This macro processes the program input files variables. More...

Detailed Description

This macro processes the program input files variables.

Change infolder access to read.
Initialize helper variables to determine study left censor date.
Initialize abort check variables.
RiskscoreCodes and RiskScoreFile checks:
- When requesting generalized riskscores, both RiskscoreCodes and RiskScoreFile input files must be provided.
- Make sure the following do not occur:
  - Some risk score in RISKSCOREFILE are not represented in the RISKSCORECODES file.
  - RISKFROMANCHOR or RISKTOANCHOR are blank and analysis type=4.
  - RISKFROMANCHOR or RISKTOANCHOR are not blank and analysis type is not 4.
  - RISKFROMANCHOR = EPISODEENDDT and RISKTOANCHOR=INDEXDT.
- Check the following in RiskScoreCodes:
  - Make sure weight exists for every entry.
  - Make sure code is populated if condid is not IN (intercept).
  - Make sure codecat is a valid value.
  - If specified, make sure sex values are valid.
  - If specified, make sure age group values are valid.
Determine length of code/CaresettingPrincipal in inputfiles.
Determine length of DX, PX, and NDC code.
Diagnosis, Procedure, and Dispensing codes:
- Assign type.
- Cohortcodes:
  - Determine if pregnancy outcome codes will be evaluated from MIL table.
  - Strength and unit can only be set for RX codes and must be in list of valid values.
  - Strength and unit should both be populated if one is populated.
  - When strength and unit are populated check the sameday value for dose calcuations.
  - If T4HOIMETHOD=binary and if indexcriteria=FUT for any code, then abort.
  - Allow pregnancy outcome codes to be separated by a space. Create obs with single code.
  - Determine max length of mois to be used throughout QRP.
    - If MIL and ("LIVE" or "MIX") outcomes are requested in the same QRP run, then abort.
  - Ensure UNIT has the same value within group/stockgroup for type 2 and within group for type 5.
  - If cohortcodes contains cause of death defining codes (CODECAT=CD), CODETYPE must be set to 10.
- InclusionCodes:
  - Do not set study start date if any condfrom exclusion window is '.'.
  - Strength and unit can only be set for RX codes.
  - Strength and unit should both be populated if one is populated.
  - When strength and unit are populated check the sameday value for dose calcuations.
  - If mincumdose is populated, then both strength and unit must be populated AND condfrom/condto cannot be missing.
  - If there is an aFDD or cFDD requirement, cannot use CODESUPPLY, strength/unit must be populated.
  - Ensure MinRXDays has same value within subcondlevel and only specified for codecat=RX.
  - Ensure UNIT has same value within subcondlevel.
  - Ensure mincfdd/maxcfdd has same value within subcondlevel.
  - Set MinRXDays to 1 where the above tests failed and continue processing.
  - Ensure only 1 codedays value per group condlevel subcondlevel.
  - Create numeric Condlevel and SubCondlevel variables for looping.
- Append inclusion and exclusion codes to work.&COHORTCODES.
If for Type 4 all 3 pregnancy lookup files aren't specified, then abort.
Pregnancy Outcomes lookup file:
- If the same code is used for multiple outcomes in PREGNANCYCODES, then abort.
- If there are duplicate codes found, then abort.
Type 4: Pregnancy Duration codes lookup file:
- Read in pregnancy duration file, combine mpriority and cpriority variabless.
- If duplicate codes found, then abort.
Type 4: Pregnancy Metadata File:
- Read in pregnancy metadata lookup file which contains pregnancy outcomes and incidence day values for applying washout.
  - As live birth outcomes can only be selected using (LIVE and MIX) or MIL, remove record and corresponding incidence day value of unused outcome if both are present in the file.
- Sort by preg_outcomenum defensively as preg_outcomenum will be re-sequenced depending upon selected source of live birth records.
- If MIL table is specified, check for ("LIVE" or "MIX") outcome and mark position of value within incidencedays.
- If MIL table is not specified, check for a MIL outcome and mark position of value within incidencedays.
- Loop through _incidencedays values and remove unused values were applicable.
- Confirm incidencedays contains same number of incidence values as outcomes.
- If outcomes not defined in both PREGNANCYCODES and PREGNANCYMETA, then abort.
- If an outcome selected in COHORTCODES is not in the PREGNANCYMETA, then abort.
Type 4: Medication of interest/drug use codes and check for ENRDAYS.
Mother-Infant Linkage Analysis:
- Create migroupnames dataset for use in t4cida summary table creation and PS/COV analysis.
