Written Responses from RRT Members (PDF: 123KB/8 pages)

Protecting, maintaining and improving the health of all Minnesotans
Date:
June 1, 2011
To:
Provider Peer Grouping Rapid Response Team members
From:
Katie Burns, Health Economics Program
Subject:
Tools for Clinic-Level Peer Grouping
Thank you for participating in the Rapid Response Team. In preparation for our next
meeting, I wanted to distribute the attached memo. The memo describes tools we are
using to conduct the PPG physician clinic analysis at a clinic level. Specifically, it
describes a measure called the Primary Care Service Index (PCSI) and outlines how
MDH will use the PCSI and its components to facilitate a clinic level of analysis. We are
also proposing to use the PCSI to perform a service mix adjustment and are soliciting
RRT input on whether a need exists for a separate adjustment to account for variations in
service mix or whether the ACG risk adjustment tool can adequately account for this type
of variation to the extent such adjustment is warranted.
We will review the memo during our meeting to ensure you have an opportunity to
clarify your understanding of the issues and to ask questions.
Response deadline: We will need your feedback on these issues by Tuesday, June 14
at 4:00 pm. Responses may be provided via email to [email protected].
MN Council of Health Plans: Sue Knudson
MDH Rapid Response Team
Peer Grouping Methodology
Tools for Clinic-Level Peer Grouping
Thank you for the opportunity to review and provide input on the tools for clinic-level
peer grouping. Our feedback is in two parts. First, regarding the Primary Care
Service Index (PCSI) creation and use to perform clinic level analysis and second, for
use in performing an added service mix adjustment on top of ACG based illness
burden adjustment.
1. PCSI Method
The PCSI method appears to have been thoughtfully constructed and potential
application for consideration is done so with good intent. While it is difficult to
provide specific comments because the source databases are not used by plans, we
are concerned the source data taxonomy for physician specialty identification will be
problematic to the degree of questioning credibility of the results. Specifically, the
databases’ recorded classification and specification do not reflect practicing specialty
and could likely result in inadvertently comingling specialties within a peer group.
For example, an Internal Medicine physician practicing in a subspecialty office may
not be offering primary care to patients and instead is acting as an extender of the
specialist. Further clarification or validation of practicing specialty is needed to
ensure accuracy of the peer group. We recommend the provider registration process
be designed robustly to account for practicing specialty.
2. PCSI for Service Mix Adjustment
We understand MDH’s intent of doing this additional “risk adjustment” step is to
mitigate a concern that multispecialty clinics attract sicker patients. Our review of
the recommended approach leads us to conclude this added step may actually result
in over-fitting the model. By over-fitting, meaningful variation and extra resource
use that increases costs will be explained away when in fact they are a relevant
factor. We see potentially the greatest benefit of adding such an adjustment
accruing to those who may use more specialty services at no added value to
outcome or quality.
The whole notion of proportioning costs among clinics sites is problematic because it
is arbitrary and it renders any information to take action to improve results
impossible or at best complex beyond interpretation for operational changes. We
recommend the unit of analysis remain at the group level as originally designed and
communicated. In our experience, it is always easier to start with an accurate and
less complex model to explain results and build refinements to the unit of analysis
incrementally for provider acceptance.
On the whole, this method is very complex and virtually non-replicable by local plans
who may wish to standardize methods in our community. Given the over-fitting
concern, we recommend the exclusive use of ACG illness burden adjustment.
Further, we continue to recommend appropriate adjustments for other relevant
factors such as excluding quaternary services (burn, trauma, etc.). This is a better
approach to ensuring accurate results based on comparable services among
providers.
Thank you for the opportunity to provide input and suggest alternative methods for
consideration.
MN Medical Association: Janet Silversmith
On behalf of the MMA, I appreciate the opportunity to provide comment on the use of
the department-developed PCSI for purposes of defining “primary care” clinics subject to
total care analysis and for use in service mix adjustment.
The MMA appreciates the department’s efforts to create a methodology to better define
the clinics that will be subject to the total care analysis. The PCSI is a thoughtful
approach, but it is new and untested. As with much of the peer grouping analysis, the
lack of a testing or “beta” phase – where the results could be developed and shared with
hospitals and clinics, and the methods subject to refinement based on feedback – is a
serious limitation. Even the initial results are intended to be used by the state and private
purchasers for high-risk purposes, such as payment reform or network development. The
MMA would appreciate conversations with the department in the near future about
whether the current mandated uses of the data are appropriate, particularly in year one.
