Continuity of care and its determinants of routine maternal and newborn health visits in Nepal: Evidence from a nationally representative household survey

Resham Bahadur Khatri (  rkchettri@gmail.com ) Health Social Science and Development Research Institute, Kathmandu Rajendra Karkee School of Public Health and Community Medicine, BP Koirala Institute of Health Sciences, Dharan, Jo Durham School of Public Health and Social Work, Queensland University of Technology, Brisbane Yibeltal Assefa School of Public Health, Faculty of Medicine, University of Queensland, Brisbane


5
Maternal and newborn health (MNH) is a priority public health issue in low-and 6 lower-middle-income countries (LMICs). Most maternal and newborn deaths occur in 7 Sub-Saharan Africa and South Asia [1]. The majority of the maternal and newborn 8 deaths could be prevented through uptake of essential antenatal, intrapartum and 9 postnatal interventions [2]. The World Health Organization (WHO) recommends 10 women should receive health interventions during routine MNH visits such as at least 11 four antenatal care (4ANC) visits, institutional delivery assisted by skilled health 12 providers [3], at least three PNC visits within the first week after childbirth [4]. 13 A study of 75 LMICs high burden of maternal and neonatal deaths estimated 14 increased access and quality of essential MNH interventions could reduce up to 71% 15 of neonatal deaths, 33% of stillbirths, and 54% of maternal deaths annually [5]. The 16 coverage of routine MNH visits during maternity period, however, is often low and 17 characterised by high rates of discontinuation along the continuum of care (CoC).

18
For example, the completion of all routine MNH visits was low in several LMICs [6-19 9], including in Ghana [10], Cambodia [6], and Tanzania [11]. In Tanzania, 90% 20 dropout was reported from first ANC visit to PNC visit; while the highest (55%) 21 proportion was seen from institutional delivery to a PNC visit [11]. 22 Maternal and newborn health is continuum of care (CoC) from the life cycle 23 perspective [8]. This perspective of CoC describes delivery of health services from 24 conception through to birth, and childhood-adolescence-adulthood period. The 25 antenatal, intrapartum, and postnatal is a shorter version of CoC. This period is vital 26 for health status of mothers and newborns, and is the combined construct from 27 survival and health service delivery [7]. It is arguably a single entity except for their 28 biological differences; for instance, interventions received by pregnant women can 29 affect health newborns. In the MNH continuum, mid-level health workers can provide 30 essential MNH interventions services up to the first level of health system (peripheral 31 health facilities) and community level. The combined coverage of 4ANC visits, 32 institutional delivery, and first PNC visit can be considered as the marker of the CoC 1 of MNH [12][13][14]. 2 Globally, the CoC in MNH has received substantial attention in research, policy and 3 programs over the past two decades [7]. For instance, the sustainable development 4 goal three (SDG3) states universal coverage of quality MNH services across the 5 CoC (target 3.8) [10,15]. Out of nine tracer services in SDG3, two are related to 6 MNH services, such as childbirth assisted by skilled health attendants and health 7 facility delivery [16]. Thus, the assessment of composite coverage of all routine visits 8 is essential to track the coverage of tracer MNH services and SDG3 target. 9 Nepal has the highest maternal and neonatal death rates within South Asia [17,18]. 10 Annually 259 (per 100,000 live births) women die due to pregnancy and childbirth-11 related issues, and 21 (per 1000 live births) newborn die within the first month of 12 birth in Nepal [19]. High maternal and neonatal deaths in Nepal may be contributed 13 by low coverage of MNH visits, thereby lack of uptake of essential interventions 14 across MNH continuum. Past evidence in Nepal showed poor access to MNH 15 services and poor completion of MNH visits. For instance, Nepal Demographic and 16 Health Survey (NDHS) 2016 reported 70% of pregnant women received 4ANC visits,17 and nearly two in five women received institutional delivery and first PNC visit within 18 48 hours of childbirth [19]. A study in 2019 reported only 40% women completed all 19 routine MNH visits, i.e., 4ANC visits, institutional delivery, and PNC visit [8]. The 20 same study reported among rural women, and those with higher birth order (more 21 than two children) had lower odds of MNH visits across the CoC [8]. Another study 22 reported women of wealth status and illiterate women low the completion of MNH 23 visits across the CoC [20]. Other studies revealed that the completion of 4ANC visits 24 contributed to uptake of institutional delivery [21] and PNC visit [22], and women who 25 received institutional delivery services were more likely to receive PNC visit [23][24][25]. disadvantaged Janajatis) and advantaged ethnicities (includes Brahmin/Chhetri, 8 advantaged Janajatis). Maternal education was categorised into illiterate (who 9 cannot read and write), and primary (up to grade eight), and secondary and higher 10 (who have education of grade nine or higher). In the NDHS 2016, wealth quintiles 11 were constructed using principal component analysis (PCA) based on more than 40-12 asset items being owned by households. In this study, these wealth quintiles were 13 merged into two groups, such as the lowest two quintiles as Poor (lower 40%), and 14 upper three quintiles as Rich (upper 60%). 15 This study had one outcome variable with four mutually exclusive categories: 16 discontinued before completing 4ANC visits=1; completed 4ANC visits but 17 discontinued before completing institutional delivery =2; completed 4ANC visits and 18 institutional delivery but discontinued before completing PNC visit=3; completed all 19 three MNH visits=0 (reference category). 20 21

