How does the number and scatter of trials vary by discipline?

Article type
Authors
Glasziou P1, Thorning S1, Erueti C1, Hoffmann T1
1CREBP, Bond University, Qld, Australia
Abstract
Background: Bradford’s law of scatter suggests that if the journals in a field are sorted by number of articles and divided into 3 equal sized groups the number of journals will be in the proportions 1:n:n2. Bradford’s Law has been shown to apply to trials overall but little work has been done for specialties of diseases.

Objectives: To assess whether and how Bradford’s law applies to trials and reviews overall and in different specialty areas.

Methods: We downloaded all trials (using publication type) in MEDLINE for 2009, and classified trials in each of 14 clinical areas and specific disorders that are major contributors to the global burden of disease: mental disorders, depression, alcohol related disorders, heart disease, myocardial ischemia, nervous system diseases, cerebrovascular, dementia, otorhinolaryngologic, hearing loss, endocrine disorders, diabetes, cancer and lung cancer.

Results: The number of journals covering all trials in an area was highly correlated with the number of trials (r = 0.97), but varied by clinical area. Depression had most with 318 trials in 214 journals - 1.5 trials per journal. Cardiology had least scatter with 1,459 trials in 374 journals - 3.9 trials per journal. The number of journals to identify half the trials in a condition ranged from 9 for lung cancer trials to 79 for stroke trials. The journal with the highest number of trials in each area was generally a journal within that speciality, e.g. Stroke for cerebrovascular disease and Diabetes Care for diabetes. However, general medical journals and journals in related subspecialties often occurred in the top 10.

Conclusions: Trials are widely scattered in the medical literature and for specialty areas and diseases the scatter follows the pattern predicted by Bradford’s law. The scatter has implications for both clinicians trying to keep up to date and for those performing systematic reviews.