3.2 Case Mix Index and Why CDI Tracks It
Key Takeaways
- CMS defines a hospital’s case mix index (CMI) as the average diagnosis-related group relative weight for the discharges in the calculation: sum of DRG weights divided by the number of discharges.
- CMS publishes both transfer-adjusted and unadjusted CMI figures; a hospital’s internal monthly CMI is the same arithmetic applied to whatever payer and discharge set the finance team chooses.
- CMI moves when the mix of base MS-DRG families changes, when CMS recalibrates relative weights, and when documentation and coding change which MS-DRG is assigned inside a family.
- CDI influences CMI only indirectly, by supporting complete and clinically valid documentation that assigns the correct MS-DRG—not by treating a CMI target as a query quota.
- Hospital-wide CMI is a blunt metric; family-level CC/MCC capture and a handful of high-volume MS-DRG families explain CDI’s contribution more honestly than a single campus number.
3.2 Case Mix Index and Why CDI Tracks It
Quick Answer: Case mix index (CMI) is the average MS-DRG relative weight of the discharges you include in the math. CMS says a hospital’s CMI is calculated by summing the DRG weights for the Medicare discharges in the file and dividing by the number of discharges. CDI changes CMI only by changing which MS-DRGs those discharges receive—not by editing a dashboard.
Relative weight is the IPPS language for expected resource intensity. A weight of 1.0 means the stay is expected to use about the same resources as the average Medicare inpatient case used to calibrate that year’s scale. Weights above 1.0 describe more resource-intensive MS-DRGs; weights below 1.0 describe less resource-intensive MS-DRGs. CMS publishes the official relative weight for each MS-DRG in the annual IPPS rule files. This section does not invent those official weights. When a number appears below, it is a teaching weight used only to show the arithmetic.
CMS’s definition, not a slogan
On its Case Mix Index file page, CMS states that a hospital’s CMI represents the average diagnosis-related group (DRG) relative weight for that hospital and is calculated by summing the DRG weights for all Medicare discharges and dividing by the number of discharges. CMS also notes that CMIs are calculated using both transfer-adjusted cases and unadjusted cases, and that the agency updates the public file to coincide with proposed and final IPPS updates.
Write that as a formula you can defend in a leadership meeting:
CMI = (sum of MS-DRG relative weights for discharges in the set) ÷ (number of discharges in the set)
If the hospital’s finance team includes only Medicare fee-for-service IPPS discharges, the number is a Medicare CMI. If the team includes all payers that group with MS-DRG logic, it is an all-payer CMI built on the same average. Those two numbers are not interchangeable. When someone says “CMI dropped,” the first clarifying question is which discharges and which weight table.
Transfer-adjusted CMI exists because IPPS transfer payment policy can change how much of a stay’s weight is recognized when a patient moves to another acute hospital under the transfer rules. CDI does not calculate that adjustment by hand. CDI does need to know that a raw internal average and the CMS public CMI file may not match, even when documentation did not change.
Teaching arithmetic (illustrative weights only)
Five discharges in a teaching set use these illustrative relative weights, not CMS-published weights for named MS-DRGs:
| Teaching stay | What the stay represents in the story | Illustrative relative weight |
|---|---|---|
| A | Low-severity medical discharge | 0.80 |
| B | Medical discharge with a CC-tier MS-DRG | 1.10 |
| C | Medical discharge with an MCC-tier MS-DRG | 1.40 |
| D | Higher-weight surgical discharge | 2.00 |
| E | Mid-range medical discharge | 0.90 |
Sum of weights = 0.80 + 1.10 + 1.40 + 2.00 + 0.90 = 6.20. CMI = 6.20 ÷ 5 = 1.24.
Now change only stay C from the CC-tier teaching weight (1.10, if that stay had grouped without an eligible MCC) to the MCC-tier teaching weight 1.40. The table already uses 1.40 for C. If C had instead grouped at 1.10, the sum would be 5.90 and CMI would be 1.18. One correctly documented, eligible MCC on a single chart moved this tiny set by 0.06. Scale that idea to a hospital with thousands of discharges and you see why finance watches CMI—and why one MCC on one chart barely moves a real campus number.
If next month the surgical stay D disappears because that service line is closed, and a second 0.80 medical stay takes its place, the sum becomes 5.00 and CMI becomes 1.00 even if every medical chart is documented perfectly. That drop is mix, not a CDI failure.
