Poster #35 - Population-level SES and child development: Using Latent Profile Analysis to identify “off-diagonal” neighbourhoods

• In Event: 3-001 - Poster Session 2 during Breakfast

Abstract

Background
There is a growing body of evidence indicating a gradient between child development outcomes and socioeconomic status (SES) at the individual level. Even though the association between development indicators and neighbourhood SES is less well studied, it also shows a similar gradient. However, there is also evidence that some communities, labelled “off-diagonal”, diverge from this trend. Some “high” SES neighbourhoods have poorer than expected child outcomes, whereas some “low” SES neighbourhoods child outcomes are better than expected.

One of the primary objectives of the Canadian Neighbourhoods Early Child Development (CanNECD, Guhn et al, 2016) project was to identify and characterize such “off-diagonal” neighbourhoods. This was done using data from the Canadian Census and Taxfiler databases for custom-defined neighbourhoods across Canada which were linked to child development outcomes collected with the Early Development Instrument (EDI; Janus & Offord, 2007).

Building on earlier approaches (Kershaw et al., 2009; Hertzman, 2011; Tanton et al., 2017), this study investigated the existence and frequency of off-diagonal neighbourhoods by identifying possible homogenous groups based on neighbourhood-level child development outcomes (EDI) and SES index. It then used the derived groups to identify and describe off-diagonal neighbourhoods.

Methods
EDI data for 2038 neighbourhoods collected between 2008 and 2013 (Webb et al., 2016) were examined. Five indicators of developmental vulnerability were derived from these data. The CanNECD SES indicators were derived from the 2006 Census or the 2005 Taxfiler database (Forer et al., in prep). We used exploratory Latent Profile Analysis (LPA) in a structural equation model framework with Mplus software (Muthén & Muthén, 1998-2015) to identify possible subgroups separately for SES and EDI vulnerability. Last, we used contingency tables analysis to identify off-diagonal groups of neighbourhoods.

Results
We derived three child development groups based on prevalence of vulnerability: low (A, 57.2%), medium (B, 35.6%), and high (C, 7.3%). We also derived four meaningful SES groups: low (1, 31.6%), low-moderate (2, 12.7%), high-moderate (3, 38.4%) and high (4, 17.4%) (Figure 1). Contingency table analysis of these groups showed that there were 209 off-diagonal neighbourhoods in the low SES and low EDI vulnerability group and no neighbourhoods in the high SES and high EDI vulnerability group (Figure 2).

Conclusion
The identification of off-diagonal neighbourhoods contributes to our understanding of modifiable and moderating factors influencing child development at the neighbourhood level. Our findings will be expanded through detailed analyses of demographic and geographic analyses of the off- and on-diagonal neighbourhoods.

Authors

• Eric Duku, McMaster University
  Presenting Author

• Magdalena Janus, McMaster University

• Simon Webb, McMaster University

• Caroline Reid-Westoby, McMaster University

• Barry Forer, University of British Columbia

• Martin Guhn, University of British Columbia

• Marni Brownell, University of Manitoba/Manitoba Centre for Health Policy

• Nazeem Muhajarine, University of Saskatchewan