Factors associated with the use and quality of antenatal care in Nepal: a population-based study using the demographic and health survey data

This study used data from the 2011 Nepal Demographic and Health Survey (DHS), a nationally representative household cluster survey of 10,826 households which interviewed 12,674 women aged 15–49 years and 4,121 men aged 15–49 years and had a response rate of 99% [33]. We restricted our analyses to the 4,079 women who gave birth in the five years preceding the survey. Among women with two or more live births in the five years preceding the survey, we referred to the most recent birth only. Full details of the survey have been published elsewhere [34].

In the Nepal DHS, all women who gave birth between 2007 and 2011 were asked a number of questions about antenatal health care. Information was collected about the person and institution providing ANC (if any), the number of ANC visits, the timing of the first ANC visit and the components included in the ANC provided. These components were: blood pressure measurement; urine testing for the detection of bacteriuria and proteinuria; blood tests for syphilis and anaemia; and the provision of iron supplementation, intestinal parasite drugs, tetanus toxoid injections and health education during pregnancy. Questions on health education were related to the provision of information on danger signs during pregnancy, where to go in case of such complications, and recommendations to use a skilled birth attendant for delivery [34].

1. We used ‘attendance at four or more ANC visits’ as our first outcome variable based on the World Health Organization recommendations on ANC [35].

2. We defined good quality ANC as that which included all seven recommended items of ANC in Nepal. These items are: the provision of iron supplementation, blood tests, urine tests, at least two tetanus toxoid injections, measurement of blood pressure, provision of intestinal parasite drugs and health education regarding their pregnancy [23]. We defined health education as the receipt of any one of the three items of health education during ANC: a recommendation to use a skilled birth attendant; education on the warning signs that may suggest significant pregnancy complications; or advice on where to seek health care, should problems develop.

Sixteen variables were considered for their potential association with attendance at four or more ANC visits. These were: women’s age at birth of the child, women’s education (no education, primary, secondary and tertiary education), women’s work status in the past 12 months (whether in paid employment at the time of the survey or the previous12 months), the quintile of wealth of women’s household (the DHS wealth quintile is a composite indicator which divides the households into five categories: poorest, poorer, middle, richer and richest; and were derived using principle component analysis based on information from housing characteristics and ownership of household durable goods) [36], religion (Hindu, Buddhist, Muslim, Kirat and Christian), smoking status of women, decision-making power (whether women participated in at least one of the decisions regarding their own health care, major household purchases or visits to their family or relatives), general media exposure (whether exposed to either radio, television or newspapers/magazines at least once a week), parity, whether the pregnancy was wanted at the time (the survey had three categories: ‘pregnancy wanted at that time’, ‘pregnancy wanted later’ and ‘pregnancy not wanted at all’ of which the later two were merged into one to form the response ‘No’ and the first one remained as ‘Yes’) and history of previous pregnancies (including history of miscarriages, stillbirths or neonatal deaths and multiple pregnancies), use of modern family planning method (the survey data had four categories: ‘No method’, ‘Folkloric method’, ‘Traditional method’ and ‘Modern method’; the first three categories were merged to form the response ‘No’ and the fourth category formed the response ‘Yes’), ecological zones (Mountains, Hills or Terai), residence (urban, rural), husbands’ education (no education, primary, secondary, tertiary and don’t know) and their occupation (agricultural, professional/technical/managerial, clerical, services, skilled manual, unskilled manual and other).

Twelve variables were considered for their potential association with receipt of good quality ANC: women’s age at birth of the child, women’s education, wealth quintile, decision-making power of women, general media exposure, parity, history of previous pregnancies, use of modern family planning methods, ecological zones, residence, place of receipt of ANC and the type of health worker providing ANC. The variable ‘place of receipt of ANC’ specified the highest place where the women got ANC and originally had 16 categories which were reduced to 9. The two responses ‘respondent’s home’ and ‘Other’s home’ were merged to form one category ‘Home’. The four main Non-Government Organisations (NGOs) providing services plus the ‘other NGO’ were merged into one response ‘NGO’. The two responses ‘Others’ and “Other Government’ were merged into one category ‘Other’. The two responses ‘Private Hospital/Clinic’ and ‘Other Private’ were merged into one category ‘Private Hospital/Clinic’. ‘Type of health worker’ had 8 categories reduced to six (Doctor, Female Community Health Volunteer, Village Health Worker, Maternal and Child Health Worker, Health Assistant/Auxiliary Health Worker and Nurse/Midwife). The two additional categories in the survey: ‘Traditional Birth Attendant’ and ‘Other’ were excluded as there were only five responses to each.

