Introduction

Worldwide, 19% of births occur by cesarean section (CS), with the highest rates found in more developed countries [1]. Between 1990 and 2014, the CS rate increased from 11% to 25% in Europe and from 22.3% to 32.3% in North America [1]. While CS may be a life-saving intervention in cases of pregnancy or childbirth complications, it is also associated with reproductive health risks. A recent systematic review showed that women who undergo CS are at a higher risk of miscarriage, stillbirth, placenta previa, placenta accreta, and placental abruption in future pregnancies [2].

Risks associated with CS may extend to children as well, possibly because CS is often conducted due to emergency indications and health complications which themselves may present challenges (e.g., fetal distress, preeclampsia). Children born by CS have been found to have an increased risk of asthma, obesity, and type-1 diabetes [3]. Furthermore, birth by CS has been associated with risks for neurodevelopmental disorders such as autism and attention-deficit/hyperactivity disorder [4], although recent evidence suggest that those associations are most likely due to unmeasured familial confounding [5]. Some studies reported delays in cognitive and motor development in children born by CS, but the results of these studies are inconsistent [6, 7].

Moreover, some existing research suggests that CS may affect child temperament. Although early perspectives on temperament [8, 9] emphasized the genetic underpinnings of early appearing traits, contemporary perspectives acknowledge the influence of both prenatal and postnatal environments on the hormonal and neural systems associated with characteristic patterns of emotional response [10, 11]. To date, only a few studies have focused on the association between CS and infant temperament. Field and Widmayer [12] observed that mothers who gave birth per emergency CS, compared to those who gave birth vaginally, rated their infants as having more optimal temperament 4 and 8 months postpartum. Maziade et al. [13], on the other hand, found no relation between birth mode and infant temperament at these same ages. Simons et al. [14] showed that infants born through CS versus vaginal birth were rated by their parents as less adaptable. In addition, examining the interaction between birth mode and infant risk status, Simons et al. [14] observed that CS-born infants in a low-risk group were perceived as more withdrawing, more negative, and less adaptable than those from high-risk and/or vaginal birth groups. The authors of this study did not, however, indicate whether their results applied to emergency or planned CS.

Related research on the associations between CS and child emotional and behavioral problems has likewise yielded conflicting results. Some studies reported adverse effects of CS on child outcomes, such as a higher risk of parent-reported internalizing problems (e.g., anxiety/depression) at 1.5 and 5 years of age [15, 16] and increased stress behavior two hours after birth [17], while others found no effect of CS on child behavior reported by parents [6, 18–20]. One large study [21] even reported that, in children aged 4–6 years, the risk of externalizing problems (e.g., aggression) was lower in those children who were born by elective CS based on maternal request rather than by spontaneous or assisted vaginal birth.

Some investigators attempted to explain the inconsistencies in existing research by identifying specific groups of children particularly vulnerable to the negative effects of CS. As noted above, Simons et al. [14] reported specific effects of CS for low-risk but not high-risk pregnancies. Several studies examined potential moderating effects of child’s sex but found no evidence for such interactions [6, 16, 19]. Although often included as a covariate or a controlled variable in the studies, parity is a hitherto overlooked possible moderator of birth practices. Since research suggests that CS has negative impact on maternal well-being (such as lower self-esteem or higher risk of post-traumatic stress symptoms) especially in primiparous women [22, 23], who are also at a higher risk of emergency CS [24], some authors speculated that the effect of birth mode on infant behavior might differ depending on parity [25]. Primiparous women have been found to report lower maternal self-efficacy [26, 27], worse mother-infant bonding [28] and higher postpartum anxiety [26] compared to multiparous women, which could, combined with their higher risk of negative psychosocial consequences of CS, lead to their tendency to perceive their child behavior as more problematic. However, to the best of our knowledge, to date no study has tested this hypothesis.

The inconsistency in existing research findings regarding the potential effects of CS on child behavioral outcomes may also be due to differing methodology. While some studies focused on maternally requested CS specifically [16, 21], others differentiated between elective and emergency CS [6, 19], and several studies failed to distinguish between different types of CS [14, 15]. Although several studies controlled for pregnancy and intrapartum complications, no study considered specific indications for CS, which might be the actual cause of children’s behavioral problems. Another gap in the previous research is that child outcomes are mostly assessed on a one-time basis later in childhood, thus yielding no information about the stability of potential effects.

