Interest in vitamins, minerals, herbs and other dietary supplements (DS) has grown markedly in the US since the early 1990's 12345678910. DS are defined by the US Food and Drug Administration as "vitamins, minerals, herbs or other botanicals, amino acids, and substances such as enzymes, organ tissues, glandulars, and metabolites." DS are commonly used 231112. In certain cases, DS are medically recommended (e.g., folic acid for women of child-bearing age 13). However, for most supplements there are few official recommendations, and patients often use DS without informing their physician 21415. Physicians have been urged to discuss with their patients the effectiveness and safety of DS, particularly when combining use of DS with prescription medications 161718192021.

In response to citizen demand and the high prevalence of DS use, the US National Institutes of Health have funded research to evaluate the safety and effectiveness of DS 22. An increasing number of research papers about DS have been published recently, and a number of journals specifically devoted to CAM research and education have appeared. However, most primary care clinicians subscribe to established and highly visible general medical journals; articles in these journals are frequently cited in professional practice guidelines and by the media to guide the general public. Many medical journals are supported in part by advertising revenues; however, we are unaware of research evaluating the relationship between pharmaceutical advertising and publication about natural health products, which may in some cases be viewed as competitive with pharmaceutical products.

In general, the aim of advertising is to increase demand for and purchase advertised products. Direct to consumer pharmaceutical advertising affects consumer behavior 2324 ; advertising in medical journals and other marketing aimed at clinicians affects clinician prescribing behavior 2526. Pharmaceutical funding of research also appears to influence authors' interpretation of results (conclusions) about the effectiveness of these products 27. It is possible that advertising funds may influence journals and their paid staff 25. The potential conflicts of interest between advertisers, editors and authors have resulted in the emergence of guidelines and rules about disclosure of potential conflicts of interest, particularly for authors writing about the effectiveness and safety of specific products 28293031323334.

Because of the importance of established medical journals in informing and influencing professional and public behavior, the question of unintended bias favoring advertisers in making editorial decisions is of great public health as well as business interest. The purpose of this pilot study was to begin to explore the relationship between pharmaceutical advertising and the publication of articles about DS. We expected that journals with more pharmaceutical advertising would publish fewer articles about DS because DS may at times be viewed as competitors to pharmaceutical products; furthermore, we expected that those with more pharmaceutical advertisements (pharmads) would publish more articles suggesting that DS were unsafe or ineffective than journals with fewer pharmads. As a secondary question, we assessed publications about DS in three journals devoted to complementary and alternative medicine (CAM); we expected CAM journals would have fewer pharmads and more DS articles; finally, we expected that fewer of the articles in CAM journals would be unfavorable about DS than articles in the general medical and primary care journals.

The number of issues reviewed for each of the 11 primary journals ranged from 12 to 52, with an average of 30 issues per journal (Table 1). The average number of pages per issue ranged from 56 to 217. The average number of all advertising pages (all-ads) ranged from 3.9 to 87.8 per issue. The average number of pharmaceutical ads (pharmads) varied from 0.15 to over 60 pages per issue and from 0.001 to 0.441 pharmads per average number of journal pages per issue. We arbitrarily grouped journals into those with less than ten pharmads in an average issue (the fewest pharmads group; two journals with an average impact factor 5.9); those with more than 40 pages of pharmaceutical ads per issue (the most pharmads group; two journals with an average impact factor 18.5), and those that had between 10–39 pages per issue (the middle ads group; seven journals with an average impact factor 6.9). Although we did not set out to review the number of pages devoted to DS advertising, reviewers noted that ads for DS such as Centrum^® and Caltrate^® were very rare among these 11 journals, and there were no ads for folate, individual vitamins or for non-vitamin/mineral DS such as glucosamine, ginkgo, or garlic in these 11 journals.

The number of articles of any type about any DS in each journal ranged from four to 61 (Table 2). The journals with the most pharmads published fewer major articles about DS per issue than journals with a middle number of ads (0.08 versus 0.21 per major articles per issue, P < 0.05); and journals with the middle number of pharmads published fewer major DS articles per issue than journals with the fewest pharmads (0.21 versus 0.43, P < 0.05), consistent with a dose-effect response. There was no apparent relationship between journals' impact factor and either their number of ads or their number of publications about DS. None of the 35 articles published about DS in the two journals with the most pharmads were clinical trials or observational cohort studies.

The DS that were covered the most often in these medical journals were (Table 3): folate and other B vitamins; calcium and vitamin D; iron; essential fatty acids such as omega three fatty acids; and caffeinated beverages such as coffee and green tea. There were fewer articles about probiotics, glucosamine, individual herbs (garlic, ginkgo, ginseng, St. Johns wort, lavender, tea tree oil) or combination herbal products.

Safety concerns were raised less often in clinical trials or cohort studies than in other articles. Specifically, concerns about unsafe DS were cited in 1/44 (2%) clinical trials or cohort studies versus 3/34 (10%) of editorials or reviews, and 20/177 (11%) of other kinds of articles. Journals with the most pharmads were significantly more likely to publish major articles concluding that DS were unsafe than journals with middle or fewest pharmads (67% versus 7% and 4%, respectively, P < 0.005 Table 4).

In the 11 primary journals, 68/254 (26%) articles about DS concluded they were ineffective; only 10 of these articles were clinical trials or cohort studies. Conclusions about ineffective DS were less common in journals with fewest pharmads than in journals with more pharmads. For example, half (50%) of the editorials from journals with the most pharmads concluded that DS were ineffective, compared with 33% of editorials from journals with middle amounts of pharmaceutical advertising and 27% of editorials from the journals with the least pharmaceutical advertising (Table 4); these differences were not statistically significant.

