Table 3 Mixed-effects ordinal logistic regression models of the association between neurodevelopmental disorders and poor academic performance on mathematics.

From: The negative impact of neurodevelopmental disorders and multiple co-occurring conditions on academic performance of school-age children and adolescents

 

Mathematics

OR (95% CI)

Model 0

Model 1

Model 2

Neurodevelopmental disorders

 Intellectual disabilities

(Ref. No)

8.24 (4.64–14.6)*

3.48 (1.68–7.21)*

6.90 (3.28–14.5)*

 Communication disorders

(Ref. No)

4.65 (3.66–5.90)*

2.92 (2.22–3.85)*

2.27 (1.71–3.01)*

 Autism spectrum disorders

(Ref. No)

2.35 (1.71–3.24)*

1.46 (1.03–2.09)*

1.35 (0.94–1.94)

 ADHD (Ref. No)

5.95 (5.30–6.68)*

4.75 (4.16–5.41)*

4.00 (3.50–4.57)*

 Specific learning disorder

(Ref. No)

5.99 (5.33–6.73)*

4.37 (3.84–4.98)*

3.68 (3.22–4.20)*

 Motor disorders (Ref. No)

2.17 (1.43–3.28)*

2.09 (1.34–3.25)*

1.28 (0.81–2.01)

Sociodemographic factors

 Gender (Ref. Girls)

1.09 (1.01–1.18)*

 

0.82 (0.74–0.90)*

 Age

1.14 (1.12–1.16)*

 

1.10 (1.08–1.13)*

 Socioeconomic status

0.96 (0.95–0.96)*

 

0.97 (0.97–0.97)*

 Ethnicity (Ref. Spanish)

1.73 (1.56–1.92)*

 

1.17 (1.04–1.31)*

 Parental divorce/separation

(Ref. No)

1.99 (1.78–2.22)*

 

1.44 (1.28–1.63)*

School-related variables

 Educational support

(Ref. No)

4.06 (3.56–4.64)*

 

1.72 (1.48–1.99)*

 Grade retention (Ref. No)

7.85 (6.12–10.1)*

 

2.42 (1.83–3.19)*

  1. OR, odds ratio; CI, confidence interval; ADHD, attention-deficit/hyperactivity disorder.
  2. Model 0: crude analysis; Model 1: adjusted for sociodemographic factors and school-related variables. Model 2: adjusted for sociodemographic factors, school-related variables, and other neurodevelopmental disorders.
  3. Covariate estimates for Model 1 are omitted, as each neurodevelopmental disorder was analyzed separately with adjustment for the same set of background characteristics, yielding distinct coefficients across disorder-specific models.
  4. *Significant at a FDR of 5% based on the number of predictors included within each model.