SEARCH

SEARCH BY CITATION

Keywords:

  • Caesarean section;
  • fertility;
  • sub-Saharan Africa

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Conclusion
  8. Contributors
  9. Funding
  10. References

Objective  To determine the impact of caesarean section on fertility among women in sub-Saharan Africa.

Design  Analysis of standardised cross-sectional surveys (Demographic and Health Surveys).

Setting  Twenty-two countries in sub-Saharan Africa, 1993–2003.

Sample  A total of 35 398 women of childbearing age (15–49 years).

Methods  Time to subsequent pregnancy was compared by mode of delivery using Cox proportional hazards regression models.

Main outcome measures  Natural fertility rates subsequent to delivery by caesarean section compared with natural fertility rates subsequent to vaginal delivery.

Results  The natural fertility rate subsequent to delivery by caesarean section was 17% lower than the natural fertility rate subsequent to vaginal delivery (hazard ratio = 0.83, 95% CI 0.73–0.96, P < 0.01; controlling for age, parity, level of education, urban/rural residence and young age at first intercourse). Caesarean section was also associated with prior fertility and desire for further children: among multiparous women, an interval ≥3 versus <3 years between the index birth and the previous birth was associated with higher odds of caesarean section at the index birth (OR = 1.4, 95% CI 1.1–1.7, P= 0.005); among all women, the odds of desiring further children were lower among women who had previously delivered by caesarean section (OR = 0.67, 95% CI 0.54–0.84, P < 0.001). Caesarean section did not appear to increase the risk of a subsequent pregnancy ending in miscarriage, abortion or stillbirth.

Conclusions  Among women in sub-Saharan Africa, caesarean section is associated with lower subsequent natural fertility. Although this reflects findings from developed countries, the roles of pathological and psychological factors may be quite different because a much higher proportion of caesarean sections in sub-Saharan Africa are emergency procedures for maternal indication.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Conclusion
  8. Contributors
  9. Funding
  10. References

Several studies have investigated the relationship between caesarean section and subsequent fertility in high-income countries. The most recent of these studies have shown that there is an association between caesarean section and reduced subsequent fertility, but that the relationship is complex, involving pathological and psychological factors.1 In low-income countries, the relationship between caesarean section and subsequent fertility remains unexplored.

The objective of our study was to investigate whether there is an association between caesarean section and subsequent fertility in sub-Saharan Africa, a region within which many countries have a low or a very low caesarean section rate.2 The precise definition of a ‘low’ caesarean section rate is the subject of debate; some authors have suggested an ‘optimum’ rate of 5% to meet obstetric need, others a ‘minimum’ rate of 1–2%.3–6 Whatever the outcome of this debate, and despite rising rates of elective caesarean sections among a very small minority of affluent women, most caesarean sections in sub-Saharan Africa are emergency operations for extreme maternal indication.3 We hypothesised that long-term pathological and psychological effects of the long and difficult labour and often severe obstetric complications that precede these emergency operations, plus the effects of the procedure itself, would be associated with reduced subsequent fertility.

The aim of our study was to inform research into the long-term consequences of severe obstetric complications in resource-poor settings, a neglected area of research, despite its relevance to safe motherhood strategies. In such settings, loss of fertility will have a huge impact on a woman's health and wellbeing, and very few if any women will have access to fertility treatment.7–10

Our study is based on a measure of ‘natural fertility’, which is affected by pathology (infertility or subfertility) and psychology (reproductive behaviour excluding the use of contraception or induced abortion). We analysed natural fertility rates and factors associated with natural fertility according to whether the index birth (most recent live birth) was by caesarean section or by vaginal delivery. We discuss our findings in the light of evidence from studies conducted in high-income countries.

