Rigor in designing and performing biomedical research and the ability to reproduce the findings are top priorities at the National Institutes of Health (NIH). The initiative to focus on rigor and reproducibility of clinical and preclinical studies is summarized in a 2014 paper in Nature by NIH’s Director and Principal Deputy Director, Dr. Collins and Dr. Tabak. In addition, an NIH Web portal offers principles, guidelines, four training modules, and other features.
At the National Center for Complementary and Integrative Health (NCCIH), one aspect of planning and implementing clinical research that we often discuss with grantees and applicants is blinding, also called masking. Ultimately, its goals are to minimize the potential for introducing bias into study results and to maximize results’ objectivity, whether preclinical or clinical. Bias can be introduced into a study by the investigators, other study personnel, and/or study participants. Many people are familiar with single-blinded and double-blinded designs, but maintaining blinding during a study requires careful planning.
When we talk with investigators about blinding, here are a few points that often come up:
- Many of the Center’s studies are not fully blinded because it’s not feasible. For example, if you want to study meditation or massage, chances are that patients will know which intervention they are receiving (i.e., active or control). However, this doesn’t mean that your study team shouldn’t go to appropriate lengths to make sure that most of the team remains blinded and the data are carefully collected and managed.
- Study blinding involves much more than just knowledge of individual participant treatment assignment. Another important aspect concerns who will do your outcomes assessments. This should be an objective, blinded observer, not the team member who delivered the intervention.
- Once you have collected your blinded data, how you handle the data is also important—e.g., people who are supposed to be blinded should not have access to the data once it is collected, at least until the study is over. Unless the team formally planned an early (or interim) analysis before starting the study, data analysis should not be done before the study ends.
- Journals are paying much more attention to the details of blinding in reports of clinical studies, beyond whether a study is just described as single- or double-blinded.
The topic of research bias and blinding is not simple, but it’s a good one to talk about. NCCIH offers resources to help you maximize the rigor of your study. First, there is our Clinical Research Toolbox. Check out the new materials under “Resources for Rigor and Reproducibility,” including two sets of current research standards: (1) a Cochrane Collaboration tool, “Assessing Risk of Bias in Included Studies,” and (2) the Web site for the CONSORT (CONsolidated Standards of Reporting Trials) 2010 guideline. I consider these resources to be very important standards for modern clinical studies. In addition, NCCIH has staff available to you in the Division of Extramural Research who are experts in the planning and implementation of clinical studies.