Prevalence and risk factors for third- and fourth-degree perineal lacerations during vaginal delivery: a multi-country study

Authors


Dr A Koyanagi, Department of Global Health Policy, Graduate School of Medicine, University of Tokyo, 7-3-1 Hongo, Bunkyo-ku, Tokyo 113-0033 Japan. Email koyanagiai@yahoo.com

Abstract

Please cite this paper as: Hirayama F, Koyanagi A, Mori R, Zhang J, Souza J, Gülmezoglu A. Prevalence and risk factors for third- and fourth-degree perineal lacerations during vaginal delivery: a multi-country study. BJOG 2012;119:340–347.

Objective  To investigate the prevalence and risk factors of third- and fourth-degree perineal lacerations in 24, mainly developing, countries.

Design  Analysis using cross-sectional data from the WHO Global Survey on Maternal and Perinatal Health.

Setting  Seven African, nine Asian and eight Latin American countries.

Population  Women at admission to hospital for delivery in 373 facilities between 2004 and 2008.

Methods  We estimated the country-wise prevalence of third- and fourth-degree perineal lacerations, and conducted region-wise multivariate logistic regression analyses to identify its risk factors.

Main outcome measures  Prevalence and risk factors of third- and fourth-degree perineal lacerations.

Results  A total of 214 599 women who underwent vaginal delivery were analysed. The prevalence of third- and fourth-degree perineal lacerations ranged widely across countries [from 0.1% (China, Cambodia, India) to 15.0% (Philippines)] and facilities (from null to 76.3%). After the deletion of facilities reporting no third- or fourth-degree perineal lacerations, and also highly outlying facilities, the range in prevalence was 0.1% (Uganda) to 1.4% (Japan). Forceps-assisted delivery, nulliparity and high birthweight were significant risk factors in all three regions. Vacuum-assisted delivery was also a significant risk factor in Africa and Asia.

Conclusions  Misdiagnosis of third- and fourth-degree perineal lacerations in developing countries may be common. Correct recognition and diagnosis may lead to timely treatment and fewer sequelae. Risk factors of third- and fourth-degree perineal lacerations in developing countries were similar to those previously reported from developed countries.

Introduction

A third- or fourth-degree perineal laceration or tear is a serious adverse outcome of vaginal delivery. If left untreated it may lead to persistent perineal pain, sexual and urinary problems, and fecal incontinence.1 These sequelae severely affect the physical and psychological well-being of postpartum women.2 Previous studies have reported the prevalence of third- and fourth-degree tears to be between 0.1 and 10.2%,1,3–13 and most studies have consistently reported instrumental delivery,4–6,8–16 macrosomia,3,4,6,8,9,13–17 nulliparity,3,6,8,9,15 and episiotomy4,6,8–10,12–14,16,17 as risk factors of third- and fourth-degree tears. All of these prevalence and risk factor studies were conducted in developed countries and, to date, there are no data from developing countries. Therefore, we conducted secondary data analysis from a large multi-country data set (WHO Global Survey, GS) to describe the prevalence and risk factors of third- and fourth-degree tears in 24 mainly developing countries in Africa, Asia and Latin America.

Methods

Study design

The detailed methodology of the GS has been published elsewhere.18–20 Briefly, the GS was a facility-based multi-country cross-sectional study, with the objective of creating a global data system on maternal and perinatal health services. The study involved 373 health facilities from 24 countries, and was conducted between 2004 and 2005 in Latin America and Africa, and between 2007 and 2008 in Asia. The 24 countries in the survey were: Algeria, Angola, Democratic Republic of Congo, Niger, Nigeria, Kenya and Uganda (Africa); Cambodia, China, India, Japan, Nepal, Philippines, Sri Lanka, Thailand and Vietnam (Asia); and Argentina, Brazil, Cuba, Ecuador, Mexico, Nicaragua, Paraguay and Peru (Latin America). A stratified multistage cluster sampling design was used to obtain a sample of health institutes. The capital city in each country was included in the project sample. In addition, two provinces or regions in each country were randomly selected using a computer. Next, a census of all facilities handling more than 1000 births per year and those conducting caesarean sections was obtained. If there were fewer than seven facilities in a country, all facilities were selected. If there were more than seven facilities, seven facilities were selected randomly. The duration of data collection was 3 months in institutions that had <6000 expected deliveries per year and 2 months in institutions that had 6000 or more expected deliveries per year.19 Written consent was obtained from all ministries of health of the participating countries and from the directors of the selected facilities.19 All individual data were obtained from medical records without the identification of participants. The Ethics Review Committee of the WHO and that of each country independently approved the protocol.19

