FSTG01 [spec added]

[barnett11]

Table Name

FSTG01

Description

Forest Plot for Odds Ratio

Analysis dataset

ADRS

Analysis filtering

Data is expected to be filtered by the user (so not in template)
eg. Overall Survival [ADRS.PARAMCD]="BOR"

Analysis population

IT

Column Variable

ARM

Analysis variables

ADRS.AVALC

Default Baseline Risk Factors, but will be expected to be updated by user
SEX, AGEGR1, RACE

Analysis concept

Statistics

The estimate of the odds ratio (OR) or hazard ratio (HR) should be represented by a green square (or other user-defined symbol). For OR, the size of the symbol is proportional to the total sample size for that subgroup. For HR, the size of the symbol is proportional to the total number of events for that subgroup. Provide option to turn off proportional sizing and use single-sized symbol for all subgroups. Show a solid vertical reference line at 1 and a dashed vertical reference line at the overall treatment effect. Allow α level for CIs to be adjusted, default is 2-sided 5%. Plot OR and HR on logarithmic scale so that left and right tails for comparable ratios have same lengths. Include an “All Patients” row on each page of output.

Allow for either dichotomous outcomes or time-to-event outcomes. Dichotomous outcomes require logistic regression analysis with odds ratios as summary statistics. Time-to-event outcomes require Cox proportional hazards regression analysis with hazard ratios as summary statistics.

For dichotomous outcomes use Proc LOGISTIC to compute treatment odds ratios and corresponding 95% confidence intervals from Wald test for each covariate subgroup. Analysis should be consistent with specifications for efficacy table LGRT02. By default present all summary statistics shown in mock-up FSTG01. Allow the option for presenting the interaction p-values alongside the odds ratio and 95% CI (refer to LGRT02 for further details). Note: The decision to include interaction p-values should be considered carefully given their associated low power. Provide an option to not present the response rates columns in the table and increase size of the plot area accordingly.
For time-to-event outcomes use Proc PHREG with discrete tie handling to compute treatment hazard ratios and corresponding 95% confidence intervals for each covariate subgroup. Analysis should be consistent with specifications for efficacy table COXT01. By default present all summary statistics shown in mock-up FSTG02. Allow the option for presenting the interaction p-values alongside the hazard ratio and 95% CI (refer to COXT01 for further details). Note: The decision to include interaction p-values should be considered carefully given their associated low power. Use Proc LIFETEST to compute median survival times for each combination of treatment group and covariate subgroup. Provide an option not to present the median survival times in the table and increase size of the plot area accordingly.

Default treatment of zero count rows

Keep zero count rows by default

Additional comments

No response