Estimating the prevalence of vaginal birth after cesarean section and repeat cesarean section: An analysis of nationally representative household surveys in 59 low‐ and middle‐income countries

Abstract

Objective

Little is known about the birthing patterns of patients having experienced a prior cesarean section (CS) in low‐ and middle‐income countries (LMICs). We sought to estimate the prevalence of and characterize trends relating to vaginal birth after cesarean section (VBAC) and repeat CS in LMICs.

Methods

We analyzed 107 nationally representative surveys across 59 LMICs. For countries with population CS rates above 8%, we analyzed repeat CS trends across two or three consecutive surveys and estimated the association between repeat CS and facility level and facility type.

Results

The prevalence of VBAC ranged from 0.5% in Turkey (95% CI: 0.1, 3.5) to 67.6% in Niger (95% CI: 53.4, 79.2). We observed an inverse association between the population CS rate and VBAC prevalence. Of the 33 countries with CS rates below 10%, 26 countries had VBAC prevalence estimates ranging from 20% to 50% with the prevalence of six countries exceeding 50%. Of the 26 countries with CS rates at or above 10%, 19 had VBAC prevalence estimates below 20%. We observed strong associations between repeat CS and births in private facilities or hospitals, and we found that repeat CS also increases over time across multiple countries. However, many of these associations were not statistically significant across multiple countries.

Conclusion

VBAC prevalence is greatest in countries with the lowest population CS rates, indicating that unplanned, unmonitored, and unsafe VBACs are occurring, necessitating shifts in birth planning and preparedness strategy. Repeat cesareans predominate where population CS rates are already high.

1. INTRODUCTION

The dual challenge of cesarean section (CS) underuse and overuse ultimately means that women and newborns risk being injured or even dying from preventable causes. Where high unmet need for emergency obstetric and newborn care (EmONC) persists, obstetric complications for which CS is indicated risk going unmanaged. ¹ , ² In places with CS overuse, surgical complications and iatrogenic injury happen all too frequently, ³ especially where health facilities that lack the equipment and human resources to consistently perform the procedure safely. Much has been done to sound the alarm on the need for safe and appropriate CS: the WHO issued a statement on the importance of providing CS to women in need, ⁴ researchers have published effective approaches for reducing unnecessary CS, ⁵ and development assistance projects have mobilized to support health systems in delivering high‐quality surgical obstetric care. ² , ⁶

However, little is known about what happens in future pregnancies after an initial CS in low‐ and middle‐income countries (LMICs). ⁷ Women may face several birthing outcomes in ensuing pregnancies, including a repeat CS (scheduled or emergency), or a supervised trial of labor after cesarean (TOLAC) which may result in vaginal birth after cesarean (VBAC) or in referral for a CS if deemed safer and/or indicated. Although cesareans are associated with a variety of postoperative morbidities, ² , ⁸ the risk of placenta accreta increases among women with repeat CS. ⁹ TOLAC, which may be appropriate for many women with one prior CS, ¹⁰ has been perceived as preferable among patients in some contexts, ¹¹ , ¹² and has been previously demonstrated as safe among low‐risk women. ¹³ However, providers have also indicated that limited human resources and equipment (e.g., clinical guidelines, cardiotocography and labor monitoring tools) impedes the provision of safe and appropriate VBAC. ¹⁴ , ¹⁵

The current evidence on repeat CS and VBAC in LMICs is generally limited to observations from specific health facilities, covering patient and provider perceptions, ¹¹ , ¹² measures of successful or failed TOLAC, ¹⁶ , ¹⁷ and risk prediction. ¹⁸ In one study, the repeat CS prevalence across five sites in Bangladesh, Guatemala, India, and Pakistan, was estimated at 83.2%. ¹⁹ As the global CS rate is expected to grow to 28.5% by 2030, ²⁰ it is important to understand how women with prior CS are delivering across LMIC.

We aimed to address this issue through a multi‐country analysis of household survey datasets from the Demographic and Health Survey (DHS) Program. First, we aimed to estimate the prevalence of VBAC and repeat CS, and describe the relationship between VBAC and the population CS rate. Second, we aimed to characterize repeat CS trends by (1) quantifying the proportion of repeat CS among all births and all CS for all countries. We then aimed to, among a subset of countries, (2) describe how the repeat CS prevalence changes over time, and (3) examine the extent to which repeat CS differs by the type of health facility (private vs public) and level of health facility (hospital vs other health facility).

2. MATERIALS AND METHODS

2.1. Survey design

DHS collect data from nationally representative samples across a variety of health areas. ²¹ Surveys generally follow a two‐stage stratified sampling design, in which primary sampling units (usually census enumeration areas) are selected in the first stage, households within each primary sampling unit are listed, and a number of households are selected from the listing in the second stage. Sampling is usually stratified by both geographic region and whether the household is in an urban or rural area. Sampling during the first stage is performed using probability proportional to size within each stratum, whereas each household is selected with equal probability during the second stage.

The surveys in this analysis cover 59 LMICs in Central Asia, Latin America and the Caribbean, North Africa/West Asia/Europe, South and Southeast Asia, and sub‐Saharan Africa, and were performed between 1991 and 2023. We used the most recent survey to estimate the prevalence of VBAC and repeat CS and to characterize how the prevalence of repeat CS differs by facility type and level. We used either the two or three most recent surveys to describe how repeat CS prevalence has changed over time. The variables in the dataset are included in the birth recode file, whose questions map to the women's questionnaire. Each row in the birth recode represents a woman‐newborn dyad. Depending on the survey, a woman's mode of delivery is captured in the birth recode file for either the past 3 or 5 years preceding the survey.