Statistical analysis 22
Multinomial logistic regression analysis was conducted, and the magnitude of 23 (dis)continuity of care was reported as relative risk ratios (RR) with 95% confidence 24 intervals (CIs). In the analysis, sampling weights (available in the NDHS 2016 25 dataset) have been calculated and applied, so results are representative at the 26 national as well as strata levels. All analyses were weighted to adjust for the two-27 staged cluster sampling used in the NDHS 2016 survey (primary sampling unit=383; 28 stratification (strata= 14; province seven with rural and urban; strata); survey weights 29 (probability weight = sample weight/1,000,000) [19]. All estimates were reported in 30 weighted value (unless otherwise indicated) including frequency, and proportion (%).

31
The clustering effect of complex sampling design was adjusted using survey 'svy' set 32 command in Stata 14.0. 33 Before running the multivariable multinomial regression model, multicollinearity was 1 checked and excluded independent variables having variation inflation factors ≥3 2 [33]. Backwards elimination multivariable multinomial logistic regression analyses 3 were conducted [34]. First, the full multivariable regression model was run, estimated 4 p-value for each independent variable. Then identified the most insignificant variable 5 was deleted comparing p values. This procedure was repeated until no insignificant 6 independent variable was left at p<0.2 [35]. The statistical significance level was set 7 p<0.05 (two-tailed) to identify the independent variables associated with the outcome 8 variable. The goodness of fit test was conducted using the Log-likelihood Ratio (LR) 9 test [10]. All analyses were conducted using Stata 14.0 (Stata Corp, 2015). 10

11
Background characteristics of women 12 Table 1 shows the background characteristics of women included in this study.

13
Among the 1,978 women, 42% were from households in the lowest two wealth 14 quintiles. More than two-thirds (69%) of women were from disadvantaged ethnic 15 groups, mostly Madhesi, Janajatis and Dalits. Nearly two in five women (42%) were 16 native Nepali speakers (the national language). Nepali is primarily spoken in the Hill 17 region, where most of the residents are from relatively advantaged ethnicities [36].

18
Male household head characterised more than two-thirds (73%) of the households. 19 More than half (55%) of women were from the Terai (Plain) Region. One in four 20 women (26%) were from province two, whereas the smallest percentage of women 21 (6%) were from province six. About half (46%) of women were from urban areas. 22 Two-thirds (67%) of women had no decision-making authority in relation to accessing 23 in health-seeking, buying something (financial decision making in the family) or 24 meeting with relatives (movement). Nearly one-third (29%) of women reported any 25 kind of perceived violence (e.g., beating when food burnt or beating if women went 26 out without asking husband). In total, four in five (79.7%) women were aged 20-34 27 years, and approximately 69% of women did not have a bank account. 28 Three in five women felt distance to a health facility was a challenge when accessing 29 health services. Further, nearly 72% of women perceived it as challenging to access 30 care when there was no available female healthcare worker. In addition, over two-31 thirds (68%) of women had no awareness of the availability of a health mothers' 32 group in their community. One in ten mothers gave childbirth through caesarean-1 section. 2