Data collection

All pregnant women who gave birth at the facilities during the period of data collection were included in the study. Data on demographics, third- and fourth-degree tears, maternal body mass index (BMI), infant birthweight, maternal age, parity, induction of labour using oxytocin, misoprostal or other prostaglandins, and mode of delivery were obtained from medical records, and a structured questionnaire was completed using this data. Maternal weight in Africa and Latin America referred to that of the last antenatal care visit, whereas in Asia it referred to the last weight before delivery.

Statistical analysis

Statistical analysis was conducted using stata 11.1 (StataCorp LP, College Station, TX, USA). The analysis consisted of three different samples (Table 1). Firstly, the caesarean section rate was based on the 290 499 women for whom information on mode of delivery was available. Secondly, a total of 214 905 women underwent vaginal delivery, but 306 had no information on third- and fourth-degree tears, and thus the prevalence of third- and fourth-degree tears was calculated based on the remaining 214 599 women. As the variation in the prevalence of third- and fourth-degree tears among facilities was extremely high, ranging from null to 76.3%, we reasoned that there is substantial misclassification, and consequently excluded facilities with no third- and fourth-degree tears, and also facilities with an exceptionally high prevalence of third- and fourth-degree tears, defined as >1.5 (interquartile range, IQR, from the 25th to the 75th percentile) above the 75th percentile of the total analytical sample after the exclusion of facilities with no third- and fourth-degree tears (Figure 1). This cut-off point corresponded to a prevalence of >4.16%. This method of identifying outliers has been described in previous literature.21 By doing so, we intended to exclude the facilities in which third- and fourth-degree tears were not recognized and facilities where third- and fourth-degree tears were over-diagnosed. This exclusion resulted in a remaining sample of 146 403 women from 204 facilities. Next, we conducted multivariate logistic regression analyses to identify the risk factors for third- and fourth-degree tears at a regional level. Adjusted odds ratios for third- and fourth-degree tears were obtained after accounting for the confounding effects of mode of delivery, induction of labour, parity, infant birthweight, maternal age and BMI, and country. Also, because 33.6% of the data on BMI from the analytical sample were missing from Africa, we included a missing category only for this region. This variable was included explicitly in the model because data is likely to be missing preferentially from resource-limited facilities in Africa, where, for example, tape measures and scales were not available. These facilities are likely to have unusually poor childbirth outcomes, and the explicit inclusion of the missing data variable in the model enabled at least a partial adjustment of the effects of these resource limits in the model. The models were also adjusted for the clustering effect within facilities, using the clustered sandwich estimator. P < 0.05 was considered to be statistically significant.

Table 1.   Prevalence of caesarean section and third- and fourth-degree perineal lacerations by country
RegionCountryTotal sample*Total analytical sample (vaginal delivery)**Analytical subsample***
nCaesarean section (%)Total no. of facilitiesNo. of facilities with no 3rd and 4th degree perineal lacerationsn3rd and 4th degree perineal lacerations (%)Total no. of facilitiesn3rd and 4th degree perineal lacerations (%)
  1. DRC, Democratic Republic of Congo.

  2. *Total sample includes caesarean section. Women missing information on mode of delivery were excluded.

  3. **Total analytical sample and subsample only include women who underwent vaginal delivery and had information on third- and fourth-degree perineal lacerations.

