SEARCH

SEARCH BY CITATION

Abstract

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. METHODS
  5. RESULTS
  6. DISCUSSION
  7. CONCLUSION
  8. Acknowledgments
  9. References

Objective  To investigate (A) the determinants of infant stress response at delivery and (B) test the hypothesis that stress at birth, as reflected by cord arterial cortisol, influences cortisol response to vaccination at two months.

Design  Prospective observational study.

Setting  Tertiary referral maternity hospital.

Population  One hundred and seventy-two primiparous women with uncomplicated singleton pregnancies.

Methods  Women were recruited antenatally. At birth, cord arterial blood and obstetric data were collected. Saliva was collected from infants immediately before and after vaccination at two months. Cortisol was analysed from cord blood and saliva by radio-immunoassay.

Main outcome measures  Stress response at birth, as demonstrated by cord arterial cortisol; association with saliva cortisol response to vaccination at two months.

Results  Cord arterial cortisol varied with mode of delivery, combined spinal/epidural use and pH. Salivary cortisol response at two months correlated with cord arterial cortisol (r= 0.24, P < 0.05). Infants with the highest and lowest cord arterial cortisol had markedly different cortisol responses at two months (P= 0.017). These groups had different modes of delivery with caesarean rates of <8% in the high cortisol response group and 83% in the low cortisol response group (P < 0.0001).

Conclusion  Babies born vaginally mount greater cortisol responses at birth than those delivered by caesarean section. Stress at delivery may influence the infant HPA axis response for up to two months.


INTRODUCTION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. METHODS
  5. RESULTS
  6. DISCUSSION
  7. CONCLUSION
  8. Acknowledgments
  9. References

Animal studies have consistently shown that perinatal stress results in long term programming of the hypothalamic–pituitary–adrenal (HPA) axis.1–3 In rodents, caesarean section especially with superimposed anoxia, compared with normal birth, results in long term reductions in corticosterone receptor affinities in the hippocampus and hypothalamus, which in turn result in increased corticosterone responses in adulthood.4

In humans, birth is a major stressor in the perinatal period, and caesarean delivery before labour is known to lower the degree of infant stress, as reflected by cord cortisol concentrations.5,6 To determine whether birth stress in humans, as in animal models, causes long term programming of the HPA axis, we reported in a pilot study that the salivary cortisol response to vaccination at two months was related to mode of delivery on univariate analysis,7 suggesting the possibility of programming as in the animal models. However, there are numerous potentially confounding influences on birth stress including fetal distress and duration of labour. The aim of this study was to investigate the determinants of cord arterial cortisol at birth and to examine the relationship between birth stress and infant salivary cortisol response to vaccination at two months of age.

METHODS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. METHODS
  5. RESULTS
  6. DISCUSSION
  7. CONCLUSION
  8. Acknowledgments
  9. References

Primiparous women with uncomplicated, singleton pregnancies were recruited antenatally from Queen Charlotte's Hospital and consented to participate in the study (n= 539) as approved by our institutional ethical committee. Exclusion criteria were delivery <36 weeks, birthweight <2500 g, known oligohydramnios or abnormal umbilical Doppler waveforms, antenatal corticosteroid treatment, moderate to severe pre-eclampsia or other maternal conditions that compromise fetal wellbeing. Any babies that were severely acidaemic at birth (cord arterial pH < 7.00) were excluded from the analysis. Thus, all babies were within the ‘normal’ range of birth stress.

Cord blood was collected from 217 deliveries, yielding 172 complete arterial samples. Both the cord artery and vein were punctured and cord blood gases compared, to confirm that arterial blood was obtained. Umbilical arterial blood has a smaller maternal cortisol contribution than venous umbilical blood. Samples were collected into lithium heparin vacutainers within 20 minutes of delivery, centrifuged immediately (or refrigerated first if taken overnight) and the plasma frozen at −80°C until assay.

