In this project, I created a dashboard allowing peer group comparisons of HCAHPS patient satisfaction scores for small, medium, and large hospitals by state. Hospitals could use the dashboard to evaluate their performance relative to peer institutions and identify targets for improvement.

The Hospital Consumer Assessment of Healthcare Providers and Systems (HCAHPS) Survey is a national, standardized, publicly-reported survey of patients' perspectives of hospital care. The 29-item instrument and data collection methodology allows valid comparisons to be made across hospitals. The survey was nationally implemented in 2006 and public reporting of hospital scores began in 2008. Since 2012, HCAHPS scores have also played a role in hospital payment through the Hospital Value-Based Purchasing (VBP) program. For additional information on the survey and methodology see  http://www.hcahpsonline.org/.   

The datasets used in this project are available through the Data Catalog on Data.CMS.gov, specifically the Patient survey (HCAHPS) - Hospital dataset. I used data from 2022, the last complete year in which data had been reported. To create groups of small, medium, and large size hospitals, I also used a table of hospital bed counts provided by healthcare data analyst Joshua Matlock.

I used PostgreSQL 16/pgAdmin 4 to clean, join, and prepare the data (code linked below) and saved the output data table to import into Tableau. I created tables for the HCAHPS and bed data, padded provider ID numbers (CCNs) that had been truncated in Excel during data gathering, changed date columns to properly formatted date types, created a CTE with the hospital bed table to assist with joining selections from that table with the HCAHPS data, and used a window function to identify the most recent bed count so only those rows would be joined with the HCAHPS table.

Using output from the PostgreSQL queries, I created a dashboard in Tableau allowing:

• comparison of hospitals in peer groups within state and hospital size

• examination of hospitals in the peer group with the highest to lowest percentage of patients rating the hospital with top-box scores (9-10)

• selection of individual hospitals to review:

  1.  survey response rate and number of completed surveys

  2.  for each question on the survey, the difference in patient responses from the mean of the cohort

  3.  for each question on the survey, a visual representation of where responses fell in the range of cohort hospital averages

I chose high contrast font colors and the color blind palette for accessibility. I used calculated fields to filter by top-box scores (9-10 or "always") and hospital size (large: >=500, medium: 100-499, small <100 beds) and used a level of detail expression to calculate the difference between a selected hospital's percentage of patients giving top-box answers for a question and the average percentage for all hospitals in the cohort. I created a parameter for hospital names and parameter action to allow users to select the hospital name in the "% of Patients Rating Hospital 9-10" chart and have the other dashboard components update to show the data for the selected hospital.