Assessing the effect of pregnancy intention at conception on the continuum of care in maternal healthcare services use in Bangladesh: Evidence from a nationally representative cross-sectional survey.

BACKGROUND: The Continuum of Care (CoC; defined as accessing the recommended healthcare services during pregnancy and the early postpartum period) is low in lower-middle-income countries (LMICs). This may be a major contributor to the high rates of pregnancy-related complications and deaths in LMICs, particularly among women who had an unintended pregnancy. With a lack of research on the subject in Bangladesh, we aimed to examine the effect of unintended pregnancy on CoC.
METHODS: Data from 4,493 mother-newborn dyads who participated in the cross-sectional 2014 Bangladesh Demographic and Health Survey were analysed. Women's level of CoC was generated from responses to questions on the use and non-use of three recommended services during the course of pregnancy: four or more antenatal care (ANC) visits, skilled birth attendance (SBA) during delivery, and at least one postnatal care (PNC) visit within 24 hours of giving birth. Global recommendations of service use were used to classify CoC as high (used each of the recommended services), moderate (used at least two of the three recommended services), and low/none (no PNC, no SBA, and ≤3 ANC visits). Women's pregnancy intention at the time of conception of their last pregnancy (ending with a live birth) was the major exposure variable, classified as wanted, mistimed, and unwanted. Unadjusted and adjusted (with individual-, household-, and community-level factors) multilevel multinomial logistic regression models were used to assess the association between unintended pregnancy and level of CoC.
RESULTS: In Bangladesh, the highest level of CoC occurred in only 12% of pregnancies that ended with live births. This figure was reduced to 5.6% if the pregnancy was unwanted at conception. The antenatal period saw the greatest drop in CoC, with 65.13% of women receiving at least one ANC visit and 26.32% having four or more ANC visits. Following the adjustment of confounders, an unwanted pregnancy was found to be associated with 39% and 62% reduced odds of women receiving moderate and high levels of CoC, respectively, than those with a wanted pregnancy. Having a mistimed pregnancy was found to be associated with a 31% reduction in odds of women achieving a high CoC than women with a wanted pregnancy.
CONCLUSION: Almost nine in ten women did not achieve CoC in their last pregnancy, which was even higher when the pregnancy was unintended. Given that the ANC period has been identified as a critical time for intervention for these women, it is necessary for policies to scale up current maternal healthcare services that provide in-home maternal healthcare services and to monitor the continuity of ANC, with a particular focus on women who have an unintended pregnancy. Integration of maternal healthcare services with family planning services is also required to ensure CoC.

Introduction

Reducing complications in pregnancy and childbirth remains a major challenge in low- and middle-income countries (LMICs). Complications are responsible for around 822 maternal deaths a day in LMICs (99% of the 830 maternal deaths worldwide) [1,2]. Moreover, of the 5.4 million under-five deaths, which occur globally every year, 4.4 million deaths could be averted if LMICs had the same under-five mortality rate (currently 69 deaths per 1,000 live births) as high-income countries (5 deaths per 1,000 live births) [3]. To do so, the current under-five mortality rate in LMICs needs a 14-fold reduction [3]. The current neonatal mortality (2.5 million annually in LMICs; 47% of the total under-five deaths) has to be reduced by around 50 times, most of which occur because of birth complications [4,5]. Therefore, maternal, newborn and child health remains a priority issue in LMICs [6-8]. Consequently, Sustainable Development Goal (SDG) 3.1 (reduce maternal mortality ratio to less than 70 per 100,000 live births) and SDG 3.2 (reduce under-five and neonatal deaths to 25 and 12 per 1,000 live births, respectively) were designed to significantly reduce maternal and under-five deaths between 2015 and 2030 [9]. Achieving these ambitious goals requires ensuring universal access to sexual and reproductive healthcare services (SDG 3.7) and universal health coverage (SDG 3.8) for women and children under five, which are still challenges in LMICs [3,9].

Elements of the Millennium Development Goals (MDG) between 2000 and 2015 focused on contraception, antenatal care (ANC), and skilled birth attendance (SBA) coverage [10]. The targets were a two-thirds reduction of under-five mortality (MDG 4) and a three-quarters reduction of maternal mortality (MDG 5) [10]. Consequently, global healthcare services coverage was increased significantly (e.g. SBA increased from 59% in 1990 to 71% in 2015) although not in LMICs, including Bangladesh [2,10,11]. Increased coverage of particular healthcare services does not ensure the prevention of maternal, newborn and child mortality [12]. Consequently, MDGs 4 and 5 were mostly unmet in LMICs [2]. Also important was an inadequate focus on postnatal care (PNC) in the MDGs, which global estimates found could be effective in preventing approximately 80% of current maternal deaths and two-thirds of current neonatal deaths [8,13]. Therefore, the World Health Organization (WHO) is now advocating for integrated service delivery for women and children from pre-pregnancy to delivery and the immediate postnatal period [14,15], whereas the SDGs are prioritizing universal health coverage [9].

Continuum of Care (CoC) focuses on the continuation of care received throughout the life cycle and the integration of health providers at various level of services (primary, secondary, and tertiary levels). CoC throughout pregnancy is focused on the use of at least four skilled ANC visits during pregnancy (later changed to eight ANC visits in 2016), SBA during delivery, and the first PNC visit for women and their newborns within 24 hours of birth. The WHO recommends these services to achieve better pregnancy outcomes and significant reductions in maternal and newborn morbidity and mortality in LMICs [9]. The pathway to such reductions is through preparing women for delivery [16], by enabling them to identify illnesses [17], by enabling healthcare facilities to provide obstetric care [18], and by referring women to a higher level of services to avoid life-threatening obstetric complications [17].

The available evidence has shown there is a low rate of CoC and little has changed in LMICs, especially during the period of the MDGs [12,18-20]. One important reason for this low rate is the lack of focus on family planning and preconception care in CoC services. This is a missed opportunity to start women on the CoC pathway, with many prevention activities needing to occur prior to conception. Women with an unintended pregnancy are at particular risk of not receiving CoC as unintended pregnancy often occurs in the absence of family planning and preconception care [21]. Despite this, it is still unclear where the gaps are in seeking care along the continuum and what factors contribute to these gaps [12,19]. The few studies conducted on this topic are equivocal, particularly in terms of estimated CoC rates and what factors contribute to a lack of CoC [12,18-20].

Two studies from retrospective cross-sectional surveys in Ghana found that around 8% of pregnant women received CoC. Here, CoC was associated with women’s geographical location, cohabiting marital status, traditional religion, transportation problems, and lack of knowledge regarding childhood illnesses [20,22]. Similar factors were found in an analysis of nine countries in sub-Saharan Africa (13.9% CoC) and South Asia (24.5% CoC), where 85.5% of the current global maternal deaths occur [12]. Meanwhile, increased age at first delivery, lower parity, increased years of education and wealth quintile, residing in urban areas, having high autonomy, and exposure to mass media were factors found to be associated with 27% and 45% completion of CoC in Pakistan [18] and Nepal [23], respectively. These factors were also found to be common in a regionally-based study conducted in Tanzania, where 10% of women had CoC [24]. However, in Cambodia, of these factors, only being in the highest wealth quintile was found to be significantly associated with a 60% CoC rate [19]. Importantly, these estimates of CoC are significantly lower than comparable countries’ weighted indicators (composite coverage index) of ANC, SBA and PNC [25]. This further suggests that achieving CoC is associated with a different set of factors than the factors associated with particular service use, as receiving a service at one level does not necessarily mean the following service is received.

The high occurrence of obstetric complications challenges the achievement of the SDGs’ targets of reducing maternal and newborn deaths in LMICs and Bangladesh [26]. Obstetric complications are even higher for women with an unintended pregnancy, who represent around 43% of total pregnancies at conception in Bangladesh (including pregnancies that are terminated) and around one-quarter of the total pregnancies that ended with live births [26-28]. This may be a result of not using components of CoC or discontinuation of CoC. However, previous research identified the negative effects of unintended pregnancy on specific service use (ANC, SBA and PNC) in LMICs, including Bangladesh [21,27,29,30]. The causes are individual correlates of unintended pregnancy (e.g. lower education, lower wealth quintile), family influences (e.g. husband and family member opposition and negligence), community context (e.g. higher poverty and lower education), as well as psychological responses following pregnancy (e.g. depression and anxiety) [21,26]. However, focus on the effect of unintended pregnancy on CoC is notably absent in the global literature. To the best of our knowledge, only two studies in Ghana have examined the effect of unintended pregnancy on CoC. Both were conducted in a regional context, where a positive association [20] and non-association [22] with CoC were found, respectively, for women reporting a mistimed and unintended pregnancy compared to women with a wanted pregnancy. These differing findings are in part due to the adjustment of different confounders and the analysis of small samples. Therefore, the objective of this study was to examine the effect of pregnancies that were unintended at conception on the level of CoC in Bangladesh. Nationally representative samples were analysed, controlling for a range of individual-, household-, and community-level factors that may potentially influence the association.