(Dis)continuity of care of routine MNH visits in antenatal, intrapartum, and 2
postnatal period 3 Figure 2 shows the continuity of routine MNH visits across the CoC. Among 1,978 4 women included in this analysis, only two in five (41%) attended all three MNH visits 5 (4ANC visits, institutional delivery, and one PNC visit within 48 hours of childbirth). 6 Almost all (96%) received at least one ANC visit, but only 71% completed 4ANC 7 visits. More than one in two women (52%) completed at least 4ANC visits and 8 received institutional delivery services. Women without 4ANC visits, however, had a 9 higher rate of home delivery. For instance, among women who were unable to 10 complete 4ANC visits, 58% of them gave birth at home, while 71% of women with no 11 ANC visits (n=72) delivered at home. Only 4% (of N=1,978) of women did not 12 receive any of ANC visits or institutional delivery, or PNC visit ( Figure 2). 13 3 Table 2 shows the women who completed/discontinued MNH visits across the CoC. 4 Over half of women completed all visits if they were from provinces three (51%) and 5 four (54%), belonged to advantaged ethnicity (54%), had secondary or higher-level 6 education (54%), had jobs (53%), had a bank account (54%), had media exposure 7 (51%), perceived the distance to the health facility was not a problem (53%) and who 8 delivered via caesarean-section (71%). However, only one in four women completed 9 all three MNH services if they were from province six (24%), a Bhojpuri speaker 10 (21%), illiterate (25%), and higher birth order (≥4) (21%) ( Table 2). 11 12 1

Determinants of the (dis)continuity of MNH visits across the continuum of care 2
Along the pathway of the antenatal-postnatal period, there were three possible points 3 of discontinuation: before completing 4ANC visits, before completing institutional 4 delivery, and/or before completing a PNC visit. In the bivariable regression analysis, 5 several factors were associated with discontinuation along the pathway 6 (Supplementary file; Table 2). Associated structural factors were language, wealth 7 status, education; and intermediary factors significantly associated were the place of 8 residence, province, region, birth order, media exposure on public health issues, 9 access to a bank account, the intention of last birth, perceived problem of the long 10 distance to the health facility, and perceived violence. In addition, health system 11 factors associated significantly with the discontinuity of care were the perceived 12 problem if not having female providers in health facilities, awareness of health 13 mothers' groups, and mode of delivery. 14 visits compared to the reference category. 10 Nine determinants were significantly associated with continuity of care until 4ANC 11 visits but discontinued before completing institutional delivery ( visits) compared to women from province one and those aged 20-34 years. 27 Two determinants were associated with continuity of care until 4ANC visits and 28 institutional delivery but discontinued before completing PNC visits (

1
The current study examined the composite of coverage routine MNH visits and had 2 low completion of all routine MNH visits across the CoC. We found more than two-in-3 five (41%) women in Nepal received all routine MNH visits across the CoC. There 4 was high proportion of discontinuation around later weeks of pregnancy (4ANC 5 visits) or around childbirth (institutional delivery). Several structural determinants 6 were found to be associated with discontinuity of care across the CoC. For instance, 7 women with structural disadvantages (e.g., disadvantaged ethnicity, women of lower 8 wealth status, illiterate women) had a higher risk of discontinuation across the CoC. 9 Intermediary and health system determinants contributed to the (dis)continuity of 10 care of MNH services (e.g., provinces six or seven, Maithali speaker women, high 11 birth order, and poor media exposure on health issues); if women had poor 12 awareness on health mothers' group, and perceived problem if not having female 13 providers in health facility found to have higher risk of discontinuation of MNH visits 14 across the CoC. 15

16
The reasons for low CoC in this study may be due to high discontinuation around 17 later gestational week of childbirth, and no PNC visit of women gave birth in health 18 facilities. Health awareness on the importance of pregnancy, childbirth, and PNC services can 31 be improved via exposure to mass media (e.g., local radios, television) and 32 dissemination of health information to current and future mothers. A past study in 33 Nepal reported mass media exposure was positively associated with maternal 34 healthcare utilisation [49]. Health awareness through mobile technology could play 1 an important role in the utilisation of health services generally [50] and the MNH 2 services particularly [51]. Context-specific strategies can be adopted to increase the 3 uptake of needed MNH visits that include outreach clinics in remote and 4 underprivileged communities, or mobilisation of local community workers for PNC 5 home visits [52]. In addition, properly functional health mothers' group in the 6 community could raise awareness among pregnant women and provide necessary 7 health information in their pregnancy and childbirth. Health mothers' groups are 8 women-led community health groups where current and future mothers can gather 9 and discuss reproductive, maternal, child health and nutrition issues [53]. Such 10 health groups could address the social taboos as talking about reproductive health-11 related issues is culturally taboo in Nepali society, and women are usually like to 12 share the provision of female providers [26,54]. Additionally, the current SDIP has 13 provisioned 4ANC visits and institutional delivery, but the provision of financial 14 incentive lacks for PNC visit. Ensuring financial incentive in the SDIP programs could 15 increase the uptake of all three MNH visits. 16

