Retrospective study of association between choice of vasopressor given during spinal anaesthesia for high-risk caesarean delivery and fetal pH☆

Introduction 

Vasopressors are frequently used to manage hypotension during spinal anaesthesia for caesarean section. Ephedrine was the traditional vasopressor of choice because it was considered to be better for uterine perfusion than pure α-adrenergic agonists such as phenylephrine. Recent studies have found advantages when using phenylephrine. In low-risk pregnancies, phenylephrine is associated with improved fetal acid-base status[1], [2], [3], [4], [5] and with reduced maternal nausea and vomiting,[2], [3], [5] compared with ephedrine. In the UK there has been a significant change in practice. In 1999, 95% of obstetric anaesthetists used ephedrine as the sole vasopressor,6 whereas in 2006, 51% used phenylephrine as the first-line vasopressor.7 In North America, a survey of Society for Obstetric Anesthesia and Perinatology members in 2007 found an approximately even split between ephedrine and phenylephrine use for scheduled caesarean delivery.8 The updated report on the Guidelines for Obstetric Anesthesia by the American Society of Anesthesiologists has recommended that, in the absence of maternal bradycardia, phenylephrine may be preferable to ephedrine because of improved fetal acid-base status in uncomplicated pregnancies.9 However, there is little evidence to inform choice in complicated pregnancies.

The aim of this study was to compare umbilical artery pH with phenylephrine and ephedrine during spinal anaesthesia where caesarean section was performed because of an increased risk of fetal compromise: high-risk caesarean section. We have studied fetal pH during a four-year time period when there was a transition from the predominant use of ephedrine to phenylephrine as the first-line vasopressor for high-risk caesarean sections at our hospital.

Methods 

This was a retrospective observational study. Following discussion with the Chairman of the local ethics committee, formal approval was not considered to be necessary for retrospective chart reviews at our institution. We identified all women from four consecutive years who had a high-risk caesarean section under spinal anaesthesia for a singleton delivery, where fetal umbilical arterial and venous blood gases had been recorded at delivery. High-risk caesarean section was defined as caesarean section for non-reassuring fetal heart rate tracing, dystocia, pregnancy-induced hypertension, growth restriction, antepartum haemorrhage, prolonged rupture of membranes, or cord prolapse. Dystocia was included because of the association between prolonged labour and fetal acidosis. We then reviewed the notes, recording maternal and fetal demographic and operative data. Blood gas values, taken from a double-clamped segment of umbilical cord at delivery, 5-min Apgar scores assessed by a midwife and admission to the neonatal unit, were recorded.

During the period of study, ephedrine was routinely given as 6-mg boluses, and phenylephrine as 100-μg boluses, at the discretion of the anaesthetist. Phenylephrine infusion was recommended only to be given according to a standard protocol, which had been developed for a prospective study completed in 2001 at our hospital.2 Phenylephrine was started at 33μg/min immediately following spinal injection and then titrated aiming to keep systolic arterial pressure at baseline. The infusion rate was doubled, or halved, as required. The maximum infusion rate was 67μg/min. If there was hypotension despite the prophylactic infusion, 100-μg boluses of phenylephrine were given. There were no guidelines for ephedrine infusion. Criteria for admission to the neonatal unit were gestation <34 weeks, weight <1800g or “poor condition”.

Results 

There were 594 high-risk caesarean sections during 2000-03, inclusive. One hundred and thirty-eight of these were not included because they did not have both umbilical arterial and venous blood gas results (46 had no blood gas value recorded, 39 had only one sample and for 53 there was <0.02 difference between the two samples). Indications for caesarean section were similar for those with or without blood gas results. Of the 456 with blood gases, 56 sets of notes could not be traced and 15 cases were excluded because they received a combination of ephedrine and phenylephrine as first-line therapy. Data from the remaining 385 cases were analyzed: 115 received no vasopressor, 122 ephedrine (group E) and 148 phenylephrine (group P) (Table 1). Sixty-six percent of group P were given an infusion as first line therapy, compared with 9.8% of group E. Seventy-four percent of group P delivered in the second two years of the four year study period (time period 2002/03), compared with 20% of group E.

Table 1. Choice of vasopressor and method of administration, by time period

		No-vasopressor group (n=115)	Ephedrine group (n=122)		Phenylephrine group (n=148)
			Bolus (n=110)	Infusion (n=12)	Bolus (n=51)	Infusion (n=97)
	Time period 2000/2001	74	87	10	29	9
	Time period 2002/2003	41	23	2	22	88

Data are numbers.

