A Warning from the OIG about Higher-Severity DRG Shift

This OIG audit is an opportunity for us to be introspective.

In February, a report came out from the U.S. Department of Health and Human Services (HHS) Office of Inspector General (OIG) regarding their concern about an apparent increase in the reporting of highest-severity Medicare Severity Diagnosis-Related Groups (MS-DRGs) (Trend Toward More Expensive Inpatient Hospital Stays in Medicare Emerged Before COVID-19 and Warrants Further Scrutiny). In this report, Medicare Part A claims for inpatient hospital stays from 2014 through 2019 were analyzed.

Here is a quick review for anyone not familiar with the way that MS-DRGs are determined. The principal diagnosis (or procedure) establishes the DRG, and the secondary diagnosis or diagnoses determine the tier. Most MS-DRGs are grouped in dyads or triads. Some secondary conditions are considered comorbid conditions or complications (CCs), and some are deemed major comorbid conditions or complications (MCCs). If an encounter has one or more CCs, it gets elevated to the middle tier, and if there are one or more MCCs (instead of or in addition to CCs), the encounter is referred to as “with MCC,” landing in the highest tier (that is, within the triad, or triplet, sets). There are also binary dyads(couplets/pairs) (e.g., no CC/MCC, with CC/MCC; no MCC, with MCC).

The relative weight, which is a number assigned to a DRG commensurate with the consumption of resources, tends to increase with the severity and complexity of the patient’s conditions. Higher-weighted DRGs are allotted longer lengths of stay, and reimbursement follows the relative weight. Sicker, more complex hospital patients with more comorbidities and complications tend to stay longer, require a higher intensity of service, utilize more resources, and cost more to take care of.

The traditional clinical documentation integrity (CDI) approach was a hunt for CCs and MCCs. It involved not inventing them, but finding verbiage that suggested the presence of those conditions, albeit in an uncodable format, and getting the providers to document to accurately represent how sick and complex each patient actually was.

When I am consulting, if I am trying to identify provider documentation opportunity, I often select records without a CC or an MCC. It stands to reason that if there are conditions that are not being picked up because documentation is suboptimal, no CC/MCC charts will be fruitful. I compile the data and educate based on the findings.

My intention is for the providers to learn from their omissions and document appropriately from that point onward. There is a danger that some providers will overcompensate and start using verbiage suggesting a sicker patient when it is not appropriate. That is what clinical validation queries are for – finding documented conditions that do not seem to be supported by the clinical indicators and getting the providers to remove them.

In this OIG audit, they found a trend towards the highest-severity DRGs over the course of the review period, and concluded that upcoding was occurring. They noted that nearly a third of these highest-severity admissions had shorter length of stays than expected, and over half of them had only a single MCC. Their bottom line was that since the COVID-19 pandemic has placed a financial strain on the country’s healthcare system, it would be optimal to ensure that Medicare dollars are being spent appropriately by focusing on inpatient hospital stays that account for a large share of Medicare spending. Their conclusion was that the Centers for Medicare & Medicaid Services (CMS) should conduct targeted reviews of MS-DRGs subject to potential upcoding.

After careful consideration, I must point out multiple flaws in the OIG’s assumptions, analysis, and conclusions:

