Institutional delivery and postnatal care utilisation among reproductive-aged women who had completed four or more antenatal care visits in Myanmar: a secondary analysis of 2015–2016 Demographic and Health Survey

Abstract

Objectives

To examine the prevalence of institutional delivery and postnatal care after home delivery and to identify their determinants in Myanmar mothers who received at least four antenatal care visits.

Design

The study used the Myanmar Demographic and Health Survey data (2015–2016), a nationally representative cross-sectional study.

Participants

The study included women aged 15–49 years who had at least one birth within the 5 years preceding the survey and completed four or more antenatal visits.

Outcome measures

Institutional delivery and postnatal care after home delivery were used as outcomes. We used two separate samples, that is, 2099 women for institutional delivery and 380 mothers whose most recent birth was within 2 years before the survey and delivered at home for postnatal care utilisation. We used multivariable binary logistic regression analyses.

Setting

Fourteen states/regions and Nay Pyi Taw Union Territory in Myanmar.

Results

The prevalence of institutional delivery was 54.7% (95% CI: 51.2%, 58.2%) and postnatal care utilisation was 76% (95% CI: 70.2%, 80.9%). Women who lived in urban areas, women who had higher education, women who had higher wealth status, women who had educated husbands and women having their first childbirth were more likely to have institutional delivery than their counterparts. The institutional delivery was lower among women who live in rural areas, poor women and women with husbands who worked in agriculture than their counterparts. Postnatal care utilisation was significantly higher among women living in central plains and coastal regions, women who received all seven components of antenatal care and women who had skilled assistance at birth than their counterparts.

Conclusions

Policymakers should address the identified determinants to improve the service continuum and reduce maternal mortality in Myanmar.

STRENGTHS AND LIMITATIONS OF THIS STUDY.

• Used the Myanmar Demographic and Health Survey data collected in 2015–2016, a nationwide cross-sectional survey with systematic sampling methods and standardised data collection procedures.

• Nationally representative and generalisable to the whole country.

• Some variables supported by the literature cannot be analysed.

• Causality is not possible for the associations found in this study.

Introduction

Maternal mortality is still an ongoing public health issue, especially in low/middle-income countries.¹ It has been estimated that approximately 810 women worldwide died every day from causes related to pregnancy and childbirth in 2017.² Southeast Asia Region accounts for nearly one-third of global maternal and child deaths annually,³ with a maternal mortality ratio (MMR) of 140 per 100 000 live births in 2013.⁴ Myanmar is one of the countries with the highest MMR in the Southeast Asia Region, with reports as high as 250 per 100 000 live births as of 2017.⁵ To achieve its Sustainable Development Goal of MMR of less than 70 per 100 000 live births by 2030, Myanmar must improve its annual reduction rate from 3.7%⁶ to 5.5%, which is the WHO’s recommendation.⁷ Nationwide surveys have shown that maternal deaths in Myanmar were most frequent during the postnatal period (42 days after delivery)^{8 9} and in mothers with home births.¹⁰ The most common cause of death is postpartum haemorrhage,^{8 9} which pointed out the immediate postpartum period as the most vulnerable time for Myanmar mothers.

Antenatal care (ANC) provides an opportunity to identify mothers who have a high risk of obstetric complications and reduce their risk factors.¹¹ ANC is the first point of contact between pregnant women and health services; it influences their care-seeking behaviours and links them to a referral system when needed.^{12 13} Pregnant women who do not receive ANC or have inadequate visits are at a greater risk of adverse pregnancy outcomes.¹⁴ Women who attend ANC are more likely to deliver in health institutions and receive postnatal care (PNC).¹⁵

Skilled attendants can effectively manage complications during delivery and reduce maternal mortality.¹⁶ To avert maternal and perinatal deaths, prompt access to emergency obstetric care and quality neonatal care, including resuscitation, which can be ensured by institutional delivery, are critical during childbirth and the immediate postpartum period.¹⁷ Likewise, the postnatal period is essential to the health of both the mother and her newborn, and PNC helps detect and manage complications that may arise after delivery.¹⁸ More than 60% of maternal deaths occurred in developing countries in the postpartum period.¹⁹ About 10%–27% of newborn deaths could be averted if PNC reached 90% of mothers and newborns.²⁰

Although ANC alone does not necessarily prevent complications at birth and the postnatal period, regular ANC visits encourage the utilisation of institutional delivery and PNC services. Moreover, it can ensure access to emergency obstetric services and increase chances of survival.^21–24 In 2016, the WHO recommended eight antenatal visits instead of four visits based on focused ANC (FANC) model to improve triage and timely referral of high-risk women,²⁵ based on evidence that the FANC model was probably associated with increased caesarean section rates and perinatal mortality.^{26 27}

