Is hyperchloraemic acidosis a problem in children with gastroenteritis rehydrated with normal saline? Authors' reply

The letter from Eisenhut¹ questions the benefits of intravenous rehydration of children with gastroenteritis with normal saline (NS) because of the possibility of hyperchloraemic acidosis and suggests that Ringer's lactate or an equivalent solution might offer advantages over NS.

Our study of 102 children with gastroenteritis judged to need intravenous fluids, compared plasma and urinary electrolyte changes after randomisation to receive either 0.9% saline+2.5% dextrose (NS, n = 51) or 0.45% saline+2.5% dextrose (N/2, n = 51).² The infusion rate was at the discretion of the treating physician according to either a rapid replacement protocol (RRP; 10 ml/kg/h for 4 h) or a standard replacement protocol (SRP; replacement of estimated percent dehydration over 24 h).

Data on acidosis and plasma concentrations of chloride were not presented in our original manuscript² but are available in all children. At baseline, the children were normochloraemic (mean+SD chloride 100.4±3.4 mmol/l) with a raised anion gap. Mean bicarbonate (a surrogate for acidosis) was low (17.8±3.0 mmol/l). Ninety of 102 children had a bicarbonate value of <22 mmol/l associated with a mean anion gap of 21±3.5 mmol/l.

After 4 h (T4) the chloride concentration increased to 105±3.0 mmol/l in the NS group vs 102±2.7 mmol/l in the N/2 group (p<0.001). At the same time both groups experienced a similar improvement in their acid base status, with a rise in bicarbonate (mean±SD T4 bicarbonate NS vs N/2 19±3.2 mmol/l vs 19±2.6 mmol/l, respectively, p = 0.12) and fall in anion gap (mean T4 anion gap NS vs N/2 16±3.8 mmol/l vs 17±2.6 mmol/l, respectively, p = 0.2). Separate analysis of the approximately three quarters of children who completed 4 h of the RRP demonstrated almost identical chloride, bicarbonate and anion gap changes (data not shown). In the 16 children (NS n = 8, N/2 n = 8) who received intravenous fluids for 24 h, the chloride changes persisted (median (range) T24 chloride NS vs N/2 107 mmol/l (105–115) vs 105 mmol/l (100–110), respectively, p = 0.04) but with continued improvement in bicarbonate and anion gap in both groups (median (range) T24 bicarbonate NS vs N/2 19 mmol/l (15–20) vs 20 mmol/l (12–24), respectively, p = 0.4; median (range) T24 anion gap NS vs N/2 16 mmol/l (10–17) vs 16 mmol/l (13–20), respectively, p = 0.4).

Infusion of normal saline was associated with hyperchloraemia but not acidosis even after 24 h of intravenous fluids compared with a lower chloride solution. The near identical changes in bicarbonate and anion gap in these two groups could have one of two explanations. Firstly, the chloride load from normal saline in the volumes given does not cause acidosis. Alternatively, the chloride load may have some impact on acidosis that is offset by the superior ability of normal saline over a hypotonic saline solution to refill the intravascular space and improve tissue perfusion, important in dehydration.

Infusion of large volumes of normal saline including during resuscitation or surgery can result in hyperchloraemic acidosis.³,⁴,⁵,⁶,⁷,⁸,⁹ This has also been reported in response to Ringer's lactate or its equivalents, albeit to a lesser extent.³,⁶,⁹ Studies comparing the two fluids have usually shown no difference in clinical outcome,³,⁵,⁶,⁹,¹⁰ except in those at high risk, such as the elderly.⁷ There are no data in children apart from the report of Durward et al,¹¹ who found that the development of hyperchloraemic acidosis in children after cardiac surgery receiving normal saline as their ”maintenance” fluid had no clinical sequelae. As far as we are aware, there are also no data comparing normal saline and Ringer's lactate in adults who are salt depleted, as is frequently the case with gastroenteritis.¹²

The improvement in plasma sodium in our study patients who received NS versus N/2, together with the similar improvements in bicarbonate and anion gap in the two groups, casts doubt on the clinical significance of the modest hyperchloraemia demonstrated. When infused in large volumes normal saline has been shown to maintain a normal plasma sodium compared to a drop with Ringer's lactate.⁵,⁸ In the setting of pre‐existing salt depletion and non‐osmotic antidiuretic hormone activity such as we have shown in gastroenteritis,¹³ this may be important.

Eisenhut also questions the safety of rapid intravenous fluid replacement in dehydrated children with gastroenteritis. It was not a primary aim of our study to identify clinical and biochemical differences between those who received RRP compared with SRP. However, in both this² and an earlier observational study of children receiving only N/2,¹³ we found that the children receiving more prolonged fluids (either SRP alone or RRP followed by SRP) rather than those receiving RRP alone experienced potentially clinically significant dilutional hyponatraemia. Further, it has been well documented that rapid replacement of fluid deficit either orally or intravenously in childhood gastroenteritis is safe and well tolerated,¹⁴,¹⁵,¹⁶ has significant health economics benefits and is widely recommended.¹⁷,¹⁸,¹⁹