Table 1 presents the socio-demographic profile of the respondents. More than two-fifths of the respondents were from households that had no viable means of transportation. However, among the respondents who reported household ownership of different transport modes, motorcycle was the dominant transport mode. Nearly half of the respondents had no formal education. Secondary education was the dominant educational level among the respondents who had formal education. More than one-fifth of the respondents belong to households either at the poorest or poorer wealth quintiles. However, the distribution of the respondents at other household wealth quintiles was nearly equal among the respondents. Slightly more than half of the respondents were high parous women. Nearly half of the respondents had four or more antenatal care visits. However, more than one-third of the respondents had no antenatal care visits during their most recent pregnancy. Less than one-fifth of the respondents had their first antenatal visits within the first trimester of the pregnancy, while nearly half of the respondents had their first antenatal visits during the second trimester of the pregnancy. Also, more than one-third of the respondents had the first antenatal visits in the last trimester of the pregnancy.
Table 1
Respondents’ Socio-demographic characteristics
Characteristic | Number of Women (n = 38,102) | Percentage | Characteristic | Number of Women (n = 38,102) | Percentage |
Means of transportation | Healthcare decision |
None | 16,838 | 44.2 | Male dominated | 24,081 | 63.2 |
Canoe | 296 | 0.8 | Not male dominated | 14,021 | 36.8 |
Bicycle | 5,237 | 13.7 | Perception of distance to health facility |
Motorcycle | 12,558 | 33.0 | Big problem | 13,422 | 35.2 |
Car | 3,173 | 8.3 | Not a problem | 24,680 | 64.8 |
Maternal education | Proportion who owns no means of transport in community |
None | 17,811 | 46.8 | Low | 12,807 | 33.6 |
Primary | 7,927 | 20.8 | Moderate | 12,532 | 32.9 |
Secondary | 10,031 | 26.3 | High | 12,763 | 33.5 |
Higher | 2,333 | 6.1 | Proportion who owns bicycle in community |
Household wealth quintile | Low | 13,287 | 34.9 |
Poorest | 8,773 | 23.0 | Moderate | 12,455 | 32.7 |
Poorer | 8,504 | 22.3 | High | 12,360 | 32.4 |
Middle | 7,252 | 19.0 | Proportion who owns motorcycle in community |
Richer | 6,878 | 18.1 | Low | 13,107 | 34.4 |
Richest | 6,695 | 17.6 | Moderate | 12,040 | 31.6 |
Birth order | High | 12,955 | 34.0 |
1 | 6,723 | 17.7 | Proportion who owns car in community |
2 | 6,287 | 16.5 | Low | 12,441 | 32.7 |
3 | 5,759 | 15.1 | Moderate | 12,423 | 32.6 |
4 or higher | 19,333 | 50.7 | High | 13,238 | 34.7 |
Number of antenatal visit | Place of residence |
No ANC visit | 15,567 | 40.9 | Urban | 12,608 | 33.1 |
1–3 visits | 4,174 | 10.9 | Rural | 25,494 | 66.9 |
4 or more visits | 18,361 | 48.2 | Geographic region |
Timing of first antenatal visits | Northern region | 24,417 | 64.1 |
First trimester | 6,475 | 17.0 | Southern region | 13,685 | 35.9 |
Second trimester | 17,231 | 45.2 | | | |
Third trimester | 14,396 | 37.8 | | | |
Healthcare decision of the majority of respondents was male dominated. The majority of the respondents did not perceived distance to health facility as a problem. Also, a majority reside in the rural areas of the country, and had a preponderance of women from the northern region. Slightly more than one-third of the respondents live in communities with either low or high proportions of households that had no viable means of transportation. Distribution of the respondents by proportions of households who had bicycle in the community revealed a nearly equal proportions of low, moderate and high proportion of households that had bicycle. This pattern was replicated with respect to distribution by proportions of households that had a motorcycle or a car in the community. Figure 1 presents the prevalence of institutional and non-institutional delivery among the respondents with dominance of non-institutional delivery.