Four forces that move CMI
| Force | Inside typical CDI control? | Why it moves the average |
|---|---|---|
| Mix of base MS-DRG families | No | Transplants, major surgery, and ventilator-heavy stays carry higher weights than routine medical families. Volume shifts change CMI overnight. |
| CMS relative-weight recalibration | No | Each IPPS year CMS recalibrates weights. The same MS-DRG can contribute more or less to CMI without any chart changing. |
| MS-DRG assignment inside a family | Partly | Principal diagnosis, procedures, and eligible CC/MCC use determine which member of a family is assigned. This is the CDI/coding zone. |
| Which stays remain inpatient | Indirect | When lower-weight stays move to outpatient or observation, the remaining inpatient average can rise even if documentation quality is unchanged. |
A useful mental model used in CDI program analysis (and expanded later in Domain IV) is to stop arguing about the hospital-wide number and compare CMI—or, better, CC/MCC capture—inside a stable family. Heart-failure-and-shock, simple pneumonia, and COPD families are common examples because they are three-tier, high volume, and sensitive to a single eligible secondary diagnosis. If capture inside those families is stable and campus CMI still falls, look at mix and weight-table changes before rewriting the query policy.
Why CDI is asked to “own” CMI
Executives use CMI because it is one number, it is comparable across months, and it sits next to net patient revenue in IPPS-heavy hospitals. A rising CMI can mean the hospital is treating a heavier mix, or that documentation and coding now describe the mix that was already there. A falling CMI can mean lighter mix, a CMS weight shift, a surgeon’s departure, a utilization-review shift toward observation, or missed severity.
CDI’s honest contribution is the third story: the chart now supports the MS-DRG the patient already earned. That includes:
- A principal diagnosis that matches the UHDDS test, not the highest-weight guess.
- Secondary diagnoses that are MCC, CC, or neither as CMS currently defines them.
- Exclusion-list awareness so a neutralized CC is not counted as a “win” in a capture rate.
- Clinical validity so an MCC that raises CMI this month does not become a recoupment next quarter.
CMI is a poor individual productivity score. A reviewer who works hospice-to-inpatient transfers, normal newborns, or low-weight surgical shorts will not “produce” the same CMI movement as a reviewer who works medical intensive care. Measuring query rate, response rate, and family-level capture is more faithful to the work.
What CMI is not
CMI is not a quality score. It is not severity of illness (SOI) or risk of mortality (ROM) under All Patient Refined DRGs (APR-DRGs). Those four-level, independently scored subclasses belong with later program-analysis and mortality-index sections. A hospital can have a high MS-DRG CMI because it does cardiac surgery and still have weak SOI documentation on medical deaths. Do not use CMI as a stand-in for mortality risk adjustment.
CMI is not a compliance ceiling or floor. CMS does not publish a “correct” CMI for a community hospital or an academic medical center. Comparing your 1.4-something to a peer’s 1.8-something without comparing service lines is how programs get pressured into leading queries. There is no published numeric CMI target in the CCDS handbook, and this guide will not invent one.
CMI is not proof of medical necessity. A correctly weighted MS-DRG on a stay that never met the Two-Midnight Rule is still a setting problem (section 3.3). Inflating CMI by assigning an unsupported MCC is a clinical-validation and DRG-downgrade problem (section 3.4).
Scenario: the orthopedic departure
Lakeside Community Hospital’s monthly all-payer CMI falls from a teaching value of 1.52 to 1.41 after the only arthroplasty surgeon leaves. CDI’s heart-failure family CC/MCC capture is unchanged. The chief financial officer asks for an emergency MCC campaign. The defensible analysis is: show the count of high-weight joint-replacement MS-DRGs before and after the departure, show stable capture inside unchanged medical families, and show that a campus CMI target would require either a new surgical mix or clinically unsupported severity. The ethical move is education and concurrent review where indicators already exist—not a quota that treats CMI as a scoreboard.
Track CMI so you can explain it. Do not let CMI become the reason a query is written.
CMS describes a hospital’s case mix index (CMI) as:
A hospital’s monthly CMI falls even though CC/MCC capture inside reviewed heart-failure charts is unchanged. Which explanation is most consistent with how CMI is constructed?
Leadership asks CDI to raise CMI by a set increment this quarter. The most compliant description of CDI’s role is: