Page 75 - 15Diarrhoeaandvomiting

Diarrhoea and vomiting caused by gastroenteritis in children under 5 years

(SD 14 months). Overall, 9.4% of children (18/192) were found to be hypoglycaemic but only
one child had serum glucose levels < 40 mg/dl (2.2 mmol/l). On comparing the characteristics of
the hypoglycaemic group of children (n = 18) with the non-hypoglycaemic group (n = 178), the
mean duration of vomiting (± SD) was found to be significantly longer in hypoglycaemic children
(3.3 ± 1.7 days versus 2.4 ± 2.6 days; P < 0.05). Of those children with hypoglycaemia and
dehydration, 94% had bicarbonate levels < 18 mEq/l and 19% had blood urea nitrogen (BUN)
levels > 18 mg/dl, while in the group of children having normal glucose levels and dehydration,
92% had bicarbonate levels < 18 mEq/l and 29% had BUN levels > 18 mg/dl. The difference
between the two groups was not statistically significant for these two parameters. [EL = 3]

The incidences of various biochemical disturbances as identified in the above five studies are
shown in Table 4.7. It is important to note that the investigators arbitrarily employed varying
definitions for biochemical abnormality, and the clinical importance of these disturbances should
be taken into account when considering the results from these studies.

Accuracy of laboratory tests in detecting dehydration
Two studies were included that evaluated the diagnostic accuracy of laboratory investigations for
assessing dehydration – a systematic review and a prospective cohort study. The methodology of
the systematic review and the results on the accuracy of signs and symptoms are described in
74
detail under Section 4.1. In this section, only the findings relevant to accuracy of laboratory tests
are given.
In the systematic review, six studies were identified that evaluated the ability of laboratory
74
tests to assess dehydration. Five studies evaluated BUN levels or BUN/serum creatinine ratio as
a test for dehydration but they used different thresholds to define an increased level. With a cut-
off value of 8, 18 and 27 mg/dl for a high BUN level, the + LRs ranged from 1.4 to 2.9, while a
single study found urea levels > 40 mg/dl to significantly increase the likelihood of at least 5%
dehydration (+LR 46; 95% CI 2.9 to 733). However, this study had a small sample population
and the confidence limits of the likelihood ratio were wide. Acidosis was evaluated in four
studies but these studies also used different cut-off values. Two studies defined acidosis as base
deficit > 7 mEq/l and they reported +LR of 1.4 and 1.8, and −LR of 0.4 and 0.7, respectively. The
other two studies used serum bicarbonate levels < 15 and < 17 mEq/l as indicative of acidosis.
Both the studies reported that bicarbonate levels lower than the cut-off values were not helpful
in increasing the likelihood of dehydration (+LR of 1.5 and 3.5, respectively), but higher levels
were found to be useful in decreasing the likelihood of dehydration (−LR of 0.18 and 0.22,
respectively). One study evaluated elevated serum uric acid levels (>600 mmol/l) and increased
anion gap (>20 mmol/l) as tests for dehydration but their likelihood ratios contained the null
value. [EL = 3]

81
The second diagnostic study from the USA evaluated the accuracy of urine specific gravity,
urine ketone levels and urine output in detecting dehydration. This study was part of a larger
study to compare the safety and efficacy of rapid IVT given over 1 hour with that of infusion
over 3 hours. The study population included 75 children aged 3–36 months admitted to the
emergency department with moderate dehydration (clinically estimated) and requiring IVT due
to failure of ORT (refusal, recurrent emesis or inadequate intake). After admission, urine samples
were collected by catheterisation or spontaneous void and, following successful rehydration
with IVT, repeat samples were collected. The reference standard for estimating the degree of
dehydration was the percentage weight loss calculated by dividing the difference between the
initial weight and final rehydrated weight with the rehydrated weight. Two-thirds of the children
(50/75) had ≥3% dehydration while 21% had ≥5% dehydration confirmed by the weight-based
criterion. No statistically significant correlation was found between urine specific gravity or urine
ketone levels with the degree of dehydration. For urine specific gravity, there was no statistically
significant increase in the likelihood of either 3% or 5% dehydration at any of the cut-off values
(with the 95% confidence limits containing the null value of 1). Similar results were seen for urine
ketone levels. Finally, urine output measured after admission and during rehydration therapy did
not correlate with the degree of dehydration, and it was not helpful in increasing or decreasing
the likelihood of dehydration. [EL = 3]

50