HI410 Unit 5 Case Mix Index Analysis Example

A Rise of 0.0258 Is Not a Documentation Win Until It Is Split: A Unit 5 Case Mix Index Analysis Separating Patient Mix, Tier Movement and One Drift Signal

[Student Name]

Purdue University Global

HI410: Advanced Reimbursement Methodology

Unit 5 Case Mix Index Analysis

[Instructor Name]

[Date]

The hospital, its discharges and its weights are composite; the decomposition method and sources cited are real.

The Question Behind the Number

The finance committee's draft report says that the hospital's case mix index rose because the clinical documentation program, launched at the start of the second year, captured severity that had been missed. That may be true, but a rising index has three possible causes. The hospital may be treating sicker or more complex patients. Documentation may be recording real severity more completely. Or documentation and coding may be recording severity that the clinical record does not support, which economists have called upcoding or creep (Carter et al., 1990). Each cause has a different meaning for the hospital, and only the data can choose among them.

The Two Years in Numbers

In the first year, the hospital discharged 2,600 Medicare inpatients whose relative weights totaled 3,178.80, giving an index of 1.2226. In the second year, it discharged 2,730 patients with weights totaling 3,408.00, giving 1.2484. The index is simply the average relative weight per discharge, so it can rise because more discharges fall in heavily weighted groups or because discharges within each family move to higher tiers (Horton, 2017). The rise of 0.0258 looks small, but at this hospital's adjusted base rate of about $6,300, it is worth roughly $444,000 a year in operating payments, which is why the committee cares.

Separating Patient Mix From Tier Movement

To separate the two effects, I built a counterfactual index: the second year's discharges by family, priced at the first year's share of cases in each tier. If nothing had changed except which patients arrived, the index would have been 1.2389. The difference between the first year and the counterfactual, 0.0163, is a change in case mix in the ordinary sense. The difference between the counterfactual and the actual second year, 0.0095, comes from movement between tiers within the same families. Most of the patient mix effect has a plain explanation. Elective joint replacement volume grew from 260 to 325 cases after the hospital recruited a second orthopedic surgeon in the spring. Joint replacement groups carry weights well above the hospital's average, so 65 more of them pull the index up without any change in documentation. Admissions for heart failure, pneumonia and urinary tract infection, which carry lower weights, held almost level. This part of the rise is real and would have occurred with or without a documentation program. About three fifths of the rise came from which patients the hospital treated, and only two fifths from how their stays were recorded.

Where the Tier Movement Came From

The remaining 0.0095 needs more care. In a healthy documentation program, tier movement would be broad: many physicians, many conditions and many families, each moving a little, as specialists clarified conditions that were already being treated. Part of the second year looks like that. In heart failure and pneumonia, the share of stays in the top tier rose from about 17.5 percent to 25.0 percent, and the conditions behind the movement were varied, including acute kidney injury, acute respiratory failure and severe sepsis, spread across four physician groups. One condition behaves differently. Stays whose only major complication was functional quadriplegia numbered 9 in year one and 47 in year two. Of the 38 added stays, 31 came from one hospitalist group of six physicians. The supplied table treats the condition as a major complication, so each of these stays moved up a tier, adding an estimated 16 weight units in total, or about 0.0058 of the 0.0095 tier effect.

Testing the Signal

A concentrated rise is not proof of a problem, so I tested it against the clinical record in three ways. First, I compared the diagnosis with physical therapy documentation for a sample of 20 of the 31 stays. In 12, a therapy note written within a day of the diagnosis described the patient walking with a walker or standing to transfer with minimal assistance. Functional quadriplegia describes a patient unable to move because of severe disability, so these notes contradict it. Second, I checked whether the hospital's patient population had changed in a way that would produce more immobile patients. Admissions from long-term care facilities and patients with advanced dementia held steady across the two years. Third, I checked whether the documentation program had trained physicians on this condition. Its tip sheets list functional quadriplegia among conditions to consider, with a short definition but no clinical indicators. The pattern fits a drift signal rather than better capture. A condition rose fivefold, almost all of the rise came from one group, the hospital's population did not change, and the therapy record contradicts the diagnosis in more than half the sampled cases. Research on hospital coding shows that this kind of concentrated, payment-linked change is what external reviewers look for, and that it varies with how strongly an organization responds to financial incentives (Silverman & Skinner, 2004).

What the Findings Do Not Show

The analysis does not show that any physician acted improperly. The tip sheet's thin definition and an enthusiastic specialist may explain most of the pattern. It also does not show that all 31 stays are unsupported; in eight of the 20 sampled stays, the record described a patient who truly could not move. And the counterfactual method assumes that the first year's tier shares are a fair baseline, which would be false if the first year itself had been under-documented. The 2,600 and 2,730 discharges are large enough for the overall index to be stable, but the hospitalist group's numbers are small, so its rate should be watched over another two quarters before anyone draws a permanent conclusion.

Recommendations

1. Report the rise to the committee in two parts: 0.0163 from patient mix, mainly orthopedic growth, and 0.0095 from tier movement, of which about 0.0058 traces to one condition.

2. Do not describe the full rise as a documentation success. The defensible claim is that the program contributed part of the tier movement in heart failure and pneumonia.

3. Commission a focused retrospective review of all 47 functional quadriplegia stays by a credentialed coder outside the program, comparing the diagnosis with nursing and therapy documentation.

4. Revise the tip sheet to include clinical indicators and an example of a patient who does not meet the definition, and send any future query about the condition with the therapy findings in the indicators section, as compliant query guidance requires (American Health Information Management Association & Association of Clinical Documentation Integrity Specialists, 2022).

5. If the review finds unsupported claims already paid, route them to compliance for correction and repayment. Errors the hospital finds and corrects itself cost less, in money and in trust, than the same errors found by an outside auditor.

6. Track the index monthly in two lines from now on, one holding tier shares constant and one actual, so the committee sees the two effects separately every time.

References

American Health Information Management Association & Association of Clinical Documentation Integrity Specialists. (2022). Guidelines for achieving a compliant query practice (2022 update). American Health Information Management Association.

Carter, G. M., Newhouse, J. P., & Relles, D. A. (1990). How much change in the case mix index is DRG creep? Journal of Health Economics, 9(4), 411-428. https://doi.org/10.1016/0167-6296(90)90003-L

Horton, L. A. (2017). Calculating and reporting healthcare statistics (5th ed.). AHIMA Press.

Silverman, E., & Skinner, J. (2004). Medicare upcoding and hospital ownership. Journal of Health Economics, 23(2), 369-389. https://doi.org/10.1016/j.jhealeco.2003.09.007