Table 2 presents the independent association between place of delivery and specific explanatory and control variable. Women from household with canoes and bicycle as transport modes were less likely to have institutional delivery. However, as household means of transportation improved from motorcycle to car, the likelihood of institutional delivery increased among the women. Women from households with car as transport mode were three times more likely to have institutional delivery. Education, household wealth quintile and antenatal care visits were positively associated with place of delivery. For instance, as educational attainment improves, the likelihood of institutional delivery increased significantly. Likewise, as household wealth quintile improves, the odds of institutional delivery also increased. In contrast, birth order and timing of first antenatal care visits and place of delivery were negatively associated with institutional delivery. For instance, as birth order increased, the odds of institutional delivery reduced among the women, and as timing of first antenatal care visits increased, the odds of institutional delivery reduced significantly.
Table 2
Unadjusted odds ratio showing association between the explanatory variables and place of child delivery
Variable | AOR | p-value | 95% CI | Variable | AOR | p-value | 95% CI |
Means of transportation | Healthcare decision |
None RC | 1.000 | - | - | Male dominated RC | 1.000 | - | - |
Canoe | 0.344 | p < 0.01 | 0.268–0.442 | Not male dominated | 3.114 | p < 0.01 | 2.981–3.254 |
Bicycle | 0.405 | p < 0.01 | 0.375–0.438 | Perception of distance to health facility |
Motorcycle | 1.068 | p = 0.01 | 1.018–1.121 | Big problem RC | 1.000 | - | - |
Car | 3.717 | p < 0.01 | 3.428–4.030 | Not a problem | 2.580 | p < 0.01 | 2.461–2.705 |
Maternal education | Proportion who owns no means of transport in community |
None RC | 1.000 | - | - | Low RC | 1.000 | - | - |
Primary | 1.150 | 0.295 | 0.885–1.494 | Moderate | 0.907 | p < 0.05 | 0.832–0.989 |
Secondary | 1.369 | p < 0.01 | 1.229–1.525 | High | 0.799 | p < 0.01 | 0.740–0.862 |
Higher | 1.530 | p < 0.01 | 1.305–1.793 | Proportion who owns bicycle in community |
Household wealth quintile | Low RC | 1.000 | - | - |
Poorest RC | 1.000 | - | - | Moderate | 0.510 | p < 0.01 | 0.484–0.536 |
Poorer | 1.300 | p < 0.01 | 1.222–1.382 | High | 0.310 | p < 0.01 | 0.294–0.327 |
Middle | 1.560 | p < 0.01 | 1.406–1.732 | Proportion who owns motorcycle in community |
Richer | 1.691 | p < 0.01 | 1.544–1.852 | Low RC | 1.000 | - | - |
Richest | 2.762 | p < 0.01 | 2.510–3.040 | Moderate | 0.770 | p < 0.01 | 0.731–0.809 |
Birth order | High | 0.763 | p < 0.01 | 0.724–0.802 |
1 RC | 1.000 | - | - | Proportion who owns car in community |
2 | 0.805 | p < 0.01 | 0.751–0.863 | Low RC | 1.000 | - | - |
3 | 0.746 | p < 0.01 | 0.695–0.802 | Moderate | 1.572 | p < 0.01 | 1.487–1.661 |
4 or higher | 0.475 | p < 0.01 | 0.449–0.503 | High | 3.363 | p < 0.01 | 3.188–3.549 |
Number of antenatal visits | Place of residence |
No ANC visit RC | 1.000 | - | - | Urban RC | 1.000 | - | - |
1–3 visits | 3.410 | p < 0.01 | 3.136–3.707 | Rural | 0.196 | p < 0.01 | 0.187–0.205 |
4 or more visits | 12.921 | p < 0.01 | 12.171–13.717 | Geographic region |
Timing of first antenatal visit | Northern region RC | 1.000 | - | - |
First trimester RC | 1.000 | - | - | Southern region | 6.037 | p < 0.01 | 5.762–6.326 |
Second trimester | 0.762 | p < 0.01 | 0.719–0.807 | | | | |
Third trimester | 0.039 | p < 0.01 | 0.036–0.043 | | | | |
Healthcare decision and perception of distance to health facility relates positively with place of delivery with higher likelihood of institutional delivery among women whose healthcare decision were not male dominated and who do not perceived distance to facility as a problem. The odds of institutional delivery were lower among rural women and higher among southern women. Women who live in communities with either moderate or high proportions of households with no viable means of transportation, and with bicycles or motorcycles as transport modes were less likely to have institutional delivery. In contrast, women who live in communities with either moderate or high proportion of households with cars as transport mode were more likely to have institutional delivery. Following the bivariate analyses, a VIF test was performed to confirm absence of serious multicollinearity among the explanatory variables. A mean score of 2.79 was obtained indicating that the explanatory variables do not have serious collinearity that could mislead the estimation of the regression coefficients at the multivariate level.
