:


• children born to mothers who were teenagers at their first childbirth (see Teen (Mother) at First Birth);
  
• children in families that were ever on Income Assistance (IA); and
  
• children in the care of Child and Family Services (CFS).

For more information, please see:

• Chapter 4: Children's Vulnerability at Age Five in At-Risk Groups in Santos et al. (2012).
  
• Figure 4.5: Percent Not Ready at Age 5 by Number of Risk Factors, Winnipeg in Santos et al. (2012).
  
  • This figure details the prevalence of children's vulnerability at age five for each of the at-risk groups and their combinations.
• Figure 4.7: Odds Ratios for At-Risk Groups for Outcomes from Age 5 to Adulthood, Winnipeg in Santos et al. (2012)
  • This figure details the odds ratios for children's vulnerability at age five for each of the varying combinations of the three at-risk groups, in comparison to the findings of Brownell et al. (2010) for poor outcomes in youth and early adulthood (not completing eight credits by the end of Grade 9, not completing high school, becoming a teen mom, receiving income assistance as a young adult).

4. Epigenetics: What Causes the Biology of Adversity in EDI Outcomes

Epigenetics is the study of changes in gene activity without any change in the structure of the DNA where the genes reside. In other words, epigenetics refers to how the environment interacts with the genome. Depending on the environmental circumstances, the expression of a gene can be activated or inhibited (turned "on" or "off"), causing different visible effects. Such changes are most profound in early development.

To read more about epigenetics and EDI outcomes see Chapter 5: Linking Biological Vulnerability and Socioeconomic Adversity in Santos et al. (2012).

Key Findings in MCHP Research

Key findings related to the EDI and outcome measures found in MCHP research include:

1. Santos et al. (2012)

In the deliverable, The Early Development Instrument (EDI) by Santos et al. (2012), there was a focus on the socioeconomic inequities in health, learning, and behavior, as well as the early life (in utero and at birth) of children and its significance for children's early developmental outcomes at age five. Santos et al. also categorized and examined the three at-risk groups of children that were especially vulnerable in early development.

For more information on the key findings in this report please see Chapter 7: Summary and Conclusion.

2. Brownell et al. (2012)

In the deliverable, How are Manitoba's Children Doing? by Brownell et al. (2012), they investigated the relationship between level of EDI vulnerability and Grade 3 vulnerability and found an association with both reading and numeracy assessments. Figure 5.74: Grade 3 Students Not Meeting or Approaching Reading and Numeracy Assessment Expectations by Number of Vulnerabilities on EDI in Kindergarten shows that with each increase in EDI vulnerability, there was an increase in the percent of students who were vulnerable on both the Grade 3 reading and numeracy assessments.

They further elaborate on the factors associated with EDI outcomes and the complex inter-relationship of these factors. It is possible for some variables to be both an outcome as well as a predictor of subsequent outcomes. These variables cannot be measured directly and therefore must be estimated by related variables known as "latent constructs".

For more information, see the section titled Factors Associated with Outcomes in Chapter 5 of this deliverable.

3. Chartier et al. (2012)

In the deliverable, Health and Healthcare Utilization of Francophones in Manitoba by Chartier et al. (2012), they investigated EDI outcomes comparing a Francophone cohort and a matched cohort. Overall, they found that children in the Francophone Cohort had a higher "not ready for school" rate. The groups were similar for the domains physical health & well-being and social competence, but Francophone children scored less in the emotional maturity, language & cognitive development, and communication skills & general knowledge domains.

For more information, see section 16.1 Early Development Instrument in this deliverable.

4. Smith et al. (2013)

In the deliverable, Social Housing in Manitoba. Part II: Social Housing and Health in Manitoba: A First Look by Smith et al. (2013), they investigated the percent of children classified as being "not ready" in at least one EDI domain over two school years: 2005-06 and 2006-07, comparing a Social Housing cohort to All Other Manitobans.

They found the provincial percentage of Kindergarten students not ready for school was higher in the Social Housing cohort (44.6%) than in the All Other Manitobans cohort (26.4%). In Winnipeg, the percentage of Kindergarten students not ready for school was higher in the Social Housing cohort (45.8%) than in the All Other Manitobans cohort (26.1%).

For more information, see the section titled Children Not Ready for School (in One or More Early Development Instrument (EDI) Domains) in the deliverable.

5. Brownell et al. (2015)

In the deliverable, The Educational Outcomes of Children in Care in Manitoba by Brownell et al. (2015), they investigated the educational outcomes of children in care, including the EDI. They looked at the percent of children who were not ready in one or more domains and not ready in two or more domains.

The proportions of children deemed not ready in one or more domains were 53.1% who were ever in care, 43.5% who were never in care but whose families had received services from CFS, and 23.8% who were never in care or never received services from CFS. The proportions of children deemed not ready in two or more domains were 36.3% who were ever in care, 26.1% who were never in care but whose family had received services from CFS, and 12.0% who were never in care or never received services from CFS.

For more information:


• on the EDI outcomes of children in care, see the section titled Early Development Instrument (EDI) in Chapter 4 of the deliverable.
  
• on the use of statistical modeling to identify the factors associated EDI outcomes of children in care, see the section titled Early Development Instrument (EDI) in Chapter 5 of the deliverable.

SAS Macro Code

There is an EDI SAS macro available in the MCHP SAS Macro Library: _EDI (internal access only). Currently, it is designed to work with the data up to and including the 2016/17 EDI data.

For more descriptive technical information on the EDI macro, see the Grade K EDI PowerPoint presentation from the January 23, 2019 Data Analyst Meeting, available in the LINKS section below (internal access only).

IMPORTANT NOTES:

• In December, 2020 a coding error was identified when determining the developmental vulnerability/not ready categories for two of the five domains (physical health & well-being, and communication skills & general knowledge). The values are coded wrong in the data / at the source (i.e., Manitoba government agency). This error is not being corrected in the current MCHP EDI data available from 2005/06 to 2016/17, but should be corrected in the next EDI data extract.
  
  The errors were small but may have affected the way some children/results were categorized. The erroneous code and the corrections are listed below, with corrected cut-offs highlighted in blue:

  Developmental Vulnerability / Not Ready:
  
  Physical Health & Well-Being
  *if phys>=0 & phys<6.9232 pprob=1.
  *above was incorrect below is correct.
  if phys>=0 & phys<7.0833 pprob=1.
  
  Communication & General Knowledge
  *if comgen>=0 & comgen<4 cprob=1.
  *above was incorrect below is correct.
  if comgen>=0 & comgen<4.3750 cprob=1.

  The EDI SAS macro that is used at MCHP has been updated to reflect this change in calculating pprob and cprob values. If you do not use the EDI macro, then make sure to manually add these two conditions into your code. Please check with your data analyst to ensure you are using the updated macro/code.
  
  Also, if your analysis looks at the prob1 var – vulnerable in 1 or more domains and/or prob2 var – vulnerable in 2 or more domains, please make sure to re-calculate these values as they are likely affected by the change in the value of pprob and cprob.
  
  
• The labels generated by the macro have also been updated to reflect the change in terminology from "school readiness" to "developmentally vulnerable".
  
  
• The EDI macro will be reviewed and updated accordingly when the next set of EDI data becomes available in the Repository.