View Post
MVC Implements a Variety of Data Updates to Episode Methodology

MVC Implements a Variety of Data Updates to Episode Methodology

Throughout the past few months, the MVC team has made several methodological updates to its claims-based episodes of care data underlying the metrics shared via MVC’s online registry and push reports. Some of these updates were part of regular claims data maintenance, whereas others were improvements identified and implemented by the MVC team.

Long-Term Acute Care Hospital Utilization Added as Post-Acute Care Category

A new category of post-acute care utilization was generated within MVC episodes of care: long-term acute care hospital (LTACH) stays. Previously, facility claims were grouped into seven major categories: inpatient, inpatient rehab, outpatient rehab, emergency department, skilled nursing facility, home health, and outpatient/other. An area of opportunity was identified by the MVC Coordinating Center and MVC members to add LTACH to this list. Formerly in MVC data, claims for stays at LTACH facilities were grouped in with inpatient claims and thus counted towards “inpatient readmissions” in the context of an MVC episode of care. LTACH is now its own category of care within MVC episodes and is assessed separately from inpatient stays at general acute care hospitals and Critical Access Hospitals. To count towards post-index LTACH care in an MVC episode, a facility claim must contain bill type 011X and the billing facility NPI for the claim must be primarily affiliated with taxonomy code 282E00000X. LTACH claims will continue to be price standardized in the same manner as other inpatient claims.

As a result of LTACH being added as a separate category of care in MVC episodes, MVC members can now also look at their patients’ use of LTACHs on the MVC registry. By index condition, members can view their attributed episodes’ rate of post-index LTACH utilization as well as their average LTACH payment per episode within the Payment by Condition reports for all payers. To do so, users must navigate to the Payment by Condition report, scroll down to the “Payment Measure” filter on the left side of the registry, and select “LTACH ($)” or “LTACH (%)” to look at average payments or utilization rates, respectively.

Updates to Hierarchical Condition Category (HCC) Identification

Another update made to MVC data this year was the application of components from the most recent specifications around hierarchical condition categories (HCC) from the Centers for Medicare & Medicaid Services (CMS). HCCs are patient comorbidities that both CMS and MVC use as part of risk-adjustment processes. When creating episodes of care, MVC uses each patient’s claims data in the 180 days prior to a given index event to retrospectively assess the comorbidities diagnosed for that patient prior to their MVC episode of care. Formerly, diagnoses indicated as “present on admission” on a patient’s index claim were also used to ascertain a patient’s HCCs, but MVC has updated its methodology such that no diagnoses from the index claim will be used in the assessment of patient HCCs going forward. MVC continues to create 79 HCCs according to HCC V22, with new diagnosis codes added each year.

Furthermore, we note that the category hierarchies created by CMS have been applied to the HCC comorbidities that MVC assesses and displays on the registry. The “hierarchical” aspect of the condition categories is applied to groups of similar diagnoses with a goal that patient comorbidities are not over-counted. For example, a patient diagnosed with diabetes may have multiple similar diagnoses reported on claims over a six-month period, such as diabetes without complications, diabetes with chronic complications, and diabetes with acute complications. Rather than describing that patient as having all three diagnoses, a hierarchy is applied so this patient will simply be described as having the most severe of the group of diagnoses (i.e., diabetes with acute complications). To look at the prevalence of HCC comorbidities among your patient population for one of MVC’s 40+ inpatient or surgical episodes of care, members can navigate to the “Comorbidities” report on the registry.

New Medicare Severity Diagnosis-Related Group (MS-DRG) Version

As part of annual maintenance to accommodate newly introduced billing codes, MVC recently updated the version of Medicare Severity Diagnosis Related Codes (MS-DRGs) being used to re-group inpatient claims into categories of similar inpatient stays. MS-DRG v40.1 is now being used by MVC to categorize all inpatient claims containing ICD-10-CM diagnosis codes and ICD-10-PCS procedure codes.

