Jump to content

Property:Specifications on analyses

From CAMIH

This is a property of type Text.

Showing 20 pages using this property.
M
The data were analyzed by SPSS 20 software. The quantitative variables were expressed as mean (standard deviation) or median (interquartile range) and qualitative variables as number (%). To compare quantitative variables, repeated measures analysis of variance (ANOVA) was used and to compare qualitative variables, chi-square  +
The specific analysis method is not stated in the study, however, according to the authors, all included patients were evaluated at the end of the study.  +
The analysis was performed on an intent-to treat basis. An interim analysis was planned every 100 randomized patients to test the efficacy (primary efficacy endpoints) and the toxicity of the different arms according to the following “early stopping rules”: the arm(s) in which the efficacy values were significantly lower (p .05), by a t-test for changes, than in the other arms would be stopped. Likewise wise, the arm(s) in which grade 3 or 4 toxicity values were significantly higher by a t-test for changes, than in the other arms would be stopped. A very low level for the significance of p-values (p .001) was chosen considering that there are 10 possible pairs of between-arm comparisons and three endpoints, implying 30 possible candidate analyses; p-values are reported including Bonferroni’s corrections for multiple comparisons.  +
- Kaplan-Meier method to estimate survival probabilities and the Mantel-Haenszel log-rank test to compare groups - Cox proportional hazards regression to estimate crude and adjusted hazard ratios (HR) and 95% confidence intervals (CI)  +
All 56 patients were evaluated according to an intent-to-treat-analysis. 8 patients finished the study prematurely at their own request, but the comparibility of the study groups was still remained.  +
- Survival curves generated (Kaplan and Meier) - Log-rank test used to compare survival distributions between the two groups - Cox proportional hazards models used to calculate the RR (relative risk) of having an end point in the b-carotene group relative to those taking placebo, adjusted for other covariates  +
- Distributions of progression-free survival estimated by method of Kaplan and Meier - Effects of treatment/patient and disease characteristics analyzed using proportional hazards regression model of Cox - Magnitude of treatment differences was expressed as the relative risk, for busulfan alone relative to the busulfan plus vitamin A  +
- Analyzed: N = 77 for primary endpoint, 78 for secondary endpoints (n=5 patients did not receive any treatment, n=1 data set missing for primary endpoint) - 1% level of significance used because of the large number of treatment comparisons  +
Three interim analyses were planned: the first after 20 patients have completed Chemo-Radiotherapy to ensure toxicity in the selenium arm was not unacceptably high and the second and third after one third and two thirds of the patients had been followed for at least 18 months  +