  - Make sure expmp variable is not missing.
- If T4HOIMETHOD = timetoevent, outcomepop can only be M, otherwise abort.
Monitoring File:
- Limiting to periodid set by user.
- Determine if query period is data driven (in the case of fixed risk window sequential analyses).
- If query period is data driven, then monitoringfile processing will occur in ms_setdatadrivenqueryperiod.
- If query period is not data driven and fixed across all scenarios:
  - Determine end of query period and assign last date of follow-up if censoring on query or DP max date.
    - Add QA DP Max Date to file.
    - Turn off CDPEND if using SURVEILLANCEMODE.
    - Assign index assessment period end date.
      - FUPDRIVEN - query period ends at the earliest of FUPENDATE/CDPEND.
      - Check to see if dp_maxdate is prior to startdate.
    - Assign censordate.
  - Take minimum StartDate and maximum EndDate.
  - Create non-overlapping looks dates.
  - Assign calendar based macro variables.
- Data driven: select start of surveillance.
Surveillance mode processing:
- Prior data existence for surveillance.
Preprare Raw extraction files:
- Adapt for different lengths to avoid warnings later.
- Compute CondIdNum to handle gaps in CondId.
- Select covariates codes (not combination of covariates).
- To prevent warning when covariatecodes contains >11 length for CC codes, reset length of code.
- Checking for the presence of optional lab variables.
- Strength and unit can only be set for RX codes.
- Strength and unit should both be populated if one is populated.
- When strength and unit are populated check the sameday value for dose calcuations.
- If there is aFDD requirement, cannot use CODESUPPLY, strength/unit must be populated.
- Do not set study start date if any covfrom window is '.'.
- Codedays variable should be specified at the covariate level.
  - Run a check to ensure value is unique within covariate.
- Covfrom and covto should be the same value for RX codes within a stockgroup.
- Ensure MinRXDays has same value within covarnum.
- Ensure MinRXDays only specified for codecat = RX.
- Ensure UNIT has same value within covarnum.
- Ensure lab code types are only numeric/character per covariate.
- Check that codepop is consistently populated within each covariate for type 4.
  - Ensure codepop has same value within covariate.
- Set MinRXDays to 1 where the above tests failed and continue processing.
- List of covarnum-studyname for labels.
  - Add dummy covariates if covarnum are not sequential.
- Count Covars and ComboCovars.
- Confirm there are not any missing covarnums.
  - Covarnums must be consecutive without gaps.
- Confirm valid covarnums for Combo Covars.
- If there are combo covars that do not meet specification, then abort.
- Check consistency with PROFILE parameter.
- Check if some covariate windows are anchored on INDEXDT_EXP and if there are unexposed reference cohorts.
- If COVARIATE windows are anchored to INDEXDT_EXP, then abort.
- Combine COHORTCODES, RiskScoreCodes and CovarCodes codes to pull data from CDM only once.
  - DH codes always pulled in ms_cidanum, so a _dth dataset is not created.
- Check for the presence of optional lab variables.
- Rename some variables if the combo file is used.
- List of encounters to extract in ms_cidanum.
Creating the number of groups to create cohorts:
- To prevent unecessary looping, separate count of groups with different age stratification.
- Determine distinct number of enrollment/demographic/death cohorts.
- Process basecohort for type 2:
  - Make sure same basecohort are grouped and assess order to process them correctly.
  - Check whether variables are the same across same basecohort.
Stratifications:
- ZipFile.
- USERSTRATA.
  - Add year when month is requested.
  - Add year when quarter is requested.
  - Add agegroupnum to strata that include age group.
  - Sort levelvars applicable to denominators.
  - If MOINAME specified, check if cohortcodes has at least one MP.
  - Check if MICOHORTFILE exists when requesting t4cida table.
  - Determine if age group and/or geog vars are output.
  - If geographic strata requested, zipfile must also be specified.
  - Determine if censor table should be produced.
  - Type 4: determine if control cohort should be produced.
ITSFile:
- Compute age group denominators.
- Create lookup file for its levels to add potential missing strata and prevent duplicate computation by ms_cidadenom.
- Separate t2cida levels from t2its. If the same levelvars is requested for both with different levelid, only keep t2cida for efficiency and assign mapping in lookup file.