  4. ***Facilities reporting no third- or fourth-degree perineal lacerations were excluded. Facilities with a prevalence of third- and fourth-degree perineal lacerations > 1.5 (IQR) above 75th percentile, after exclusion of facilities without third- and fourth-degree perineal lacerations, were excluded.

AfricaAlgeria15 88713.818813 6547.2755421.1
Angola64101.7201162980.7948860.9
DRC900913.021678340.91456391.0
Kenya20 33416.120817 0631.11112 1120.4
Niger84355.411179762.8960470.7
Nigeria918514.9211078131.4824150.5
Uganda14 10413.9201512 1320.6479690.1
AsiaCambodia564214.75248120.1337150.2
China14 70946.5211978670.1211480.4
India24 97717.920720 5190.11314 8680.2
Japan335520.810126561.7821951.4
Nepal857520.58268170.5555940.4
Philippines13 43219.017110 87915.0762161.1
Sri Lanka15 15730.814310 4940.41186720.4
Thailand983834.412164540.91162940.9
Vietnam13 41235.815886070.3647340.3
Latin AmericaArgentina10 86935.814367530.31157330.4
Brazil15 36130.119610 7200.71278120.4
Cuba12 76935.817881950.4861010.3
Ecuador12 48440.418774372.2947561.2
Mexico21 05338.121513 0280.81610 7350.9
Nicaragua567531.18439120.4422760.8
Paraguay351442.26220241.6312591.0
Peru16 26334.517410 6550.31396850.4
Figure 1.

 Facility-specific third- and fourth-degree perineal laceration rates by country and region.

Results

Table 1 illustrates the prevalence of caesarean section and third- and fourth-degree perineal lacerations by country using the three different sample populations (i.e. all women, including those with caesarean section; only women with vaginal delivery; and a subsample of women who underwent vaginal delivery after the exclusion of facilities suspected of over- and under-diagnosis of third- and fourth-degree tears). Based on all 214 599 women who underwent vaginal delivery, the countries with the lowest prevalence of third- and fourth-degree tears were China (0.1%), Cambodia (0.1%) and India (0.1%), whereas those with the highest prevalence were the Philippines (15.0%), Algeria (7.2%) and Niger (2.8%) (Table 1). After the exclusion of facilities with suspected over- and under-diagnosis of third- and fourth-degree tears, the countries with the lowest prevalence were Uganda (0.1%), Cambodia (0.2%) and India (0.2%), and the highest prevalences were observed in Japan (1.4%), Ecuador (1.2%), Algeria (1.1%) and the Philippines (1.1%) (Table 1).

Table 2 presents the prevalence of third- and fourth-degree perineal lacerations by maternal and infant characteristics by region in women who underwent vaginal delivery after the exclusion of facilities suspected of over- and under-diagnosis of third- and fourth-degree tears. The overall mean maternal age (SD) was 25.8 (6.0) years, and a high maternal BMI was common in Latin America (BMI ≥ 30 kg/m2 25%). Women in Africa tended to have higher parity than in other regions. Prevalence of birthweight ≥ 4000 g ranged from 1.1% (Asia) to 4.3% (Africa). Forceps and vacuum-assisted delivery together constituted between 1.5% (Africa) and 3.6% (Asia) of all vaginal delivery. Induction of labour was common in Latin America (12.3%) and Asia (12.8%), but was not so frequent in Africa (3.5%). The overall prevalence of third- and fourth-degree tears among women who gave birth to babies weighing <4000 and ≥4000 g were 0.6 and 1.1%, respectively. The corresponding figures for nulliparous and multiparous women were 0.8 and 0.4%, respectively. Across regions, prevalence was particularly high among women who underwent forceps-assisted delivery (ranging from 2.4% in Asia to 4.0% in Africa and Latin America), and vacuum-assisted delivery in Africa (4.0%) and Asia (2.3%).

