Introduction

At a meeting of the New York Obstetrical Society in December 1945, Curtis Mendelson first described his studies on aspiration of stomach contents associated with general anaesthesia (GA) in pregnant women – these findings were published the following year [1]. Mendelson had noted 66 cases of aspiration of stomach contents into the lungs in over 44 000 pregnancies at the New York Lying-In Hospital from 1932 to 1945, an incidence of 1.5 per 1000 deliveries. In 45 of these cases, the aspirated material was recorded; 40 mothers had aspirated liquid and 5 had aspirated solid food. Only two of the mothers actually died, both dying on the delivery table as a result of suffocation after ingesting solid material. The surviving women went on to develop an aspiration pneumonitis, thought to be due to the aspiration of gastric acid. So the risks of gastric aspiration were noted to be two-fold: the aspiration of solid particles sufficiently large to obstruct the airway, and the pneumonitis secondary to acidic gastric contents.

The significance of this seminal paper lies not only in the description of this eponymous condition for the first time, but also on the profound effect it had on the subsequent management of labouring women. Nil-by-mouth (NBM) policies were introduced into many Western labour wards in the 1940s and 1950s in the belief that this would reduce the incidence of pulmonary aspiration of acidic gastric contents should GA be required.

Pulmonary aspiration is increasingly rare, and in the years 1994–2012 only one maternal death from the aspiration of gastric contents under general anaesthesia has been recorded in the UK, an 18-year period when approximately 12 million women delivered [2–7].

The fact that maternal death from pulmonary aspiration of gastric contents has virtually disappeared has led to relaxations in the ‘NBM rule’, in Europe, at least. In a postal survey of 351 maternity units in England and Wales, almost 33% allowed labouring women to both eat and drink [8]. In a smaller study of clinical practice in the Netherlands, only 14% of the obstetricians surveyed had a restrictive policy during normal labour [9]. This non-restrictive policy was not associated with a higher mortality due to Mendelson’s syndrome.

By contrast, the practice of restricted oral intake in normal labour appears to persist in the United States. In an anonymous survey of obstetricians and anaesthetists, approximately 90% restricted patients to clear fluids throughout labour – although this view represented a response rate of only 33% of those surveyed [10]. In April 2007, the American Society of Anesthesiologists, in a practice guideline for obstetric anaesthesia, recommended that ‘solid foods should be avoided in labouring patients’ [11]. This advice was echoed two years later in an American College of Obstetricians and Gynecologists (ACOG) Committee Opinion published in September 2009 [12].

Practical Obstetric Considerations

A number of anatomical and physiological changes occur in pregnancy and labour which increase the risk of pulmonary aspiration, if GA is required. Many of these have been exacerbated by the rising rates of maternal obesity witnessed in the last decade.

The anatomical changes that can increase the difficulty of tracheal intubation in obstetric patients include raised body mass index (BMI) and enlarged breasts.

National data published for the UK in 2010 identified the growing numbers of obese pregnant women [13]. The study found that almost 5% of the UK maternity population were severely obese (BMI ≥35), while just over 2% were morbidly obese (BMI ≥40). An increased risk of obstetric complications, including operative delivery, was also noted in obese women.

This obesity and enlarged breasts can make laryngoscopy more difficult, particularly if the situation is exacerbated by the laryngeal oedema that may complicate common obstetric conditions, such as pre-eclampsia.

These considerations may also jeopardize effective airway protection by making it more difficult to apply cricoid pressure correctly.

In pregnancy, symptomatic gastro-oesophageal reflux is much more likely to occur. This occurs due to a combination of rising intra-abdominal pressure caused by the gravid uterus, and the steady decline in lower oesophageal sphincter tone as a result of high progesterone levels. The tone of the sphincter is also reduced by GA – increasing the likelihood of reflux of gastric contents into the oropharynx, and their subsequent pulmonary aspiration.

Labour also causes significant depression of gastric motility and delays in gastric emptying in proportion to its duration [14]. Following a standard ‘meal’ of 750 ml of water containing a dye, the volume of fluid remaining in the stomach after 30 min was compared in three groups of women: non-pregnant, late pregnancy (pregnancy of at least 34 weeks’ gestation) and labouring. The mean volume remaining during labour was significantly higher than in non-pregnant women and pregnant women before labour. The authors proposed that these results might be attributed to the pain and emotional disturbances that accompany labour.

Finally, narcotic analgesics exacerbate the problem of delayed gastric emptying in labour, and women who have received intramuscular narcotic analgesics such as pethidine or diamorphine in labour are more likely to vomit or to have increased volumes of gastric residuals at delivery [15].

A postal survey of UK obstetric practice published in 2008 found that the use of such intramuscular opioids is still widespread, at least in consultant-led maternity units. A total of 234 consultant-led obstetric units in the UK were questioned about their use of intramuscular opioids, and of units responding to the survey questionnaire over 84% were using pethidine and 34% used diamorphine [16].

Oral Intake in Labour

Labour is a metabolically challenging time for mother and fetus. Non-diabetic women in late pregnancy have been shown to exhibit a state of ‘accelerated starvation’ if denied food and drink [26]. This results in the increased production of ketones – in particular β-hydroxybutyrate and aceto-acetic acid – and the non-esterified fatty acids from which they are derived, and significant reductions in plasma glucose and insulin levels. The changes seem to occur equally in lean and obese women. These physiological changes are exacerbated by the metabolic changes of labour, and the authors caution against the common practice of skipping breakfast in pregnant women.