2.2. Variables

We constructed the VBAC and repeat CS variables using the birth index and the mode of delivery (a binary variable describing whether the child was born by CS). These two variables can be used to classify each childbirth into five mutually exclusive groups: first time CS (first cesarean), first time vaginal birth (first vaginal), repeat CS, repeat vaginal birth, and VBAC. In this analysis, vaginal birth or CS for women reporting a prior CS were classified as either VBAC or repeat CS. If a vaginal birth occurred after both a vaginal and a cesarean birth, that specific birth was classified as a VBAC as opposed to a repeat vaginal birth. Furthermore, multiple births were counted as one birth. If a multiple birth comprised of a vaginal and a first‐time CS, the birth was classified as a first‐time CS. If one of the births in a multiple birth was vaginal and the woman had a history of CS, then the birth would be classified as a VBAC.

We characterized repeat CS across multiple explanatory variables, including survey year, facility level, and facility type. We recoded facility level as a binary variable: “hospital” for facilities with the word “hospital” in their categorization, and “other health facility” for all other facilities. We recoded facility type as a binary variable as well. Any facilities with the words “public” or “government” in their description were classified as public facilities. Facilities with the words “private,” “NGO,” or “religious” were categorized as private facilities.

2.3. Analytical methods

The primary analysis involved estimating the prevalence of VBAC and repeat CS in all 59 LMICs. For this analysis, we used the most recent survey since 2012 and accounted for the complex survey design by applying the survey weights, stratum, and primary sampling unit variables included in the datasets. We present the prevalence estimate and 95% confidence intervals (95% CI) for all births and for births among women reporting prior CS. We adopt the latter as the denominator for our working prevalence definitions. In addition, we present estimates of first cesarean, first vaginal, repeat cesarean, and repeat vaginal deliveries as a proportion of all births. As part of this analysis, we plotted (1) the relative contribution of first time and repeat cesarean to the overall population CS rate, and (2) the VBAC prevalence against the population CS rate.

To characterize how the repeat CS prevalence changes over time, we plotted repeat CS prevalence estimates across consecutive surveys. We limited this analysis to 28 countries with a population CS rate exceeding 8% in their most recent survey to ensure adequate sample sizes. Depending on what surveys were publicly available, we either compared two or three consecutive surveys. For surveys spanning multiple years (e.g., India 2019–2021), we used the first year of the survey as the time point. In addition to the plotting the point estimate, we denote the 95% CI using error bars. We tested for trends between surveys using t‐tests, using a P value of <0.05 to denote a statistically significant change in prevalence between surveys.

Lastly, for a subset of 16 countries with a population rate exceeding 8% and more than 10 VBAC in their most recent survey, we explored how repeat CS differs across health facilities by level (hospital versus non‐hospital) and type (private versus public). This analysis was performed using simple logistic regression. The odds ratio (OR) and 95% CI are reported for each country.

All statistical analyses were performed in Stata 18. Data visualizations were generated in R following the post‐processing of analytical output in Microsoft Excel.

2.4. Ethical review and informed consent

This study was a secondary analysis of publicly available data. As such, this study was not considered human subjects research per EngenderHealth institutional standard operating procedures and ethical review was not sought. Because no data were collected from human subjects as part of this secondary analysis, informed consent was not sought. Per DHS documentation, data collection protocols and instruments are reviewed by the ICF institutional review board, informed consent is sought from each participant prior to data collection, and ethical review documentation can be made on request. ²²

3. RESULTS

A total of 563 605 women spanning 325 480 deliveries were included in the 59 most recent surveys and 1 053 086 women spanning 589 876 deliveries were included across the 107 surveys used across all analyses (Table S1). Most births were either first‐time vaginal delivery, first‐time CS, or repeat vaginal deliveries, as opposed to repeat CS or VBACs. The weighted number of births among women with prior CS (i.e., repeat CS and VBACs) ranged from six in Burkina Faso to 6936 in India, compared to 4830 and 90 886, respectively, for the weighted number of all births (Table 1).

TABLE 1.

Prevalence of first and repeat cesarean and vaginal births as a proportion of all births and VBAC prevalence among women with a history of prior cesarean section.

