Until now, abductive/inductive reasoning is conventional when mitigating medical malpractice. Abductive/inductive reasoning presents a daunting task over a single observation: the injury. The observation could be a medical error caused by the medical intervention, but it could also be an error-of-nature caused by random chance.
Four variables measure this observation to determine which is which:
- Background risk: It is the probability of random chance as found in scientific literature. The background risk represents the standard of care because any complication associated with the standard of care can only be random. Medical errors are impossible in standards of care. Ignoring background risk is a “heuristic bias.”
- Relative risk: Being empiric, the standard of care allows for a risk up to the point the risk becomes a breach of duty; hence, there is a calculated risk. Relative risk measures calculated risk. If there is no difference between the medical intervention and the standard of care, the relative risk is 1.0. If there is a difference, the relative risk is greater than 1.0. How much greater depends on the “risk of harm” (ROH). ROH = 100 percent (the maximum calculated risk) ÷ background risk. If the background risk is 15 percent, ROH = 100 percent ÷ 15 percent = 6.67. Relative risk is a choice proportional to ROH. Again, there are biases: “selection bias” and “avoidance bias.” A “rationale” must accompany the assignment of relative risk.
- Incident risk: Incident risk = background risk x relative risk. Each phase in the medical intervention has an incident risk; therefore, the medical intervention is represented by a “test sample” of 10 incident risks. Some are the background risk and some are greater than the background risk.
- Level of significance: It is the tipping point, which separates medical interventions that comport with the standard of care from those that do not. In “the law,” it is 50 percent probability plus a scintilla. In “science,” it is 95 percent probability. Clinical medicine is science. Because it is a choice by the investigator, there is a bias: “confirmation bias.”
CCC+C is a decision-making algorithm based on deductive reasoning. There are four steps in CCC+C.
Step 1: Collate. The standard of care and the medical intervention are organized into 10 discrete phases, each with a respective duty. Fundamental to all phases in the medical intervention is the first duty, “primum non nocere,” the duty to avoid a calculated risk greater than 100 percent.
Real physician voices, twice a week
Free, and one click to unsubscribe.
Step 2: Compare. Each phase in the standard of care is compared to its counterpart in the medical intervention using the variables described above.
Step 3: Conclude. This step is hypothesis testing, which proves statistical significance. When the differences in incident risks are statistically significant, the difference is not random chance, and there is a medical error and a departure from the standard of care. When there is no statistical significance, the difference is by random chance, and there is no departure. The one-sample t-test, which is standard in most statistical software, is used to determine statistical significance.
The level of significance is alpha, either 0.5 or 0.05. The p value is the result. When the p value is greater than alpha, the null hypothesis is retained, and there is no departure from the standard of care. When the p value is equal to or less than alpha, the null hypothesis is rejected, and there is a departure. The p value determines whether or not there is proximate cause.
Step 4: Certify. Any investigator prepares and notarizes a report. The report includes six quantifiable results:
- Error rates: The probability of type I error and type II error. Type I error is rejecting a true null hypothesis. It is “alpha,” and it is not calculated. Type II error is retaining a false null hypothesis. It is “beta.” Beta is calculated: Beta = 1 – power. For convenience, when type I error is 5 percent, as in science, type II error is 20 percent. When type I error is 50 percent, as in law, type II error is 0.65 percent.
- Selection bias: The probability that the p value is unreliable because of selecting capricious relative risks. When the null hypothesis is retained, selection bias is type I error. When rejected, it is 100 percent minus type I error.
- Avoidance bias: The probability that the p value is unreliable because valid relative risks are avoided. When the null hypothesis is retained, avoidance bias is type II error. When rejected, it is 100 percent minus type II error.
- Confirmation bias: The probability that the p value is unreliable because of the choice of alpha. Confirmation bias = alpha – p value. When the null hypothesis is retained, confirmation bias is a negative number. When the null hypothesis is rejected, it is a positive number.
- Heuristic bias: The probability that the p value is unreliable because of the choice of background risk. When background risk is correct, heuristic bias is 0. Otherwise, it is 100 percent.
- Merit: The probability that all the data an examiner uses to determine the p value are reliable. Merit = 100 percent (certainty) minus the average of selection bias, avoidance bias, confirmation bias, and heuristic bias.
Because standards of care are empiric, when using abductive/inductive reasoning, an observation can be exploited to fit a bias. An error-of-nature can be interpreted as a medical error and vice versa. The slightest difference will serve as a departure from a standard of care, even though the discrepancy is so trivial that it cannot be the proximate cause because there is no significant increase in relative risks. Nevertheless, a preponderance of evidence can be presumed, and causation is fabricated. This is all that is necessary to mitigate a lawsuit without any demonstrable merit. Error-prone conclusions often result.
CCC+C ends this. Both abductive/inductive and deductive reasoning operate on probabilities and have the same function, but deductive reasoning, which follows a discrete heuristic, is more effective and is more transparent.
Howard Smith is an obstetrics-gynecology physician.