Methods

Study design

This study analysed secondary data extracted from the 2014 Bangladesh Demographic and Health Survey (BDHS). The details of the survey have been published elsewhere [11,31]. In brief, the BDHS was a nationally representative cross-sectional survey conducted in all sampled households, where women aged 15–49 years were eligible to participate in an individual questionnaire. The households were selected based on a stratified two-stage cluster design whereby the first stage was the sample enumeration areas or clusters drawn from the 2011 Bangladesh national census framework. A sample of households was selected in the second phase through stratified random sampling of the list of households in each selected cluster. The National Institute of Population Research and Training in Bangladesh conducted this survey in collaboration with MEASURE DHS, USA. Financial and technical support was provided by international development organizations, including The United States Agency for International Development (USAID) and The United Nations Children’s Fund (UNICEF) [11].

Data

Data from a subsample of 4,493 of the 17,863 women who took part in the survey were analysed. Details of this sample selection procedure are presented in Fig 1. The conditions of inclusion were as follows: i) provided a response to the retrospective question regarding giving live birth in the three years prior to the date of the survey; ii) responded to the questions on the use of maternal healthcare services during their pregnancy (most recent pregnancy if more than one pregnancy occurred); and iii) reported on pregnancy intention at conception of the included pregnancy (which was reported retrospectively).

Fig 1

Schematic of analytic sample selection process for unintended pregnancy and continuity of care of maternal healthcare services.

Outcome variable

The outcome variable, CoC, was generated from women’s responses to questions on the use or non-use of ANC, SBA, and PNC. Two survey questions regarding ANC services were used to determine if women had received ANC, and if so, how many times. Each eligible woman was first asked, “Did you see anyone for antenatal care for this pregnancy?”. Women who responded “Yes” to this item were then asked, “How many times did you receive antenatal care during this pregnancy?”. SBA status was derived by combining women’s responses to the two questions that were asked to determine who had provided care during delivery (“Who assisted with the delivery of [name of the most recent child born in the past three years])?” and where delivery had occurred (“Where did you give birth to [name of the most recent child born in the past three years])?”. Four additional questions collected information on the timing of the first PNC visit for women and newborns, as well as providers of PNC. Each eligible woman was first asked, “Did anyone check on you and your baby’s health following birth?” Further questions were asked to determine the timing (“How long after delivery did the first check take place?”), place (“Where did this first check take place?”), and providers (“Who checked on your health at that time?”) of PNC. These items were recoded to generate the CoC variable depending upon the continuity of using maternal healthcare through to PNC. The WHO guidelines in 2014 (which were in use at the time of survey completion) on maternal healthcare service use and providers were followed to generate the CoC variable. All recommendations, except the number of ANC visits, remain in the current WHO guidelines. These were, during the course of a pregnancy, each woman should have the following: i) receive at least four skilled ANC visits (this was changed to eight ANC visits in 2016, however, four skilled ANC visits is the current Bangladesh government’s recommendation); ii) be assisted by SBA during delivery; and iii) receive at least one PNC visit (for women and their newborns) within 24 hours of delivery from skilled healthcare personnel [32]. The variable was classified (see Table 2) as: high (women received each of the three recommended services (four or more ANC visits, birth assisted by SBA, and received at least one PNC visit within 24 hours of birth); moderate (women received at least two of the three recommended services), and low/none (women were not assisted by SBA, received no PNC, and had three or fewer ANC visits). We included women with no use of healthcare services as low/none level of CoC due to low response (20/4493).

Table 2

Grouping of maternal healthcare indicators (ANC visits, delivery assisted by SBA, and PNC visits) into the level of continuum of care (highest, moderate, low/none), Bangladesh Demographic and Health Survey, 2014 (N = 4,493).

Delivery and postnatal care	Antenatal care
	Fewer than 4 ANC visits (n = 3305; 100%)	4 or more ANC visits (n = 1188; 100%)
Delivery by skilled birth attendant
Yes (n = 1,990; 100%)	1,137 (34.4%)	853 (71.8%)
No (n = 2,503; 100%)	2,168 (65.6%)	335 (28.2%)
Postnatal care for mother-newborn dyads within 24 hours of birth
Yes (n = 1,275; 100%)	700 (21.2%)	575 (48.4%)
No (n = 3218, 100%)	2;605 (78.8%)	613 (51.6%)
Prevalence of levels of healthcare-seeking behaviours
Level of the continuum of maternal healthcare	Number	Percent (95% CI)
Low/no CoC	2,129	47.4 (45.8–52.1)
Moderate CoC	1,801	40.1 (36.9–42.0)
Highest CoC (WHO’s recommended level)	564	12.5 (10.1–13.4)

*Classifications reflect current World Health Organization guidelines; N = total number of women included in the study; n = number of women who used a particular service(s); percentage in the parentheses is column total.

Explanatory variables

Women’s intention at conception of their most recent pregnancy that ended with a live birth was taken as the major exposure variable. Eligible women were asked i) “When you got pregnant with (name of last child that occurred within 3 years of survey date), did you want to get pregnant at this time?”. The response option was either “Yes” or “No”. If women responded “No”, they were asked, “Did you want to have a baby later on or did you not want any (more) children?”. Responses were then categorized as wanted (if the response was “yes” to the first question), mistimed (if the response was “later” to the follow-up question), and unwanted (if the response was “no more” in the follow-up question). The general validity of this measure of data collection and classification at the aggregate level has been repeatedly demonstrated in different settings, including Asian and African countries [33-38].

We adjusted the association between level of CoC and women’s intention at conception of their most recent pregnancy with a range of individual-, household-, and community-level factors. The factors were selected by using forward regression analysis on the list of factors found in our literature search [12,18-20,22,39]. The individual-level factors included were women’s age at the birth of their last child, decision-making about women’s healthcare, and women’s education. Partner’s education, partner’s occupation, exposure to mass media, and household wealth quintile were included as household-level factors. The BDHS’ reported variables, as well as the authors’ generated variables (based on individuals’ responses at the cluster level), were included as community-level factors. The BDHS’ reported factors were respondents’ place of residence and region at the time the survey was conducted. The authors’ generated factors were community-level illiteracy, community-level non-use of four or more ANC visits, community-level non-use of professional delivery care, and community-level non-use of PNC within 24 hours of birth. The details of the procedure to make these variables at the cluster level have been published elsewhere [27,30].

Statistical analysis

The study was designed and reported in accordance with Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines [40]. Basic sociodemographic characteristics of the women included were presented with means (±SD; for continuous variables) and frequency (for continuous and categorical variables). The different combinations of maternal healthcare services within the continuum of maternal healthcare was presented by percentages; the same was done separately for total pregnancies and across intention of pregnancy at conception. Dropout across the CoC (ANC through to PNC) was also calculated. Unadjusted and adjusted multilevel multinomial logistic regression models were carried out to assess the association between pregnancy intention and level of CoC. The unadjusted model was carried out to determine the raw effect of women’s pregnancy intention at conception on level of CoC. The adjusted model included individual-, household-, and community-level adjustment factors to determine the net effect of women’s pregnancy intention at conception on level of CoC. These models were employed using Stata’s generalized linear latent and mixed models (GLLAMM) command developed by Rabe-Hesketh and colleagues in 2005 [41]. The benefit of using this model is it can analyse the clustering structure of BDHS’ data (individuals nested within a household and households nested within a cluster), whereas previous studies found a multilevel model was deemed appropriate [12,41]. Sampling weights were also applied. Results were reported as odd ratios (OR) with 95% confidence intervals (CI). All analyses were done using Stata software version 15.1 (Stata Corp, College Station, Texas, USA).

Results

Characteristics of the sample

Table 1 shows the weighted sociodemographic characteristics of the women included. Of the 4,493 participants, most were aged 20–34 years (M = 23.63, SD = ±5.69), possessed a secondary level of education (M = 6.31, SD = ±3.83), and had ≤2 children (M = 2.16, SD = ±1.41). Around 26% of women reported that their last pregnancy that ended with a live birth was unintended at conception, 15.1% reported it was mistimed, and 10.9% reported it was unwanted.

