Page 60 - 15Diarrhoeaandvomiting

Diagnosis

Clinical question
How accurate are laboratory blood tests in distinguishing bacterial from viral gastroenteritis?

There is variation in practice regarding the use of laboratory blood tests in distinguishing between
bacterial and viral causes of gastroenteritis, hence their accuracy in detecting these causes was
sought.

Evidence overview

Four diagnostic studies were included in this section, one with EL = 2 and the rest with EL = 3.
In the first three studies, the accuracy of acute-phase proteins was evaluated for detecting
bacterial gastroenteritis, and C-reactive protein (CRP) was assessed in all the studies, while
ESR, interleukin-6 (IL-6) and interleukin-8 (IL-8) were assessed in one study each. The last study
evaluated the diagnostic ability of total and differential blood count in differentiating bacterial
from viral causes of gastroenteritis.
The first study, from Italy, looked at the diagnostic accuracy of CRP and ESR measurements in
65
the differentiation of bacterial and viral gastroenteritis. Over a 4 year period, it recruited 111
children aged between 1 and 60 months admitted to a hospital with acute diarrhoea lasting
more than 12 hours but less than 15 days . Children with chronic gastrointestinal diseases such
as cow’s milk protein intolerance, Crohn’s disease, gastro-oesophageal reflux or chronic diseases
were excluded. After admission, all children had blood taken for the measurement of CRP and
ESR levels, while stool culture was performed to detect bacterial aetiology and viruses detected
by ELISA testing on the stool specimens. The accuracy of CRP in detecting bacterial or viral
gastroenteritis was calculated at the cut-off values of 12, 20 and 35 mg/l, while elevated ESR
was taken as value ≥ 25 mm/hour. Of the 111 children, 53 (48%) were diagnosed with bacterial
gastroenteritis (mainly non-typhoidal salmonella), 35 (32%) had viral gastroenteritis and the
remaining 21% had culture-negative infections. The mean CRP level in children with bacterial
infections was significantly higher than in those with viral infections (P < 0.001) and culture-
negative infections (P < 0.01). CRP levels were strongly associated with bacterial infections at all
three cut-offs: 12 mg/dl (OR 25.8; 95% CI 7.6 to 87.9), 20 mg/l (OR 46.4; 95% CI 5.9 to 365) and
35 mg/l (OR 27; 95% CI 3.4 to 212). The specificity of CRP in detecting bacterial gastroenteritis
was high at all the cut-off levels (89% at 12 mg/l and 97% at both 20 mg/l and 35 mg/l) but
the highest sensitivity was 77% at 12 mg/dl, compared with 58% and 44% at the other two
cut-off values, respectively. The area under receiver operating characteristic curve (AROC) at
12 mg/l was 0.83. Raised ESR levels (≥25 mm/hour) were also strongly associated with bacterial
infections (OR 3.5; 95% CI 1.2 to 9.9) and showed a sensitivity of 42%, specificity of 83% and
AROC of 0.62 for detecting them. Raised total leucocyte count did not show any statistically
significant association with any of the three infections. [EL = 2]

The second study, from Taiwan/China, aimed to determine whether IL-6, IL-8 and CRP were
66
useful diagnostic markers in differentiating bacterial from viral gastroenteritis. The study included
56 children (mean age 2.5 years) admitted with acute gastroenteritis, of whom 21 had rotavirus
(by Rotaclone® test), 18 had bacterial infections (by stool culture with salmonella species
isolated predominantly) while 17 children were recruited as controls. Children with chronic
disease or history of persistent/intractable diarrhoea were excluded. No details were provided
about the control group or exclusion criteria. The concentration of both CRP and IL-6 were
significantly higher in children with bacterial gastroenteritis than in those with viral infections
(P < 0.001) and the control group (P < 0.001). IL-8 concentrations were elevated in both bacterial
and viral infections and there was no statistically significant difference in the levels between the
two groups. Diagnostic accuracy results were analysed using ROC curves and it showed best
results for CRP, with the AROC being 0.90 at the cut-off value of 2 mg/dl, followed by IL-6 with
an AROC of 0.83 at the cut-off value of 10 pg/ml. At these cut-off values, the sensitivity and
specificity of CRP in detecting bacterial gastroenteritis were 83% and 76%, respectively, while
those of IL-6 were 78% and 86%, respectively. IL-8 was found to be of less diagnostic value, with
an AROC of 0.68, sensitivity of 50% and specificity of 67% at the cut-off value 70 pg/ml. [EL = 3]
67
In the third study, from Israel, the ability of the quick-read CRP (QR-CRP) test to detect bacterial
gastroenteritis was determined in a convenience sample of 44 children (age range 4 days to
17 years, median age of 2.4 years) admitted to the emergency department of a tertiary hospital. All

35