The number of articles (squares) and the number of alternative medicine articles (triangles) cited in MEDLINE, 1966 through 1996. Trend lines (curved solid lines), based on mean values, are included for illustrative purposes only.
The percentage of alternative medicine articles of clinical trials (squares) and articles published in "focus" journals (triangles), as cited in MEDLINE, 1966 through 1996. Trend lines (curved solid lines), based on mean values, are included for illustrative purposes only.
The number of articles (diamonds) and of clinical trial–type articles (triangles) indexed under traditional medicine, acupuncture, homeopathy, and chiropractic in the MEDLINE database, 1966 through 1996. For clinical trial–type articles, a trend line (curved solid lines), based on mean values, is included for illustrative purposes only.
Barnes J, Abbot NC, Harkness EF, Ernst E. Articles on Complementary Medicine in the Mainstream Medical LiteratureAn Investigation of MEDLINE, 1966 Through 1996. Arch Intern Med. 1999;159(15):1721–1725. doi:10.1001/archinte.159.15.1721
To investigate the growth of interest, if any, in complementary or alternative medicine by the professional scientific community from the number of MEDLINE-listed and clinical trial–type articles for January 1, 1966, through December 31, 1996.
Systematic literature searches of the MEDLINE database, using the expanded terms "alternative medicine," "traditional medicine," "acupuncture," "homeopathy," and "chiropractic," were conducted in January 1998 to evaluate the number of all articles. The number of clinical trial–type articles on the above was obtained by conducting searches for those indexed as 1 or more of the following publication types: clinical trial; clinical trial phase 1, 2, 3, or 4; controlled clinical trial; meta-analysis; randomized controlled trial; and limited to "human" trials only.
Articles indexed as alternative medicine formed a small proportion (0.4%) of the total number of MEDLINE-listed articles throughout the period studied. From 1966 through 1996, the total number of articles listed in MEDLINE rose significantly to a peak of 400,000 additions per annum in 1996 (r=0.97; P<.001). By contrast, the number of articles indexed under alternative medicine rose progressively only from 1972 through 1986 and since then has been relatively stable at around 1500 additions per annum. For this period, the proportion of clinical trial–type alternative medicine articles was low (mean, 2.1% per annum) but increased significantly from 1987 through 1996, reaching around 10% of the total in 1996 (r=0.79; P<.001). Patterns of growth in the number of publications for individual therapies have varied during the period studied, and clinical trial–type articles form only a small part of any increase.
Interest in and awareness of complementary medicine among orthodox health care professionals has increased in the past 30 years. The increase in the number and proportion of reports of clinical trials indicates an increasing level of original research activity in complementary medicine and suggests a trend toward an evidence-based approach in this discipline. The cumulative number of clinical trial–type articles is small, however, and more high-quality original research in complementary medicine is required.
AS THE PUBLIC demand for complementary (or "alternative") medicine increases, there are more calls for the integration of complementary medicine into national health care systems. With the increased emphasis on evidence-based medicine,1 decisions on integrating complementary medicine into routine clinical practice should be made on the basis of reliable evidence of efficacy and safety and not public demand.
If the evidence base of complementary medicine is increasing, one would expect to see, first, an increasing awareness of complementary medicine among orthodox health care professionals and, second, increased research activity in complementary medicine in the form of clinical trials. Both are likely to be associated with an increase in the number of articles on complementary medicine appearing in the mainstream medical literature, especially in journals with higher impact factors. The aim of this study was to investigate the growth of professional scientific interest in complementary medicine from the appearance of relevant publications in the MEDLINE database from January 1, 1966, through December 31, 1996, and to determine whether the number of clinical trials (the mainstay of evidence-based medicine) has increased.
Systematic computerized literature searches were conducted in January 1998 of MEDLINE (using Ovid software; Ovid Technologies Inc, New York, NY) for 1966 through 1996. Data for 1997 were excluded because indexing would have been incomplete at the time of the study.
