Most recognize me from posts about deductive reasoning on KevinMD, but I never explain deductive reasoning’s effect on a conventional medical malpractice lawsuit. Here is how it affects such a lawsuit.

All medical malpractice lawsuits have four things in common:

• All have an underlying complication, either an error-of-nature, which is caused by random chance, or a medical error, which is caused by the medical intervention.
• All test the premise that there is a difference between the standard of care, which is the epitome of excellence that never departs from duty, and the medical intervention, which is the facsimile of excellence that potentially departs from duty.
• All have a measurable standard for decision-making.
• All authenticate evidence.

Medical malpractice is conventionally argued using inductive reasoning. There is nothing wrong with inductive reasoning. However, when a robust method is missing, inductive reasoning is manipulated by “cherry-picking.” Cherry-picking occurs because of selection bias, which is selecting evidence supporting a conclusion, and avoidance bias, which is avoiding evidence contradicting a conclusion. This is also referred to as a confirmation bias. Biases are also possible using deductive reasoning but less so because, as will be shown, deductive reasoning and hypothesis testing, in particular, are more robust and open to examination.

In this hypothetical medical malpractice case, the plaintiff’s medical expert uses inductive reasoning. Presume that this expert finds three departures from the standard of care. Using “preponderance of evidence” as the decision-making standard, collectively, these departures are the proximate cause of the complication. Therefore, inductive reasoning infers that the complication is a medical error and concludes that there is medical malpractice.

Alternatively, the defense’s medical expert uses deductive reasoning (hypothesis testing). This expert finds that there are 10 phases in the medical intervention and 10 counterpart phases in the standard of care. Only one phase in the medical intervention differs from its counterpart in the standard of care. There is a complication, which is either an error-of-nature or a medical error, which must be proven by hypothesis testing.

The background risk for this complication is 15 percent. Each phase in the standard of care is represented by the background risk. To create a test sample to represent the medical intervention, the medical expert uses 15 percent as the yardstick for the complication’s occurrence rate in all 10 phases of the medical intervention. In the one phase of the medical intervention, which is different from the standard of care, there is an increase in the risk of harm and the occurrence rate is 99.7 percent. Nine other phases are the same. Hence, the test sample is 15 percent, 15 percent, 15 percent, 15 percent, 15 percent, 15 percent, 15 percent, 15 percent, 15 percent, and 99.7 percent. The decision-making standard is alpha, which is 0.05 (the sine qua non for hypothesis testing).

Using the single sample T-test, the p-value = 0.171718. The p-value is greater than alpha (0.05) and the null hypothesis is retained. Although there is a difference between the version of the medical intervention in the null hypothesis and the one in the standard of care, the difference is not statistically significant. Hence, the complication is an error-of-nature, and the conclusion is there is no medical malpractice.

Type I error, or the probability of rejecting a true null hypothesis, is alpha or 5 percent. Type II error, or the probability of retaining a false null hypothesis, is more complicated to determine but presume it is 20 percent. Because the null hypothesis is retained, type I and type II errors are both consistent with a null hypothesis that is true.

In deposition testimony, the plaintiff’s medical expert agrees with 10 counterpart phases in the standard of care and the medical intervention and specifies which three phases are different. However, the expert provides no evidence of why or how they are different. Hence, selecting three differences and avoiding evidence for the three differences is cherry-picking.

The defense attorney adapts inductive reasoning to hypothesis testing to examine the plaintiff’s argument for the effects of cherry-picking. This adaptation uses the same robust method as the defense’s expert. However, the null hypothesis, which is tested, refers to the plaintiff expert’s version of the medical intervention having three departures of duty. Alpha remains 0.05. The test sample of 10 phases, representing the medical intervention, uses the same yardstick as the defense’s expert. Because the plaintiff’s expert uses no method to determine how or why three phases are different, the defense attorney adopts the defense expert’s method. The test sample is 15 percent, 15 percent, 15 percent, 15 percent, 15 percent, 15 percent, 15 percent, 99.7 percent, 99.7 percent, and 99.7 percent. The background risk is 15 percent. The p-value is 0.040563.

Since the p-value is less than alpha (0.05), the null hypothesis is rejected. This adaptation to deductive reasoning is completely compatible with the plaintiff expert’s use of inductive reasoning. However, hypothesis testing quantifies authentication of evidence with error rates; inductive reasoning method does not.

The adaptation however does. Type I error is 5 percent and type II is 20 percent. Because the plaintiff’s version of the null hypothesis is rejected, both type I and type II errors are consistent with a null hypothesis that is false.

Deductive reasoning complements inductive reasoning. Deductive reasoning, in the guise of hypothesis testing, is a robust methodology that exposes the unsoundness of premises by quantitatively measuring type I and type II errors. Therefore, the adaptation of inductive reasoning to hypothesis testing casts doubt on the conclusion by the plaintiff’s expert of medical malpractice, which is precisely what a defense strategy is intended to do. How this ultimately is contextualized in legal proceedings remains to be seen. Jurors are not expected to be statistically literate; however, all medical experts and officers of the court are and should be in order to conduct business. Deductive reasoning levels the playing field.

Howard Smith is an obstetrics-gynecology physician.