Patterns and determinants of effective coverage of routine maternal and newborn health visits in Nepal: Analysis of the 2016 Demographic and Health Survey

Antenatal care (ANC) visits, institutional delivery, and postnatal care (PNC) visits are 35 vital for improved health of mothers and newborns. Access of these routine maternal 36 and newborn health (MNH) visits have increased in the last few decades in Nepal; 37 however, little is known on the effective uptake (including timely, skilled, frequent, 38 and adequate care) of essential MNH interventions during those visits. This study 39 examined the patterns of effective coverage (EC) of routine MNH visits and their 40 determinants in Nepal. A secondary analysis was conducted taking data from the Nepal Demographic and 43 Health Survey (NDHS) 2016. The study included 1,978 women aged 15–49 years 44 who had a live birth in the two years preceding the survey. Three outcome variables 45 were EC of i) at least 4ANC visits, ii) institutional delivery, and iii) first PNC visit for newborns and mothers within 48 hours of childbirth. The independent variables 47 included several structural, intermediary and health system factors. Binomial logistic 48 regression analysis was conducted, and the magnitude of EC was reported as odds 49 ratio (OR) with 95% confidence intervals (CIs). The statistical significance level was 50 set at p<0.05 (two-tailed).

Despite increased access to routine maternal and newborn (MNH) visits in many 91 LMICs [11,12], the reduction of the Maternal Mortality Ratio (MMR) and the 92 Neonatal Mortality Rate (NMR) is slow [13]. Poor uptake of essential interventions 93 during routine MNH visits have contributed to the stagnant or slow rate of reduction 94 of NMR and MMR [3]. Only reaching at HFs, however, does not guarantee that 95 women receive all the recommended interventions [14]. Programs and research 96 need to also examine the actual receipt of essential MNH interventions and patterns 97 and determinants of EC of health services including MNH services [15,16]. 98

Measuring EC requires information about what happens at HF after visits, and can 99
provide insights into the quality of health services [17,18]. Estimation of EC 100 incorporates the population in need of health services, the proportion of HFs contact 101 for services, and the composite coverage of uptake of essential MNH interventions at 102 the HFs and reflects the performance of health systems [2,19]. The EC of health 103 services, therefore, provides a proxy measure of the quality of health services [20, 104 21] and is being given a high priority in policies and programs, especially in 105 reproductive, maternal and child health programs [8,22,23]. 106 Nepal has a high MMR (259 per 100,000 live births) and NMR (21 per 1000 live 107 births) within South Asia [24]. Further, while significant progress has been made, the 108 rate of reduction has slowed substantially compared to the rates of increases in MNH 109 visits over the past two decades in Nepal. For instance, from 2006 to 2016, 110 institutional delivery increased from 18% to 59%, while MMR decreased from 281 to 111 259 in the same period [25]. Evidence suggests that high death rates among 112 disadvantaged groups who usually have low coverage of routine MNH visits [25]. 113 The slow progress in women and newborns' survival can be explained by either 114 disadvantaged women having poor access to recommended MNH interventions or 115 the health system is inefficient in providing essential MNH interventions. 116 In the current health policy document, Nepal has prioritised quality of care as the 117 health system's overarching principle, within the Nepal health sector strategy (NHSS) 118 (2015-2020) [26], the national health policy 2019 [27] and the Nepal Safe 119 Motherhood and Newborn Health Roadmap 2030 [28] emphasizing the universal 120 quality of health services for achieving Sustainable Development Goal 3(SDG3), 121 including MNH services [29]. The current system, however, lacks a healthcare quality 122 monitoring mechanism. Neither the routine health information system such as health 123 management information system (HMIS) [30] nor do any strategic documents 124 describe the measurement of EC of health services, including for MNH [31,32]. 125 While studies have examined contact coverage of routine MNH visits [33,34] and 126 perceived quality of care [35,36], little is known evidence is published in the EC of 127 routine MNH visits [37]. This study examined about the patterns and determinants of 128 EC of routine MNH visits. This study begins to address this gap findings and may 129 inform programs and policies to target women groups with a high burden of maternal 130 and newborn mortalities but with poor access to quality MNH services in Nepal and 131 similar settings. 132

