One of the best-selling supplements in the UK as well as several other countries is evening primrose oil (EPO). It is available via all sorts of outlets (even respectable pharmacies – or is that supposedly respectable?), and is being promoted for a wide range of conditions, including eczema. The NIH website is optimistic about its efficacy: “Evening primrose oil may have modest benefits for eczema.” Our brand-new Cochrane review was aimed at critically assessing the effects of oral EPO or borage oil (BO) on the symptoms of atopic eczema, and it casts considerable doubt on this somewhat uncritical view.
Here is what we did: We searched six databases as well as online trials registers and checked the bibliographies of included studies for further references to relevant trials. We corresponded with trial investigators and pharmaceutical companies to identify unpublished and ongoing trials. We also performed a separate search for adverse effects. All RCTs investigating oral intake of EPO or BO for eczema were included.
Two experts independently applied eligibility criteria, assessed risk of bias, and extracted data. We pooled dichotomous outcomes using risk ratios (RR), and continuous outcomes using the mean difference (MD). Where possible, we pooled study results using random-effects meta-analysis and tested statistical heterogeneity.
And here is what we found: 27 studies with a total of 1596 participants met our inclusion criteria: 19 studies tested EPO, and 8 studies assessed BO. A meta-analysis of results from 7 studies showed that EPO failed to improve global eczema symptoms as reported by participants and doctors. Treatment with BO also failed to improve global eczema symptoms. 67% of the studies had a low risk of bias for random sequence generation; 44%, for allocation concealment; 59%, for blinding; and 37%, for other biases.
Our conclusions were clear: Oral borage oil and evening primrose oil lack effect on eczema; improvement was similar to respective placebos used in trials. Oral BO and EPO are not effective treatments for eczema.
The very wide-spread notion that EPO is effective for eczema and a range of other conditions was originally promoted by the researcher turned entrepreneur, D F Horrobin, who claimed that several human diseases, including eczema, were due to a lack of fatty acid precursors and could thus be effectively treated with EPO. In the 1980s, Horrobin began to sell EPO supplements without having conclusively demonstrated their safety and efficacy; this led to confiscations and felony indictments in the US. As chief executive of Scotia Pharmaceuticals, Horrobin obtained licences for several EPO-preparations which later were withdrawn for lack of efficacy. Charges of mismanagement and fraud led to Horrobin being ousted as CEO by the board of the company. Later, Horrobin published a positive meta-analysis of EPO for eczema where he excluded the negative results of the largest published trial, but included results of 7 of his own unpublished studies. When scientists asked to examine the data, Horrobin’s legal team convinced the journal to refuse the request.
The evidence for EPO is negative not just for eczema. To the best of my knowledge, there is not a single disease or symptom for which it demonstrably works. Our own review of the data concluded ” EPO has not been established as an effective treatment for any condition”
Our new Cochrane review might help to put this long saga to rest. In my view, it is a fascinating tale of a scientist being blinded by creed and ambition. The results of such errors can be dramatic. Horrobin misled all of us: patients, health care professionals, scientists, regulators, decision makers, businessmen. This caused unnecessary expense and set back research efforts in a multitude of areas. I find the tale also fascinating from other perspectives; for instance, it begs the question why so many ‘respectable’ manufacturers and retailers are still allowed to make money on EPO. Is it not time to debunk the EPO-myth and say it as clearly as possible: EPO helps only those who financially profit from misleading the public?
This post has an odd title and addresses an odd subject. I am sure some people reading it will ask themselves “has he finally gone potty; is he a bit xenophobic, chauvinistic, or what?” I can assure you none of the above is the case.
Since many years, I have been asked to peer-review Chinese systematic reviews and meta-analyses of TCM-trials submitted to various journals and to the Cochrane Collaboration for publication, and I estimate that around 300 such articles are available today. Initially, I thought they were a valuable contribution to our knowledge, particularly for the many of us who cannot read Chinese languages. I hoped they might provide reliable information about this huge and potentially important section of the TCM-evidence. After doing this type of work for some time, I became more and more frustrated; now I have decided not to accept this task any longer – not because it is too much trouble, but because I have come to the conclusion that these articles are far less helpful than I had once assumed; in fact, I now fear that they are counter-productive.
In order to better understand what I mean, it might be best to use an example; this recent systematic review seems as good for that purpose as any.