However, the recommended number of antenatal visits in Myanmar is yet to be achieved, with only 18% of mothers receiving eight ANC visits²⁸ and 59% receiving four ANC visits.²⁹ In recent years, the Maternal Death Surveillance and Response System of Myanmar reported significant maternal deaths even among women who took four antenatal visits.^{9 10} This finding pointed out that women who received adequate antenatal visits still have a risk of maternal death and need to answer whether four antenatal visits were enough for the continued uptake of institutional delivery and PNC. Hence, we focused on these women and assessed the continuum of maternal healthcare utilisation among women with at least four antenatal visits using a methodology similar to the Ethiopian study by Fekadu et al.³⁰

Specific objectives

To examine the prevalence of institutional delivery and PNC after home delivery and to identify their determinants in Myanmar mothers who received at least four antenatal visits.

Methods

The study used the Myanmar Demographic and Health Survey (MDHS) data (2015–2016), the latest nationwide cross-sectional survey in Myanmar. The detailed methodology of MDHS can be found elsewhere.²⁹ In brief, the MDHS used systematic sampling methods and standardised questionnaires. Due to the non-proportional sample allocation and the possible differences in response rates across states/regions, we used the sample weights to get representative national and state/regional-level results.

The inclusion criteria for this study were Myanmar women aged 15–49 years who had at least four antenatal visits in their most recent birth within at least 5 years before the survey. A total of 2099 women were eligible for the institutional delivery outcome. According to the MDHS (2015–2016), only 8.2% of mothers who delivered at a health facility did not receive PNC.²⁹ Therefore, we excluded the women who had institutional delivery, assuming that all these women received PNC. We analysed the PNC utilisation only in women with at least four antenatal visits and home delivery. The sample for PNC utilisation after home delivery was a subset of these 2099 women. From 2099 women, we selected 380 women who had home delivery during the recent 2 years for PNC utilisation outcome, since this information was available only for women who delivered 2 years before the survey to reduce recall bias. We described the detailed population flow diagram in figure 1.

Figure 1.

Population flow diagram. Flow chart of study population selection. The grey boxes show the subjects excluded from the study. Green boxes show the samples analysed for each outcome. MDHS, Myanmar Demographic and Health Survey. IR, Individual Recode.

Patient and public involvement

No patient was involved.

Variables

Dependent variables

This study used institutional delivery and PNC 42 days after home delivery as dependent variables. The dependent variable, ‘institutional delivery’, was dichotomous. This study defined institutional delivery as delivering in a health facility for the most recent birth, including public, private, non-governmental organisation or government health facilities, based on the MDHS definition.

This study defined PNC utilisation as receiving a postnatal check by a healthcare provider within 42 days after home delivery. The outcome variable, ‘postnatal care after home delivery’, was also a dichotomous variable, categorised into ‘yes’ and ‘no’.

Independent variables

We selected the variables associated with institutional delivery and PNC utilisation in prior Myanmar studies and international studies with socioeconomic settings comparable with Myanmar as independent variables. These included women’s background characteristics, husband and child-related, household and service-related factors.

We categorised women’s age into five groups: ‘15–19 years’, ‘20–24 years’, ‘25–29 years’, ‘30–34 years’ and ‘35 years and above’³¹; education into four groups: ‘no formal education’, ‘primary education’, ‘secondary education’ and ‘more than secondary education’³²; and employment status into ‘currently employed’ and ‘currently unemployed’.³³ For women’s involvement in decision-making, we calculated a score based on three questions: ‘if the woman participates in decisions regarding her own healthcare’, ‘if the woman participates in decisions regarding large household purchases’, and ‘if the woman participates in decisions regarding visits to family or relatives’. The study used the same scoring as the final report of the MDHS.²⁹ A score of 0 was categorised as no involvement in decision-making, 1–2 as some involvement and 3 as full involvement.³³

We assigned the husband’s highest level of educational attainment to ‘no formal education’, ‘primary education’, ‘secondary education’ and ‘more than secondary education’. The birth order of the child was recoded into ‘first’, ‘second’, and ‘third and above’. We categorised the husband’s occupation into ‘managerial/clerical’, ‘sales/services’, ‘agricultural’ and ‘household/manual labour’. Pregnancy wantedness was categorised into ‘wanted at that time’, ‘wanted later’ and ‘not wanted’.³⁰

The region of residence was categorised into the hilly region (Kachin, Kayah, Kayin, Chin, Shan States), delta region (Yangon, Ayeyarwaddy, Bago regions), coastal region (Tanintharyi, Mon, Rakhine States) and central plains (Sagaing, Magway, Mandalay, Naypyitaw regions).³⁴ Place of residence was categorised into ‘urban’ and ‘rural’. The MDHS data measured the wealth status of the study population into the poorest, poorer, middle, richer and richest. Our study grouped the wealth index into poor (poorest and poorer), middle and rich (richer and richest).³³ We also categorised the family size into ‘five members or less’ and ‘more than five members’.

