Issue Description to RRT Members (PDF: 125KB/13 pages)

 Protecting, maintaining and improving the health of all Minnesotans Date:
June 14, 2010
To:
Provider Peer Grouping Rapid Response Team members
From:
Katie Burns, Health Economics Program
Subject:
Second set of issues for your consideration
Thank you for participating in the Rapid Response Team. In preparation for our second
meeting, I wanted to distribute the attached memos from Mathematica Policy Research,
Inc. They describe the next set of issues for which we would like your input:
1) Should the peer grouping analysis include people without any medical claims in a
given calendar year? If so, what method should be used to attribute those patients to
providers?
2) What specific attribution rules should be used to attribute patients to physician clinics
or medical groups?
We will walk through the memos at our meeting tomorrow afternoon 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 Friday, June 25th
at 4:00 pm. Responses may be provided via email to [email protected].
Thanks very much.
MEMORANDUM
P.O. Box 2393
Princeton, NJ 08543-2393
Telephone (609) 799-3535
Fax (609) 799-0005
www.mathematica-mpr.com
TO:
Katie Burns
FROM:
Aparajita Zutshi, Eric Schone, Mary Laschober, Randy Brown
SUBJECT:
Recommendations for Attribution of Non-users to
Clinics/Groups for Total Care Cost and Quality Measures
DATE: 6/13/2010
Purpose of Memo
This memo presents the Mathematica team’s recommendations for attribution of patients
without any medical claims in a given calendar year— (“non-users”) to provider entities (PEs)
such as clinics and medical groups for purposes of reporting on their total care cost and quality
measures. Attributing non-users can be important, because ignoring them may penalize providers
who are conservative about office visits and those whose proficient care reduces the likelihood
that care will be needed in the subsequent year. Also, failing to include non-users in the
denominator of some preventive care measures may lead to overestimates of a provider’s
performance. On the other hand, including non-users can introduce biases if it is difficult to
identify the appropriate provider(s) to which such patients should be attributed. Below, we
discuss three options for attributing non-users for claims-based cost and quality measures, but
ultimately recommend that non-users not be attributed to any PE or included in the peer grouping
reports. We also discuss the Advisory Group’s recommendation of using up to three years of
historical claims data for attributing non-users and prorating non-users across all providers until
historical data become available and present our reasons for departing from this
recommendation.
The first section of this memo describes the options, the second section discusses how they
perform in relation to various goals of the peer grouping analysis and performance reporting, and
the third section contains our recommendation.
Options for Attributing Non-users
Below we describe three options for attributing non-users to PEs – 1) never attribute nonusers, 2) attribute non-users for claims-based cost and quality measures using historical claims
data, and 3) attribute non-users for claims-based cost and quality measures using historical
claims data and, until such data become available, prorate non-users across all providers.
Option 1: Never Attribute Non-Users
Under this option, we would not attribute non-users to PEs and they would be excluded
from all rounds of peer grouping analysis and performance reporting.
An Affirmative Action/Equal Opportunity Employer
MEMO TO: Katie Burns
FROM:
Aparajita Zutshi, Eric Schone, Mary Laschober, Randy Brown
DATE:
6/13/2010
PAGE:
2
Option 2: Attribute Non-Users for Cost and Quality Measures Using Historical Claims Data
Under this option, we would attribute non-users to PEs using one year of historical claims
data, (for example, for 2009 non-users with 2008 claims data, we would use 2008 claims data to
attribute 2009 non-users). Because Medicare claims will only be available for 2008, for the
reports to be produced in 2010, in our estimates for calendar year 2008 we would not be able to
use this method to attribute non-users among Medicare beneficiaries. Therefore, for the sake of
consistency across all payers, we would exclude non-users from the first round of the peer
grouping analysis. We would be able to implement Option 2 in the second round of peer
grouping.
Under this option, once historical claims data become available, we would use the multipleproportional rule to attribute non-users in order to be consistent with our recommendation of the
multiple-proportional rule for attribution of patients with care utilization as proposed in our
previous memo dated May 26, 2010. We would include non-users in the calculation of cost
measures and selected claims-based quality measures for which non-users are eligible. Examples
of quality measures include rates of hospital admissions and ER visits for certain conditions, for
example, rate of hospital admissions for asthma. Measures that are conditional on an inpatient
hospitalization, such as rates of re-admission or 30-day mortality post-discharge, would by
definition still exclude non-users.