The production of ketones is a normal physiological adaptation to generate an alternative energy supply when glucose supply is limited. Despite this, concerns about the detrimental effects of maternal starvation and subsequent ketosis in labour in the 1960s and 1970s led to the intrapartum administration of high-dose intravenous dextrose solutions as a preventative measure. However, it is now clear that ketones are not so detrimental to the mother and fetus as once thought. In addition, it quickly became apparent that while strategies employing high-dose intravenous dextrose throughout labour could swiftly correct maternal ketosis, the practice was associated with adverse consequences for the mother and fetus, so the practice was abandoned [27].

Attention subsequently switched to intrapartum measures that might attenuate the metabolic consequences of eating and drinking in labour. These have been evaluated in a limited number of randomized controlled studies of labouring women considered as at low risk of needing a GA (there are no studies looking at high-risk women in labour).

Unrestricted Diet in Labour

Only one randomized study has assessed the effect of unrestricted oral intake in labour compared with complete restriction [28]. In 2005, Tranmer et al. reported on the effect of oral intake in 328 low-risk nulliparous women in Canada. Women randomized to the ‘usual care’ arm of the study (N = 165) were permitted only ice chips, popsicles or sips of fluid during the active phase of labour. Those in the intervention group (N = 163) were allowed unrestricted access to their choice of foods and fluid during labour, although encouraged to eat ‘easily digestible’ foods or fluids. The primary outcome of this study was the incidence of labour dystocia and there were no significant differences in this, or any other labour and delivery outcomes, between the two groups. The reported rates of thirst, hunger and nausea were also comparable. Interestingly, only 56% of those allocated to the unrestricted diet group reported that they actually ate or drank some form of carbohydrate in labour.

Restricted Diet in Labour

Two studies have assessed the comparison of low-residue food in labour with restriction to water only.

In 1999, Scrutton et al. published the results of a prospective study examining the effect of light diet on the metabolic profile, outcome of labour and the risk of aspiration in a group of 94 women delivering in a university teaching hospital [29]. Women presenting in early labour (cervical dilatation less than 5 cm) were randomly allocated to one of two groups; those in the eating group (N = 48) were permitted to select from a low-residue diet throughout their labour, whereas those in the starved group (N = 46) were permitted water only. The two groups were similar with respect to age, parity, induction status and cervical dilatation at the time of randomization. By the end of labour, plasma β-hydroxybutyrate and non-esterified fatty acids were significantly lower in the eating group. Conversely, the plasma glucose and insulin levels were higher in the eating group. However, those women who ate had significantly higher gastric volumes within an hour of delivery, and were twice as likely to vomit at or around delivery. The volumes vomited by women in the eating group were also significantly larger than volumes vomited by women in the starved group (309 ml vs. 104 ml). The vomit contained a considerable amount of solid and semi-solid residue.

There were no significant differences between groups with respect to duration of labour, mode of delivery or neonatal outcome. The rates of assisted delivery in this relatively small study were only 25% in each group, implying that those studied were genuinely representative of a low-risk population.

The restricted diet was generally well tolerated by those in the intervention group, though there was a progressive decrease in food consumption as labour advanced.

These results suggest that those women allowed a light diet in labour are at reduced risk of developing ketosis without any discernible benefit to the labour progress or outcome for mother or fetus. However, the residual gastric volume around the time of birth is significantly higher in the eating group, who are almost certainly at greater risk of pulmonary aspiration should they require an emergency GA.

A far larger prospective, randomized study from the same unit was published in 2009 [30]. A total of 2426 nulliparous, low-risk women were randomized on admission to either the ‘eating’ group (N = 1219) or to the ‘water-only’ group (N = 1207). The two groups were comparable with respect to their age, ethnic origin, prelabour food intake, need for intravenous fluids and use of prostaglandin and oxytocin.

Both groups of women had free access to water, but the women in the ‘water-only’ group received water or ice chips only and were encouraged not to eat if they requested to do so. Those randomized to the ‘eating’ group were encouraged to consume a low-fat, low-residue diet at will during their labour, and the foods they consumed included fruit juice, soup, cereal, biscuits, fruits, chocolate, toast, vegetable stew, sandwiches, burgers, chicken and rice.

Over 5% of all women recruited to this study consumed nothing during labour. Of the women allocated to the ‘water-only’ group, 20% failed to adhere to the study protocol and consumed food in their labour. Conversely, 29% of those who had been actively encouraged to eat during their labour chose not to do so.

There were no significant differences between the two groups in any of the following outcome measures:

This study found that eating in low-risk labour did not influence obstetric outcomes, including mode of delivery and the duration of labour. The study was not powered to show any difference in rates of gastric aspiration, since the prevalence of this condition is now so low that any such study would need to recruit vast numbers of women.

Although the women recruited to the study were all ‘low risk’, the intervention rates were quite high, with CS rates of 30% in both groups – implying quite a highly medicalized birthing environment. Notwithstanding this, the findings from this large study dominate the findings in subsequent meta-analysis of the benefits and harms of restricting oral intake during labour [31].