The number of articles on complementary medicine was retrieved by expanding the medical subject heading "alternative medicine" (AM) (Table 1). Also, specific literature searches were conducted separately using expanded medical subject heading terms for 4 major complementary therapies listed under the heading AM: "traditional medicine" (which includes mainly traditional forms of herbal medicine), "acupuncture," "homeopathy," and "chiropractic."
For AM in general and each of the 4 therapies in particular, searches were conducted to identify the number of articles in the following categories: all MEDLINE-listed journals; articles indexed as "clinical trial–type articles" (ie, as ≥1 of the following publication types: clinical trial; clinical trial phase 1, 2, 3, or 4; controlled clinical trial; meta-analysis; and randomized controlled trial) and limited to "human" trials only; and a subset of "focus" journals (n=33) comprising general or internal medicine journals with an impact factor of 1 or higher (chosen as they are likely to be read and taken seriously by a large number of health care professionals) (Table 2).
Each of the searches described above was used to identify the number of articles per publication year for 1966 through 1996. For each search, the following were calculated: the total number of articles, the number and proportion of clinical trial–type articles (proportions were calculated as a percentage of the total number of articles), and the number and proportion of articles in the focus journals.
To compare the increase in the number of AM articles with that of MEDLINE-listed articles, a further search was performed to identify the number of all MEDLINE-listed articles per publication year.
All numbers retrieved were taken at face value; no attempt was made to determine if, for example, each acupuncture "hit" was actually an article on acupuncture or if a clinical trial "hit" was a report of a clinical trial.
Data are presented as mean (SD).
Figure 1 shows the number of all MEDLINE-listed articles and MEDLINE-listed AM articles per publication year for 1966 through 1996. Despite year-to-year variations, there is a significant upward trend with time in the number of all new MEDLINE-listed articles (r=0.97; P<.001), providing a cumulative total of 8.55 million articles for 1966 through 1996. The number of new MEDLINE-listed AM articles was low and relatively stable for 1966 through 1971 (mean, 293.7 [29.7] new articles per year), but from 1972 through 1986, there was a progressive increase in the number of new articles—from 423 new AM articles in 1972 to 1847 in 1986. For 1986 onward, the number of new AM articles was again relatively stable (1558.2 [60.4] new articles per year from 1987-1996). The cumulative total of AM articles for 1966 through 1996 was 33,602.
The normalized rate of increase in new AM articles from 1972 through 1986 was significantly greater (unpaired 2-tailed t =11.97; P<.001) than that for all new MEDLINE-listed articles. In absolute terms, however, the rate of increase in the number of new AM articles (mean, 93.9 per year) was low in comparison to the rate of increase of all new MEDLINE-listed articles (mean, 6622.8 per year). By contrast, from 1987 through 1996, the normalized rate of increase was significantly greater for all new MEDLINE articles (t=2.15; P<.05) than for new AM articles.
The proportion of AM articles (as a percentage of all new AM articles per year) that were clinical trials from 1966 through 1996 is shown in Figure 2. From 1972 through 1986 (when the rate of new AM articles was increasing annually), this proportion was low (2.3% for the 15-year period). From 1987 through 1996, however (when the number of new AM articles was not increasing annually), there was a significant increase in the proportion of new AM articles that were indexed as clinical trials (r=0.79; P<.001). Figure 2 also shows that the proportion of AM articles in the focus journals has varied, peaking at 9.2% in 1974, but from 1976 onward, varying between 1.9% and 7.3% per year.
Figure 3 shows the number of new MEDLINE-listed articles for 4 major complementary therapies and the number of these that were indexed as clinical trials for 1966 through 1996.
The number of new articles indexed as "traditional medicine" (ie, various types of herbal medicine) increased markedly (51-719) from 1966 through 1986. From 1987 to 1996, the mean number of new traditional medicine articles per year has been 383.2 (65.4). The number of these articles that were indexed as clinical trials has shown a significant upward trend with time since 1984 (r=0.46; P<.01), but the absolute number is still relatively small: from 1966 through 1996, 115 new clinical trial–type articles on traditional medicine appeared in MEDLINE.
