Discussion

Principal findings and interpretation
Hyponatraemia defined as plasma sodium ≤130 mmol/l was found in 21 women after delivery. The reduction in plasma osmolality confirms the dilutional origin of hyponatraemia. Hyperglycaemia might have contributed to hyponatraemia in the six hyperglycaemic women.15 All study participants drank similar amounts per hour and received moderate hourly volumes of intravenous fluids with a mean sodium content of 69 mmol/l. Intravenous fluids were thus hypotonic, but with a sodium content higher than the most common glucose solutions. Two-thirds of all fluids were orally ingested, and these fluids are invariably hypotonic. In women with longer duration of labour, the cumulative effect of approximately 300 ml of hourly fluid intake resulted in a significant reduction of plasma sodium levels. The duration of labour itself cannot explain the development of hyponatraemia. On the contrary, longer lasting labour without administration of fluids would result in dehydration and hypernatraemia. Hyponatraemia may possibly have a negative influence on the process of labour, as hyponatraemia was significantly correlated with longer duration of second stage, instrumental delivery, and emergency caesarean for failure to progress. In women delivered by emergency caesarean section, hyponatraemia was not caused by intraoperative fluids as this was present before initiating anaesthesia. Earlier studies analysed cord sodium concentrations and found no difference in maternal and cord sodium concentration.1,3 We analysed both umbilical arterial and venous sodium concentrations and found that umbilical arterial concentrations were higher than maternal levels but significantly correlated, indicating fetal equilibration with maternal hyponatraemia. In infants, large weight loss, hyperbilirubinaemia, and respiratory distress have been believed to be caused by fluid overload.4,5,12 In our study, weight loss in excess of 10% of birthweight was most frequent in fluid group 3 (P = 0.05), indicating a possible relationship with maternal fluid overload. The few cases of hyperbilirubinaemia and respiratory problems were evenly distributed, with no differences between the groups (Table 2).

Hyponatraemia in the obstetric patient
Sodium and the corresponding anions are the main osmoles in extracellular fluid, and sodium concentration mainly reflects body water content. Reabsorption of water in the collecting ducts of the kidney is controlled by vasopressin, in turn regulated by plasma osmolality.15 Vasopressin is secreted in response to increased osmolality, causing water retention that will restore osmolality. Reduction of plasma osmolality by approximately 10 mOsm/kg occurs in early pregnancy mainly as a result of a decrease in plasma sodium by 3–5 mmol/l. By 12 weeks of gestation, this adaptation has stabilised, and plasma osmolality remains low until after delivery. Release of vasopressin in response to adequate physiological stimuli corresponds to the nonpregnant state, but a lower osmolality is maintained.16 The maximum capacity of excreting a water load at rest is reported as approximately 900 ml per hour in healthy women but is reduced by one-third in late pregnancy.17 Pain, stress, and fear are nonosmotic stimuli for vasopressin secretion. Hypovolaemia is, however, the most potent stimulus for vasopressin secretion and may cause water retention even in the presence of hypoosmolality.18 Oxytocin, a short peptide with a structure similar to vasopressin, may also cause water retention by stimulation of the specific V2 vasopressin receptors in the kidney. This receptor is not downregulated during prolonged stimulation as is the case with the oxytocin receptor. However, the infusion rate of oxytocin must exceed 20 mU/minute to cause antidiuresis in humans.19 Hyponatraemia during labour caused by oxytocin administered in electrolyte-free solutions is well recognised, but despite this, hypotonic solutions are often used as a vehicle for oxytocin.20 In our study, fluid volumes needed to administer oxytocin were modest, whereas remaining fluid administration was unrestricted (Table 1). Multivariate logistic regression showed that total fluid volume correlated significantly with hyponatraemia (P < 0.001), whereas oxytocin administration did not significantly correlate with hyponatraemia (P =0.072) (Table 3). Women in fluid groups 2 and 3 received oxytocin at rates necessary for oxytocin to express antidiuretic effect but probably during too short a period of time for significant antidiuretic effect to develop. The administration of hypotonic fluids in conditions with increased vasopressin activity can cause dilutional hyponatraemia.15,18,21 Typically, this may occur postoperatively, but during labour stimuli for vasopressin release are abundant. Labour itself is, however, not a situation causing inappropriate vasopressin secretion.22 The women in our study received fluid volumes well below their predicted maximum capacity of renal excretion at rest; therefore, the development of hyponatraemia indicates increased vasopressin activity during labour. Our results indicate that fluid volume is the major determinant of hyponatraemia, but the antidiuretic effects of endogenous vasopressin and oxytocin administration increase the susceptibility of women to develop hyponatraemia during labour. Fetal vasopressin levels are high during birth, thus also exposing the overhydrated fetus to the risks of hyponatraemia.12
Hyponatraemia is usually defined as a decrease in plasma sodium level below 136 mmol/l.15 More important than the absolute level of hyponatraemia is the speed with which it has developed. Hyponatraemia causes oedema and cellular swelling; initial symptoms of cerebral oedema are irritability, headache, nausea, and vomiting. Subsequently, convulsions and coma can occur, and severe hyponatraemic encephalopathy can cause respiratory arrest and death.15,21 Diagnostic difficulties during labour are obvious as initial symptoms of hyponatraemic encephalopathy are nonspecific and may easily be confused with symptoms of pre-eclampsia.
In the pregnant woman, symptoms might possibly occur at a lower level of plasma sodium due to the pregnancy-induced reduction in plasma sodium. Images by computed tomography show reversible reduction in brain size during pregnancy.23 Although the significance of this reduction remains to be explained, it could be a result of intracellular adaptation to the pregnancy-induced reduction of extracellular osmolality.
The highest morbidity and mortality rates in hyponatraemic encephalopathy are found in women of fertile age.18,21 When life-threatening symptoms occur, therapy must aim at quick restitution of plasma osmolality. Some authors advocate even the use of hypertonic saline and diuretics.15 In all other cases, the simple measure of water depletion will allow plasma levels to return to normal.

