As the volume of encounter data collected by CMS grows — reaching an anticipated 775 million records for 2017 — the agency is working to ensure that data are accurate and complete through analysis, communications with Medicare Advantage organizations (MAOs) and other efforts that it will continue to refine, officials attested during a session of the CMS Medicare Advantage and Prescription Drug Plan Fall Conference, held Sept. 7 in Baltimore.
CMS in 2012 began collecting data from MAOs through the encounter data system and in 2016 started phasing in EDS-based payments. For 2018, 15% of payments will be based on EDS scoring while the other 85% will come from the old risk adjustment payment system (RAPS). But use of the system for payment purposes has received criticism from plans — which have reported differences between RAPS and EDS scoring — and the Government Accountability Office, which earlier this year asserted that CMS has made “limited progress” in validating encounter data and that it should implement GAO’s earlier recommendation that CMS fully assess data quality before use.
Though it is still processing records for 2017, CMS expects to collect roughly 775 million submissions this year, compared with 500 million collected in the first year that submissions were required (see chart below), and estimates it will have collected more than 3.2 million records between calendar years 2013 and 2017. While some of that growth has been related to rising MA enrollment, CMS has also observed growth in submissions per beneficiary as the collection process matures, said Shruti Rajan, an official with the Division of Encounter Data and Risk Adjustment Operations in the Center for Medicare’s Medicare Plan Payment Group, during the session.
The most commonly recorded “front-end” edit from the first quarter of 2017 as well as 2015 and 2016 has been the diagnosis code, occurring at a rate of 0.70% in the most recent quarter. Meanwhile, of the five most frequently occurring “back end” edits (i.e., those reflected in MAO-002 reports sent back to sponsors), chart review duplication (1.84%) tops the list for professional records and exact inpatient duplicate encounters (1.22%) occurs most frequently for institutional records.
Through various communications, CMS has heard from plans about the latter. “The reasons for the concerns are varied and many times they include plans attempting to resubmit an encounter but with a change in diagnoses or some of the other data fields,” said Rajan. “Our analysis indicates that the data elements on these duplicate records might differ between the duplicate record and the initial submission, and so we are considering options related to this edit.”
Analysis Includes Record-Level Review
To improve the EDS, which began capturing records in 2012, CMS in 2015 implemented a two-pronged Medicare Data Integrity Plan that includes (1) validating the completeness and accuracy of encounter data and (2) communicating with MAOs on the best ways to improve data submissions. This involves four “interrelated” activities: analysis of encounter data, communication with MAOs, monitoring and compliance, said officials.
“Analysis is really the crux…that’s supporting the other pieces,” said Monica Reed-Asante, deputy director with the Division of Payment Policy within the Plan Payment Group. This includes taking a close look at edits to see if they are centralized and if they are affecting certain submitters, and to make sure they are functioning as intended, she told attendees. In addition, the agency is conducting record-level analysis, looking at data elements for validity and completeness, record level counts and submission patterns. “That data element-level review is really to analyze the accuracy of the information that is submitted in specific fields,” she said.
Moreover, CMS has tried to control for enrollment in its analysis, so in order to assess record volume, for example, it has been comparing encounter data records per 1,000 enrollees to fee-for-service (FFS) claims per 1,000 beneficiaries on a national and regional level. Reed-Asante said CMS recognizes that FFS may not always be the most appropriate benchmark for this analysis and is looking at other options now that encounter data is maturing. For instance, it may be possible to compare records for a specific submitter over the course of several years.
A couple of the findings of CMS’s analysis are:
Over the course of four service years (2013-2016), inpatient, professional and outpatient claims per 1,000 MA enrollees are consistently lower than for FFS beneficiaries. As a percent of FFS claims, MA encounter data records for inpatient claims were particularly low, from a range of 61% in 2013 to a high of 72.5% in 2015. Analysts thought the large differences could be due to submission issues, lower utilization or a combination of both, and wanted to dig deeper, said Reed-Asante. They arrived at a January 2017 study that looked at MA and FFS claims for 2010 and found that for MA, the inpatient days were roughly 16% lower in MA than FFS. CMS used that as a basis for adjusting the FFS benchmark, which lowered the differences so that MA increased as a percentage of FFS from 61% to 73% in 2013, for example, “making it look much more comparable to FFS,” said Reed-Asante.
CMS also conducted an analysis of inpatient encounter data records to compare “no-pay” claims (which most hospitals are required to submit as part of the Disproportionate Share Hospital Payment program). Theoretically, these claims would have an encounter data submission in RAPS and in EDS, but when matching them on four fields, analysts found that there is some difference in submission patterns for RAPS and encounter data for inpatient records, although the occurrence of a no-pay event found in RAPS but not in EDS has dropped from 20% in 2013 to 12% in 2015, she observed.
In addition to ongoing communications with MAOs (e.g., site visits, one-on-one calls and monthly user group calls), and monitoring and compliance activities, CMS is looking at next steps that include updating and streamlining its guidance website to make the information for MAOs more clear and easily accessible, added Reed-Asante.
To view a web replay of the session, visit the CMSHHSgov channel at www.youtube.com.
MA Encounter Data Submissions Continue to Rise
SOURCE: CMS, presented at the Medicare Advantage and Prescription Drug Plan Fall Conference, held Sept. 7 in Baltimore.