Option 3: Attribute Non-Users for Cost and Quality Measures Using Prorating in the Short-term
and Using Historical Claims when Data become Available
Under this option we would prorate non-users across all PEs for the first round (and
potentially future rounds) of the peer grouping analysis until historical claims data become
available, following which we would attribute non-users based on care provided historically.
For prorating non-users across all PEs, we propose using a consumer choice model to
predict from which providers a non-user is most likely to seek care. We would develop this
model using patients who have positive claims during the model year. Once we have attributed
these patients to clinics/groups, we would run a conditional logit (choice) model to capture the
relationship between the choice of provider(s) and various patient and provider characteristics.
The choice set would be a set of clinics/groups defined by criteria such as distance from the
attributed patient’s zip code (a radius approach). The characteristics determining this choice
might include patient demographic characteristics such as age and gender, distance from the
patient’s zip code to the clinic’s zip code, number of a clinic’s patients that year with a given
type of insurance, and number of a clinic’s patients in different age groups. After estimating the
choice model, we would use the estimated coefficients to predict from which clinics/groups nonusers would most likely seek care.
MEMO TO: Katie Burns
FROM:
Aparajita Zutshi, Eric Schone, Mary Laschober, Randy Brown
DATE:
6/13/2010
PAGE:
3
Under this option, once historical claims data become available, we would use the multipleproportional rule to attribute cost and quality measures to PEs, if that this is the allocation rule
agreed upon for peer grouping in general.
Comparison of Options for Attributing Non-Users
Below we describe how the three options for attributing non-users perform in relation to
various goals of the peer grouping analysis. Table 1 summarizes the pros and cons of each
option.
Table 1. Tradeoffs among Options for Attributing Non-Users
Option
Face validity
Potential for
Immediate
implementation
Option 1
Populationwide
measurement
of costs and
quality
none
+++
+++
Option 2
+
++
+
Option 3
+
+
++
Note: ‘+’ signifies strength of criterion.
Population-wide measurement of costs and quality: The total care measure for a provider
is intended to capture the value a provider delivers to his/her patient population over a period of
time. For this reason, the Advisory Committee recommends including the cost and quality of
care provided to non-users in the total care measure to enable population-wide measurement of a
provider’s performance, including potential patients. By including non-users in both cost and
quality measures, both Options 2 and 3 enable population-wide comparisons among providers,
with Option 3 enabling such comparisons in the very first round of the peer grouping analysis.
By excluding non-users from the total care measure, Option 1 only supports provider
comparisons based on a subset of the provider’s potential patient population; that is, those
patients who the provider actually treated over the measurement period. We note, however, that
even though Options 2 and 3 are more inclusive than Option 1, they still fail to include all
segments of a provider’s potential patient population; for instance, they fail to include uninsured
patients who are non-users. Also, since some nonusers in a given year have gone several years
without using services. Option 2 will only capture nonusers in 2009 who used some services in
2008, since 2008 is the first year for which data are available.
Potential for bias: The cost and quality measures calculated for a provider are unbiased
estimates of the true cost and quality of care that the provider delivers to its patients if the
measures are based on the provider’s true patient population (or a random sample of the
population). By using only current year claims filed by a provider to attribute patients to the
MEMO TO: Katie Burns
FROM:
Aparajita Zutshi, Eric Schone, Mary Laschober, Randy Brown
DATE:
6/13/2010
PAGE:
4
provider, we exclude from the calculation of cost and quality measures both insured and
uninsured patients who have zero claims in the current year, and thereby introduce two potential
sources of bias in the provider’s estimated measures. While the direction of the bias created by
excluding the uninsured is uncertain, exclusion of non-users would cause a provider’s average
costs and adverse event rates (for instance, hospital avoidance rates) to be overestimated. By
using historical claims data (or the prorating method) to include non-users we can eliminate this
bias in the estimated cost and quality measures. 1
At the same time, the methods used for including non-users in provider peer grouping can
introduce biases of their own. Attribution based on historical claims is based on the assumption
that if the non-user were to seek care in the current year he/she would visit the same group of
providers to the same degree as the previous year and hence these providers should be attributed
his current year’s cost and quality measures in proportion to his previous year’s E&M visits. If
the patient were to switch one or more of his previous year’s providers in the current year then
by attributing non-users based on historical care we would be underestimating these providers’
cost and hospital avoidance measures. This kind of downward bias would be especially large for
providers that delivered poor care and for providers that delivered incidental care such as urgent
care centers.
