Dehydration in acute gastroenteritis

Three ways of assessing the degree of dehydration in children with diarrhoeal disease are clinical assessment, weight gain with rehydration and laboratory tests. The study by Yilmaz et al. in this issue attempts to improve the reliability of assessing dehydration by using laboratory tests.1 This study did not use clinical assessment on admission, but laboratory tests are compared to percentage weight change after recovery. However, the latter is an imperfect gold standard in our experience, and the rehydration period was extended beyond 24 h in this study so the degree of dehydration ranged up to 19%[sic]. In spite of these limitations, the authors found that urea and bicarbonate were useful in assessing the degree of dehydration in acute gastroenteritis. Clinical assessment of dehydration has low sensitivity because clinical signs only become present with moderate to severe dehydration (≥ 5%), and the actual degree of dehydration may be underestimated with obesity and overestimated with wasting or sepsis. An American study found the four signs which best predicted dehydration were a capillary refill time > 2 s, absent tears, dry mucous membranes and ill general appearance.2 A Melbourne study found that poor capillary return was the most reliable clinical sign of dehydration.3 Although capillary refill time is a useful sign of dehydration, it can be affected by factors such as fever, ambient temperature and age.4 Other studies have found laboratory tests to be generally insensitive in assessing hydration, but bicarbonate, urea, creatinine and uric acid have tended to be the most helpful tests.5, 6 The percentage weight gain with rehydration seems a relatively objective measure of dehydration which can be used to verify the clinical assessment after therapy, but it is obviously no help in assessing the degree of dehydration on presentation. It does require two accurate measurements, preferably on the same scale with the child undressed and taken at the same time of day (24 h apart) as longer periods are affected by weight changes from loss of subcutaneous tissues due to the catabolic state or from refeeding. The initial weight is subtracted from the rehydration weight and taken as a per cent of the rehydration weight. The per cent weight change tends to underestimate the degree of dehydration with ongoing stool losses and could potentially overestimate the degree of dehydration if the child were given excessive IV fluids (e.g. developed puffy eyes). Urinary output and specific gravity are helpful to confirm that a child has been adequately rehydrated and is passing frequent dilute urine. In our studies at Royal Darwin Hospital,7 we have compared clinical assessment (using the World Health Organization protocol) with per cent weight change after 24 h rehydration in 324 acute gastroenteritis admissions of whom 91% were Aboriginal children. The mean per cent dehydration (95% confidence intervals) by clinical assessment on admission for the 268 Abori­ginal children from remote communities was 5.0% (4.6−5.3), compared to 3.9% (3.6−4.2) for weight change after 24 h rehydration. Although these two different measures of dehydration appeared to correlate well (Fig. 1), the kappa agreement was only 0.30 and the Pearson correlation 0.56. On multiple linear regression, clinical assessment correlated better than weight change with bicarbonate (P < 0.001 vs. P = 0.62) or venous blood pH (P = 0.001 vs. P = 0.33) and also with urea (P = 0.04 vs. P = 0.55), on adjusting for age and the other measures of dehydration. This model with four degrees of free­dom explained 33.9% of the variability in clinical dehydration whereas only 22.7% was explained in the comparable weight change dehydration model. Assessment of dehydration: clinical versus weight change (means and 95% confidence intervals). Not surprisingly, the degree of clinical dehydration tended to reflect how sick the child appeared. A low serum bicarbonate on admission was the best predictor of dehydration, correlating strongly (P < 0.001) with worse clinical dehydration, higher urea, greater diarrhoea severity (as a score) and younger age, but not with weight change assessment of dehydration (P = 0.48), and these five factors explained 43.2% of the variability in bicarbonate. The same regression model but with venous blood pH on admission replacing bicarbonate had similar findings. Creatinine was significantly less sensitive than urea as a measure of dehydration. From our results, we conclude that weight change after rapid rehydration is not a reliable measure of dehydration, and that plasma bicarbonate (or venous pH) is a better laboratory indicator of significant dehydration than urea. Oral rehydration with an appropriate solution is a highly effective means of