- Get all required levels.
- Keep unique levelvars per table type (prevalent or not).
Never Exposed Cohort:
- Utilization File and RiskScoreFile.
- To prevent unecessary looping for prevalent data, separate count of groups with different age stratification.
Secondary Episode Analysis:
- Determine the highest adherenceid value across all analysisgrps.
- List of primary episode groups and analysisgrps when prev cohort is requested.
- For type 5:
  - Confirm dose categories are defined when t5dose or t5disp with cfdd_output_cat requested on userstrata.
  - Confirm the same number of categories are requested per group.
  - Check if cfdd is requested for at least one group.
- Must meet both strength/unit and levelvars conditions.
Concomitant Episode Analysis.
Switchinclusion file:
- Check to see if the age stratas are the same within analysis group values.
Check if DRUGCLASSFILE is specified when MFUFILE is also specified and contains RX codes.
Determine if PS or Covariate Stratification analysis to be performed.
HDPS Integrity Check:
- Get groups that are not compared with a predefined model.
- Create dataset to id which group for HDPS.
- Check if the same group is there more than once (then it is associated with more than one HDPS window).
- If HDPS window is anchored to INDEXDT_EXP, reference cohort must be exposed.
Tree Lookup File:
- If T4HOIMETHOD timetoevent and tree analysis both specified, then abort.
- Check if groups specified do not exist in treefile.
- Check if denominator is missing for PSMATCH/T3.
- Check if weight is specified for PSSTRAT/IPTW.
- Check if denominator is specified for PSSTRAT/IPTW.
- Check if groups meet outcome anchoring criteria.
- Check if poisson dataset is requested in USERSTRATA.
- Outcome evaluation for infants (CODEPOP=I/MI) must begin on or after delivery date for type 4.
- Identify groups for ps_match in treefile and check denominator values.
- Poisson analysis:
  - Identify groups for ps_stratification in treefile and check denominator values.
- PSMATCHFILE, STRATIFICATIONFILE or IPTWFILE required for SI.
- Check if required conditions are filled if the query is not DATADRIVEN.
- For type 4:
  - Check if denominator=person is specified in treefile for PSSTRAT/IPTW SI analyses and verify if ReqDaysAftPreg is sufficient.
Determine study start date for claim extraction - left censor date.
- Take the maximum value of all possible lookback parameters to substract from &STARTDATE.
- If any parameter has an "any" lookback, do not set left censor date.
- Check renaming parameters for "any" lookback if &NOSTUDYSTARTDATE.=N, then take minimum value.
- Exclusion and Covariate lookbacks have already been checked.
- When cohort includes MIL deliveries then linkagestatus cannot be blank.
- Look for max lookback exclusion and covariate codes.
- Compute studystartdate:
  - For types 1, 2, 3, 5, 6: determine maximum lookback date.
    - For type 6: select minimum product dates to product date is after studystartdate.
  - For type 4: query period binds the delivery date, however need to consider pregnancy duration.
    - If all condfromanchor values are set to episodeenddt, no need to add _maxduration.
Stockpilingfile:
- Determine which dataset to use and set group/analysis group variable.
- Expand stockpiling groupings to reference and exposure groups.
- In the case of stockgroups only being in covariatecodes file, need to create group/analysisgrp stockpiling row.
- Since there is no key variable to link this to, a cross join where combinations of all requested baseline tables and stockgroup will be created.
- Remove duplicate stockgroup values prior to cross join.
- De-duplicate rows to correctly assign stocknumber based on group.
- Join stocknumber back onto original dataset.
- Set defaults and merge in information from original stockpiling file.
- De-duplicate rows to correctly assign stocknumber based on group.
- Join stocknumber back onto original dataset.
- List of stockgroups used for dose calcuations.
CIDACOV pre-processing:
- Rename group as primary to make treatmentpathways compliant with ms_cidacov analysisgrp logic.
- Set createbaseline to Y when a baseline table is requested for any switch within an analysis group.
- Confirm CREATEBASELINE = Y for all groups in level 2 and 3 analyses.
  - List all EOI/REF groups.
- Mark any groups to update CREATEBASELINE and write warning to the log.
If strength or unit parameters do not meet requirements, then abort.
INCLUSIONCODES CONDFROMANCHOR and CONDTOANCHOR:
- Joining on &COHORTFILE. to ignore potential MIL groups.
Perform additional checks on inputfile variables.

