Person A – If you use this method you can double the risk of problems.
Person B – We rarely have problems.
Person A – How would you know? And you still are doubling the risk.
Fortunately, the major risks associated with Deep Brain Stimulation lead implantation surgery are relatively rare. For example, while post-operative scanning may reveal an intracerebral hemorrhage rate of nearly 10%, only approximately 2% result in a clinically significant hemorrhage. The problem is that such rare events make surveillance and judgments difficult. Statistical analysis and judgment depends on the Theory of Large Numbers. The Central Tendency of a data set, such as the mean or median of the sample, stabilizes as the number of observations increases. Prior to some critical number of observations, one cannot “trust” measures of Central Tendency.
Consider the situation where the risk of significant hemorrhage with optimal methods is 1%. This means that with each surgery, the risk is 1% or 0.01. Assuming that the risk for each subsequent patient is independent, then the cumulative risk of witnessing a significant hemorrhage is n * 0.01, where n is the number of persons having surgery. For example, the average risk for a significant hemorrhage occurring after 10 subjects is only 10%, or 0.1. However, this number is misleading. An individual patient cannot have a 10% significant hemorrhage. Even if 49 patients were observed, the probability of having seen a significant hemorrhage in one patient still is less than 50%, or 0.5. Even this assumes that the sample from which patients are drawn is representative of the population at risk. Next, consider the situation where a surgeon adopts a method that increases the risk of significant hemorrhage to 2%, or 0.02. On average, it would take at least 25 subjects to have a 50% chance of detecting a single significant hemorrhage, and again more than 50 subjects to detect the excess hemorrhage over what would be expected under optimal conditions.
The scenarios described above are best-case, or idealized, situations. Decision-making becomes even more problematic when the actual occurrence is a stochastic process. Thus, in one group of 50 patients, no hemorrhages are encountered. In the next group of 50, 2 patients with significant hemorrhages are encountered. Though the overall risk is unchanged, the physician’s experience is very different. If one wants to have confidence in the rate of significant hemorrhages, typically a much larger number of patients is required to conform to the Theory of Large Numbers that underlies statistical judgments.
The great danger is that the physician will make judgments based only on the group of 50 patients that he or she experiences. If it is the first group, then the physician likely will underestimate the risk. If it is the second group, there is the chance of overestimating the risk. The scenario just described is also idealized by assuming complete knowledge of all the patients. It is not clear if patients are followed in such a manner as to have complete knowledge.
The risk that it may not be possible to know
As can be seen, judgments based on small numbers are very difficult. One cannot have confidence in the various measures upon which physicians and healthcare professionals depend. Yet the patient is in front of the physician and healthcare professional, and the latter are obliged to help. The responsibilities and interactions between the patient, physician and healthcare professional can be understood in terms of the ethical principles of beneficence, autonomy, non-malfeasance (meant as not harming rather than suggesting incompetence), and justice.
The obligation to beneficence is fundamental to medicine in the presumption that persons become physicians and healthcare professionals just to help. The issue of non-malfeasance is more difficult. In the general sense, malfeasance is unavoidable. It harms the patient to make an incision in the scalp, yet we counterbalance that necessary harm with the benefit hoped to be obtained. It is an age-old problem of trying to strike a balance between necessary harm and the greater hope of benefit. One approach to the situation is the Principle of Double Effect, which explains how an action causing harm may be justified when it is inseparable from the good effect that is desired. One of the clearest explications of this was described by Thomas Aquinas (1225 – 1275) in his work Summa Theologica. According to this principle, a key determinant in balancing harm versus good is the intention of the physician and healthcare professional. The intentions relate not only to the good, but also to the harm; thus the harm must be minimized.
As we, physicians and healthcare professionals, must intend to minimize harm, how do we gauge our efforts? What is the basis for our accountability? Is it based on actuarial experience, that is, how many patients we may have harmed with the judgment based on more harm than is standard? In this case, accountability is retrospective. Given the difficulty of ascertaining harm, holding strictly to a retrospective accountability may risk allowing us a “get out of jail free” card. Alternatively, should not accountability be prospective? Clearly, the large majority of physicians and healthcare professionals implicitly display prospective accountability by their judicious considerations prior to causing harm in order to obtain benefit. But the question is how to construct the prospective accountability? What is the basis for determining that certain methods should or should not be used? As discussed above, subsequent demonstrations of harm are not likely helpful in a great many decisions physicians and healthcare professionals must make.
Obligation to use reason
Much of the argument against “cookbook” medicine is based on the skepticism that any “recipe” could address the variability among patients. Most every physician and healthcare professional recognizes the necessity of individualizing the care of a particular patient. Thus, by necessity, physicians and healthcare professionals must exercise reason, even if reasoning is not valued by Evidence-Based Medicine. Indeed, courts of law have been defining standards of medical care, not by what similarly- situated physicians would do, but rather based on what reasonable physicians would do.
That same obligation to use reason is inherent in virtually every decision. Fortunately, many of these issues have already been settled by consensus or consilience. Nevertheless, the individual physician and healthcare professional are expected to use their own reasoning. It is no less the case in the decisions about what methods to use for DBS lead implantation surgery. The use of microelectrode recordings clearly increases the risk of hemorrhage, even if that increased risk does not show up in clinical trials. To appreciate how this must be so, consider making the contrary statement that it does not matter how many times you puncture the brain in terms of risk. To maintain this position, one would have to argue that the second microelectrode penetration somehow has less risk and the next one, if necessary, has even less risk. This clearly does not make sense. Thus, the obligation is to minimize the risks while maintaining the benefit. This is also true of the type of microelectrodes used and how they are used. In the absence of unobtainable data, one has to reason using principles and physics.