Studying perinatal influences on infant temperament is of great clinical relevance as a large body of research implicates temperament as a risk factor for later behavior problems and psychopathology, with negative mood forecasting both internalizing and externalizing problems [29, 30]. The underlying mechanisms through which CS may affect child behavioral and temperamental characteristics are not yet determined, but several hypotheses have been proposed, including altered microbiota in the infant gut [31–33], programming of hypothalamic–pituitary–adrenal (HPA) axis functioning [17, 34, 35], and “iatrogenic prematurity” in infants born before optimal maturation of the fetal brain [36].

The effects of CS may additionally be mediated by elements of the infant’s social context following birth. An early meta-analysis of relations between delivery mode and psychosocial outcomes [37] indicated that, in comparison to mothers who had delivered vaginally, mothers who underwent CS experienced less satisfaction with the birth, responded less positively to their newborns, were less likely to breastfeed, and interacted less with their infants in the home. More recent studies continue to support these findings [38, 39], and it has been proposed that increased risk for postpartum depression, changes in neural responses to infant crying, and feelings of failure among mothers who have undergone CS, may affect their maternal behaviors [22, 40, 41]. Indeed, CS has been associated with parenting that is characterized by less stimulation, caretaking and intimate play [42]. Because maternal depression, self-efficacy, and parenting behavior have been related to difficult temperament [43, 44]; and because increased levels of infant negativity and fear have been related to low maternal responsivity and sensitivity [45–47]; it is reasonable to expect CS to affect infant difficulty through its effect on mothers’ psychology and actions. A final mechanism potentially relating CS to temperament also concerns maternal psychology. Feelings of disappointment, diminished self-efficacy and depression are associated with cognitive biases, including increased attention to perception of negative information [48–50]. As previous studies operationalize temperament and behavioral features through parent report questionnaires, maternal psychology associated with CS may result in higher ratings of difficulty by biasing the reports of mothers, causing them to overestimate the degree to which their child exhibits characteristics associated with difficulty [14–16].

In this study, we aimed to obtain a more complete picture of the effects of CS than previous studies by distinguishing between emergency and planned CS, controlling for indications for CS, assessing infant temperament repeatedly (6 weeks and 9 months postpartum), and investigating parity as a moderator. We hypothesized that infants born through planned or emergency CS, versus those born vaginally, would display more difficult temperament. Furthermore, we anticipated that the effect of CS on infant temperament would be moderated by parity. In particular, we hypothesized that CS would be most strongly associated with less favorable infant temperament ratings in primiparous women. We also expected to replicate the finding of earlier studies that concluded that the effect of CS on infant behavior is not moderated by infant’s sex.

Materials and methods

Results

Sample description

The characteristics of the mothers and infants included in this study are summarized in Table 1. Of the 452 women who completed the ICQ at 6 weeks postpartum, 77% had vaginal birth, 12% planned CS for medical reasons, and 11% had undergone emergency CS. The numbers of female and male infants were approximately even (50,2 and 49,8%, respectively). A total of 43% women were primiparae. Most mothers (75%) were married, 22% were single, and 3% were divorced. The mean age was 31 years (range 18–45). The women who completed the ICQ at 9 months postpartum (n = 258) had similar characteristics except that the percentage of women with a high educational level increased slightly. No differences in infant sex, parity, maternal age or educational level were found by type of delivery.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Characteristics of the sample.

https://doi.org/10.1371/journal.pone.0255367.t001

Infant temperament at 6 weeks of age

In both the unadjusted (S2 Table) and adjusted (Table 2) analyses, children born by emergency or planned CS for medical reasons did not have significantly different ICQ scores at 6 weeks from those born by vaginal delivery. In the adjusted analysis, only marital status was significantly associated with ICQ scores, with single women rating their infants as less difficult than married women. The overall model fit was not significant.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Association between the mode of delivery and infant temperament—Adjusted for sociodemographic factors, parity and infant’s sex.