Samples were missed or insufficient in some cases due to a clinical need for Rh serotyping or the pressures of a busy labour ward. Some samples were found to be venous blood and therefore discarded. Assays were done in duplicate in batches using a standard solid-phase RIA (DPC, Los Angeles, CA, USA). Intrasssay and interassay coefficients of variation were 4.2% and 6.7%, respectively. Obstetric data including cardiotocograph (CTG) abnormalities, meconium-stained amniotic fluid, length of labour, analgesia and infant Apgar scores were extracted from the maternal notes immediately post delivery, prior to assay blind to this data.

A subgroup of the initial cohort (n= 79) agreed to return at two months for follow up. Reasons given for not returning were mainly having moved from the area or inconvenience. Routine infant vaccinations (diphtheria, tetanus, pertussis, haemophilus influenza type B and meningitis C) were administered using a standard 2-injection technique by the same practitioner. Infant saliva was collected before and 20 minutes after vaccination for cortisol estimation using ‘Salivettes’ (Sarstedt, Leicester, UK).

Cortisol assays were done in duplicate using the same RIA for plasma. To increase accuracy in the estimation of the lower cortisol levels found in saliva, extra control values were added to the standard curve at the lower end of the scale. The pre-vaccination cortisol level was subtracted from the post-vaccination level to give delta (Δ) cortisol (i.e. the cortisol response).

Statistical analysis used SPSS 11 (Chicago, IL, USA) and GraphPad Prism 3.0 (San Diego, CA, USA). Parametric statistics were used for normally distributed data, which for some parameters (arterial and venous cord cortisol) first required ln transformation. All statistical analyses were two tailed.

RESULTS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. METHODS
  5. RESULTS
  6. DISCUSSION
  7. CONCLUSION
  8. Acknowledgments
  9. References

Table 1 shows baseline patient characteristics. Cord artery cortisol varied according to mode of delivery (Fig. 1). Median (range) values were for assisted delivery 624 (297–1146) nmol/L, normal vaginal delivery 581 (184–1178) nmol/L, intrapartum caesarean 363 (167–1078) nmol/L and prelabour caesarean 334 (181–619) nmol/L (P < 0.0001, ANOVA). Bonferroni analysis showed that both vaginal groups had significantly higher values than both caesarean groups (P < 0.001). Cord venous cortisol levels were mostly lower than arterial levels, median (range) values for the four delivery groups were 516 (298–1302) nmol/L, 601 (216–1293) nmol/L, 351 (183–1077) nmol/L and 268 (110–531) nmol/L, respectively (P < 0.0001, ANOVA).

Table 1.  Baseline characteristics of birth cohort and of those who returned at two months.
VariableBirth cohort (n= 172)2-month cohort (n= 79)
  • *

    Denotes not normally distributed, median and range are given, all others are normally distributed and means [SD] are given.

Birthweight3412 [456] g3468 [524] g
Sex83:89, F:M37:42, F:M
Gestation*40 (36–42)40 (36–42)
Arterial cord pH7.25 [0.066]7.24 [0.069]
Apgar (1 minute)*9 (5–10)9 (5–10)
Normal delivery58 (34%)25 (32%)
Assisted delivery34 (20%)15 (19%)
Emergency caesarean41 (24%)20 (25%)
Elective caesarean39 (23%)19 (24%)
Epidural anaesthesia89.50%87.00%
Prostin induction33.70%38.00%
Syntocinon augmentation51.70%57.00%
Cord arterial cortisol512.7 [237] nmol/L512.5 [234] nmol/L
Age at vaccination* 8 weeks (8–10) (n= 79)
2-month baseline cortisol* 3.84 (0–34.6) (n= 79)
2-month delta cortisol 10 [7.9] nmol/L (n= 79)
image

Figure 1. Cord arterial cortisol by mode of delivery. Comparison of all groups; P < 0.0001, ANOVA. Bonferroni analysis reveals significant differences between either vaginal group and either caesarean group, P < 0.001. nvd = normal vaginal delivery; em.luscs = emergency caesarean delivery; el.luscs = elective caesarean delivery.

