ms_createmicohorts.sas File Reference

This macro creates exposure and control (unexposed, comparator) cohorts, process according to MICOHORTFILE including additional inclusion/exclusion criteria, and evaluate outcomes. More...

Detailed Description

This macro creates exposure and control (unexposed, comparator) cohorts, process according to MICOHORTFILE including additional inclusion/exclusion criteria, and evaluate outcomes.

• Check if MIDAYSDIFF is specified, if so and MILTABLE exists,then merge in ADATE.
  
• Assign libname for &RUNID._concepi_mstr_preg_tri dataset used.
  
• Determine unique combination of GROUPNAME-OUTCOMEPOP-GRACEDAYS to determine if a consolidated enrollment span has been previously computed
  
• Loop through milgrp.
  
  • MICOHORTFILE parameters.
    
    • Infant sex values applicable to demographics.
      
    • When linked infant parameters are missing, assign default values.
      
  • COHORTFILE for enrollment parameters.
    
    • Age stratification.
      
    • Enrollment gap.
      
    • Pre-pregnancy period.
      
  • Assess correct HDPS window for this specific milgrp.
    
  • For data-driven query period.
    • Determine query period.
      
    • Add MILGRP end date to monitoring file.
      
    • Recreate look dataset.
      
    • Output METADATA file.
      
  • Determine maximum treewashper to assess enrollment.
    
  • Custom warning if control cohort is unexposed and index date is INDEXDT_EXP.
    
  • Custom warning if control cohort is unexposed and outcome anchor is INDEXDT_EXP.
    
  • If OUTCOMEPOP = I or OUTCOMEPOP = MI, use gracedays to determine if infant enrollment start should be reassessed as the delivery date or remain as its original value.
    
  • If OUTCOMEPOP = MI, create consolidated mother-infant enrollment spans by bridging enrollment spans using the ENROLGAP parameter.
    
    • Find all infant enrollment dates from pregcohort (with gracedays applied) and also find those from worktemp.dem_infant_enr&ENRNUM. that are not in pregcohort.
      
    • Get all unique patid and linkedid so we can retrive all enrollment spans for this combination.
      
    • Join all enrollment from mothers and infants.
      
    • Consolidate mother-infant enrollment spans by bridging enrollment spans using the ENROLGAP parameter.
      
    • Change episode if maximum allowable gap is reached.
      
    • Reconciliation of eligible episodes.
      
      • outenr&ENRNUM. is the consolidated enrollment dataset formed by bridging the enrollment episodes of mothers and infants.
      • Add in consolidated enrollment dates (outcome_enr_start and outcome_end_end) to concepi_mstr_preg_tri dataset.
        
  • Create exposure (EOI) and reference (REF) cohorts. Cohorts are created by calling macro create_eoiref_cohorts for each cohort
    • For the EOI and exposed ref cohort, identify all pregnancies with the exposure during the exposure evaluation window
    • For the REF cohort, identify all pregnancies without the exposure during the exposure evaluation window. Because dplocal.&RUNID._concepi_mstr_preg_tri contains 1 row per MOI, roll up to 1 row per delivery
    • Exclude any pregnancy episodes if the linked infant exceeded maximum allowable days required between birth date and delivery date if only the MIL pregnancy outcome is requested for the cohort
    • Exclude any pregnancy episodes if the linked infant is not the desired sex if only the MIL pregnancy outcome is requested for the cohort
    • Create dataset to use for further exclusions:
      • Define all relevant dates (exposurefrom/exposureto, outcomefrom/outcometo).
        
      • Reassign index date if begins prior to exposure from period.
        
      • Assign ID to delivery in order to link Mothers-Infants during Incl/Excl processing.
        
      • Compute difference between date of enrollment and date of birth for linked infants.
        
      • Compute length of enrollment for linked infants.
        
      • For time to event analysis outcome assessment should only begin from the day of index exposure and later for cohorts with MOI.
        • Any outcome occurring before index exposure will be ignored.
        • For unexposed control cohort, exposure start window or start date should Not be considered.
          
      • Evaluate enrollment for HDPS window from indexdate and episode end date.
        
      • Re-evaluate enrollment for baseline covariates/HDPS anchored on index exposure for exposure cohort and exposed reference cohort.
        
      • Enrollment must be satisfied during the exposure and pre-pregnancy outcome assessment period.
        • When outcomepop=M, evaluate Treewash enrollment from outcome start date.
          
      • Compute followuptime and ttc for treescan.
        
      • Check if enrollment is met on the first day of the outcome window.
        
      • Recompute age, year, month and quarter using the indexdt defined by INDEXDATE.
        
      • Check if the indexdt defined by INDEXDATE is recorded in the query period.
        
      • Check if required post-pregnancy enrollment criterion is met.
        
    • Assess whether delivery date is during current look period (analysis =TREE)
    • Reassess age group based on user defined index date and exclude pregnancy episodes that no longer meet the age group criteria
    • Exclude episodes where the user defined index date does not overlap the query period
    • Select all enrollment spans for mothers and infants.
      
    • Process additional codes: Inclusion/Exclusion and outcomes.
      
      • Break out CaresettingPrincipal into Caresetting and Principal.
        
      • Extract milgrp-specific records.
        
      • Restrict to mother and infant records applicable to cohort
      • Make sure all claims overlap enrollment span and only keep eligible exposures (overlaping Enr_Start Enr_End).
      • Stockpile RX claims
        
      • Need to reshave because Stockpiling may have pushed claims outside enrollment period.
      • Create _InclExcl _FUPEvent, and _FUTforTrunk datasets.
        
      • Evaluate inclusion/exclusion criteria.
        