For "acupuncture" (Figure 3), few new articles appeared in MEDLINE from 1966 through 1971 (16.7 [6.9] per year), but after 1971, there was a rapid increase to 244 new articles in 1974. Interest has been maintained, with a mean of 240.9 (51.2) new articles per year for 1976 through 1996. This increased interest in acupuncture is also reflected in a small increase in the number of new clinical trial–type articles: from 0.2 (0.4) per year for 1966 through 1972, to 8.4 (3.5) per year for 1973 through 1988. From 1989 through 1995, the number of such articles has averaged 28.2 (7.7) per year but fell to 12 in 1996.
The number of new articles indexed as "homeopathy" was low for 1966 through 1983 (Figure 3; 16.5 [13.2] per year), but after 1983, there was a significant linear rise in the number per year (r=0.80; P<.001). The absolute number is still small, and there have been few clinical trial–type articles on homeopathy (3.5 [2.4] per year for 1984-1996).
A similar pattern can be seen for "chiropractic" (Figure 3), for which there were 21.1 (8.6) MEDLINE-listed articles per year from 1966 through 1983, but for which there has been a significant increase in the number of articles (7.3 per year) from 1984 through 1996 (r=0.75; P<.001). From 1988 through 1996, 102.9 (9.4) new articles on chiropractic per year appeared in MEDLINE, although the mean number of new clinical trial–type articles for the same period was 6.7 (2.1) per year.
This study investigated the inclusion of articles on complementary medicine in the mainstream medical literature MEDLINE database during a 31-year period. The number obtained is dependent on the structure and accuracy of MEDLINE indexing and may not include all MEDLINE-listed articles relevant to complementary medicine: eg, the medical subject heading terms "plants, medicinal" and "drugs, Chinese herbal" are not included under the expanded term "alternative medicine." Some non–complementary medicine articles, incorrectly indexed under AM, may even have been included in the total, and not all clinical trial–type articles may have been reports of clinical trials.
The number of AM articles, scientific or otherwise, retrieved by our searches—1682 clinical trial type, and 33,602 total for 1966 through 1996—is almost certainly an underestimate of the number of articles that exist. For example, almost 4000 randomized trial reports, the sources of which include unpublished theses to simple conference reports, are held on the Cochrane Collaboration Complementary Medicine Registry.2 Previous research has shown that 81% of these are listed on MEDLINE, but that of these only about one third are indexed with both "randomized controlled trial" and an AM key word.2 Our study was concerned, however, with the integration of valid, peer-reviewed information on complementary medicine in the mainstream medical literature, and for this purpose, MEDLINE was the premier single source, despite some recognized limitations.3
Nevertheless, the findings show that new articles indexed as AM have entered the database in 3 phases from 1966 through 1996, in contrast to the number of all new MEDLINE articles, which has risen steadily. The greater number of AM articles being listed in MEDLINE annually (about 1500 per annum for 1986-1996 compared with 300 per annum for 1966-1971) may reflect an increase in indexing activity with regard to articles on AM, an increasing awareness of and interest in complementary medicine among health care professionals and journal editors, or both. The cumulative number of AM articles (33,602 in 31 years), however, is only 0.4% of the total number of MEDLINE-listed articles (8.55 million to the end of 1996). Certainly, that the editors of the American Medical Association's journals have selected complementary medicine and AM as the subject for coordinated theme issues4 demonstrates the need for knowledge in this area and indicates that the subject has not been adequately addressed so far, given the number of consultations (425 million in 1990) estimated to occur between AM practitioners and adults in the United States.5 The significant increase in AM articles annually from 1972 through 1986 (when the rise in the number of AM articles was greater than the rise in the number of articles cited by MEDLINE generally) is due only partly to an increase in the number of clinical trial–type articles (ie, those representing original clinical research). It is, perhaps, more likely to be due to a large increase in the number of opinion or general interest articles on complementary medicine and basic research articles on specific complementary therapies—eg, herbal or traditional medicines for which a large body of scientific literature on analytical, in vitro, and in vivo (but not human) studies exists.