Table 2.   Prevalence of third- and fourth-degree perineal lacerations by maternal and infant characteristics*
CharacteristicsCategoryAfricaAsiaLatin America
n%n%n%
  1. Regions of study: Africa includes Algeria, Angola, Democratic Republic of Congo, Kenya, Niger, Nigeria and Uganda; Asia includes Cambodia, China, India, Japan, Nepal, Philippines, Sri Lanka, Thailand and Vietnam; and Latin America includes Argentina, Brazil, Cuba, Ecuador, Mexico, Nicaragua, Paraguay and Peru.

  2. *Facilities with no third- or fourth-degree perineal lacerations were excluded. Facilities with a prevalence of third- and fourth-degree perineal lacerations > 1.5 (IQR) above the 75th percentile after exclusion of facilities without third- and fourth-degree perineal lacerations were excluded.

  3. **Maternal BMI in Africa and Latin America refers to that of the last antenatal care visit, whereas in Asia it refers to the last weight measured before delivery.

  4. ***No missing category, as analysis was restricted to women with no missing variable for mode of delivery.

  5. ****Induction by oxytocin, misoprostal or other prostaglandins. Missing category also includes women who had induction by other methods.

Age (years)<2070370.940540.693560.9
20–3431 9710.645 0010.534 6160.5
≥3551280.443810.743660.6
Missing4740.40NA190.0
BMI (kg/m2)**<3025 2500.748 1040.528 8450.7
≥3043910.635980.594880.6
Missing14 9690.417341.410 0240.5
Parity015 5291.025 9360.719 8910.9
1 or 216 7800.523 8390.422 0090.4
≥312 0430.336350.564090.5
Missing2581.2260.0482.1
Birthweight (g)<250046430.580780.336390.6
2500–399937 7320.644 7990.542 7460.6
≥400018851.05591.419101.1
Missing3501.00NA620.0
Mode of delivery***Spontaneous43 9380.651 5410.447 2420.5
Forceps4464.06142.410664.0
Vacuum2264.012812.3490.0
Induction of labour****No42 9300.645 5590.542 2240.6
Yes15331.266760.759200.8
Missing1471.412010.82131.4

Table 3 illustrates the results of the multivariate logistic model of predictors of third- and fourth-degree perineal lacerations. Maternal age and BMI were not associated with risk of third- and fourth-degree perineal lacerations, but nulliparity, high birthweight and forceps-assisted delivery significantly increased the risk in all three regions. Compared with nulliparous women, multiparous women (parity 1 or 2) had a significant 58%, 46%, 61% lower risk of third- and fourth-degree tears in Africa, Asia and Latin America, respectively. The risk of third- and fourth-degree tears was between 1.98 (Africa) and 2.99 (Asia) times higher among women with infants with birthweight ≥ 4000 g compared with women with normal birthweight babies (2500–3999 g). Forceps-assisted delivery was also a significant risk factor, where the OR ranged from 3.72 (Africa) to 9.28 (Latin America), and so was vacuum-assisted delivery [ORs 5.59 (Africa) and 4.17 (Asia)]. The OR for Latin America could not be calculated because there were no third- and fourth-degree tear cases in the vacuum-assisted delivery group. Finally, a non-significant trend for higher risk of third- and fourth-degree tear was observed for those who underwent induction of labour (Africa, OR 1.53 95% CI 0.79–2.98; Asia, OR 1.38, 95% CI 0.90–2.11; Latin America, OR 1.38, 95% CI 0.88–2.15).

Table 3.   Multivariate logistic model of predictors of third- and fourth-degree perineal lacerations*
CharacteristicsAfricaAsiaLatin America
CategoryAdjusted OR (95% CI)**CategoryAdjusted OR (95% CI)**CategoryAdjusted OR (95% CI)**
  1. DRC, Democratic Republic of Congo.

  2. *Facilities with no third- or fourth-degree perineal lacerations were excluded. Facilities with a prevalence of third- and fourth-degree perineal lacerations > 1.5 (IQR) above 75th percentile, after the exclusion of facilities without third- and fourth-degree perineal lacerations, were excluded.

  3. **Adjusted for all covariates in the model and clustering within facility.

