Factors associated with health facility delivery among reproductive age women in Nepal: an analysis of Nepal multiple indicator cluster survey 2019

The data for this study were extracted from an individual woman and household record of the Nepal Multiple Indicator Cluster Survey (MICS) 2019 datasets. This dataset is publicly available for researchers. Permission to access and use this dataset is obtained from UNICEF/MICS website (http://​mics.​unicef.​org/​surveys). The Nepal Multiple Indicator Cluster Survey is a nationally representative cross-sectional survey carried out in 2019 by the Central Bureau of Statistics with technical and financial support from the United Nations Children's Fund (UNICEF). This survey is a part of the Global MICS program. The objective of MICS was to monitor the health status of women and children by various indicators. The survey collected information on socio-economic and demographic characteristics of households and household members, reproductive and maternal health, child mortality, child health, nutrition, literacy and education, child protection, attitudes towards domestic violence, knowledge of HIV/AIDS, access to media, and use of information and communication technology. Hence, the Nepal MICS 2019 provides internationally comparable data on a wide range of indicators about the situation of children and women. In this context, the findings of the study could help concerned authorities to promote maternal health services to achieve SDG 3.

The Nepal MICS employed a two-stage, stratified cluster sampling approach. In the first stage, sampling strata were identified, and then census enumeration areas (EAs) were selected from each sampling strata. In the second stage, a sample of households was selected from each selected area using a listing of households. A detailed description of the sample design is available in the Nepal MICS report [31].

In Nepal MICS, 12,800 households were selected for the sample, of which 12,655 households were interviewed. Likewise, 15,019 women aged 15–49 years were identified as eligible for interviews. However, 14,805 women aged 15–49 years were successfully interviewed, yielding a 98.6 percent response rate. For this study, 1950 married women aged 15–49 years who had at least one live birth in the last two years preceding the survey were included in the study population.

The dependent variable of this study is the place of delivery for the most recent birth in the last two years before the survey. This variable is divided into two categories. Women who delivered in public and private health facilities with skilled providers are coded 1 (or 'health facility delivery with skilled providers'), and those who delivered at home or in a health facility without skilled providers are coded 0 (or 'home/elsewhere). Based on the Nepal MICS, doctor, nurse or midwife and auxiliary nurse midwife are considered as skilled providers [31].

The independent variables included in this study are: age of women, place of residence, province, women's education, parity, mother's age at most recent birth, caste/ethnicity, access to media, internet use, health insurance, migration status, household wealth quintiles, and ANC visits. The independent variables were selected based on their significance in the previous studies, empirical findings and theoretical explanations from previous studies, and variable listed in Nepal MICS report. As presented in Table 1, age of women was categorized into three groups (15–19, 20–29, and 30–49 years), women's education was broadly categorized as no education, basic (grade1-8), secondary (grade 9–12), and higher), age of mother at most recent live birth was also categorized into three groups (< 20, 20–34, and 35–49 years), caste/ethnicity was grouped into five categories— Brahman/Chhetri, other Terai caste, Dalit, Janajati and Muslim. Access to media was constructed from questions on whether a woman reads newspapers, listens to the radio, or watches television. If a woman has exposure to at least one of the three media, she is coded as 1 for exposure to media and if she has no exposure to media to any of three, she is coded as 0 for not exposed. The variable, internet use, is constructed from a question on the ever-used internet. The variable, migration status, was constructed from questions such as how long have you been continuously living in the current place and did you live in an urban or a rural area just before you move here. This variable is categorized into four groups (urban migrants, rural migrants, urban non-migrants, and rural non-migrants. The household wealth index is a composite measure of household wealth status. The household wealth index is constructed using data on dwelling characteristics, ownership of consumer goods, household assets, and housing characteristics, and ANC visits were categorized into three groups (no, 1–3 visits, and 4 + visits).

Data were analyzed using STATA version 15.1. Univariate analysis is carried out to analyze selected socio-demographic characteristics of women aged 15–49 years. Chi-square tests were used to check the factors that were associated with outcome variables. Bivariate logistic regression was also carried out to examine the crude association between dependent and independent variables. The backward elimination method is used to select specified variables for the logistic model based on the Bayesian Information Criterion (BIC). Based on the BIC criteria, the final logistic model included the age of women, residence, province, women's education, parity, internet use, wealth index quintile, and ANC visits. Multi-collinearity assessment is performed before the multivariate analysis. A pairwise correlation was used to examine the multicollinearity among the independent variables. After assessing multicollinearity, it was found that none of the variables were highly correlated. The results of multivariate analysis are shown in odds ratios (OR). The critical level was set at a 95% confidence interval (CI). Sample weights were applied in all analyses. The complex survey design of MICS (primary sampling units and strata) was taken in logistic regression analysis.

