We have all heard it, time and time again. In fact, I recall telling my hospital’s chief medical officer that my patients were sicker than others to explain why my average length of stay was longer than those of other internists on the medical staff. The difference, of course, was that for my patients, it was true, but not for the other doctors.
When I review Program for Evaluating Payment Patterns Electronic Reports (PEPPERs) with hospitals, I always point out that PEPPER data is comparative data, and not a measure of whether one is doing things right or wrong. The PEPPER includes measures of the percentage of medical and surgical inpatients whose claims included a CC or MCC (comorbid condition and complication and major comorbid condition and complication).
A hospital with a “sicker” patient mix, with service lines that care for more complex patients, or a tertiary or quaternary care center, can expect to have a more complex patient population and therefore have more patients with CCs and MCCs. They also would expect to have a higher case mix index (CMI).
CMI, though, is simply an average of the case weights for all inpatients: a number derived from the weight assigned to each Diagnosis-Related Group (DRG), which in turn is derived from the principal diagnosis and the presence or absence of just one CC or MCC on the claim. Hospital administrators like the CMI because it translates directly to revenue.
Improve documentation, get DRGs to move to higher-weighted DRG, as with moving a simple pneumonia to a complex pneumonia, or add a CC or MCC to a baseline DRG, and the payment from Medicare goes up.
But does CMI adequately stratify hospitals well enough to be able to say, “our inpatients are sicker than your inpatients, so we have more CCs and MCCs?” Of course not. The CMI does not change if the patient has one MCC or three MCCs, but clearly the patient with three MCCs is going to be sicker and need more resources.
Well, thanks to Dr. James Kennedy, a frequent contributor to ICD10monitor, we now have a much better measure of patient “sickness.” Dr. Kennedy found a database published by the Centers for Medicare & Medicaid Services (CMS) that indicates the average Hierarchical Condition Category (HCC) score of every inpatient admitted in 2021 at almost all hospitals. For any who have tried to venture into data.CMS.gov and survived, you know it can be a harrowing experience. Nonetheless, knowing that Dr. Kennedy took the excursion and found useful information provided me the confidence to try myself.
And I was rewarded with a .csv file with 3,134 lines and 45 columns. But of those 141,030 data points, I discovered the few that provided useful data. Those were column B (the hospital name), column E (the city name), column F (the state name), column J (the number of discharges), and column AS (the average HCC score of all inpatients). I promptly hid the rest of the data, providing me enough information to be of value. And of paramount importance, I followed Dr. Kennedy’s advice and saved the file as an .xlsx file (but only after first forgetting to do that and losing all my edits).
Because the HCC score includes demographic factors along with most diagnoses, whether they are a CC/MCC or not, and not just a single CC or MCC, as with CMI, this data is as close to the most accurate publicly available measure of patient acuity as we may be able to get.
From this initial effort to make the data more manageable, I was then able to sort the remaining data to find the hospitals with the highest average HCC score (a specialty hospital with 50 discharges has a score over 8), the national-average HCC score (1.986), the highest and lowest in a state, while also being nosy and looking at some prominent facilities throughout the country. I am sure people with Excel skills better than mine (representing the majority of the population) will undoubtedly find other interesting factoids.
So, what can you do with this? Aside from comparing your facility to your competitors for the fun of it, this may help you understand whether your high percentage of patients with CCs and MCCs (dare I say “to the point of being an outlier”) is justified by a sicker patient population, or perhaps whether it may be an indication that you should increase your clinical validation efforts. If you have access to length-of-stay data, this higher average HCC score may help explain why yours is longer than others, or whether your doctors really are keeping patients longer than necessary, or doing evaluations of incidental findings since it is more convenient, among other reasons.
In this era of constant cost pressure and consolidation, the addition of data such as this should be welcome – with, of course, the caveat that there are many factors that influence hospital finance beyond how sick your traditional Medicare patients are.
But more granular data certainly beats the alternative.