We generated the variable ‘problems in accessing healthcare’ based on the following questions in the MDHS:

1. Is there any problem in getting permission to go for treatment?

2. Is there any problem in getting the money needed for treatment?

3. Is there any problem due to the distance to the health facility?

4. Is there any problem in getting medical help due to not wanting to go alone?

If women responded with no problem to all four questions, we recoded them as ‘no’ (do not have difficulty accessing healthcare). If they had a problem with any question, we assumed it was ‘yes’ (difficult to access healthcare).³³

We recoded the ‘components of received ANC’ variable into a dummy variable based on seven components of ANC: tetanus toxoid injections, iron/folate supplementation, deworming, measured blood pressure, urine test, blood tests and received counselling about pregnancy-related complications during antenatal visits.³⁵ If women received all components, we recoded them as ‘yes’, and if otherwise, ‘no’. We considered this variable as a proxy indicator for service quality received by pregnant women.

If women delivered with doctors, nurses/midwives or female health visitors, we recoded them as birth with skilled assistance and vice versa.²⁹ We used this variable as an independent variable only for PNC utilisation outcome.

Statistical analysis

We checked the missing data, outliers and data completeness. We reported the prevalence of outcomes by error bars and the background characteristics of the study population using frequency distribution tables. We used the Pearson’s Χ² test to assess the bivariate association between dependent and independent variables. We also performed the multivariable binary logistic regression analysis to evaluate the determinant of institutional delivery and PNC utilisation. We used the manual backward deletion method for precise estimation of the ORs. We ran the initial model, including all variables whose p values were less than 0.2 during bivariate analysis, and we built the final model using the manual backward deletion method.^{33 36} We assessed the multicollinearity using the variance inflation factor and model fitness based on Hosmer-Lemeshow’s goodness-of-fit test. We used the sample weights using the survey data analysis command (svy) to account for design effect and non-response rate for all estimates to represent the whole population in the nation using STATA software (V.15.1). We set p value at 0.05 as statistically significant.

Results

The women from the delta region (38.3%) constituted most of the 2099 women analysed for ‘institutional delivery’. Among 380 women sampled for the outcome ‘PNC utilisation after a home delivery’, the majority came from central plains (34.6%). Rural women were more than urban women in both samples. The women who had home births were primarily poor (54.1%), while more rich women were included (44.8%) for institutional delivery. Most women in both samples had not received all seven components of ANC. Nearly half (43%) of the women who had home births did not receive skilled assistance at birth (table 1).

Table 1.

