A couple of years ago, I worked with ICD10monitor cohost Penny Jefferson on a project about admission type and how it impacts performance on quality measures. That data gap is one of the biggest takeaways from the project: the lack of cohesion among measure requirements and data elements that lets some hospitals outperform their peers on paper without necessarily providing better health outcomes.
For example, from the name alone, one would think that “admission type” is a data element limited to hospital inpatients, because the term “admission” refers to inpatient care. It is true that it is a required element on inpatient claims, but its reported value does not reflect the admission type at the time the admission order is written. Instead, the National Unform Billing Committee (NUBC) requires the admission type to be determined at the time the patient starts hospital services for the billed episode of care.
This distinction matters; consider a patient who arrives at the hospital for an elective outpatient procedure, like a diagnostic cardiac catheterization, in preparation for a future elective procedure. If the results reveal critical multivessel disease, what was expected to be an elective inpatient admission at a later date could become emergent admission requiring immediate intervention.
Although this patient is subsequently admitted emergently, the admission type on this claim will be elective. Most quality measures (such as Patient Safety Indicators, or PSIs) exclude the non-elective inpatient population. However, even though this patient now requires emergent care (i.e., an emergent admission), the reported admission type remains “elective,” and the patient is eligible to be part of PSI population.
This is just one of many quirks that can arise when existing data requirements from one area (i.e., NUBC definitions) are not considered in the context of quality reporting. Unfortunately, there are few quality control measures to ensure that all hospitals are collecting and reporting each data element the same way, or according to existing requirements, which can influence their performance. This is especially problematic when most outcome measures are comparative to other similar facilities.
As hospitals tried to improve their performance on PSI measures, many erroneously thought they could create their own standards and definitions for admission type, unaware that they need to consider how NUBC advises admission type is reported in this situation (defaulting to elective). More often than not, these organizational standards allow the facility to exclude more cases than what should have been excluded if NUBC definitions were followed.
Using less stringent criteria translates into better performance. This is where that opening statistic comes from, an ICD10monitor article by Penny Jefferson, where she reported that the top-ranked hospital in a study had an elective admission rate of 4 percent, compared to the 10th-ranked hospital with an elective inpatient population of 42 percent. Data with this much discrepancy suggests variation in how the data is being collected or reported. If the goal is to compare hospital performance, shouldn’t all hospitals be using the same rules?
How implausible is a 4-percent elective inpatient rate? Or, put the other way, a 96-percent non-elective inpatient rate? External benchmarks suggest this is very implausible. Data from the Emergency Department (ED) Benchmarking Alliance found in 2024 that about 19 percent of ED patients were admitted as inpatients. Vizient reports a higher number at 23 percent, which is not surprising, because their membership was derived from the University HealthSystem Consortium, a company that supported Academic Medical Centers. Neither figure is a perfect stand-in for hospital-wide elective/non-elective admission type mix, because the ED is not the only entry point for patients – but it is the source for most non-elective inpatient admissions. Even direct admissions that are transferred from outside hospitals are frequently evaluated in the ED before being admitted. A 96-percent non-elective rate implies that almost every admission arrived with ED-level urgency, which is difficult to square against ED-to-inpatient conversion rates that are at best in the low 20s.
Admission type is not the only place where measurement methodology has recently been shown to diverge from clinical reality. Not only does there appear to be gamesmanship when it comes to admission type, but irregularities in quality data reporting is more common than most hospitals want to admit – unless they are the ones with low scores. Case in point, five Tenet-owned Florida hospitals won a lawsuit against Leapfrog when they received a failing grade for not participating in a voluntary survey. The ruling judge found that Leapfrog’s methodology had “no scientific basis” and “misrepresented hospital safety.”
That ruling matters beyond Leapfrog, because the financial and reputational stakes behind these measures are real. Performance on quality measures can indirectly impact hospital revenue through Centers for Medicare & Medicaid Services (CMS) penalties. Hospitals that perform below their peers can have Medicare payments cut by as much as 6 percent. Hospital reputation and market share are also on the line. Plaintiffs in the Leapfrog case claimed lower “safety” grades led prospective patients to avoid these hospitals out of fear of a higher risk of injury or death.
Clinical documentation integrity (CDI) can play a vital role in performance on many quality measures. In fact, many CDI departments were moved from revenue cycle verticals and into clinical verticals, reporting to the chief quality officer. But that shift raises its own question. Are hospitals putting too much emphasis on avoiding negative outcomes? Has the pendulum swung too far, to the point that hospitals are losing the intent of quality measures, promoting patient safety? PSIs were designed to identify potential hospital complications and preventable adverse events as a tool to help hospitals improve patient safety and quality care. The pressure on CDI and other departments may be so great that these measures are losing clinical value.
Vizient released themes from a recent roundtable discussion with quality executives who recognize the “growing tension between work that supports rankings and work that directly advances patient care improvement.” An emerging concern is that current ranking systems may be driving organizational behavior in unintended ways.
CDI professionals who conduct quality reviews should review their organization’s performance against their peers and verify the accuracy of how data elements are collected if they are an outlier. They can also push back when a documentation query is being driven by “getting an exclusion,” rather than clinical accuracy. Those are not the same goal, and only one of them serves the patient.
Quality measures should not be data that needs to be managed; they should be viewed as learning opportunities for the medical team, coding teams, and quality teams to protect those who entrust them with caring for their loved ones.
