Research Article

Language Delay Is Not Predictable from Available Risk Factors

Table 2

Multivariate analysis. Effects of candidate predictor variables, reported as odds ratio for language delay with 95% confidence interval and value.

PredictorModel 1Model 2
Estimate (95% CI), valueEstimate (95% CI), value

Attends Nursery
 Yes versus no0.53 (0.20, 1.44),
HPI status at start of visit
 Additional versus core0.82 (0.25, 2.70),
 Intensive versus core1.02 (0.21, 4.93),
Involvement with non-SW services
 Yes versus no4.58 (1.16, 18.10), 4.31 (1.25, 14.86),
Gender
 Female versus male2.90 (1.06, 7.89), 2.66 (1.00, 7.11),
Child's behavioural and developmental problems
 Yes versus no8.26 (1.73, 39.43), 8.02 (1.89, 33.97),
Family behavioural and developmental problems
 Yes versus no6.06 (0.87, 42.40), 6.85 (1.07, 43.82),
Bilingual Family
 Yes versus no5.62 (1.76, 18.01), 5.89 (1.87, 18.57),

Model 1: all predictors with at univariate analysis. Model 2: best fitting model found by backwards selection, starting with model 1, with stepwise exclusion of terms with .