In this paper, I present the view that dichotomous thinking by physicians often does a disservice to clear thinking about patients. Two examples are given. In the first, a commonly used dichotomy masks a continuum of the quantity of tissue damage after myocardial infarction. In the second, a dichotomy replaces a more appropriate thought process: probabilistic diagnostic assessment.

The electrocardiogram has been a valued tool for the diagnosis of myocardial infarction for more than 60 years. Early efforts were made to categorize infarction dichotomously as "transmural" or "nontransmural" on the basis of the assumption that abnormally large Q waves imply transmural infarction. More recently, it has been recognized that pathologically nontransmural infarctions may be associated with abnormal Q waves and that transmural infarctions may not. "Transmural" and "nontransmural" were subsequently replaced by Spodick [1] with "Q-wave infarction" and "S-T infarction." The latter term has evolved to become "non-Q-wave infarction." Although the Q wave/non-Q wave terminology may be more descriptively accurate than the transmural/nontransmural terminology, it maintains the comfortable but misleading tradition of dichotomous classification of disease and disorders.

Infarction removes the infarcted tissue's contribution to the cardiac electrical field, altering the electrical field during the QRS complex. If that electrical contribution is directed inferiorly (as, for example, with an inferior-wall myocardial infarction), one would expect the intensity of the electropositive field to be reduced in the caudad regions of the body, as detected in the vertically directed lead axis, aVF.

From the viewpoint of biological reality, infarction of the left ventricular myocardium does not occur in digital bytes. Infarction is a variable process that can affect any amount of myocardium from less than 1 g to about 300 g. If loss of myocardium can be envisioned as a continuum, does it not follow that alteration in the electrocardiogram should also be a continuum? Accordingly, one can envision an electrocardiographic spectrum from R R R R R R r q Q Q Q Q Q in any given lead. Movement from left to right along this continuum occurs with infarction. Large infarctions may replace large R waves with large Q waves. Small infarctions do not do this and may only reduce an R wave or produce a small Q wave.

Figure 1 shows excerpts from lead aVF of five consecutive electrocardiograms. The patient was a 50-year-old man who presented with severe oppressive retrosternal pain, nausea, and diaphoresis. Myocardial infarction was confirmed enzymatically; echocardiography showed left ventricular inferior-wall hypokinesis. The electrocardiogram did not show an abnormal Q wave, but it did show significant alteration. The amplitude of the R wave diminished from 1.3 mV to 0.4 mV with accompanying appropriate ST-T wave evolution. Clearly, infarction occurred. Clearly, the cardiac electrical field was altered. Clearly, there was movement to the right in the R-Q spectrum. Does it really matter whether an abnormal Q wave appeared?

View larger version (56K): [in this window] [in a new window]   Figure 1. One complex from lead aVF of five sequential electrocardiograms. The patient was a man with an enzymatically and echocardiographically proven inferior-region myocardial infarction. The 1.3-mV R-wave initially present diminishes to less than 0.4 mV as a result of myocardial loss.

Speculation about pathophysiology derived from an imprecise procedure such as electrocardiography makes no sense. Coronary arteriography provides far better insights into the nature of the coronary lesion and its dynamics. Spodick [1] and many others [2] have pointed out that patients with Q-wave infarction differ in many respects from patients whose infarctions are not accompanied by pathologic Q waves. Many of these differences have been shown to be clinically useful, and some have indicated possible differences in the pathogenesis of the two types of infarction [2]. An alternate hypothesis is, of course, that most of these differences are attributable to the size of the infarction rather than to whether the electrocardiogram happens to show a Q wave. Thus, if persons with Q-wave infarctions are more likely to have congestive heart failure, a higher early mortality rate, and a lower rate of recurrent infarction, it is likely that this is simply because more ventricle has been damaged and less ventricle is at risk for subsequent events.

What is the alternative? Is not the best alternative simply to diagnose infarction on the basis of all available information, including the clinical presentation, the electrocardiogram, the enzymatic evidence, and the data provided by an array of imaging techniques that distinguish damaged from normal myocardium? For descriptive labels, tiny, small, medium, and large might be a good start.

