A Primer on Hypothesis Testing for Craniofacial Surgeons: How to Obtain Valid Conclusions From Observations and Experiments

Abstract

“Ability and judgment” are the only terms that appeared twice in the Hippocratic Oath and are the pillars of medicine and surgery, as clinical decision making and treatments depend on physicians’ judgments and abilities. Sound judgment requires knowledge obtained and accumulated through individual and collective experience, but more importantly from studies employing scientific methods. In a compendium article to this current one, the death of George Washington within days of onset of respiratory infection in 1799 exemplified the futility of ineffective, if not dangerous and harmful, medical treatments. The key question is how can anyone know with certainty that a particular therapy is or is not effective? Since individual differences exist in responses to corrective procedures and medications, absolute certainty is uncommon. However, as the number of patients increases, so does the confidence, at times approaching certainty. Statistical testing allows for the extraction of valid, useful information from data collected and is a critical part of all surgical disciplines. While this mini-review will not replace the need for expert statistical inputs from statisticians, it will provide a deep appreciation of and a working knowledge for these tests. We focus on hypothesis testing, the foundation to synthesize knowledge to guide, improve new treatments, and confirm or repudiate existing ones. The primer begins with hypothesis testing and systematically examine parametric and non-parametric statistical tests, ending with Mendelian randomization.