There are always concern for calculating power and sample size in economic evaluation, whether the number of samples is representative enough and the result could be generalized to the population. Is there any standard method for sample size calculation in economic evaluation? And whether the method could be applied to any type of economic evaluation (e.g. cost-effectiveness, cost-utility, etc)?
Associate Professor |
Replied: 20 Mar 2018 at 23:37
With the increasing collection of economic data alongside conduct of clinical trial, there is now indeed a growing concern in the power and sample size in economic evaluation. Most often, clinical trials are underpowered to detect differences in cost.
The most common approach is based on the net benefit confidence limit equation to test the hypothesis about expected cost effectiveness ratio or net monetary benefit e.g. hypothesized that the incremental ratio of therapy XX compared with therapy YY will be lower than W per QALY1-4. This approach test the hypothesis by assessing whether net benefit, calculated by use of W, is significantly different from 0. Information usually required for the calculation include the variance of cost, variance of effectiveness, society’s willingness to pay for a benefit gain, difference in cost, difference in effectiveness and the correlation between the difference in cost and effect. More details can be found in the reference list below.
The methods are applicable to both CEA and CUA though the estimated sample size is applicable to the chosen effect. Hence, it is recommended to calculate based on the primary outcome that is most relevant in the study.
1. Briggs AH, Gray AM. Power and sample size calculations for stochastic cost-effectiveness analysis. Med Decis Making. 1998;18(2 Suppl):S81-92.
2. Laska EM, Meisner M, Siegel C. Power and Sample Size in Cost- Effectiveness Analysis. Med Decis Making. 1999;19(3):339-343.
3. Walter SD, Gafni A, Birch S. Estimation, power and sample size calculations for stochastic cost and effectiveness analysis. Pharmacoeconomics. 2007;25:455-466.
4. Glick HA, Doshi JA, Sonnad SS, Polsky D. Economic evaluation in clinical trials. Oxford: Oxford University Press; 2007.
Replied: 12 Apr 2018 at 15:39
Thank you Prof Asrul and Prof Alec for the answers.
However, it is difficult to calculate based on the net benefit confidence limit equation and clinicians/experts are more interested to do calculation based on the clinical primary outcome.
Thanks for all the references, I will look in more details.
Professor of Management Science |
United Kingdom |
Replied: 26 Mar 2018 at 03:21
That sounds like good advice to me. The only thing I would observe is that often it can be difficult to get trial designers to power their trials based on economic criteria, typically they like to power based on their clinical primary outcome. See Petrou and Gray (2011) - 2nd page, first column - on this issue.
1. Petrou S, Gray A. Economic evaluation alongside randomised controlled trials: design, conduct, analysis, and reporting. BMJ. 2011;342:d1548.