Program inputs

dplprior.&RUNID._adjusted_eval(&PERIODIDSTART.-1) (Dataset for covariate stratification and propensity score estimations, ps stratum weighting analysis, inverse probability of treatment weights (IPTW) analysis, and matching strategies and time period.)
dplprior.&RUNID._mstr (Dataset with one record per individual per index date for every GROUP.)
dplprior.&RUNID._signature (Dataset containing metadata associated with the request, including request identifiers, program identifiers, database version, run time metrics, and SAS environment information.)
indata.&DISTABLE. (Dataset with dispensing data.)
indata.&LABTABLE. (Dataset with lab result data.)
indata.&PROCTABLE. (Dataset with procedure data.)
infolder.&COHORTCODES. (Dataset with codes used to define exposures and outcomes of interest.)
infolder.&COHORTFILE. (Dataset used to define enrollment and demographic requirements, type of cohort identification strategy.)
infolder.&COMBOFILE. (Datasets that specifies how combination items should be created.)
infolder.&COMBOFILE.codes (Datasets with codes used to specifie combination items.)
infolder.&CONCFILE. (Dataset to specify GROUP values from a Type 2 analysis and perform additional analyses.)
infolder.&COVARIATECODES. (Dataset for request of covariates presence and analytic adjustment.)
infolder.&COVSTRATFILE. (Dataset describing parameters for a covariate stratification analyses.)
infolder.&INCLUSIONCODES. (Dataset with codes used to define additional cohort inclusion/exclusion criteria.)
infolder.&IPTWFILE. (Dataset specifying the parameters for an IPTW analysis.)
infolder.&ITSFILE. (Dataset with ITS analysis specifications.)
infolder.&MFUFILE. (Dataset for most frequent utilization (MFU) assessment.)
infolder.&MICOHORTFILE. (Dataset for Type 4 analysis with pregnancy cohorts among pregnancies matched to an infant specification.)
infolder.&MONITORINGFILE. (Dataset with monitoring period(s) for descriptive, inferential and sequential analyses.)
infolder.&MULTEVENTFILE. (Dataset to specify COHORTGRP values from a Type 2 analysis and perform additional analyses.)
infolder.&MULTEVENTFILE_ADHERE. (Dataset to specify multiple criteria to determine overall adherence for a Type 2 multiple events analysis.)
infolder.&OVERLAPFILE. (Dataset characterizing an overlap of primary and secondary treatment episodes during the observation window.)
infolder.&OVERLAPFILE_ADHERE. (Dataset to specify multiple criteria to determine overall adherence for a concomitant use analysis.)
infolder.&PREGNANCYCODES.(Dataset for Type 4 cohort identification strategy.)
infolder.&PREGNANCYDURATION. (Dataset for Type 4 cohort identification strategy.)
infolder.&PREGNANCYMETA. (Dataset for Type 4 cohort identification strategy.)
infolder.&PSCSSUBGROUPFILE. (Dataset with all subgroups and subgroup levels for each analysis group.)
infolder.&PSESTIMATIONFILE. (Dataset with the parameters for estimating a PS model and is required for PS-based analyses.)
infolder.&PSMATCHFILE. (Dataset with parameters for a PS matching analysis.)
infolder.&RISKSCORECODES. (Dataset for risk score calculation if requested.)
infolder.&RISKSCOREFILE. (Dataset required for calculating one or more risk scores.)
infolder.&STOCKPILINGFILE. (Dataset with valid dispensings selection used by the stockpiling algorithm to create exposure episodes.)
infolder.&STRATIFICATIONFILE. (Dataset with parameters for a PS stratification analysis and required for PS-based analyses.)
infolder.&TREATMENTPATHWAYS. (Dataset with identification and computation of switch pattern episodes.)
infolder.&TREEFILE. (Dataset with parameters required to execute multiple SI analyses from a basic QRP execution.)
infolder.&TREELOOKUP. (Dataset with hierarchical tree of codes that are eligible to be HOI.)
infolder.&TYPE1FILE. (Dataset required for a background rate calculation cohort identification strategy.)
infolder.&TYPE2FILE. (Dataset required for an exposures and follow-up time cohort identification strategy.)
infolder.&TYPE3FILE. (Dataset required for a self-controlled risk interval design cohort identification strategy.)
infolder.&TYPE4FILE. (Dataset required for a pregnancy episodes identification strategy.)
infolder.&TYPE5FILE. (Dataset required for medical product utilization cohort identification strategy.)
infolder.&TYPE6FILE. (Dataset required for evaluating manufacture level product utilization and switching patterns.)
infolder.&USERSTRATA. (Dataset with ouput tables and stratifications specification.)
infolder.&UTILFILE. (Dataset with medical or drug utilization metrics specification.)
infolder.&ZIPFILE. (Lookup table required if a request requires stratification of results by geographic location.)

