Concept: MCHP De-Identified Home Care Client File

Creating A Useable File

Data for the Home Care Study came from the Manitoba Health Home Care Utilization - Manitoba Support Services Payroll (MSSP) database. The raw MSSP client data had to be worked with to create a useable file. A MCHP version was needed for several reasons:


• observed inconsistencies around the prompt "closing" of the home care records.
  
• the possibility of being registered in more than one home care program at the same time, making double counting an issue.
  
• when setting together multiple years, the existence of records with the same program and registration date, but differing termination dates.
  
• other reasons, such as the complexities around hospitalizations.

MCHP De-identified Home Care Client File

MCHP created a new, de-identified, home care client file with the following standard exclusions / definitions:


• Exclude temporary client numbers from the client root data.

  SAS CODE: if substr(client,1,2)='TM' then delete;

• Exclude program codes not beginning with 1 (the 1 is for home care)

  SAS CODE: if substr(program,1,1) ne '1' then delete;

• Exclude records for clients whose postal code does not begin with 'R', as given in the client root data (i.e. out of province residents)

  SAS CODE: if substr(postal,1,1) ne 'R' then delete;
  NOTE : for analyses by sub-areas of Winnipeg, Public Trustee's postal code 'R3C 3L6' was also excluded.

• After setting all '00000000' termination dates to '99999999', exclude records with termination date prior to registration date:

  SAS CODE: if termdt < regdt then delete;

• Because records with the same program and registration date could be repeated in multiple years, only the most recent year was kept (likely to have the most correct termination date) for records with matching client number, program and registration date.

  SAS CODE:
  
  proc sort data=XXX;

  by client program regdt descending claimyr termdt;
  run;

  data XXX;

  set XXX;
  by client program regdt descending claimyr termdt;
  if first.regdt then output;
  run;

  ** where claimyr is a created variable with the value of the year (4 digit so that the 2000 years sort properly) of data that was set;

• Inconsistencies in closing dates were then fixed up (see below for details).
  
  
• If a program/registration date was opened in one year (e.g. 97/98) AND not terminated, but the information was not repeated in the following year (98/99), then a termination date of the fiscal year end was imputed (March 31, 1998).
  
  
• Because records could contain overlapping dates, the records were combined into episodes. For the home care report, any records being separated by 7 days or less were combined into an episode. However, this is not done in the MCHP version of the data so that individual PIs can decide what is best for their purposes.

Inconsistencies in prompt closing of home care records

Inconsistencies had been observed around the prompt "closing" of the home care records. A de-identified MCHP home care client file was constructed that augmented and updated MSSP information with data from Personal Care Home files and vital statistics. The MSSP data identified start and end dates for episodes of home care and the end date for each home care episode was compared against files that held similar information on admission to Personal Care Homes, coverage cancellation and death and, where necessary, was updated.