Public Comments

Status: Original and derived artifacts are available for this document.

Download: pdf

Andrew G. Huff, Ph.D., M.S.
Risk Factor Research, LLC, Michigan, USA
[email protected]

October 13, 2026

Centers for Medicare & Medicaid Services, Department of Health and Human Services
Docket CMS-2026-2773 (document CMS-10346 (OMB control number 0938-1129))
Federal Register document 2026-16455 (91 FR 52303, August 13, 2026)
Submitted via Electronically via www.regulations.gov (docket CMS-2026-2773), or by regular
mail to CMS, Office of Strategic Operations and Regulatory Affairs, Division of Regulations
Development, Attention: Document Identifier/OMB Control Number CMS-10346 / 0938-1129,
Room C4-26-05, 7500 Security Boulevard, Baltimore, Maryland 21244-1850; comments due
October 13, 2026
Re: Agency Information Collection Activities: Proposed Collection; Comment Request.
Quality Bonus Payment Appeals, Form CMS-10346, OMB control number 0938-1129,
extension of a currently approved collection. Docket CMS-2026-2773; FR Doc. 2026-16455;
91 FR 52303 (August 13, 2026).
I write in my individual capacity as an epidemiologist and former population health executive
who has designed, operated, and been held accountable to quality measurement systems. This
comment addresses only the information collection: whether it has practical utility, whether the
burden estimate is accurate, how the quality and clarity of the information could be improved,
and how burden could be reduced. It takes no position on Star Ratings policy. The notice
describes the collection as an extension without change. I ask CMS to use this extension to make
the collection more useful to the agency itself, at little or no added cost to respondents.
Summary of recommendations
1. Practical utility: report annually, in aggregate and without identifying plans, the number of
reconsideration requests and hearing requests filed, the measures appealed, the share upheld at
each stage, and the dollar amounts at stake, as the vital signs of the Star Ratings measurement
system.
2. Burden accuracy: state in the supporting statement the basis for 120 respondents, 20 annual
responses, and 160 hours, because those three figures cannot be reconciled on the face of the
notice.
3. Quality and clarity: require each appeal to identify, in structured fields, the measure, the
specific data element disputed, the asserted error type (calculation error or incorrect data), and
the correction sought, so that CMS can aggregate root causes across years.
4. Minimizing burden: accept appeals electronically through the Health Plan Management
System and publish a decision log with reasons, which would obviate repeat appeals on
questions CMS has already settled.
The case for the collection as it stands
The collection is necessary and the design is sound. Sections 1853(n) and 1853(o) of the Social
Security Act tie Quality Bonus Payments to a 4-star threshold, and section 1854(b)(1)(C) ties
beneficiary rebates to the same rating level; a plan that is rated incorrectly loses money it earned
1

and its enrollees lose rebates they were owed. The two-step review at 42 CFR 422.260(c), a
reconsideration followed by an informal hearing on the record, is proportionate. Limiting the
permitted bases to calculation error and incorrect data under 422.260(c)(3)(ii) keeps the process
from becoming a venue for relitigating measure design. Releasing preliminary ratings each
November so plans can check them before they become final is good practice. Nothing below
argues against any of that.
Practical utility: the appeals record is a validity instrument
Under 44 U.S.C. 3506(c)(2)(A) and 5 CFR 1320.8, the agency must show that the information
collected has practical utility. The notice describes the form's utility as adjudicating individual
appeals, which it plainly has. It has a second utility that the notice does not mention and that
CMS should claim. Every Star Ratings measure is an instrument: it takes raw claims, survey, and
administrative data and produces a score that is then used to make a financial decision. The
appeals process is the only place where that instrument is systematically tested against a
motivated, well-informed adversary with the data in hand. An appeal that succeeds on
calculation error or incorrect data is direct, probative evidence that the instrument misfired for
that plan in that year. The rate at which appeals succeed, by measure, is therefore a
measurement-validity signal that no other part of the program produces.
Today that signal is collected and then, as far as a member of the public can tell, not reported. I
ask CMS to publish annually, in aggregate and without identifying plans, the number of
reconsideration and hearing requests, the measures appealed, the share upheld at each stage, and
the dollar amounts at stake. These are the vital signs of the measurement system in the same
sense that the CDC's surveillance evaluation attributes are the vital signs of a surveillance
system. A measure that draws repeated successful appeals needs attention to its data pipeline; a
measure that draws none is either working or not worth appealing, and the two cases are
distinguishable by the dollars involved. Reporting this costs respondents nothing, because the
data already exist inside the collection.
Burden accuracy
The burden table reports 120 respondents, 20 total annual responses, and 160 total annual hours.
These figures cannot be reconciled on their face. If 120 organizations each file once a year,
responses should be 120, not 20. If the agency means that roughly 20 appeals arrive each year
from a universe of about 120 eligible organizations, the respondent count is a population and not
a count of filers, and the table should say so. The implied 8 hours per response is plausible for a
reconsideration request that identifies a measure and describes the issue, but it seems low if the
same collection also covers hearing requests and the written evidence that accompanies them. I
ask CMS to state in the supporting statement what each figure counts, whether hearing requests
are included, and what the historical annual filing counts have been for the last three years, so
that commenters and OMB can judge the estimate against observed data rather than against each
other.
Quality and clarity of the information collected
The form asks organizations to identify the measure they believe was miscalculated or based on
incorrect data and to describe the issue. A free-text description serves the individual case but
defeats aggregation. I recommend adding structured fields: the measure identifier; the specific
2