Methods

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Conclusion
  8. Contributors
  9. Funding
  10. References

We used Demographic and Health Surveys (DHS) data from 22 sub-Saharan African countries covering a 10-year period (1993–2003), focusing on questions concerning the previous 3 or 5 years of reproductive history, which were asked of all women aged 15–49 years in sample households.11 Some countries were surveyed more than once in the 10-year period. Because each survey is a random sample independent of previous surveys, we included multiple surveys from single countries to increase the power of our study. Questionnaires and estimates of sampling errors for each survey, together with details of general survey methodology and data processing, are published by DHS.12

We calculated a time since last live birth interval for each woman for whom the mode of delivery at last live birth (the index birth) had been recorded (corresponding to the question ‘Was (NAME) delivered by caesarean section’). The interval began at the time of the index birth and was censored at the time of interview. The ‘failure’ event for survival analysis was subsequent onset of a current or a terminated pregnancy. For women who were currently pregnant, the duration of pregnancy was subtracted from the interval. For women who reported a terminated pregnancy since the index birth (corresponding to the question ‘Have you ever had a pregnancy that miscarried, was aborted, or ended in a stillbirth?’), the interval was recalculated as the time between the index birth and the date of termination minus the duration of the terminated pregnancy if reported or the mean duration (4 months) if unreported. Women who reported more than one terminated pregnancy since the index birth were excluded because dates of termination were only recorded for the most recent terminated pregnancy.

Women with a caesarean section before the index birth were excluded as were women currently breastfeeding, currently abstaining from sexual intercourse or currently amenorrhoeic. Women who were currently using contraception were also excluded (unless otherwise stated) because preliminary analysis of fertility rates showed that subtraction of the reported duration of use of the current method from the interval was unlikely to adjust sufficiently for long-term contraceptive use. South Africa, with a caesarean section rate of 15% in 1998, was excluded from our study on the assumption that this rate was likely to include a large number of elective caesarean sections.

We used Cox regression (maximum-likelihood proportional hazards, Breslow methods for ties) to estimate, for each survey and for all surveys pooled in a single data set, the hazard ratio (HR) for pregnancy following birth by caesarean compared with vaginal delivery, adjusting for age (5-year intervals from 15 to 49), parity (1, 2, 3, 4 and 5+), level of education (none, primary, secondary or higher), urban or rural residence and young age (<14 years) at first intercourse (a known risk factor for infertility in sub-Saharan Africa).13 We used multiple logistic regression on the pooled data set, adjusting for these same factors, to estimate odds ratios for pregnancy in the 5-year period following the index birth and to explore other factors associated with fertility and mode of delivery. All methods incorporated adjustments for complex survey designs (sample weights and clustering). All analyses were performed using Stata v8 (StataCorp LP, College Station, TX, USA).

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Conclusion
  8. Contributors
  9. Funding
  10. References

Surveys

DHS sample sizes ranged from 3040 (Côte d’Ivoire, 1998) to 15 367 women (Ethiopia, 2000)—1Table 1. Surveys were conducted over a period of several months; the earliest data in our study were collected in Kenya during 1993 and the most recent in Burkina Faso during 2003. The caesarean section rate ranged from 0.47% (21/4523) in Chad (1996) to 7.60% (206/2497) in Zimbabwe (1999). The number of women in each sample contributing to the analysis ranged from 376 (Ghana, 1993) to 2730 (Mali, 2001). Of the 270 353 women in the pooled data set, 35 398 contributed to the analysis (1Figure 1). Among these 35 398 women, there were 16 876 pregnancies (at time of interview or terminated since the last live birth) in 84 662 years at risk.

