Electronic Thesis and Dissertation Repository

Thesis Format

Monograph

Degree

Master of Science

Program

Epidemiology and Biostatistics

Supervisor

Martin, Janet

Abstract

Introduction: Lack of study publication leads to bias in the scientific literature. It is important to better understand this phenomenon and find methods for mitigation.

Research Question: How many clinical trials registered on ClinicalTrials.gov in London, Ontario are started, completed, and published?

Methods: Data from all studies in the ClinicalTrials.gov registry associated with London, Ontario were collected, from registry conception until the end of 2017. We determined whether these registered studies were published by July 2020 and whether their first publication included their planned primary outcome at all. Main factors associated with non-publication were assessed using multivariable log-binomial regression. Multivariable modified Poisson regression was used to assess the association between enrollment size and publication. Time to publication was assessed using multiple linear regression.

Results: Of the registered studies (n = 2446), only 38% were published and 30% with their planned primary outcome. Median time to publication post-start was 53 months [IQR: 36, 75]. Factors associated with publication were randomized design, prospective registration, industry funding, drug study, and enrollment size (p < 0.05). Factors associated with shorter time to publication were positive results, prospective registration, and industry funding, while drug studies were associated with longer time to publication (p < 0.05). Surgical studies seemed to have decreased chances of publication and lengthened time to publication but was not statistically significant in either case.

Conclusions: A substantial proportion of clinical trials from London, Ontario remained unpublished. The factors predictive of non-publication and time to publication suggest potential avenues for increasing publication rates.

Summary for Lay Audience

An inherent limitation of scientific literature is that we only know the information that is publicly accessible through publications. Having as much information as possible is important when making decisions, especially when it comes to treating life-threatening illnesses. When studies are not published, the amount of information available is reduced.

Even worse is when certain factors make studies less likely to be published, such as unfavourable results or the type of treatments studied. Publication bias occurs when the amount of information available is skewed because some studies are less likely to be published than others. Additionally, there may be selective outcome reporting, where some studies are published with only a subset of their results. These issues are highly prevalent and hinder our ability to make reliable scientific decisions. Unfortunately, there is no way of knowing about every single study that ever existed. The closest thing we have are study registries like ClinicalTrials.gov, where studies are ideally registered before they are conducted or published.

The purpose of our study was to look at registered trials affiliated with London, Ontario, Canada to determine what proportion of these registered studies were ultimately published and to determine factors that may predict a study not being published. We found that fewer than 40% of these studies were published, which suggests that the published literature affiliated with London, Ontario represents fewer than half of all of its registered studies. We were able to quantify several factors associated with publication that could be addressed to increase publication rates in the future.

Creative Commons License

Creative Commons Attribution-Noncommercial 4.0 License
This work is licensed under a Creative Commons Attribution-Noncommercial-Share Alike 4.0 License.

Share

COinS