Descriptive Biostatistics and Epidemiology for Craniofacial Surgeons: The What, Who, When, Where, How, and Why in Data Description, Analysis, Presentation, and Interpretation

Abstract

Physicians often make diagnosis and treatment decisions based on incomplete data. That is why we practice medicine. We use accumulated knowledge and prior experience, individual and collective, to restore form and function of our patients to return them to normalcy with continued, durable homeostasis. However, due to the complex nature, our diagnosis may be wrong and our treatments ineffective or even harmful. The history of medicine and surgery is replete with such examples from snake oil and bloodletting to Halstedian radical mastectomy. Without knowing the governing dynamics, the cause-effect relationship is often obscure. Prospective, blinded, placebo-controlled trials provide the highest level of evidence, like the COVID vaccine trials, to determine if a treatment works. However, trials cannot explain in detail how and why a treatment works, or why it does not. Variabilities and uncertainty abound and require the correct mathematical methods to tease out the signal from the noise, causality from association. Collectively, statistics is the science of uncertainty and the extraction of reliable, useful information from raw data. The objectives of this review are to provide craniofacial surgeons with a primer in descriptive statistics: how to design investigations, collect, prepare, present, and interpret clinical data. Since large datasets at regional and national depositories represent powerful and valuable resources, and that their proper use requires a working knowledge in epidemiology, we included sections on incidence, prevalence, sensitivity, and specificity regarding diagnosis, treatments, and testing.