Assessment of Systematic Reviews Concerning Adverse Effects of Herbal Medicines
Reviewed: Posadzki P, Watson LK, Ernst E. Adverse effects of herbal medicines: an overview of systematic reviews. Clin Med. February 2013;13(1):7-12.
For decades, members of the herbal medicine and conventional medical communities have debated the relative safety of herbs and phytomedicines as a class of consumer goods. Herbal advocates have generally stated that herbs and their medicinal preparations are relatively safe, as evidenced by low adverse events (AEs) in clinical trials, as well as infrequent reporting of AEs to poison control centers and other pharmacovigilance vehicles.
A group of European researchers have attempted to gain further insight into the relative safety of herbal medicines (HMs) by reviewing published systematic reviews (SRs) of controlled clinical trials on numerous herbs and phytomedicines. The authors examined the evidence regarding AEs of HMs used in humans for any clinical condition, as identified in SRs. A search of electronic databases identified 4,366 potentially relevant publications; of those, the authors identified 50 SRs of 50 different HMs that met the study criteria. If an HM was the subject of more than one SR, then the most recent, sound, and independent SR was selected. SRs of multi-herb mixtures were excluded.
SRs were independently rated for quality by two authors using a modified Oxman score, which is an assessment of SRs using a checklist published by Andrew Oxman aimed at evaluating sources of bias involved in a review process.1 Most SRs had only minor methodological weaknesses, although the authors state that quality was “frequently not optimal.” Of SRs with scores < zero (n=five), all scored negative five, the lowest possible Oxman score.
The selected SRs originated from Canada, Denmark, Germany, Italy, South Africa, the United Kingdom, and the United States. They raised a range of safety concerns. The number of primary studies included in the SRs varied from one to 216. The types of primary reports used also varied widely. Some used other SRs as primary reports. One SR used a meta-analytical method.
Mild AEs cited ranged from the most-mentioned AEs, including gastrointestinal upset, nausea, and allergic reactions, to comparatively rare AEs, including pain, constipation, dry mouth, poor appetite, menstrual disorders, muscle spasm, muscle weakness, fatigue, and others. The SRs found that 31 HMs were safe, with only mild, transitory AEs. These include the herbs in Table 1.
Moderately severe AEs cited for 15 HMs included anorexia, blood coagulation abnormalities, severe nausea, electroencephalogram (EEG) changes, inhibition of platelet aggregation, and other conditions. HMs with moderately severe AEs are listed in Table 2.
Severe AEs noted for four HMs included acute psychosis, cerebral hemorrhaging, coma, respiratory arrest, tachycardia, hallucinations, convulsions, renal failure, acute lung injury, hepatitis, hyperkalemia, liver damage or failure, congestive heart failure, carcinoma, seizures, and other conditions up to and including death. The SRs for the following HMs concluded that they were unsafe: belladonna (standardized herb powder; Atropa belladonna, Solanaceae), (Larrea tridentata, Zygophyllaceae), kava (Piper methysticum, Piperaceae), and senna (syn. Cassia senna, Fabaceae). Chaparral was associated with cases of liver dysfunction; kava was presumably included due to reports of liver toxicity associated with some kava preparations; and curiously, despite being included here, senna leaf and/or fruit extracts are widely approved as safe and effective nonprescription stimulant laxative ingredients by regulatory authorities in many developed countries, including the United States.
Some included SRs were not designed as SRs, but as monographs. Nevertheless, according to the authors, an SR of poor quality may have value in raising potential risks. In justifying their use of SRs only, the authors note that a review of all information to date about AEs and HMs would be far beyond the scope of a single journal article. They propose that SRs should also include AEs mentioned in anecdotal data like case reports. While of no use in establishing cause and effect, such evidence may be relevant in assessing safety issues.
- Oxman AD. Checklists for review articles. BMJ. 1994;309:648-651.