Driver macro to compute characteristics and produce L1 baseline table. More...

Detailed Description

Driver macro to compute characteristics and produce L1 baseline table.

Initialize variables:
- Number of unique stockgroup parameters.
- Indicator that a covariate includes an AFDD requirement.
  - Used to trigger additional variable creation.
- List lab covariates with additional variables to create.
- Count number of covariates anchored to indexdt_exp.
- Assign type specific variables to keep on ADS dataset.
  - Check if any covariate anchored to INDEXDT_EXP.
- Count number of groups/analysis groups to process.
- List groups to create baseline table.
- Check if stockpiling differs by group and stockgroup.
- Create dataset with combo covariate rules.
  - Initialize rule and label for all covariate codes.
  - Reset rules for combo covars only.
Determine squaring conditions for sex, race, hispanic and restrict to applicable cohortgrp.
For type 6:
- Subset group by switch evaluation step.
  - If the switch is 1 or 2, subset it and reassign the index dates and agegroups.
Create sample and list of patient.
- List of patients and sort dataset to facilitate merges.
Compute baseline characteristics:
- Extract minimum and maximum index date values for data extraction censoring.
Compute utilization variables.
Extract claims for covariates and requeted risk scores.
Risk scores computation.
Prepare Analytic Dataset (ADS):
- Add dummy indexdt_exp variable (if necessary).
- Sort _base for subsequent merge.
- For type 5:
  - Restrict to index episode.
Evaluate covariates:
- Get list of dummy covariates (if any).
- For type 4:
  - Compute additional variables to correctly assess MinExtractDt and MaxExtractDt.
  - Check if covariates are anchored on EPISODEENDDT and if so take the maximum episodeenddt value.
  - For MI cohorts:
    - Consider the pre-pregnancy period and the maximum indexdt_exp.
    - Check if covariates are anchored on INDEXDT_EXP and if so take the maximum indexdt_exp value.
  - _CombineRecords stacks extracted DX/PX, dispensing, encounter and lab codes.
    - Consider ExpireDt in addition to ADate to process data correctly when we have presence of dateonly='N'.
  - Determine which covar# have MinRxDays > 1 and have an aFDD requirement.
Finding claims in covariatewindow and evaluate additional covariate parameters.
- If combo covariates are defined using at least a covariate anchored on indexdt_exp, consider it as such.
- Build dynamic where clause with requested covfromanchor/covtoanchor combinations:
  - covfromanchor = INDEXDT and covtoanchor = INDEXDT
  - covfromanchor = INDEXDT and covtoanchor = EPISODEENDDT
  - covfromanchor = EPISODEENDDT and covtoanchor = EPISODEENDDT
  - covfromanchor = INDEXDT_EXP and covtoanchor = INDEXDT_EXP
  - covfromanchor = INDEXDT_EXP and covtoanchor = INDEXDT
  - covfromanchor = INDEXDT and covtoanchor = INDEXDT_EXP
  - covfromanchor = INDEXDT_EXP and covtoanchor = EPISODEENDDT
Process covariates where MinRxDays = 1:
- If lab covariates are requested determine anchordate to keep the one closest to covtoanchor.
- Evaluate aFDD requirement:
  - Select covarnum with afdd requirement.
  - Remove index dates that do not meet afdd requirement for each covar.
Process covariates where MinRxDays > 1:
- If lab covariates are requested determine anchordate to keep the one closest to covtoanchor.
- Extract start and end date within COVFROM-COVTO window.
- Evaluate afdd requirement among covariates with a MINRXDAYS requirement.
  - Select covarnum with afdd requirement.
- Remove index dates that do not meet afdd requirement for each covar.
- Determine number of stockgroups:
  - If >1, then need to consolidate potential overlapping dispensings.
  - If =1, then there are no overlapping dispensings after stockpiling.
- Only keep claims on _Covar1 if minrxdays satisfied (total days in window >= minrxdays).
Process lab covariates:
- Check if additional lab variables should be created.
- For lab covariates, keep the closest to covtoanchor date.
  - If there is a tie, select random.
- Compute additional lab variables.
Count number of ADate per covariate and only keep covariate when freq >=codedays.
- Square.
- Create combo covariates.
- Surveillance mode:
  - Only compute covariates for new patients.
- Add covariate booleans to _base and dplocal master file.
  - Fill covars with 0s or missing if anchored on indexdt_exp and unexposed reference cohorts, pregnant cohorts, or non-pregnant cohorts.
Create final ADS dataset:
- One dataset period PERIODID.
- Add additional variables related to lab covariates.
- For type 2:
  - Adjust followuptime for episodes going beyond the PERIODID.
  - Add event variable.
- For type 4:
  - Add event variable.
- For type 6:
  - Add switchstep variable to differentiate switches.
- Assign correct variable names for correct merging and inferential/tree analysis.
- Carry forward information from previous look in ads_mstr.
Compute baseline table.
- If there are no patients in query, create empty baseline table.
  - Create baselinetable with patient, episode and group variables.
  - Rename group variable to either group/analysisgrp.

Program inputs

dplocal.&RUNID._&DATA. (Dataset containing one record per individual per index date for every cohort specified in the CIDA module run.)
infolder.&COHORTFILE. (Dataset used to define enrollment and demographic requirements, type of cohort identification strategy.)
infolder.&COVARIATECODES. (Dataset for request of covariates presence and analytic adjustment.)
infolder.&DRUGCLASSFILE. (Dataset required for the use of drug utilization metrics.)
infolder.&RISKSCOREFILE. (Dataset required for calculating one or more risk scores.)
infolder.&UTILFILE. (Dataset with medical or drug utilization metrics specification.)

Program outputs

dplocal.&RUNID._ads_&DATA. (Dataset containing one record per GROUP, PATID, and INDEXDT.)
msoc.&RUNID._baseline&OUTCOHORT._&I. (Dataset including metrics for cohorts of interest during baseline.)
msoc.&RUNID._baseline_&MFU._&I. (Dataset with MFU assessment (NDCs, diagnoses, and procedures before and/or after the index date).)
msoc.&RUNID._profile_&I. (Dataset containing patients and episodes by their unique combination of covariates observed.)

Usage

%ms_cidacov(DATA=mstr_mi,
            OUTCOHORT=_mi,
            GROUPVAR=group,
            GROUPDEFDATA=migroupnames,
            GROUPDEFVAR=cohortgrp, 
            COHORTGRPVAR=groupname,
            SUFFIXVARNUM=, 
            LOOKDT=indexdt_orig);

Parameters

[in]	DATA	Indicates input the patient-level dataset.
[in]	OUTCOHORT	Indicates the suffix for output baseline table if additional cohorts are created.
[in]	GROUPVAR	Name of variable in &data. that differentiates groups/analysis groups.
[in]	GROUPDEFDATA	Dataset including group/analysisgrp/milgrp definitions.
[in]	GROUPDEFVAR	Name of variable in &groupdefdata. that differentiates groups/analysis groups.
[in]	COHORTGRPVAR	Name of cohort group variable that will be used to assess squaring conditions.
[in]	SUFFIXVARNUM	Used in type 6 for switch step.
[in]	LOOKDT	Date variable used for creating individual PeriodId files (indexdt/episodeenddt).

SAS Macros Dependencies

Author: Sentinel Coordinating Center (info@.nosp@m.sent.nosp@m.inels.nosp@m.yste.nosp@m.m.org)

	QRP