Table 5 Multilevel logistic regression analysis for underweight (referent: normal weight) (N = 12,222).

From: Application of single-level and multi-level modeling approach to examine geographic and socioeconomic variation in underweight, overweight and obesity in Nepal: findings from NDHS 2016

Variables

model 1

model 2

model 3

model 4

 

OR

LCL

UCL

OR

LCL

UCL

OR

LCL

UCL

Intercept

−1.28(0.14)

−1.03(0.15)

  

−1.04(0.17)

  

−0.82(0.18)

  

Age (years)

          

15–25

 

ref

  

ref

  

ref

  

25–35

 

0.45

0.40

0.52

0.448

0.4

0.51

0.63

0.53

0.74

35–45

 

0.48

0.41

0.55

0.46

0.4

0.53

0.60

0.50

0.73

45–55

 

0.59

0.50

0.68

0.566

0.5

0.66

0.71

0.58

0.86

55–65

 

1.05

0.91

1.21

1.005

0.9

1.16

1.19

0.98

1.46

>65

 

1.74

1.51

1.99

1.68

1.5

1.93

1.85

1.50

2.29

Sex

Male

 

0.89

0.81

0.97

0.899

0.8

0.98

0.92

0.83

1.01

Female

 

ref

  

ref

     

Education

No education, preschool

          

Primary

       

0.83

0.72

0.96

Secondary

       

0.69

0.59

0.80

Higher

       

0.52

0.42

0.63

Marital status

Formerly/ever married

       

1.20

1.02

1.42

Never married

       

2.30

1.99

2.67

Currently married

       

ref

  

Residency

Urban

       

0.87

0.76

0.99

Rural

          

Wealth quintile

Poorest

    

ref

  

ref

  

2

    

0.672

0.6

0.78

0.70

0.61

0.81

3

    

0.784

0.7

0.9

0.80

0.70

0.92

4

    

0.562

0.5

0.67

0.60

0.51

0.72

Richest

    

0.448

0.4

0.54

0.48

0.39

0.59

Error variance

Level 2 intercept*

0.12(0.07)

0.14(0.08)

  

0.16(0.09)

   

0.16(0.09)

 

Level 3 intercept*

0.07(0.02)

0.10(0.02)

  

0.14(0.03)

   

0.13(0.03)

 

ICC-level 2

3.50

3.40

  

4.60

   

4.60

 

ICC-level 3

2.10

2.90

  

4.10

   

3.80

 

PCV-level 2

 

2.9%

  

−31.4%

   

−31.4%

 

PCV-level 3

 

−38.1%

  

−95.2%

   

−81.0%

 

AIC

13053.4

12640.9

  

12563.8

   

12403.0

 

SBIC

13053.2

12622.9

  

12537.7

   

12364.9

 
  1. Model 1: empty model, model 2: adjusted for age and sex, model 3: model 2 plus wealth quintile, and model 4: model 3 plus education, marital status, and residency. The level two intercepts are for provinces and level 3 are for the districts. Abbreviation: AIC: akaike information criterion, LCL: lower confidence limit, SBIC: schwarz bayesian information criterion, UCL: upper confidence limit.
Back to article page