  4. ***Reference category is country with prevalence closest to overall 3rd and 4th degree perineal laceration prevalence in that region.

  5. ****Maternal BMI in Africa and Latin America refers to that of the last antenatal care visit, whereas in Asia, it refers to last weight measured before delivery. The missing category for BMI was only included for Africa, as 33.6% of the data on BMI was missing.

  6. *****Induction by oxytocin, misoprostal or other prostaglandins.

  7. Statistical significance: aP < 0.050; bP < 0.010; cP < 0.001.

Country***Algeria1.13 (0.34–3.78)Cambodia0.37 (0.14–1.01)Argentina0.37 (0.17–0.82)a
Angola1.28 (0.38–4.31)China0.80 (0.35–1.84)Brazil0.44 (0.19–1.05)
DRC1.25 (0.40–3.86)India0.52 (0.23–1.16)Cuba0.23 (0.11–0.50)c
Kenya0.43 (0.11–1.61)Japan2.54 (0.99–6.52)Ecuador1.80 (0.89–3.67)
Niger1.00Nepal0.84 (0.31–2.25)Mexico1.03 (0.51–2.06)
Nigeria0.50 (0.17–1.47)Philippines3.03 (0.99–9.27)Nicaragua1.00
Uganda0.15 (0.05–0.40)cSri Lanka1.00Paraguay1.00 (0.33–3.05)
  Thailand1.94 (0.70–5.44)Peru0.46 (0.25–0.84)a
  Vietnam0.74 (0.24–2.25)  
Age (years)20–341.0020–341.0020–341.00
<201.02 (0.68–1.52)<200.64 (0.40–1.03)<200.99 (0.70–1.40)
≥350.94 (0.53–1.64)≥351.12 (0.74–1.71)≥351.19 (0.70–2.02)
BMI (kg/m2)****<301.00<301.00<301.00
≥300.87 (0.59–1.28)≥300.86 (0.52–1.44)≥300.86 (0.66–1.13)
Missing0.91 (0.51–1.64)    
Parity01.0001.0001.00
1 or 20.42 (0.24–0.74)b1 or 20.54 (0.38–0.77)b1 or 20.39 (0.26–0.58)c
≥30.22 (0.14–0.35)c≥30.56 (0.25–1.24)≥30.47 (0.27–0.79)b
Birthweight (g)2500–39991.002500–39991.002500–39991.00
<25000.65 (0.41–1.02)<25000.43 (0.23–0.82)a<25001.02 (0.57–1.83)
≥40001.98 (1.29–3.03)b≥40002.99 (1.29–6.91)a≥40002.54 (1.45–4.46)b
Mode of deliverySpontaneous1.00Spontaneous1.00Spontaneous1.00
Forceps3.72 (1.64–8.45)bForceps4.21 (2.01–8.84)cForceps9.28 (6.56–13.11)c
Vacuum5.59 (2.07–15.13)bVacuum4.17 (2.48–7.02)cVacuumNA
Induction of labour*****No1.00No1.00No1.00
Yes1.53 (0.79–2.98)Yes1.38 (0.90–2.11)Yes1.38 (0.88–2.15)

Discussion

To the best of our knowledge, this is the first multi-country study to investigate the prevalence and risk factors of third- and fourth-degree perineal lacerations in developing countries. The strength of the study is the large sample size from multiple countries, which allowed a direct comparison of prevalence and risk factors across regions. Also, the large sample size allowed for the investigation of multiple risk factors, which cannot be achieved with small sample size because of the generally low prevalence of third- and fourth-degree tears.

When all facilities were included, the prevalence of third- and fourth-degree tears ranged from 0.1% (China, Cambodia, and India) to 15.0% (Philippines). Out of the 373 facilities included in our study, 142 reported no cases of third- and fourth-degree tears. Suspected under-reporting was highly prevalent in countries such as China and Uganda, where 90.5 and 75.0% of the facilities, respectively, did not report any cases of third- and fourth-degree tears. On the other hand, suspected over-reporting was common in the Philippines, where three out of 17 facilities reported more than 60% of women having third- or fourth-degree tears. The overall prevalence in our study was 0.6% (range 0.1–1.4%) when facilities with suspected under- and over-reporting were excluded, and this figure is comparable with previous studies that included both nulliparous and multiparous women, where results ranged from 0.1% to 10.2% (UK,3 USA,4–9 Japan10 and Finland11).