Program outputs

&COHORTCODES. (Dataset that includes the codes for outcomes.)
&COHORTCODES._po (Dataset that includes the codes for pregnancy outcomes.)
&INCLUSIONCODES. (Dataset containing inclusion/exclusion codes for the cohort.)
&MONITORINGFILE. (Dataset that defines the relevant query periods for execution.)
&PREGNANCYCODES. (Dataset that includes the codes for pregnancy outcomes.)
&PREGNANCYDURATION. (Dataset for type4 with pregnancy duration codes.)
&PREGNANCYMETA. (Dataset for Type 4 analysis containing pregnancy outcome metadata.)
msoc.&RUNID._runtimes (Dataset containing metrics on total run time, cohort creation run time, and run time for several program processes.)
work._&COVARIATECODES. (Dataset containing covariates codes excluding combination of covariates.)
work._aniv (Dataset containing a list of age anniversary to extract in ms_cidanum.)
work._checklink (Dataset where cohort(s) do have not linkagestatus specified.)
work._cod (Dataset containing a list of cause of death to extract in ms_cidanum.)
work._codedays_check (Dataset containing multiple codedays value per group condlevel subcondlevel.)
work._cohortfile (Dataset with separate count of groups with different age strat.)
work._cohorttype (Dataset for never exposed cohort.)
work._diag (Dataset containing diagnosis codes to extract.)
work._dte (Dataset containing a list of death to extract in ms_cidanum.)
work._enc (Dataset containing a list of encounters to extract in ms_cidanum.)
work._incorrect_combo_covars (Dataset containing combo covars that do not meet specification.)
work._lab (Dataset containing lab result codes to extract.)
work._mil (Dataset containing a list of mother-infant linkage to extract in ms_cidanum.)
work._ndc (Dataset containing dispensing codes to extract.)
work._proc (Dataset containing procedure codes to extract.)
work._tmpdupcode (Dataset containing the same code for multiple outcomes in PREGNANCYCODES.)
work.adherencefile (Dataset to determine overall adherence for a Type 2 multiple events analysis.)
work.cohortfile (Dataset containing cohort group.)
work.cohortfile_for_enr (Dataset where same basecohort are grouped and assess order to process them correctly.)
work.covarcodes (Dataset with covariate code list.)
work.dummycovars (Dataset containing dummy covariates if covarnum are not sequential.)
work.hdps (Dataset containing combination items that behave like DX &DPLOCALLIB.&outname._diag and worktemp._diagextract.)
work.hdpssettings (Dataset to id which group for HDPS.)
work.labcodetype_check (Dataset containing lab code types that are used to define a single covariate.)
work.looks (Dataset containing non overlapping looks dates.)
work.migroupnames (Dataset for PS/COV analysis.)
work.monitoringfile_agg (Dataset containing minimum startdate and maximum enddate.)
work.riskscorefile (Dataset containing unique risk scores.)
work.riskscorescodes (Dataset containing risk score codes for raw extraction.)
work.secondaryinputfile (Dataset for secondary episode analysis.)
work.stockpile_covar (Dataset containing variables to stockpile (where stockgroups are only in covariatecodes file).)
work.stockpile_noncovar (Dataset containing variables to stockpile.)
work.studynames (List of covarnum-studyname for labels.)
work.type2file (Dataset for never exposed cohort.)
work.type6groups_forbaseline (Dataset with identification and computation of switch pattern episodes sorted by analysisgrp and descending createbaseline.)
work.unit_check (Dataset containing subcondlevel with different unit value.)
work.userstrata (Master list of all strata.)
work.userstrata_denom (Dataset containing unique levelvars per table type (prevalent or not).)
work.userstrata_its_lookup (Dataset containing levelvars per table type prevalent or not.)
work.utilfile (Dataset with medical or drug utilization metrics specification.)

Usage

%ms_processinputfiles();

Parameters: None.

SAS Macros Dependencies

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

	QRP