By GlobalMHSummit.com Research Team. Sources checked October 6, 2026.
You can compare a trial's registry record with its published paper, and doing so shows you something a headline never will: which outcomes the researchers said they would measure, and which ones the paper actually reports. This guide does that for one well-documented example, a Colorado study of cannabis flower and anxiety. It is a lesson in reading research, not an answer about whether CBD works.
The short version for this example: the registry record listed three primary outcomes in April 2018, and the paper reports two of them. Part of the paper's CBD finding, the short-term tension and paranoia results, rests on mood measures that the registry record does not name. That does not mean anything improper happened. It means a careful reader has questions to ask, and this page shows you where to find the answers.
What we compared, and what kind of study it is
The registry record is NCT03491384 on ClinicalTrials.gov, sponsored by the University of Colorado, Boulder. The paper is Bidwell et al., Cannabis and Cannabinoid Research, 2024 (PubMed listing), which cites that registry number.
The two documents describe the study design differently, and the differences matter:
- The registry record lists the study type as observational, with a cohort model and a prospective time perspective, not as an interventional clinical trial. Source: current record. ClinicalTrials.gov's own guide to reading a record explains that observational studies are those where participants are not assigned an intervention by the researchers.
- The paper calls its design quasi-experimental, with a nonequivalent control group. Adults who chose to use cannabis were randomly assigned to one of three legal-market cannabis flower products (THC-dominant, about equal THC and CBD, or CBD-dominant), while a comparison group of non-users was self-selected. There was no placebo, and participants knew which product they bought. Source: paper, Methods and Limitations.
- The paper uses several labels for itself: “quasi-experimental” in the title, “randomized” in the discussion, and “quasirandomized” in the conclusion. The accurate summary is that product assignment among users was randomized, but this was not a placebo-controlled randomized trial.
This is also legal-market cannabis flower bought by participants in Colorado, not CBD gummies, oils, or isolate products. The CBD-dominant product in the paper was about 24% CBD with under 1% THC flower. Source: paper, Abstract.
Registration-to-paper outcome comparison
The record's History tab shows nine versions. Version 1 was submitted 2018-04-05 and listed three primary outcomes. The change list for versions 2 through 7 shows no edits to outcome measures, and we read version 7 (2023-10-13) to confirm its outcomes match version 1. The changes to outcome measures appear in the versions submitted 2024-12-03 and 2025-08-25, both after the paper's January 2024 online publication. Source: record history; version 1; version 7.
Here is the comparison, outcome by outcome. Each item gives what the registry said and what the paper reports.
Registered primary outcomes
- Anxiety on the DASS-21 questionnaire. Registered in 2018 as a change in anxiety and negative affect across all three DASS subscales, to be tested against blood THC and CBD levels. The current record narrows the primary outcome to the anxiety subscale. The paper reports all three subscales. All groups, including non-users, improved over four weeks. Among cannabis users, the anxiety score fell more in the CBD-dominant group than in the THC-dominant group, a difference of about 1 point (reported p=0.02) on a scale the registry describes as running 0 to 42. When non-users were included, the paper reports no significant group or group-by-time effect on any subscale. The paper analyzes results by assigned product group; in the text we read, we did not see DASS change modeled against blood levels. Source: paper, Results.
- Inflammation (blood cytokine levels). Registered as a primary outcome from version 1 onward, and the registry's Results tab lists it among the posted outcomes. The paper's text does not mention cytokines or inflammation. We did not open the registry's posted data for this outcome or the paper's supplementary tables, so we cannot say what the data show. Source: results tab.
- Patient Global Impression of Change (PGIC), a one-item rating of how much anxiety has changed. Registered as primary. The paper reports it. All groups reported improvement, and the three cannabis groups reported more improvement than non-users but did not differ from each other. The authors note that expectancy may explain the user versus non-user gap, since non-users made no study-related change. One detail differs: the current record says PGIC was given only once, at four weeks, while the paper says it was measured at two and four weeks. Source: paper, Methods and Discussion.
Measures the paper emphasizes that the registry does not name
- Profile of Mood States (tension, paranoia, elation), measured right after use. The paper's acute finding, that tension fell after CBD-dominant use but not after the other products, comes from this measure. The registry record versions we read do not list it by name. In versions 1 through 7, a secondary outcome called Depression and Mood refers generally to self-reported mood states, so whether this counts as covered is a judgment call. The protocol document posted on the record (version 21, dated June 1, 2023) does describe this instrument as part of data collection. Source: protocol PDF.
- Addiction Research Center Inventory (subjective drug effects) and blood cannabinoid levels after use. Reported in the paper as acute outcomes. The protocol describes the inventory; the registry versions we read do not name it as an outcome.