133
Data sources and sampling design 134 The NDHS 2016 data were used for this study, extracted data of individual women 135 record from the NDHS. The NDHS 2016 is a nationally representative cross-136 sectional survey conducted to assess the health services performance, especially 137 health status of reproductive, newborn, child health and nutrition [25]. 138 A more detailed sampling method is described in the NDHS 2016 report [25]. In brief, 139 the NDHS 2016 identified rural and urban sampling strata from seven provinces 140 producing 14 strata. The survey adopted a two-stage sampling design. who had a live birth in the two years preceding the survey as this sample included 149 information on pregnancy, childbirth, and postnatal care interventions. 150

Conceptual framework of the study 151
Based on the review of previous conceptual frameworks [38][39][40], a conceptual 152 framework was developed for this study (Figure 1). This conceptual framework 153 comprises inputs that include several determinants, their contexts and mechanisms 154 at system, institutional and individual level. These contexts and mechanisms can 155 produce the output of effective uptake of MNH interventions. Improved output as EC 156 of MNH visits may lead to the survival of mothers and newborns, economic benefit, 157 and confidence in the system. Inputs are broadly categorised into three domains: 158 structural, intermediary, and health system. Structural factors cover all basic factors 159 (e.g., wealth status, ethnicity, gender) that can influence the intermediary and health 160 system factors. Intermediary factors are non-health sector factors, which generally 161 affect the conditions of health via influencing the family/community contexts (living 162 and working conditions) and individual characteristics. Health system variables 163 include several variables that involve in the provision and delivery of quality health 164 services. on Social Determinants of Health [41]. 177 178

Study variables 179
Independent variables included characteristics of women and their healthcare 180 experience (Supplementary Table S1). Based on available information in the NDHS 181 2016 data, and as guided by the conceptual framework ( Figure 1). MNH visit [2]. Based on these woman-specific coverage scores of each MNH visit, 227 population-level EC was calculated using formula EC=Q×U|N [46]. In this formula, 228 EC refers to effective coverage of MNH visit at the national level, Q relates to the 229 average quality score of all interventions of particular MNH visit (e.g.4ANC visits), 230 and U refers to the utilisation of MNH visit (the contact coverage of the MNH visit), 231 and N is the number of population of need [2,23]. 232

233
For the regression analysis to identify determinants of EC of MNH visits, coverage 234 scores were dichotomised, taking woman-specific score (of respective MNH visit).

235
Although there are no gold standards of cut-off points for categorisation of poor or 236 good effective coverage, this study took reference cut-off point of 0.80 to a previous 237 study undertaken in Kenya [47] and Nepal [48]. analyses [49]. Backward elimination multivariable logistic regression analyses were 261 conducted [50]. For which, firstly, a full multivariable regression model was run, and 262 then estimated p-values for each independent variable and the insignificant variable 263 were identified. This procedure was repeated until no insignificant independent 264 variable was left at p<0.2 [51]. The adjusted odds ratios (aOR) with 95% CIs for all 265 independent variables retaining p<0.05 were reported. The goodness of fit tests was 266 conducted using the Hosmer Lemeshow test (non-significant results (p>0.05) 267 indicated an adequate fit) [52]. All analyses were conducted using the survey 268 command function and considering the clustering effect in Stata 14.0 (Stata Corp, 269 2015). 270

271
Descriptive characteristics of women 272 Table 1 shows the background characteristics of the women included in the analysis.

273
Among 1,978 women, 42% were from households in the lowest two wealth quintiles. 274 More than two-thirds (69%) of women were from disadvantaged ethnic groups, 275 mostly Madhesi, Janajatis and Dalits. Nearly two in five women (42%) were native 276 Nepali speakers (the national language). Male household head characterised more 277 than two-thirds (73%) of the households. More than half (55%) of women were from 278 the Terai (Plain) Region. One in four women (26%) were from province two, whereas 279 one in twenty women (6%) were from province six. About half (46%) of women were 280 from urban areas. Two-thirds (67%) of women had no decision-making authority (or 281 empowerment) in relation to access in health-seeking, buying something (financial 282 empowerment) or meeting with relatives (movement authority). Nearly one-third 283 (29%) of women reported any kind of perceived violence (e.g., beating when food 284 burnt or beating if women went out without asking husband). In total, four in five 285 (79.7%) women were aged 20-34 years, and approximately 69% of women did not 286 have a bank account. Three in five women perceived distance to a health facility was 287 a challenge when accessing health services. Further, nearly 72% of women 288 perceived it as challenging to access care when there was no available female 289 healthcare worker. In addition, over two-thirds (68%) of women had no awareness of 290 the availability of a health mothers' group in their community. One in ten mothers 291 delivered babies via C-section (Table 1). 292