Download figure to PowerPoint

Cord artery cortisol correlated negatively with cord arterial pH (r=−0.37, P < 0.001, n= 158) (Fig. 2a). Babies with meconium-stained amniotic fluid had higher umbilical artery cortisol levels than those without, median (range), 627 (193–1078) nmol/L and 417 (167–1178) nmol/L, respectively (P= 0.003, n= 172). Visual grading of meconium staining from 0 (clear amniotic fluid) to 3 (thick meconium staining) also gave different cord artery cortisol levels with median (range) for meconium grades 0 to 3 being 417 (167–1178) nmol/L, 614 (193–860) nmol/L, 559 (193–1078) nmol/L and 802 (513–1004) nmol/L, respectively (P= 0.008, n= 172, Kruskal–Wallis) (Fig. 2b). As would be expected incidence of meconium varied according to delivery mode, being higher in interventional deliveries (P < 0.001, χ2). Combined spinal/epidural (CSE) use was associated with a significantly lower cord arterial cortisol than no analgesia or use of Entonox (P < 0.001).

image

Figure 2. (a) Cord arterial cortisol and arterial cord pH at delivery, r=−0.37, P < 0.001. (b) Cord arterial cortisol and grade of meconium-stained amniotic fluid present. Comparison of all groups; P= 0.008, Kruskal–Wallis.

Download figure to PowerPoint

Factors significantly associated with cord arterial cortisol in univariate analyses (mode of delivery, arterial pH, CSE use and meconium passage) were included in a multiple regression analysis. This showed that the most important determinant of cord artery cortisol was vaginal or caesarean delivery (β= 0.42, P < 0.001), but arterial pH (β=−0.22, P= 0.001) and CSE use (β= 0.19, P= 0.004) were also significant, while meconium was a weaker predictor (β= 0.13, P= 0.05); overall R2= 0.40.

None of the prelabour caesarean babies had high arterial cord cortisol (95% < median cortisol in the normal vaginal delivery group), as shown in Fig. 1. However, some in the intrapartum caesarean group did have raised cortisol levels, and this correlated positively with the amount of labour experienced, as determined by cervical dilatation at the time of caesarean (r= 0.45, P= 0.003, n= 41). The length of second stage in vaginal deliveries also correlated positively with cord arterial cortisol (r= 0.24, P= 0.02, n= 92).

Sex, gestation, prostaglandin induction, length of first stage and use of syntocinon augmentation (used in 52% of the cohort), did not influence cord arterial cortisol levels, although there was a trend towards higher cortisol with increasing birthweight (r= 0.18, P= 0.08, n= 92).

Within the vaginal group, multiple regression analysis showed the most important determinants of cord arterial cortisol to be CSE use (β=−0.43, P < 0.001), length of second stage (β= 0.31, P= 0.001) and fetal heart rate baseline, on CTG, at the start of the second stage (β= 0.23, P= 0.02), overall R2= 0.26.

The cohort that returned at two months had similar baseline characteristics to the non-returners (Table 1). The higher the cord arterial cortisol at birth, the greater the rise in salivary cortisol in response to vaccination at eight weeks of age, r= 0.24, P < 0.05 (Fig. 3). There was no relation between cord arterial cortisol and the baseline, pre-inoculation, saliva cortisol levels. The cortisol response to vaccination at two months was not related overall to delivery method (mean Δ values [SD] for assisted delivery 10.4 [10.2] nmol/L, vaginal 9.8 [7.5] nmol/L, emergency 10.1 [9.3] nmol/L and elective caesarean 10.1 [7.7] nmol/L). A negative trend between cord arterial pH and cortisol response to vaccination at two months (r=−0.22, P= 0.06, n= 72) fell short of statistical significance.

image

Figure 3. Cord artery cortisol (ln) and cortisol response to vaccination at two months (nmol/L) [y=−16.9 + 4.3X, r= 0.24, n= 79; 95% CI for slope, 0.017–8.53].