Frequency distribution of background characteristics of two samples

Variables	Institutional delivery (n=2099) n (%)	PNC after home delivery (n=380) n (%)
Age (years)
15–19	41 (2.0)	9 (2.2)
20–24	314 (15.0)	87 (23.0)
25–29	559 (26.6)	112 (29.6)
30–34	552 (26.3)	91 (24.0)
35 and above	633 (30.1)	81 (21.2)
Highest education level of the women
No formal education	183 (8.7)	54 (14.3)
Primary education	853 (40.6)	187 (49.3)
Secondary education	792 (37.8)	125 (32.8)
More than secondary education	271 (12.9)	14 (3.6)
Currently employed or not
No	896 (42.7)	202 (53.1)
Yes	1203 (57.3)	178 (46.9)
Involvement in decision-making*
No involvement	87 (4.3)	25 (6.8)
Some involvement	589 (29.2)	96 (25.9)
Full involvement	1344 (66.5)	249 (67.3)
Husband’s education
No formal education	225 (10.7)	66 (17.4)
Primary education	734 (35.0)	147 (38.6)
Secondary education	931 (44.3)	159 (41.9)
Higher than secondary education	209 (10.0)	8 (2.1)
Occupation of the husband†
Managerial/clerical	212 (10.2)	16 (4.2)
Sales/services	216 (10.4)	28 (7.5)
Agricultural	463 (22.3)	119 (31.5)
Household/manual labour	1189 (57.1)	214 (56.8)
Birth order of the index child
First	852 (40.6)	103 (27.1)
Second	582 (27.7)	123 (32.5)
Third and above	665 (31.7)	154 (40.4)
Regions
Hilly	386 (18.4)	88 (23.2)
Coastal	234 (11.1)	49 (12.8)
Delta	803 (38.3)	112 (29.4)
Central plains	676 (32.2)	131 (34.6)
Residence
Urban	705 (33.6)	63 (16.5)
Rural	1394 (66.4)	317 (83.5)
Wealth status
Poor	782 (37.2)	206 (54.1)
Middle	377 (18.0)	75 (19.7)
Rich	940 (44.8)	99 (26.2)
Family size
≤5 people	1200 (57.2)	200 (52.6)
>5 people	899 (42.8)	180 (47.4)
Problems in accessing healthcare
No problem	1125 (53.6)	173 (45.6)
Any problem	974 (46.4)	207 (54.4)
Received all seven antenatal components
No	1419 (67.6)	249 (65.5)
Yes	680 (32.4)	131 (34.5)
Received ≥2 times of tetanus injections before birth	1701 (81.1)	323 (85.2)
Received iron supplements during pregnancy	2050 (97.7)	373 (98.2)
Received deworming medication during pregnancy	1355 (64.6)	281 (73.9)
Blood pressure taken during antenatal visits	2012 (95.9)	367 (96.7)
Urine sample taken during antenatal visits	1508 (71.8)	239 (63.0)
Blood tests done during antenatal visits	1491 (71.1)	251 (66.2)
Received health education about pregnancy complications during antenatal visits	1740 (82.9)	317 (83.5)
Pregnancy wantedness
Wanted at that time	1924 (91.7)	343 (90.4)
Wanted later	90 (4.3)	23 (5.8)
Not wanted at all	85 (4.0)	14 (3.8)
Skilled assistance at birth
No		164 (43.0)
Yes		216 (57.0)

Weighted numbers are presented.

*Missing values present. Seventy-nine women had no data in ‘involvement in decision-making’.

†Missing values present. Nineteen women had no data for the variable ‘occupation of the husband’.

PNC, postnatal care.

Prevalence

A total of 1148 women had institutional delivery, and the prevalence for the union was 54.7% (95% CI: 51.2%, 58.2%). The delta region had the highest institutional delivery (59.8%; 95% CI: 53.5%, 65.8%) and the hilly region had the lowest utilisation (48.9%; 95% CI: 40.6%, 57.4%). Two hundred and eighty-eight women who delivered at home received PNC, and the union prevalence for PNC utilisation among these women was 76% (95% CI: 70.2%, 80.9%). The central plains was the highest (89.3%; 95% CI: 81.5%, 94%) and the hilly region was the lowest (61%; 95% CI: 49.6%, 71.4%) for PNC utilisation (figure 2).

Figure 2.

Prevalence of institutional delivery and PNC utilisation after home delivery in reproductive-aged Myanmar women who had at least four antenatal visits for their most recent birth. PNC, postnatal care.

Bivariate analyses

Women’s education and employment, husband’s education, husband’s occupation, birth order of the child, residence, wealth status, problems in accessing healthcare and components of ANC received were significantly associated with institutional delivery in bivariate analysis. The educational level of the women, region of residence, components of ANC received and skilled assistance at birth were significantly associated with PNC utilisation (table 2).

Table 2.

Bivariate analysis for outcome variables (institutional delivery and PNC utilisation) using background characteristics of reproductive-aged Myanmar women who had at least four antenatal visits for their most recent birth

