This macro computes denominators. More...

Detailed Description

This macro computes denominators.

Only compute denominators if OUTPUTDENOM ne N and USERSTRATA file is specified and contains relevant tables.
- Original enrollment periods.
  - If death censoring is required, it was already applied (Enr_End was previously truncated by CIDAnum).
  - Restrict eligibility to study period (minus maximum of required lookbacks).
  - Initially truncate start eligibility enrollment start by max of lookback periods.
- Inclusion required (POV3 Incl) for eligibility.
  - For each condition, shave periods for all subconditions.
  - Determine how many levels of subconditions needed and if condition is an inclusion.
  - Create &cond specific dataset.
  - Determine if subcond is an inclusion or exclusion.
  - Merge time periods for each subcondition and shave for each.
    - When cumdose or afdd specified create a de-duped inclexcl dataset.
      - Necessary for codedays computation.
    - Only keep time periods that meet codedays criteria.
  - If mincumdose or average filled cumulative dose is specified then restrict eligibility periods.
    - Drop refstart and refend since _Denom&Group can be looped.
    - When cumdose or afdd specified create a de-duped inclexcl dataset.
      - Necessary for codedays computation.
    - Only keep time periods that meet codedays criteria.
  - Output dataset for inclusion=1 and for exclusions=0 and then (a) shaveoutside (b) shaveinside denom.
    - For inclusions, need to merge periods.
    - First shaveoutside denom file but need to merge inclusion periods together.
      - _Denom&Group._STACK_INC includes all time periods where the potential index dates will meet at least one of the cond inclusions.
    - Merge all periods of ineligibilities.
      - _Denom&Group._STACK_EXCL includes all time periods where the potential index dates will meet at least one of the cond exclusions.
    - Remove all these periods from the ELIG.
- Create separate prevalance and incidence cidadenom datasets.
  - Group Index (POV1) to remove from eligibility.
    - Need to use washper2 when autoprev = Y.
  - Follow-up Event (POV5) to remove from eligibility.
    - Only apply to type 2 analysis.
  - Follow-up Washout (POV6) to remove from eligibility.
  - This macro is to avoid having to output one record per interval per stratification level which for certain partners, can become quite large.
  - Make sure the number of enrollment days remaining after each potential elig days is sufficient.
    - Get first Index Dates and Event Dates to handle member days correctly according to CohortDef.
    - Adjust eligibility end date.
    - Secondary episode observation window.
      - Remove invalid time due to obsfrom-obsto enrollment/data availability criteria.
    - Because DenomEnrEndDt can have been truncated, keep only the observations with positive member days.
  - Add dob/sex/race/hispanic.
    - Demographics:
      - Process zip codes.
      - Truncate periods such that all interval lies in MinAge-MaxAge.
        
        Member not meetin minimum age yet.
        
        Member has reached or will reach will reach MaxAge.
      - At this point, each day in the DenomEnrStartDt-DenomEnrEndDt intervals represents a potential index date if the patients would have started one of the study drugs.
        
        Restrict only to the ExpPeriodStartDt-ExpPeriodEndDt period.
        
        This will simplify accumulating days for year and year/Month strata, especially if the ExpPeriodStartDt day is not the first day of the month and or the first month of the year.
  - Save denominators for future processing.
  - Create individual denom contribution to final table.
  - If no observations in _denom&group, square the table.
  - Sum denominators for each stratification.
  - Check if level is required in userstrata file.
    - Check if level needs the CreateSums macro output.
      - Quarter periods.
      - Geographic should not produce month/quarter output.
  - Break down periods by age categories.
  - Geographic categories.
    - Produce uncertain level stratifications, regardless of geographic region specified.
      - Adjust denominator for zip_date.
      - 1. If zip_date < DenomEnrStartDt, full enrollment is certain.
      - 2. If zip_date is missing, full enrollment is uncertain.
      - 3. If zip_date > ZipEnrEnd, then no uncertain time.
      - 4. If DenomEnrStartDt <= zip_date <= DenomEnrEndDt, then calculate uncertain time.
        
        Adjust denom enrollment.
  - Verify if at least one stratum was computed.
  - Store patient denoms to dplocal.
  - For type 2:
    - Create Prevalent and Incident dataset.

Program inputs

infolder.&ZIPFILE. (Lookup table required if a request requires stratification of results by geographic location.)
work._fupevent (Follow-up Event (POV5) to remove from eligibility.)
work._fupwash (Follow-up Washout (POV6) to remove from eligibility.)
work._groupindex (Dataset containing all potential outcomes.)
work._inclexcl (Dataset with a list of control patients and demo data.)
work._inclexcl_cumdose (Dataset with eligible periods meeting the cumulative dose or average filled daily dose criteria.)
work._it&INCLUSIONCODES. (Dataset with inclusion/exclusion definition.)
work.userstrata_denom (Dataset containing unique levelvars per table type (prevalent or not).)
worktemp.enr_&ENROLLMENTNUM. (Dataset containing enrollment episodes.)

Program outputs

dplocal.&RUNID._denomcounts&DS_SUFFIX. (Dataset for eligible members and member-days metrics for the T1_CIDA and T2_CIDA tables in the msoc output folder.)
work._eligincl_&SUBCOND. (Dataset containing eligible time periods for each subcondition.)
work._overlap_spans_&SUBCOND. (Dataset containig only the time periods that meet codedays criteria.)
worktemp._nvrexp&GROUP. (Dataset containing never exposed cohort.)

Usage

%ms_cidadenom();

Parameters: None.

SAS Macros Dependencies

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

	QRP