Inpatient Claim Outlier Length of Stay Methodology

MVC updated the method by which inpatient claims with a particularly long length of stay are identified and price standardized. MVC price standardizes each inpatient claim by adding up three components: a standard DRG-based payment, an inpatient transfer payment (if applicable), and a length of stay-based outlier payment (if applicable). An outlier payment is added to the total price-standardized payment amount for a given inpatient claim if the covered patient remained in the hospital significantly longer than an average patient with the same DRG. In the past, MVC identified these “outlier” long length of stay inpatient hospitalizations using publicly available national long length of stay thresholds for every DRG from TRICARE, the uniformed services healthcare program. MVC’s updated outlier methodology uses Medicare Fee-for-Service (FFS) claims to identify the 99th percentile in length of stay (days) among inpatient claims for each MS-DRG. The hospitalization length of stay on each inpatient claim is then compared against the newly identified 99th percentile threshold for the corresponding DRG. Claims with stays exceeding that length threshold are considered outliers. The outlier payment added to that claim’s price-standardized payment amount is then calculated with an unchanged formula as follows: Outlier Payment = (Number of Days Over DRG-Specific Length of Stay Threshold) * $2,500.

All-Cause Readmissions Assessed for All MVC Conditions

New this year, all-cause inpatient readmissions following index hospitalizations will be assessed for all MVC conditions whenever readmission metrics are shown. Specifications around the identification of readmissions will not vary by index condition.

Episodes Containing COVID-19 Care Now Identified by Primary Diagnosis Codes Only

Finally, MVC has modified the identification of episodes containing care for COVID-19. Episodes are now flagged as containing significant COVID-19 care if they meet the following criteria: at any point during the 30- or 90-day episode, a COVID-19 diagnosis (U07.1) was found in the primary diagnosis code position on a facility claim categorized as inpatient, inpatient rehab, skilled nursing facility, or LTACH. These episodes are often excluded from metrics displayed in MVC push reports. To exclude episodes containing COVID-19 care from metrics shown on the registry, members can use the registry filter called “COVID Cases.” Users should select “Exclude 30-Day COVID” to exclude episodes in which COVID-19 was found within the index event or 30 days post-index. Selecting “Exclude 90-Day COVID” will exclude episodes where a primary COVID-19 diagnosis was found within the index event or 90 days post-index.

For more information on MVC episodes of care data, please refer to MVC’s data guide. MVC members with questions not covered within the data guide are welcome to reach out to the Coordinating Center at Michigan-Value-Collaborative@med.umich.edu.

View Post
Understanding MVC Episode Creation and the Meaning of Data

Understanding MVC Episode Creation and the Meaning of Data

The Michigan Value Collaborative (MVC) houses a wealth of administrative claims data for Blue Cross Blue Shield of Michigan, Blue Care Network, and Medicare Fee-For-Service beneficiaries in Michigan. These data provide valuable insights about health services utilization, patient outcomes, and the value of care received at facilities across Michigan. To gain meaningful insight from large sets of unprocessed claims data, MVC transforms these claims into 30- and 90-day episodes of care for over 40 different medical conditions and surgical procedures. These episodes are the core of MVC analytics and result in the data available to MVC members through the online data registry and push reports. In order to best understand the meaning of MVC analytic outputs, it’s important to know what an MVC episode actually includes and how it is created.

One of the ways that MVC provides insights into episodes of care is with data on payments for professional and facility healthcare services. A foundational aspect of MVC data is that payments are price standardized and risk adjusted. In other words, the payments that MVC shares in its reports and on the registry are not representative of actual dollars; they represent utilization of healthcare services that account for patient, provider, and payer differences for fairer comparisons between members and over time. Price standardization removes variation due to various factors that impact insurer payments, such as the payer, contractual agreements, geographic location, and time (see example scenario in Figure 1).

Figure 1.

On the other hand, risk adjustment accounts for differences between patients, because some patients have more healthcare needs than others and may experience a worse outcome or require greater amounts of care than another patient. Risk adjustment helps with fair comparisons between hospitals that see more complex patients compared to those with fewer complex patients. MVC’s risk adjustment always accounts for patient age, gender, payer, history of high healthcare expenditures, and comorbidities such as end-stage renal disease. It also sometimes accounts for condition-specific factors. Together, price standardization and risk adjustment allow for patients and hospitals to be compared more accurately and fairly in MVC reporting.