Survey	Birth categorization as a percentage of all births						Among women with prior cesarean
	First vaginal	First cesarean	Repeat vaginal	Repeat cesarean	VBAC	Weighted N	VBAC	Repeat cesarean	Weighted N
	% [95% CI]						% [95% CI]
Kyrgyz Republic 2012	70.2 [68.9, 71.5]	5.3 [4.4, 6.3]	23.9 [22.8, 25.1]	0.4 [0.2, 0.7]	0.2 [0.1, 0.4]	1778	34 [17.4, 55.7]	66 [44.3, 82.6]	24
Tajikistan 2017	64.7 [63.6, 65.9]	4.2 [3.7, 4.8]	29.8 [28.7, 31.0]	0.8 [0.6, 1.1]	0.4 [0.2, 0.6]	2670	31.3 [19.9, 45.5]	68.7 [54.5, 80.1]	78
Colombia 2015–16	49.0 [47.5, 50.6]	38.5 [37.0, 40.0]	7.9 [7.2, 8.6]	4.4 [3.6, 5.4]	0.3 [0.2, 0.4]	4392	5.5 [3.3, 9.1]	94.5 [90.9, 96.7]	501
Dominican Republic 2013	35.0 [32.8, 37.2]	49.1 [46.9, 51.4]	8.1 [6.9, 9.5]	7.3 [6.2, 8.5]	0.5 [0.2, 1.4]	1442	6.3 [2.1, 17.0]	93.7 [83.0, 97.9]	276
Guatemala 2014–15	55.8 [54.8, 56.8]	23.0 [21.9, 24.2]	17.7 [16.9, 18.7]	3.0 [2.6, 3.4]	0.5 [0.3, 0.6]	4942	13.2 [10.0, 17.1]	86.8 [82.9, 90.0]	424
Haiti 2016–17	74.4 [72.9, 75.8]	4.9 [4.2, 5.7]	20.3 [18.9, 21.7]	0.4 [0.2, 0.6]	0.1 [0.1, 0.2]	2501	23.4 [9.3, 47.7]	76.6 [52.3, 90.7]	31
Honduras 2011–12	66.4 [65.4, 67.4]	16.7 [15.8, 17.6]	15 [14.3, 15.7]	1.5 [1.2, 1.8]	0.4 [0.3, 0.6]	4274	20.8 [14.2, 29.3]	79.2 [70.7, 85.8]	191
Albania 2017–18	59.6 [57.2, 62.0]	27.8 [25.4, 30.3]	9.4 [8.0, 11.1]	3.0 [2.3, 3.8]	0.2 [0.1, 0.7]	1064	6.4 [1.9, 19.3]	93.6 [80.7, 98.1]	80
Armenia 2015–16	67.7 [65.5, 69.9]	15.4 [13.7, 17.3]	14.4 [12.9, 16.1]	2.3 [1.6, 3.2]	0.2 [0.1, 0.5]	698	7.2 [2.5, 18.9]	92.8 [81.1, 97.5]	41
Egypt 2014	36.0 [34.9, 37.0]	40.4 [39.3, 41.5]	12.2 [11.5, 12.9]	11.0 [10.4, 11.7]	0.4 [0.2, 0.5]	6627	3.1 [2.2, 4.3]	96.9 [95.7, 97.8]	1746
Jordan 2023	51.5 [49.2, 53.8]	36.7 [34.5, 38.9]	7.6 [6.6, 8.8]	3.8 [3.1, 4.7]	0.4 [0.2, 0.8]	2825	9.1 [4.4, 17.8]	90.9 [82.2, 95.6]	180
Turkey 2018–19	39.0 [36.9, 41.1]	42.9 [40.5, 45.4]	10.1 [8.8, 11.6]	7.9 [6.9, 9.0]	0 [0.0, 0.3]	969	0.5 [0.1, 3.5]	99.5 [96.5, 99.9]	201
Yemen 2013	62.5 [61.6, 63.4]	4.0 [3.6, 4.5]	32.5 [31.6, 33.5]	0.6 [0.5, 0.8]	0.3 [0.2, 0.5]	6496	34.9 [25.5, 45.8]	65.1 [54.2, 74.5]	146
Afghanistan 2015	60.5 [59.6, 61.4]	2.3 [2.0, 2.7]	36.6 [35.7, 37.6]	0.3 [0.2, 0.5]	0.3 [0.2, 0.4]	12 064	48.4 [38.1, 58.8]	51.6 [41.2, 61.9]	181
Bangladesh 2022	53.3 [51.2, 55.4]	42.8 [40.6, 45.0]	2.6 [2.2, 3.1]	1.2 [0.9, 1.6]	0 [0.0, 0.1]	3691	2.1 [0.5, 8.6]	97.9 [91.4, 99.5]	67
Cambodia 2021–22	77.2 [75.6, 78.8]	17.8 [16.2, 19.5]	4.5 [3.8, 5.2]	0.4 [0.2, 0.7]	0.1 [0.1, 0.3]	3322	27.3 [11.0, 53.4]	72.7 [46.6, 89.0]	26
India 2019–21	58.7 [58.4, 58.9]	18.6 [18.4, 18.9]	19.6 [19.4, 19.9]	2.7 [2.6, 2.8]	0.4 [0.3, 0.4]	90 886	12.2 [11.2, 13.3]	87.8 [86.7, 88.8]	6936
Indonesia 2017	74.0 [73.1, 74.9]	15.8 [15.0, 16.6]	8.9 [8.4, 9.4]	1.2 [1.0, 1.4]	0.1 [0.1, 0.2]	6731	11.2 [7.3, 16.9]	88.8 [83.1, 92.7]	221