Table 3 presents results of the fitted adjusted models. In Model 1, while the odds of institutional delivery reduced by 72.7% and 44.7% among women from households with canoes (AOR = 0.273, p < 0.01; CI: 0.198–0.378) and bicycles (AOR = 0.553, p < 0.01, CI: 0.499–0.613) as mode of transport respectively, the odds increased by 37.0% among women from households with motorcycles as means of transportation (AOR = 1.370, p < 0.01; CI: 1.282–1.465). Likewise, women from households with cars as means of transportation were four times more likely to have institutional delivery compared to women from households that had no viable means of transportation (AOR = 4.037, p < 0.01; CI: 3.585–4.546). At the community level, the likelihood of institutional delivery were significantly lower among women who live in communities where the majority of households either had no viable means of transportation or had the dominance of bicycles or motorcycles. In contrast, women who live in communities with high proportion of households with cars as the transport mode were nearly four times more likely to have institutional delivery compared to women in the reference category (AOR = 3.644, p < 0.01; CI: 2.797–4.746).
Table 3
adjusted odds ratio showing further association between the explanatory variables and place of child delivery
Characteristic | Model 1 | Model 2 | Model 3 |
AOR | p-value | 95% CI | AOR | p-value | 95% CI | AOR | p-value | 95% CI |
Means of transportation |
None RC | 1.000 | - | - | 1.000 | - | - | 1.000 | - | - |
Canoe | 0.273 | p < 0.01 | 0.198–0.378 | 0.230 | p < 0.01 | 0.164–0.324 | 0.317 | p < 0.01 | 0.221–0.456 |
Bicycle | 0.553 | p < 0.01 | 0.499–0.613 | 0.867 | p < 0.05 | 0.774–0.969 | 0.884 | p < 0.05 | 0.785–0.995 |
Motorcycle | 1.370 | p < 0.01 | 1.282–1.465 | 1.114 | p = 0.05 | 1.033–1.201 | 1.292 | p < 0.01 | 1.143–1.460 |
Car | 4.037 | p < 0.01 | 3.585–4.546 | 1.225 | p < 0.05 | 1.077–1.395 | 2.044 | p < 0.01 | 1.781–2.345 |
Proportion who owns no means of transport in community |
Low RC | 1.000 | - | - | 1.000 | - | - | 1.000 | - | - |
Moderate | 0.852 | p = 0.01 | 0.777–0.934 | 0.130 | p < 0.05 | 0.006–0.254 | 0.861 | p < 0.05 | 0.784–0.946 |
High | 0.318 | p < 0.01 | 0.237–0.426 | 0.401 | p < 0.01 | 0.291–0.511 | 0.872 | p = 0.01 | 0.788–0.967 |
Proportion who owns bicycle in community |
Low RC | 1.000 | - | - | 1.000 | - | - | 1.000 | - | - |
Moderate | 0.537 | p < 0.01 | 0.421–0.685 | 0.859 | p < 0.05 | 0.748–0.986 | 0.962 | 0.649 | 0.815–1.135 |
High | 0.352 | p < 0.01 | 0.261–0.476 | 0.889 | 0.073 | 0.782–1.011 | 0.858 | 0.145 | 0.698–1.054 |
Proportion who owns motorcycle in community |
Low RC | 1.000 | - | - | 1.000 | - | - | 1.000 | - | - |
Moderate | 0.874 | 0.285 | 0.683–1.119 | 1.215 | p < 0.05 | 1.024–1.441 | 1.111 | 0.219 | 0.939–1.314 |
High | 0.821 | 0.203 | 0.605–1.113 | 1.507 | p < 0.01 | 1.220–1.862 | 1.319 | p < 0.05 | 1.071–1.625 |
Proportion who owns car in community |
Low RC | 1.000 | - | - | 1.000 | - | - | 1.000 | - | - |
Moderate | 1.809 | p < 0.01 | 1.413–2.315 | 1.216 | p < 0.05 | 1.013–1.459 | 1.448 | p < 0.01 | 1.264–1.658 |
High | 3.644 | p < 0.01 | 2.797–4.746 | 1.915 | p < 0.01 | 1.748–2.083 | 3.146 | p < 0.01 | 2.621–3.777 |
Place of residence |