How are MVC episodes created?

Episodes are initiated by one of three types of index events: inpatient admission to a hospital, an emergency department visit resulting in an inpatient admission to a hospital, or an outpatient procedure at a hospital. To become an MVC episode, these initiating events must have corresponding billed and paid insurance claims that fit into the episode definition of an MVC condition. These definitions are typically comprised of inclusion and exclusion codes for ICD-9-CM and ICD-10-CM diagnosis codes, ICD-9-PCS and ICD-10-PCS codes, and/or CPT codes. Generally speaking, episodes of medical conditions have an initiating facility or professional inpatient claim with a qualifying diagnosis code in the primary diagnosis code position. For example, a patient admitted to an inpatient hospital with a primary diagnosis code of R65.20 (severe sepsis without septic shock) may have a sepsis episode in MVC data. The criteria for initiating surgical episodes are more varied and depend on the procedure. For more details about condition-specific episode definitions and rules, please refer to the MVC Episode Definitions document available on the resources page of the MVC registry.

Once an episode is initiated, the claims that follow are attached to that episode and categorized into payment components (Figure 2). The span of the initiation of an episode through initial discharge is called the index event, for which MVC aggregates facility claims and associated price-standardized facility payments for the base payment as well as outlier and transfer payments, if applicable. The claim categories after index discharge make up the bulk of an episode. This post-discharge aspect of MVC episodes allows hospitals to follow their patients after they leave the hospital's four walls. For the 30 or 90 days after discharge, facility claims for that patient’s episode are price-standardized and categorized into post-acute care and inpatient readmission components. Post-acute care is further sub-categorized into the following categories: emergency department (ED), home health care (HH), skilled nursing facility (SNF), rehab (inpatient and outpatient), and outpatient facility-based services. For greater detail about MVC episode creation, see the MVC Data Guide on the resources page of the MVC website.

Figure 2.

While facility claims and their respective price-standardized payments are grouped into various categories, the price-standardized professional payment spans the entirety of the episode. Since claims for facility and professional-based services are billed separately, they can be brought together by service dates for a more comprehensive summary of care. The resources page on the MVC website contains a more detailed breakdown of episode payment components.

How are episodes used?

Once episodes are created, the MVC Coordinating Center analyzes these data to answer a variety of questions about health services utilization. For example, what proportion of a hospital’s joint replacement patients are going to a SNF after their procedure? How does that compare to the statewide average? How long is the average length of stay for a hospital’s sepsis patients? What proportion of sepsis patients are admitted to the ICU/CCU during their hospitalization? There are many questions surrounding utilization and outcomes that MVC utilizes its data to help answer. MVC episodes can help inform a wide variety of quality improvement initiatives for numerous conditions.

In addition to some of the measures mentioned above, MVC frequently includes comparison groups in reporting as a point of reference, which allows hospital and physician organization (PO) members to compare their performance, utilization, or outcomes with the collaborative average, regional average, or other individual hospitals or POs. MVC comparison groups are sometimes further tailored by size or type, such as by critical access hospital status or by PO size. See Figure 3 for a sample caterpillar plot from a recent chronic obstructive pulmonary disease (COPD) report for POs.

Figure 3.

While claims-based analytics can provide insights into health services utilization, it doesn’t always reveal the full picture. MVC encourages members to use its push reports and custom reports in conjunction with electronic medical record data and as conversation starters with staff and clinicians working on particular conditions or service lines. Additionally, many MVC conditions align with other collaborative quality initiatives (CQIs). As the MVC team collaborates with other CQIs to combine clinical and claims data, MVC encourages cross-collaboration of site champions to foster partnerships and information sharing.

The MVC Coordinating Center is always open to comments and suggestions to help improve its portfolio of analytic offerings. If you or your team has any feedback on existing reports, suggestions for new reports, or interest in new MVC conditions, please contact the Coordinating Center at Michigan-Value-Collaborative@med.umich.edu.