The Right Observation Rate – I Have the Numbers, Sort of…

If you have heard me speak or have read my articles you know that one of the questions most asked is “what is the target/benchmark/right observation rate?”

It was seven years ago, in June 2016, that I published my article in Compliance Today, describing for the first time “Hirsch’s Law” to determine a hospital’s correct observation rate for Medicare patients. For those unfamiliar, Hirsch’s Law states that if every patient requiring the use of a hospital bed is reviewed by case management for proper admission status, with the use of a secondary physician review as appropriate, and every patient is placed in the right status, and observation services are only ordered for the patients where observation services are appropriate per regulations, and every patient goes home as soon as their need for hospital care has finished, and every patient who requires a second midnight stay is admitted as an inpatient, then your observation rate is at your benchmark.

When written, Hirsch’s Law only applied to traditional Medicare patients as Medicare Advantage (MA) plans, along with all the other payer types, made up their own arbitrary, ever-changing rules on the use of observation. That will change in 2024 when the MA plans will be required to follow 42 CFR 412.3 and can no longer require patients to linger as outpatient receiving observation services for days on end.

Despite my repeated pleas to stop looking for that elusive target observation rate, hospital administrators still compare observation rates and urge their staff to strive for a lower rate. To that end, I sought to not define a benchmark rate but to determine the actual observation rate for almost every hospital in the country. That way I could state “There is no benchmark rate but here is the national average rate,” hoping that would either appease, anger, or delight those who ask, depending on how that average rate compares to their hospital’s rate.

But obtaining this data was no easy task.

The first obstacle was the fact that many use observation as a catch-all category for all sorts of patients. A recent denial of a claim for an inpatient admission for a scheduled surgery by a major national payer was followed by a notation that the hospital may submit the claim as observation and be paid at that rate. This is not the appropriate use of observation by anyone’s standard yet if the hospital followed the payer’s suggestion, that might be considered an observation stay. In some hospitals, patients who are “abandoned” in the ED without acute medical needs, but no safe discharge plan are placed in the hospital as outpatient with observation services. That is also not proper but nonetheless, such patients would be included in the hospital’s observation count.

In addition to the improper uses, observation can also be properly used for pediatric patients, obstetric patients, and psychiatric patients, making data interpretation difficult. In my experience, when the observation rate is discussed at hospitals, the rate available to the utilization review staff and the finance staff never come close to matching, and neither knows how that rate is even derived other than “it is displayed on the report I get in my inbox every day.”

After considering all these factors, I determined the “cleanest” way to determine the observation rate for a hospital that can be used for comparison was to do the following: (1) Look only at fee-for-service Medicare, (2) Determine how many claims each hospital had paid for Comprehensive Ambulatory Payment Classification (C-APC) 8011, the C-APC for Observation, and (3) determine how many total DRG payments the hospital was paid, all for the same time period. The numerator would be the number of C-APC 8011 claims and the denominator would be the number of C-APC 8011 claims plus the number of DRG payments. Formulaically, that is Observation/Observation plus Inpatient.  

Now before my colleagues rupture an aneurysm objecting to my technique, let me provide the caveats. First, C-APC 8011 is not paid if the patient has a status indicator T or J1 procedure during their outpatient stay. This could commonly occur with a patient placed outpatient with observation services for abdominal pain or gastrointestinal bleeding. Such a patient could stay under two midnights and receive eight or more hours of observation services but the performance of an EGD and/or a colonoscopy would mean C-APC 8011 is not paid. (As an aside, I have repeatedly tried to convince the Centers for Medicare & Medicaid Services (CMS) that this policy makes no sense, as such a patient would receive more services, but the hospital would be paid less than if the procedure was not done, without any success.) This patient type would not be included in this observation rate determination and of course, would also not be in the denominator.

Likewise, C-APC 8011 is only paid if the patient has eight or more hours of observation services. This again would result in some undercounting since some patients could be placed outpatient with observation services and be able to be discharged in under eight hours. But anecdotally this is unusual so the effect would be small.

On the other hand, as noted above, “custodial” patients may be hospitalized with observation services ordered despite the lack of medical necessity, so in some cases, the number of C-APC 8011 claims may be inflated in hospitals with more such patients.