The no-vasopressor, ephedrine and phenylephrine groups were similar for demographic data, but there were differences for diabetes, labour, pregnancy-induced hypertension, spinal local anaesthetic dose and hypotension (Table 2). We were unable to find accurate records for maternal oxygen administration, or the time from uterine incision to delivery.

Table 2. Maternal and fetal demographic data, indications for delivery and operative data

		No-vasopressor	Group E	Group P	P
		(n=115)	(n=122)	(n=148)
	Maternal age (years)	28 (22 – 30)	29 (25 – 33)	29 (24 – 32)	0.30
	height (m)	1.62 (1.60 – 1.69)	1.62 (1.58 – 1.68)	1.62 (1.57 – 1.65)	0.57
	weight (kg)	65 (54 – 73)	64 (57 – 74)	66 (58 – 76)	0.58
	Gestation (weeks)	38 (34 – 40)	39 (36 – 41)	37 (35 – 40)	0.075
	< 37 weeks	38%	30%	39%	0.25
	< 34 weeks	21%	13%	18%	0.41
	Birthweight (g)	2820 (1900 – 3530)	3140 (2300 – 3550)	2920 (1910 – 3500)	0.14
	< 2500g	38%	29%	40%	0.13
	< 1500g	15%	10%	15%	0.057
	Previous caesarean section	10%	17%	16%	0.19
	Maternal diabetes	0%⁎	4.9%	2.0%	0.042
	In labour	49%⁎	62%#	41%⁎	0.004
	Indication(s) for caesarean section
	Non-reassuring fetal heart rate	37%	38%	33%	0.71
	Dystocia	27%	37%	24%	0.050
	Pregnancy induced hypertension	31%⁎,#	16%	16%	0.003
	Growth restriction	16%	13%	22%	0.16
	Antepartum haemorrhage	6.1%	6.6%	12%	0.21
	Prolonged rupture of membranes	2.6%	0.8%	3.4%	0.37
	Cord prolapse	0.9%	0.8%	1.4%	0.89
	Operative data
	Bupivacaine dose (mg)	12.0 (11.0 – 12.5)#	12.0 (11.0 – 12.5)#	12.5 (11.5 – 12.5)⁎	0.006
	Spinal-delivery (min)	22 (18 – 27)	22 (16 – 28)	22 (17 – 28)	0.96
	Blood pressure <90mmHg	6.1%⁎,#	17%	20%	0.005
	Anaesthetic consultant directly involved	13%	18%	21%	0.20
	Obstetric consultant directly involved	16%	10%	24%	0.095

Data are median (IQR), or percent. P value column refers to comparison of the three groups.

⁎ P<0.05 compared with ephedrine. # P<0.05 compared with phenylephrine.

The median total dose of ephedrine in group E given before delivery was 12mg (IQR 6-18); the median total dose of phenylephrine in group P given before delivery was 200μg (IQR 100 – 400). We were unable to find accurate records of the dose of vasopressor given by infusion. Thirteen percent of group E were given a second-line vasopressor (all phenylephrine boluses, median total dose 200μg [IQR 100 – 400]) compared with 5% of group P (all ephedrine boluses, median total dose 9mg (IQR 3 - 15)) (P=0.014).

There was no difference in umbilical artery pH between the three groups on direct comparison (Table 3). Table 4 shows the results of separately regressing umbilical artery pH on each potentially explanatory variable. Following forward stepwise multiple regression analysis, the only variable that was associated with altered pH was non-reassuring fetal heart rate trace (Table 5).

Table 3. Fetal outcomes

		No-vasopressor	Group E	Group P	P
		(n=115)	(n=122)	(n=148)
	Arterial blood gas values:
	pH	7.26 (7.21 – 7.30)	7.27 (7.22 – 7.30)	7.28 (7.22 – 7.32)	0.21
	PO2 (kPa)	1.5 (1.0 – 2.2)	1.5 (1.0 – 2.1)	1.6 (1.0 – 2.0)	0.93
	PCO2 (kPa)	6.8 (6.2 – 7.9)	7.1 (6.3 – 7.9)	6.9 (6.1 – 7.3)	0.29
	Base excess (mEq/L)	−2.3 (−4.8 to −1.1)	−2.4 (−4.9 to −1.0)	−2.8 (−4.5 to −1.4)	0.68
	Venous blood gas values:
	pH	7.32 (7.28 – 7.35)	7.33 (7.29 – 7.36)	7.34 (7.29 – 7.37)	0.28
	PO2 (kPa)	2.7 (2.3 – 3.3)	3.0 (2.4 – 3.7)#	2.7 (2.2 – 3.4)⁎	0.042
	PCO2 (kPa)	5.7 (5.3 – 6.3)	5.8 (5.1 – 6.3)	5.7 (5.2 – 6.2)	0.83
	Base excess (mEq/L)	−2.7 (−4.6 to −1.1)	−2.7 (−4.9 to −1.2)	−2.3 (−4.1 to −1.1)	0.63
	Other fetal outcomes:
	Umbilical artery pH < 7.20	23%	20%	15%	0.20
	5-min Apgar score < 7	2.6%	0%	4.1%	0.089
	Admission to neonatal unit	33%	23%#	37%⁎	0.040

