Value-Based Purchasing and the LTC physician: We are high cost and low qualityOctober 1st, 2013 by
Today’s blog post is a preliminary analysis of how LTC physicians will fare under the Value-Based Purchasing (VBP) program.
I believe we are marked for failure. LTC physicians will look like the least competent group of Medicare providers in the medical community, and we will be penalized with lower payments and public shaming. This is an unavoidable consequence that arises solely from serving the nursing home population.
VBP affects groups larger than 99 this year, 10 or more in 2014, and by 2015 every group of two or more will be included. VBP measures on two performance dimensions: quality and cost. Each dimension has three performance levels: low, medium and high.
Prior to implementing VBP, CMS is privately showing medical groups how they will fare under the program. This is done by use of a group specific 2012 Quality and Resource Use Report and Physician Quality Reporting System Feedback Report. This QRUR report uses Medicare Part B claims to associate patients with the group. Patients are associated with the medical group using logic similar to ACO attribution. The report uses the universe of Medicare Claims (A, B and D) to create cost and quality measures for each associated patient.
It is important to understand the Clinical Quality Measures in QRUR are indirect; they use patient-specific ICD-9 codes submitted by the medical group and look for other claims-based activities (medications, tests, procedures) to validate that the expected quality related activity was performed.
This is the essence of the Administrative Claims Option available for PQRS GPRO beginning in 2013.
Now for some specifics:
Our LTC physician group in North Carolina received our QRUR report two weeks ago. (Here is a link to a decent discussion of these reports from ASC Surgery News). These reports are now available for any medical group that submitted claims to Medicare Part B on behalf of 25 or more providers during 2012. Previously, the report was only available to large groups (more than 99) who participated in the GPRO option for PQRS during 2011.
While preliminary, the systematic problems for our group that I found were due to the structural nature of our LTC practice, and not to our clinical behavior.
Here is an analysis of what I can deduce:
Quality: All groups receive a composite quality score derived from a family of up to 26 measures. The actual number of applicable measures is based on the group’s universe of patients and ICD-9 codes. The patient claim data is “mined” from the universe of 2012 CMS claims. Only measures that have a patient population of at least 20 individuals are used for computing the score. The QRUR report for our physician group identifies 3,400 attributed patients out of 6,800 unduplicated patients in this practice during 2012; so 50 percent were attributed.
The group’s composite “quality score” for the 24 measures that applied was 1.15 standard deviations below the mean score for the benchmark (approximately 3,900 primary care groups). We were subjected to a large number of measures because of an intentional effort to document acuity (submitting multiple ICD-9s per claim). That is an important strategy in any Medicare Shared Savings program, but it does have consequences.
(At this point, it is worthwhile pointing out that this particular practice is led by a five-physician board, each of whom are Board Certified in FP or IM, and are all AMDA CMDs. The shortest tenure in LTC medicine for any of them is more than 12 years. I presume they have a solid grasp on the concept of “best practices” in LTC.)
Issues with the quality measures fall into several different categories.
- Industry practice: We know that 100 percent of our patients with Warfarin therapy are supposed to have routing INR monitoring, and that state nursing home surveyors routinely check for this. However, our score showed that 55 percent of eligible patients had no reported INR test. Why does our report show such a dismal performance rate? If a PT/INR test is reported by “claim,” it is done by the facility not the MD. All tests for Part A patients are bundled in the per diem payment, so no line item lab tests are reported. We, as a group, and our state surveyors have pushed facilities to do point of care PT/INR testing. The economics of having facility nursing staff generate individual claims data for these bedside tests is daunting, and we assume many facilities find this an uneconomical activity to report on an individual claim.
- What is better care? The group was deficient in the use of statins for CAD, but scored 100 percent on that measure in the PQRS system (the PQRS system allows reporting that statins weren’t prescribed for clinical reasons). In other words, the vast majority of our CAD patients have a documented assessment of their suitability for statin therapy, but we were judged as “non-compliant” based on this simple claims data analysis.
- High-risk pain medications: Any time a drug on the Beers list is prescribed counts against you. I understand that in LTPAC settings there are situations where a Beers list drug is the correct therapeutic agent.
- Acute Condition Hospitalization Rate: We were more than one standard deviation (unfavorable) for two of the three acute conditions, not a surprising finding based on the fragility of our population.
A comment on hospitalization/rehospitalization rates: Our hospital region (Memorial Mission, Asheville N.C.), has an all cause rehospitalization rate that is among the lowest in the Southeast, and we have third-party case managers working with our physicians, facilities and hospital to manage rehospitalizations. I’m confident our stats are great compared to any FFS community.
There are strategies we will employ to change our Quality Scores, primarily by using the option to report on PQRS measures for VBP via a Qualified Registry. That way we get to control our measures and the data reporting.
Cost: This is the biggest and most intractable problem. The group is 1.85 Std. Dev. above the mean cost for our benchmark population. This is after an acuity-based cost adjustment that recognized our population was at the 87th acuity percentile.
This file (cost pages from the ECPM 2012 QRUR report) shows the systematic problem. Our group had deviant cost profiles for both hospital costs (our individual cases were more expensive than average) and for post-acute costs. Obviously, if we work exclusively in LTPAC and all of our patients enter the SNF setting directly from a hospital stay, it is impossible to avoid this problem.
In Medicare Advantage plans, CMS created a separate risk category, ISNPs, to recognize the systematically higher cost of the LTPAC population. We need that same recognition for LTPAC physicians.
Without a CMS recognition of our systematically higher costs, LTPAC physicians can never escape the “high cost” tier. That dooms us to failure under VBP. It is already difficult to recruit LTPAC physicians and extenders; why does CMS saddle them with a payment penalty for treating this population?