Biostatistical Trends in Pediatrics: Implications for the Future

All physicians face the challenge of keeping abreast with a body of medical knowledge that is growing at an exponential rate. On this account, physicians often spend countless hours each month reading journals to "keep current" with recently described techniques of diagnosis and therapy. To gain maximal benefit from their reading, physicians must be able to assess the scientific merit of published research. They must evaluate the various claims and conclusions and then decide which are valid and applicable to their own clinic settings. This type of critical insight requires familiarity with basic principles of good study design and with biostatistical logic and procedures. Unfortunately, recent studies have demonstrated that physicians' concepts regarding statistics are often inaccurate, and even more disturbing, that readers are often willing to draw conclusions unsupported by the available data.1,2 Some authorities have therefore recommended remedial statistical training for physicians by means of increased attention to statistical issues in biomedical journals. The question of exactly which statistical concepts and techniques need to be mastered, however, remains largely unanswered. On this account, I reviewed several volumes of Pediatrics to determine which statistical techniques have appeared regularly, and whether the frequency and intensity of statistical analysis have recently changed enough that a familiarity with biostatistics that was adequate a few years ago may no longer be sufficient. I surveyed the scientific reports in volumes 9 (1952),29 (1962), 49(1972), and 69 (1982). I excluded editorials, book reviews, correspondence, and the like, and I classified remaining articles as review articles, case reports, or research reports.