Coding-Clinical Disconnect Part I: Type 2 Myocardial Infarction

Type 2 MI is not a coding construct.

 I received some correspondence regarding my Type 2 myocardial infarction article last week which prompted me to ponder the coding-clinical disconnect. Let us address the Type 2 MI concerns this week, and next week, I will tackle the broader picture.

Historically, the verbiage of “Type 2 MI” was coded with the ICD code for an unspecified site ST-segment elevation MI. Then we were instructed to use the NSTEMI (a subclass of Type 1 MI) code. Both applications of an inferior coding solution were examples of the coding-clinical disconnect (CCD). This is when the coding and clinical practice do not match or make sense, especially from the clinician’s point of view.

In the 2018 code updates, we got the unique, specific code for Type 2 MI, which resolved the coding-clinical disconnect. Unlike Type 1 MIs which have a corresponding subsequent Type 1 MI code (I21- vs. I22-), there is a single code for Type 2 MI, whether it is a first or recurrent infarction. Providers do not use I25.2 Old myocardial infarction in reference to a previous Type 2 MI. 

Since the publication of my Type 2 MI article, a physician suggested that hospitalists should come up with their own definition of myocardial infarction because he doesn’t believe Type 2 MI is a legitimate diagnosis. He is entitled to his opinion, but my personal approach is to reference evidence-based consensus statements which are developed by task forces of societies and organizations, after their careful consideration of the facts and research. I believe my job, as a clinical documentation integrity (CDI) expert, is to reconcile current clinical practice with the current ICD-10-CM code set, to ensure that the patient encounter is accurately represented.

Integrity must play a crucial role in what we do. I do not want patients to be represented as being sicker than they are, but I also feel an obligation to clients to prevent patients from falsely appearing less ill. This is not an issue of intentional up- or down-coding. In our poll last week, thirty-four percent (34%) of our respondents note under-documentation of Type 2 MI compared to eight percent (8%) who find the diagnosis is used excessively.

My dissenter does make one excellent point. We need to consider unintended consequences when we do CDI. Documentation of Type 2 MI may result in denials and quality metric issues.

When we retrospectively query for a condition which seems to be supported by clinical indicators, we run the risk of a clinical validation denial based on insufficient documentation. This is why CDI experts encourage documentation of a condition multiple times in a medical record. Finding a random diagnosis tacked onto a discharge summary raises suspicions. It is better to have an uncertain diagnosis early which evolves and is ruled in or out as the encounter unfolds.

Picking up a major comorbid condition (MCC) might consequently trigger a patient safety indicator (PSI) or core measure. If that is truthfully telling the story of the patient encounter, so be it. If accurate coding suggests that a surgeon has a much higher than expected rate of postoperative complications, it may reveal the need for an intervention by quality. CDI’s role is to avoid unjust punishment of quality medical care with excellent outcomes, merely due to suboptimal documentation.

This used to be the objection of physicians to diagnosing a Type 2 MI which resulted in the undesirable consequence of triggering the Acute MI core measures. The MI core measures are intended to standardize and optimize care for Type 1 MIs. If a provider believes Type 2 MIs do not need to be treated with aspirin, ACEI, beta-blockade, etc., he can assert that as an exclusion criterion. The core measures include, “Other reasons documented by a physician/advanced practice nurse/physician assistant for not…” performing that action. If I had a say in this set of core measures, I would add I21.A1 as a specific exclusion criterion, but that is not currently in place.

Another consequence which really concerns my antagonist is that insurance companies may blackball patients who have been labeled as having had a MI. This was, indeed, a problem prior to October 1, 2017. The solution to this conundrum is not to eliminate Type 2 MI designation. If they haven’t already done so, insurers should adapt their actuarial tables to take into consideration the real-world implications of the third universal definition of MI. Type 2 MI is not clinically congruent to Type 1 MI.

My detractor’s recommendation is for providers to never use the expression “Type 2 MI” and to substitute “troponin leak,” instead. If the entire medical community made this shift, the comparative quality metrics would remain constant. But if Hospital 1 intentionally stopped documenting and coding Type 2 MIs, it would seem as though its case mix index (CMI) had dropped. If all its competitors continued to diagnose and code Type 2 MI in contradistinction, maverick Hospital 1 would then appear to have a less sick and complex patient population in comparison.

No institutional policy can avert the code, if documented, despite the one percent (1%) in our survey noting they do so. It is not compliant for coders to unilaterally discount a diagnosis that a provider has made; if a physician documents “Type 2 MI” the coder must pick up I21.A1, unless they obtain clinical validation repudiation.

I believe the most important reason for wanting to capture Type 2 MI as its own entity is to not underestimate the severity of illness. I strongly believe that our documentation has genuine impact on the medical care we provide to our patients, not just on quality metric representation. As noted in the original Type 2 MI article, the import of the Type 2 MI is that it portends a worse prognosis for the underlying condition.

Join me next week when I will discuss the coding-clinical disconnect in general. Type 2 MI represents just a tiny example of a larger problem which CDI professionals must negotiate.

Comment on this article