The previously reported risk factors of third- and fourth-degree tears in developed countries were mostly similar in our study in mainly developing countries. We focused on six main risk factors of third- and fourth-degree tears, based on past findings: maternal age,8–10,13,14 BMI,22 parity,3,6,9,10,15 infant birthweight,3,6,14,15 instrumental delivery (forceps or vacuum),4–6,8–16 and induction of labour.10 First, we did not find any significant associations between maternal age and third- and fourth-degree tears, although there was a tendency for older women to have higher ORs in Asia and Latin America. The results of previous studies are mixed but most have reported a significant or non-significant higher risk of third- and fourth-degree tears among older women,9,13,14 with the exception of one study.8 Secondly, no significant associations between BMI and third- and fourth-degree tears were identified in our study, although there was a non-significant trend for those with higher BMIs to have a lower risk in all three regions. Although most previous studies have reported no association between BMI and third- and fourth-degree tears,23 lower risks of third- and fourth-degree tears have been reported in a large study from the USA, where a dose-dependent significant protective effect was observed with higher BMIs at admission to labour/delivery only among nulliparous women.24 The authors of this study postulate that higher caesarean section rates among obese women and extra soft tissue around their perineum might explain why obesity is protective against third- and fourth-degree tears. Thirdly, multiparity was associated with a significant dose-dependent protective effect against third- and fourth-degree tears, compared with nulliparity in all three regions. The results accord with previous studies from the USA and UK that have reported a significant 3.0–7.3 times higher risk of third- and fourth-degree tears among nulliparous women, compared with multiparous women.3,6,9,15,24 Nulliparity is a well-known risk factor for third- and fourth-degree tears,8 and the most biologically plausible mechanism is the lack of elasticity of the perineum among nulliparous women.3,10,15 Fourthly, birthweights of ≥4000 g were associated with a significant 1.98, 2.99 and 2.54 times higher risk of third- and fourth-degree tears, compared with normal birthweights in Africa, Asia and Latin America, respectively. These findings are comparable with previous literature, which have all reported lower birthweights to be protective against perineal laceration.3,6,15,24 Next, forceps-assisted delivery was associated with a 3.72–9.28 times higher risk of third- and fourth-degree tears compared with spontaneous delivery in all three regions. Vacuum-assisted delivery was also associated with a significant 4.17 and 5.59 times higher risk in Asia and Africa, respectively. These results are consistent with numerous studies that report forceps and vacuum-assisted delivery to be risk factors for third- and fourth-degree tears (OR 3.0–15.5 for forceps;6,9,10,14,15 OR 2.6–9.5 for vacuum-assisted delivery9,10,14). Finally, we observed a 1.53, 1.38 and 1.38 times higher risk of third- and fourth-degree tears among those who underwent induction compared with those who did not in Africa, Asia and Latin America, respectively, but the results were not significant in any of the three regions. There are conflicting results regarding induction of delivery in the literature. A large study in Japan reported a significant 2.19 times higher risk of third- and fourth-degree tears among women who underwent induction of labour,10 but others, including a large study in the USA, have found no such effect.3,24

Although results were similar across regions, some differences in the magnitude of the association between the risk factors and third- and fourth-degree tears were observed. This may be attributable to the different caesarean section rates ranging from 1.7% in Angola to 46.5% in China. If women with a high risk of third- and fourth-degree tears were more likely to undergo caesarean section in some regions, and thus not be under risk of third- and fourth-degree tears, this could at least partially explain the regional differences observed.