Determinants of good EC of routine MNH visits
In the bivariable logistic analysis, 16    Model I: outcome variable: good EC of 4ANC visits. Independent variables: wealth status, language, ethnicity, religion, maternal education, maternal occupation, perceived violence, household head, residence, province, maternal age, child sex, access to a bank account, media exposure, index child, empowerment, distance to HFs, the perceived problem of not having female providers, awareness on health mothers' group, c-section delivery.

Summary of the findings
The effective coverage of 4ANC visits, institutional delivery, and PNC visit was 52%, 33% and 23%, respectively. These findings provide a baseline for the future estimation of the quality of MNH services in Nepal. Women with disadvantaged ethnicity or those from province six, speaking Maithili as the first language, and those who had high birth order (≥4) had lower odds of EC of MNH visits. Women who had access to a bank account, or women who had C-section delivery had higher odds of good EC of MNH visits. Women who perceived problem if not seen by female providers had poor EC of MNH visits compared to who did not care about the gender of health care providers.

EC of routine MNH visits and determinants
This study found low EC of all forms of MNH visits (4ANC visits, institutional delivery, and PNC visit) compared to contact coverage of respective coverage of MNH visit.
For example, in Nepal, contact coverage of 4ANC visits, institutional delivery and first PNC visits was 71%, 64% and 51% respectively [53], but EC was lower than these figures. These suggest that women are reaching to HFs but not receiving all essential interventions. Yet, while contact coverage of routine MNH visits is essential on its own it is inadequate to avert MMR and NMR.
Poor EC of MNH visits identified in the current study are consistent with the studies conducted in Bangladesh [54], Cambodia [55], and other LMICs of South Asia and Sub Saharan Africa [18,46]. Studies conducted in Nepal reported poor quality of ANC [48] and had low uptake of recommended antenatal interventions (e.g., ANC counselling or iron or tetanus toxoid injections) [56]. An analysis of multi-country data revealed low EC of facility delivery [57] and 4ANC visits and PNC visit [58] despite high contact coverage of respective MNH visits. Few studies also reported poor EC of other health services such as treatment of sick children (e.g., treatment of diarrhoea, pneumonia) [46] and family planning services [23,46]. In addition to challenges in reaching to HFs, poor EC coverage may have contributed by the health system's weakness, such as poor readiness, including a lack of trained health workers, or shortages of medicine for MNH services.
The current study identified several determinants associated with EC of routine MNH services including ethnic and geographic disadvantage, Maithili speaking women, birth order, and gender of health care providers. First, in Nepal, ethnicity and socioeconomic status are associated with power and position in the society [59,60].
Women with upper-caste groups usually have higher wealth status, usually are Nepali native speaker, they get a good opportunity for education and employment and are likely to have better awareness of health matters and the availability of health services. Disadvantaged ethnicities belong to the lower strata of the hierarchical caste system of Nepal [61]. Women with economic and ethic disadvantages have poor access to health services in Nepal [62]. Health programs should focus on disadvantaged ethnicities and design context-specific strategies to provide effective health services to those groups.
Secondly, women of province six had poor EC that could be contributed by scattered settlements, poor transportation facilities with limited access to reach HFs [63]. Reaching HFs is problematic due to limited road networks and poor transportation facilities [64,65]. Women's access to HFs can be improved via local development approaches such as constructing HFs in strategic locations and connecting community and health facilities by constructing bridges and road networks. However, these are non-health sector interventions that require multisectoral actions. Third, non-native Nepali speaker women, for instance, Maithili speaking women, usually have poor access to information. Poor access to health information might have contributed due to their low literacy rate [66]. Health education and information programs are not available in the local language, which affects health literacy among disadvantaged groups. In some cases, women cannot express their health care needs to health services providers if they could not understand Nepali or if health providers are unable to speak local languages.