All analyses used sampling weights and adjusted for the sampling design (clustering and stratification). We used linearization to obtain valid estimates of standard errors [37]. Firstly, we calculated descriptive statistics (frequencies or means and standard deviations as appropriate) and their 95% confidence intervals. We then calculated unadjusted odds ratios and 95% confidence intervals using logistic regression models to examine the association between the study factors and the receipt of: 1) four or more ANC visits; and 2) good quality ANC. For the variables mother’s age and parity, we checked their linearity using a lowess smoother [37]. The lowess smoother indicated a quadratic relationship between mother’s age and both outcomes, and so this variable was modelled using a quadratic term in both of the models. Because of the inclusion of a quadratic term, we centred mother’s age at its mean (25.4 years) to avoid problems with collinearity. We then fitted multivariable logistic regression models as follows:

We initially retained all variables that were significant at P < 0.25 in the univariable models. We then used a backwards elimination procedure, by removing the least significant variable in the model. This was repeated until all variables in the model had a P-value of < 0.05. For the model of quality of ANC, we retained the variable mother’s education in the final model as its significance was slightly above 0.05. In both models, we checked whether a quadratic term for mothers’ age was still required; in both cases, the model with a linear term only was adequate.

The analyses were undertaken in SAS 9.3 and Stata 12.1.

Half the women (n = 2078, 50.0%, 95% CI = 46.1 to 53.8%) had four or more ANC visits, whereas the other half had fewer than 4 ANC visits, including 15% of women who had no ANC at all. A total of 3,468 (3520 after weighting) women who had at least one ANC visit were included in the analysis for quality of ANC. A larger proportion of women attending four or more ANC visits received good quality ANC, compared to those who attended fewer than four ANC visits (84% vs 16%).

Three variables were highly significant (P < 0.001) for the association with four or more ANC visits in the multivariable model: women’s education, the wealth quintile to which their households belonged and their parity. With increasing education levels of women, their odds of receiving four or more ANC visits also increased, which was as high as seven times for women with tertiary education compared to those with no education (OR = 7.11; 95% CI: 3.28 to 15.44). Increasing wealth of women’s household also increased the odds of women getting four or more ANC visits, as women in the richest quintile had three times the odds of receiving four or more ANC visits than women in the poorest quintile (OR = 3.00; 95% CI: 1.95 to 4.60). Women of higher parity had lower odds of receiving four or more ANC visits. Women’s decision-making power (P = 0.005), and their husbands’ education (P = 0.010) and occupation (P = 0.002) were also significant in the model (Table 1). Women who participated in at least one household decision-making had higher odds of receiving four or more ANC compared to women who did not participate in any decision-making. The levels of husbands’ education increased the odds of women getting four or more ANC visits. Women whose husbands were involved in agriculture had much lower odds of having four or more visits compared to those involved in other occupations, including professional/technical/managerial jobs, clerical, skilled or unskilled manual jobs and services (Table 1).

Other variables significantly associated with four or more ANC visits in the same model were women’s age at birth of the child (P = 0.021) and their smoking status (P = 0.036). Non-smoking women had greater odds of attending four or more ANC visits, as did older women. However, variables including whether women were employed in the past 12 months, their religion, their general media exposure, whether the pregnancy was wanted, history of previous pregnancies, whether women resided in rural or urban area, which were all significant in the univariable model, lost their significance in the multivariable model (Table 1).

Among women who had at least one ANC visit, some components of ANC were more commonly received than others. Health education, iron supplementation, blood pressure measurement and two or more tetanus toxoid injections each were received by at least three quarters of the women while 64% were given intestinal parasite drugs, 56% had a urine sample taken and 45% had a blood sample taken (Table 2).

Two variables were highly associated (P <0.001) with receipt of good quality ANC: the health worker who provided the care and the woman’s parity (Table 2). The type of health worker providing ANC was a very strong predictor for the receipt of good quality ANC. Compared to women who received ANC from a doctor, those who received ANC from a Village Health Worker had 96% lower odds of receiving good quality ANC (OR = 0.04; 95% CI: 0.01 to 0.29); those being served by a Female Community Health Volunteer had 90% lower odds (OR = 0.10; 95% CI: 0.01 to 0.62), those receiving service from a Maternal and Child Health Worker had 84% lower odds (OR = 0.16; 95% CI: 0.07 to 0.41), those receiving service from a Health Assistant or an Auxiliary Health Worker had 75% lower odds (OR = 0.25; 95% CI: 0.37 to 0.46) and those being served by a nurse had 26% lower odds (OR = 0.74; 95% CI: 0.58 to 0.94) of getting good quality ANC. With each unit increase in parity, the odds of receiving good quality ANC decreased by 21% (OR = 0.79; 95% CI: 0.70 to 0.88; Table 2).

Women’s age at birth of the child (P = 0.041), their education (P = 0.057), wealth quintile to which their households belonged (P = 0.001), their residence (P = 0.013), the place where ANC was received (P = 0.005) and modern family planning use by the women (P = 0.045) were all significant in the model. Older women received good quality ANC compared to younger women. Women with a tertiary education had one-and-half times greater odds of receiving good quality ANC compared to those with no education (OR = 1.53; 95% CI: 0.97 to 2.41). Women in the richest quintile had three times the odds of receiving good quality ANC compared to women in the poorest quintile (OR = 2.88; 95% CI: 1.70 to 4.89). Women residing in urban areas had greater odds of receiving good quality ANC compared to those living in the rural areas. Women using modern family planning had decreased odds of receiving good quality ANC (Table 3).