Maldives 2016–17	51.9 [49.5, 54.2]	36.0 [33.9, 38.3]	7.6 [6.3, 9.0]	3.9 [3.2, 4.9]	0.6 [0.3, 1.2]	1096	13.4 [7.3, 23.5]	86.6 [76.5, 92.7]	123
Myanmar 2015–16	68.6 [66.9, 70.3]	16.4 [14.7, 18.3]	14.3 [12.7, 16.1]	0.5 [0.3, 0.8]	0.1 [0.0, 0.3]	1707	15.6 [4.7, 40.7]	84.4 [59.3, 95.3]	26
Nepal 2021–22	75.5 [73.5, 77.4]	18.0 [16.1, 20.0]	6.1 [5.1, 7.3]	0.4 [0.2, 0.7]	0 [0.0, 0.3]	1977	9.2 [0.6, 62.8]	90.8 [37.2, 99.4]	12
Pakistan 2017–18	50.2 [48.8, 51.6]	16.1 [14.7, 17.6]	27 [25.6, 28.4]	6.1 [5.3, 7.0]	0.6 [0.4, 0.9]	4207	9.2 [6.2, 13.3]	90.8 [86.7, 93.8]	694
Papua New Guinea 2016–18	68.1 [66.9, 69.3]	2.7 [2.0, 3.6]	28.8 [27.5, 30.1]	0.1 [0.1, 0.2]	0.2 [0.1, 0.6]	3806	67 [38.0, 87.1]	33 [12.9, 62.0]	33
Philippines 2022	75.9 [73.8, 77.8]	17.4 [15.5, 19.4]	6.3 [5.5, 7.2]	0.5 [0.3, 0.8]	0 [0.0, 0.1]	2942	9 [1.5, 39.4]	91 [60.6, 98.5]	22
Timor‐Leste 2016	69.2 [67.9, 70.5]	3.4 [2.8, 4.1]	27 [25.8, 28.3]	0.2 [0.1, 0.5]	0.1 [0.0, 0.3]	2990	32.3 [11.8, 63.1]	67.7 [36.9, 88.2]	25
Angola 2015–16	61.8 [60.9, 62.7]	3.1 [2.6, 3.6]	34.1 [33.2, 35.0]	0.5 [0.3, 0.8]	0.4 [0.2, 0.8]	5652	46.2 [27.8, 65.7]	53.8 [34.3, 72.2]	125
Benin 2017–18	64.5 [63.7, 65.2]	4.3 [3.9, 4.7]	30.3 [29.5, 31.1]	0.6 [0.4, 0.8]	0.4 [0.3, 0.5]	5744	39.2 [30.1, 49.1]	60.8 [50.9, 69.9]	125
Burkina Faso 2021	88.4 [87.4, 89.3]	6.1 [5.4, 6.9]	5.4 [4.8, 6.1]	0.1 [0.0, 0.2]	0 [0.0, 0.1]	4830	35.8 [0.0, 100.0]	64.2 [0.0, 100.0]	6
Cameroon 2018–19	64.4 [63.4, 65.4]	3.0 [2.7, 3.5]	32 [30.9, 33.0]	0.3 [0.2, 0.5]	0.3 [0.1, 0.5]	4099	42.7 [26.9, 60.0]	57.3 [40.0, 73.1]	60
Chad 2014–15	60.0 [59.5, 60.5]	1.1 [0.9, 1.3]	38.4 [37.9, 39.0]	0.2 [0.1, 0.4]	0.3 [0.2, 0.4]	6994	54.8 [36.1, 72.2]	45.2 [27.8, 63.9]	85
Comoros 2012	58.0 [56.3, 59.6]	8.1 [6.8, 9.6]	31.6 [29.9, 33.4]	1.5 [1.0, 2.3]	0.8 [0.5, 1.4]	1394	34.9 [21.5, 51.3]	65.1 [48.7, 78.5]	74
Congo (Brazzaville) 2011–12	69.0 [67.5, 70.4]	5.1 [4.3, 6.0]	25.1 [24.0, 26.2]	0.5 [0.3, 0.9]	0.3 [0.2, 0.6]	3564	38.7 [22.6, 57.6]	61.3 [42.4, 77.4]	70
Congo Democratic Republic 2013–14	58.2 [57.3, 59.0]	4.2 [3.6, 4.9]	36.1 [35.2, 37.1]	0.8 [0.6, 1.0]	0.8 [0.6, 1.1]	7536	51.6 [43.0, 60.1]	48.4 [39.9, 57.0]	280
Cote d'Ivoire 2021	84.9 [83.5, 86.2]	8.0 [7.0, 9.2]	6.6 [5.9, 7.5]	0.2 [0.1, 0.3]	0.2 [0.1, 0.6]	3967	61 [23.0, 89.1]	39 [10.9, 77.0]	23
Ethiopia 2016	68.3 [66.6, 69.9]	1.7 [1.4, 2.0]	29.8 [28.1, 31.6]	0.2 [0.1, 0.3]	0.1 [0.0, 0.2]	4469	26 [12.2, 47.0]	74 [53.0, 87.8]	24
Gabon 2019–21	66.6 [65.2, 67.9]	7.2 [6.3, 8.3]	24.8 [23.3, 26.4]	1.0 [0.6, 1.5]	0.4 [0.2, 0.8]	2611	30.9 [16.2, 51.0]	69.1 [49.0, 83.8]	83
Gambia 2019–20	68.9 [68.0, 69.9]	3.0 [2.5, 3.6]	27.6 [26.5, 28.6]	0.3 [0.2, 0.6]	0.2 [0.1, 0.3]	3246	35 [16.9, 58.8]	65 [41.2, 83.1]	37
Ghana 2022–23	74.9 [73.3, 76.5]	19.0 [17.5, 20.5]	5.4 [4.8, 6.1]	0.5 [0.3, 0.8]	0.2 [0.1, 0.4]	3638	24.5 [9.6, 49.7]	75.5 [50.3, 90.4]	35
Guinea 2018	69.1 [68.2, 70.0]	2.4 [2.0, 2.8]	28.1 [27.2, 29.0]	0.2 [0.1, 0.4]	0.2 [0.1, 0.3]	3122	47.2 [31.4, 63.5]	52.8 [36.5, 68.6]	33