Download figure to PowerPoint

The two-month data were also analysed to compare those in the highest and lowest 15th centiles for cortisol response at birth. At two months of age, these two groups of babies had markedly different cortisol responses to vaccination. Mean Δ cortisol (95% confidence interval) in the low and high cord arterial cortisol groups was 5.0 (1.0–8.9) nmol/L and 14.8 (7.6–21.9) nmol/L (P= 0.017, n= 12 and 13, respectively) (Fig. 4). These low and high stress groups also differed both in arterial pH at delivery [mean 7.28 (7.26–7.31) and 7.20 (7.17–7.24), respectively, P= 0.001] and in mode of delivery (83% and 8% caesarean, respectively, Fisher's Exact Test, P < 0.0001).

image

Figure 4. Cortisol response to vaccination at two months in those in the 15% most and least stressed at birth. Error bars denote SEM. P= 0.017, Students t test.

Download figure to PowerPoint

DISCUSSION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. METHODS
  5. RESULTS
  6. DISCUSSION
  7. CONCLUSION
  8. Acknowledgments
  9. References

This study has carried out the most thorough examination to date of the determinants of cord arterial cortisol levels at delivery. Our study is larger than previous ones, and we have examined in more detail, which of several possible delivery components are the major determinants of cord arterial cortisol level. We have shown that mode of delivery is the most important factor, particularly vaginal versus caesarean, but that use of CSE, fetal acidaemia and length of labour, specifically in the second stage, were independent determinants.

Previous studies have demonstrated differences in cord cortisol between vaginal and caesarean births.8,9 Many have used mixed or venous cord blood, which will contain a higher percentage and varying proportions of maternal cortisol. Although we cannot exclude a maternal cortisol contribution (indeed there will be a component), by using strictly cord arterial blood, we have minimised this and kept the proportion more constant between mother–infant pairs. In our previous study we showed that while there was a large difference in infant cortisol response between caesarean and vaginal deliveries, there were no such differences demonstrated in mothers.5 For the first time we have looked at not just prelabour caesarean sections, but also those done after a varying amount of labour. This has demonstrated that it is not just the final passage down the birth canal that raises cortisol, but that contractions themselves stress the fetus.

We found that CSE use in labour was associated with reduced cord arterial cortisol level independent of cord pH and mode of delivery. Previous studies have demonstrated that both fentanyl and bupivacaine (the standard drugs used in this hospital) cross the placenta in small amounts and are detectable in fetal plasma.10 We know from intrauterine needling studies that fentanyl can suppress the fetal cortisol response to stress.11 So, does fentanyl or bupivacaine in the fetal circulation suppress fetal cortisol production, or is it the maternal cortisol response that is being suppressed by these drugs? We know that fentanyl is highly lipophilic and levels rapidly equilibrate between mother and fetus10; therefore, the answer could well be that both maternal and fetal cortisol production are suppressed. Further research is necessary to answer this question.

There was quite a high attrition rate in this study. However, despite this, there was no difference in the proportion of different deliveries between the cohort that returned at two months and those who did not. Failure to collect cord blood mostly reflected pressures on midwives in a busy labour ward. Non-collection was more common at instrumental deliveries, but rare at elective caesareans (failure rate 68% and 25%, respectively). This may have resulted in some of the most difficult, potentially high fetal stress deliveries being missed, and may help to explain why we did not find a higher level of cord arterial cortisol in those born by assisted delivery compared with normal vaginal delivery, as we had previously.5

In the second part of this study, we have shown that there is a link between cord arterial cortisol levels at delivery and the saliva cortisol response to inoculation at two months. This supports the possibility that an early stressful experience can programme the HPA axis in humans, as in animal models, although more research is necessary to establish this.