Data are median (IQR) or proportion. P-value column refers to comparison of the three groups.

⁎ P<0.05 compared with ephedrine; # P<0.05 compared with phenylephrine.

Table 4. Results of separately regressing umbilical artery pH on each potential explanatory variable

	Potential explanatory variable	Coefficient b	Standard error (b)	t	P
	Ephedrine group	−0.0096	0.0092	−1.0	0.30
	Phenylephrine group	0.016	0.0088	1.81	0.071
	Infusion	0.017	0.0094	1.84	0.066
	Time period 2002/03	0.018	0.0085	2.15	0.032
	Maternal age (years)	0.00084	0.00072	1.18	0.24
	Maternal height (cm)	0.00052	0.00066	0.79	0.43
	Maternal weight (kg)	0.000051	0.00031	0.16	0.87
	Gestational age (weeks)	−0.0023	0.0011	−2.11	0.035
	Fetal weight (kg)	−0.0031	0.0044	−0.70	0.48
	Previous caesarean section	0.002	0.012	0.17	0.87
	Diabetes	−0.046	0.028	−1.64	0.10
	Labour	−0.0077	0.0086	−0.89	0.37
	Non-reassuring fetal heart rate	−0.031	0.0088	−3.51	<0.001
	Dystocia	0.0072	0.0094	0.77	0.44
	Pregnancy-induced hypertension	0.015	0.011	1.40	0.16
	Growth restriction	0.010	0.011	0.90	0.37
	Antepartum haemorrhage	0.024	0.015	1.55	0.12
	Prolonged rupture of membranes	0.020	0.028	0.70	0.48
	Cord prolapse	−0.034	0.042	−0.81	0.42
	Bupivacaine dose (mg)	0.0023	0.0030	0.77	0.44
	Spinal-delivery interval (min)	0.00071	0.00054	1.30	0.19
	Hypotension	−0.0042	0.012	−0.36	0.72
	Anaesthetic consultant directly involved	−0.00099	0.011	−0.09	0.93
	Obstetric consultant directly involved	0.010	0.011	0.89	0.37

Table 5. Result of forward stepwise multiple regression analysis for umbilical artery pH on possible explanatory variables from Table 4

	Variable	Coefficient b	Standard error (b)	t	P
	Constant	7.26	0.009
	Non-reassuring fetal heart rate	−0.031	0.0088	−3.51	<0.001

	Source of variation	Degrees of freedom	Sum of square	Mean squares	F	P
	Regression	1	0.084	0.084	12.3	<0.001
	Residual	383	2.6	0.0068

r2=0.03.

On direct comparison, there was no difference in the incidence of umbilical artery pH <7.20, or 5-min Apgar score <7, between the groups, but there was a difference in the incidence of admissions to the neonatal unit (Table 3). However, if neonates who automatically fulfilled the criteria for neonatal unit admission because of gestation <34 weeks or weight <1800g were excluded from analysis, there was no difference between the groups: no-vasopressor: 11 of 85 (13%); group E: 12 of 104 (12%); group P: 21 of 112 (19%) (P=0.29).

Discussion 

Our observations for umbilical artery pH differ from those in low-risk patients, which show a higher pH with phenylephrine.[1], [2], [3], [4], [5] Prophylactic ephedrine infusion during low-risk caesarean section has been associated with a high incidence of fetal acidosis.[2], [4], [5], [11], [12]

We have previously suggested that ephedrine can cause a β-adrenergic-mediated increase in fetal carbon dioxide production leading to respiratory acidosis.2 Prophylactic ephedrine infusion has also been associated with fetal metabolic acidaemia.[3], [4], [5], [11] This may be secondary to an increase in fetal metabolic rate with ephedrine, to the extent that fetal oxygen demand exceeds its limited supply, thereby increasing anaerobic metabolism. Ephedrine-induced fetal acidosis appears to be associated both with the total dose of ephedrine given before delivery and with the duration of fetal exposure to ephedrine, but not with hypotension.4 This is probably because hypotension is unlikely to lead to adverse fetal biochemical changes if it is rapidly treated. This is supported by a study of low-risk caesarean sections in which all patients received bolus doses of phenylephrine if hypotension developed, but 50% also received a prophylactic phenylephrine infusion.13 Although 88% of the bolus-only group had hypotension, compared with only 23% of the prophylactic phenylephrine infusion group, umbilical artery pH was similar.

