Binomial Distribution Sample Confidence Intervals Estimation 8. Number Needed to Treat/Harm
|
 |
|
Post a Comment
|
|
|
|
|
|
ABSTRACT:
Nowadays, the number needed to treat became the most important parameter in reporting the treatment effects in clinical trials, from binary outcomes such as ˇ°positiveˇ± or ˇ°negativeˇ±. Defined as a reciprocal of the absolute risk reduction, the number needed to treat is the number of patients who need to be treated to prevent one additional adverse even. In medical literature, the number needed to treat is reported usually with its asymptotic confidence intervals, method that is used by the most software packages even if it is knows that is not the best method. The aim of this paper is to introduce three new methods of computing confidence intervals for number needed to treat/harm.
Using PHP programming language was implementing the proposed methods and the asymptotic one (called here IADWald). The performance of each method, for different sample sizes (m, n) and different values of binomial variables (X, Y) were asses using a set of criterions: the upper and lower boundaries; the average and standard deviation of the experimental errors; the deviation of the experimental errors relative to imposed significance level (¦Á = 5%). The methods were assess on random binomial variables X, Y (where X < m, Y < n) and random sample sizes m, n (4 ˇÜ m, n ˇÜ 1000).
The performances of the implemented methods of computing confidence intervals for number needed to treat/harm are present in order to be taking into consideration when a confidence interval for number needed to treat is used.
|
|
|
|
STATISTICS
|
|
Click on # to view
|
|
Citations
|
|
3
|
|
References
|
|
2
|
|
Comments
|
|
0
|
|
Quality
|
|
0/0.00
|
|
Interest
|
|
0/0.00
|
|
View(er)s
|
|
2/446
|
|
|
|
|
|
|
| Prev |
Next |
|