We searched the PubMed MEDLINE database in April 2010 using the methods reported by Arangeui et al.4 The search strategy used for the different groups was similar (Table 1).

Using the filter terms Dermat*, Reumat*, and Rheumat* (as appropriate) in the first author affiliation field, we searched for research articles by Spanish, British, and French dermatology departments and Spanish rheumatology departments that were indexed in Medline in 2008 or published online in the same year in the case of electronic-only journals.

Articles which clearly did not deal with dermatology or rheumatology topics were excluded. This exclusion was justified because the filters used also yield publications by large, multidisciplinary departments containing the terms Dermat*, Reumat*, or Rheumat*.4 Also excluded were articles that were classified as non-research articles.

The documents retrieved were initially grouped into 3 categories:

• 1.

  Clinical research articles

• 2.

  Basic research articles

• 3.

  Non-research articles

Clinical research articles were further classified by level of scientific evidence using a reproducible and previously validated definition of clinical research.4Table 2 shows the definitions used to distinguish between clinical and basic research articles. Articles that did not correspond to either of these categories were classified as non-research articles.

The observers were not blinded to the names of the authors as it has been indicated that this does not produce bias in the analysis of manuscripts.5 The research articles were identified using the PubMed identifier. The following details were recorded for each article: a) research group (specialty/country); b) formulation of a question relevant to clinical practice; c) focus on patients, healthy individuals, samples, or health care systems; d) level of scientific evidence (according to the Oxford Centre for Evidence-Based Medicine (CEBM)6; e) area of clinical research, where relevant (prevalence, etiology, diagnosis, prevention, pharmacological and nonpharmacological treatments, quality of life, economics, and health care systems); f) financial disclosures; g) presence and type of funding; and h) details of the role of the sponsor (ie, level of control over study data and publication).

According to the CEBM classification system used, articles with a score of 3 or less were considered to have a high level of scientific evidence. A score of 4 was assigned to epidemiological studies and case series, and a score of 5 was assigned to case reports, commentaries, nonsystematic reviews, etc. Because these articles are more prone to random errors and errors due to bias and confounding factors, they were classified as non-research articles.

We analyzed the presence and type of funding by studying different sections of the articles:

• 1.

  Affiliations. It was considered that funding existed if at least 1 of the authors worked at a public institution, a foundation-type institution, or a pharmaceutical company. We considered that the production of all articles involves a cost (staining of a biopsy sample, use of disposable surgical material etc.) that would be identical for public and private hospitals and for university departments. What we were interested in assessing, however, was the presence of additional, or external, funding, ie, funding other than that used to cover costs associated with routine practice. Accordingly, external funding was not considered to exist if all the authors worked at a public or private hospital or university.

• 2.

  Financial disclosure section. This was considered to be any part of the article that included the term funded by (or similar). When specific mention of funding was made, a record was made of the type of funding organization.

• 3.

  Conflicts of interest. Funding was considered to exist if there was mention of a relationship with a public institution, a foundation-type institution, or industry, except when it was explicitly stated that this relationship had had no direct bearing on the study.

• 4.

  Acknowledgments. Funding was considered to exist if support from a public institution, a foundation-type institution, or industry was acknowledged.

• 5.

  Material and methods. Funding was considered to exist if the material and methods section mentioned that the study material, or any of the material used to conduct the study, had been donated by a public or foundation-type institution or by industry.

Financial disclosure was regarded as adequate when there was an explicit financial disclosure statement in addition to a statement regarding the role played by the sponsor.7,8

The different types of funding were recorded on a Microsoft Excel spreadsheet together with links to the websites of the sponsors or to pages containing information about them. All sponsors were analyzed independently by 2 observers blinded to the corresponding articles. A third person intervened in the few cases in which were there discrepancies regarding classification. Three categories of funding were used:

• a)

  Public funding: grants or funding received from government institutions such as national health institutes or local, regional, national, or international bodies, or affiliation with these institutions.

• b)

  Foundation-type funding: grants or funding received from independent not-for-profit bodies such as foundations, institutes, scientific societies, patient associations, or similar, or affiliation with these bodies.

