Occasionally, the NEJM publishes a scary article that gets a lot of press, but doesn’t pan out to be valid. The recent example is the PESIT study, which seemed to imply that everyone admitted with syncope had PE (don’t worry- they don’t).1 Another example was a report correlating azithromycin use with sudden death.2 Despite numerous subsequent articles refuting this3 (including two meta-analyses of RCTs4,5), mythological fear of azithromycin persists. It looks like NEJM might be at it again, with this week’s review article:6
Background: MI as a complication of non-cardiac illness
- Troponin elevation is extremely common among critically ill patients. In most cases, patients don’t meet the universal definition of MI.7 These non-MI troponin elevations (myocardial injury, rather than infarction) correlate with mortality, but don’t require specific therapy.
- Among patients who do meet the definition for MI, most will have type-II myocardial infarction (demand ischemia with stable coronary stenosis; see below). Treatment of type-II MI consists primarily of treating the underlying disease.
- Relatively few patients have a type-I MI, reflecting rupture of an unstable coronary plaque. Only these patients will benefit from advanced coronary interventions such as catheterization.
Diagnosis of MI in the context of non-cardiac illness is incredibly challenging, because we’re searching for a needle (Type-I MI) in a haystack (non-MI troponin elevations and Type-II MI). We obviously don’t want to miss any Type-I MI’s. However, the vast majority of patient’s don’t have a type-I MI. Broadly deploying anticoagulation and cardiac catheterization in such patients will cause only iatrogenic harm.
Acute MI & Infection
Before jumping into this article, let’s parse out what is already widely known on this topic versus clinically useful questions:
What we already know:
- Any source of physiologic stress will increase the risk of myocardial infarction (e.g. emotional stress, extreme exertion, infection, surgery, critical illness).
- Myocardial infarction and troponin elevation correlate with mortality.
Clinically useful questions:
- Are critically ill patients with infection more likely to have MI compared to non-infected patients with the same level of physiologic stress (e.g. matching SOFA scores)? Is there something specific about infection which causes MI, or is this just another form of physiologic stress?
- Among patients admitted for infection, how many have a Type-I MI requiring specific therapy? How do these patients manifest clinically, and how can we detect them?
Acute MI & infection: What is the risk, actually?
Musher et al. cite fifteen articles as evidence that infection increases the risk of MI. This may seem alarming, but it’s nothing we don’t already know. Any source of physiologic stress will cause an increase in MI (for example, running a marathon8). The real issue is the actual risk of Type-I MI. Unfortunately, none of these fifteen articles attempt to parse out Type-I vs. Type-II MI. In fact, none of them even include the words “Type-I” or “Type-II.”
The authors attempt to sidestep this issue, by suggesting that most of the MIs diagnosed were actually Type-I MI:
This statement that Type-II MI is uncommon in infection is a genuinely weird claim. I don’t buy it and neither does twitter:
Most episodes of myocardial ischemia among patients hospitalized for acute infection are due to plaque rupture (Type-I MI). Demand ischemia (Type-II MI) causes only a small proportion of infection-related MI events.
— josh farkas (@PulmCrit) January 12, 2019
My experience has been that among ICU patients admitted for pneumonia, type-II MIs greatly outnumber type-I MIs. Autopsy studies and chart-review studies show that Type-II MIs are more common than Type-I MIs.9,10 Thus, most of the MIs reported in these studies were likely Type-II MIs.
Only one of the studies cited provided enough granular detail to begin to sort out the issue of Type-I vs. Type-II MIs. Musher 2007 performed a retrospective chart review of 170 patients who were hospitalized and eventually diagnosed with pneumococcal pneumonia.11 Twelve of these patients had evidence of myocardial ischemia upon admission:
Most of these patients probably had Type-II MIs. Type-I MI does occur, but it’s rare (likely closer to 1-2% of patients, rather than the 7-8% statistic quoted in the review article). Most importantly, all of these patients had EKG changes, supporting the concept that screening for myocardial ischemia with an H&P and EKG is entirely adequate.
How should this information affect our management?
The article mostly contains information about epidemiology and possible molecular mechanisms. After pages of buildup, the only thing it says about clinical management is this:
This is really weird. Let’s start with the “troponin leak” bit. Patients who have a pneumonia with elevated troponin but no other evidence of ischemia don’t have a myocardial infarction and don’t require any specific therapy for the elevated troponin. This shouldn’t be controversial – it’s written into the Universal Definition of Myocardial Infarction.7 Available evidence and guidelines suggest that for pneumonia patients with an isolated troponin leak, we should focus on treating their pneumonia:
The article suggests that we should initiate statins in patients with pneumonia, to prevent myocardial infarction. The concept of using statins in pneumonia or sepsis has been around for years, but RCTs haven’t supported it.12
The article further suggests that we should initiate aspirin in patients with pneumonia. Like statins, the concept of using aspirin in sepsis or pneumonia patients is nothing new. However, it also lacks evidentiary support (for example an RCT of aspirin in ARDS detected no benefit13). Given that patients with severe infection can develop stress ulceration, blanketing them with aspirin might not be wise.
Ultimately, no evidence is presented to indicate that we should treat these patients differently from any other ill patients. Thus, initial evaluation should include an EKG and a thorough history and physical. For patients with evidence of ischemia, further evaluation and treatment should be pursued as per routine. If the EKG and H&P don’t arouse genuine suspicion for ischemia, then just treat their pneumonia.
Troponin has caused more harm then good ever since it was introduced. LDH-AST-CK-CKMB-Troponin… it has been getting worse ever since. More and more patients exposed to harm (heparin, cath) or noise (cardiology consult). Instead of type 2 “MI” better term is “acute heart injury”
— Ognjen Gajic (@ogi_gajic) January 12, 2019
- Myocardial infarction can be caused by any source of physiologic stress. Infection is no exception to this rule.
- Myocardial infarction can be commonly diagnosed among patients with acute infection. However, most of these patients likely have type-II MI (demand ischemia), which doesn’t mandate specific therapy (e.g. anticoagulation, cardiac catheterization).
- For patients presenting with infection, appropriate screening tools to look for myocardial ischemia include a history, physical examination, and EKG. If these tests reveal no evidence of ischemia, there is no indication to check a troponin.
- Allow me to repeat this for emphasis: Neither influenza nor pneumonia are indications to check a troponin level.
- Aggressively screening every influenza patient with a troponin will only degrade the positive predictive value of the test, leading to lots of false-positive results. This may trigger a cascade of inappropriate therapies (anticoagulation, beta-blockade, cardiac catheterization) with iatrogenic harm.
- Troponin elevation in non-cardiac critical illness (IBCC)
- Avoiding over-diagnosis of MI in critically ill patients (PulmCrit)
- Non-MI troponin elevation (EMDocs by Christina Smith et al.)