Variables	Institutional delivery (N=2099)			PNC utilisation (N=380)
	n	%	95% CI	n	%	95% CI
Age group			p=0.805			p=0.671
15–19	41	63.4	44.5, 78.8	9	64.1	27.5, 89.3
20–24	314	54.5	47.5, 61.3	87	73.6	62.4, 82.4
25–29	559	55.9	50.0, 61.6	112	73.9	63.9, 82.0
30–34	552	53.0	47.6, 58.3	91	76.5	64.0, 85.6
35 and above	633	54.8	49.9, 59.5	81	82.0	71.1, 89.4
Highest education level of women			p<0.001			p=0.036
No formal education	183	29.0	21.6, 37.7	54	59.8	43.5, 74.1
Primary education	853	42.9	38.0, 48.0	187	79.2	71.8, 85.0
Secondary education	792	62.7	57.7, 67.4	125	76.8	67.4, 84.1
Higher than secondary education	271	86.0	81.1, 89.8	14	89.0	58.7, 97.9
Employment of the women			p=0.040			p=0.673
Not currently employed	896	58.0	53.3, 62.5	202	77.0	69.9, 82.8
Currently employed	1203	52.3	48.1, 56.5	178	74.8	65.7, 82.1
Involvement in decision-making			p=0.478			p=0.107
No involvement	87	51.6	39.0, 64.0	25	59.7	37.5, 78.6
Some involvement	589	57.2	51.8, 62.5	96	72.1	59.8, 81.8
Full involvement	1344	54.0	50.1, 57.9	249	79.2	72.8, 84.4
Husband’s education			p<0.001			p=0.144
No formal education	225	31.5	25.1, 38.7	66	78.1	64.0, 87.7
Primary education	734	45.2	40.0, 50.6	147	70.3	60.2, 78.7
Secondary education	931	60.2	55.8, 64.4	159	79.4	72.2, 85.2
Higher than secondary education	209	88.7	83.4, 92.4	8	94.1	80.0, 98.5
Occupation of the husband			p<0.001			p=0.339
Managerial/clerical	212	81.1	74.7, 86.1	16	60.9	33.5, 82.8
Sales/services	216	71.0	61.9, 78.6	28	64.5	36.7, 85.0
Agricultural	463	36.8	30.7, 43.4	119	80.5	69.8, 88.0
Household/manual labour	1189	53.8	49.4, 58.0	214	75.8	68.9, 81.6
Birth order of index child			p<0.001			p=0.682
First	852	70.6	65.9, 74.8	103	79.3	69.0, 86.9
Second	582	49.9	44.5, 55.3	123	74.7	65.1, 82.4
Third and above	665	38.6	33.9, 43.6	154	74.7	66.1, 81.7
Regions			p=0.102			p<0.001
Hilly	386	48.9	40.6, 57.4	88	61.0	49.6, 71.4
Coastal	234	51.6	43.7, 59.4	49	81.3	68.9, 89.5
Delta	803	59.8	53.5, 65.8	112	69.7	56.3, 80.4
Central plains	676	53.0	47.2, 58.7	131	89.3	81.5, 94.0
Residence			p<0.001			p=0.557
Urban	705	78.9	73.5, 83.6	63	71.8	52.6, 85.4
Rural	1394	42.4	38.1, 46.9	317	76.8	71.0, 81.7
Wealth status			p<0.001			p=0.404
Poor	782	35.8	30.5, 41.4	206	74.8	67.1, 81.2
Middle	377	46.6	40.0, 53.3	75	77.8	65.4, 86.6
Rich	940	73.7	69.6, 77.5	99	77.0	64.9, 85.8
Family size			p=0.188			p=0.077
≤5 people	1200	53.2	48.9, 57.5	200	71.6	62.5, 79.2
>5 people	899	56.7	52.2, 61.1	180	80.8	74.0, 86.2
Problems in accessing healthcare			p<0.001			p=0.104
No problem	1125	62.3	58.1, 66.2	173	80.2	72.4, 86.3
Any problem	974	46.0	41.2, 50.9	207	72.4	64.7, 78.9
Received all antenatal components			p<0.001			p=0.011
Yes	1419	62.3	56.8, 67.5	249	85.4	75.6, 91.7
No	680	51.1	47.0, 55.2	131	71.0	63.9, 77.2
Pregnancy wantedness			p=0.739			p=0.348
Wanted at that time	1924	54.9	51.3, 58.4	343	77.1	71.2, 82.1
Wanted later	90	50.2	38.0, 62.3	23	65.6	40.3, 84.4
Not wanted at all	85	55.6	43.0, 67.5	14	64.9	38.6, 84.4
Skilled assistance at birth						p<0.001
No				164	60.8	51.8, 69.2
Yes				216	87.4	81.0, 91.8

Weighted numbers are presented. P values were calculated using Pearson’s Χ² tests.

PNC, postnatal care.

Multivariable binary logistic regression

We included the variables with p<0.2 in bivariate analysis for multivariable binary logistic regression. Women with higher than secondary education were 2.63 times (95% CI: 1.49, 4.64; p=0.001) more likely to have institutional delivery than women with no formal education. Compared with women who had husbands working in agricultural jobs, women who had husbands working in managerial/clerical jobs were 2.33 times more likely (95% CI: 1.42, 3.82; p=0.001), women who had husbands working in sales/services were 1.81 times more likely (95% CI: 1.10, 2.98; p=0.019) and women who had husbands working in household/manual labour were 1.37 times more likely (95% CI: 1.00, 1.88; p=0.048) to have institutional delivery (table 3).

Table 3.