rehydration which uses the principle of glucose-facilitated sodium transport.8-10 The optimal concentration of an oral rehydration solution (ORS) is approximately 60mmol/L of sodium, 20 mmol/L of potassium, 110 mmol/L (2.5%) of glucose and an osmolality of about 220.11, 12 This hypo-osmolar ORS has also been used safely in malnourished children without developing hyponatraemia.13 Cereal-based oral rehydration solutions (e.g. rice) have not been shown to have a definite benefit in non-cholera diarrhoea compared to glucose-based oral rehydration solution,12 but there is anecdotal evidence that it may be more palatable. Many studies in industrialized countries have shown that oral rehydration is under-utilized in children with diarrhoea, with excessive reliance on IV fluids in well hydrated children.14, 15 The most likely reason for a child refusing to drink ORS because of the taste is that he/she is not dehydrated as dehydrated children will not refuse to drink ORS because of taste. Oral rehydration may be time consuming for caregivers, particularly with vomiting. Vomiting can be managed with small, frequent amounts of oral rehydration solution (e.g. 5 mL every 2 min),8 but this requires a compliant and motivated caretaker. Paediatric wards in developing countries often supply a large central container of ORS so that mothers can serve themselves with a ladle and rehydrate their children using a cup and spoon. Vomiting is often an early symptom of gastro­enteritis before there is dehydration, so a short history of vomiting and refusal to drink ORS because of taste are not good indications for IV therapy. Some clinicians promote the use of nasogastric tubes for rehydration, but these are unpleasantly invasive, and if admission investigations require a venesection, why not use this IV route if invasive rehydration is required? Reducing unnecessary IV rehydration by resorting to nasogastric tubes seems pointless, unless there are difficulties with IV access. The European Society of Pediatric Gastroenterology, Hepatology and Nutrition (ESPGHAN) has recently published new guidelines for the management of gastroenteritis,16 which include the following ‘6 pillars of good practice’: Use of ORS to correct estimated dehydration in 3−4 h Use of hypo-osmolar solutions (ORS as in 1) Continuation of breastfeeding Commencement of early refeeding after 4 h rehydration Prevention of further dehydration by giving additional ORS (10 mL/kg/watery stool) No administration of unnecessary medication The duration of diarrhoea can be reduced by 0.43 (0.12−0.74) days by early feeding of children with acute gastroenteritis,17 which also has added nutritional benefits. Small frequent feeds (12 per day) also appear to speed up recovery from diarrhoea.18 The best foods to be introduced in the treatment of acute gastroenteritis are complex carbo­hydrates (e.g. rice, wheat, bread, and cereals), yoghurt, fruit and vegetables. The so-called BRAT diet (banana, rice, apple and black unsweetened tea) was a progenitor to this approach. Fatty foods or high sugar foods such as sweetened tea, juices or soft drinks should be avoided.9 Soft drinks, juices and similar solutions tend to be too hypertonic and low in electrolytes. An important advance in the treatment of dehydration in acute gastroenteritis has been the change to rapid rehydration over 4 h. This was introduced as best practice by WHO in the 1980s and has been used by the author in the developing world since then, where it proved safe and effective with limited nursing care (since the only IV flow rates without pumps tend to be wide open and blocked). Rapid rehydration has only been accepted in developed countries in more recent years.8, 9, 16, 19 Rapid intravenous rehydration with Ringer’s lactate (Hartmann’s solution) is now best practice for moderate to severe dehydration or when oral rehydration is inappropriate or fails. The only contraindications to a trial of oral rehydration therapy are shock, coma, ileus and severe hypokalaemia. Rapid re­hydration aims to correct the child’s deficit over 4 h. Thus, a 10-kg child who is 10% dehydrated would receive one litre of Ringer’s lactate over 4 h. Hypernatraemic dehydration is a relative contraindication to rapid rehydration, although a bolus of Ringer’s lactate or normal saline is unlikely to drop the serum sodium fast enough to induce seizures. Rapid rehydration is highly successful in children with failure to thrive or moderate malnutrition, so it is only in severe malnutrition (marasmus or especially kwashiorkor) that there is a risk of heart failure.20 Intraosseus infusions can be life-saving in an emergency, but are not without risks (e.g. osteomyelitis).21 There is no evidence of a short-term benefit from the addition of bicarbonate to rehydration solutions,22-24 but