A Null Quality–Effect Correlation Did Not Resolve Missing Safeguards in the THC Meta-Analysis
Only six of the 31 publications in the THC cardiovascular review reported randomization and masked outcome assessment. The lack of a significant quality–effect correlation did not establish that missing safeguards were harmless.
The review assessed several methodological features
Its eight-point framework included randomization, blinding, multiple outcomes, dose response, timing and completeness of data. The authors found no statistically significant association between the overall score and the estimated blood-pressure, heart-rate or blood-flow effects. That was a result of the quality analysis, not confirmation that every study had an equally reliable design.
An overall score can obscure specific uncertainties
Different safeguards address different possible biases. A similar total score does not mean two studies have the same strengths, and a correlation across scores may have limited ability to identify the influence of a particular missing safeguard. The null analysis cannot supply evidence that an unreported procedure was performed adequately.
Keep appraisal separate from dismissal
The methodological limitations do not prove that every reported physiological effect was biased. They constrain confidence in the size and generalizability of the synthesis. A useful evidence appraisal reports the low frequency of key safeguards and the null correlation together, rather than using either as a reason to ignore the other. This preserves the review’s caution without inventing a numerical correction for bias.
Related analysis: the study’s broader findings and limitations.