The "plateau" in the number of new AM articles at an average of 1558 per year from 1987 through 1996 could be a result of several factors. It may reflect a reduction in the rate of submission of AM papers to MEDLINE-listed journals in favor of more specialized AM journals, many of which are not MEDLINE-listed. Also, it may suggest that mainstream medical journals, while maintaining a level of coverage on AM, are unwilling or unable to devote more coverage to the issue.
The linear increase in the number of new clinical trial–type AM articles since 1987 may indicate movement toward an evidence-based approach, possibly driven by an increased requirement for evidence by health care providers and health insurance companies and/or an increase in funding available for clinical research in complementary medicine, eg, to the Office of Alternative Medicine at the National Institutes of Health, Bethesda, Md. The cumulative number of clinical trial–type articles (1682 for 1966-1996), however, is small (around 5% of all AM articles for 1966-1996), indicating that the integration of evidence-based principles into complementary medicine is still a long way off.
The relative lack of clinical trials in complementary medicine or AM may reflect general difficulty in obtaining funding in this area for researchers not associated with institutions such as the National Institutes of Health. For example, for many manufacturers of complementary remedies, there is little commercial incentive to carry out clinical trials. Most such products are sold as unlicensed supplements for which evidence of efficacy in the form of clinical trials is not required, and many types of complementary remedies cannot easily be patented.
There is no evidence of a trend toward an increasing number of new articles indexed as AM appearing in the focus journals from 1966 through 1996 (Figure 2), although the peak in the early 1970s is prominent (an increase in the number of articles indexed under acupuncture during this period—probably reflecting the interest generated after President Richard Nixon's visit to China in the 1970s—is almost entirely responsible for the peak).6 An explanation for the paucity of AM articles appearing in the focus journals may be that these papers are subject to publication bias from mainstream medical journals during the peer review or editorial process.7 The quality of a paper, however, rather than its subject, has been shown8 to be the prime consideration of orthodox reviewers of manuscripts on complementary medicine, and because it is likely that many clinical trials (and their resulting reports) of complementary therapies are of poor methodological quality,9,10 such papers are rejected by mainstream medical journals, especially those with a high impact factor, on the basis of the quality of the research. It is also possible that researchers of complementary medicine, in the expectation of a fairer review, may submit their work for publication to newer, specialty complementary medicine journals not (at the time of this study) listed on MEDLINE.
The patterns of growth in the number of articles indexed under each of the 4 therapeutic categories have varied during the period studied. The number of articles is highest for "traditional medicine, " a term that, when expanded, includes several terms relating to traditional systems of medicine based largely on the use of herbal remedies (Table 1). For acupuncture, interest has been maintained since the early 1970s but has tailed off since 1994, possibly in favor of other, more outlandish, therapies. For homeopathy and chiropractic, the number of new articles has been highest since 1988 (excepting the number for homeopathy in 1966 when MEDLINE was introduced). This increase could be patient driven or could reflect increasing interest among orthodox health care professionals and increasing professional organization of complementary therapists. Although there has been an increase in the number of clinical trial–type articles from 1966 through 1996 for all 4 therapies, the number of clinical trials is still low, offering little comfort for those demanding evidence of efficacy from well-conducted, randomized controlled trials.
Patterns of inclusion of articles on complementary medicine in the mainstream medical literature suggest an increasing interest in and awareness of complementary medicine among orthodox health care professionals from 1972 through 1986, which has since plateaued. In this plateau, however, there has been an increase in the number of clinical trial–type articles, reflecting some small increase in the level of original clinical research activity. The absolute number of clinical trial–type articles indexed as AM is, however, still relatively small, suggesting that more evidence from clinical trials is required for an objective assessment of complementary medicine.
Reprints: Joanne Barnes, BPharm, MRPharmS, Department of Complementary Medicine, Division of Community Health Sciences, School of Postgraduate Medicine and Health Sciences, University of Exeter, 25 Victoria Park Rd, Exeter EX2 4NT, England (e-mail: firstname.lastname@example.org).
Accepted for publication December 21, 1998.
Ms Barnes holds the Boots Research Fellowship. Dr Abbot was funded by the Wellcome Trust (grant 050836/Z/97)
We thank the reviewers for their suggestions and helpful comments on this article.