2019 COST Measure #001: Medicare Spending Per Beneficiary (MSPB)

Measure Description

The Medicare Spending Per Beneficiary (MSPB) clinician measure assesses the cost to Medicare as a result of services performed by an individual clinician during an MSPB episode, which comprises the period immediately prior to, during, and following a patient’s hospital stay. An MSPB episode includes all Medicare Part A and Part B claims falling in the episode “window,” specifically claims with a start date between 3 days prior to a hospital admission (also known as the “index admission” for the episode) through 30 days after hospital discharge.
The MSPB measure is attributed to individual clinicians, as identified by their unique Medicare Taxpayer Identification Number/National Provider Identifier (TIN-NPI). MSPB measure performance may be reported at either the clinician (TIN-NPI) or the clinician group (TIN) level.

Rationale

Medicare is transforming from a system that rewards volume of service to one that rewards effective care and reduces delivery system fragmentation. To advance this transformation, the Centers for Medicare & Medicaid Services (CMS) provides financial incentives to clinicians based on their performance on selected quality measures as well as cost measures.
The MSPB clinician measure, a cost measure, was originally developed for use in the Physician Value-Based Modifier program, and was updated (see Section 4.0 for details) for inclusion in the Merit-based Incentive Payment System (MIPS) cost performance category.

Overview of Measure Calculation

Using Medicare Part A and B claims data, with certain exclusions attached, the MSPB measure compares observed and expected episode costs, as follows.

Measure Numerator:
The numerator for the measure is the sum of the ratio of payment-standardized observed to expected MSPB episode costs for all MSPB episodes for the TIN-NPI or TIN. The sum of the ratios is then multiplied by the national average payment-standardized observed episode cost, to convert the ratio to a dollar amount.

Measure Denominator:
The denominator for the MSPB measure is the total number of MSPB episodes for the TIN-NPI or TIN.

Eligibility and Exclusion Criteria:
The MSPB measure assesses costs during episodes of care initiated by acute inpatient hospital stays. Episodes for a beneficiary are excluded from the MSPB measure if they meet any of the following conditions:

• the beneficiary was not continuously enrolled in both Medicare Parts A and B from 93 days prior to the index admission through 30 days after discharge
• the beneficiary’s death occurred during the episode
• the beneficiary is enrolled in a Medicare Advantage plan or Medicare is the secondary payer at any time during the episode window or 90-day lookback period
• the index admission for the episode did not occur in a subsection (d) hospital paid under the Inpatient Prospective Payment System (IPPS) or an acute hospital in Maryland
• the discharge of the index admission occurred in the last 30 days of the performance period
• the index admission for the episode is involved in an acute-to-acute hospital transfer (i.e., the admission ends in a hospital transfer or begins because of a hospital transfer)
• the index admission occurs within the 30-day post-discharge period of another MSPB episode
• the index admission inpatient claim indicates a $0 actual payment or a $0 standardized payment

After applying the exclusions outlined above, all remaining episodes are included in the calculation of the MSPB measure.

Data Source:
The MSPB measure is calculated based on all Medicare Parts A and B final action claims during the performance period, including: inpatient hospital; outpatient; skilled nursing facility; home health; hospice; durable medical equipment, prosthetics, orthotics, and supplies; and Medicare Part B Carrier (non-institutional Physician/Supplier) claims. Beneficiary enrollment data and Part A and B claims are used to determine eligible episodes and to attribute MSPB episodes to clinicians, as described in Section 3.0. Prescription drug costs covered by Part D are not included in calculation of the MSPB measure. MSPB does not require any additional measure submission by clinicians or clinician groups.

Detailed Measure Calculation Methodology

Calculation of the MSPB measure is divided into seven steps:

1. Define the population of index admissions,
2. Calculate payment-standardized episode costs
3. Calculate expected episode costs
4. Exclude outliers
5. Attribute episodes to a TIN-NPI
6. Calculate the MSPB measure for the TIN-NPI or TIN
7. Report the MSPB measure for the TIN-NPI or TIN

The following sections explain these steps in more detail.