Table 1.  HR for natural fertility rates subsequent to caesarean section compared with vaginal delivery (HR adjusted for age, parity, level of education, urban/rural residence and young age at first intercourse)
CountryYearSample sizeCaesarean section rate (%)HR (caesarean section vs vaginal delivery)
nHR95% CIP
Benin199654912.35020.30.11.50.16
200162193.910970.70.41.30.24
Burkina Faso199964451.110420.60.31.20.15
200312 4770.818861.00.33.00.99
Cameroon199855012.74261.20.43.60.79
Central African Republic199458841.95001.30.62.70.57
Chad199674540.515880.90.32.70.85
Cote d’Ivoire199480991.88811.00.34.10.95
199830402.44650.30.11.40.14
Ethiopia200015 3670.719680.60.14.00.60
Gabon200061836.38770.80.51.40.41
Ghana199345624.83761.20.52.60.73
199848434.06471.20.52.70.66
200356914.47391.00.52.10.94
Guinea199967532.213390.60.31.30.19
Kenya199375406.08951.10.71.80.60
199878817.05461.00.71.70.86
200381954.99161.10.71.80.78
Malawi200013 2203.019640.60.41.00.07
Mali199697040.911570.90.42.40.86
200112 8491.127300.90.51.90.87
Mozambique199787792.38101.60.64.30.38
Niger199875770.610640.50.21.70.26
Nigeria199998103.97200.80.41.60.58
200376202.012960.90.23.70.92
Rwanda200010 4212.210621.20.52.50.71
Tanzania199681204.95801.20.52.70.67
199940293.76070.70.21.90.43
Togo199885692.26090.70.22.00.46
Uganda199570702.811610.80.41.60.56
200172463.111510.70.41.30.24
Zambia199680212.115080.50.21.60.26
200276582.412390.70.31.40.29
Zimbabwe199461286.03810.70.21.90.45
199959077.66690.80.41.50.44
Pooled data270 3532.835 3980.80.71.00.01
image

Figure 1. Numbers of women in the pooled data set (all surveys combined) included in the final analysis.

Download figure to PowerPoint

Subsequent fertility

The difference in the fertility rate subsequent to caesarean section compared with vaginal delivery (adjusted for age, parity, level of education, urban/rural residence and young age at first intercourse) was not statistically significant at the 5% level for any of the individual country surveys (Table 1). For the pooled data set (all surveys combined), the fertility rate subsequent to caesarean section was 17% lower than the natural fertility rate subsequent to vaginal delivery (HR = 0.83, 95% CI 0.73–0.96, P < 0.01; adjusted for age, parity, level of education, urban/rural residence and young age at first intercourse). The effects of each variable in this model are shown in 2Table 2.

Table 2.  Factors associated with natural fertility rates subsequent to index birth (multivariate model—HR adjusted for all other factors)
FactorHRP value95% CI
Caesarean section (vs vaginal delivery) at index birth 0.830.0090.730.96
Age group (vs age 15–19 years)20–240.670.0000.620.73
25–290.470.0000.430.51
30–340.290.0000.260.32
35–390.190.0000.170.22
40–440.090.0000.080.10
45–490.050.0000.040.06
Parity (vs primiparous)21.280.0001.201.36
31.450.0001.351.56
41.700.0001.581.83
5+2.200.0002.032.37
Education (vs none)Primary1.080.0001.041.13
Secondary or higher0.980.4890.921.04
Rural residence (vs urban) 1.360.0001.291.43
Age <14 years at first intercourse (vs ≥14 years) 0.840.0000.780.90

The unadjusted odds ratio for becoming pregnant within 5 years of delivering by caesarean section compared with vaginal delivery was 0.71 (95% CI 0.60–0.85, P < 0.001); adjusted for age, parity, level of education, urban/rural residence and young age at first intercourse, the odds ratio was 0.75 (95% CI 0.62–0.89, P= 0.002). The odds of taking >1 year to conceive were higher among women of parity ≥2 who had delivered by caesarean section compared with vaginal delivery (OR = 1.9, 95% CI 1.1–3.1), but this effect was not found among primiparous women, among whom the effect for subsequent pregnancy was not significant (HR = 1.0, 95% CI 0.8–1.2).