Recruitment of local health workforce and health education interventions conducted
in local language can increase the uptake of MNH interventions.
Fourthly, women with high birth order (four or more) had poor EC of MNH visits in their recent pregnancy and childbirth. If women have more children, they may not prioritise the latest pregnancy and might not take the recommended MNH interventions in their pregnancy and childbirth [67]. Effective family planning services could reduce the number of birth order [67,68]. Women with wanted pregnancy visit health facility and take recommended MNH interventions for a healthy pregnancy and healthy childbirth. Fifth, some women prefer female providers for the uptake of MNH services in Nepal. Female health care providers can understand the need for health service users, and evidence indicates the provision of female providers can improve access to health services [69]. Similarly, if women receive good quality ANC, they are more likely to receive good quality institutional delivery services or PNC visit [37]. Improved quality of care is also likely to increase subsequent utilisation of MNH visits [56]. In nutshell, both access and health system response approaches care likely to improve EC of routine MNH visits. These approaches include equipping HFs with adequate supplies, trained health workers (e.g., local health workers who can understand local language and culture) and respectful maternity care and improved awareness on uptake of quality pregnancy, childbirth, and postnatal care from HFs.
This study demonstrated the measurement of the EC of routine MNH visits using NDHS data. The quality of care is emphasised in global health policies and universal access to quality MNH services SDG3 [9]; evaluating EC however, requires a population level measurement of quality of care. The EC of MNH visits can provide a proxy indicator of quality of care and go beyond simply contact coverage. This is important as evidence suggests access to health services alone cannot result in the intended MNH outcomes [13,70]. Measurement of quality of care requires multiple sources of data, such as users, providers, facility inventory, and observation of interaction between providers and users [9,71]. This study also demonstrates, however, a single source of data can be used to assess the health care quality. This study assessed the EC of routine MNH visits taking information on services users' engagement with the health system and using data of household survey and information on adequate care, timely care, and frequent and skilled care while women visit health facilities.

Policy and programmatic implications of the study
This study has implications for policies and programs. Firstly, the EC has been an important concept for its applicability in health system performance and could be instrumental in tracking the progress towards universal health coverage of MNH services [17]. Secondly, in Nepal, monitoring systems have given limited attention to measuring the EC of health services including MNH services [72]. The methods used to estimate to assess the EC could provide ideas for the measurement of quality of health services. Thirdly, the universal coverage of quality MNH services are essential and need to focus on the quality of care for the reduction of NMR and MMR. To achieve SDG3 targets, the programmatic and policy focus should be beyond the contact coverage of MNH visits as only improving access to health services or contact coverage could not result in intended better MNH outcomes. Finally, this study identified disadvantaged women who were getting poor EC of MNH visits and identified the need for targeted approaches for effective uptake of MNH interventions among those groups. Programs and policies should focus on women with poor wealth status and ethnic disadvantages, and women speaking Maithili as the first language. Context-specific program approaches require to receive essential MNH interventions and ultimately improve the EC of MNH visits.

Limitations of the study
This study has some limitations. Firstly, data were self-reported, and thus, recall bias may occur. However, this study restricted to analysis among women who had a live birth in the two years prior to the survey (2014-2016), which is a relatively shorter recall period than other studies published using NDHS datasets in the past. Secondly, outcome variables were self-reported by women during face-to-face interviews, which may lead to recall bias and social desirability bias. Thirdly, assessment of EC was limited by the information of MNH interventions available in the NDHS 2016. The NDHS did not collect information on all the WHOrecommended essential MNH interventions [4,5,45]. This study emphasises the importance of collecting information on MNH interventions in NDHS or routine HMIS system.

Conclusions
The uptake of recommended MNH interventions was low in all routine MNH visits (4 ANC visits, institutional delivery, and the first PNC visit) in Nepal. Poor EC of MNH   However, the first author took approval for the download and use of the dataset for his doctoral thesis and this publication.