Table 1

Weighted descriptive characteristics of women who gave at least one live birth within 3 years prior to the date of the 2014 Bangladesh Demographic and Health Survey (N = 4,493).

Variables	Percent
Exposure variable: pregnancy intention at conception
Wanted	74.1 (71.3–76.7)
Mistimed	15.1 (13.4–16.8)
Unwanted	10.9 (9.4–12.6)
Sociodemographic and reproductive characteristics
Women’s age (in years) at birth of last child
Mean (±SD)	23.63 (±5.69)
≤19	28.0 (26.3–29.7)
20–34	67.7 (65.9–69.4)
≥35	4.4 (3.4–5.2)
Women’s education
Mean years of schooling (±SD)	6.31 (±3.83)
Illiterate	14.2 (12.3–16.3)
Primary2	28.0 (26.1–29.9)
Secondary3	47.7 (45.2–50.3)
Higher4	10.2 (9.0–11.6)
Parity
Mean number (±SD)	2.16 (±1.41)
≤2	70.0 (67.8–72.2)
>2	30.0 (27.8–32.2)
Partner’s education
Mean years of schooling (±SD)	5.84 (±4.66)
Illiterate	23.9 (21.5–26.4)
Primary2	30.0 (28.0–32.0)
Secondary3	31.8 (29.4–34.3)
Higher4	14.4 (13.1–15.9)
Partner’s occupation
Agricultural worker	25.8 (23.0–28.8)
Physical lobourerb	44.0 (41.6–46.4)
Services	5.9 (5.0–7.0)
Business	61.6 (19.9–23.4)
Other	2.8 (2.0–3.8)
Household wealth quintile
Poorest	21.7 (19.0–24.6)
Poorer	18.9 (17.3–20.7)
Middle	19.1 (17.1–21.2)
Richer	20.6 (18.6–22.9)
Richest	19.7 (17.2–22.4)
Place of residence
Urban	26.1 (23.5–28.9)
Rural	73.9 (71.1–76.5)

Note: N = total number of women included in the study.

Level of the continuum of maternal healthcare

Table 2 contains information on elements of CoC defined following the WHO’s guidelines. Around half of all women (47.4%) who had a live birth in the three years before the survey had the lowest level of CoC and only 12.5% of women had the highest level of CoC (met WHO’s recommended levels).

Table 3 presents different combinations of maternal healthcare services received along the CoC by intention at conception of their last pregnancy. Of all the women, 27% had none of the maternal healthcare services, which increased to 37% for women who reported their last pregnancy was unwanted at conception. Around 14% of women with a wanted pregnancy received all services, and only 5.6% of women with an unwanted pregnancy received all services. SBA (3.7%) and PNC (0.5%) were uncommon without at least one ANC visit. After receiving at least one ANC visit, 7.2% of women had four or more ANC visits but no SBA and or PNC, 6.4% of women had four or more ANC visits and SBA but no PNC, and only 0.4% of women had PNC but did not have four or more ANC visits or SBA. These proportions were even lower among women who had a pregnancy which ended with a live birth that was mistimed or unwanted at conception.

Table 3

Proportion of women who received each combination of the continuum of maternal healthcare services by intention of pregnancy at the time of conception, Bangladesh Demographic and Health Survey, 2014 (N = 4493).

Continuum of maternal healthcare				Overall (n = 4,493)	Intention of pregnancy at conception
ANC (1+)	ANC (4+)	SBA	PNC (women and newborns)		Wanted (n = 3,362)	Mistimed (n = 670)	Unwanted (n = 461)
				27.4	25.16	31.7	37.3
Yes				20.0	19.2	18.2	28.4
		Yes		3.7	3.6	3.9	4.2
			Yes	0.5	0.4	0.4	0.7
Yes	Yes			7.2	8.1	5.2	3.5
		Yes	Yes	3.2	3.1	3.9	3.5
Yes	Yes	Yes		6.4	7.0	6.0	2.8
Yes		Yes		6.9	7.3	6.1	4.8
Yes	Yes		Yes	0.24	0.26	0.2	0.2
Yes			Yes	0.4	0.4	0.6	0.0
Yes		Yes	Yes	11.5	11.6	12.5	9.1
Yes	Yes	Yes	Yes	12.5	13.8	10.7	5.6

Note: ANC (1+) = at least one ANC visit; ANC (4+) = at least 4 ANC visits; SBA = skilled birth attendance; PNC = postnatal care for women and newborns within 24 hours of delivery; N = total number of women included in the study; n = number of women with a particular type of pregnancy.

Fig 2 shows the changes in the proportion of women who attained CoC from one service to the next by pregnancy intention at conception. The proportion of dropout from one level to the next is also presented. Around 14% of women who had a wanted pregnancy at conception received all services, which declined from 68% use of at least one ANC. The proportion of women who received all services declined further if the pregnancy was mistimed or unwanted at conception. Only 5.6% of women who reported their last pregnancy was unwanted at conception received all services, which declined from 54.5% use of at least one ANC. The highest proportion of dropout in CoC occurred in the translation of one ANC visit to four or more ANC visits, followed by translation of SBA to PNC, and having four or more ANC visits to SBA. The trends associated with dropout were more pronounced for pregnancies that were unwanted or mistimed than those that were wanted at conception.

Fig 2

Women who discontinue the continuum of maternal healthcare through to PNC by women’s pregnancy intention at conception of their last pregnancy that ended with a live birth.

Effect of pregnancy that was unintended at conception on level of continuum of maternal healthcare services use

Table 4 shows the unadjusted association between level of CoC and intention of pregnancy at conception. Likelihoods of using the moderate or highest level of CoC were found to be significantly lower among women who reported their last pregnancy was mistimed or unwanted at conception than women with a wanted pregnancy at conception. This association persisted following multivariate adjustment (see Table 5). In the fully adjusted model, compared to a wanted pregnancy, an unwanted pregnancy was found to be associated with 39% (OR 0.61, 95% CI 0.47–0.78) and 62% (OR 0.38, 95% CI 0.23–0.64) lower likelihoods of the moderate and highest levels of CoC, respectively. A pregnancy that was mistimed at conception was found to be associated with a 31% (OR 0.69, 95% CI 0.49–0.97) reduction in risk of achieving high CoC than a wanted pregnancy. However, the association between mistimed pregnancy and the moderate level of CoC was no longer statistically significant relative to low/no level.

Table 4

Multilevel multinomial logistic regression model with intention at conception of most recent pregnancy as the sole correlate of the level of continuum of maternal healthcare, Bangladesh Demographic and Health Survey, 2014.

	Level of continuum of maternal healthcare
	Moderate vs. low/none level		Highest vs. low/none level
	Odd ratio (95% CI)	p-value	Odd ratio (95% CI)	p-value
Most recent pregnancy intention at conception: wanted conception (ref)
Mistimed conception	0.81 (0.64–1.00)	0.05	0.67 (0.50–0.90)	<0.01
Unwanted conception	0.42 (0.33–0.54)	<0.01	0.25 (0.16–0.38)	<0.01
Constant	1.10 (0.97–1.25)	0.12	0.37 (0.31–0.44)	<0.01
Random effectsa
Cluster-level variance (SE)b	1.51 (0.17)***
Log-likelihood for fixed effects to random effects model	1114.94***
Log-likelihood ratio test for the null model to the random effects model (chi-square)c	74.06***

aWe assume that the within cluster-level random effects are equal for the ‘moderate’ and ‘highest’ levels; therefore, only between cluster-level variance estimates are reported.

bSignificance of random effects evaluated by comparing the model with a similar one in which random effects were constrained to zero.

cCompared to the null model with no-covariates.

***p<0.01.

Table 5

Odd ratios from multilevel multinomial logistic regression to assess the association between level of continuum of care and women’s pregnancy intention at conception, adjusting for individual-, household-, and community-level factors, Bangladesh Demographic and Health Survey, 2014.

