(P17-026-20) Application of P-Curve Analysis to Dietary Supplement Clinical Trials: Case Study of Trials of Curcumin Products for Arthritis
Objectives: Turmeric root (Curcuma longa) and its constituents, the curcuminoids, are widely reported to have anti-inflammatory effects. A recent meta-analysis found that 8 studies that met the review criteria "provide scientific evidence that supports the efficacy of turmeric extract...in the treatment of arthritis" but notes the small number of trials and "the methodological quality of the studies were not sufficient to draw definitive conclusions." The objective of this project was to apply p-curve analysis to these trials, to assess the likelihood of biased data reporting or p-hacking, a common source of false positive reports.
Methods: PubMed was searched for English language reports of randomized, placebo-controlled clinical trials of turmeric- or curcuminoid-containing products for osteoarthritis or rheumatoid arthritis (henceforth arthritis) pain or other inflammatory conditions in arthritis patients. Trials in which curcumin was combined with other supplements or drugs were included. Data for primary outcomes from the retrieved papers were used to generate a p-curve. Where multiple primary outcomes were reported, or no primary outcome was specified, results for a well-validated arthritis scale were used for the analysis.
Results: Twelve randomized, double-blind, placebo-controlled clinical trials of curcumin for arthritis were identified and analyzed. Dates of publication ranged from 1991 to 2019, group sizes ranged from 12 to 80, doses, where specified, ranged from 180 to 1500 mg curcuminoids or turmeric extract per day, and treatment duration ranged from 6 to 36 weeks. The p-curve analysis did not reveal a left-skewed curve which would be diagnostic of p-hacking (more p-values concentrated around .04 and fewer low values) and thus does not suggest that p-hacking is present.
Conclusions: Inspection of the distribution of p values for placebo-controlled clinical trials of curcumin-containing products for osteo- and rheumatoid arthritis in the PubMed database do not show evidence of extensive p-hacking.
Funding Sources: NIH Office of Dietary Supplements
Barbara C. Sorkin
Program Director National Institutes of Health North Bethesda, Maryland
Jaime J. Gahche
Nutritional Epidemiologist National Institutes of Health Office of Dietary Supplements Bethesda, Maryland