In addition, some scheduled outpatient surgery patients will require observation services for delayed recovery or minor complications. These outpatient claims will be paid under the C-APC for the surgery itself and not C-APC 8011. But this number would be even smaller and more difficult to weed out from the claims data so it will not be considered.  

I also included all DRG payments rather than limiting the DRG count to only the medical Major Diagnostic Category (MDC) DRGs and excluding the surgical MDCs. The patient who is admitted as inpatient with heart failure and undergoes a placement of a defibrillator will fall into a surgical MDC but the reason for admission was a medical diagnosis. Eliminating all surgical MDCs just did not seem appropriate.

I will also note that in calculating the average rate, I excluded critical access hospitals (CAHs) since they are not paid under the Inpatient Prospective Payment System payment (IPPS) structure and there would be no easy way to use the data accessible to me to calculate their observation rate.

With all those caveats, and certainly more which I did not mention, including my reliance on an outside supplier of Medicare claims data without access to the actual MedPAR data, here is what I found for the fiscal year 2021.

1. There were 164 hospitals that were not paid for any C-APC 8011 hospital stays. Many of these were surgical hospitals, where no such claim would be expected. But many were large full-service medical centers, including four with more than 5,000 inpatient admissions and 22 with over 1,000 inpatient admissions.

• There was great variability in observation rates. The highest rate was 59.2 percent at a very small hospital that had 71 C-APC 8011 payments and 49 DRG payments. The lowest hospital, excluding the zero C-APC 8011 stay facilities, had a rate of 0.25 percent, with 19 C-APC 8011 payments and 7,651 DRG payments.

• With no disrespect to smaller hospitals, I limited my average observation rate calculation to hospitals with 50 or more beds. This eliminated 529 of the 2,788 hospitals on the list but just felt right to avoid larger hospitals (inappropriately) claiming that “those little guys are not in the same league as we are.” That left an average observation rate of 12.18 percent.

There you have it. 12.18 percent. A national average, full of caveats. Not the right rate, not the benchmark rate, not your goal rate. Simply the average. This (probably unreadable) table shows the distribution of percentages.

Now how do you compare?

All you need to do is ascertain how many C-APC 8011 claims and how many DRG claims were paid in FY 2021 and compare it to that 12.18 percent. Or you can contact me, and I can look you up on the data I have.

And now my comments.

First, I have no explanation for some of the 164 hospitals that were not paid a single C-APC 8011. Were these hospitals without emergency departments where patients do not present with syncope or chest pain or shortness of breath and need hospital care that would not exceed two midnights? Is it possible that all patients are admitted as inpatient and then the short stays are reviewed for self-denial after discharge? That would certainly delight the physicians to not have to even think about status. But I must wonder if this is a compliant process since the patient’s financial liability is affected favorably if their short inpatient stay is self-denied rather than if they were placed outpatient with observation services and the claim paid as C-APC 8011.

It is also baffling how many hospitals had C-APC rates in the low one-digit percentages, with 13 hospitals with rates below 1 percent. A 1,000+ bed prominent Midwest hospital with more than 80,000 ED visits yet only 164 C-APC 8011 payments is as difficult to rationalize as those with zero.

Likewise, several moderately large hospitals, up to 400+ beds, had C-APC 8011 percentages over 30%. Is the threshold for discharge home from the ED so high that one of every three Medicare patients gets placed in the hospital as outpatient with observation services? Are these facilities so damaged from past RAC audits and recoupments that only the critically ill get admitted as inpatient?

In conclusion, I really do not know what to do with this data.

But at least there is a number – 12.18 percent for fiscal year 2021. As I have said repeatedly, it is not the right rate by any means. I continue to believe in Hirsch’s Law and perhaps the very wide range of rates we are seeing here suggests that there is little adherence to the law.

Perhaps calculating a true observation rate is just so complex that any attempt to do so is a fool’s task. And perhaps this data adds support to the contention of many that the differentiation of payment for inpatient and outpatient hospital care is so convoluted that it is time for CMS to reconsider the whole part A, part B, IPPS, and OPPS payment paradigm.

Programming Note: Listen to Dr. Ronald Hirsch when he makes his Monday rounds during Monitor Mondays with Chuck Buck.