Determinants of institutional delivery among reproductive-aged Myanmar women who had at least four antenatal visits for their most recent birth (N=2099)

	Initial model		Final model
Variables	aOR (95% CI)	P value	aOR (95% CI)	P value
Highest education level of women
No formal education	1.0		1.0
Primary education	1.34 (0.9, 1.99)	0.151	1.37 (0.92, 2.04)	0.121
Secondary education	1.47 (0.95, 2.30)	0.085	1.54 (0.99, 2.39)	0.053
Higher than secondary education	2.58 (1.46, 4.54)	0.001	2.63 (1.49 to 4.64)	0.001
Employment of the women
Not currently employed	1.0
Currently employed	0.84 (0.66, 1.07)	0.166
Occupation of the husband
Agricultural	1.0		1.0
Managerial/clerical	2.34 (1.42, 3.85)	0.001	2.33 (1.42, 3.82)	0.001
Sales/services	1.74 (1.05, 2.88)	0.032	1.81 (1.10, 2.98)	0.019
Household/manual labour	1.37 (1.00, 1.88)	0.049	1.37 (1.00, 1.88)	0.048
Husband's education
No formal education	1.0		1.0
Primary education	1.66 (1.14, 2.42)	0.008	1.68 (1.14, 2.49)	0.010
Secondary education	1.59 (1.10, 2.30)	0.014	1.64 (1.12, 2.40)	0.010
Higher than secondary education	2.86 (1.58, 5.18)	0.001	2.90 (1.61, 5.22)	<0.001
Birth order of index child
First	1.0		1.0
Second	0.44 (0.33, 0.58)	<0.001	0.44 (0.33, 0.58)	<0.001
Third and above	0.39 (0.28, 0.53)	<0.001	0.38 (0.28, 0.52)	<0.001
Region
Hilly	1.0
Coastal	1.16 (0.75, 1.80)	0.511
Delta	1.21 (0.78, 1.87)	0.398
Central plains	0.98 (0.65, 1.46)	0.910
Residence
Rural	1.0		1.0
Urban	2.31 (1.59, 3.35)	<0.001	2.49 (1.71, 3.62)	<0.001
Wealth status
Poor	1.0		1.0
Middle	1.19 (0.83, 1.70)	0.343	1.18 (0.82, 1.69)	0.369
Rich	1.96 (1.39, 2.75)	<0.001	1.97 (1.40, 2.77)	<0.001
Family size
≤5 people	1.0
>5 people	1.05 (0.82, 1.34)	0.691
Problems in accessing healthcare
No problem	1.0
Any problem	0.83 (0.65, 1.06)	0.140
Received all antenatal components
No	1.0
Yes	1.22 (0.91, 1.63)	0.176

aOR, adjusted OR.

Compared with women whose husbands had no formal education, women with husbands who achieved higher than secondary education were 2.9 times more likely (95% CI: 1.61, 5.22; p<0.001), women who had husbands with primary level education were 1.68 times more likely (95% CI: 1.14, 2.49; p=0.010) and women with husbands who achieved secondary level education were 1.64 times more likely (95% CI: 1.12, 2.40; p=0.010) to have institutional delivery. Compared with women who were having their first child, women having their 3rd–12th child were 62% less likely (95% CI: 0.28, 0.52; p<0.001) and women having their second child were 56% less likely (95% CI: 0.33, 0.58; p<0.001) to deliver at a health facility (table 3).

Urban women were 2.49 times more likely than rural women to use institutional delivery (95% CI: 1.71, 3.62; p<0.001). Women from rich households were 1.97 times more likely to have institutional delivery than women from poor households (95% CI: 1.40, 2.77; p<0.001) (table 3).

In the analysis of PNC utilisation, women living in the central plains region were 4.12 times more likely (95% CI: 1.90, 8.93; p<0.001) and women living in the coastal region were 3.04 times more likely (95% CI: 1.36, 6.82, p=0.007) to receive PNC after home delivery compared with women living in the hilly region. Women who had received quality ANC were 2.02 times more likely to use PNC than those who did not receive all seven components (95% CI: 1.01, 4.03; p=0.046). Compared with women who did not receive skilled assistance at birth, women who received skilled assistance had 3.42 times higher chances of receiving PNC after home delivery (95% CI: 1.81, 6.46, p<0.001). See details in table 4.

Table 4.