Kenya 2022	76.8 [75.3, 78.1]	15.5 [14.3, 16.9]	6.8 [6.3, 7.4]	0.8 [0.6, 1.1]	0.1 [0.1, 0.2]	7101	12.7 [6.4, 23.4]	87.3 [76.6, 93.6]	94
Lesotho 2023–24	73.7 [71.0, 76.2]	22.6 [20.0, 25.5]	3 [2.0, 4.6]	0.4 [0.2, 1.0]	0.2 [0.1, 0.8]	998	37.6 [6.1, 84.8]	62.4 [15.2, 93.9]	9
Liberia 2019–20	73.8 [72.4, 75.1]	4.5 [3.7, 5.6]	21 [19.6, 22.4]	0.3 [0.1, 0.7]	0.4 [0.2, 0.8]	2155	57.8 [26.1, 84.1]	42.2 [15.9, 73.9]	38
Madagascar 2021	73.4 [72.3, 74.4]	2.3 [2.0, 2.8]	24.1 [23.0, 25.2]	0.1 [0.1, 0.3]	0.1 [0.1, 0.2]	5057	44.2 [25.4, 64.8]	55.8 [35.2, 74.6]	29
Malawi 2015–16	74.4 [73.7, 75.2]	5.2 [4.8, 5.7]	19.4 [18.7, 20.2]	0.6 [0.4, 0.7]	0.3 [0.2, 0.4]	6871	35.7 [25.1, 47.9]	64.3 [52.1, 74.9]	148
Mali 2018	63.8 [62.8, 64.7]	2.1 [1.8, 2.5]	33.7 [32.8, 34.7]	0.2 [0.2, 0.4]	0.2 [0.1, 0.3]	4308	40.5 [21.6, 62.8]	59.5 [37.2, 78.4]	41
Mauritania 2019–21	63.2 [62.3, 64.0]	4.5 [4.1, 5.1]	31.1 [30.3, 32.0]	0.8 [0.6, 1.2]	0.3 [0.2, 0.5]	4657	28 [17.8, 41.1]	72 [58.9, 82.2]	128
Mozambique 2022–23	87.8 [86.8, 88.7]	4.8 [4.2, 5.5]	7.3 [6.4, 8.2]	0.1 [0.0, 0.2]	0 [0.0, 0.1]	3926	23.1 [1.1, 89.2]	76.9 [10.8, 98.9]	8
Namibia 2013	69.4 [67.9, 70.8]	13.0 [11.8, 14.2]	15.8 [14.7, 17.0]	1.6 [1.2, 2.0]	0.3 [0.2, 0.6]	2009	16.8 [9.9, 27.1]	83.2 [72.9, 90.1]	88
Niger 2012	60.2 [59.5, 60.9]	1.1 [0.9, 1.4]	38.4 [37.6, 39.1]	0.1 [0.1, 0.2]	0.2 [0.1, 0.3]	5302	67.6 [53.4, 79.2]	32.4 [20.8, 46.6]	39
Nigeria 2018	63.4 [62.9, 63.9]	2.3 [2.0, 2.5]	33.7 [33.2, 34.3]	0.4 [0.3, 0.5]	0.2 [0.2, 0.3]	13 363	41.1 [32.4, 50.3]	58.9 [49.7, 67.6]	199
Rwanda 2019–20	65.3 [64.3, 66.3]	12.4 [11.6, 13.3]	19.6 [18.7, 20.6]	2.3 [1.9, 2.7]	0.4 [0.2, 0.5]	3369	13.5 [9.5, 18.9]	86.5 [81.1, 90.5]	217
Senegal 2023	83.1 [81.7, 84.4]	10.5 [9.3, 11.8]	6 [5.4, 6.8]	0.2 [0.1, 0.4]	0.2 [0.1, 0.4]	3892	43.6 [18.8, 72.1]	56.4 [27.9, 81.2]	22
Sierra Leone 2019	73.6 [72.6, 74.6]	3.5 [2.8, 4.3]	22.4 [21.6, 23.3]	0.4 [0.2, 0.8]	0.1 [0.0, 0.2]	4081	19.3 [8.2, 39.3]	80.7 [60.7, 91.8]	49
South Africa 2016	65.4 [63.2, 67.6]	21.6 [19.8, 23.6]	10.3 [9.1, 11.6]	2.1 [1.5, 3.0]	0.6 [0.3, 1.1]	1414	21.1 [11.8, 34.8]	78.9 [65.2, 88.2]	94
Tanzania 2022	80.8 [79.3, 82.1]	9.9 [8.8, 11.1]	8.6 [7.8, 9.5]	0.4 [0.2, 0.6]	0.3 [0.2, 0.6]	4506	44.4 [28.9, 61.2]	55.6 [38.8, 71.1]	45
Togo 2013–14	69.2 [68.1, 70.2]	5.7 [5.0, 6.5]	24.3 [23.2, 25.3]	0.6 [0.4, 0.8]	0.3 [0.1, 0.4]	2789	30.3 [18.3, 45.8]	69.7 [54.2, 81.7]	55
Uganda 2016	63.1 [62.3, 63.8]	5.3 [4.8, 5.8]	30.4 [29.6, 31.3]	0.8 [0.7, 1.0]	0.4 [0.3, 0.6]	6178	34.5 [27.0, 42.9]	65.5 [57.1, 73.0]	189
Zambia 2018–19	71.7 [70.7, 72.7]	4.3 [3.7, 5.1]	23.3 [22.4, 24.2]	0.5 [0.3, 0.8]	0.2 [0.1, 0.4]	4017	31.5 [16.9, 51.0]	68.5 [49.0, 83.1]	68
Zimbabwe 2015	74.6 [73.4, 75.8]	4.9 [4.3, 5.6]	19.8 [18.7, 20.9]	0.5 [0.3, 0.8]	0.2 [0.1, 0.4]	2524	27.3 [13.7, 47.1]	72.7 [52.9, 86.3]	46