Table 2 shows the background characteristics of women aged 15–49 years. The majority of the women (68%) are aged 20–29 years. About two third of women lived in an urban area while one-fifth of women were in Madhesh. About 40% of women had attained secondary while 31% received a basic education. Forty-four percent of women belonged to the first parity. About 78% of women had their recent birth at the age 20–34 years. More than one-third of women are Janajati. Sixty-two percent of women reported having access to media and 44% of women use the internet. Ninety-five percent of women have no health insurance. Seventy-one percent of women are rural migrants. More than two-thirds of women fall below the middle household wealth quintile. Seventy-eight percent of women attended at least four ANC visits for their most recent birth.

Table 3 reveals that three-fourths (75%) of women in Nepal delivered their most recent birth at a health facility with skilled providers . Delivery in a health facility with a skilled provider was higher among women aged 15–19 years (78%) and women residing in urban areas (82%). More than 80% of women of Gandaki, Bagmati, and Sudurpaschim provinces are delivered in a health facility with the assistance of a skilled provider. Similarly, the proportion of health facility delivery with a skilled provider was higher among women with secondary and higher education, who had a lower parity, whose age at most recent birth was less than 20 years, who belonged to the Brahman/Chhetri, who were exposed to media, who used the internet, who had health insurance, who were urban migrants, with highest wealth quintile, and who attended ANC at least four visits. The percentage of delivery in a health facility with skilled providers increased with the education level of women and the household wealth quintiles, and with the number of ANC visits but decreased with increasing parity.

The results of the bivariate analysis showed that place of residence, province, women's education, parity, caste/ethnicity, exposure to media, internet use, health insurance, migration status, household wealth quintile, and ANC visits were significantly associated with place of delivery at 95% confidence level (p < 0.001). Age of women and age at most recent birth was not found to be significantly associated with place of delivery.

The results of the logistic regression analysis of health facility delivery with the skilled provider are presented in Table 4. Place of residence, province, women's education, parity, internet use, wealth quintiles, and ANC visits were significantly associated with health facility delivery with skilled providers. After adjusting the effects of other variables, all variables that were significantly associated in the unadjusted model remained significantly associated with health facility delivery with skilled providers. The adjusted odds ratio indicated that women living in urban areas were 1.72 times [AOR: 1.74, 95% CI: 1.24–2.43] more likely to deliver at a health facility with skilled providers than women living in rural areas. The odds of delivering in a health facility with skilled providers were 5.64 [AOR:5.64, 95% CI: 3.11–10.22] times higher among women from Sudurpaschim, 3.06 times [AOR: 3.06, 95% CI: 1.65–45.66] higher among women form Karnali, 2.32 times [AOR: 2.32, 95% 1.35–3.98] higher among women form Province 1, 2.18 times [AOR: 2.18, 95% CI: 1.25–3.80] higher among women from Gandaki and 2.07 times [AOR: 2.07, 95% CI: 1.31–3.27] higher among women from Lumbini Province compared with women from Province 2. Regarding women's education, the odds of receiving at a health facility delivery by skilled providers were 3.42 times [AOR:3.42, 95% CI:1.15–10.15] higher for women have higher education than no education. Parity was also significantly associated with health facility delivery with skilled providers in adjusted odds ratio. The odds of delivering at a health facility with skilled providers were 2.82 times [AOR:2.82, 95% CI:1.652–4.82] higher for women of one parity than women of four or more parities. Women who used the internet had significantly higher odds of delivery in a health facility by skilled provider [AOR: 1.49, 95% CI: 1.07–2.08] than women who did not use the internet. Women in the poor household status [AOR: 2.67, 95% CI: 1.69–4.22], middle household status [AOR:4.78, 95% CI: 2.89–7.90], rich household status [AOR: 5.50, 95% CI: 3.16–9.56] and richest household status [AOR: 4.60, 95% CI: 2.20–9.761] have significantly higher odds of delivery in a health facility with skilled providers than women in poorest household status. Women who attended 1–3 ANC visits and at least four ANC visits during pregnancy have 4.08 times [AOR: 1.08, 95% CI: 2.10–7.91] and 10.81 times [AOR: 10.81, 95% CI: 5.75–20.31] higher odds of delivery in a health facility with skilled providers than women who do not attend ANC visits.