Attribution based on the prorating method is based on the assumption that the control
variables used in the prediction model completely capture all differences between users and nonusers so that non-users can be attributed based on the characteristics of users. However, it is very
likely that non-users differ from users in unmeasured ways that cannot be captured by the model
such as the propensity to use certain providers. In this case, including non-users would bias
downward the cost and hospital avoidance measures of the providers that delivered care to users
relative to the providers actually managing non-users.
Since Option 3 bases attribution of non-users on the prorating method in the first round and
the historical claims method in future rounds, it can cause providers’ rankings to change over
time in ways that are unrelated to actual changes in the providers’ relative performance over time
due to the different types of biases that can be introduced by the two methods. This false
variation in providers’ rankings over time will be especially problematic when providers’
payments are based on a tiered-pricing method as envisioned in the MN health care reform law.
Overall, there are biases resulting from both exclusion and inclusion of non-users that can
cause performance variation between a provider and his/her peers that is unrelated to true
1
The claims-based approach to peer grouping leaves no scope for including uninsured members of the
provider’s potential population in quality or cost measures.
MEMO TO: Katie Burns
FROM:
Aparajita Zutshi, Eric Schone, Mary Laschober, Randy Brown
DATE:
6/13/2010
PAGE:
5
differences in their costs and quality of care. Since the primary objective of peer grouping is to
compare providers on the quality and efficiency of care the care they actually deliver to patients,
it seems best to exclude non-users. Under this conceptualization of the objective and measures,
non-users would not be considered a part of any provider’s relevant population and their
exclusion will not bias the provider’s cost and quality measures. Under this goal, Option 1 would
result in unbiased provider comparisons while Options 2 and 3 would introduce increasing
degrees of bias in provider comparisons.
Face validity: Including non-users in a provider’s cost and quality measures is likely to
have less face validity than not including non-users irrespective of which method is used to
attribute non-users. It may be difficult to convince providers why they are being attributed a
patient’s costs and quality of care that they did not treat in the current year. Providers may be
wary of the assumptions underlying the methods for attributing non-users: that is, they may not
agree that a patient’s historical pattern of visits (Options 2 and 3) is a credible method for
identifying his current responsibility for those patients; or that non-users would choose the same
providers as users with the same general characteristics (Option 3’s prorating method). Inclusion
of non-users will be confusing at best and highly controversial at worst. A methodology that is
difficult to explain or controversial is less likely than alternatives to achieve provider acceptance.
Face validity of attribution will be crucial when rankings from the peer grouping analysis are
used to pay providers.
Immediate implementation: Option 1 can be implemented immediately; option 2 requires
data from a prior year, which is not available for the first peer grouping analysis and reporting.
However, reporting and peer grouping for the second and subsequent years could implement
option 2.
Recommendations for Attributing Non-Users
Based on the comparison of different options, we recommend Option 1 - excluding nonusers from current and future rounds of peer grouping and performance reporting. If MDH feels
that it is important to use the total care measure as a population-wide measure, we recommend
Option 2, that is using only historical claims data for attributing non-users’ cost and quality of
care, and doing so only in future years.
cc: The project team
MEMORANDUM
P.O. Box 2393
Princeton, NJ 08543-2393
Telephone (609) 799-3535
Fax (609) 799-0005
www.mathematica-mpr.com
TO:
Katie Burns
FROM:
Aparajita Zutshi, Eric Schone, Randy Brown, and Mary Laschober DATE: 6/13/2010
SUBJECT:
Recommendations for Attribution of Total Care and Chronic
Condition Cost Measures to Provider Clinics/Groups
Purpose of Memo
This memo presents recommendations for attributing patients to provider entities (PEs) such
as clinics or medical groups for purposes of reporting on their annual cost measures for both total
care (that is, all patients) and care for people with targeted chronic conditions. Please note that
for purposes of the physician peer grouping analysis, patients will ultimately be attributed to
physician clinics or medical groups rather than to individual physicians. 1 Our recommendations
are guided by lessons learned from our CMS Resource Use Report (RUR) project, and by the
twin goals of the attribution task – to maximize sensitivity and specificity. Sensitivity, for this
purpose, is defined as the probability of detecting a physician’s role in a patient’s care when it is
present. Specificity is defined as the probability of not attributing a patient’s care to a physician
who was not involved in that care. Maximizing sensitivity increases the likelihood that providers
are held accountable for care decisions that were under their control. Maximizing specificity
increases the likelihood that providers are not incorrectly held accountable for care decisions that
were not under their control.