17
The

Policy and programmatic implications 31
This study has some implications for programs and research. First, the creation and 32 execution of a composite indicator provided insight into MNH CoC and should be 33 included in routine health management information system (HMIS) and periodic 34 health survey (e.g., demographic and health survey). The quality of MNH services is 1 prioritised in SDG3, which focuses on universal coverage of MNH services [58]. 2 Universal coverage of quality MNH services is crucial for better MNH outcomes and 3 in achieving SDG3. The government of Nepal should focus its programs targeting 4 women living in difficult geographical areas (e.g., province six) and women with 5 social disadvantages (e.g., poor, marginalised ethnicity). Supply-side approaches 6 found to improve the better health services delivery included strengthening birthing 7 center (e.g., health logistics, human resources and training), and establishing 8 maternity waiting home [59]. Availability of childbirth services in all rural health 9 facilities could increase the facility childbirth and first PNC visit. The home visits 10 through community health workers in hard to reach community could also improve 11 PNC visit home visit. 12

13
Second, the provision of female health providers could improve the counselling on 14 the importance of MNH services in the MNH continuum. Other potential strategies 15 could be mass media mobilisation to raise awareness on the COC and focused MNH 16 services to disadvantaged population groups, such as women of lower wealth status, 17 who speak Maithali speak, living in remote areas. The health system could formulate 18 focussed service delivery packages to women with structural disadvantages, for 19 instance, provision of focus incentive to those groups who are most marginalised 20 women, Dalits, Karnali province, Maithali speaking women. The current safe delivery 21 incentive program (SDIP) includes separate incentives for 4ANC visits, and 22 institutional delivery program [60]. This SDIP should be reformed by providing 23 incentive for women those who complete 4ANC visits, delivered babies in health 24 facilities and complete first PNC visit. Additionally, such incentive program needs to 25 be designed for specific groups based on marginalisation status. The composite 26 coverage indicator employed in this study could help to reform the SDIP. 27 28 Third, health services need to focus on remote areas, including improving family 29 planning services. Better family planning services could improve the birth spacing, 30 thereby reducing unintended pregnancy and reduced numbers of parity. Women of 31 wanted pregnancy may complete all routine MNH visits and receive all essential 32 antenatal, intrapartum, and postnatal interventions for their and newborns better 33 health. 34 1

Strengths and limitations of the study 2
This study has some strengths and limitations. Strengths included; first, this study is 3 based on a nationally representative survey with higher response rate (98%), and 4 findings could be generalised at the national level. Second, this study considered the 5 PNC visit for mothers and newborns rather than previous studies that examined PNC 6 visit for newborn or PNC mothers separately. This study has the following limitations. 7 First, inferences drawn from this study are based on an observational and cross-8 sectional design, which allows the study of correlations rather than causality. 9 Second, the NDHS 2016 collected information based on recall of women who had a 10 live birth five years prior to the survey (2011-2016); however, we included a short 11 recall period of two years restricting study sample of women who had a live birth two 12 years preceding the survey (2014-2016). Third, this study is based on secondary 13 data analysis; we were unable to include important variable such as obstetric 14 complications that could contribute to discontinuation along the pathway. Fourth, the 15 outcome variable was self-reported after face-to-face interviews with women, which 16 may have social desirability bias (e.g., over-reporting of good behaviours and 17 underreporting of bad behaviours) and misclassification. Finally, from the research 18 perspective, this study has not explored stories of why women discontinued health 19 services utilisation across the CoC. The qualitative study could provide a deeper 20 understanding of real stories of the underlying reasons for discontinuation across the 21 CoC. 22