Registered secondary outcomes the paper does not report
- The registry lists cognitive testing, a negative-affect induction task, balance and motor testing, physical activity, and general health and wellbeing among secondary outcomes. The paper does not report them, although it states that it reports all measures in the study. The record's publication list includes later papers from the same study, and their titles suggest they cover some of these outcomes. We have not read those papers. Source: record, Study Plan and Publications.
What this comparison can and cannot tell you
It can show you whether a paper's headline result comes from an outcome the researchers named in advance, whether the registered design matches the published one, and whether outcomes disappeared or appeared along the way. It is a transparency check. A mismatch is not automatically a problem: one study can produce several papers, and registry records can be updated over time. It is a prompt to ask questions.
It cannot tell you whether CBD reduces anxiety. For this example specifically:
- With no placebo group and participants who knew their product, expectations could shape self-reported results. The authors say so in their limitations section.
- The non-user comparison group was self-selected, so the groups may differ in ways the study did not measure.
- Use was self-directed (“ad libitum”), so doses varied, and long-term exposure was self-reported.
- The study tested cannabis flower in adults who wanted to use cannabis for anxiety (the registry lists ages 21 to 70). It says nothing about gummies, oils, or isolates, or about people outside that group.
- A registry listing is not an endorsement. ClinicalTrials.gov states that the U.S. government does not review or approve the safety and science of all studies listed there, and this record lists no FDA-regulated drug product. A seller citing a registry number has not shown that its product works. Source: record banner and Drug and device information.
What is still missing or unresolved
- Timing. The paper says recruitment began in March 2017. The registry record was first submitted on 2018-02-28 and first posted on 2018-04-09, and the paper calls the study preregistered. The documents do not explain the gap, which could reflect screening versus formal study start. Only the authors can say.
- Enrollment. The record lists 210 (estimated) in 2018, 503 (actual) in the October 2023 version, and 361 (actual) now. The paper analyzed 300 people. We did not find the reasons in the record.
- The inflammation outcome. Where, if anywhere, those results are published is not clear from the paper.
- The PGIC timing difference between the record and the paper.
- Supplementary tables and the registry's posted result tables were not reviewed for this page.
How to run this check on any CBD study
- Find the registry number (it starts with NCT) in the paper's abstract or methods. If there is no number, that is worth asking about.
- On ClinicalTrials.gov, check Study Type. Interventional and observational studies answer different questions.
- Read the Primary Outcome Measures, which the site describes as the most important measures. See the how-to-read guide for what each section means.
- Open the Record History tab and read the earliest version. Note the date and whether outcomes changed later, and whether the changes came before or after the paper.
- List which registered outcomes the paper reports, which it leaves out, and which new ones it adds. Then ask whether the paper's main claim rests on a registered outcome.
For a broader checklist, see our guide to reading mental-health research headlines and our explainer on which measure a well-being study used.
Who to ask next
If you have anxiety symptoms or are wondering about a CBD or cannabis product for them, that is a conversation for a licensed clinician. Please do not start, stop, or change any medicine on the basis of a study summary. If you take prescription medicines, a pharmacist or prescriber can check for interactions; our questions to ask a pharmacist and appointment preparation guide can help you get ready. Questions about whether a product is lawful to sell or what its label may claim belong with the relevant regulator, such as the FDA or your state agency. Questions about a specific trial's missing outcomes are best sent to the paper's corresponding author or the journal; ClinicalTrials.gov does not list a study contact for completed studies, and this one lists none.
Sources
- ClinicalTrials.gov record NCT03491384, current version (last update posted 2025-08-27). Checked October 6, 2026.
- Record version 1 (submitted 2018-04-05) and version 7 (submitted 2023-10-13). Checked October 6, 2026.
- Record results tab. Checked October 6, 2026.
- ClinicalTrials.gov, How to Read a Study Record. Checked October 6, 2026.
- Bidwell LC, Martin-Willett R, Skrzynski C, et al. Acute and Extended Anxiolytic Effects of Cannabidiol in Cannabis Flower: A Quasi-Experimental ad libitum Use Study. Cannabis Cannabinoid Res. 2024;9(4):1015-1027. doi:10.1089/can.2023.0187. Full text; PubMed. Checked October 6, 2026.
- Study protocol and statistical analysis plan, version 21 (June 1, 2023), posted on the record. Checked October 6, 2026.
This article is educational and is not medical advice. It does not diagnose or treat any condition, recommend any product, or suggest starting or stopping any medicine. Global MH Summit (GlobalMHSummit.com) is an independent publication. It is not affiliated with any university, hospital, clinic, government agency, or event organizer, and it does not provide medical care. See our medical disclaimer and editorial standards.