Based on these two criteria, we recommend the multiple-proportional rule for attributing
patients to PEs. We also recommend using the percentage of Evaluation and Management
(E&M) visits to identify the PEs involved in a patient’s care and their level of involvement in
that care over the year. While it may not be the case that a provider’s share of E&M visits is the
best measure of the extent to which that provider was responsible for a given patient’s care,
under the assumption that it is the best indicator of the relative contribution of a provider to a
patient’s overall cost and quality outcomes for the year, the multiple-proportional rule is
perfectly sensitive and specific. As we show below, the rule assigns some responsibility to a
provider for every patient it touches (sensitivity) and a share of responsibility proportional to its
level of involvement (as measured by E&M visits) in the patient’s care (specificity).
1
The attribution approaches recommended here apply only to peer grouping of physician practices, not to peer
grouping of hospitals. All hospital stays will be attributed solely to the hospital where the admission occurred.
An Affirmative Action/Equal Opportunity Employer
MEMO TO: Katie Burns
FROM:
Aparajita Zutshi, Eric Schone, Randy Brown, and Mary Laschober
DATE:
6/13/2010
PAGE:
2
We acknowledge the Advisory Group’s recommendation to use attribution rules that support
credible attribution (that is, maximize specificity), rather than rules that simply maximize the
number of patients attributed to a provider (that is, maximize sensitivity). We also note the
committee’s preference for attributing patients to a single provider. In this memo, we describe
and contrast three rules - the plurality-minimum, multiple-even, and multiple-proportional rules,
show how the multiple-proportional rule improves upon both specificity and sensitivity
compared to the other rules, and present our reasons for departing from the committee’s
recommendation for attribution to a single provider. 2
Definitions of Attribution Rules
Plurality-Minimum Rule
This rule attributes sole responsibility for a patient’s care to the provider who billed for the
greatest number of E&M visits for the patient over the period of measurement, with the added
criterion that the provider billed for at least 30 percent of annual E&M costs for that patient.3 For
example, if Providers A, B and C billed for 20, 50, and 30 percent of E&M visits for the patient
over the year, respectively, we would exclusively assign Provider B the total annual costs for that
patient as long as Provider B billed for at least 30 percent of the annual E&M costs for that
patient.
Multiple-Even Rule
This rule attributes equal responsibility for a patient’s care to all providers who billed for
any E&M visits for the patient over the measurement period, with the added criterion that the
provider(s) billed for at least 30 percent of the annual E&M costs for the patient. In the example
above, we would assign all three providers the total annual costs for that patient, as long as each
provider billed for at least 30 percent of the annual E&M costs for that patient.
2
Our reason for choosing these rules as points of comparison is because these were the most common single
and multiple attribution rules we discovered during an environmental scan of attribution rules we conducted for the
CMS RUR project.
3
The reason for choosing a 30 percent minimum cost threshold is that it was the most common cost threshold
used across single and multiple attribution rules in an environmental scan of attribution rules we conducted for the
CMS RUR project. Instead of E&M costs we can also use E&M visits as a minimum threshold.
MEMO TO: Katie Burns
FROM:
Aparajita Zutshi, Eric Schone, Randy Brown, and Mary Laschober
DATE:
6/13/2010
PAGE:
3
Multiple-Proportional Rule
This rule attributes responsibility to all providers that billed for any E&M visits for the
patient over the measurement period, in proportion to the percentage of the patient’s E&M visits
for which they billed. Unlike the previous rules, this rule does not require the provider to fulfill a
minimum E&M cost (or visit) criterion since the provider is only being held accountable for the
patient’s costs and quality of care in proportion to the amount of the provider’s involvement (as
measured by E&M visits) in the patient’s care. The purpose of a minimum cost threshold is to
enhance the specificity of rules that attribute all of the patient’s total care costs to a single or
multiple providers and is therefore largely unnecessary for the multiple-proportional approach.
Table 1 shows how this rule attributes a patient’s total care costs of $1,000 across three
different clinics that the patient visited over the course of a year.
Table 1: Application of the multiple-proportional rule to total care costs
Proportion of annual E&M visits at each clinic
Costs attributed to each clinic ($)
0.2
200
0.3
300
0.5
500
Rationale for Recommending the Multiple-Proportional Rule
Below we describe how the three rules perform on key goals of the attribution task, on
general goals of a peer grouping effort, and on specific goals of the present effort, and argue
why we recommend the multiple-proportional rule over other attribution approaches for the
present effort.