Step 1: Define Population of Index Admissions
An MSPB episode includes all Medicare Part A and Part B claims with a start date between 3 days prior to a hospital admission (“index admission”) and 30 days after hospital discharge. The episode exclusions described in Section 2.3 are applied to arrive at the final population of index admissions.

Step 2: Calculate Payment-Standardized MSPB Episode Costs
Calculate the standardized cost of each MSPB episode by summing all standardized Medicare claims payments made during the episode window. Episodes include the 30 days after discharge in order to emphasize the importance of care transitions and care coordination for improving patient care. The MSPB measure is payment-standardized to take into account payment factors unrelated to provision of care, such as add-on payments for medical education and geographic variation in Medicare reimbursement rates. More information on the payment standardization algorithm is available in an overview document titled “CMS Price (Payment) Standardization – Basics” and a more detailed document titled “CMS Price (Payment) Standardization – Detailed Methods” at the webpage referenced in Section 4.0.

Step 3: Calculate Risk-Adjusted Expected MSPB Episode Costs
To estimate the expected cost for each episode, the MSPB methodology uses an ordinary least squares regression model to risk adjust for beneficiary age and comorbidities.
Specifically, expected costs for each episode are calculated using a model based on the CMS Hierarchical Condition Category (CMS-HCC) risk adjustment methodology for the Medicare Advantage program.8 However there are several differences. For example, in the MSPB methodology a separate risk adjustment model was estimated for episodes within each major diagnostic category (MDC). The MDC is determined by the Medicare Severity Diagnosis Related Group (MS-DRG) of the index hospital stay. The Medicare Advantage Program risk adjustment model includes 24 age/sex variables, while the MSPB methodology does not adjust for sex and includes 12 age categorical variables. In addition, the MSPB methodology includes individual indicator variables for history of ESRD, long-term care status, and whether the beneficiary qualifies for Medicare through disability or age, in contrast to the stratification and interaction variables used in the Medicare Advantage model.
As mentioned in Section 2.3, episodes where the beneficiary is not enrolled in both Medicare Part A and Medicare Part B for the 90 days prior to the episode are excluded. The 90-day lookback period is needed to capture beneficiaries’ comorbidities for use in risk adjustment. Comorbidities are measured using (i) 79 HCC indicators derived from the beneficiary’s claims during the 90 days prior to the start of the episode, (ii) an indicator of whether the beneficiary recently required long-term care, and (iii) the MS-DRG of the index hospitalization. The 79 HCC indicators are specified in Version 22 (V22) of the HCC model.
The MSPB risk adjustment model also accounts for the impact of comorbidities by including interactions between HCCs and/or enrollment status variables included in the Medicare Advantage model. Interaction terms are included because the presence of certain comorbidities increases costs more substantially for particular beneficiaries than predicted by HCC indicators alone.9 Tables 1 through 6 in Section 5.0 present the final set of risk adjustment variables.

Step 4: Exclude Outliers
Statistical outliers are excluded from MSPB measure calculation to mitigate the effect of high- and low-cost episodes on each TIN-NPI or TIN’s MSPB measure score. Step 4.1 below describes the winsorization on the lower bound of expected episode costs, and Step 4.2 explains the process for excluding outlier episodes.

Step 4.1: Winsorize Expected Costs
To prevent the creation of extreme expected cost values, the MSPB methodology winsorizes expected values at the 0.5th percentile.10,11 Expected cost values are also renormalized, by multiplying the winsorized expected values by the ratio of the average standardized cost within each MDC and the average winsorized expected cost within each MDC. Renormalizing the expected values ensures the average expected episode cost for each MDC is the same before and after winsorizing.

Step 4.2: Exclude Outliers
The residual for each episode is calculated as the difference between the standardized episode cost and the winsorized expected episode cost. Outlier episodes are defined as MSPB episodes whose residuals fall above the 99th percentile or below the 1st percentile of the distribution of residuals across all MSPB episodes. Excluding outlier episodes as determined by the distribution of the residuals eliminates episodes that deviate most from their expected values in absolute terms. Next, renormalize expected values by multiplying the expected episode costs (after excluding outliers) by the ratio of the average standardized episode cost and the average winsorized expected episode cost (after excluding outliers). Renormalization ensures that the average expected episode costs are the same as the average standardized episode costs after outlier exclusions.