• They consistently referred to these secondary conditions only as “complications.” In coding, “complication” implies preventability, like postprocedural complications. However, we know that the terms CC and MCC really refer to “comorbidities and complications.” If the patient has a condition present on admission, it is a comorbidity. If a condition develops during the stay, it makes the stay more complicated, and it makes the patient’s case more complicated, but it isn’t necessarily the provider’s fault.
• What happened in 2015?! We transitioned to ICD-10-CM from ICD-9-CM. This has got to be confounding on its own merits.
• You have to compare the DRG distribution to something. Personally, I will compare a hospital’s DRG distribution to Medicare Provider Analysis and Review (MEDPAR) data, but I always keep in the back on my mind that this may not be a fair comparison. Quaternary care institutions with robust CDI programs are not really comparable to small rural community hospitals with no CDI support. Once an in-depth review and analysis is done, you can compare your data to your own optimized data in the future, but until then, you need to start benchmarking somewhere. The point is, who says all the rest of them have been doing it right? Maybe in 2014, all the hospitals were missing legitimate CCs and MCCs, and this oversight was rectified during the OIG’s review period. Perhaps 2019’s data was not skewed nefariously; perhaps the prior comparison years were falsely depressed by faulty documentation and insufficient CDI intervention.
• The OIG found that as the number of cases with increased severity rose, the number of cases with lower severity decreased. Isn’t this obvious? If the patient has that principal diagnosis, they have to fall into one of the tiers – the tiers are mutually exclusive. If the providers improve at recognizing and documenting comorbid conditions or complications in a codable format, the percentage of patients in the lower tiers fall as the higher tiers swell.
• On page 5, there is a table where they show a change in the length of stay (LOS) from FY 2014 to FY 2019. I don’t understand the methodology. If each separate cohort fell from between 0.3 to 0.5 days, how could the LOS for all stays only drop by 0.1 days? Are hospitals being more militant about placing patients who previously were designated inpatients in outpatient status for observation services? This would skew the distribution to higher levels of severity.
• Many facilities have implemented programs to improve efficiency and reproducibility of best-practice healthcare. Maybe the length of stay has dropped due to improvements in process. Is the discharge procedure for hospitals being streamlined and expedited? The cited drop was from 6.9 days to 6.4 days. The hospital administrator would be delighted in aggregate, but the individual patient had an almost seven-day stay in both cases.
• The report notes that stays during which the beneficiary died were excluded from the analysis. This could affect the LOS in either direction. Patients might die rapidly if they were extremely ill, causing a very short stay, which could skew the LOS shorter, but some patients who expire have prolonged stays, which could skew the LOS longer. The impact could vary depending on the number of patients in a given population who died.
• The report gives an example of a patient who had a principal diagnosis of pneumonia and 23 additional diagnoses, only one of which was an MCC, which elevated the tier to the highest severity. Yes, that is exactly how this works. Having a single MCC has the same impact as having 10 MCCs, and it flips the patient into the w/MCC tier. If they had acute hypoxic respiratory failure and required aggressive treatment and resource utilization, they had an MCC, whether it resolved rapidly or took days. They belong in the w/MCC tier.
• The OIG report goes on to say that “if that single diagnosis was not billed appropriately, the stay should have been assigned to the lower-level severity MS-DRG 194.” That might not even be true – it could be that the patient would drop two tiers if none of the other secondary diagnoses were a CC. But that is not the point. If a patient has a legitimate MCC, they belong in the w/MCC bucket.
• The real question would not be if the single diagnosis had been billed appropriately, but whether it had been diagnosed and documented appropriately. This OIG report was entirely claims-based. In my opinion, it is problematic for them to draw conclusions about the appropriateness of single MCCs without performing record reviews.

The OIG’s recommendation was that targeted reviews of MS-DRGs and stays that are vulnerable to upcoding were indicated, and that hospitals that frequently bill them should be investigated. CMS did not concur with the recommendation.

As a society, we do not condone fraud. The healthcare community does not suborn for upcoding. There is, however, potential for overzealous diagnosing, querying for, and coding of risk-adjusting comorbid conditions and complications. The CDI community should not, and does not, advocate that providers inflate the severity and complexity of patients’ diagnoses.

This OIG audit is an opportunity for us to be introspective. If your facility has the bandwidth to do so, it would be prudent to perform focused reviews. Assess your Program for Evaluating Payment Patterns Electronic Report (PEPPER) and see if there are red flags. Look at your DRG distribution and if there is excessive variance; perform chart reviews. Set up alerts for DRGs with a single CC or MCC, especially the DRG conditions listed in the report: COPD, simple pneumonia, heart failure, renal failure, urinary tract infections, and sepsis. Monitor your own trends. Make sure your providers are well-educated regarding CCs and MCCs.

Being proactive can save you a world of trouble from third-party payer denials, Recovery Audit Contractors (RACs), and the OIG. Heed the shot fired across our bow.

Programming Note: Listen to Dr. Remer report this story live today during Talk Ten Tuesdays, 10 a.m. Eastern.