When only the electrocardiogram and the clinical presentation were available for the assessment of patients with infarction, Q waves made a valuable contribution. We have moved on. Labeling patients on the basis of such a simplistic dichotomy should no longer be permitted to obscure the continua of size, of disability, and of therapeutic and secondary preventive options that characterize myocardial infarction.

The classification of serous cavity effusions as "transudate" or "exudate" is the second example of obfuscation by dichotomization. Abnormal amounts of fluid in the pleural cavities or other serous spaces can result from increased production of pleural fluid, decreased fluid absorption, or both. Increased production can occur with increased interstitial hydrostatic pressure as a sequel of increased intracapillary pressure, as is commonly seen in congestive heart failure. It can also occur with injury to the pleural surfaces. Decreased absorption may reflect increased intravenous pressure or impaired lymphatic drainage from the lungs. Pleural effusion is often a valuable clue to the presence of cardiac or pulmonary pathology.

More than five decades ago, it became apparent that certain characteristics of the fluid probably reflected changes in hydrostatic factors, whereas other characteristics were more likely to reflect tissue injury or damage. Paddock [3] reinforced the dichotomous classification of transudate or exudate to describe serous fluids. Criteria that would both sensitively and specifically separate these two categories of fluid have been proposed. Light [4] modified existing criteria with a focus on serous fluid protein and lactate dehydrogenase (LDH) levels as the mainstays for determining whether fluid is a transudate or exudate.

Light and coworkers [5], analyzing protein and LDH levels in pleural effusions, reinforced the notion of a dichotomy between transudates and exudates. Careful analysis of the figures that they presented, however, indicates that 17 of 46 protein levels in "transudates" were greater than the lowest levels encountered in "exudates," whereas 28 of 103 "exudates" had protein levels that were lower than the highest levels found in the "transudates." Similarly, 26 of 46 "transudates" had LDH levels greater than the lowest levels encountered in the "exudates," and 30 of 103 "exudates" had levels that were lower than the highest levels found in the "transudates." Labeling a clinical case according to this dichotomy conveys an unjustified aura of diagnostic certainty. The data do not support such a clear distinction. With reference to ascitic effusions, Runyon and colleagues [6] showed the diagnostic superiority of the serum-ascites albumin gradient over the traditional transudate/exudate categories.

An alternate approach to the uncertainties of biology and medicine is to use data to influence probabilistic assessment. Thus, the greater the fluid protein level and the greater the LDH level, or both, the greater the probability of parietal or visceral damage leading to pleural fluid accumulation. At the extreme, the probability will approach unity. The lower these values are, the lower the probability of tissue damage and the higher the probability that deranged hydrostatic forces are the pathophysiologic source of the effusion.

These probabilities, based on available data, are altered by the circumstances in which they are found. Thus, the post-test probability is influenced not simply by the protein and LDH levels but also by the pretest probability of any given disorder. Accordingly, pleural fluid with a protein value of 40 g/L in a person with known severe left ventricular dysfunction and exertional dyspnea may still carry a high composite probability that congestive heart failure is the cause of the effusion. Conversely, an effusion with a protein level less than 25 g/L in the context of a radiographically visible subpleural mass in an elderly smoker may still be judged to have a high probability that malignancy is the source of effusion. A valuable alternate construct, well-described by Sackett and colleagues [7], is to multiply the pretest probability of a given diagnosis by a likelihood ratio (as derived, for example, from data such as those reported for effusion protein and LDH levels [5]) for a diagnostic test to calculate the post-test probability of the diagnosis.

Two examples of a medical tradition of dichotomous thought have been presented. In the first, it is postulated that the continuum of myocardial infarction size is better characterized by continua of clinical and laboratory characteristics that permit qualitative judgments of infarction size rather than by an unrealistic Q wave/non-Q wave dichotomy. Similarly, rather than an arbitrary dichotomous decision, a continuum of probabilistic thought is preferable for the diagnostic assessment of serous effusions.

Most dichotomous classifications do not add information; they simply cloak uncertainty in the guise of pseudoclarity. Dichotomous thinking often does a disservice to medical logic and to patients.