Levels of evidence in otolaryngology journals

To identify trends in clinical research and levels of evidence in otolaryngology journals.

We reviewed all original research articles from 1993, 1998, and 2003, in 4 major otolaryngology journals. Levels of evidence were graded 1 (strongest) through 5 (weakest).

Of 2584 total articles, 1924 (75%) were clinical research. During the study period, there was increased median sample size (from 22 to 30, P=0.06), more planned research (from 30% to 37%, P=0.023), more internal control groups (from 36% to 43%, P=0.011), and more articles with P values (from 26% to 45%, P<0.001) or confidence intervals (from 1.2% to 7.6%, P<0.001). Most evidence was level 4 (57%), but median levels increased slightly over time (P=0.027). Therapy articles had the weakest evidence (80% levels 3 to 5) and diagnostic test assessments had the strongest (75% levels 1 and 2).

Although clinical research increased in quantity and quality, sample sizes were modest, most articles lacked controls, and confidence intervals were rare. Therapy articles would benefit from higher evidence levels.

By defining the current levels of evidence in otolaryngology journals, this overview should help guide future efforts.