Abbreviations: CI, confidence interval; VBAC, vaginal birth after cesarean section.

3.1. Prevalence of VBAC and repeat CS

As a proportion of all births, the percentage of repeat CS ranged from 0.1% (95% CI: 0.0–0.2) in Mozambique and Burkina Faso to 11.0% (10.4–11.7) in Egypt (Table 1). VBAC as a proportion of all births ranged from 0.0% (0.0–0.1) in Bangladesh to 0.8% (0.6–1.1) in the Democratic Republic of Congo (Table 1).

We observed noteworthy variability in the prevalence of VBAC and repeat CS across countries, as a proportion of births among women with a prior CS. The countries with the lowest prevalence of VBAC and highest prevalence of repeat CS included Turkey (VBAC prevalence: 0.5%; 95% CI: 0.1–3.5), Bangladesh (2.1%; 0.5–8.6), Egypt (3.1%; 2.2–4.3), Colombia (5.5%; 3.3–9.1), and Dominican Republic (6.3%; 2.1–17.0, Table 1). The countries with the highest prevalence of VBAC and lowest prevalence of repeat CS include Niger (VBAC prevalence: 67.6%; 95% CI: 53.4–79.2), Papua New Guinea (67%; 38.0–87.1), Cote d'Ivoire (61%; 23.0–89.1), Chad (54.8% 36.1, 72.2), and Liberia (57.8%; 26.1–84.1, Table 1).

Countries with higher population CS rates tend to have lower VBAC rates and countries with low population CS rates tend to have higher VBAC rates (Figure 1). Most countries with population CS rates at or below 10% had VBAC prevalences ranging between 30% and 50%. Apart from Senegal, Tanzania, and Lesotho, with VBAC prevalence of 43.6% (18.8–72.1), 44.4% (28.9–61.2), and 37.6% (6.1–84.8), respectively, all countries with CS rates above 10% had VBAC prevalence below 30% and most countries' prevalence were below 20%. Countries with population CS rates between 10% and 25%, such as India and Rwanda, had higher VBAC prevalence (12.2% and 13.5%, respectively) than countries with population CS rates exceeding 40% (i.e., Bangladesh, Egypt, Dominican Republic), whose VBAC prevalence approached zero. Considering the diametric relationship between VBACs and repeat CS among women with a history of prior CS, these findings also signal that countries with the highest population CS rates tended to have the highest repeat CS rates as well.

FIGURE 1.

Association between the country population cesarean section (CS) rates and vaginal birth after cesarean section (VBAC) prevalence.

3.2. Interpreting repeat CS trends: As a share of births (overall and by CS), over time, and across facility level and type

Having observed an inverse association between VBAC and the population CS rate, we explored differences in repeat CS across countries in three ways: as a share of total births and as a share of the overall CS rate, over time, and in relation to health facility type and health facility level. These data indicate that repeat procedures make up a small share of all births (Figure 2a). However, the share of repeat CS among all births is greater for countries with CS rates above 20% than for countries with CS rates below 20%. Although there is cross‐country variation in repeat CS as a share of overall CS (Figure 2b), repeat CS as a proportion of all CS does not appear to vary among countries with higher versus lower population CS rates.

FIGURE 2.

Repeat cesarean section (a) as a proportion of all births and (b) as a proportion of all cesarean sections. Countries in this figure are sorted by population cesarean section (CS) rate (shaded gray), with the highest being at the top and the lowest being at the bottom of the chart.

Of the 28 countries with a population CS rate of ≥8% and with data available from two or three surveys, the proportion of women experiencing a repeat CS increased over time for 26 countries (Figure 3). Although the increases across successive surveys are not statistically significant for most countries, statistically significant differences were observed in Colombia, Armenia, Egypt, India, Pakistan, Kenya, Guatemala, and Namibia.

FIGURE 3.

Temporal trends in repeat cesarean section prevalence by country survey year. Labeled red if change between current survey and prior survey P < 0.05.

Although the odds of repeat CS were higher in private sector facilities relative to public sector facilities in 13 countries assessed, statistically significant associations were only noted across four countries (Table 2): Egypt (2.85; 95% CI: 1.27–6.38), Guatemala (40.38; 5.30–307.97), India (2.77; 2.21–3.48), and Pakistan (3.41; 1.36–8.56). Similarly, the odds of repeat CS were higher in hospitals relative to non‐hospital facilities, but such associations were only statistically significant for six countries: Indonesia (21.60; 95% CI: 7.33–63.64), Pakistan (94.38; 7.86–1133.57), Rwanda (4.59; 1.84–11.45), Tanzania (17.44; 1.83–166.50), Namibia (20.32; 1.50–274.62), and South Africa (13.01; 1.16–145.72).

TABLE 2.

Association between facility level (hospital/non‐hospital) and facility type (private/public) and repeat cesarean delivery.

