Analysis results metadata tables may be generated for submissions to regulatory agencies as part of the Define XML. The examples here illustrate information that could be produced to describe primary and secondary endpoint outputs. They are presented as examples of the types of analyses that may be produced to explore the endpoints presented in the preceding section. These are not currently required by most regulatory agencies, although they are considered a "nice-to-have" feature. For more details on ARM, see the ADaMIG and the Analysis Results Metadata (ARM) v1.0 for Define-XML v2.0 (https://www.cdisc.org/standards/foundational/adam).

Table 1. A simple comparison of change by treatment groups over 1 year.

Display

Table 14.xx.xx Change in Left Ventricular Ejection Fraction Percent Over 1 Year

Analysis ResultComparison of Change of Ejection Fraction Percent for the Treatment Groups (1 Year)
Analysis Variables(s)CHG
Analysis ReasonPrimary efficacy endpoint as specified in the SAP
Analysis PurposePrimary outcome measure

Data References

(incl. selection criteria)

PARAMCD = "LVEF_C"

Where ITTFL = "Y" and AVISITN = 6;

Documentation

Comparison of means with 95% confidence interval

Programming Statements 

(Add programming language statements here: SAS, R, etc.)

Proc means DATA=ADEFNTP N MEAN clm STDERR;

/* Subset the data. Can run again selecting for paramcd="RVEF_C"*/

Where PARAMCD = "LVEF_C" and ITTFL = "Y" and AVISITN = 6;

Class TRT01P;

Var CHG; 

run;


Table 2. This table demonstrates a model over multiple time points.

Display

Table 14.xx.xx Percent Change in Left Ventricular Ejection Fraction Percent Over Time (Yearly)

Analysis ResultComparison of Percent Change of Ejection Fraction for the Treatment Groups Over Time (Yearly)
Analysis Variables(s)CHG
Analysis ReasonPrimary efficacy endpoint as specified in the SAP
Analysis PurposePrimary outcome measure

Data References

(incl. selection criteria)

PARAMCD = "LVEF_C"

Where ITTFL = "Y"

Documentation

The mixed model using lsmeans to compare treatment groups

Programming Statements 

(Add programming language statements here: SAS, R, etc.)

/* Note that CHG is only computed post-baseline, so subset for the yearly visits, i.e. 6, 12, 18, etc. */

Proc mixed DATA=ADEFNTP;

Class TRT01P AVISITN;

/* Subset by param, as there are 2 parameters with CHG calculated*/

Where ITTFL = "Y" and AVISITN IN (6, 12, 18, 24, 36, 48) and PARAMCD = "LVEF_C"; 

Model PCHG=AVISITN*TRT01P/Solution;

RANDOM INTERCEPT / SUBJECT=STUYDID TYPE=UN;

LSMEANS TRT01P*AVISITN/ CL PDIFF;

run;


Table 3. This table demonstrates a model over multiple time points.

Display

Table 14.xx.xx Percent change in NT-proBNP over time (1 year)


Comparison of Percent Change of NT-proBMP for the Treatment Groups Over Time (Yearly)
Analysis Variables(s)CHG
Analysis ReasonPrimary efficacy endpoint as specified in the SAP
Analysis PurposePrimary outcome measure

Data References

(incl. selection criteria)

PARAMCD = "BNPPRONT"

Where ITTFL = "Y"

Documentation

The mixed model using lsmeans to compare treatment groups

Programming Statements 

(Add programming language statements here: SAS, R, etc.)

/* Note that CHG is only computed post-baseline, so subset for the yearly visits, i.e. 6, 12, 18, etc. */

Proc mixed DATA=ADEFNTP;

Class TRT01P AVISITN;

/* Subset by param, as there are 2 parameters with CHG calculated*/

Where ITTFL = "Y" and AVISITN IN (6, 12, 18, 24, 36, 48) and PARAMCD = "BNPPRONT"; 

Model PCHG=AVISITN*TRT01P/Solution;

RANDOM INTERCEPT / SUBJECT=STUYDID TYPE=UN;

LSMEANS TRT01P*AVISITN/ CL PDIFF;

run;


Table 4. This example uses the dataset ADEFMRI, with AGE, SEX, RACE, and BNPPRONT as covariates.

Display

Table 14.xx.xx Percent change in NTproBNP over 1 Year

Analysis ResultComparison of Percent Change of NTproBNP for the Treatment Groups Over a period of Time.
Analysis Variables(s)PCHG
Analysis ReasonPrimary efficacy endpoint as specified in the SAP
Analysis PurposePrimary outcome measure

Data References

(incl. selection criteria)

PARAMCD = "L_VEFC";

ITTFL = "Y" and AVISITN GE 6;

Documentation

Multiple regression model


Proc reg data=ADEFMRI;

Where PARAMCD = "L_VEFC" and ITTFL = "Y" and AVISITN GE 6;

By TRT01P;

Model PCHG = AAGE SEX RACE BNPPRONT;

run;