Fluids during labour
Practice and policies regarding oral intake during labour show large variations throughout the world. A liberal attitude seems to prevail in the UK, many European countries, and Australia, and some countries even allow solids during labour.9 Clear oral fluids are recommended during uncomplicated labour in the USA.24‘Nil per mouth’ policies are often challenged, particularly by midwives, and create the need of intravenous fluid administration. Maternal starvation may cause fetal acidosis, and several studies address the safety of glucose administration.25,26 Solutions containing glucose are usually hypotonic despite electrolyte addition, although near isotonic solutions have been used for study purposes.26
Two studies indicate that women receiving 250 ml per hour of Ringer's lactated solution intravenously had shorter duration of labour and less need for oxytocin than those receiving 125 ml per hour.7,8 The assumption regarding beneficial influence of larger volumes of fluids during labour is largely based on literature in the field of sports medicine.7,8,27 For optimal muscle performance, athletes are recommended to drink ‘the maximal amount tolerated’ as the sensation of thirst is believed to underestimate the real fluid requirements during exercise.27 Thirst is, however, a strong physiologic stimulus, effectively protecting against dehydration when fluids are readily available. No physiologic warning system protects the body against over-hydration. Therefore, suppression of thirst by abundant drinking implies exposure to the risk of hyponatraemia. Since 1991, several deaths due to hyponatraemic encephalopathy have occurred during endurance competitions. Common recognition of excessive drinking as the main cause of these disasters was, however, delayed until 1995.28,29
The duration of labour equals a marathon for many women, hence the importance of careful administration of fluids. Many authors consider 150–200 ml per hour safe to drink during labour, but with simultaneous intravenous administration, this could well be an excessive amount as illustrated by the present study.1,12,13 Also, the tonicity of fluids determines their potential for causing hyponatraemia, and clear oral fluids are invariably hypotonic. Sport drinks are somewhat confusingly described as isotonic. However, their osmolality, even when similar to that of plasma, is largely made up of carbohydrates. Their content in sodium is less than half compared with plasma, rendering these drinks hypotonic. Some prospective randomised trials have been designed to study the impact of sport drinks and carbohydrate intake on labour duration and outcome.30–33 These studies show conflicting results, but all have in common the lack of control of electrolyte status in the study participants.

Strengths and limitations of the study
A study published in 1981 described iatrogenic hyponatraemia to be caused mainly by electrolyte-free intravenous fluids administered as vehicle for oxytocin.1 The authors suggested that oxytocin should be administered at higher concentrations, and fluid balance supervised closely during labour.
The first advice changed practice of oxytocin administration, but more recent case reports of hyponatraemia during labour imply that the second advice was forgotten.34,35 Hyponatraemia during labour, now less frequently iatrogenic, is more often caused by overdrinking.12–14,31 Our study was therefore designed to evaluate the prevalence of hyponatraemia during modern management of labour. The women in our study received oxytocin in moderate fluid volumes, and most of the fluids administered were orally ingested. Nonetheless, 21 women in our study developed hyponatraemia during labour, thus illustrating that the oral route of administration does not diminish the risks of excess water. As in other studies, we found hyponatraemia in women receiving oxytocin as well as epidural analgesia.1,6 The multivariate regression analysis performed in our study shows that hyponatraemia was probably caused by fluid intake, but only associated with oxytocin and epidural analgesia. Modern low-dose epidural regimens do usually not call for large volumes of intravenous fluids to preserve blood pressure.
Our results are all the more important as the habit of drinking large quantities of water has become quite common in the general population.10 Also, a more liberal attitude to fluid administration during labour might have developed. The comparison of fluid volumes administered during labour as reported in scientific papers indicates that such could be the case. In a study published in 1991, evaluating saline or glucose as vehicle for oxytocin, one study group received a mean of 710 ± 640 ml of intravenous glucose.2 A more recent work, published in 2005, studied the effect of unrestricted oral carbohydrate intake.32 The intervention group received 3234 ± 1473 ml of intravenous fluids.
The observational study design has limitations when compared with a randomised controlled trial. However, it can be doubted whether an ethical committee would permit a deliberate exposure to the degree and risks of hyponatraemia observed in the present study. Analysis would have benefited from hourly registration of fluid intake and urine output. This option was considered, but although desirable, the work load imposed could have jeopardised the realisation of the study. Also, such close monitoring of behaviour would have introduced an observational bias. However, it is possible that the women who developed hyponatraemia had larger hourly oral fluid intake leading to symptomatic hyponatraemia. Tiredness and irritability, initial symptoms of hyponatraemia, may have prompted energy supplement by intravenous glucose infusion. This alterative interpretation strengthens the need of registering all oral fluid intakes during labour and remembering hyponatraemia as possible diagnosis. Not all consecutive women were included, omitting many women in advanced labour. The proportion of nulliparas is therefore large in our study population, probably increasing the incidence of hyponatraemia. Infants were examined and treated according to departmental routine; therefore, only clinically significant symptoms were investigated. The results regarding neonatal outcome should be interpreted with these limitations in mind.