Step 5: Attribute Episodes to TIN-NPI
Each MSPB episode is attributed to the TIN-NPI responsible for the plurality of Part B Physician/Supplier services during the index admission. Costs of services are measured by Medicare standardized allowed amounts, and services must be performed by MIPS-eligible clinicians during the episode’s index admission (the period between admission date and discharge date of the hospital stay, inclusive). Part B services are defined as all clinician services billed on non-institutional claims. To determine attribution, Part B services billed by MIPS-eligible clinicians are considered if they are (i) on the admission date and in a hospital setting, with place of service restricted to inpatient, outpatient, or emergency room hospitals, (ii) during the index hospital stay, regardless of place of service, or (iii) on the discharge date with place of service restricted to inpatient hospital.12 If more than one TIN-NPI has the plurality of Part B services standardized payment, the episode will be attributed to the TIN-NPI with the plurality of Part B services bill lines. If more than one TIN-NPI also have the same count of services during a given episode’s index hospitalization, the MSPB episode is randomly attributed to one TIN-NPI.

Step 6: Calculate MSPB Measure for Each TIN-NPI or TIN
The MSPB measure is calculated for each TIN-NPI or TIN by (i) calculating the ratio of standardized observed episode costs to winsorized expected episode costs and (ii) multiplying the average cost ratio across episodes for each TIN-NPI or TIN by the national average episode cost. This method of cost ratio calculation allows for comparison of differences in observed and expected costs at the level of each individual episode before comparison at the provider level. The content of each cost calculation depends on the desired reporting level, as noted in the following steps.

Step 6.1: Calculate risk-adjusted episode cost ratio
For each non-outlier episode, divide the episode’s total standardized cost (calculated in Step 2) by the episode’s total winsorized and renormalized expected cost (as determined by Step 4).

Step 6.2: Calculate the MSPB measure for each TIN-NPI or TIN
Under MIPS, the MSPB measure is an average of ratios. After calculating each episode’s risk- adjusted cost ratio (in Step 6.1), the cost ratios for all episodes are averaged for each TIN-NPI or TIN (depending on reporting choice). Multiplying this average cost ratio by the national average episode cost (all total standardized costs averaged over the universe of attributed, non- outlier episodes) gives the MSPB measure for each TIN-NPI or TIN.

Step 7: Report MSPB Measure for Each TIN-NPI or TIN
Though MSPB is attributed at the TIN-NPI level, reporting can be at either the TIN or TIN-NPI level. Under MIPS, the case minimum for reporting is 35 episodes, regardless of reporting level.

Additional References

For more information about the methodology used in payment standardization, please refer to the CMS Price (Payment) Standardization documents on QualityNet.

More details on the MSPB measure as defined by MIPS may be found in the CY 2017 (81 FR 77166–77171) and CY 2018 (82 FR 53643–53647) QPP Final Rules, including updates from previous versions of MSPB, and in the CY 2019 (83 FR 59767 – 60079) Physician Fee Schedule Final Rule.

Final details of MSPB episode construction and original application in the Hospital Value-Based Purchasing (VBP) Program are in the FY 2012 IPPS/LTCH PPS Final Rule (76 FR 51618– 51626) and the FY 2013 IPPS/LTCH Final Rule (77 FR 53583–53596).

Tables

The following tables present the final set of risk adjustment variables used to calculate expected MSPB episode costs, as referenced in Section 3.0.

Table 1. Age Variables

Table 1. Comorbidity Measures: Hierarchical Condition Categories (HCCs) Included in the CMS-HCC Risk Adjustment Model

Table 2. Enrollment Status Variables

Table 3. Long-Term Care Variables

Table 4. Variable Interaction Terms

Table 5. Indicator Variables