The place where ANC was provided was associated with receipt of good quality ANC. Compared to those receiving ANC from a government hospital, those who visited a Non-Government Organisation had more than one-and-a-half times greater odds of receiving all seven ANC items (OR = 1.60; 95% CI: 1.02 to 2.52). However, women who received care at any other places [including Government Primary Health Care (PHC) outreach, Government Sub-Health Post, Government Health Post, Government PHC, private hospital/clinic and home] had lower odds of receiving good quality ANC compared to those who received ANC at a government hospital.

On the other hand, the decision-making power of women, the ecological zones where they lived and history of previous pregnancy lost their significance in the multivariable model (Table 3).

Women with higher levels of educational attainment were more likely to have four or more ANC visits. A study in Nepal on women’s empowerment and maternal health utilisation, analysing the 2001 DHS, also found that women's education was strongly associated with greater use of maternal health care [38]. Female education improves wealth, reduces gender disparity and empowers women [39]. Previous studies in South Asia suggest that education fosters new values and attitudes that are favourable to the use of modern health care, increases the chances of women desiring skilled care and empowers them to access such care [38]. Therefore, policies aimed at improving female education are likely to also improve ANC use. This is particularly important in Nepal, where only a very small proportion of Nepalese girls (6%) complete secondary school [35], only 18% of Nepalese women have a secondary education and the adult female literacy rate is 47% [40].

Educational levels of the husbands also showed a similar effect on the receipt of four or more ANC visits. This is similar to the results of a systematic review undertaken in low-income countries [41]. A randomised controlled trial on ANC in urban Nepal found that educating women and their husbands together resulted in women retaining more maternal health knowledge than if they attended alone, concluding that educational strategies become more effective when men are incorporated [42].

Women from higher income households had better odds of attending four or more ANC visits. This is similar to another study analysing data from 2006 Nepal DHS which found that women from higher wealth index household used ANC earlier during pregnancy and more frequently than those from lower wealth index households [29]. Women whose husbands were involved in agriculture had lower odds of four or more ANC visits than those whose husbands were involved in other occupations. In a study in rural Nepal, women whose husbands were involved in farming were less likely to have ANC than those who had any other jobs, linking farming with poverty and lack of disposable income [43].

The finding from this study that parous women have decreased odds of attending four or more ANC visits is consistent with findings from similar studies [41, 44, 45]. The second or later pregnancy carries lower risks for complications if the first pregnancy and birth were uncomplicated [33]. The experience of a previous pregnancy without complications might influence the perception of women about the need to seek care during pregnancy [46]. Some women might be occupied with the responsibilities of other children and hence find it difficult to attend ANC [41, 46].

In this study, non-smoking women had higher odds of four or more ANC visits. A similar conclusion was drawn in Brazil [45]. Previous studies have shown use of fewer health care services in general and ANC in particular among smokers [35]. Nonetheless, a previous study in Nepal depicted no association between women’s smoking status and use of ANC when controlled for other socio-demographic factors [29].

Women who participated in at least one household decision-making factor had higher odds of receiving four or more ANC visits compared to women who had no decision-making participation. This is consistent with the findings of a previous study in Nepal in which women who had a say in decision-making regarding large household purchases had significantly higher chances of receiving ANC than those who did not [38].

The type of health worker providing ANC plays a vital role in determining the quality of ANC received by women [19]. This study highlights the importance of skilled providers (doctors, nurses or midwives) in the provision of ANC. The odds of receiving good quality ANC were higher for women attending skilled providers compared with those who were relatively less skilled (Female Community Health Volunteers, Village Health Workers, Maternal and Child Health Worker and Health Assistant/Auxiliary Health Worker).

Women with higher levels of education were more likely to receive good quality care. A Ugandan study demonstrated the importance of women’s education for the receipt of all the recommended ANC components [31]. Women from higher income households had higher odds of receiving good quality ANC. This is consistent with studies from India and Nigeria which showed that higher socioeconomic status was associated with higher chances of using all ANC components [31, 46, 47]. More affordability and access to health information among women of higher socioeconomic status compared to women from lower socioeconomic status may explain these associations [29].

This study shows that the rurality of the place where women reside impacts on the quality of ANC. Women in urban areas had higher odds of receiving good quality ANC compared to those in the rural areas. This may be due to the fact that the rural areas have significantly less or even absent transportation infrastructure, making access to health care difficult [48].

Various ANC interventions can contribute to improving maternal and neonatal health. The majority of women in this study had their blood pressure measured, allowing for the detection of hypertension, which may represent an early sign of pre-eclampsia [35]. However, only just over half of the women had their urine tested for protein and infection, and even fewer (45%) had their blood taken to test for anaemia. These findings are comparable to a previous study which related such low utilisation with the lack of capacity of the health centres to carry out such tests [49]. Ninety-one per cent of women received iron supplements. The World Health Organization recommends providing iron supplementation to all pregnant women in developing countries at high risk of iron deficiency [13]. The proportion of women receiving tetanus toxoid injections, intestinal parasite drugs, and health education during pregnancy are similar to those from a Zambian study that used the 2007 DHS, but much higher than those from a Ugandan study [19, 31].