Fertility prior to caesarean section

For women of parity ≥2, the risk of caesarean section at the index birth was associated with the length of the interval between that birth and the birth preceding it. Higher odds of caesarean section were found if the most recent between-birth interval was ≥3 versus <3 years (OR = 1.4, 95% CI 1.1–1.7, P= 0.005; adjusted for age, parity, level of education, urban/rural residence and young age at first intercourse). The risk of caesarean section at the index birth was associated with neither earlier between-birth intervals nor the interval between date of first intercourse and date of index birth among women for whom the index birth was their first.

Miscarriage, abortion or stillbirth

Delivery by caesarean section did not appear to increase the risk of a subsequent pregnancy ending in miscarriage, abortion or stillbirth (5.8% [42/730] caesarean section delivery versus 3.8% [1285/34 079] vaginal delivery, adjusted OR = 1.3, 95% CI 0.8–1.9) irrespective of whether women currently using contraception were included or not. In a separate analysis, we found no association between caesarean section and risk of multiple subsequent pregnancy terminations.

Contraception and sterilisation

Current contraceptive use was much more prevalent among women whose previous delivery had been by caesarean section (43% [568/1319] versus 21% [9360/44 339], P < 0.001). The method with the largest difference was female sterilisation (10% [134/1319] versus 1% [365/44 339], P < 0.001). Some surveys recorded the date of sterilisation; among women whose previous delivery was by caesarean section, 42% (100/240) had their sterilisation in the same month as the delivery. Among women whose previous birth was vaginal, this proportion was 4% (122/3239).

Women reporting current use of a contraceptive method were excluded from our analyses because most surveys reported only the duration of use of the current method. The fertility rate for women who reported current use of contraception was much lower than that for women who were not currently using contraception even when the duration of use of the current method was subtracted from the time since last live birth interval (data not shown), suggesting that current users were more likely to have used contraception for other (unreported) periods since the last birth. Their inclusion would therefore exaggerate the effect of a previous caesarean section on subsequent fertility, given the higher prevalence of current contraceptive use (including sterilisation) among women with a previous delivery by caesarean. Indeed, the effect of caesarean section on fertility when women currently using contraception were included (HR = 0.80, 95% CI 0.69–0.91, P= 0.001) was stronger than that found with the exclusion of these women.

Fertility preferences

DHS provide data on women's fertility preferences; we found no significant differences by mode of delivery in the proportions of women who did not desire further children (caesarean section 29% versus vaginal 26%) or who desired children within 2 years (caesarean section 26% versus vaginal 27%) or after 2 years (caesarean section 33% versus vaginal 35%). However, after controlling for age, parity, level of education, urban/rural residence and young age at first intercourse, the odds of desiring further children were lower among women who had previously delivered by caesarean section (OR = 0.67, 95% CI 0.54–0.84, P < 0.001). Inclusion of women currently using contraception did not alter the magnitude or significance of this odds ratio (OR = 0.68, 95% CI 0.57–0.81, P < 0.001).

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Conclusion
  8. Contributors
  9. Funding
  10. References

The present study is the first to investigate the relationship between caesarean section and fertility across sub-Saharan Africa. Our objective was to find out whether, all else being equal and subject to the constraints of retrospective survey data, natural fertility rates were different according to the mode of delivery at the most recent live birth. Our key finding is that delivery by caesarean section is associated with reduced natural fertility.

There is strong evidence to support the assertion that caesarean section is a proxy indicator of severe obstetric complication in resource-poor settings.3,14 It is reasonable to suppose that the pathological and psychological trauma of a long and difficult labour ending in an emergency operation, together with any adverse effects of the operation itself, would lengthen the time before next pregnancy. However, before we can assert that reduced fertility associated with caesarean section is unique to such settings, we must compare our findings with evidence from settings where caesarean section is not such a strong indicator of severe obstetric problems.