data element or intermediate value disputed; the asserted error type, which the regulation already
limits to two; and the correction sought, expressed as the value the plan believes is correct.
Structured appeals let the reconsideration official see immediately what is in dispute, and they let
CMS aggregate root causes across plans and years, which is the only way to distinguish a one-off
data feed failure from a measure specification that is being read two ways. This is a modest
change in form design, not in scope, and it makes the collection more perspicuous to everyone
who touches it.
Minimizing burden through information technology
Two steps would reduce burden on both sides. First, accept appeals electronically through the
Health Plan Management System, which every MA organization already uses, with the
structured fields above as the intake form, rather than through a document that must be drafted,
transmitted, and re-keyed. Second, publish a running log of reconsideration and hearing
decisions with reasons, redacted of plan identity where necessary. A published decision log
attenuates the single largest avoidable source of burden in any appeals system, which is repeat
appeals on questions the agency has already resolved. A plan that can see that an identical dataelement argument was rejected last year, and why, will not spend eight hours filing it again, and
the reconsideration official will not spend time rejecting it again. The epistemic gain is shared:
the agency learns which rulings plans find unclear, and plans learn the standard they are being
held to.
Scope and confidence
I did not address the substance of the Star Ratings methodology, the 4-star threshold, or the
limits on appealable grounds in 422.260(c)(3); those are policy questions outside a PRA notice.
My confidence that the burden figures as published cannot be reconciled without an explanation
is high; my confidence that an annual aggregate report would be valuable to CMS is high and
rests on general measurement principles rather than on knowledge of CMS's internal data; my
confidence that HPMS is the right electronic channel is moderate, and CMS should substitute
whatever system is already used for Star Ratings data exchange if HPMS is not.
Disclosure: I have no financial interest in any vendor or contract involving this program, and I
was not compensated for this comment.
Thank you for the opportunity to comment.
Respectfully,
Andrew G. Huff, Ph.D., M.S.
Risk Factor Research, LLC
References
1. Centers for Medicare & Medicaid Services, Agency Information Collection Activities:
Proposed Collection; Comment Request (CMS-10346 and CMS-10453), 91 FR 52303, FR Doc.
2026-16455, August 13, 2026. https://www.federalregister.gov/documents/2026/08/13/202616455/agency-information-collection-activities-proposed-collection-comment-request

3

2. 42 CFR 422.260, Appeals of quality bonus payment determinations.
https://www.law.cornell.edu/cfr/text/42/422.260
3. 44 U.S.C. 3506(c)(2)(A), Federal agency responsibilities (Paperwork Reduction Act).
https://www.law.cornell.edu/uscode/text/44/3506
4. 5 CFR 1320.8, Agency collection of information responsibilities.
https://www.law.cornell.edu/cfr/text/5/1320.8

4