Country	Private sector facility	N	Hospital‐level facility	N
	OR (95% CI)		OR (95% CI)
Egypt	2.85 (1.27–6.38)	1740	1.10 (0.50–2.46)	1740
Gabon	2.16 (0.26–18.16)	74	1.85 (0.09–40.16)	74
Guatemala	40.38 (5.30–307.97)	411	0.59 (0.19–1.80)	411
Honduras	1.99 (0.24–16.76)	184	1.81 (0.40–8.20)	184
India	2.77 (2.21–3.48)	6206	1.37 (0.97–1.93)	6206
Indonesia	1.28 (0.49–3.35)	250	21.60 (7.33–63.64)	250
Jordan	0.45 (0.07–2.84)	201	1.87 (0.42–8.31)	201
Kenya	2.93 (0.27–32.28)	72	6.37 (0.73–55.33)	72
Comoros	NA a		1.91 (0.34–10.79)	61
Lesotho	1.29 (0.07–25.11)	25	1.18 (0.13–11.03)	25
Maldives	3.33 (0.31–35.77)	126	NA a
Namibia	NA a		20.32 (1.50–274.62)	91
Pakistan	3.41 (1.36–8.56)	512	94.38 (7.86–1133.57)	512
Rwanda	4.69 (0.60–36.81)	211	4.59 (1.84–11.45)	211
Tanzania	2.91 (0.21–40.02)	38	17.44 (1.83–166.50)	38
South Africa	6.65 (0.63–70.60)	77	13.01 (1.16–145.72)	77

Abbreviations: CI, confidence interval; CS, cesarean section; OR, odds ratio; VBAC, vaginal birth after cesarean section.

^a NA signifies that the odds ratio could not be modeled because all births belonged to one category (e.g., all births at private sector facilities were repeat CS and none were VBAC).

4. DISCUSSION

To our knowledge, this is the first study to report nationally representative VBAC and repeat CS prevalence estimates across multiple LMICs. The results highlight the sheer variability of VBAC across countries: whereas up to two‐thirds of eligible deliveries were VBACs in Papua New Guinea and Niger; virtually no VBACs occurred in Turkey and Bangladesh. This variation may be explained, in part, by the population CS rate: countries with lower population CS rates had higher VBAC rates, and countries with higher population CS rates had lower VBAC rates. In countries with higher population CS rates, the prevalence of repeat CS tended to increase over time; and repeat CS tended to happen more frequently at private sector facilities and hospitals – although these relationships were not statistically significant across many countries.

Where population CS rates have reached or exceeded a certain point, it unfortunately appears as if the saying “once a cesarean, always a cesarean” describes common practice in many LMICs. ²³ The predominance of repeat CS we see in these data among countries with high population CS rates is consistent with what has been previously documented at sites in India, Guatemala, Pakistan, and Bangladesh. ¹⁹ Our observation of increasing repeat CS over time across multiple countries reflects trends that may imply CS overuse in birthing populations. ²⁰ The increased proportion of repeat CS in private sector facilities, albeit not statistically significant across many countries, is not surprising given how extensively similar trends have been documented for CS in general (i.e., primary and repeat). ²⁴ , ²⁵ , ²⁶

For countries with low population CS rates and high VBAC rates, it is important to question whether these VBACs are occurring in safe settings. A safe TOLAC requires supervision or labor monitoring coupled with fetal heart rate monitoring so that any early signs of uterine rupture can be detected and managed. ²⁷ This ultimately means that higher‐level facilities offering comprehensive emergency obstetric and newborn care (CEmONC) are likely better positioned to perform TOLAC safely as they are more likely to offer 24‐h surgical and anesthesia services. ²⁷ Assuming that population CS rates are low because access to the staffing, equipment, and supplies needed for CS (and thus CEmONC) provision are limited, ²⁸ it stands to reason that many or most of VBACs occurring in these countries may be happening by default, not following a supervised TOLAC with active labor monitoring, but in facilities lacking the skilled personnel and equipment needed to manage complications.

Interestingly, the countries with the highest VBAC prevalence, for which we posit that VBACs are occurring by default, also tend to have higher maternal mortality ratio (MMR) estimates. Niger, Papua New Guinea, Cote d'Ivoire, Liberia, and Chad, which were the countries with the highest prevalence of VBAC in our analysis, had 350, 189, 359, 628, and 748 maternal deaths per 100 000 child births, respectively. Whereas Turkey, Bangladesh, Egypt, Colombia, and Dominican Republic, the countries with the lowest prevalence of VBAC in our analysis, had 15, 115, 17, 59, and 124 maternal deaths per 100 000 live births, respectively. ²⁹ This correlation, though illustrative, does not necessarily quantify the specific contributions of VBAC complications to the maternal death rate. That said, some of the challenges in CEmONC provision described above may be impeding both the safe provision of TOLAC/VBAC as well as maternal mortality prevention efforts at‐large.

4.1. Strengths and limitations

The strengths of the present analysis include the use of nationally representative data whose measurement is consistent across countries. In addition to reporting prevalence estimates, this analysis attempts to describe national‐level prevalence using explanatory variables. Lastly, we introduce a novel way of categorizing childbirths into five mutually exclusive groups, which can open further analyses into birth patterns.

Despite these strengths, many limitations persist. TOLAC remains a black box because these data originate from household surveys. The dataset sample sizes were either prohibitively small for some analyses (e.g., relationships between facility level/type and repeat CS), or the datasets themselves lacked the variables needed to perform deeper analyses (e.g., facility readiness to perform supervised TOLAC). Because the survey's recall period covers either 3 or 5 years preceding the administration of the survey, it is also possible for births coded as either first‐time cesarean or first‐time vaginal deliveries to be misclassified.