Subsequent fertility

In the UK, Murphy et al. found that caesarean section was associated with an increased risk of taking >1 year to conceive from the time of planning a pregnancy compared with vaginal delivery (OR = 1.5, 95% CI 1.1–2.1).15 We found no effect for this outcome for the same interval. However, among women of parity ≥2, we found higher odds of taking >1 year to conceive (OR = 1.9, 95% CI 1.1–3.1) as did Murphy et al. (OR = 3.0, 95% CI 1.7–5.1).15 Our adjusted odds ratio for becoming pregnant within 5 years of delivering by caesarean section compared with vaginal delivery (OR = 0.75, 95% CI 0.62–0.89) is consistent with a study in Brazil, which looked at the odds of becoming pregnant within 35–52 months (OR = 0.80, 95% CI 0.68–0.94).16

Most of the studies that have investigated the association between caesarean section and subsequent pregnancy have recruited primiparous women. Among this group, we found no association between primary mode of delivery and subsequent pregnancy, in contrast with recent findings of Mollison et al.17 That we found an association for multiparous, but not for primiparous, women could reflect greater desire for further children among the latter group, regardless of the first birth experience.

Fertility prior to caesarean section

Murphy et al. also reported an increased risk of delivery by caesarean section among nulliparous women who had been infertile for >3 years (OR = 2.3, 95% CI 1.6–3.3).15 We found no effect among nulliparous women, but found a higher risk of caesarean section among women of parity ≥2 who had not conceived for >3 years compared with <3 years (OR = 1.4, 95% CI 1.1–1.7). Also, in the UK, Pandian et al. reported a higher risk of caesarean section among women with unexplained infertility compared with the general population (OR = 1.5, 95% CI 1.1–1.8).18

Miscarriage, abortion or stillbirth

Several studies in the UK have investigated the association between primary mode of delivery and subsequent pregnancy outcomes: Mollison et al. found a higher risk of ectopic pregnancy but fewer induced abortions;17 Tower et al. found no significant difference in the risk of miscarriage or ectopic pregnancy;19 Smith et al. found higher odds of unexplained stillbirth at ≥34 weeks of gestation during second pregnancy.20 A review by Hemminki of cohort studies, mostly among primiparous women, found no increased risk of induced abortion but found some evidence of higher risks of miscarriage and ectopic pregnancy.21 We found no association between caesarean section and subsequent pregnancy termination (miscarriage, abortion or stillbirth), but found that the accuracy and reliability of DHS data on terminated pregnancies is uncertain and, together with our exclusion of multiple terminated pregnancies, would have constrained our ability to detect an effect.

Contraception and sterilisation

Current contraceptive use and sterilisation were much more prevalent among women whose last birth had been caesarean. The exclusion of current users from our analyses minimises but does not eliminate the effect of more contraceptive use among women who had caesareans. Sterilisation during caesarean section may represent an unmet need for women to plan their family size or a desire by clinicians to prevent future obstetric complications, rather than raise issues of informed consent, but this question merits further research, given the increasing caesarean section rates in developing countries.22–25 We could find no evidence to suggest that women who give birth by caesarean section in sub-Saharan African countries are routinely advised to delay conception.

Fertility preferences

Women who had delivered by caesarean were less likely to desire further children. Whether this reflects a desire to avoid another traumatic labour and delivery could not be determined from the DHS data. More importantly, desire to have further children does not necessarily mean that the woman is trying to conceive. However, our finding is consistent with a study in the UK, which reported that fear of future childbirth was more common among women who had a primary caesarean section.26

Infertility and subfertility

The proportions of women in our study who experienced an infertile period of duration ≥5 years between their last live birth and a current pregnancy (range 4–22%) are consistent with the proportions of women categorised as ‘currently fertile’ who have had at least one long childless interval (≥5 years) followed by at least one birth reported in a study based on DHS data from 27 sub-Saharan African countries (range 5–17%).13 Hence, even with a 5-year follow-up period, we would be cautious in labelling women as ‘secondarily infertile’. However, another study based on DHS data reported high prevalence of secondary infertility in sub-Saharan Africa (range 5–25%).27 A matched case–control study from Zimbabwe found that previous delivery by caesarean section was more common among women with secondary infertility than among fertile controls (OR = 2.8, 95% CI 1.4–5.7); in the same study, secondary infertility was strongly associated with postoperative sepsis (OR = 11.5, 95% CI 3.4–38.4).8