Prevention, early detection and treatment of third- and fourth-degree tears are essential. The large variability in rates of third- and fourth-degree tear among facilities warrants attention. Indeed, 38.1% (142/373) of the facilities in our study did not report any cases of third- and fourth-degree tears, and some facilities reported a prevalence of over 60%, which is far from the prevalence of third- and fourth-degree perineal laceration considered to be normal. Under-recognition of third- and fourth-degree tears is unacceptable, as if left untreated, it may result in persistent perineal pain, urinary and fecal problems, and sexual dysfunction.1,25 On the other hand, over-diagnosis may lead to unnecessary treatment. In order to overcome this problem, clinicians must be aware of the devastating consequences of untreated third- and fourth-degree tears, and the introduction of standardised guidelines to promote correct diagnosis at the clinical scene may be important. In terms of prevention, avoidance of unnecessary instrumental delivery and possibly induction of labour during delivery may contribute to a reduction in third- and fourth-degree tears, as high rates of unnecessary instrumental delivery and induction of labour have been reported,26,27 at least in developed countries. The magnitude of this problem in developing country settings remains largely unknown, and is an area for further research.

Some limitations of this study should be noted. First, as this was a facility-based study, the results may not be generalisable to the general population, especially in countries where the rate of hospital delivery is low. Second, there was probably substantial over- or under-reporting of third- and fourth-degree tears in some countries, resulting in unreliable estimates. We excluded facilities with no third- and fourth-degree tears and facilities with a prevalence higher than 1.5 (IQR) above the 75th percentile in our analytical sample in order to exclude facilities with suspected over- and under-reporting. We cannot be certain about how much the misclassification was reduced by this procedure. However, despite potential misclassification, the effects of most risk factors were in line with previous publications. Future studies should be conducted with strict diagnostic criteria. Thirdly, we did not have information on some reported risk factors of third- and fourth-degree tears, such as episiotomy,25,28 inherent predisposition (e.g. short perineal body),25 prolonged second stage of labour,25,28 previous anal sphincter tear25 and female genital mutilation.29 Therefore, the independent and confounding effects of these factors remain unknown. Finally, the inclusion of a missing category is known to induce biases in parameter estimates, and thus the inclusion of a missing category for BMI in Africa may have induced some level of bias. However, the analysis without a missing category yielded similar results (data not shown), and thus we believe that this is not a major limitation.

Conclusion

In conclusion, over- and under-diagnosis of third- and fourth-degree perineal lacerations in developing countries may be common. Training of medical personnel to facilitate the early detection and treatment of postpartum women with third- and fourth-degree perineal lacerations may be important in reducing sequelae. Most risk factors previously reported from developed countries were similar in our context of mainly developing countries. Studies using strict diagnostic criteria for third- and fourth-degree perineal lacerations are warranted in developing countries to obtain more accurate estimates on prevalence and risk factors.

Disclosure of interests

We declare that we have no conflicts of interest.

Contribution to authorship

HF wrote the article. AK analysed the data and contributed to writing the article. RM, JPS and AMG contributed to data collection and provided advice. JZ contributed to data analysis and interpretation. All authors read and approved the final version of the manuscript.

Details of ethics approval

The Ethics Review Committee of the World Health Organization (WHO) and that of each country independently approved the protocol. Individual informed consent was not obtained because this study was a cluster-level study in which data was collected from medical records without any individual identification. WHO/RHR Scientific and Ethical Review Committee, 25 April 2003, ref. no. A25176.

Funding

This study was financially supported by: the Department of Making Pregnancy Safer, the United States Agency for International Development (USAID); UNDP/UNFPA/WHO/World Bank Special Programme of Research, Development and Research Training in Human Reproduction (HRP), Department of Reproductive Health and Research (RHR), WHO, Geneva, Switzerland; Ministry of Health, Labour and Welfare of Japan; Ministry of Public Health, Beijing, China; and the Indian Council of Medical Research, Delhi, India.

Acknowledgements

The authors thank all those who made contributions to the study design and implementation. The authors also thank the study coordinators, data collectors and staff of the Ministries of Health and WHO offices, the participants of the study and Stuart Gilmour for his statistical advice.

Ancillary