Use of Robson classification to assess cesarean section rate in Brazil: the role of source of payment for childbirth

Source of data and subjects

The “Birth in Brazil” study is a national hospital-based study of postpartum women and their newborns that was carried out from February 2011 to October 2012. This study included a complex sample of 266 hospitals with 90 postpartum women interviewed in each hospital. These hospitals were selected among those which had ≥ 500 births in 2007 (19 % of all them) and where occurred 78.6 % of all births in Brazil that year [17]. Some of facilities’ characteristics are included in Additional file 1 and further informations are presented in Azevedo Bittecourt et al. [18].

The sample was selected in three stages. In the first stage hospitals were stratified according to geographical region (North, Northeast, South, Southeast and Midwest), location (in or outside a state capital), and type of hospital health care (private, public or mixed), generating 30 stratums. Hospitals were selected with probability proportional to the number of births in each of the 30 stratums. In the second stage an inverse sampling method was used to select the number of days (minimum of seven) necessary to carry out 90 interviews of postnatal women in each hospital. In the third stage, all women who had given birth to a live newborn, regardless of weight or gestational age, or to a stillbirth with birth weight ≥ 500 g and/or gestational age ≥ 22 weeks of pregnancy in one of the sampled hospitals in the period of the data collection, were invited to participate. A calibration procedure was used to ensure that the distribution of the puerperal women interviewed was similar to that observed among the births in the population for the year 2011. Further information on the data collection [17] and the design of the sample is detailed elsewhere [19].

In the current analysis, we included all 23,894 women interviewed for the “Birth in Brazil” study.

Robson groups and covariables

The variables necessary for applying the Robson classification are: number of fetuses (single or multiple); fetal presentation (cephalic, breech or transverse); previous obstetric record (nulliparous or multiparous, with or without uterine scar); onset of labor and delivery (spontaneous, induced or prelabor CS); and gestational age at the time of delivery.

We defined as having a “public source of payment” those women who delivered in public health care facilities or in mixed health care facilities (private facilities financed by both public and private funds) that were not paid by a health insurance plan. “Private source of payment” included women who delivered in a mixed health care facility that was paid for by a health insurance plan and those who delivered in a private facility, regardless of whether the delivery was covered by a health insurance plan. We used the terms “public sector” and “private sector”, respectively, to refer to these definitions.

The socioeconomic, demographic and obstetric characteristics investigated were: “age” (12–19, 20–34 or ≥ 35 years); “self-reported skin color”: White, Black, Pardo/Mixed, Yellow, and Indigenous); “marital status” (living with partner or not); “duration of education” (≤7, 8–10, 11–14 and ≥ 15 years); “parity” (0, 1 or ≥ 2); “number of previous CS” (0, 1, 2, or more); “type of pregnancy” (single, multiple); “induction of labor” (yes/no); “labor (induced or spontaneous)” (yes/no); “type of delivery” (vaginal, forceps/vacuum or CS); and high obstetric risk. High obstetric risk covered the following complications: hypertensive disorders, eclampsia, preexisting diabetes, gestational diabetes, severe chronic diseases, infection at hospital admission for birth (including urinary tract infection and other sever infection such as chorioamnionitis and pneumonia), placental abruption, placenta previa, intrauterine growth restriction and major newborn malformation (including anencephaly, hidrocephaly, spina bifida, gastrosquisis and other abdominal wall defects, cardiac malformations and multiple malformations).

All data were collected from women’s and newborn medical records, except data regarding sociodemographic characteristic, such as maternal age, skin color, marital status and years of schooling, which were collected through face-to-face interviews with the mothers during their hospital stay. The gestational age was calculated using an algorithm that primarily relied upon ultrasound estimates (74 % of all women) [20].

Statistical analysis

Differences in proportions of maternal characteristics between the public and private source of payment of birth were analyzed by chi-square statistical test with a significance level of < 0.05.

Differences in the relative size of the Robson groups by source of payment for birth (public or private) were analyzed by Z-test with Bonferroni adjustment with a significance level of < 0.05. We used the same to analyze differences in the CS rate by source of payment for birth and by obstetric risk within each Robson group.

We took into consideration the complex sampling design in all statistical analyses. The statistical program used for analysis was SPSS, version 20.0 (SPSS Inc., Chicago, IL, USA).