The low sample sizes mean that the precision of many estimates was low across countries. Confidence bounds were wide for prevalence estimates and associations that were otherwise strong in magnitude were not adequately powered to detect a statistically significant difference. Where population CS rates are low, most cesareans are occurring among women experiencing the procedure for the first time. Therefore, we can infer that the share of women eligible for either a repeat cesarean or VBAC is low. However, several country datasets with population CS rates below 10% have large enough sample sizes to support the observations being made (e.g., Angola, DRC, Nigeria, and Uganda). Where population CS rates are high, the sub‐population of women with prior CS may be substantive; however, most of them experience repeat cesareans. Pooling estimates at the regional or even global levels may have yielded statistically significant trends. However, the utility of pooled estimates may be limited given the extensive cross‐country variability seen in these data.

4.2. Implications for research and practice

These findings underscore the importance of strengthening access to safe and appropriate birthing options for pregnant women having delivered via cesarean in the past. This first involves actively identifying pregnant women with prior CS and following‐up with them to ensure they reach a hospital for safe delivery – regardless of whether it is a VBAC or a planned repeat CS. Discussion of the importance of TOLAC or planned cesarean during antenatal care may currently be a missed opportunity. Recent research demonstrates the impact of such early identification in reducing emergencies and births. ³⁰

In terms of birthing options, TOLAC potentially leading to VBAC may be viable if a variety of conditions are met, including appropriate supervision. ¹⁰ , ³¹ If the enabling environment permits the safe provision of supervised TOLAC, providers ought to counsel patients on the risks and benefits of TOLAC and those of elective repeat cesarean delivery (ERCD), develop TOLAC plans, and document both any counseling notes and management plans in the patient record. ¹⁰ , ³¹ Existing guidelines indicate that most women with one (and possibly two) previous low‐transverse incision(s) may be eligible for TOLAC, whereas women with a high risk of uterine rupture (for example: previous uterine rupture, T‐incision, classical uterine incision, extensive transfundal uterine surgery) will require ERCD. The safe provision of TOLAC requires providers to hold specialized training, that protocols are available in facilities, that providers conduct continuous electronic fetal heart monitoring, that blood be available in the facility if needed for transfusion, and that providers be able to perform emergency CS on a 24‐h basis if required. ²⁷ , ³¹ Moreover, the importance of adequate facility readiness becomes more acute among higher‐volume facilities as the additional monitoring needs of women undergoing supervised TOLAC draw on a strained pool of resources. Ensuring that facilities are ready to perform either supervised TOLAC or repeat CS safely will indeed require investments into human resources, infrastructure, and policy development and implementation.

Research efforts ought to focus on characterizing whether an enabling environment is available to support health facility readiness to perform supervised TOLAC. Where unmet need persists, it is important to understand whether supervised versus unsupervised TOLACs are happening, why, and what factors incentivize repeat CS over TOLAC. It would be useful to explore the extent to which providers are trained in monitoring women under a supervised TOLAC protocol, what equipment and supplies facilities have, and whether facilities are able to handle complications. Some of these questions may be answered through the ecological linkage of Service Provision Assessment and DHS datasets, as was done previously for CS. ³² Future studies may also address post‐VBAC or post‐repeat CS follow‐up or continuity of care.

For countries with higher population CS rates, more research is needed to understand why repeat CS is going up over time, or why repeat CS tends to occur more frequently at private facilities. Studies ought to explore whether the mechanisms driving repeat CS trends follow those of CS in general (e.g., fear of litigation as a driver of primary CS), ¹⁵ or represent their own distinct phenomena (e.g., additional readiness requirements needed to perform TOLAC safely). In these contexts, it is important to understand what policies and procedures are in place to support the provision of supervised TOLAC (e.g., counseling expecting mothers on VBAC as a birthing option and its risks and benefits). More research is also needed to understand the patient and provider biases and incentives favoring repeat CS (e.g., time convenience and potential income from CS over labor monitoring, preferences and mistrust in systems among patients). ⁵

The various safety considerations and evidence gaps for VBAC and repeat CS in LMICs described above, however, highlight the importance of preventing unnecessary CS use in the first place: fewer primary CS will result in fewer subsequent pregnancies requiring either TOLAC or repeat CS.

AUTHOR CONTRIBUTIONS

FK, SR, VT, and RS co‐developed the research questions. SR developed the analysis plan, pulled data from the DHS, and led statistical analyses. FK contributed to statistical analyses and data visualization. FK and SR co‐wrote the first draft of this article, with clinical inputs from RS. All authors reviewed the manuscript and approved the final version.

FUNDING INFORMATION

The research activities described in this paper were originally funded by the United States Agency for International Development (USAID) through the MOMENTUM Safe Surgery in Family Planning and Obstetrics and Demographic and Health Survey Program awards (cooperative agreement nos. 7200AA20CA00011 and 7200AA18C00083, respectively). This paper was subsequently completed independently by the authors. The contents do not necessarily reflect the views of USAID or the United States Government.

CONFLICT OF INTEREST STATEMENT

None declared.

Supporting information

Data S1.

DATA AVAILABILITY STATEMENT

Data used in this study are available to registered users from the Demographic and Health Survey (DHS) Program website and code used to perform the analyses are available on request to the authors.