Mayo Clinic (PDF)

TO:
Anne Krohmer, Project Coordinator
[email protected]
FROM:
James Naessens, Sc.D
Mayo Clinic Division of Health Services Research
Randy Schubring
Mayo Clinic Division of Health Policy and Government Relations
[email protected]
DATE:
May 5, 2015
RE:
State-based risk adjustment model study
We are writing in response to the Minnesota Department of Health’s request for information
on the study of state-based risk adjustment in Minnesota’s small group and individual health
insurance markets. Accurate risk adjustment is crucial to ensuring that plans’ premium
differences are not solely the result of favorable or unfavorable risk selection or choices by high
risk enrollees in the individual and small group markets. Further, accurate risk adjustment must
help ensure enrollee access, preventing plans from cherry-picking enrollees and avoiding more
complex, sicker patients.
We are concerned that the short timeline between now and October 2015 to complete the
study and make recommendations may result in a hurried decision. We are also cognizant that
results of the federal risk adjustment model have yet to be confirmed, and no other state has
fully implemented and put into operation a state-based model.
In Massachusetts—the only state currently pursuing a state-based model—specific concerns
with inaccuracies in the data being extracted from the all payer claims database resulted in
delays in implementation. We believe greater confidence is needed in the capacity of the
Minnesota All Payer Claims Database (APCD) to provide accurate and timely datasets.
Last year’s Minnesota APCD work group recommendations underscored the need to build
confidence in the accuracy and validity of the APCD data inputs with stakeholders. In particular,
the work group recommended that a technical advisory committee be established to ensure
the accuracy and completeness of the data input as well as the construction, design and input
process of the current APCD. This advisory committee has yet to be convened. We believe
greater transparency of APCD inputs and methodologies is needed before stakeholders will
have the confidence in the APCD to serve as the dataset for building a risk adjustment model.
Acknowledging these limitations, we suggest that the Minnesota Department of Health request
an extension of the October 2015 deadline for this study. Specifically, an extended timeline
would greatly expand the information available to the Minnesota Department of Health when
advising the legislature on this important topic. Recommendations on the best route forward
should incorporate learning from both the Massachusetts and federal experiences, and
importantly should incorporate assessment of the integrity - the reliability and validity - of the
APCD. The formation of the APCD Advisory Group is a crucial, and yet to be realized,
component of this process.
Under the provisions of the Affordable Care Act, states retain the option to propose a statebased alternative methodology in the future. If the State does move forward with this study,
we believe that it would be wise to invest the time to carefully research and evaluate options
that recognize the uniqueness of the state, as opposed to working to move forward quickly on a
decision to adopt a state model. Specifically, we suggest that the study focus on the following
issues:

Examine whether the geographic cost factor sufficiently and fairly accounts for
reasonable variation in input prices throughout the state, particularly in the case of
reasonable differences in input pricing across primary, secondary, tertiary and
quaternary care sectors.

Evaluate how the removal of reinsurance and risk corridors will impact the ability of
plans to include major referral centers and quaternary care facilities in their networks,
and assess whether continuation of one of these programs would alleviate narrow
network problems in the state. Without reinsurance or risk corridors, more pressure will
be placed on risk adjustment to accommodate the potential influence of a few extreme
outliers. Many statistical models will not be able to address this issue.

Focus on areas of the state where market failure or limited plan offerings have been
observed, and identify whether a more sophisticated risk adjustment model and risk
transfer formula, or a sophisticated form of reinsurance, could alleviate this problem.

Work to better differentiate levels of disease severity in the risk adjustment model and
risk transfer formula, particularly in the case of multi morbidity and highly complex
disease burden requiring referral services. An examination of diagnosis coding in ICD-10
may be helpful in this effort.
Thank you for the opportunity to comment on the feasibility of a state-based risk adjustment
model. If we can be of further assistance, please don’t hesitate to contact us.