Limitations of our study

DHS record only live births; hence, caesarean sections coincident with perinatal death are not recorded. The exclusion of stillbirths would tend to underestimate the effect of caesarean section on subsequent fertility, given the association between stillbirth and emergency caesarean section and the more severe long-term physical and psychological consequences of a long and painful labour ending in stillbirth. However, this confounding may be countered by a greater desire of women who have had a stillbirth to become pregnant again as soon as possible. We used current or terminated pregnancies as endpoints, but early miscarriages are unlikely to be recorded accurately by DHS. The reliability of reported caesarean section rates derived from DHS data was not an issue for our study, and a recent study shows that DHS estimate caesarean section rates with reasonable accuracy.28 DHS data did not allow us to differentiate spontaneous from instrumental vaginal deliveries.

DHS surveys are vulnerable to reporting and recall error.11 The magnitude and randomness of these errors are unknown, but fertility rates derived from DHS retrospective survey data have been shown to be reasonably accurate when compared with data from a prospective longitudinal study.29 The potential biases introduced by missing data and our inclusion criteria (Figure 1) are also unknown. Women were selected into the survey at random, and a woman is effectively selected at a random point in her reproductive history, meaning that longer birth intervals may tend to be selected more. With an effect of caesarean on interval length, caesarean intervals will tend to be overselected. For our primary objective, the establishment of a relationship between caesarean and interval length, this introduces no bias. But for analyses of risk factors for caesarean, there may be bias where these factors are directly associated with time between pregnancies.

General

Our findings that, in sub-Saharan Africa, caesarean section is associated with lower subsequent natural and overall fertility, lower prior fertility and less desire for further children are not inconsistent with the studies conducted in developed countries. Comparison of our findings with those from studies conducted in the Europe and the USA must be mindful of the differences that exist between and within high-income and low-income settings. In a low-income country, lack of resources and late referral may magnify the risks of morbidity and mortality associated with operative delivery.14 Conversely, any adverse impact of postoperative complications on a woman's long-term health may be greatly outweighed by the risks associated with home delivery without a professional attendant. The impact on psychological factors of differences in fertility, such as childbearing at younger ages, the need to produce many children and ostracisation of infertile women, will be greatly amplified by social and cultural norms, and caesarean section itself may be stigmatised.4 Patterns of infertility, mechanisms for coping with infertility and factors such as contraceptive use subsequent to obstetric complications may be very dissimilar.8,30,31

Conclusion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Conclusion
  8. Contributors
  9. Funding
  10. References

Although caesarean section in low caesarean settings is an indicator of severe obstetric complications, the long-term impact of the complications, and the operation itself, on a woman's fertility do not appear to differ in magnitude from effects found in settings with much higher caesarean section rates. In both the settings, the relationship between caesarean section and fertility is likely to be complex and involve factors such as prior fertility. Further research in low-income settings would be needed to elucidate the relative contributions of pathological and psychological factors to lower fertility following delivery by caesarean section.

Contributors

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Conclusion
  8. Contributors
  9. Funding
  10. References

V.F. and T.M. raised the research question. S.M.C. conducted the research, performed the data analyses and wrote the paper. V.F. and T.M. provided guidance, input and comments during the research process and at each revision of the paper.

Funding

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Conclusion
  8. Contributors
  9. Funding
  10. References

This work was undertaken as part of an international research programme—IMMPACT (Initiative for Maternal Mortality Programme Assessment, http://www.abdn.ac.uk/immpact, funded by the Bill & Melinda Gates Foundation, the Department for International Development, the European Commission and USAID). The funders have no responsibility for the information provided or views expressed in this manuscript. The views expressed herein are solely those of the authors.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Conclusion
  8. Contributors
  9. Funding
  10. References