Determinants of PNC utilisation among reproductive-aged Myanmar women who had at least four antenatal visits and had home delivery in their most recent birth (N=380)

Variables	Initial model		Final model
	aOR (95% CI)	P value	aOR (95% CI)	P value
Highest education level of women
No formal education	1.0
Primary education	2.65 (1.15, 6.11)	0.023
Secondary education	2.09 (0.82, 5.35)	0.123
Higher than secondary education	1.42 (0.13, 15.1)	0.773
Involvement in decision-making
No involvement	1.0
Some involvement	1.29 (0.42, 4.01)	0.653
Full involvement	1.71 (0.58, 5.04)	0.327
Husband's education
No formal education	1.0
Primary education	0.66 (0.27, 1.65)	0.376
Secondary education	0.84 (0.35, 1.99)	0.691
Higher than secondary education	2.83 (0.29, 7.29)	0.365
Region
Hilly	1.0		1.0
Coastal	2.69 (1.20, 6.04)	0.017	3.04 (1.36, 6.82)	0.007
Delta	1.36 (0.61, 3.02)	0.450	1.45 (0.67, 3.16)	0.344
Central plains	4.96 (2.15, 11.44)	<0.001	4.12 (1.90, 8.93)	<0.001
Family size
≤5 people	1.0
>5 people	1.85 (0.98, 3.49)	0.056
Problems in accessing healthcare
No problem	1.0
Any problem	0.97 (0.52, 1.82)	0.931
Received all antenatal components
No	1.0		1.0
Yes	1.78 (0.87, 3.63)	0.111	2.02 (1.01, 4.03)	0.046
Skilled assistance at birth
No	1.0		1.0
Yes	2.75 (1.40, 5.41)	0.004	3.42 (1.81, 6.46)	<0.001

aOR, adjusted OR; PNC, postnatal care.

Discussion

Institutional delivery prevalence observed was higher than union figures stated in the 2015–2016 MDHS report at 37%²⁹ and Myanmar Public Health Statistics Report (2014–2016) at 51%,³⁷ due to analysing only women with at least four antenatal visits in this study. However, the prevalence was below the target set in Myanmar’s Five Years Strategic Plan for Reproductive Health (2014–2018) at 60%.³⁸ Regarding PNC utilisation, while there is no specific target for women after a home delivery, the prevalence seen was approaching but fell short of the general target of 80% set by the WHO.³⁹ Thus, both institutional delivery and PNC prevalence in Myanmar, even in the group of women who had four antenatal visits, did not meet their respective target set. The PNC prevalence was significantly higher than the Ethiopian study, that is, 8%; however, the institutional delivery prevalence was close to Ethiopia’s prevalence of 56%, respectively, as observed by Fekadu et al’s research,³⁰ which had a similar methodology.

Educational attainment, both in women and husbands, was predicted to be a significant predictor of the place of delivery. This finding is consistent with Fekadu et al’s 2018 Ethiopian study³⁰ and previous Myanmar and international studies.^40–43 People with higher educational attainment are generally more knowledgeable and cautious of pregnancy-related complications. They can also communicate with healthcare providers more efficiently and inquire more about delivery care services. Another reasonable supposition is that they had higher socioeconomic status and were thus financially more accessible to deliver at a health facility.

Our current study found that several variables, directly or indirectly representing socioeconomic status, significantly influenced Myanmar women’s choice of delivery place. For instance, the household’s wealth status was a significant predictor of institutional delivery, which coincides with Fekadu et al’s 2018 Ethiopian study.¹⁶ The financial constraints that led to their home delivery decision could explain this finding. The husband’s occupation indirectly indicates socioeconomic status and significantly influences institutional delivery. Women whose husbands were working in agriculture were less likely to have institutional delivery than those with any other occupation type, and this finding was consistent with previous studies.⁴⁴ Agricultural workers generally have lower educational attainment and socioeconomic status and usually reside in rural areas. These factors might be obstacles to the institutional delivery of their wives. Programmes that promote maternal healthcare services in educationally and economically disadvantaged women, such as voucher schemes, might improve institutional delivery.

Our study found urban–rural variation in institutional delivery utilisation, that is, urban women were 2.5 times more likely to have institutional delivery than rural women. This finding coincides with Fekadu et al’s 2018 study³⁰ and previous Myanmar studies.^{29 37 45} This might be due to difficulties in accessing health facilities,⁴⁶ considerable disparities in health facilities and health workforce between urban and rural,⁴⁷ or traditional practices of rural women, which make them prefer home delivery.⁴⁸ Until rural women incline towards institutional deliveries and the health facilities are ready for them, basic emergency obstetric care services, which can be provided at the primary level of healthcare, should be prioritised in rural areas. Local midwives and auxiliary midwives (AMWs) should be given basic emergency obstetrics and newborn care training. Moreover, emergency lifesaving commodities, such as oxytocin, misoprostol and clean delivery kits, should be readily available at the grass-roots level. The referral system needs to be strengthened to avoid delays in reaching care in case of an emergency during a home delivery.