Specificity: Because the multiple-proportional rule attributes responsibility in proportion
to a given provider’s share of contacts with a patient, it is more specific than pluralityminimum since plurality is more likely to give providers exclusive responsibility for patients
for whom the provider is not necessarily the primary manager and ignore their role in
secondary management of (and collaboration concerning) other patients. It is also more
specific that the multiple-even rule since the multiple-even rule does not distinguish between
providers with different levels of interaction with the same patient.
The greater specificity of the multiple-proportional rule becomes even more pertinent in
the context of using annual per capita costs as the cost measure for peer grouping
MEMO TO: Katie Burns
FROM:
Aparajita Zutshi, Eric Schone, Randy Brown, and Mary Laschober
DATE:
6/13/2010
PAGE:
4
comparisons for treatment of patients with particular chronic diseases. 4 Since all providers
the patient saw over the course of a year are eligible for attribution of the patient’s annual
total costs, the plurality approach can give rise to situations where a clinic that started seeing
the patient for the first time in the second half of the year could potentially be held solely
responsible for the patient’s costs incurred during the first half of the year as well. For
instance, if the patient has CHF and his/her first claim for CHF care is during the second half,
but he/she was being treated for cancer at the beginning of the year, under plurality
attribution, the cardiology clinic treating CHF could get attributed all of the cancer costs as
well, despite having had no involvement in this care. 5 Under multiple-proportional
attribution, both the cancer providers and the cardiology clinic would get a share of total
annual costs, proportionate to the number of E and M visits.
Sensitivity: Because the multiple-proportional rule assigns some portion of a patient’s
cost for all patients seen by the provider to the provider over the measurement period, it
maximizes the number of patients that can be attributed to a provider. As a result, this rule
allows for the broadest, and hence the most statistically precise, assessment of a given
provider’s efficiency. Using this rule will potentially minimize concerns regarding the
reliability of provider rankings by increasing case sizes compared to single attribution rules.
Another positive feature of the multiple-proportional rule is that it will enable reporting on
the cost of specialty groups and specialists who typically do not get attributed any (or
enough) patients under single attribution rules based on E&M visits.
Face validity: The potential of the multiple-proportional rule in performing better on both
sensitivity and specificity compared to the other rules should increase the face validity of this
rule for providers. This was confirmed during formative testing in the CMS RUR project where
providers preferred the multiple-proportional rule above other single and multiple attribution
approaches.
Applicability to different patient populations: Attribution is most difficult for patients
with multiple chronic illnesses, who often receive care from a large number of providers.
Compared to other rules, the multiple-proportional rule is best suited to assigning the appropriate
degree of responsibility to the relevant providers for this group of patients. This rule will be most
useful for elderly Medicare patients, but we expect that the multiple-proportional rule will
perform equally well in capturing care relationships for all patient populations of interest.
4
5
For details see MDH’s memo to the Rapid Response Team on this issue, dated 5/24/10.
The multiple-even rule would also be unfair to clinics since it would assign all of the patient’s costs to all
clinics involved in the patient’s annual care, irrespective of the degree and timing of involvement in the care.
MEMO TO: Katie Burns
FROM:
Aparajita Zutshi, Eric Schone, Randy Brown, and Mary Laschober
DATE:
6/13/2010
PAGE:
5
Possible Perverse Incentives: We must also consider whether the proposed attribution
methodology is likely to create more perverse incentives than other approaches. For example,
one might think that providers would not want to participate in the care of an expensive patient,
because that patient would increase their average cost. This would potentially be true under any
attribution approach, but for providers who are not the primary provider for a given patient, this
incentive could be especially pronounced under multiple attribution, because the practice knows
it would be attributed some of the cost for that patient.
This potential incentive to avoid high cost patients is offset by two factors, however. First,
risk adjustment will control for the fact that some patients are expected to be more expensive
than others. The provider will only be adversely affected by having a high resource use patient if
the provider uses more resources in treating that patient than other patients with the same
conditions and characteristics. Second, a resource-intensive patient, if treated efficiently given
his or her diagnoses and comorbidities, actually could be highly beneficial to the provider. Thus,
while it might seem that multiple attribution could create some perverse incentives for providers
to treat only healthy patients, a clear understanding of how peer grouping works should help
clarify that no such perverse incentives are present.