With respect to the PCSI, the MMA is concerned about the increased complexity this
approach adds to an already complex methodology. Adequately explaining the approach
to clinics and patients will be a challenge. That said, the development of the PCSI relies
on a couple of sources that include self-reported data for which advance knowledge of
how such data might be used was not provided, so it is not clear how much care was put
into accurate self-reporting. In addition, there may be differences in approach in terms of
how physicians chose to classify themselves, particularly in the NPPES. The MMA
strongly urges the department to clearly inform all clinics subject to the total care
cost analysis the basis for their selection as a primary care clinic – that is, providing
clinics with the names and specialty designations of the physicians and other
providers assigned to that clinic. Reporting this information will allow clinics to
correct any errors in specialty designations or in assigned providers, which is critical to
assuring the provider community of the accuracy of the data. Medica’s recent profiling
program included significant numbers of inaccurate specialty designations that led to
further inaccuracies in terms of quality measure attribution; in fact, one large clinic
system identified a 20% error rate in the specialty designations!
With respect to the question of the application of the PCSI for service mix adjustment
purposes, the MMA appreciates the department’s concerns that the use of ACGs alone
will provide insufficient risk adjustment; we share some of those same concerns. The
MMA appreciates the department’s good-faith and thoughtful efforts in the development
of the PCSI, but the complexity of the approach makes it impossible to know in advance
how well the proposed index will actually perform; it is also not clear – absent any data –
what approaches to risk and service mix adjustments in addition to ACGs would be
desirable and appropriate and to what degree the PCSI meets such needs. Again, systems
of this complexity are usually put to real-world testing and analysis, followed by
methodological revisions and improvements based on those real world analyses. The
MMA urges the department to conduct further analyses and drill-downs on specific
clinic results and provide further information on the results of such analyses to the
RRT to better understand whether application of the PCSI provides reasonable and
useful refinements.
Thanks for your consideration.
MN Hospital Association: Mark Sonneborn
Hi Mark. Thank you for the opportunity to comment on this issue. The [Hospital] team
looked at this information and asked me to respond.
While the PCSI method may introduce some classification issues, the method to decide
what is and is not a primary care clinic and the method to allocate docs to sites seem to
be reasonable.
However, the PCSI cost adjustment did not seem appropriate to us and will lead to
confusing and skewed results. The ACG adjustment is specifically designed and
validated to account for patient illness burden differences. Applying the PCSI cost
adjustment builds in an inherent bias toward specialists. If risk adjustment using ACGs
accounts for expected cost differences based on patient care needs don’t we want to
compare different clinic structures on an apples to apples basis?
If specialist dominated practices use more resources to manage similarly ill patients we
should know that and published results should reflect that. If specialists are not
adequately coding comorbidities to reflect the actual illness of their patients, they
should be encouraged to code more completely for both quality and accuracy purposes.
If specialist provided care quality is better we should seek to identify that through
quality measures but even if care quality was better in those sites, we should not
assume that results in higher costs. For complex chronically ill patients, higher quality
care should theoretically result in the reduction of complications, ER visits and admits,
actually leading to reductions in total cost. Specialists often contend that their special
expertise can lower cost through more appropriate use of imaging/labs/surgeries, etc.
so we shouldn’t assume that their costs would be greater.
None of us were aware of other local or national comparisons of care cost that have
taken this added step, nor are we aware of any previous studies validating this
approach. What seems certain however, is if used, this extra step will make it
impossible to compare CMS, plan and provider analyses to PPG results. Adding this
additional complexity to the analysis will be likely to impede the important process of
gaining provider acceptance of PPG results so we can move past the “arguing about the
data” phase and onto using it for improvement .
Thanks for your attention to this and please let me or [Hospital] know if you have any
questions.
[consultant]
MN Business Partnership: Beth McMullen
No response submitted.
AARP: Michelle Kimball
No response submitted.
DHS: Marie Zimmerman
No response submitted.