A novel use for the word "trend" in the clinical trial literature

Background: A new meaning of the word "trend" is appearing in reports of clinical trials. Methods: Abstracts of all clinical trials in PubMed with English abstracts that contained the word "trend" for each decade from 1971 to 2001 were reviewed. Results: "Trend" was used 36 times in 1981, 170 times in 1991, to 375 times in 2001, most often to refer to a judgment about statistical significance. When the expression "significant trend" was accompanied by a P value, the P value was always less than 0.05; when the expression "nonsignificant trend" was accompanied by a P value, all P values were greater than 0.05; 25% of the time, P values were greater than 0.19, and 5% of the time, they were greater than 0.45. When the unmodified word "trend" was accompanied by a P value, most P values were greater than 0.05. However, on 16 occasions, the P value was less than 0.05, and 5% of the time, P values were greater than 0.20; on 1 occasion it was used with a P value of 0.6. About 30% of the time, the exact meaning of the word could not be determined from the abstract. Conclusion: A novel use of the word "trend" has developed in the medical literature over the last several decades. This novel use should be documented in technical dictionaries, and further discussion should occur about the implications of this new connotation. In the meantime, clinical trials reporting can be improved by accompanying the word "trend" with associated P values or point estimates and 95% confidence intervals.