Page 67 - 15Diarrhoeaandvomiting

Diarrhoea and vomiting caused by gastroenteritis in children under 5 years

+LR 4.1; 95% CI 1.7 to 9.8), abnormal skin turgor (four studies, +LR 2.5; 95% CI 1.5 to 4.2)
and abnormal respiratory pattern (four studies, +LR 2.0; 95% CI 1.5 to 2.7). Sunken eyes and
dry mucous membranes showed a small increase in the likelihood of dehydration (+LR for both
1.7) and the lower limit of their 95% CI was close to the null value. Results for weak pulse as a
predictor for dehydration were variable, with one study showing it to be a fair predictor (+LR 3.1;
95% CI 1.8 to 5.4) while another did not (+LR 7.2; 95% CI 0.4 to 150). The presence of cool
extremities as a test for dehydration was examined in two studies and both reported imprecise
point estimates for the +LR (95% CI too wide to draw conclusions). The 95% CI for the positive
and negative LRs for increased heart rate, sunken fontanelle in young infants, and an overall poor
appearance included the null value.
A second prospective cohort study aimed to determine whether CRT measured using a digital
75
device (DCRT) could determine the presence of significant dehydration. The study population
comprised 83 children (aged 1 month to 5 years) with acute gastroenteritis admitted to an
accident and emergency department in Canada. Following admission and enrolment, the degree
of dehydration was estimated using a seven-point Likert scale, CRT was clinically assessed in the
conventional way by the paediatric medical staff, and DCRT measured using a small digital video
camera with customised graphics software. The reference standard (degree of dehydration) was
calculated by measuring the difference between the pre- and post-rehydration weight of the child.

Thirteen (16%) children met the WHO definition of dehydration (≥5%), with 12 estimated to have
a fluid deficit between 5% and 8% and one with 11% deficit. For these children, there was a
strong correlation between the child’s fluid deficit and the DCRT (Pearson’s correlation coefficient
0.75; P < 0.001). The AROC for detecting presence of dehydration ≥ 5% was 0.99 for DCRT and
0.88 for clinical assessment. DCRT showed the best result for predicting dehydration more than
5%, with 100% sensitivity, 91% specificity and a +LR of 11.4 (95% CI 5.4 to 22). Compared with
the clinical assessment scale, conventional CRT showed better results for specificity (88% versus
81%) and for +LR (4.5 versus 4.1), but poorer results for sensitivity (54% versus 77%). [EL = 2]

Clinical assessment of the severity of dehydration
Four guidelines had classified degrees of dehydration by using a combination of signs and
symptoms. These are summarised in Tables 4.2–4.4.

Evidence summary
Results from a systematic review [EL = III] suggest that prolonged capillary refill time, abnormal
skin turgor and abnormal respiratory pattern are the signs most useful to detect 5% or worse
dehydration in a child with gastroenteritis. Sunken eyes and dry mucous membrane were also
found to be useful although their predictive value was less than the above three signs. For the
other signs and symptoms, either the pooled likelihood ratios were statistically not significant or
there was wide variation in the results from individual studies. Results also show that there was
generally a poor agreement between clinicians on the presence of these clinical signs. Another
study [EL = II] showed that CRT measured using a digital video technique (DCRT) had better
accuracy in detecting dehydration of 5% or worse than the conventional clinical CRT and the
clinical assessment scale.
Although the published guidelines employed different methods of classifying the severity of
dehydration, they all used similar symptoms and signs (individually or in combination) for these
classification methods.

GDG translation from evidence to recommendations
Clinical detection of dehydration
A range of symptoms and signs have traditionally been considered useful in the detection of
dehydration. The GDG found that many of these did not have evidence with regard to their
reliability, particularly in those children with less severe degrees of dehydration.
The GDG considered that the identification of symptoms useful for the detection of dehydration
would be important, particularly because they could be employed as part of the remote
assessment process. However, the only symptom of possible value identified from the evidence
was a report of ‘normal urine output’ and the evidence between studies was inconsistent. The
GDG considered that enquiry should be made about this matter, and that some reassurance

42