        • Rename variables related to episodes nomenclature to make them consistent with cida cohort episodes and easily usable in the ms_createpov3 macro.
        • Reassign original variable names.
        • Identify all conditions on inclusioncodes file.
        • If one exposure is not excluded then all exposures related to that pregnancy are not excluded.
    • Level 7: Exclusion - Pregnancy episodes must satisfy the pre-pregnancy enrollment requirements within the query period.
      
    • Level 8: Exclusion - Pregnancy episodes must satisfy the post-pregnancy enrollment requirements within the query period.
      
    • Level 9: Exclusion - Pregnancy episodes must satisfy the inclusion and exclusion criteria.
      
    • Check if FUT codes truncation needs to be applied.
      
    • Evaluate outcomes (HOIs).
      • Check if pregnancy outcomes should be assessed as HOIs and add them to the list of potential HOIs if this is the case
      • Extract cohort specific codes (mother and infant).
      • Assess incusions/exclusions criteria for events if necessary.
      • Select 1st event date in outcome window and merge with master episode list.
        
    • Apply micohortdef criterion and compute followuptime for time to event analysis.
      
      • Timetocensor should start from User-defined Indexdate.
        
    • Level 10: Exclusion - Must have at least 1 day of followup in the risk window.
      
      • For L2 analyses, this exclusion will be performed in ms_covariateadjustment after PS model.
        
    • Level 11: Exclusion - Pregnancy episodes must meet the micohortdef criterion.
      
    • Output attrition table to msoc folder.
      
  • Combine _eoi and _ref cohorts.
    
  • Compute MOI metrics for final cohorts for baseline/table 1.
    
    • Summarize.
      
    • Save to dplocal.
      

Program inputs

• &LIB_CONCEPI_MSTR.&RUNID._concepi_mstr_preg_tri (Dataset containaing trimester statistics for each pregnancy outcome and medical product of interest.)
• dplocal.&RUNID._concepi_mstr_preg_tri (Dataset for Type 4 analysis for the pregnant cohort containing trimester statistics for each pregnancy outcome and medical product of interest.)
  
• infolder.&&PSFILE_CURRENT. (Dataset for PS matching (&PSMATCHFILE.), IPTW (&IPTWFILE.) and/or PS stratification (&STRATIFICATIONFILE.).)
• infolder.&COHORTFILE. (Dataset used to define enrollment and demographic requirements, type of cohort identification strategy.)
• infolder.&INCLUSIONCODES. (Dataset with codes used to define additional cohort inclusion/exclusion criteria.)
• infolder.&IPTWFILE. (Dataset specifying the parameters for an IPTW analysis.)
• infolder.&MICOHORTFILE. (Dataset for Type 4 analysis with pregnancy cohorts among pregnancies matched to an infant specification.)
• infolder.&PSESTIMATIONFILE. (Dataset with the parameters for estimating a PS model and is required for PS-based analyses.)
• infolder.&STRATIFICATIONFILE. (Dataset with parameters for a PS stratification analysis and required for PS-based analyses.)
• infolder.&TREEFILE. (Dataset with parameters required to execute multiple SI analyses from a basic QRP execution.)
• infolder.&TYPE4FILE. (Dataset required for a pregnancy episodes identification strategy.)
• &COHORTCODES. (Dataset that includes the codes for outcomes.)
• &MONITORINGFILE. (Dataset that defines the relevant query periods for execution.)
• &INCLUSIONCODES. (Dataset that includes codes for inclusion criteria)
• work._attrition (Temporary dataset with number of mother-infant pairs/pregnancy episodes excluded and remaining for each cohort.)
• work._attrition_cond (Dataset containing defined group pregnancy episodes.)
• work.cohortfile_for_enr (Dataset where same basecohort are grouped and assess order to process them correctly.)
• work.hdps (Dataset containing evaluation window dates and anchor dates for high-dimensional propensity score estimation on selected analysisgrp.)
• work.stockpile_noncovar (Dataset containing variables to stockpile.)
• worktemp.dem_infant_enr_&ENRNUM. (Dataset containing infant enrollment spans.)
• worktemp.drugs (Dataset containing RX to extract.)
• worktemp.drugscomb (Dataset containing combination items that behave like RX.)
• worktemp.enr_&ENRNUM. (Dataset containing enrollment spans.)
• worktemp.labextract (Dataset containing labs to extract.)
• worktemp.meds (Dataset containing combination of _diags and _procs.)
• worktemp.milextract (Dataset with mother-infant linkage data.)

Program outputs

• dplocal.&RUNID._mstr_mi (Dataset for Type 4 analyses produced for exposed and comparator/unexposed pregnant cohorts.)
• msoc.&RUNID._metadata_for_time_period_&PERIODIDSTART. (Dataset used when doing sequential SI analysis.)
• msoc.&RUNID._mil_attrition (Dataset containing the number of mother-infant pairs/pregnancy episodes excluded and remaining for each cohort for Type 4 analysis.)

Usage

%ms_createmicohorts();

Parameters

None.

SAS Macros Dependencies

• ms_agestrat.sas
• ms_attrition_cidacompute.sas
• ms_caresettingprincipal.sas
• ms_createpov3.sas
• ms_extractlabs.sas
• ms_extractmeds.sas
• ms_loopmeds.sas
• ms_outputruntimes.sas
• ms_periodsoverlap.sas
• ms_processwildcards.sas
• ms_shaveoutside.sas
• ms_starttimer.sas
• ms_stoptimer.sas
• ms_stockpiling.sas

Note

When creating the cohorts, the MICOHORTFILE.INDEXDATE parameter is used to assign one of the following date to the INDEXDT variable: PREGSTARTDT, INDEXDT_EXP, PREGENDDT. This is used for purposes of:

• calculating maternal age
• computing signal identification TTC
• computing zip_uncertain
• performing inferential/tree analyses.

Author

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