Skip to main content

Table 5 Multilevel linear regression model of body mass index Z-score using the WHO classification system, in the analytical dataset, El Salvador, 2015/2016

From: Overweight and obesity of school-age children in El Salvador according to two international systems: a population-based multilevel and spatial analysis

CharacteristicIntercept only modelStudent fixed predictorsSchool predictorsMunicipal predictors
Full model
 Coefficient (95% CI)Coefficient (95% CI)Coefficient (95% CI)Coefficient (95% CI)
Intercept0.39 (0.37, 0.42)a1.54 (1.44, 1.65)a1.36 (1.26, 1.46)a1.36 (1.15, 1.56)a
Student level
 Sex (ref.: girls) 0.15 (0.13, 0.16)a0.14 (0.13, 0.16)a0.14 (0.13,0.16)
 Age −0.16 (−0.17, −0.15)a−0.15 (−0.16, − 0.13)a− 0.15 (− 0.16, − 0.13)a
School level
 Location (ref.: rural)  0.23 (0.20, 0.26)a0.22 (0.20, 0.25)a
 Type (ref.: public)  0.33 (0.29, 0.37)a0.32 (0.29, 0.36)a
Municipal level
 Human development index (ref.: 0.68 or more)   0.12 (0.08, 0.17)a
 Logarithm of population size   −0.007 (−0.026, 0.012)b
Random effects
 Student level variance1.6461.6351.6361.63
 School/municipal level variance0.09590.09040.06400.063
Municipal level variance0.02850.02710.01300.009
Student ICC92.9%93.3%95.5%95.4%
School ICC7.03%6.7%4.5%4.5%
Municipal ICC1.6%1.5%0.7%0.5%
Deviance3909.49 (Reference)886.541642.891680.45
P-value<  0.001<  0.001<  0.001<  0.001
  1. Ref.: category of reference. Numbers in parentheses represent the 95% confidence interval. ICC intra-class correlation coefficient. ap-value < 0.01, bp-value = 0.478
\