GEMINI-Medicine / Rgemini

A custom R package that provides a variety of functions to perform data analyses with GEMINI data
https://gemini-medicine.github.io/Rgemini/
Other
3 stars 0 forks source link

Function deriving total inpatient cost #136

Open loffleraSMH opened 4 months ago

loffleraSMH commented 4 months ago

New Feature Request

Function that derives cost of hospitalization based on Resource Intensity Weights (RIW), which represents weighted cost relative to average inpatient cost. The actual cost amount can be inferred based on CMG methodology (by year, based on when data were pulled) and needs to be adjusted for (healthcare-specific) inflation.

This is the suggested general approach (may need to be refined), which is used in the Pediatric Complexity project:

  1. Create a csv file for CPWC(CHSC) for each hospital number & year, based on the info here: https://yourhealthsystem.cihi.ca/hsp/inbrief?lang=en#!/indicators/015/cost-of-a-standard-hospital-stay-cshs/;mapC1;mapLevel2;provinceC9001;trend(C1,C5001);/
  2. This then needs to be merged with other GEMINI tables as follows (preliminary plan suggested by Surain):
  3. Merge ipcmg into your cohort by genc_id, assess missingness of methodology_year and riw_15
  4. Methodology_year has some missing data. If methodology_year is empty, but there is another row that is otherwise identical (in terms of cmg, diagnosis_for_cmg_assessment, comorbidity_level, riw_inpatient_atypical_indicator), impute the methodology_year from that row.
  5. Methodology_year also has some values that are >1 year different than calendar year, calling into question data quality. We use the same process as above to impute methodology years in rows where methodology_year and calendar year are >1 year apart.
  6. Report all records excluded due to missing values required for total cost calculation (i.e. RIW).
  7. Merge in CHSC data by hospital_num
  8. Multiply riw_15 by CHSC
  9. Adjust for health-care specific inflation.

Once the method has been developed, we should triangulate the results with some publicly reported values (probably from CIHI) to make sure we are in the ballpark.