Care coordination: One drawback of the multiple-proportional rule (or any multiple
attribution approach) is that it does not concentrate responsibility for a patient’s care on a single
PE, potentially weakening the lead provider’s motivation to influence the care decisions of other
physicians treating the same patient. As a result, the multiple-proportional rule may be less
effective in encouraging care coordination—a serious need particularly among older patients
with multiple chronic conditions—than a rule that assigns responsibility to a single PE.
The problem with the argument for sole attribution leading to stronger incentives to manage
patients well and coordinate their care across providers is that this approach gives the secondary
providers no incentive to participate in care coordination. On the other hand, by attributing some
portion of the total cost of a given patient to every provider who treated the patient, the multiple
attribution approach gives even providers with limited involvement in a patient’s care an
incentive to cooperate and coordinate with other providers involved in that patient’s care.
Simplicity: Another possible drawback of the proposed approach is that, of all three rules,
the multiple-proportional rule may be the most difficult to explain to providers given its lack of
familiarity. According to the RUR environmental scan, plurality and multiple-even rules are
more common in public and private peer grouping and performance reporting efforts.
Nonetheless, the RUR project also found that it was difficult for providers to grasp that all costs
a patient incurred—not just those resulting from the attributed provider’s direct care—were
included in the total cost measure under a plurality rule. We believe that with careful attempts at
understanding providers’ concerns and engaging in provider education, we will be able to
harness the advantages of this rule to achieve both efficient peer grouping and provider
acceptance.
MEMO TO: Katie Burns
FROM:
Aparajita Zutshi, Eric Schone, Randy Brown, and Mary Laschober
DATE:
6/13/2010
PAGE:
6
Other Recommendations
We propose the following recommendations regarding other aspects of the attribution task.
Attribution Based on all E&M visits: We will use all E&M visits that a patient made over
the course of a year to determine the PEs to which the patient should be attributed. While all
types of encounters contribute to costs, we favor the use of E&M visits, which represent a faceto-face provider/patient interaction, an approach also commonly used by public and commercial
peer grouping efforts such as MedPAC, CMS’s Physician Group Practice Demonstration, and
Aetna. We further propose the use of all E&M visits over alternatives such as using only
ambulatory E&M visits and using E&M visits instead of E&M costs.
Using all E&M visits instead of only ambulatory E&M visits recognizes and accounts for
the possibility that the provider had an opportunity to affect the care of his patient irrespective of
care setting. While providers may feel that this approach lacks a certain amount of face validity
because certain E&M services, particularly emergency department visits, are not strong
indicators of a provider’s ability to manage and coordinate the patient’s care, this concern
becomes less of an issue when using a multiple-proportional approach. No one provider is solely
responsible for managing a patient’s care and each provider is only attributed a share of the
patient’s care commensurate with its involvement.
Using E&M visits for attribution better captures the patient-provider relationship than E&M
costs, where the latter may reflect unintentional distortions in the established relative value of
different types of E&M services. For instance, different stages of E&M visits such as new patient
visits vs. established patient visits vs. follow-up visits are typically linked to different payment
rates and it is not entirely clear that the payment rates correlate with the amount of decisionmaking involved in these visits. 6
Attribution to Managed Care PEs: We will attribute managed care patients using the
multiple-proportional rule described above. An alternative to this approach is to “attribute” these
patients to the clinic or group that is their designated primary care provider. The primary
advantage of the second approach is that it may be more familiar to physicians that have
managed care patients. The problem with the second approach is that it creates inconsistency in
the rules used to attribute managed care versus fee-for-service patients, which may cause
6
At the same time, we believe that both E&M visits and E&M costs can be used to create minimum thresholds
for rules such as plurality-minimum and multiple-even since both can be useful in measuring whether the PE(s)
being given full responsibility for the patient’s total care cost had at least a minimum role in managing the patient’s
care over the year.
MEMO TO: Katie Burns
FROM:
Aparajita Zutshi, Eric Schone, Randy Brown, and Mary Laschober
DATE:
6/13/2010
PAGE:
7
provider rankings to vary not due to real differences in cost-effectiveness of care but due to
differences in methodology. Another problem with the second approach is that managed care
patients would rarely end up being attributed to specialists, which would result in specialists not
being rated for the costs of their managed care patients. While we recommend the first approach,
we will work with MDH to assess the relative pros and cons of the two approaches and arrive at
the best decision for the state.
cc: The project team