The child’s birth order also influenced the choice of institutional delivery utilisation. This finding is consistent with previous studies^{30 49} and is assumed to be contributed by the Ministry of Health’s guideline of delivering primiparous mothers at a health facility. Another possible reason was that mothers and their families tended to give more attention to the first childbirth. Grand multipara mothers with a high risk of complications⁵⁰ should be urged for institutional deliveries during antenatal visits.

We also found regional variation in PNC utilisation. The women from the hilly region were less likely to receive PNC after home delivery. It may be due to the hilly region being less developed, suffering from war and conflicts, and its scarcity of health workforce.⁴⁷ Further research in this region, using a qualitative or mixed-methods approach, is recommended to explore the barriers healthcare providers face in paying PNC visits to mothers who had home births.

One notable finding was that nearly half (43%) of women who chose to deliver at home did not receive skilled assistance at birth. Moreover, those women who did not receive skilled assistance at birth are also less likely to receive PNC after home delivery. Although these women had access to services and got four antenatal visits, they chose not to deliver with skilled birth attendants (SBAs), missing healthcare during the intrapartum and postpartum periods when maternal mortality is the highest.⁸ Previous qualitative studies explored why some mothers delivered with traditional birth attendants (TBAs). These studies described many reasons: (1) community acceptance of TBAs and the influence of older women within the community,⁵¹ (2) preference for the traditional methods that TBAs offer, such as applying the turmeric paste,⁴⁸ (3) negative perception that it is more costly to deliver with SBAs,⁵¹ and (4) experiences of being asked for unaffordable payments by some SBAs and receiving inadequate care when they could not pay enough.⁵² Further generalisable research is needed to identify the issues that ought to be resolved by the stakeholders.

Women who received all seven components of ANC were more likely to receive PNC after home delivery. This finding is consistent with Fekadu et al’s 2018 study.³⁰ The possible explanation for this finding is that these mothers received counselling about pregnancy-related complications and understand the need to deliver at a health facility and the benefits of PNC. Another plausible explanation is that it reflects the quality of ANC the women received and the competency of their service provider, which is assumed to play a significant role in PNC utilisation after home delivery. Well-trained midwives would give all essential components of ANC and pay PNC visits to mothers who had home births.

However, descriptive statistics revealed that, although the receipt of each component well surpassed over 70%, except for deworming, only one-third of the women received all components. This finding might stem from several causes: midwives not being well trained, midwives being overworked, recall bias or a combined effect of all. The literature stated that some midwives deviate from antenatal guidelines due to insufficient time for the procedures,⁵¹ since they are overburdened with various vertical project activities, spending less time on maternal and reproductive health activities.^{51 53 54} Besides giving regular training to strengthen their technical skills, the Ministry of Health should develop strategies for recruiting new midwives and task-shifting their workload onto AMWs and public health supervisors.

This study has some limitations. As the study used the 2015–2016 MDHS data, the available variables were limited; therefore, the analysis could not include some variables supported by the literature. In addition, causality is not possible for the associations found in this study since we used cross-sectional data. However, as the MDHS data were collected nationwide, with systematic sampling methods and standardised data collection procedures, the results were nationally representative and should be generalisable to the whole country.

Conclusion

The institutional delivery and PNC prevalence in Myanmar did not meet the national targets in the group of women who had four antenatal visits. Myanmar women’s choice of delivery place largely depended on their socioeconomic characteristics. Once the socioeconomic status of Myanmar mothers advances, institutional delivery rates are likely to increase altogether. The maternal and child health programmes should improve the institutional delivery and PNC utilisation rates by promoting the quality, availability and accessibility of the services provided, especially in the hilly region. Further research using a mixed-methods approach is warranted in mothers who dropped out along the continuum of care to explore the reasons for not using institutional delivery and PNC.

Data availability statement

Data are available upon reasonable request at "The DHS Program" website.https://dhsprogram.com/data/available-datasets.cfm

Ethics statements

Patient consent for publication

Not required.

Ethics approval

We used the required dataset of the MDHS (2015–2016) with permission from ICF International. We received ethical approval from the University of Public Health Institutional Review Board, Yangon (UPH-IRB) (UPHIRB/MPH/7). The MDHS protocol was approved by the Ethics Review Committee on Medical Research, including Human Subjects in the Department of Medical Research, Myanmar Ministry of Health and Sports, and by the ICF Institutional Review Board.