Analysis code: measuring hospital acquired harms associated with low-value care

Kelsey Chalmers, Lown Institute. 2020-2021.

Create cohorts and outcomes

top_run.sas: top level file to set up the SAS environment, run the cohort selection and outcomes. Calls the following files and macros they create.

Low-value care

lvc_label.sas: applies criteria for overuse cases in outpatient and inpatient.

This program requires lvc_definitions.xslx, the definition code list of the included low-value services.

Criteria: principal procedure label

pri_procedure.sas: Creates the label for the principal procedure for each admission.

Harms: HACs

hac_label.sas: applies HAC codes/criteria on admissions.

This program requires hac_{xx}_{name}.xlsx The HAC codes from CMS. Includes:

hac_01_foreignobject.xlsx
hac_02_airembolism.xlsx
hac_03_bloodinc.xlsx
hac_04_pressureulc.xlsx
hac_05_falls.xlsx
hac_06_cauti.xlsx
hac_07_vcai.xlsx
hac_09_glycemic.xlsx
hac_12_ssi_ortho.xlsx
hac_13_ssi_cardio.xlsx
hac_14_pneumothorax.xlsx

Harms: PSIs

psi_label.sas: applies the PSI codes/criteria on admissions.

This is from the AHRQ PSI Safety Indicators SAS software.

It requires the formats created by psi_hosp_formats.sas, as well as the macros created in psi_macros.sas.

Outpatient 7-day hospital admissions

hospitalvisit_outpatient_lavel.sas: finds the unplanned inpatient hospital admissions within 7-days from a low-value outpatient procedure.

This program requires the CCS codes, which are created as a format from ccs_dx_icd10cm_2019_1.csv (diagnoses) and ccs_pr_icd10pcs_2019_2.csv (procedures).

These codes are combined based on the Yale New Haven measures, which are recorded in the table unplanned_admissions.xslx. This is converted to formats in css_formats.sas.

Create outputs for paper

These functions create the required output from the CMS Virtual Research Data Center.

Cohort description

figure_1_cohort.sas: creates cohort counts used in figure 1.

Outpatient procedure counts and outcomes

table_1_outpatient.sas: creates counts by service for the unplanned admissions, HACs and PSIs for the included outpatient admissions. Creates confidence intervals based on bootstrapped random samples.

Inpatient procedure counts and outcomes

table_2_inpatient.sas: creates counts by service for the HACs and PSIs for the included inpatient admissions. Creates confidence intervals based on bootstrapped random samples.

Individual PSI counts by inpatient procedures

table_3_psi.sas: creates counts by PSI and service for included inpatient admissions. Creates confidence intervals based on bootstrapped random samples.

Costs and LOS for inpatient procedures

table_4_costs_los.sas: creates the estimated length of stay (LOS) and costs for admissions with and without a HAC.

Uses AHRQ programs to create Elixhauser comorbidity labels:

comorb_icd_10cm_analy_v20211.sas
comorb_icd10cm_format_v20211.sas

Uses the cost to charge ratio file for providers, output__CCR.csv (can be created using the R functions in the medicare_cost folder).

Medicare cost to charge ratios

Within medicare_cost folder, there are three R functions.

medicare_cost.R takes the downloaded cost reports for each year and finds the relevant information to calculate cost to charge ratio.
ccr_year.R creates the annualised CCRs.
ccr_threeyears.R outputs the provider tables with full CCRs for 2016 to 2018, the study period.

Other outputs

output_paper.sas: creates other counts reported in main text.

Miscellaneous

do_over.sas: Very handy macro from Ted Clay and David Katz that makes SAS coding over different variables much easier.

kelsey209 / hac-lvc-osf