• c)

  Industry funding: funding received from pharmaceutical companies or from manufacturers of the material studied.

The general study data were recorded on Microsoft Excel spreadsheets and analyzed using Stata 10 (StatCorp LP, 2009). The tests used were χ2, the Fisher exact test, and logistic regression.

The search strategy retrieved 918 articles. After applying the inclusion and exclusion criteria, the final number of articles was 348, of which 96 were publications by Spanish dermatology departments (Fig. 1). The breakdown of clinical and basic research articles published by each group is shown in Table 3.

Figure 1.

Flow diagram showing articles retrieved by search strategy and reasons for exclusion.

(0.6MB).

There was evidence of external funding in 33.7% (31/92) of clinical research articles and in 100% (4/4) of basic research articles. A breakdown of the types of funding is given in Table 3.

The differences between clinical and basic research were statistically significant (P=.016, Fischer exact test).

External funding was present in 36.4% (35/96) of all research articles produced by Spanish dermatology departments. The corresponding percentages for the other groups were 80% for Spanish rheumatology departments (40/50 articles), 44.9% (44/98) for French dermatology departments, and 69.2% (72/104) for British dermatology departments. The differences were statistically significant for Spanish rheumatology departments and British dermatology departments compared to Spanish dermatology departments (P<.001, χ2). The breakdown of percentages by clinical and basic research for each group is shown in Table 3. The percentages for public and industry funding were similar for clinical research. Foundation-type funding was associated with public or industry funding in 38.1% of articles (8/21) published by Spanish dermatology departments. The corresponding figures were 74.2% (23/31) for Spanish rheumatology departments (P=.02, Fischer exact test), 85.7% (18/21) for French dermatology departments (P=.004, Fischer exact test), and 45.3% (29/64) for British dermatology departments (not significant, Fischer exact test).

There was a clear relationship between the presence of external funding and level of scientific evidence of clinical research articles. In the case of articles produced by Spanish dermatology departments, 76.9% (10/13) of the studies with the highest level of evidence (score of ≤3) had received external funding compared to just 26.5% (21/79) of level-4 studies. The differences were statistically significant (P=.003, χ2). The differences persisted when the groups were analyzed together, with a higher prevalence of external funding (overall and specific types) observed in articles with a high level of evidence (P<.001, χ2) (Figs. 2 and 3).

Figure 2.

Percentage of clinical research (all groups) that received external funding according to level of scientific evidence.

(0.09MB).

Figure 3.

Type of funding allocated to clinical research (all groups) according to level of scientific evidence.

(0.1MB).

Based on the assumption that the production of articles with a high level of scientific evidence is a desirable goal, the fact that studies with a higher level of evidence were found to have received more funding led us to investigate in more detail the factors that were associated with high levels of evidence. To do this, we performed logistic regression. The first factor included in the model was foundation-type funding, which yielded an odds ratio of 2.5 (P<.001)

On adding public and industry funding, we saw that the model improved significantly (P<.001, likelihood ratio test), indicating that while foundation-type funding exerted an independent effect on quality of scientific evidence, it was also associated with funding by public institutions or industry. The model again improved significantly with the addition of the research group, and this improvement was independent of type of funding. No improvements were seen with the addition of area of research or with interactions between the previous factors.

The final model explained 22% of the variance in level of evidence (pseudo R2) (Fig. 4).

Figure 4.

Factors associated with a high level of scientific evidence. Logistic regression model. The figure shows the effect of successively adding different factors to the model. OR indicates odds ratio.

(0.29MB).

On analyzing the clinical research articles published by Spanish dermatology departments, we observed a higher percentage of funding allocated to studies dealing with quality of life (P=.01, Fisher exact test).

Because of the small number of articles in each research area, we decided to analyze the groups as a whole, and observed a significant association between a higher percentage of funding by industry and research into pharmacological treatment and quality of life. The different areas of research received a comparable proportion of public and foundation-type funding (Table 4).

Table 5 shows the percentage of articles that included a specific financial disclosure statement for all groups and types of funding.

Table 6 shows the percentage of articles that specified the role played by the sponsor for the different groups studied.