Main findings

The CS rate in Brazil was more than two-fold higher in women covered by private health care than in women who delivered in the public sector. The groups with the greatest impact on Brazil’s CS rate in both public and private sectors were group 2 (nulliparous, term, cephalic with induced or cesarean delivery before labor), group 5 (multiparous, term, cephalic presentation and previous cesarean section) and group 10 (cephalic preterm pregnancies), which accounted for more than 70 % of CS carried out in the country.

The prevalence of obstetric risk was not different despite the discrepancies in sociodemographic characteristics of women from the public and private sectors. High-risk women had significantly higher CS rates when compared with low-risk women in almost all Robson groups only in the public sector, but not in the private sector, which suggest a liberally overuse of CS in women with private health care.

Strengths and limitations

This study is important for many reasons. First, it was based on a national survey that covers all Brazilian states and was representative of 2,337,475 births (80 %) occurring in 2011 [19]. To our knowledge, it is the third study that used the Robson classification to assess CS rates at a national level and the second to use primary data [21, 22]. We collected all essential information included in the Robson classification, and only a few women could not be classified into one of the Robson groups. This minimized the problem of using routine data, which are not always accurate. Second, we estimated gestational age using an algorithm, based primarily on obstetric ultrasound, which confers certain advantages over last menstrual period, as the latter tends to overestimate the rate of preterm birth in the Brazilian population [20]. Finally, we also used a clear definition to classify women who went into labor, which is commonly lacking in previous reports [23].

Because of the sample design, our results can only be extrapolated to the 80 % of the population who give birth in hospitals with more than 500 deliveries per year, and not to the entire Brazilian population. In addition, this study had limited power to compare differences between the public and private sectors for categories of Robson groups of very low frequency, such as categories 6, 7, 8, 9 and induction groups in the private sector (2a and 4a). Another limitation of the study is the potential misclassification of some women who belonged in Groups 1 and 3 and were erroneously classified as Groups 2 and 4 because of the definition used for labor induction. It is possible that some nulliparous and multiparous women admitted with spontaneous onset of labor (Groups 1 and 3) received oxytocin during the latent phase, before reaching 4 cm dilation, for augmentation of labor. However, this probably will not affect the main findings of the study, considering the underuse of labor induction in this study.

Interpretation

CS rates continue to increase around the world without a clear understanding of the main drivers and consequences. The CS rate found in the “Birth in Brazil” study (51.9 %) is among the highest in the world along with China (52.5 %), Cyprus (52.2 %), the Dominican Republic (56.4 %) and Egypt (51.8 %) [24, 25]. There is evidence that it continues to grow [6].

Our results showed that women who delivered in the private sector were more frequently white, older and with higher education, conditions associated with CS in previous studies [26, 27]. Although there were more multiparous women, and fewer twin pregnancies and previous CS in women in the public sector, it is unlikely that these factors alone can explain the difference in CS rates. The low use of labor induction in the private sector (only 3.5 %) was also remarkable, reinforcing the preference for CS before labor as a form of immediate delivery. Even in the public sector, the rate of induced deliveries was lower than in countries with low CS rates, such as France and the Netherlands [21, 22], and also lower than previously reported in Latin America [28].

The current analysis of CS by Robson classification revealed, as with other studies, that the nulliparous group, term, cephalic presentation is one that contributes most to the total rate of CS [21, 29, 30]. Analyzing nine institutions, Brennan et al. [29] showed that 98 % of institutional variation in the CS rate may be attributed to this group, which contributed to over 30 % of CSs performed in France and the Netherlands [21, 22]. The same authors also pointed out that the proportion of this group in the population was similar between institutions, reinforcing the hypothesis that there are variations in the CS rate in this group that affect the overall rate. In our study, the proportion of groups 1 and 2 combined was 39 %, similar to that found in Latin America (36.4 %) [31], France (38.2 %) [21], Canada (39.7 %) [30] and the Netherlands (39.9 %) [22]. However, in Brazil, we found that the group of CSs before labor (group 2b) impacted more on the contribution of term nulliparous women (14.9 %). In European countries, the proportion of this group (2b) is around 1 % of the obstetric population [21, 22], but even in the Brazilian public sector, this group included 9 % of women in our study. As the number of nulliparous women is almost the same, the proportion of group 1 (18.9 %) was below what is commonly found in other studies that have reported it at above 25 % of the obstetric population [21, 22, 29, 31]. When we analyzed the women with private payment, this percentage was even lower (6.4 %), despite the higher proportion of nulliparous, term, cephalic women in the private (45.7 %) than the public (36.1 %) sector.