Although cord arterial cortisol correlated significantly with both mode of delivery and cortisol response at two months, the two-month response did not differ by mode of delivery per se, as in our pilot study.7 To clarify the effect at two months we subcategorised babies into those with high or low cortisol response at birth, as defined by 15th centile cutoffs of cord artery cortisol level, and showed that the former group had an approximately threefold greater response (Fig. 4). Although the 15th centile cutoff was chosen arbitrarily, delta cortisol, mode of delivery and pH were also significantly different using other cutoffs such as the 20th centile (data not shown). The current finding of a positive correlation between cord arterial cortisol at birth and later cortisol response suggests that it is the degree of stress experienced at birth, rather than the mode of delivery per se, that is important. Nevertheless, caesarean delivery still predominated in the low stress group and vaginal delivery in the high stress group.

One possibility to consider is that infants have genetically programmed different degrees of cortisol response to a particular stressor, which we have observed both at birth and again two months later. Hence, babies with high cord cortisol would have a high response to vaccination. However, the high cortisol response group were mostly vaginal deliveries with a low mean cord arterial pH, while the low cortisol response group were predominantly caesarean deliveries with a higher mean pH. This suggests that, although there may well be a genetic component, factors associated with birth itself, at least at the extremes, do affect the HPA response at two months. The trend towards a correlation between cord arterial pH and cortisol response at two months adds further weight to this argument.

Stress is a term commonly used with negative implications. However, a degree of stress at birth is probably beneficial. We know that babies born by elective caesarean section have higher rates of respiratory morbidity, due to less catecholamine stimulation of lung liquid resorption.12 However, high cortisol levels may be detrimental either if they reflect concomitant hypoxaemia, or adversely programme the HPA axis.

If the HPA axis was to be programmed permanently in humans, as in animal models, the implications are considerable. HPA hyperactivity is associated with melancholic depression (indeed antiglucocorticoids are being developed as antidepressant drugs),13 while conversely hypoactivity is associated with atypical depression.14 It remains to be determined whether the apparent HPA up-regulation we have shown at two months in those most stressed at birth continues beyond this age. Further investigation of the cortisol response later in infancy however is likely to be confounded by maturational changes of the HPA axis, such as the development of diurnal variation and the cortisol response hypoactivity observed by one year.15

Even if the effects demonstrated here last for only two months, they could still be clinically relevant. It is well established in adults that high levels of cortisol are associated with suppression of the immune response and greater susceptibility to infection.16 Further, increased cortisol responsivity in infancy has been associated with a failure to cope with new situations, such as maternal separation,17,18 although other studies show inconsistent links between cortisol reactivity and infant behaviour.7,19

This study adds weight to evidence suggesting that prenatal stress may have longer-term implications as infants develop. Delivery itself may well be a sufficient stressor to cause longer lasting changes in the HPA axis; more research is necessary to gauge the long term implications of early life stressors on physical and mental health.

CONCLUSION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. METHODS
  5. RESULTS
  6. DISCUSSION
  7. CONCLUSION
  8. Acknowledgments
  9. References

Cord artery cortisol levels vary by mode of delivery, with caesarean-delivered babies having considerably lower levels than vaginally delivered ones. Cord arterial cortisol is related to the rise in cortisol in response to vaccination at two months. This suggests that stress at birth could alter physiological responses in the infant, at least for the first two months of life.

Acknowledgments

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. METHODS
  5. RESULTS
  6. DISCUSSION
  7. CONCLUSION
  8. Acknowledgments
  9. References

The authors would like to thank the Sports Aiding Medical Research for Children (SPARKS) who funded this study. We are grateful to Diana Adams for practical help in the clinics, to the midwives for their cord blood collections and to the mothers and babies, without whom this study would not have been possible. We acknowledge infrastructural support from the Henry Smith Charity and the Institute of Obstetrics and Gynaecology Trust.

References

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. METHODS
  5. RESULTS
  6. DISCUSSION
  7. CONCLUSION
  8. Acknowledgments
  9. References

Accepted 6 December 2004