Urban RC | | | | 1.000 | - | - | 1.000 | - | - |
Rural | | | | 0.550 | p < 0.01 | 0.499–0.607 | 0.665 | p < 0.01 | 0.601–0.736 |
Geographic region |
Northern region RC | | | | 1.000 | - | - | 1.000 | - | - |
Southern region | | | | 2.486 | p < 0.01 | 2.140–2.888 | 2.092 | p < 0.01 | 1.804–2.425 |
Maternal education |
None RC | | | | 1.000 | - | - | 1.000 | - | - |
Primary | | | | 1.768 | p < 0.01 | 1.637–1.909 | 1.930 | p < 0.01 | 1.753–2.125 |
Secondary | | | | 2.547 | p < 0.01 | 2.347–2.763 | 2.980 | p < 0.01 | 2.673–3.321 |
Higher | | | | 7.343 | p < 0.01 | 6.220–8.670 | 9.402 | p < 0.01 | 7.650-11.554 |
Household wealth quintile |
Poorest RC | | | | 1.000 | - | - | 1.000 | - | - |
Poorer | | | | 1.489 | p < 0.01 | 1.339–1.657 | 1.587 | p < 0.01 | 1.397–1.803 |
Middle | | | | 1.974 | p < 0.01 | 1.772–2.199 | 2.277 | p < 0.01 | 1.987–2.610 |
Richer | | | | 2.584 | p < 0.01 | 2.297–2.907 | 3.184 | p < 0.01 | 2.729–3.715 |
Richest | | | | 4.004 | p < 0.01 | 3.478–4.608 | 5.596 | p < 0.01 | 4.638–6.751 |
Birth order |
1 RC | | | | 1.000 | - | - | 1.000 | - | - |
2 | | | | 0.740 | p < 0.01 | 0.666–0.823 | 0.705 | p < 0.01 | 0.630–0.788 |
3 | | | | 0.732 | p < 0.01 | 0.656–0.816 | 0.687 | p < 0.01 | 0.612–0.770 |
4 or higher | | | | 0.700 | p < 0.01 | 0.641–0.765 | 0.645 | p < 0.01 | 0.587–0.708 |
Number of antenatal visits |
No ANC visit RC | | | | | | | 1.000 | - | - |
1–3 visits | | | | | | | 0.950 | 0.504 | 0.817–1.105 |
4 or more visits | | | | | | | 1.735 | p < 0.01 | 1.515–1.986 |
Timing of first antenatal visits |
First trimester RC | | | | | | | 1.000 | - | - |
Second trimester | | | | | | | 0.864 | p = 0.01 | 0.796–0.939 |
Third trimester | | | | | | | 0.129 | p < 0.01 | 0.109–0.152 |
Healthcare decision |
Male dominated RC | | | | | | | 1.000 | - | - |
Not male dominated | | | | | | | 1.438 | p < 0.01 | 1.338–1.546 |
Perception of distance to health facility |
Big problem RC | | | | | | | 1.000 | - | - |
Not a problem | | | | | | | 1.198 | p < 0.01 | 1.109–1.295 |
In Model 2, the addition of the selected background characteristics slightly altered the pattern of association between means of transportation and place of delivery, though means of transportation remained significantly associated with place of delivery at both household and community levels. For instance, the odds of institutional delivery among women from households that had motorcycle as the transport mode reduced from 37% in Model 1 to 11.4% in Model 2. Also, the odds of institutional delivery among women from households that had cars as the transport mode reduced from more than four times likelihood in Model 1 to 22.5% in Model 2 indicating that the background characteristics played strong mediating role in the likelihood of institutional delivery. Additionally, the odds of institutional delivery increased significantly as education and household wealth quintiles improved, while the odds reduced significantly as birth order increased. Furthermore, the likelihood of institutional delivery reduced significantly for rural women (AOR = 0.550, p < 0.01; CI: 0.499–0.607), whereas southern women were more than twice more likely to have institutional delivery compared to northern women (AOR = 2.486, p < 0.01; CI: 2.140–2.888).