The group that singly most contributed to CS in Brazil was multiparous, term with previous CS (group 5). Recently, a WHO analysis found that CS rate and the absolute contribution of group 5 has increased in recent years [16]. These data show the domino effect of CS use: rising CS rates, especially in nulliparous women, increase the number of women with previous CS, who are more likely to undergo a repeat CS [16]. As a result of the history of high CS rates in Brazil, group 5 constitutes almost 20 % of Brazil’s population; combined with the high rate of repeat CS, this makes it responsible for almost a third of CS carried out in the country both in the public and private sectors. Our data are consistent with the WHO Global Survey of Latin America [31], where group 5 accounted for 26.7 % of CS. The CS rate for this group, although not different from that found in countries with very high and high human development index in the WHO surveys (from 78.1 to 79.4 %) [19], is considerably higher than that found in France (61 %) [21] and the Netherlands (47 %) [22]. While the success of vaginal birth after cesarean (VBAC) reaches 70 % in many studies [32], an incentive to this practice would be essential to reduce CSs in Brazil. In addition, repeat CSs increase the chance of placenta accreta and placenta previa, which can result in increased risk in subsequent pregnancies [32, 33].

The multiparous groups without CS (groups 3 and 4) contributed to just over 10 % of CSs. Noteworthy is the high CS rate in group 4 (61 %), even in the public sector (55 %), which is related to the number of women undergoing CS before delivery (group 4b) that is greater than those undergoing induction (group 4a). While in Brazil group 4b corresponds to 3.2 % of women, in other countries it does not exceed 1 % [21, 22, 30]. These numbers may again reflect the preference of CS to induction of labor in high-risk pregnancies, but also the use of CS for concomitant tubal ligation, as mentioned in other reviews [8, 11, 34].

The third group that contributed most to the CS rate in both sectors was the preterm birth group, contributing to nearly 10 % of CSs performed in Brazil. This number is slightly higher than that found in countries with low rates of prematurity. In the Netherlands, group 10 corresponds to 7.1 % of CS [22], while in France the percentage is 8.3 % [21]. In Brazil, both the group size (9.7 %) and its CS rate (50.1 %) affected the CS overall rate.

Finally, the groups of non-cephalic presentations (groups 6, 7 and 9) and twins (group 8) together contributed only 8.9 % of CSs. This number is lower than the WHO Survey of Latin America (14 %) [31] and considerably lower than that observed in France (20.5 %) [21] and the Netherlands (27.2 %) [22]. Even excluding twins, whose prevalence in these countries is greater, and considering only non-cephalic presentations, the gap remains large (Brazil: 7 %; France: 16.5 %; Netherlands: 22.5 %).

In Brazil, there was a clear difference in both the distribution of women and CS rates into Robson groups according to the source of payment. The two largest relative size groups in the public sector (groups 3 and 1) had little importance in the private sector. Additionally, there was a clear concentration of nulliparous women in group 2b and multiparous women in group 5, which represented > 70 % of CSs in the private sector, where > 80 % of women did not go into labor, reinforcing the saying “once a cesarean, always a cesarean.”

Analyzing the increase in the number of CSs in the period between the two WHO surveys, Vogel et al. [16] concluded that the threshold for medically indicated CS has become lower over time, or the use of elective CS has risen, or both occurred together. This appears to be what has occurred in Brazil over recent decades. While the CS rate is higher than those found in other countries in groups with low probability of CS (term nulliparous and multiparous with spontaneous labor and multiparous with induced labor), the widespread use of elective CS in nulliparous and multiparous women, regardless of obstetric risk, even in the public sector, was also observed. Indeed, 84.2 % of all CS in Brazil are performed before the active phase of labor (data not shown).

In the private sector, it is very likely that CS was not related to the presence of obstetric risk, since CS rates according to the risk of pregnancy were different only in group 10. Furthermore, CS rates were also extremely high in low-risk women. Despite women with private funding having a greater preference for CS (36.1 % of nulliparous and 58.8 % of multiparous in early pregnancy) [8], this fact alone does not explain such high rates of CS.

The high rates of elective CS in Brazil, especially in the private sector, are of concern, because they may bring unnecessary harm to women’s and babies’ health if performed without indication [5], including increased maternal [3] and neonatal morbidity, especially when performed before 39 weeks [35]. Our data revealed a great difference in CS rates in the low-risk preterm group according to source of payment (25.4 % public and 71.4 % private), which raises questions about whether this practice may be leading to iatrogenic prematurity.