Table 2.
Multilevel logistic regression models of correlates of mosquito net use among women in Nigeria
| Variable | Model 1a | Model 2b | Model 3c | Model 4d |
|---|---|---|---|---|
| aOR (CrI) | aOR (CrI) | aOR (CrI) | aOR (CrI) | |
| Individual-level factors | ||||
| Age | ||||
| 15–24 | 0.97 (0.82–1.14) | 0.94 (0.79–1.11) | 0.93 (0.78–1.09) | |
| 25–34 | 1.07 (0.92–1.23) | 1.07 (0.91–1.26) | 1.05 (0.89–1.23) | |
| 35+ | 1 (reference) | 1 (reference) | 1 (reference) | |
| Education | ||||
| No education | 1.01 (0.82–1.24) | 0.87 (0.69–1.08) | 0.86 (0.69–1.06) | |
| Primary | 1.23 (1.01–1.48) | 1.17 (0.95–1.41) | 1.16 (0.94–1.41) | |
| Secondary/higher | 1 (reference) | 1 (reference) | 1 (reference) | |
| Household wealth index | ||||
| Poor | 3.00 (2.33–3.87) | 2.40 (1.81–3.19) | 2.33 (1.76–3.05) | |
| Middle | 2.32 (1.92–2.77) | 2.09 (1.71–2.56) | 2.09 (1.69–2.56) | |
| Rich | 1 (reference) | 1 (reference) | 1 (reference) | |
| Mosquito causes malaria | ||||
| No | 1 (reference) | 1 (reference) | 1 (reference) | |
| Yes | 2.10 (1.71–2.53) | 2.09 (1.73–2.51) | 2.08 (1.71–2.58) | |
| Exposed to malaria messages | ||||
| No | 1 (reference) | 1 (reference) | 1 (reference) | |
| Yes | 1.39 (1.22–1.59) | 1.40 (1.22–1.61) | 1.39 (1.21–1.59) | |
| Chances of getting malaria are the same | ||||
| No | 1 (reference) | 1 (reference) | 1 (reference) | |
| Yes | 1.46 (1.25–1.67) | 1.43 (1.25–1.65) | 1.46 (1.25–1.69) | |
| Drugs for preventing malaria in pregnancy are effective | ||||
| No | 1 (reference) | 1 (reference) | 1 (reference) | |
| Yes | 2.39 (1.89–2.99) | 2.43 (1.93–3.06) | 2.38 (1.91–3.04) | |
| Tests are a good way to detect malaria | ||||
| No | 1 (reference) | 1 (reference) | 1 (reference) | |
| Yes | 3.17 (2.29–4.19) | 3.23 (2.44–4.23) | 3.29 (2.54–4.33) | |
| ACT is effective in treating malaria | ||||
| No | 1 (reference) | 1 (reference) | 1 (reference) | |
| Yes | 1.04 (0.87–1.25) | 1.05 (0.91–1.23) | 1.05 (0.87–1.22) | |
| Number of household members | ||||
| < 5 | 1.18 (1.02–1.36) | 1.21 (1.05–1.38) | 1.21 (1.05–1.39) | |
| 5+ | 1 (reference) | 1 (reference) | 1 (reference) | |
| Community-level factors | ||||
| Residence | ||||
| Urban | 1 (reference) | 1 (reference) | ||
| Rural | 1.001 (0.72–1.29) | 1.05 (0.77–1.39) | ||
| Region | ||||
| North Central | 1 (reference) | 1 (reference) | ||
| North East | 1.29 (0.62–3.79) | 1.05 (0.55–2.15) | ||
| North West | 1.44 (0.61–4.03) | 1.02 (0.43–2.40) | ||
| South East | 0.46 (0.18–1.21) | 0.76 (0.39–1.39) | ||
| South South | 1.02 (0.42–2.17) | 2.10 (1.05–3.84) | ||
| South West | 0.76 (0.33–1.58) | 1.40 (0.77–2.46) | ||
| Socioeconomic disadvantage | ||||
| Tertile 1 (least disadvantaged) | 1 (reference) | 1 (reference) | ||
| Tertile 2 | 2.07 (1.51–2.76) | 1.95 (1.39–2.59) | ||
| Tertile 3 (most disadvantaged) | 2.73 (1.70–4.14) | 2.41 (1.35–4.49) | ||
| State-level factors | ||||
| Socioeconomic disadvantage | ||||
| Tertile 1 (least disadvantaged) | 1 (reference) | |||
| Tertile 2 | 2.41 (1.33–4.01) | |||
| Tertile 3 (most disadvantaged) | 3.80 (1.37–9.05) | |||
| Measures of variation | ||||
| State level | ||||
| Variance (SE) | 0.582 (0.322–0.989) | 0.551 (0.289–0.989) | 0.280 (0.125–0.523) | 0.192 (0.055–0.391) |
| Explained variation (%) | Reference | 5.28 | 51.8 | 67.0 |
| ICC (%) | 13.18 | 12.42 | 6.75 | 4.72 |
| MOR | 2.07 | 2.03 | 1.66 | 1.52 |
| Community level | ||||
| Variance (SE) | 0.539 (0.409–0.689) | 0.596 (0.434–0.782) | 0.581 (0.432–0.762) | 0.581 (0.430–0.768) |
| Explained variation (%) | Reference | −10.5 | −7.73 | −7.69 |
| ICC (%) | 25.40 | 25.84 | 20.74 | 19.01 |
| MOR | 2.01 | 2.09 | 2.07 | 2.07 |
| Model fit statistics | ||||
| Bayesian DIC | 7424 | 6525 | 6510 | 6514 |
SE standard error, DIC deviation information criterion, CrI credible interval, ICC intra-cluster correlation, MOR median odds ratio
aModel 1 is the empty model with no independent variables
bModel 2 is adjusted for age, education, household wealth index, knowledge about causes of malaria, exposure to malaria messages, knowledge about efficacy of mosquito nets, knowledge about efficacy of malaria prevention drugs, knowledge about importance of test to detect malaria, knowledge about efficacy of ACT and number of household members
cModel 3 is additionally adjusted for residence, region and community socioeconomic factors
dModel 4 is additionally adjusted for state socioeconomic factors