In this study of high-risk caesarean sections, the main reason that we did not find a difference in umbilical artery pH between ephedrine and phenylephrine was probably that the majority of the ephedrine group received a relatively low dose of ephedrine, thereby reducing its fetal metabolic effects. The median dose of ephedrine given before delivery to the 90% of the ephedrine group who received boluses as first-line treatment was only 12mg (IQR 6-18). This contrasts with the approximately four-fold greater dose of ephedrine (median 52mg [IQR 32-71]) given before delivery in a low-risk caesarean section study at our hospital, which was associated with a low umbilical artery pH (median 7.20 [IQR 7.10 – 7.28], compared with 7.27 [IQR 7.22 – 7.30] in this high-risk study).4

Prematurity and labour may have contributed to the lack of difference between the vasopressor groups in this high-risk study by reducing hypotension and, therefore, vasopressor requirements. Seventy-eight percent of those who received vasopressor were either not at term or in labour. Both of these variables have been associated with reduced hypotension during spinal anaesthesia,[14], [15] and prematurity has been associated with reduced ephedrine requirements.14 The urgent nature of the surgery for many of the high-risk cases may also have reduced the difference between the groups by reducing the spinal-delivery interval and, therefore, the total dose of ephedrine given before delivery and the duration of fetal exposure.

A recent prospective study of non-elective caesarean sections by Ngan Kee et al. comparing the fetal effects of phenylephrine and ephedrine, supports our observations, finding a similar umbilical artery pH with ephedrine and phenylephrine.16 Analogous to our observations with low- and high-risk studies, they gave the ephedrine group in their non-elective study relatively low doses of ephedrine (median 10mg [IQR 10 – 30]) and recorded a relatively high median umbilical artery pH (7.28 [IQR 7.24 – 7.31]). This contrasts with the ephedrine group in their low-risk study, who received higher doses of ephedrine (median 54mg [IQR 42 – 71]), but had a low umbilical artery pH (median 7.21 [IQR 7.12 – 7.25]).5 In the non-elective study premature babies were excluded, but labour, the urgent nature of the surgery, and the use of reactive boluses rather than a prophylactic infusion, probably explained the low doses of ephedrine used.

We observed umbilical venous PO2 to be greater with ephedrine than with phenylephrine, but we cannot draw any conclusions from this observation because the difference was small and we do not have an accurate record of maternal oxygen administration, which can affect umbilical venous PO2.17

The ephedrine and phenylephrine groups were not matched for potential confounding variables, such as time period of operation, method of vasopressor administration, labour and bupivacaine dose. This could have biased the univariate analysis. Because of the retrospective observational design of the study, we used multiple regression analysis to take into account the effect of potential explanatory variables, other than choice of vasopressor, on umbilical artery pH. However, it is possible that the effects of other variables were not fully accounted for. The median bupivacaine dose was 1.0mg greater in 2002-03 than in 2000-01, while the proportion of women receiving no vasopressor decreased. There was, however, no difference in the incidence of hypotension, diabetes or pregnancy-induced hypertension in the two time periods. The proportion of women receiving no vasopressor in the second time period probably decreased because the use of prophylactic infusions increased. Bupivacaine dose and infusion were entered into the regression analysis. As there may have been other changes in practice over time that were less clear, time period was included as a potential explanatory variable.

The incidence of hypotension was remarkably low in both our vasopressor groups. Arterial pressure was documented by hand, so there may have been a degree of selective recording, or rounding up of readings. Other factors contributing to the low incidence of hypotension may have been the high proportion that were either in labour or not at term.[14], [15]

In conclusion, this study found similar fetal umbilical artery pH values in high-risk caesarean sections performed under spinal anaesthesia, whether phenylephrine or ephedrine were used to maintain arterial pressure. This contrasts with low-risk studies, which have found a lower umbilical artery pH with ephedrine. This may be explained by the relatively low doses of ephedrine that were used in our high-risk study, thereby minimizing the fetal metabolic effects of ephedrine.