3.7 L2 Comparison File

The L2 Comparison File is optional. This file must be specified if creating an L2 (inferential analysis) report. This file defines the model parameters and the output to include in the report.

Table 3.13: L2COMPARISONFILE specification
Parameter Field Name Description
Group Name for Analysis ANALYSISGRP Name to differentiate exposure of interest/reference pairs.
Note 1: ANALYSISGRP must correspond to values of ANALYSISGRP included in distributed input files.
Note 2: all ANALYSISGRP values must be unique from all other ANALYSISGRP and GROUP values. The same ANALYSISGRP cannot be used in more than one run of the query request package.
Note 3: if any subgroup analyses are requested in QRP for an ANALYSISGRP, those subgroup analyses will automatically be output.
Format: SAS character $40; no special characters (e.g., commas, periods, hyphens, spaces, etc.) allowed, and underscores must be used to mark spaces.
Example: drug_a_and_drug_b
Group Display Order ORDER Unique number for each ANALYSISGRP. This is used to determine the order with which each ANALYSISGRP will be output for requested tables and/or figures. All ANALYSISGRPs should receive a unique ORDER value. ORDER values should start at 1 and be contiguous.
Format: Numeric
Example: 1
RunID Corresponding to ANALYSISGRP RUNID RunID assigned to the run that corresponds to the specified ANALYSISGRP.
Format: SAS character $5
Example: r01
Output PS Distribution Histograms OUTPUTPSDISTRIBUTION Indicator specifying whether to output histograms of PS distribution for the ANALYSISGRP specified between exposure and control cohorts, before and after adjustment, when applicable.
Note 1: to output figures and PS histograms for each ANALYSISGRP, both INCLUDEINREPORT (in the Figure Lookup File) and OUTPUTPSDISTRIBUTION must equal Y. If INCLUDEINREPORT = N and OUTPUTPSDISTRIBUTION = Y, a warning will be written to the log and no histograms will be output. A histogram will not be output for the specified ANALYSISGRP if INCLUDEINREPORT = Y and OUTPUTPSDISTRIBUTION = N.
Valid values are:
  • Y: Output PS histograms
  • N: Do not output PS histograms
Format: SAS character $1
Example: Y
Output Conditional Analysis OUTPUTCONDITIONAL Indicator specifying whether to output conditional analysis for the ANALYSISGRP specified.
Valid values are:
  • Y: Output conditional analysis in report
  • N: Do not output conditional analysis
Note 1: specification of OUTPUTCONDITIONAL is relevant only to PS fixed ratio matching analyses (conditional analysis will be automatically output for a PS variable ratio matching, PS stratification, or covariate stratification report).
Format: SAS character $1
Example: Y
Output Unconditional Analysis OUTPUTUNCONDITIONAL Indicator specifying whether to output unconditional analysis in report for the ANALYSISGRP specified.
Valid values are:
  • Y: Output unconditional analysis
  • N: Do not output unconditional analysis
Note 1: specification of OUTPUTUNCONDITIONAL is relevant only to PS fixed ratio matching analyses
Format: SAS character $1
Example: Y
Output Forest Plot OUTPUTFORESTPLOT Indicator specifying whether to output forest plots in report for the ANALYSISGRP specified.
Valid values are:
  • Y: Output forest plots
  • N: Do not output forest plots
Note 1: if requesting forest plot(s), Figure F2 must be included in the Type 2 or Type 4 Figure File.
Format: SAS character $1
Example: Y
Convergence Status to Consider CONVRULE Enter the convergence status values for considering whether the model has converged.
Valid values are:
  • 0: Convergence satisfied
  • 0: Missing initial values for some parameters
  • 1: Quasi-complete separation
  • 1: Complete separation
  • 2: Ridging has failed to improve the likelihood
  • 2: Iteration limit reached without convergence
  • 2 or 3: Missing initial values for some of the parameters
  • 2 or 3: Line search has failed to improve the likelihood
  • 2 or 3: Improper initial values for the intercept estimates
  • 3: Improper initial values for estimating the model parameters
Note 1: specify multiple values via a comma-separated or a space-separated list
Note 2: valid values based on variable STATUS from dataset [runid]_estimates_[periodid].sas7bdat
Note 3: only DPs meeting convergence status requirement will be considered for effect estimates and Kaplan-Meier plots, if requested.
Note 4: if FIRTH=Y in either PS Estimation file or PS Subgroup file, then the resulting ESTIMATE value in the MSOC Estimates file will have value of 2. This is because correctly implementing FLIC will result in warnings in the SAS log indicating that convergence was not achieved, particularly “WARNING: Convergence was not attained in 0 iterations.” Thus, to output estimates for models where FIRTH=Y, CONVRULE should be set to include 2.
Format: SAS character - length can vary
Example: 0,1
Number of HDPS variables to select from each DP TOPNHDPS Enter the number of high-dimensional propensity score (HDPS) variables to select from each Data Partner.
Format: Numeric
Example: 10
Kaplan-Meier Reference Population KMREFPOP Indicator specifying whether to include weighted comparison group, unweighted real population, or both in the Kaplan-Meier plot.
Valid values are:
  • Weighted: Include weighted comparison group
  • Unweighted: Include unweighted real population
  • Both: Include both
Note 1: only applicable for variable-ratio matched analyses and conditional plots.
Note 2: if individual level data is not returned (INDLEVEL = N in QRP Parameters file) then KMREFPOP must be set to ‘unweighted’.
Note 3: 95% confidence intervals (CI) will automatically be output on the Kaplan-Meier plot for all unadjusted and unconditional analyses, and for conditional plots where KMREFPOP is set to ‘unweighted’. If KMREFPOP is set to ‘weighted’ or ‘both’ for conditional plots, 95% CI will NOT be output.
Note 4: KMREFPOP should be left blank if conducting an IPTW analysis.
Format: SAS character $10
Example: Weighted