this may not be true in Top End Aboriginal children who have extremely high rates of acidosis with acute gastroenteritis. Note that oral rehydration solutions have 10 mmol/L of citrate, and Ringer’s lactate contains 40 mmol/L of lactate as base. A rela­tive contraindication to the use of Ringer’s lactate is alkalosis (e.g. pyloric stenosis) when normal saline is more appropriate for rehydration. No study has compared Ringer’s lactate to normal saline for rehydration, but most paediatric recommendations prefer Ringer’s lactate, although the Advanced Paediatric Life Support and ICU guidelines tend to favour normal saline,25 and ESPGHAN favours 20 mL/kg of 0.8% saline for shock followed by 0.45% saline in 5% dextrose.16 There is a potential danger of rapid IV rehydration if maintenance solutions (Na+ 30−60 mmol/L) are used instead of rehydration solutions (Na+ 130−155 mmol/L), particularly in a severely dehydrated child in whom severe hyponatraemia with seizures and death could be induced. Many clinicians treating dehydration are aware of the recent change to rapid rehydration protocols but are not aware of the need to use Ringer’s lactate nor the danger of using maintenance solutions for rapid rehydration. One cannot help wondering whether some unexpected deaths in gastroenteritis might have been related to this misunderstanding. There is a continuing controversy about the use of colloid (e.g. albumin) versus crystalloid (e.g. Ringer’s lactate) solutions for volume replacement in critically ill patients. A systematic review26, 27 did not support the use of colloids for volume replacement and this issue is now the subject of an Australian multicentre trial in the adult intensive care unit (ICU) setting. Colloids have never been standard treatment for rehydration of children with acute gastroenteritis so are best avoided. In our Darwin cohort, IV rehydration (mostly rapid with Ringer’s lactate) was carried out in 56% of diarrhoeal admissions. Those requiring IV fluids had a mean degree of clinical dehydration of 5.9% (5.5−6.4%) and weight change dehydration of 4.5% (4.1−4.9%), whereas those who were orally rehydrated were assessed to be a mean of 3.6% (3.2−4.0%) and 3.0% (2.6−3.4%) dehydrated, respectively. However, in these Aboriginal children, hypokalaemia was documented in 69.5%, and acidosis (bicarbonate < 18 mmol/L) in 64.4%, with a mean blood pH of 7.31 (7.29−7.32) on presentation. Rapid IV rehydration is very safe in our hospital setting, and it is convenient to give IV fluids after venesection for laboratory tests, especially since ongoing osmotic diarrhoea from lactose intolerance is very common.28 But these complications and degrees of dehydration are uncommon in non-Aboriginal children with acute gastroenteritis, who can mostly be rehydrated with ORS even in hospital. There are some interesting novel therapies for diarrhoea under investigation. Racecadotril is an inhibitor of enkephalins (endogenous opioid peptides) which causes decreased intestinal hypersecretion. A Peruvian study showed a decreased duration and severity of diarrhoea in children in the racecadotril-treated group.29 Gum arabic is a soluble polysaccharide fibre with proabsorptive properties which affects intestinal nitric oxide and potassium channels, and may improve sodium absorption in diarrhoea.30, 31 Nitazoxanide is a new broad-spectrum anti­microbial agent which has been shown to be effective against giardiasis, amoebic dysentery and cryptosporidiosis.32, 33 Rifaximin is a nonabsorbable antibiotic which is effective in the treatment of small bowel bacterial overgrowth,34 which may complicate diarrhoea in malnourished children as well as those with short gut syndrome. Finally, probiotics such as Lacto­bacillus GG have been shown to shorten the course of diarrhoeal disease, particularly for rotavirus infection.35 This is currently the subject of a trial at Royal Darwin Hospital, since the benefits in partially breastfed Aboriginal children with mostly bacterial diarrhoea are uncertain. The accurate assessment of dehydration in childhood diarrhoeal disease remains a difficult challenge for clinicians. Experience in the developing world and with Top End Aboriginal children suggests that clinical assessment supplemented by plasma bicarbonate or venous blood gases are the most useful tools and that weight gain with rehydration is not a ‘gold standard’. Rapid IV rehydration with Ringer’s lactate over 4 h is safe and effective, but maintenance solutions must not be used for rapid IV rehydration. For moderate to severe dehydration, the combination of rapid IV rehydration and early feeding has reduced our use of ORS in Aboriginal children. But hypo-osmolar ORS remains the key therapy for the prevention and treatment of milder degrees of dehydration in acute gastroenteritis.