Guest Blog: Clinical Validation

By Howard Rodenberg, MD, MPH, CCDS for ACDIS CDI Blog

I think we’re all familiar with the Law of Unintended Consequences. That’s the concept that something begun with the best of intentions can wind up going horribly awry. Examples include “New Coke” and anything ending in the word “Kardashian.” (Although I’m not sure anything the Kardashians do start with the best of intentions.) Medicine is no stranger to this—think about those eras where bleeding, purging, and “laudable pus” were all the rage. Personally, the Law of Unintended Consequences also explains my first two marriages.

CDI Strategies recently cited an article that exemplified this point. All of us in the CDI world are familiar with CMS’ Hospital Readmission Reduction Program. The idea is to incentivize hospitals to minimize inpatient re-admissions within 30 days of a prior hospital discharge. However, work in the online journal JAMA Cardiology suggests that in patients admitted with heart failure, declines in the 30-day readmission rate were also associated with an increase in mortality within the same patient group at both 30 days and one year after admission.

Association is certainly not the same as causation; the fact that I’m associated with two failed marriages does not mean that I caused them to fail, although I’m sure other researchers (ex-wives) would come to a different conclusion. But it does seem like there’s some relationship between decreased readmission rates and increased deaths in the congestive heart failure (CHF) population, and a perfectly reasonable attempt to decrease inpatient costs and improve the quality of care has backfired, perhaps at the expense of lives.

How does this link into CDI world? I would argue that on many occasions, we’ve also subjected ourselves to that same law; things that seemed a good idea at the time turn into metaphorical albatrosses around our necks. (FYI, I don’t know the plural of albatross. Albatrosses? Albatri? Albatree?)

Which leads us to the clinical validation query. This is our attempt to clarify diagnoses that may not be supported by clinical indicators within the medical record. On paper, it’s a great idea. We don’t want to code things that are not there. If we do, that puts us at risk of audits and takebacks and accusations of fraud. It is altogether right and proper that we as a CDI community want to do something proactive about this.

So we’re going to sit at our desks and zap off an e-mail to the doctor saying, hey, I know I’m not seeing the patient, I have no responsibility for the welfare of the case, and I didn’t go to medical school and all, but you wrote down that the patient is septic and I don’t think they are, so are you sure about that?

You will write it much more politely and use many big Latin words, but that’s really what you’re asking. And you would be perfectly within your rights to ask, for it’s a perfectly legitimate question.

But let’s think for a moment how a doctor’s going to receive that one. Let’s look for the smiles of thanks, the glowing acknowledgement that you, as the CDI specialist, know more about this patient…in fact, more about medicine as a whole…than I do. Because that’s how it’s going to be read.

Still looking.

Physicians see the clinical validation query as a Nastygram. They view it as a personal affront, being second-guessed by someone lower on the food chain with little knowledge and no responsibility. (Like it or not, clinical medicine is still a hierarchy with physicians at the top.) And I say this not just to build a case for my point of view, but because that was my visceral reaction every time I got a memo about my ER turnaround times or a patient complaint or some administrative dictate that I violated in order to do what I thought was best for the patient.

The Law of Unintended Consequences means that no matter how good the reason for the clinical validation query, all it does is make enemies of the very physicians your CDI program is trying to support. No amount of explanations regarding the need for accurate coding or audit protection is going to change that. This kind of query undermines our efforts in nearly every way. And if you ask the query but the physician still sticks to their guns, you’re stuck in the pickle of being forced to code what isn’t there.

So how do we find our way out of the dilemma?

I think there are three potential answers.

One is simply to not send the query and code what you think is most appropriate based on the clinical indicators present. I would never actively advocate for breaking the coding rule that mandates what the physician has documented in the chart must be coded. But one can use the definition of the secondary diagnosis to help support a decision not to code. Did the stated condition lead to any changes in care? If not, it doesn’t meet the definition, no query is needed, and the diagnosis may legitimately be excluded from the final list.

The second solution is to develop institutional definitions. An institutional definition is essentially a statement by the medical staff of a hospital or healthcare system of what a clinical condition “looks like to us.”

Developing institutional definitions takes work; literature reviews, position papers and drafts, gathering specialist opinions, running them through medical staff committees and adapting them into policies and procedures. But once established, the definition can serve as a basis for physician and coder education. It solidifies the defense against audits, as it’s much harder for an insurance company’s retired psychiatrist to argue a diagnosis of acute kidney injury against a hospital’s six board-certified nephrologists. And while Coding Clinic rightly suggests that coding departments cannot establish policies or procedures that dictate whether or not to code an unsupported diagnosis, I’m personally not so sure that prohibition applies when the policy is dictated by the medical staff itself. My own opinion…not the rule of law, by any means…is that you can build a policy around the institutional definition, with such policy approved by the medical staff, establishing what CDI and coding professionals should do when the physician documents a condition which is not supported with institutional definitions of particular clinical indicators. The effort needs to be led by a clinician, either your CDI physician advisor or a similar champion. Clinical definitions from a non-physician will be accepted by medial staff with the same enthusiasm as they greet the clinical validation query.

The third approach is to reframe the clinical validation query as a force for good. Present the idea in an open-ended format to physicians as, “You’ve seen the patient, I believe you, please give me more information so I can keep auditors and quality folks off your back.” It’s a variation of the idea I espoused a few posts back, that the message doctors respond to best is that we help eliminate the “hassle factor” from their practice. The clinical validation query then becomes a mutual project in “fighting the good fight” against the outside world.

These latter two approaches are admittedly more complex ways to manage the problem than benign neglect, and take more long-term effort than simply issuing clinical validation queries. But I think they’ll show greater benefits in the long run, and I think there’s a powerful synergy when both are combined. It’s the approach we’re taking to resolve this issue, and I’ve been surprised at how accepting physicians are of our efforts when framed in the context of keeping them hassle-free.

Maybe I need an institutional definition of marriage for the third round so I can be compliant too.