Aanesthetic Agents for Day Surgery - NIHR Health Technology ...
Aanesthetic Agents for Day Surgery - NIHR Health Technology ...
Aanesthetic Agents for Day Surgery - NIHR Health Technology ...
You also want an ePaper? Increase the reach of your titles
YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.
CESA models<br />
Differences in the use of prophylactic<br />
anti-emetics<br />
Approach<br />
Decision analysis was used to estimate the<br />
expected costs, rates of PONV and ICERs if<br />
prophylactic anti-emetics were added to the<br />
anaesthetic regimen used in the CESA RCT<br />
(adults), or different anaesthetic agents were<br />
used <strong>for</strong> children. The structures of the decision<br />
trees (models) were validated, ensuring that the<br />
branch representing practice in the CESA RCT<br />
replicated the results of the trial.<br />
Probabilistic sensitivity analysis was used to<br />
generate mean expected costs and outcomes<br />
and statistical measures of expected variance<br />
around the mean. This allows estimation of the<br />
probability and extent to which uncertainty and<br />
variation in the data used affect the absolute and<br />
relative costs and outcomes. 257 For this analysis<br />
each variable was assigned a base case or average<br />
value and a distribution of possible values. The<br />
probabilistic analysis summed the results of<br />
multiple analyses (iterations). Each iteration<br />
sampled the values <strong>for</strong> the variables at random<br />
from the specified distributions. The simulation<br />
package contained a component to determine<br />
the most appropriate distributional <strong>for</strong>m to<br />
model the results of the empirical study. In<br />
some cases the package did not identify the<br />
distributional <strong>for</strong>m that predicted the results<br />
of the empirical study. In these cases, a manual<br />
fit was obtained by identifying the distributional<br />
<strong>for</strong>m that best predicted the empirical data<br />
available from the trial. Examples of the distributional<br />
<strong>for</strong>ms that were explored are triangular,<br />
normal, truncated normal, uni<strong>for</strong>m and<br />
beta subjective.<br />
Patient anaesthesised <strong>for</strong><br />
day surgery<br />
No anti-emetic<br />
prophylaxis (CESA RCT)<br />
No anti-emetic<br />
prophylaxis (model)<br />
© Queen’s Printer and Controller of HMSO 2002. All rights reserved.<br />
<strong>Health</strong> <strong>Technology</strong> Assessment 2002; Vol. 6: No. 30<br />
The sampling method used was Latin Hypercube,<br />
expected value. The simulation software used<br />
was @RISK, 258 as an add on to Microsoft Office<br />
Excel v. 7.0 ® . Every simulation requires sufficient<br />
iterations to ensure that each variable is sampled<br />
over the full distribution of values specified and<br />
that the statistics generated are reliable. As the<br />
number of iterations increases, the distribution<br />
of the outcomes is described in more detail and<br />
becomes more stable. The amount of change in<br />
the percentile values, mean and SD decreases<br />
with each subsequent iteration. The number of<br />
iterations <strong>for</strong> each simulation was determined by<br />
the software, which halted the simulation when<br />
convergence at less than 1.5% in percentile values,<br />
mean and SD was achieved. The simulation<br />
analyses gave estimates of the probability that the<br />
expected ICERs calculated could occur. The<br />
decision tree <strong>for</strong> the model was structured to<br />
mirror the trial. The literature review indicated<br />
that the likely sequence and structure of events<br />
was the same with and without anti-emetic<br />
prophylaxis (i.e. practice in the CESA RCT).<br />
What was expected to differ was the incidence<br />
and intensity of PONV, and the subsequent<br />
impact on costs.<br />
The decision tree is presented in Figure 1. One<br />
branch emanating from the decision node of the<br />
tree represents the CESA RCT when a day patient<br />
was anaesthetised with one of the four anaesthetic<br />
agents (trial) and not given anti-emetic prophylaxis.<br />
The second branch of the decision model<br />
represents the scenario when a day patient was<br />
anaesthetised with one of the four anaesthetic<br />
agents and received a prophylactic 4 mg intravenous<br />
dose of ondansetron (model). Four<br />
decision trees, one <strong>for</strong> each anaesthetic<br />
regimen in the CESA RCT, were defined.<br />
PONV<br />
No PONV<br />
PONV<br />
No PONV<br />
FIGURE 1 The decision analytic model used to simulate the use of prophylactic anti-emetics<br />
Cost path 1<br />
Cost path 2<br />
Cost path 3<br />
Cost path 4<br />
43