	Level of continuum of maternal healthcare
	Moderate level vs. low/none level		Highest vs. low/none level
	Odd ratio (95% CI)	p-value	Odd ratio (95% CI)	p-value
Most recent pregnancy intention at conception: wanted conception (ref)
Mistimed conception	0.85 (0.68–1.06)	0.15	0.69 (0.49–0.97)	<0.05
Unwanted conception	0.61 (0.47–0.78)	<0.01	0.38 (0.23–0.64)	<0.01
Women’s age at birth of their last child: ≤19 years (ref)
20–34 years	0.92 (0.77–1.10)	0.34	1.00 (0.76–1.33)	0.99
≥35 years	1.18 (0.79–1.76)	0.43	1.73 (0.83–3.58)	0.14
Decision-making about women’s healthcare: women involved (ref)
Women not involved	1.08 (0.93–1.26)	0.30	1.30 (1.00–1.68)	<0.05
Women’s education: illiterate (ref)
Primary2	1.40 (1.08–1.82)	<0.01	2.36 (1.14–4.86)	<0.05
Secondary3	1.84 (1.39–2.43)	<0.01	4.21 (2.04–8.71)	<0.01
Higher4	2.86 (1.87–4.35)	<0.01	9.85 (4.35–22.34)	<0.01
Partner’s education: illiterate (ref)
Primary2	1.13 (0.91–1.39)	0.27	0.94 (0.60–1.48)	0.79
Secondary3	1.38 (1.09–1.74)	<0.01	1.47 (0.93–2.32)	0.09
Higher4	2.24 (1.59–3.17)	<0.01	2.48 (1.43–4.30)	<0.01
Partner’s occupation: agricultural workera (ref)
Labourerb	1.09 (0.89–1.32)	0.41	1.20 (0.82–1.76)	0.35
Services	1.16 (0.73–1.83)	0.53	1.74 (0.95–3.21)	0.08
Business	1.35 (1.06–1.71)	<0.01	1.50 (1.00–2.28)	<0.05
Other	1.29 (0.83–2.01)	0.26	1.02 (0.43–2.42)	0.96
Exposure to mass media: not exposed (ref)
Moderately exposed	1.13 (0.94–1.36)	0.21	2.28 (1.55–3.35)	<0.01
Highly exposed	1.43 (1.05–1.94)	<0.05	2.90 (1.77–4.74)	<0.01
Wealth status: poorest (ref)
Poorer	1.49 (1.18–1.88)	<0.01	1.53 (0.79–2.96)	0.21
Middle	1.50 (1.17–1.92)	<0.01	2.10 (1.11–3.98)	<0.05
Richer	1.88 (1.41–2.51)	<0.01	3.34 (1.77–6.28)	<0.01
Richest	2.52 (1.81–3.51)	<0.01	5.19 (2.60–10.34)	<0.01
Place of residence: urban (ref)
Rural	1.10 (0.92–1.32)	0.32	1.04 (0.76–1.41)	0.82
Region of residence: Barisal (ref)
Chittagong	0.91 (0.70–1.19)	0.49	0.94 (0.58–1.52)	0.80
Dhaka	0.93 (0.70–1.22)	0.59	1.19 (0.73–1.92)	0.48
Khulna	1.37 (1.04–1.81)	<0.05	1.41 (0.83–2.37)	0.20
Rajshahi	1.28 (0.96–1.70)	0.09	1.46 (0.88–2.42)	0.14
Rangpur	1.41 (1.05–1.89)	<0.05	1.56 (0.91–2.66)	0.11
Sylhet	0.90 (0.66–1.23)	0.51	1.31 (0.77–2.25)	0.32
Community-level illiteracy: low (<25%, ref)
Moderate (25–50%)	1.14 (0.96–1.36)	0.15	0.98 (0.75–1.27)	0.86
High (50%)	0.96 (0.71–1.29)	0.76	0.97 (0.57–1.64)	0.91
Community-level four or more antenatal care services non-use: Low (0–49%; ref)
High (50–100%)	0.55 (0.43–0.70)	<0.01	0.26 (0.20–0.35)	<0.01
Community-level professional delivery care services non-use*: Low (0–49%; ref)
High (50–100%)	0.26 (0.22–0.32)	<0.01	0.46 (0.32–0.64)	<0.01
Community-level not appropriate postnatal care visits: Low (0–49%; ref)
High (50–100%)	0.79 (0.59–1.05)	0.09	0.22 (0.15–0.30)	<0.01
Random effectsa
Cluster-level variance (SE)b	0.03 (0.04)***
Log-likelihood for fixed effects to random effects model	609.44***
Log-likelihood ratio test for the null model to random effects model (chi-square)c	1624.59***

*Professional personnel to provide delivery care include qualified doctors, nurses, midwives, paramedics, family welfare visitors, and community skilled birth attendants. aWe assume that the within cluster-level random effects are equal for the ‘moderate’ and ‘highest’ levels; therefore, only between cluster-level variance estimates are reported.

bSignificance of random effects evaluated by comparing the model with a similar one in which random effects were constrained to zero.

cCompared to the null model with no-covariates.

***p<0.01.

Of the different factors adjusted, women’s involvement in their healthcare decision-making and increased years of education were found to be associated with higher likelihoods of achieving the highest and the moderate levels of CoC than the lower level of these corresponding characteristics. Similarly, at the household-level, partner’s engagement in business (reference group was agricultural worker), higher exposure to mass media (reference group was not being exposed to mass media), and improved household wealth quintile (reference group was household poorest wealth quintile) were the factors that were found to be associated with increased likelihoods of achieving the highest and the moderate levels of CoC. Turning to the community-level factors, we found up to an 82% lower likelihood of achieving the highest level of CoC among women residing in a community where more than 50% of women did not use four or more ANC visits, SBA, or PNC visits than women residing in a community with at least 50% use of these services. These reduced odds were approximately similar for the moderate level of CoC.

Discussion

This study provides a systematic analysis of CoC completion through to PNC, based on the most recent nationally representative household survey data in Bangladesh. Overall, we found that 12.5% of women had CoC across all three services. This figure declined by around half (5.6%) among women whose pregnancies were unwanted at conception. The highest dropout from CoC was found in the translation of at least one ANC visit to four or more ANC visits. We found these low levels of CoC, despite Bangladesh being acclaimed globally for its increase of ANC and SBA in accordance with the MDGs. Therefore, the current progress of maternal healthcare service use will minimally contribute to the SDG targets of universal healthcare coverage for optimizing maternal and newborn adverse health outcomes. Importantly, this challenge of improving CoC cannot be overcome with the current supply-oriented and disjointed healthcare provision in Bangladesh. Moreover, our study indicates that a high occurrence of pregnancies that are unintended at conception (at around 26%) is a barrier to CoC and therefore will further challenge Bangladesh’s aim of achieving its SDG targets. The findings suggest a need for policies to integrate healthcare services with a particular focus on women who have pregnancies that are unintended at conception.

This study found a dropout rate of around 80% for women who started with one ANC visit (65%) to completion of CoC (12.5%). This dropout rate was higher than in Pakistan (72.6% dropout, 12% CoC) [18], the average of three South Asian countries (Bangladesh, Nepal, Pakistan; 68.67% dropout, 24.5% CoC) [12], and in Cambodia (50% dropout, 60% CoC) [19]. However, Bangladesh had a lower dropout rate than several countries, including the average of six sub-Saharan African countries (Ethiopia, Malawi, Rwanda, Senegal, Tanzania, and Uganda; 83.9% dropout, 13.9% CoC) [12], Ghana (92% dropout, 8% CoC) [20], and Lao PDR (91.34% dropout, 6.8% CoC) [39]. Importantly, such differences of CoC between Bangladesh and other countries are reflective of the differences in uptake of individual maternal healthcare services use, including ANC, SBA, and PNC. These services are always found to be lower in Bangladesh than in other Asian countries (e.g. Pakistan, Nepal, and Cambodia) [19,42,43], however, higher than African countries [12]. Further increases in the dropout of completion of CoC were found among women who had unwanted (62%) and mistimed (31%) pregnancies at conception than wanted pregnancies. Previous studies in Bangladesh found likelihoods of using at least four ANC visits, SBA, and first PNC visit within 24 hours of delivery declined by around 33%, 35%, and 42%, respectively, among unwanted pregnancies than wanted pregnancies [27,29,30].

