Abstract
Purpose :
In 2014, CMS released provider utilization data for 2012 . Data for 2014 was subsequently released in 2016 and it thus offers the opportunity to compare utilization pattern changes over this period of time. This project attempts to identify a relation between public data release and provider practice patterns.
Methods :
The dataset covers 900,000 providers and 9 million transactions. Data for all three years was compiled into a single dataset using a combination of MongoDB, R and Hadoop. Additional data fields were added to reflect normalized data, such as number of treatments per patient for each code.
Results :
1. Impact on code utilization
Of the 765 available CPT codes in 2014:
446 codes decreased
319 codes increased
Of the 760 available CPT codes 2013:
393 codes decreased
367 codes increased
2. Between 2012 and 2014, ophthalmology ranked second among all specialties in
medicine in its cost to CMS. When part D is removed, ophthalmology ranked fifth
in 2012, and sixth in 2013 and 2014.
3. Paired t-test for comparison of means of number of treatments per patient
Shapiro test failed to reject the null hypothesis that the samples came from a normal
distribution.
Difference of mean was statistically significant (t = -5.8012, df = 3895, p-value <2.2e-16).
95% confidence interval was -0.14 to -0.12. Mean of the differences was -0.13.
Conclusions :
Comparing pre-treatment (2012 to 2013) to post-treatment data (2013-2014), there is an overall decrease in the number of codes whose utilization increased, and an increase in the number codes whose utilization decreased. The overall cost for ophthalmology increased over this time period; the cost per provider,however, decreased when excluding part D medications. The impact of Part D medications on ophthalmology-related expenditure is noteworthy.
This is an abstract that was submitted for the 2017 ARVO Annual Meeting, held in Baltimore, MD, May 7-11, 2017.