Statistics in clinical practice.

Most of us tolerated our required statistics course and passed it (although not necessarily on the optimal limb of the bell curve). Some of us even retained a little bit of information. I remember clearly Dr. Hyman Menduke's opening salvo to my freshman class at Jefferson Medical College. He looked out into 223 pairs of somber eyes and pointed out that, intellectually, we were five standard deviations from the center of the bell curve. He also noted that if we had been five standard deviations on the other side of the bell curve, we would have been institutionalized. However, the societal acceptability of our status did not make us any less abnormal than our counterparts on the other side of the bell curve. I wish I had remembered more of his course; I would not have had to spend so much time relearning the material.

Medical literature is complex, and frequently it is misleading. Those of us who care for patients depend upon published materials to stay current so that we may offer our patients the best, most advanced clinical care. The peer-review process is designed to weed out misleading articles or those with conclusions not supported by the data presented. However, as hard as editorial boards try, we fail routinely. Hence, it is essential for all of us to maintain a working knowledge of statistics so that we can read the literature incisively and decide what to believe.

Knowledge of statistics requires more than remembering the difference between the median and the mean. (For those who have forgotten, the median is the middle value when all values are ranked from high to low, or low to high. The mean, or average, is the sum of all values divided by the number of values. The median is not sensitive to outliers, but the mean is.) Understanding statistics helps us to assess study design and analysis, and to evaluate the validity of conclusions that may sound quite credible at first reading but often are unsupported upon closer assessment. Since we use the literature to guide our patient care, such distinctions may be critical (literally).

A comprehensive review of statistics is beyond the scope of this editorial, of course. However, for readers who are interested in becoming reacquainted with the topic but lack the time to take a course or wade through a statistics text, I would recommend an excellent review by Guller and DeLong. (1) This article of slightly more than 15 pages was written with the practicing surgeon in mind. I have made it required reading for residents who rotate with me (most of whom remember more statistics than most of us), and I recommend it for any otolaryngologist interested in a convenient refresher on this exceedingly important subject.

Reference

(1.) Guller U, DeLong ER. Interpreting statistics in medical literature: A vade mecum for surgeons. J Am Coil Surg 2004;198:441-58.