Page 167 - 15Diarrhoeaandvomiting

Diarrhoea and vomiting caused by gastroenteritis in children under 5 years

Table A.1 Meta-analysis of rehydration and complication rates for ORT and IVT; data from
Hartling et al. 83
Outcome Rate
ORT IVT
Failure to rehydrate a 0.065 0.029
Hyponatremia 0.113 0.079
Hypernatremia 0.002 0.002
Paralytic ileus a 0.027 0.000
Phlebitisa 0.000 0.027
Peri-orbital oedema 0.026 0.026
Abdominal distension 0.017 0.000
Seizures 0.000 0.000
a Indicates statistical significance at the 5% level.

‘further IV’. The model assumes that all patients eventually rehydrate on IVT and therefore all
terminal nodes assume rehydration. This assumption essentially means that the two treatments
do not differ in their effectiveness and under such a scenario a cost-minimisation analysis may
be deemed appropriate. The implications of this will be addressed through a threshold sensitivity
analysis and in the discussion.
The decision analysis included only those complications reported in the Cochrane review where
there was a difference between the two treatments that was statistically significant at the 5% level.
As a result, the only complications modelled were phlebitis and paralytic ileus. It is important
to note, however, that statistical significance was not the sole justification for excluding other
reported complications. Of the five excluded non-significant complications reported in the meta-
analysis, three of them had no difference in mean rates (hypernatremia, peri-orbital oedema
and seizures). For hyponatremia although there was a difference in mean rates, the data using
a random effects model was highly consistent with a null hypothesis of no difference (95% CI
−0.13 to 0.15; P = 0.9). The meta-analysis may not have been powered to detect a difference but
given the sample size of the meta-analysis this suggests that any ‘real’ difference that may exist
is likely to be very small. It is most unlikely that research would be commissioned that would be
powered to detect this and decision makers are likely to have to live with this uncertainty in the
foreseeable future. Finally, it was thought that abdominal distension, although more borderline
in terms of statistical significance, was probably not of clinical significance.

Table A.2 lists the probabilities used in the analysis. The second column in the table shows the
probability values used in the base case analysis. In addition to the base case, a ‘worst case’
*
analysis for ORT relative to IVT was also undertaken. Column 3 gives the range of probabilities
used in both analyses.
The Cochrane review also compared ORT and IVT outcomes in terms of weight gain at discharge
and duration of diarrhoea. The differences between the two treatments for these outcomes were
not statistically significant at the 5% level and therefore these were not incorporated within the
economic model.
The review also found that ORT resulted in a statistically significant reduction in length of hospital
stay. This was not explicitly included in the model because the costing undertaken as part of this
analysis took into account the different resource implications, in terms of hospital stay, of the two
treatment options.

Costs
There is variation in how ORT and IVT are delivered, particularly with respect to staff input. In
the base case analysis, the aim was to cost up a ‘standard’ or ‘typical’ treatment for both ORT

* Derivations of the values used in the ‘worst case’ analysis are explained later in this appendix.

142