https://doi.org/10.1371/journal.pone.0255367.t002

The analysis of interactions suggested that parity moderates the effect of emergency CS on infant temperament ratings at 6 weeks postpartum (B = -7.57, SE = 3.07, p = 0.014) (Table 3; Fig 1) such that primiparae rated their infants’ temperament as more difficult if they had an emergency CS compared to vaginal birth, a finding that did not apply to multiparae. We found no significant interaction between birth mode and infant sex related to temperament ratings at age 6 weeks.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. The effect of parity on the association between the mode of delivery and infant temperament at 6 weeks and 9 months postpartum.

https://doi.org/10.1371/journal.pone.0255367.g001

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. Association between the mode of delivery and infant temperament—Interaction with parity.

https://doi.org/10.1371/journal.pone.0255367.t003

The unadjusted analysis did not show any significant associations between fetal hypoxia, previous CS, fetal macrosomia, breech presentation, failure to progress, or other labor dysfunctions and ICQ scores at 6 weeks (S3 Table). The results remained unchanged in the adjusted analysis (Table 4).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 4. Association between indications for cesarean section and infant temperament—Adjusted for sociodemographic factors, parity, and infant’s sex.

https://doi.org/10.1371/journal.pone.0255367.t004

Discussion and conclusions

In this study, we investigated the effects of cesarean section (CS) on difficult infant temperament assessed 6 weeks and 9 months postpartum. When considered in the absence of moderators, we found no association between either emergency or planned CS and infant temperament in our healthy sample. This null finding is consistent with several other studies, especially those that used large population-based samples and controlled for maternal health status in pregnancy. Such studies, too, reported no relation between CS and behavioral problems [6, 18, 19].

Our analyses suggested that primiparae rated their infants’ temperament as more difficult if they had an emergency CS compared to vaginal birth, a finding that did not apply to multiparae. However, the interaction of emergency CS and parity was significant only at 6 weeks. The fact that the interaction term for emergency CS and parity at 9 months postpartum was no longer significant may be due to a smaller sample size at 9 months compared to 6 weeks, which could have resulted in lower statistical power. Alternatively, it is possible that with time, primiparous women who gave birth through CS recovered from the stress related to emergency operative delivery, which resulted in less severe assessments of their children’s behavior.

Our finding that emergency CS may increase the risk of difficult temperament in infants born to first-time mothers, but not in those born to more experienced mothers, suggests that primiparae might be more vulnerable to the effects of emergency CS in terms of lower maternal self-esteem [64] and parenting competences [39], which may lead to higher levels of negative affectivity in their infants [45–47]. Multiparous women, on the other hand, have been found to have higher maternal self-efficacy and more optimal mother-child bonding [26–28], which enables them to interact with their infants more positively even in the aftermath of emergency CS. Relatedly, primiparous women tend to experience more anxiety in the postpartum period [26], which, combined with CS-related stress, may increase cortisol levels in maternal milk with potential effects on their infants’ temperament [65].

An alternate possibility is that these results may be confined to maternal perceptions of infant temperament, rather than (or in addition to) reflecting infants’ actual behaviors. Higher maternal self-efficacy in multiparous women [26, 27] may lead to their impressions of their infants being less open to influence from the experience of an unplanned medical procedure involving their newborn. In contrast, the ratings provided by first-time mothers may reflect their higher vulnerability to negative psychosocial impacts of CS including lower parenting self-esteem and higher anxiety [39, 64]. Moreover, higher anxiety in primiparous women who undergo CS may also be related to the fact that primiparae are more affected by CS in terms of their reproductive choices; existing studies show that delivery via CS is associated with a lower number of subsequent pregnancies [66].

Our finding that the effect of emergency CS on infant behavior may differ in primiparae and multiparae may help explain the inconsistent results of previous studies which evidenced both more positive [12], more negative [14], and equal [13] temperament ratings of infants born via CS in comparison to those born vaginally. Nevertheless, the effect of CS was not moderated by infant sex, which agrees with the results of previous studies [6, 16, 19].