Its Chinese authors “hypothesized that the eligible trials would provide evidence of the effect of Chinese herbs on bone mineral density (BMD) and the therapeutic benefits of Chinese medicine treatment in patients with bone loss“. Randomized controlled trials (RCTs) were thus retrieved for a systematic review from Medline and 8 Chinese databases. The authors identified 12 RCTs involving a total of 1816 patients. The studies compared Chinese herbs with placebo or standard anti-osteoporotic therapy. The pooled data from these RCTs showed that the change of BMD in the spine was more pronounced with Chinese herbs compared to the effects noted with placebo. Also, in the femoral neck, Chinese herbs generated significantly higher increments of BMD compared to placebo. Compared to conventional anti-osteoporotic drugs, Chinese herbs generated greater BMD changes.
In their abstract, the part on the paper that most readers access, the authors reached the following conclusions: “Our results demonstrated that Chinese herb significantly increased lumbar spine BMD as compared to the placebo or other standard anti-osteoporotic drugs.” In the article itself, we find this more detailed conclusion: “We conclude that Chinese herbs substantially increased BMD of the lumbar spine compared to placebo or anti-osteoporotic drugs as indicated in the current clinical reports on osteoporosis treatment. Long term of Chinese herbs over 12 months of treatment duration may increase BMD in the hip more effectively. However, further studies are needed to corroborate the positive effect of increasing the duration of Chinese herbs on outcome as the results in this analysis are based on indirect comparisons. To date there are no studies available that compare Chinese herbs, Chinese herbs plus anti-osteoporotic drugs, and anti-osteoporotic drug versus placebo in a factorial design. Consequently, we are unable to draw any conclusions on the possible superiority of Chinese herbs plus anti-osteoporotic drug versus anti-osteoporotic drug or Chinese herb alone in the context of BMD.“
Most readers will feel that this evidence is quite impressive and amazingly solid; they might therefore advocate routinely using Chinese herbs for the common and difficult to treat problem of osteoporosis. The integration of TCM might avoid lots of human suffering, prolong the life of many elderly patients, and save us all a lot of money. Why then am I not at all convinced?
The first thing to notice is the fact that we do not really know which of the ~7000 different Chinese herbs should be used. The article tells us surprisingly little about this crucial point. And even, if we manage to study this question in more depth, we are bound to get thoroughly confused; there are simply too many herbal mixtures and patent medicines to easily identify the most promising candidates.
The second and more important hurdle to making sense of these data is the fact that most of the primary studies originate from inaccessible Chinese journals and were published in Chinese languages which, of course, few people in the West can understand. This is entirely our fault, some might argue, but it does mean that we have to believe the authors, take their words at face value, and cannot check the original data. You may think this is fine, after all, the paper has gone through a rigorous peer-review process where it has been thoroughly checked by several top experts in the field. This, however, is a fallacy; like you and me, the peer-reviewers might not read Chinese either! (I don’t, and I reviewed quite a few of these papers; in some instances, I even asked for translations of the originals to do the job properly but this request was understandably turned down) In all likelihood, the above paper and most similar articles have not been properly peer-reviewed at all.
The third and perhaps most crucial point can only be fully appreciated, if we were able to access and understand the primary studies; it relates to the quality of the original RCTs summarised in such systematic reviews. The abstract of the present paper tells us nothing at all about this issue. In the paper, however, we do find a formal assessment of the studies’ risk of bias which shows that the quality of the included RCTs was poor to very poor. We also find a short but revealing sentence: “The reports of all trials mentioned randomization, but only seven described the method of randomization.” This remark is much more significant than it may seem: we have shown that such studies use such terminology in a rather adventurous way; reviewing about 2000 of these allegedly randomised trials, we found that many Chinese authors call a trial “randomised” even in the absence of a control group (one cannot randomise patients and have no control group)! They seem to like the term because it is fashionable and makes publication of their work easier. We thus have good reason to fear that some/many/most of the studies were not RCTs at all.
The fourth issue that needs mentioning is the fact that very close to 100% of all Chinese TCM-trials report positive findings. This means that either TCM is effective for every indication it is tested for (most unlikely, not least because there are many negative non-Chinese trials of TCM), or there is something very fundamentally wrong with Chinese research into TCM. Over the years, I have had several Chinese co-workers in my team and was invariably impressed by their ability to work hard and efficiently; we often discussed the possible reasons for the extraordinary phenomenon of 0% negative Chinese trials. The most plausible answer they offered was this: it would be most impolite for a Chinese researcher to produce findings which contradict the opinion of his/her peers.
In view of these concerns, can we trust the conclusions of such systematic reviews? I don’t think so – and this is why I have problems with research of this nature. If there are good reasons to doubt their conclusions, these reviews might misinform us systematically, they might not further but hinder progress, and they might send us up the garden path. This could well be in the commercial interest of the Chinese multi-billion dollar TCM-industry, but it would certainly not be in the interest of patients and good health care.