The results point to several pathways through which pregnancy intention can influence CoC completion; however, the majority involve healthcare system level factors, individual and household level factors (such as lower education and household wealth quintile) associated with unintended pregnancy, and women’s characteristics following unintended conception. In its current health care structure in Bangladesh, family planning services and maternal healthcare services are provided separately through the Directorate General of Family Planning (DGFP) and the Directorate General of Health Services (DGHS), respectively, along with focus on supplying these services rather than considering demand-side factors (such as behavioural change of people in the community) [44,45]. Inadequate or no counselling by family planning providers (which plays a role in the number of unintended pregnancies) and their current minimal (or lack of) coordination with maternal healthcare service providers could, therefore, contribute to lower CoC among women with unintended conceptions. Moreover, the occurrence of unintended conceptions are usually found to be higher among disadvantaged women in terms of individual- and household-level factors and women residing in disadvantaged communities (including those with lower education and higher poverty) [46,47]. These factors have been found to be independently associated with the lower use of ANC, SBA, and PNC in Bangladesh and LMICs [48-50]. Post-conception characteristics, including household-level response (family members and partner opposition) and the psychological consequences following an unintended pregnancy (depression and anxiety) are other possible factors impacting on CoC among women with unintended pregnancies [21]. Accessing all components of CoC requires multiple visits to healthcare facilities which requires additional travel money and more assistance with visiting healthcare facilities. Previous studies in Bangladesh [27,29,30] and LMICs [21,51] reported these as significant deterrents to the use of ANC, SBA, and PNC following the occurrence of an unintended pregnancy. Late pregnancy detection and decisions around pregnancy termination may be other important reasons for lower CoC among women with an unintended pregnancy [21]. As these women have fewer ANC visits, there are also missed opportunities for healthcare providers to motivate them to access SBA and PNC [30]. Consequently, the progress in CoC in general, and among unintended pregnancies in particular, requires policies that address issues related to the country’s healthcare system (e.g. the disjointed approach to providing healthcare), women’s place within society, as well as sociodemographic and post-conception characteristics.

The Alma-Ata Declaration of Health in 1978 by the WHO has emerged as a major milestone in Bangladeshi healthcare policy to ensure primary healthcare at the grass-roots level [52]. Therefore, the focus has always been given to making a pluralistic healthcare system [53], developing community-based approaches by ensuring partnerships with non-governmental organizations [54], the establishment of community clinics [27], and rounds of sector-wise health and nutritional programs [55]. Consequently, significant improvements have been reported in the provision of reproductive and maternal healthcare services [56] and family planning services [54]. However, what is still lacking are policies to ensure the continuity of using family planning services to PNC services through integrated healthcare services. For instance, many women use family planning services, however, drop out without accessing maternal healthcare services [57,58]. Importantly, a significant number of women use maternal healthcare services when required because of complications (need-based) rather than access all services for better pregnancy outcomes (merit-based) [57,58]. Moreover, a significant proportion of women who start ANC services drop out later from using the recommended number of ANC, SBA, and PNC services [57]. Therefore, integration of family planning services with maternal healthcare services and among several streams of maternal healthcare services along with strong monitoring for continuity could be effective options to reduce this dropout after starting access of services. However, such integration and monitoring for continuity in Bangladesh requires several challenges to be addressed at the national level.

First, there is an acute shortage of human resources in healthcare sectors in Bangladesh (with around 800,000 workers), which has existed for decades [59], and poor sustainability and retention of community health workers (current dropout rate is 12–44%) [60,61]. Ensuring health workers in the primary level stay is therefore a challenge for the coordination of services at home and in the community and monitoring for continuity. Previous studies in Bangladesh [62] and Tanzania [63] found these are significant for initiating reproductive healthcare services, as well as continuing services use through to PNC. Second, there is an inadequate focus on the quality of services given, which is now recognized as an important deterrent to CoC [64,65]. This might be a reason for the high dropout after having at least one ANC visit that was found in this study. Third, ignorance of the demand-side factors in healthcare policies, including community support systems, mobilization of community participation, and people’s behavioural change [54,62]. Establishing public healthcare facilities through supply-oriented healthcare policies, even at the community-level, would therefore have little effect on CoC completion. The roots for this could be the conservative cultural environment, cultural norms, and traditional beliefs in Bangladesh [66,67]. For instance, the patriarchal system in Bangladesh often restricts women’s decision-making capability and the economic affordability of visiting healthcare facilities. Besides, cultural norms within Islam (the religion of more than 90% of the total population in Bangladesh) do not permit women to leave their homes while pregnant, with pregnant women believed to be highly vulnerable to evil spirits (one reason why their mobility is restricted, particularly during delivery and soon after childbirth) [68]. These practices are even more prevalent among women with disadvantaged individual-, household-, and community-level characteristics (e.g. lower education, socioeconomic status) [27,67,69]. These factors are also associated with lower contraception use (which might be a major factor for the higher occurrence of unintended pregnancy) [28] and even more restriction of women’s movement during pregnancy, predominantly in early pregnancy and in the early postpartum period [70,71]. Essentially, these are the stages where this study found higher dropout from CoC, findings that are consistent with the literature on LMICs [12,20,22,24,27,31,39]. Therefore, further progress in achieving CoC for Bangladeshi women, particularly among women with unintended pregnancies, requires the availability of healthcare services, consideration of behavioural change of people in the community, and meeting the current shortage of workforce.

Moreover, such supply-oriented policies and disjointed healthcare services in Bangladesh make maternal healthcare services mainly visit-based (women and newborns visit health facilities to receive services), even for healthcare facilities available at the community-level. However, previous studies in Bangladesh have found that in order to increase the use of maternal, newborn and child health, a visit-based approach is less effective than a services-based approach (healthcare providers visit women’s homes and provide services there) [62,66]. Similarly, a study in Ghana found higher CoC with a services-based approach (10.3%) rather than a visit-based approach (7.9%) [22]. Integration of family planning workers with mainstream maternal healthcare services could help Bangladesh to provide a services-based like maternal healthcare service. For instance, in an integrated service, family planning workers would be given additional responsibility for educating and monitoring maternal healthcare service use along with their current responsibility of family planning and distributing contraception. Currently, there is no governmental effort at the field level (in which family planning operates) to ensure maternal healthcare services use or monitoring of service use [31]. Therefore, this integration could contribute to increasing average maternal healthcare services use along with continuity. It also reduces the differences in maternal healthcare services use that occur because of the sociodemographic characteristics of women and their partners along with demand-side factors (being disadvantaged in any of these characteristics is found to be associated with a higher occurrence of unintended pregnancy) [46,47]. Therefore, family planning workers’ responsibility to focus on these groups and ensure proper education about the importance of maternal healthcare services use and monitoring for continuity is important for improving CoC in the country.

There were several limitations to our study. First, as it was a cross-sectional study, the relationship between pregnancy intention at conception and CoC is correlational only. Second, all components of CoC and pregnancy intention data were collected based on women’s retrospective (recall) responses, which may be subject to reporting errors (e.g. recalling the exact number and timing of healthcare services use). Moreover, women’s feelings about pregnancy can change in many ways over a 3 year period, which could lead to considerable ambivalence in reporting pregnancy status [37]. For instance, a woman’s opinion about pregnancy and timing might change after the birth of her child; therefore, intentional pregnancies may be reported as “unwanted” or “unintended”, and unintentional pregnancies may be reported as “wanted” [37]. An intended pregnancy could also be reported as “unintended” (or “unwanted”) if the outcome of the pregnancy (the child’s sex) is the opposite of the parents’ desire for a child of a particular sex. For instance, an Indian study reported a higher likelihood of unintended pregnancy among women whose child was female than male [72], though, no study in LMICs reported the age-specific changes of unintended pregnancy reporting. This could be because of parents’ higher preference of a male than a female child- a context which is also common in Bangladesh [11]. Moreover, retrospective responses make distinguishing between mistimed and unwanted conception challenging because of the potential nuance of stating a pregnancy is mistimed or unwanted following upto 3 years of conception occurred due to elapsed time. Third, we did not have data on the distance of the healthcare facilities from the women’s homes or the quality of services, which are important for CoC in visit-based services in Bangladesh. Moreover, the timing of pregnancy detection and the first ANC visit are important for CoC although they were not adjusted for because of lack of data. Moreover, as we weighted the forms of maternal healthcare services equally, it is important to acknowledge that the risk factors for dropout at each stage may differ. However, previous studies of this kind in Bangladesh found around consistent results in respect to the determinants of maternal healthcare services use with the current study [27,29,30]. Despite these limitations, the current study is the first in Bangladesh and LMICs that assesses the effect of unintended pregnancy at conception on CoC completion based on data from a unique representative survey. Use of the CoC framework, appropriate statistical adjustment for the survey design, and adjustment of a comprehensive range of factors enabled us to make the strongest recommendations for future policies in efforts to improve maternal health.