For the secondary question about the three CAM journals, the number of advertisements ranged from 5.7 to 24 pages per issue (Table 1). There were no advertisements for prescription pharmaceuticals in any of the CAM journals. However, there were ads for DS; the number of pages of ads for DS ranged from 0 to 12.3 with an overall average of 5.5 ads for DS per issue. The CAM journals carried significantly more articles about DS than any group of the primary journals (P < 0.01). The CAM journals published over twice as many clinical trials or cohort studies as the journals with the fewest pharmads (0.5 versus 0.22 per issue); the number of major articles per issue was significantly higher among CAM journals than among the general medical journals with the fewest ads (0.86 versus 0.43 P < 0.05). Although CAM journals published more original science about DS, the proportion of major articles concluding that DS were unsafe or ineffective was not significantly different from general journals with middle or fewest pharmads (Table 4).

This is the first study to examine the relationship between advertising in medical journals and the publication of articles in those journals about DS. The results are consistent with the hypothesis that pharmaceutical advertising biases journals against non-drug therapies. Journals with the most pharmads per issue published significantly fewer major articles about DS than journals with fewer pharmads. For example, the two journals with the most pharmads published no clinical trials or cohort studies about DS. Furthermore, journals with the most pharmads were more likely to publish articles concluding that DS were unsafe than journals with fewer pharmads. CAM journals had no pharmaceutical ads, and they published significantly more original science articles about DS; however, they were not significantly less likely to publish major articles suggesting that DS were unsafe or ineffective than general medical journals with middle or few pharmads.

These results are generally consistent with a substantial body of research suggesting that advertising influences behavior. Advertising influences consumer behavior 2324. Advertising, particularly in the forms of gifts and samples, also influences physician prescribing behavior 263536. This study extends those findings, and it supports growing efforts to minimize potential conflicts of interest or at least make them more transparent, if eliminating them entirely is not feasible 283738.

It is possible that there might have been greater apparent impact on publication decisions if the analysis had been restricted to DS that are typically used to replace pharmaceutical products. The great majority of the articles about DS reviewed here were about caffeinated beverages and vitamins or other essential nutrients (minerals and essential fatty acids) which are not competitors of prescription medications. For the most part, recent research suggests that the DS most often used, including multivitamins, calcium, iron and folate, are those often recommended by professionals or those known to be deficient in the diet 394041. Furthermore, many vitamins are manufactured by pharmaceutical companies, thus reducing these companies' potential bias against DS. For example, Centrum^® products and Caltrate^® are manufactured by Wyeth, the maker of Advil^®, Alavert^®, Effexor^® and Embrel^® among many other products. These facts would tend to favor the null hypothesis.

On the other hand, non-vitamin/mineral supplements, such as glucosamine, echinacea, garlic, ginseng, and ginkgo are also frequently used by the public, often concurrently with prescription medications 1220, yet the primary 11 journals included in the survey published few research studies, editorials or reviews on these DS. Future research will need to focus on publication about DS that may be viewed as greater threats to the pharmaceutical industry (such as non-vitamin/mineral products that are used by the public to promote weight loss, improve sexual or athletic performance, relieve joint or back pain, or to improve memory, cardiovascular health or GI health). Such research will need a large sample size of many types of journals; furthermore, such research will also need to account for potential confounding by research funding by DS manufacturers.

This pilot study has several limitations. It included only a few of the many clinical medical journals available and used only the American library version of these journals, but it did focus on those with very high readership that are often quoted by professional organizations, the media and policymakers. Results from the smaller secondary analysis of the CAM journals would be enhanced by larger sample of more diverse CAM publications. The review did not specifically address the funding for different trials or evaluate the methodologic quality of the published trials. This study was not a survey of journal editors regarding the number of submissions of different types they receive or how they decide which articles to include. An alternative hypothesis that might explain these findings is that journals with a high number of pharmaceutical ads receive few submissions about dietary supplements; furthermore, it is possible that a large number of articles about dietary supplements are of poor quality and do not deserve publication. It is also possible that the observed associations are due to another underlying factor ; for example, one recent study reported that in high impact journals, European journals were approximately twice as likely as American journals to publish positive articles about complementary therapies 42. In addition, this study used the number of issues per journal as the study denominator rather than the total number of articles per issue; future studies could use the number of articles per issue as the denominator to address the related question of whether journals with more advertising per issue also have fewer articles of all types and hence fewer articles about dietary supplements. Future studies on this topic to explore alternative hypotheses for our findings should include a larger number of journals and more diverse journals; additional studies might also survey journal editors to determine the number of articles submitted for consideration for publication and ask about the percentage of submissions related to dietary supplements.

These findings support the hypothesis that in major medical journals, more pharmaceutical advertising is associated with publishing fewer articles about DS and having more negative conclusions about DS safety. While awaiting future definitive studies to confirm these findings in a larger, more diverse sample of journals and to explore alternative explanations, these data support current efforts to reduce conflicts of interest in medical publishing and to make any such conflicts more transparent. The impact of advertising on publications appears to be non-trivial; the ultimate impact of this bias on professional guidelines, health care, and health policy is a matter of great public concern and underscores the need for additional health services research on this topic.

The author(s) declare that they have no competing interests.

KK conceived of the project, designed the data collection form, reviewed data analysis, drafted and revised the manuscript

KH reviewed journals, entered data, assisted with data analysis and manuscript preparation

Both authors read and approved the final manuscript.