Particular indications for CS had no effect on infant temperament ratings except for breech presentation that was associated with less favorable infant temperament 9 months postpartum. However, this finding is not conclusive as it only occurred 9 months and not 6 weeks postpartum and the overall model fit was not significant. Moreover, as we excluded women with serious health problems in pregnancy as well as seriously ill newborns, we explored only the effects of a limited range of indications. It is possible that other indications, especially those related to suboptimal health status in newborns, would predict infant temperamental difficulty. Indeed, Sirvinskiene et al. [15] observed that suboptimal newborn physiological functioning predicted emotional and behavioral problems at 1.5 year of age.

Investigation of the effects of CS on infant outcomes is methodologically challenging due to potentially confounding factors associated with both a higher risk of CS and infant behavioral outcomes. It is therefore a strength of our study that we accounted for such confounding by including only healthy mothers whose pregnancy resulted in the birth of a healthy newborn. Moreover, we excluded women who had CS performed on their own request without medical reasons. Those women may have a specific psychological profile which could be associated with behavioral characteristics of their children [67]. The strengths of our study also include differentiating between emergency and planned CS, taking into account the underlying indications for CS, examining infants’ temperament prospectively at two different time points, and considering parity as a potential moderator of effects.

Our study does, however, also have certain limitations. Its perhaps most important weakness is linked to the fact that infant temperament was rated by mothers, which obscures the degree to which the results reflect mother’s perceptions of their infant, or the infant’s behavior per se. The use of parent report to assess child temperament has been criticized with respect to issues of bias [68]. This concern may be particularly salient for the ICQ, in comparison to other temperament questionnaires, as it was explicitly designed to measure perceptions of “difficultness”. A related shortcoming is the lack of detail regarding temperament that is provided by the ICQ. More recently developed instruments, such as the Infant Behavior Questionnaire—Revised [69], provide a wider perspective on temperament, including dimensions associated with approach tendencies and capacities for behavioral control, to complement the emphasis on negative affectivity inherent to the concept of difficultness. Despite these critiques, however, questionnaires such as the ICQ are consistent with the literature, as, to best of our knowledge, all previous studies investigating the association between delivery mode and child temperament or behavioral problems relied on parental report [6, 12–16, 18, 19, 21]. Such frequent use of parent-administered questionnaires is due to the relative ease with which they can be employed, but also because they make use of parents’ vast experience with their children in contexts that are ethically or logistically difficult to measure in brief and artificial laboratory observations and require less interpretational inference than psychobiological measures of temperament [70]. Although bias is a real concern, contemporary temperament scholars contend that parent-report measures contain both subjective and objective components [71, 72], and parent-report measures have been shown to demonstrate superior predictive validity than laboratory observation indices [73]. To disentangle the degree to which CS has implications for temperament, future research should employ multiple methods.

Another limitation concerns sample attrition: although 1,190 mothers completed the questionnaire at baseline, only 661 completed the questionnaire at 6 weeks and 392 at 9 months postpartum. However, women who dropped out of the study differed from those who remained only with respect to educational level. Given the demographics of the Czech Republic, our sample was ethnically homogenous, including only White European population. Although this does not limit possible generalizations for the Czech population, it may limit the generalization potential of our findings to other countries or ethnically diverse populations.

In conclusion, we largely failed to find evidence linking surgical birth with adverse behavioral outcomes. An interaction between emergency CS and parity was found, suggesting that primiparous women rated their babies born by unexpected CS as more fussy, unadaptable, and unpredictable than infants born vaginally. However, based on our data, we cannot determine whether it is CS per se that affects firstborns’ behavior through biological pathways such as altered infant gut microbiota, or whether our results indicate that first-time mothers undergoing emergency surgical delivery tend to perceive their children’s behavior as more difficult. Given our null findings regarding the main effect of CS, it is yet premature to draw firm conclusions to inform clinical practice. Further research is needed to shed light on the potential risks associated with CS, preferably using observational methods for temperament assessment, investigating biological pathways through which may CS exert effects on children, while controlling carefully for indications for CS and other relevant confounders.

Supporting information