UCSF

Importance

The practice of using medical writers to communicate scientific information has gained popularity, but it may affect how and what information is communicated.

Objective

To assess characteristics of oncology trials that use medical writers and whether there is an association between the use of medical writers and trial success or the primary outcome evaluated.

Design, setting, and participants

This cross-sectional study included oncology trials testing a tumor-targeting intervention that were published in The Lancet, The Lancet Oncology, JAMA, JAMA Oncology, Journal of Clinical Oncology, and The New England Journal of Medicine between May 1, 2021, and May 1, 2022.

Exposures

Assistance of medical writers or no assistance.

Main outcomes and measures

The main outcomes were the percentage of studies with medical writers, the percentage of trial successes reported with medical writers, the association between trial success and medical writer use, and the association between a primary end point and medical writer use.

Results

Among 270 studies, 141 (52.2%) included a medical writer and 129 (47.8%) did not include a medical writer. Of the studies that included a medical writer, 83 (58.9%) were successful. Of the studies that did not include a medical writer, 64 (49.6%) were successful (P = .16 for difference). Studies with medical writers were less likely than studies without medical writers to have the end point of overall survival (15 [10.6%] vs 17 [13.2%]) and disease-free or event-free survival (16 [11.3%] vs 29 [22.5%]), whereas studies with a medical writer were more likely to have the end point of progression-free survival (32 [22.7%] vs 17 [13.2%]). Use of medical writer was associated with the conclusions being presented favorably in all studies (113 [80.1%] vs 89 [69.0%]; odds ratio [OR], 1.81 [95% CI, 1.04-3.19]), but when adjusted for other variables, there was no association (OR, 1.84 [95% CI, 0.92-3.72]).

Conclusions and relevance

In this cross-sectional study, trials using medical writers were more likely to report surrogate end points, such as progression-free survival, and favorable conclusions, but when adjusted for trial phase, randomization, and study funding, there was no association with favorable conclusions. These findings suggest that journals need heightened scrutiny for studies with medical writers and that authorship should be properly acknowledged.