Dear Stata users,
I conducted a cost-effectiveness analysis with a sample of 200 participtans and for missing values I used simple regression imputation. Seemingly unrelated regression equations models were bootstrapped (5000 replications) to allow for correlated residuals of the cost and effect equations and bootstrapping was used to obtain 95% confidence intervals for the ICERs. Herefore I used the ado-file bscicer.ado in Stata version 13.
Now my question:
The formula for the ICER is the following: ICER= incremental costs : incremental effects
I calculated (bootstrapped) incremental costs and incremental effectes as well as ICERs with the mentioned ado-file. Here, the bootstrapped ICER is not the exact result of "bootstrapped incremental costs : bootstrapped incremental effects".
In my example bootstrapped costs are 5 (95% CI: -850 - 900) and bootstrapped effects are 0.01 (95% CI -89000- 105000). In Stata I got the following result for the bootstrapped ICER: 3030 (instead of 5 : 0.01= 500). A reviewer asked, why the ICER is 3030 (instead of 500).
Does anyone of you know, what leads to this „different“ ICER or how this value is calculated? Does Stata consider any errors or the big confidence interval when calculating the mean bootstrapped ICER?
Thank you in advance!
Sarah
I conducted a cost-effectiveness analysis with a sample of 200 participtans and for missing values I used simple regression imputation. Seemingly unrelated regression equations models were bootstrapped (5000 replications) to allow for correlated residuals of the cost and effect equations and bootstrapping was used to obtain 95% confidence intervals for the ICERs. Herefore I used the ado-file bscicer.ado in Stata version 13.
Now my question:
The formula for the ICER is the following: ICER= incremental costs : incremental effects
I calculated (bootstrapped) incremental costs and incremental effectes as well as ICERs with the mentioned ado-file. Here, the bootstrapped ICER is not the exact result of "bootstrapped incremental costs : bootstrapped incremental effects".
In my example bootstrapped costs are 5 (95% CI: -850 - 900) and bootstrapped effects are 0.01 (95% CI -89000- 105000). In Stata I got the following result for the bootstrapped ICER: 3030 (instead of 5 : 0.01= 500). A reviewer asked, why the ICER is 3030 (instead of 500).
Does anyone of you know, what leads to this „different“ ICER or how this value is calculated? Does Stata consider any errors or the big confidence interval when calculating the mean bootstrapped ICER?
Thank you in advance!
Sarah
