DRG drift: catching the silent capture loss.
Case-mix index does not crash. It drifts. A 0.02 quarterly slip never trips an operating review, yet four of those quarters compound into eight figures of capture loss. This issue maps where the drift comes from and how to claw it back.
Issue #1 framed why CMI belongs at the top of every monthly operating review. Readers asked the same thing a dozen ways: we see CMI move, we have no forensic vocabulary for what is moving inside it. Issue #2 starts that vocabulary, with three leads, one cohort chart, and a 10-item operator's checklist.
CMI is drifting by cohort, not by case mix.
The 300-bed-plus cohort lost 0.13 of CMI between 2024 and 2026 year to date, sliding from 1.62 to 1.49. The 100 to 299 bed cohort moved from 1.41 to 1.34. Critical access fell from 1.18 to 1.12. None of these hospitals saw their patient population shift in any meaningful way. The drift came from inside.
Four causes show up in every audit we run. Documentation rigor decay as long-tenured CDI staff leave and replacement training lags. Query non-response from physician groups that never adopted a 24-hour SLA. MS-DRG v42 re-weighting of normal newborn, uncomplicated OB, and certain peds DRGs. Surgical DRG capture loss when procedure-to-diagnosis linkage is not documented at discharge.
Compression compounds. A 0.02 quarterly slide reads like noise. Eight quarters of it costs a 300-bed facility roughly $3.4M to $3.8M in annual capture at typical DRG payment levels. The number is invisible until someone benchmarks against the same cohort two years back.
Sepsis-3 documentation is where the capture math breaks.
Sepsis-3 requires a SOFA score increase of two or more plus suspected infection. Sepsis-2, the older Surviving Sepsis Campaign definition, only required SIRS plus suspected infection. Both clinical conversations still happen in the same ED. The coded record cares about exactly one of them.
The most common drift pattern looks like this. The chart documents SIRS criteria met, suspected infection treated, antibiotics started. No SOFA is calculated or written down. POA flagging of acute organ dysfunction is missing. The encounter codes to MS-DRG 871 without MCC instead of 871 with MCC or 870 with septic shock. The relative weight gap can be five times. The payment delta on a single case can exceed $40,000.
The CDI fix is not abstract. Build the SOFA into the ED templated H and P. Train hospitalists to document acute organ dysfunction as POA on day one when present. Pre-audit the top 20 sepsis charts every week. The capture lift shows up in 60 days.
Medical necessity is the highest-leverage CDI intervention.
The single highest-leverage move CDI leaders can make is to harden the medical necessity rationale in the admission H and P. Without it, the DRG drops a level, the ALOS expectation slips, and observation conversion risk goes up. Practices using a structured medical necessity template show CMI 0.08 to 0.12 higher than free-text shops on the same case mix.
The fix is template plus discipline. A structured H and P that names the clinical indicator, the severity, the comorbid burden, and the inpatient rationale. A two-midnight statement when applicable. Physician-advisor review of every observation conversion within 24 hours. Daily DRG validation by CDI before discharge summary lock.
Data deep diveCMI by hospital cohort. Where the compression lives.
One picture, four cohorts, three time points. The 300-bed-plus cohort owns the largest absolute compression. Critical access owns the smallest absolute slide but the largest as a share of base CMI. The 100 to 299 bed cohort is where most denial dollars sit on the table, because volume meets sepsis and surgical capture loss in the same building.
Operator's checklistTen moves for CDI leaders this quarter.
Compiled from the CMI gap audits we ran in 2025 and 2026. Ordered by leverage, not by ease.
CMI from 1.36 to 1.47 in six months, $1.8M recovered.
The hospital was carrying 1.36 CMI, drifting from a 1.43 baseline 18 months earlier. ASP-RCM ran the CDI playbook, rolled the medical-necessity H and P template across all service lines, fixed sepsis capture mix, and instituted a 24-hour query SLA. Six months in, CMI sat at 1.47 with sepsis-2 versus sepsis-3 capture corrected and $1.8M of incremental DRG revenue annualized.
Reader Q&AThree questions from this month's inbox.
How fast does a 24-hour query SLA show CMI movement?
Inside 60 days, typically a 0.02 to 0.04 lift. Concurrent queries land before discharge summary lock, so CCs and MCCs make it into the coded record instead of leaking to DNFB. The lift compounds for the first two quarters then plateaus.
Is OB and newborn CMI compression structural?
Yes. MS-DRG v42 re-weighted normal newborn and uncomplicated OB downward. If your service-line mix tilts heavy on OB and peds, your CMI will move regardless of patient population. Benchmark against your own service-line mix, not the system average.
What is the right CDI staffing ratio?
For mid-size hospitals running concurrent CDI, one CDI clinician per 1,500 to 1,800 inpatient discharges per year. Critical access can run leaner at 2,000 to 2,400. Below that, query volume drops. Above it, ROI per CDI hire flattens fast.
Want the same CMI gap audit on your data?
Free 30-day audit. Send 90 days of inpatient charges by MS-DRG, your query response log, your top 20 DRGs, and a denial sample. Inside 30 days you receive a 4-page written audit. Yours to keep.