Conclusion

This study highlighted the importance of an integrated approach to maternal healthcare services, from pregnancy to the early post-partum stage. Findings suggest only 12.5% of women had a CoC in their last pregnancy, which decreased by around half (5.6%) among women who had a pregnancy that was unwanted at conception. Higher dropout from the CoC was reported in translating at least one to four or more ANC visits. Following adjustment for a comprehensive range of individual-, household-, and community-level factors, unintended pregnancy at conception was found to be strongly associated with the level of CoC attained. Different individual-, household-, and community-level factors, including lower women’s education, lower household socio-economic status, and lower use of ANC and delivery care services at the community level were also found to be influential determinants of lowering CoC. To ensure CoC, particularly among women who have unintended conceptions, policies are required to enhance healthcare services at the community level, along with the focus on demand- and supply-side interventions together. It is also important for policies to integrate family planning and maternal healthcare services.

20 Jul 2020

PONE-D-20-07773

Assessing the effect of pregnancy intention at conception on the continuum of care in maternal healthcare services use in Bangladesh: evidence from a nationally representative cross-sectional survey

PLOS ONE

Dear Dr. Khan,

Thank you for submitting your manuscript to PLOS ONE. After careful consideration, we feel that it has merit but does not fully meet PLOS ONE’s publication criteria as it currently stands. Therefore, we invite you to submit a revised version of the manuscript that addresses the points raised during the review process.

Both reviewers highlight a number of minor suggestions and clarifications.  Please review and respond ​to the suggestions made.  Please ensure that a flow-chart is included per the second reviewer's recommendation.

Additionally, there are some minor editorial and grammar issues.  PLOS ONE does not support editorial services so your manuscript should be reviewed and edited prior to re-submission to address occasional English language mistakes.

Also ensure that decimal points are consistent throughout the manuscript.  At different points, numbers are presented with zero, one, or two decimal points in the text and tables. Unless using language such as "about 15%", please ensure consistency.

Please submit your revised manuscript by Sep 03 2020 11:59PM. If you will need more time than this to complete your revisions, please reply to this message or contact the journal office at plosone@plos.org. When you're ready to submit your revision, log on to https://www.editorialmanager.com/pone/ and select the 'Submissions Needing Revision' folder to locate your manuscript file.

Please include the following items when submitting your revised manuscript:

A rebuttal letter that responds to each point raised by the academic editor and reviewer(s). You should upload this letter as a separate file labeled 'Response to Reviewers'.

A marked-up copy of your manuscript that highlights changes made to the original version. You should upload this as a separate file labeled 'Revised Manuscript with Track Changes'.

An unmarked version of your revised paper without tracked changes. You should upload this as a separate file labeled 'Manuscript'.

If you would like to make changes to your financial disclosure, please include your updated statement in your cover letter. Guidelines for resubmitting your figure files are available below the reviewer comments at the end of this letter.

If applicable, we recommend that you deposit your laboratory protocols in protocols.io to enhance the reproducibility of your results. Protocols.io assigns your protocol its own identifier (DOI) so that it can be cited independently in the future. For instructions see: http://journals.plos.org/plosone/s/submission-guidelines#loc-laboratory-protocols

We look forward to receiving your revised manuscript.

Kind regards,

Linnea A Zimmerman, Ph.D, MPH

Academic Editor

PLOS ONE

Journal Requirements:

When submitting your revision, we need you to address these additional requirements.

1. Please ensure that your manuscript meets PLOS ONE's style requirements, including those for file naming. The PLOS ONE style templates can be found at

https://journals.plos.org/plosone/s/file?id=wjVg/PLOSOne_formatting_sample_main_body.pdf and

https://journals.plos.org/plosone/s/file?id=ba62/PLOSOne_formatting_sample_title_authors_affiliations.pdf

2. Thank you for stating the following in the Acknowledgments Section of your manuscript:

'Funding: The first author is supported by a PhD research scholarship from The University of Newcastle, Australia. Dr Melissa L. Harris is supported by an Australian Research Council Discovery Early Career Researcher Award.'

We note that you have provided funding information that is not currently declared in your Funding Statement. However, funding information should not appear in the Acknowledgments section or other areas of your manuscript. We will only publish funding information present in the Funding Statement section of the online submission form.

a. Please remove any funding-related text from the manuscript and let us know how you would like to update your Funding Statement. Currently, your Funding Statement reads as follows:

'The author(s) received no specific funding for this work.'

b. Please include your amended statements within your cover letter; we will change the online submission form on your behalf.

[Note: HTML markup is below. Please do not edit.]

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #1: Yes

Reviewer #2: Yes

**********

2. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #1: Yes

Reviewer #2: Yes

**********

3. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #1: No

Reviewer #2: Yes

**********

4. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #1: Yes

Reviewer #2: Yes

**********

5. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #1: This paper investigates the association between having an unintended birth and maternal use of healthcare – specifically continuity of care during and after pregnancy – among women in Bangladesh. The strengths of the paper lie in the large sample, completeness of the measure of unintended births (unwanted and mistimed), and the clear writing. I also want to commend the authors on Figure 1, which provides a very nice depiction of the percentages of women who completed CoC or dropped out. Despite these strengths, some areas of the paper would benefit from additional revision. I include several questions and suggestions below in the aim of strengthening the paper.

One limitation of the current study is that the authors rely solely on retrospective data. This needs to be clarified when describing the data (pg. 8). Additionally, there is a robust literature on the biases inherent in retrospective measures that should be considered in the discussion section (pg. 17) to ascertain what type of bias may have arisen due to this measurement strategy. Many studies from a variety of global settings have found that shifts in women’s reports of intentions are predictable – and tied to intendedness - rather than random as is suggested in the paper. Another potential source of bias is selection into unintended pregnancy. How might this influence the results? Other studies have used analytic strategies like propensity score weighting to control for selection into unintended pregnancy. Although I do not believe that is necessarily required here, the authors should at least acknowledge this source of bias and discuss how it might influence their results.

In the Outcome Variable section, it is important to include the actual survey questions used to create the CoC measure – similar to what is done in the Explanatory variables section (see pg. 8). This would clear up several of my questions about the items used to construct this measure, such as how a skilled birth attendant is defined.

In the literature review, the unintended fertility rate in Bangladesh is described as being 46% of all pregnancies (pg. 6), but the current study finds that only 26% of births are unintended. Please include a discussion of why you believe there is such a large difference between these percentages (selection? Measuring pregnancies vs. births?), and describe how this may have affected your results and therefore your conclusions.

The authors claim that they are able to make the “strongest recommendations for future policies” (pg. 18), but from my reading, many of the policy recommendations are rather vague. This is the area of the discussion that needs the most attention. For example, one of the authors’ recommendations is to create policies that “integrate healthcare services” (pg. 14). Can the authors provide some concrete suggestions on how Bangladesh – or LMICs more broadly – might do this and/or what these policies might look like? Similar questions apply to the recommendation for policies to “consider a range of factors…” (pg. 15). Addressing these questions and providing concrete policy recommendations would greatly strengthen the paper and its potential to contribute to positive changes in maternal health services and utilization in Bangladesh.

One of the study’s findings is that Bangladesh falls in the middle in terms of its dropout rate. I’d like to see the authors offer some explanations of why they think this is the case. Also, what could Bangladesh do better to emulate the South Asian countries mentioned on page 14?

Cultural factors are mentioned as a barrier to increasing CoC among women in Bangladesh. How might these be addressed in policy creation and implementation? Addressing this would strengthen the discussion on page 16.

Reviewer #2: This study investigates an important issue – continuum of pregnancy-related care (CoC) in low and lower-middle income countries and the impact that unintended pregnancy has on receipt of care. This work is particularly important for recognizing the potential disparities and opportunities for intervention in improving receipt of recommended maternal health services, particularly among women who experience unintended pregnancy. Below are additional comments that hopefully will improve the contribution that this article can make, mostly in terms of the presentation and framing of the dependent variable and reporting results. Overall, this is a thoughtful article, which extends the current body of maternity care research beyond simple measures of ANC, SBA and PNC.

Abstract

1. Methods: It would be helpful if the authors could note the data source (DHS Bangladesh 2014) in the abstract, so readers understand the type of dataset used for the analysis.

2. Results: The language about “maintaining” vs. “achieving” CoC is confusing. Per the definition of the CoC (high, moderate, and low/none), the measure is complete when all components of pregnancy-related care are considered. Use of the word “maintain” here makes it seem as though the study examines CoC as a measure over time, instead of within the context of a single index pregnancy. Suggest to reword any use of “maintain” as “achieve/attain” throughout the abstract and manuscript for clarity.

Methods

1. Data:

a. The subsample of women included in this analysis must have had a livebirth in three years prior to the survey; responded to the question on maternity care, and reported retrospective pregnancy intention. What proportion of women aged 15-49 had a birth in the prior three years, but did not have complete measures on all three components of CoC? It might be interesting, or at least provide some insight, if the authors could note which components of those eligibility criteria resulted in exclusion (i.e. some kind of flowchart with numbers of how they reached the final sample, for reference). Unclear if the jump from 17,863 to 4,493 women was due to a relatively low number of women who had a birth within the three years prior; incomplete data on maternal health services; or no pregnancy intention reported at conception, which are relevant for understanding the broader study sample from which this analytic sample was derived.

2. Outcome variable:

a. The authors indicate that they followed the WHO’s 2014 guidelines for maternity care, including receipt of at least four ANC visits. They also note that this changed to eight in 2016. Why did the authors choose to assess 4+ visits instead of 8+ visits, given the change in recommendations? Were any sensitivity analyses conducted to see how this altered the distribution of outcomes? I suspect far fewer women would have achieved a “high” level of continuum of care, but might provide an interesting comparison at least to address in the discussion.

b. While the authors note that their development of the outcome variable aligns with the WHO’s guidance, I would like to know more about their decision to “weight” all three components of care equally. For example, is a woman who received ANC and PNC, but no SBA likely to encounter different health risks than women who received a different combination of care (e.g., ANC, SBA, no PNC)? Do the authors anticipate that this would have altered the outcome?

c. Throughout the paper, the authors use the terminology “dropout” to describe women who received early components of CoC, but did not receive later components. This language, which reframes the dependent variable to consider each “stage” of the CoC, should be described in this part of the methods section. As a reviewer, it was not clear in the methods that the analysis would include more nuanced assessments of the CoC, other than the three-level categorical variable that is described.

3. Explanatory variables

a. It would be helpful if the authors clearly stated that “women’s intention at conception” was the woman’s fertility/pregnancy intention. The term “intention at conception” is a bit abstract; “index pregnancy intention” may be a better term for this concept throughout the manuscript.

b. Authors should also indicate that this is a retrospectively reported measure, one that may be reporting on a pregnancy from up to three years prior. Limitations of such a measure should be noted.

c. Were women who responded “Don’t know” to the question of “whether they wanted to get pregnant at the time of conception” excluded from the analysis or coded as “no”. Please clarify in this section, as ambivalence about pregnancy intentions may also play a role in women’s receipt of CoC and could potentially be explored separately (if distinct from the women included in this analysis).

d. The authors note, “The factors were selected by using forward regression analysis on the list of factors found in our literature search.” Could they please reference the specific literature from which they determined the initial set of covariates for exploration? Other work on this topic that has found significant associations or conceptual models that indicate the relevance of these factors? Particularly for the partner factors, such as occupation?

Results

1. Table 2: I find the presentation of data in Table 2 confusing, particularly because the authors use the terminology “dropout” to describe women’s trajectories through the CoC with ANC as the starting point. As it is currently presented, the authors have provided row totals by delivery and postnatal care status. To improve clarity:

a. Keep the current table format and use column totals, but simplify the row labels. For example, instead of “Delivery care” the section header would be “Delivery by skilled birth attendant”, and the rows beneath it would be “Yes” and “No”. The two rows in that section would sum to 100% for each the ANC <4 and ANC 4+ sub-populations represented by each of the columns. This would provide a clearer narrative to compare women who began the CoC trajectory with 4+ ANC visits, relative to those who did not.

b. If the authors do not want to explicitly compare these two populations (by ANC status), they should simplify the table by just including overall proportions of ANC, SBA, and PNC care for the entire sample.

c. I also suggest removing the “level of CoC” from each cell in the table, as this is confusing. The second portion of the table that lays out the levels of the continuum is a much clearer presentation. Once this table is revised, the prose explaining the descriptive findings of Table 2 and this part of the analysis should be expanded.

2. Figure 1: This figure includes a lot of important and helpful information for understanding how women “drop out” from receiving one component of the CoC to another; however, there are a lot of words included in each of the transitional squares and arrows. If possible, simplify the figure, so the proportions of women remaining at each stage are clear and do not get lost among all of the words.

a. Check grammar throughout the paragraph of results that outlines this figure.

b. The authors conclude the paragraph with, “These trends of dropout were similar across pregnancies that were wanted, mistimed and unwanted at conception.” I would suggest emphasizing that the dropout was more pronounced for pregnancies that were unwanted, as illustrated in the figure.

3. Table 5: Prose needs to be reviewed. For example, the authors state, “A pregnancy that was mistimed at conception was found to be 31% (RR 0.69, 95% CI 0.49-0.97) preventive to the highest level of CoC than a wanted pregnancy.” Should be restated in terms of being a reduced risk of achieving or attaining “high CoC” relative to women with wanted pregnancy. Similarly, in stating, “However, the association between mistimed pregnancy and the moderate level of CoC was no longer statistically significant.” The reference for the multinomial model—i.e. relative to no/low level—should be included.

Discussion

1. Overall, this discussion section feels incredibly disjointed and disconnected from the paper. It needs substantial work to orient the findings of the paper within the broader context of this research.

a. This is particularly needed for those results related to the household and community level factors that were found to have significant associations with women’s risk of achieving CoC. For example, the findings of this study could be more explicitly woven into or contrasted with the findings of other studies noted in the second paragraph.

2. There are some incomplete sentences in the discussion that make it a bit confusing to follow. Suggest a thorough re-read with a focus on grammar.

3. In acknowledging the limitations of the retrospective pregnancy intention measure, used as the primary explanatory variable in this analysis, the authors state, that “any such bias is likely to be random.” I don’t think I agree with this statement, particularly because the authors do not adjust for time since pregnancy in their analysis.

a. For example, women with index pregnancies occurring earlier than others may be more likely to retrospectively report the pregnancy as intended, now that they have a young child. In contrast, women who have a newborn or infant and are reporting on a much more recent pregnancy may be more likely to report that pregnancy as initially unintended, since the event is more proximate. I think the authors should expand here or in the methods section about how and why this measure is considered valid, despite some of these limitations.

Conclusion

1. In the conclusion, the authors claim, “To ensure CoC, particularly among women who have unintended conceptions, early detection of pregnancy is required, as well as policies to enhance healthcare services at the community level, along with the focus on demand and supply-side interventions together.” However, in the discussion, they emphasize a lack of data about the detection of the pregnancy to be a limitation. How are the authors making a claim about the importance of “early detection of pregnancy”, if this was not central to or even captured in the survey or analysis?

2. The claim that, “Following adjustment of a comprehensive range of individual- , household-, and community-level factors, unintended pregnancy at conception was found to be the strongest preventive predictor of CoC,” seems unsubstantiated by the findings. While pregnancy intention was strongly associated with level of CoC attained, many other factors at the household-level, for example wealth, were found to be the strongest predictors of achieving highest CoC (i.e. RR=3.34 for richest vs. poorest women). Please rephrase the conclusion to reflect the findings more accurately.

**********

6. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: No

Reviewer #2: No

[NOTE: If reviewer comments were submitted as an attachment file, they will be attached to this email and accessible via the submission site. Please log into your account, locate the manuscript record, and check for the action link "View Attachments". If this link does not appear, there are no attachment files.]

While revising your submission, please upload your figure files to the Preflight Analysis and Conversion Engine (PACE) digital diagnostic tool, https://pacev2.apexcovantage.com/. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Registration is free. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email PLOS at figures@plos.org. Please note that Supporting Information files do not need this step.

18 Aug 2020

Please see the relevant attached file- one file added there to respond the reviewers' comments.

Submitted filename: Response to Reviewers.docx

Click here for additional data file.

9 Oct 2020

PONE-D-20-07773R1

Assessing the effect of pregnancy intention at conception on the continuum of care in maternal healthcare services use in Bangladesh: evidence from a nationally representative cross-sectional survey

PLOS ONE

Dear Dr. Khan,

Thank you for submitting your manuscript to PLOS ONE. After careful consideration, we feel that it has merit but does not fully meet PLOS ONE’s publication criteria as it currently stands. Therefore, we invite you to submit a revised version of the manuscript that addresses the points raised during the review process.

While one of the reviewer's found the previous edits acceptable, the second felt that there were minor revisions that were still required, with which the editor agrees.  Please review the comments attached.

Particular attention should be paid to ensuring that the concerns about the proper interpretation and use of literature are addressed.  Secondly, the reviewer suggests removing the following language from the limitations section "However,  any such bias is likely to be random".  The editor strongly agrees.  It is not clear what empirical evidence  suggests this would be the case and theoretically, this assumption seems tenuous as education, age, parity, wealth, and autonomy may all be related to both experiencing an unwanted/unintended pregnancy and how the question is understood and retrospectively  reported.

Please submit your revised manuscript by Nov 23 2020 11:59PM. If you will need more time than this to complete your revisions, please reply to this message or contact the journal office at plosone@plos.org. When you're ready to submit your revision, log on to https://www.editorialmanager.com/pone/ and select the 'Submissions Needing Revision' folder to locate your manuscript file.

Please include the following items when submitting your revised manuscript:

A rebuttal letter that responds to each point raised by the academic editor and reviewer(s). You should upload this letter as a separate file labeled 'Response to Reviewers'.

A marked-up copy of your manuscript that highlights changes made to the original version. You should upload this as a separate file labeled 'Revised Manuscript with Track Changes'.

An unmarked version of your revised paper without tracked changes. You should upload this as a separate file labeled 'Manuscript'.

If you would like to make changes to your financial disclosure, please include your updated statement in your cover letter. Guidelines for resubmitting your figure files are available below the reviewer comments at the end of this letter.

If applicable, we recommend that you deposit your laboratory protocols in protocols.io to enhance the reproducibility of your results. Protocols.io assigns your protocol its own identifier (DOI) so that it can be cited independently in the future. For instructions see: http://journals.plos.org/plosone/s/submission-guidelines#loc-laboratory-protocols

We look forward to receiving your revised manuscript.

Kind regards,

Linnea A Zimmerman, Ph.D, MPH

Academic Editor

PLOS ONE

[Note: HTML markup is below. Please do not edit.]

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. If the authors have adequately addressed your comments raised in a previous round of review and you feel that this manuscript is now acceptable for publication, you may indicate that here to bypass the “Comments to the Author” section, enter your conflict of interest statement in the “Confidential to Editor” section, and submit your "Accept" recommendation.

Reviewer #1: All comments have been addressed

Reviewer #2: (No Response)

**********

2. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #1: Yes

Reviewer #2: Yes

**********

3. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #1: Yes

Reviewer #2: Yes

**********

4. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #1: Yes

Reviewer #2: Yes

**********

5. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #1: Yes

Reviewer #2: Yes

**********

6. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #1: (No Response)

Reviewer #2: Overall, the authors have done a good job revising the manuscript according to reviewers’ feedback and have addressed the vast majority of my initial concerns. I have a few final comments/suggestions that I hope will improve the quality of this work.

One remaining concern I have is about the framing and acknowledgement of limitations of the main exposure variable (unintended pregnancy), which both reviewers flagged as an issue in the original submission. In the revised submission, the authors have tried to integrate such limitations into the methods and limitations, but I still have a few concerns with how this is addressed.

In the methods, they state, “Importantly, these questions were asked together with contraception use or non-use at that time. Therefore, potential ambivalence associated with a retrospective approach to data collection was reduced greatly as women were able to classify more accurately whether this pregnancy was due to not using contraception because of hoping for a child, contraception failure, or lack of access or use of contraception. The general validity of this measure of data collection and classification at the aggregate level has been repeatedly demonstrated in different settings, including Asian and African countries [33-37].”

• I’m not sure I follow the rationale about how questions about pregnancy intention being asked in the context of contraceptive use would actually improve their validity as measures of pregnancy intendedness. In fact, multiple studies have found that pregnancy intention is inconsistently related to contraceptive use—that is, women may not be using contraception consistently aligned with whether and when they want to become pregnant.

• Additionally, the five articles referenced do not accurately address the issue of validity and potential limitations of the pregnancy intention measure. Other articles that specifically examine the validity of different measures of pregnancy intention should be included instead.

o The NIPORT (2016) report and the Hubacher (2008) article referenced do not include any assessment of validity of measures of pregnancy intention (to my knowledge) and should be removed.

o The Curtis (2011) article referenced includes a much more nuanced discussion of the ways that pregnancy intendedness relates to contraceptive use, thus, this should be reflected in the current paper if cited.

Specifically, in the limitations, they state, “An intended pregnancy could also be reported as “unintended” (or “unwanted”) if the outcome of the pregnancy (the child’s sex) is the opposite of the parents’ desire for a child of a particular sex. Moreover, retrospective responses make distinguishing between mistimed and unwanted conception challenging as many women might be unable to determine whether an unintended pregnancy was mistimed or unwanted. However, any such bias is likely to be random.”

• The first sentence: Could the authors expand on or provide literature to justify the mention of sex-specific changes in unintended pregnancy reporting? While valid, it would be helpful to have some literature and/or further explanation for this possible outcome.

• The middle sentence: Could the authors clarify what they mean about how mistimed and unwanted pregnancies may be more difficult due to the retrospective nature of the measure? This feels unclear as currently described. I think the point that they’re trying to make is about the potential nuance of stating a pregnancy is mistimed vs. unwanted many months or years after the pregnancy occurred, simply due to lapsed time, but this is not clearly stated.

• The last sentence: Ending in, “such bias is likely to be random” should be removed, as we do not know this is the case and both reviewers emphasized that this point about measures—especially since this is the main exposure in the analysis—should be framed very clearly as a limitation.

**********

7. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: No

Reviewer #2: No

[NOTE: If reviewer comments were submitted as an attachment file, they will be attached to this email and accessible via the submission site. Please log into your account, locate the manuscript record, and check for the action link "View Attachments". If this link does not appear, there are no attachment files.]

While revising your submission, please upload your figure files to the Preflight Analysis and Conversion Engine (PACE) digital diagnostic tool, https://pacev2.apexcovantage.com/. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Registration is free. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email PLOS at figures@plos.org. Please note that Supporting Information files do not need this step.

10 Oct 2020

A relevant file is attached where we have provided item by item response to the editor and reviewer's comments.

Submitted filename: Response to Editor and Reviewer.docx

Click here for additional data file.

9 Nov 2020

Assessing the effect of pregnancy intention at conception on the continuum of care in maternal healthcare services use in Bangladesh: evidence from a nationally representative cross-sectional survey

PONE-D-20-07773R2

Dear Dr. Khan,

We’re pleased to inform you that your manuscript has been judged scientifically suitable for publication and will be formally accepted for publication once it meets all outstanding technical requirements.

Within one week, you’ll receive an e-mail detailing the required amendments. When these have been addressed, you’ll receive a formal acceptance letter and your manuscript will be scheduled for publication.

An invoice for payment will follow shortly after the formal acceptance. To ensure an efficient process, please log into Editorial Manager at http://www.editorialmanager.com/pone/, click the 'Update My Information' link at the top of the page, and double check that your user information is up-to-date. If you have any billing related questions, please contact our Author Billing department directly at authorbilling@plos.org.

If your institution or institutions have a press office, please notify them about your upcoming paper to help maximize its impact. If they’ll be preparing press materials, please inform our press team as soon as possible -- no later than 48 hours after receiving the formal acceptance. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information, please contact onepress@plos.org.

Kind regards,

Linnea A Zimmerman, Ph.D, MPH

Academic Editor

PLOS ONE

Additional Editor Comments (optional):

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. If the authors have adequately addressed your comments raised in a previous round of review and you feel that this manuscript is now acceptable for publication, you may indicate that here to bypass the “Comments to the Author” section, enter your conflict of interest statement in the “Confidential to Editor” section, and submit your "Accept" recommendation.

Reviewer #2: All comments have been addressed

**********

2. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #2: Yes

**********

3. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #2: Yes

**********

4. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #2: Yes

**********

5. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #2: Yes

**********

6. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #2: No remaining concerns to address. Authors have made appropriate changes to improve the manuscript through their revisions.

**********

7. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #2: No

11 Nov 2020

PONE-D-20-07773R2

Assessing the effect of pregnancy intention at conception on the continuum of care in maternal healthcare services use in Bangladesh: evidence from a nationally representative cross-sectional survey

Dear Dr. Khan:

I'm pleased to inform you that your manuscript has been deemed suitable for publication in PLOS ONE. Congratulations! Your manuscript is now with our production department.

If your institution or institutions have a press office, please let them know about your upcoming paper now to help maximize its impact. If they'll be preparing press materials, please inform our press team within the next 48 hours. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information please contact onepress@plos.org.

If we can help with anything else, please email us at plosone@plos.org.

Thank you for submitting your work to PLOS ONE and supporting open access.

Kind regards,

PLOS ONE Editorial Office Staff

on behalf of

Dr. Linnea A Zimmerman

Academic Editor

PLOS ONE