In Model 3, there were noticeable changes in the association between means of transportation and place of delivery. The odds of institutional delivery remained lower among women from households that had canoes (AOR = 0.317, p < 0.01; CI: 0.221–0.456) and bicycles (AOR = 0.884, p < 0.05; CI: 0.785–0.995) as the transport modes. Though, the odds were also significantly higher among women from households with motorcycle as the means of transportation, the odds of institutional delivery were twice more likely among women from households that had cars compared to women from households that had no viable means of transportation (AOR = 2.044, p < 0.01; CI: 1.781–2.345). At the community level, women who live in communities with moderate or high proportions of households with no viable means of transportation were less likely to have institutional delivery. Though, women who live in communities with either moderate or high proportions of households who owns bicycle were also less likely to have institutional delivery, but these were without statistical significance. Women who live in communities with high proportions of households who owned motorcycle were 31.9% more likely to have institutional delivery compared to women in communities with low proportion (AOR = 1.319, p < 0.05; CI: 1.071–1.625). Among women who live in communities with high proportion of households who had cars as mode of transport, the odds of institutional delivery were three times more likely (AOR = 3.146, p < 0.01; CI: 2.621–3.777). Rural women were 33.5% less likely to have institutional delivery compared to urban women (AOR = 0.665, p < 0.01; CI: 0.601–0.736), while southern women were twice more likely to have institutional delivery compared to northern women (AOR = 2.092, p < 0.01; CI: 1.804–2.425).
Maternal education and household wealth maintained significant associations with institutional delivery with increasing odds of institutional delivery as the characteristics improved progressively. However, the likelihood of institutional delivery reduced significantly as birth orders increased. For instance, women who had the fourth or higher births were 35.5% less likely to have institutional delivery compared to women who had the first birth (AOR = 0.645, p < 0.01; CI: 0.587–0.708). Women who had four or more antenatal visits were 73.5% more likely to have institutional delivery compared to women who had no antenatal visit (AOR = 1.735, p < 0.01; CI: 1.515–1.986) but in contrast, women who had first antenatal visits during the second or third trimester of their most recent pregnancy were significantly less likely to have institutional delivery. Women whose healthcare decisions were not male dominated were 43.8% more likely to have institutional delivery compared to women whose healthcare decisions were male dominated (AOR = 1.438, p < 0.01; CI: 1.338–1.546), and women who did not perceived distance to health facility as a problem were more 19.8% more likely to have institutional delivery (AOR = 1.198, p < 0.01; CI: 1.109–1.295).
Table 4 presents the importance of means of transportation at the community level. In the empty model, there were significant variation in the prevalence of institutional delivery among women in the absence of means of transportation (ICC = 0.451). With the inclusion of means of transportation only in the model, the ICC of 0.368 indicate that households ownership of specific means of transportation at the community level accounts for 36.8% of the variation in institutional delivery among the women. This proportion however declined to 18.3% in model 2 with the inclusion of the background characteristics. The variation in institutional delivery attributable to means of transportation at the community level further declined to 17.0% in the full model.
Table 4
Multilevel logistic regression model showing effects of means of transportation at the community level
Parameter | Empty Model | Model 1 | Model 2 | Model 3 |
Community-level variance (S.E.) | 2.674 (0.165) | 1.919 (0.122) | 0.735 (0.055) | 0.676 (0.053) |
ICC | 0.451 | 0.368 | 0.183 | 0.170 |
Log-likelihood | -20273.5 | -19571.9 | -16234.5 | -14679.7 |
LR test | χ2 = 247.4; p < 0.001 | χ2 = 141.4; p < 0.001 | χ2 = 53.2; p < 0.001 | χ2 = 24.9; p < 0.001 |