Skip to main content
. 2017 Nov 16;14(11):1396. doi: 10.3390/ijerph14111396

Table 2.

Results from the multilevel logistic regressions for the odds of ARI symptoms among Nigerian preschool children from the North-Western and South-Southern communities, 2013 Nigeria Demographic and Health Survey (NDHS).

Explanatory Variables Crude Model Adjusted Model
uOR (95% CI) aOR (95% CI)
Regional characteristic
 Region of residence (ref = South-Southern)
  North-Western 2.94 (2.04–4.24) *** 2.86 (1.62–5.04) ***
Seasonal characteristics
 Seasonal factor (ref = wet season)
  Dry season 1.46 (1.06–2.03) * 1.42 (1.02–1.97) *
Household characteristics
 Cooking method (ref=in house kerosene/charcoal)
  In-house biomass fuel 4.85 (2.38–9.89) *** 1.76 (0.84–3.92)
  Outdoor or separate use of biomass fuel 3.02 (1.49–6.14) ** 1.55 (0.75–3.40)
  Outdoor use kerosene/charcoal/others 2.60 (1.14–5.91) * 2.25 (1.00–5.05) *
  Others 1.84 (0.47–7.20) 2.86 (0.87–8.25)
 Household poverty (ref = no)
  Yes 1.67(1.29–2.18) *** 1.42 (1.01–1.99) *
Maternal characteristics
 Maternal education (ref = no education)
  Primary education 1.36 (0.93–2.00) 1.07 (0.75–1.54)
  Secondary and above 0.76 (0.56–1.03) 1.22 (0.81–1.83)
 Mother’s religion (ref = Christian)
  Islam 2.16 (1.61–2.90) *** 0.91 (0.56–1.45)
  Traditionalist 2.85 (0.97–8.34) * 1.23 (0.28–5.35)
Child’s characteristics
 OVC status (ref = no)
  Yes 1.36 (0.88–2.09) 1.41 (0.88–2.26)
 Gender (ref = male)
  Female 1.06 (0.85–1.30) 1.12 (0.88–1.42)
 Age (ref = 3–5 years)
  <1 1.49 (1.14–1.94) ** 1.50 (1.11–2.02) **
  1–2 2.04 (1.59–2.61) *** 2.16 (1.64–2.84) ***
 Birth order (ref = 1–3)
  4–6 1.21 (0.95–1.54) 1.16 (0.88–1.54)
  >6 1.41 (1.07–1.86) * 1.58 (1.15–2.15) **
Neighborhood Characteristics
 Community OVC rate (ref = low 25%)
  High 25% community OVC rate 1.36 (0.93–2.00) 1.30 (0.95–1.78)
  Model statistics Coeff (SE)
  Log likelihood −1201.7363
Random variance
 Intra-class correlation (ICC) 0.08 *
 Variance between neighborhoods 0.56 *** (0.09)

Note: Intra-class correlation (ICC) measures the degrees of clustering with random intercepts. The correlation of the two-level multilevel logistic regressions was calculated by σμ2/(σμ2 + π2/3), where σμ2 denotes neighborhood-level variance. uOR refers to unadjusted odds ratios; aOR refers to adjusted odds ratios for covariates and CI refers to confidence interval; * p < 0.05; ** p < 0.01; *** p < 0.001.