Different combinations of the key words “Dupuytren’s contracture,” “Dupuytren’s surgery,” and “Dupuytren’s therapy” were used to identify websites in English only by 5 most popular [21] search engines: Google, Bing, Yahoo, Ask, and AOL. The geographic option in the search engines was switched off to avoid selection bias. For further assessment, we selected the first 100 search results for each search engine based on the assumption that Internet users limit their search to a number far below 100 [15]. To the selected 500 websites, the following noninclusion criteria were applied: websites not specific for Dupuytren disease, those related to articles in scientific journals, duplicates, and websites in language other than English. This resulted in the selection of 118 websites for further assessment.

To assess each website, we used the modified EQIP tool [12], which is a checklist consisting of 36 items and evaluates data in three different categories: (1) content data, (2) identification data, and (3) structure data (Table 1).

The EQIP tool was developed by rating the quality of 73 documents describing medical care procedures used at the University Hospital of Geneva, Switzerland. The assessment rules were defined on 25 documents, and two assessors independently rated the remaining 48 documents. The interrater reliability was very good (kappa statistic=.84), and the intraclass correlation coefficient was as high as .95 [16]. Although the EQIP tool included a 4-option rating scale of “yes,” “partly yes,” “no,” and “NA” (not applicable) in its native form, we decided to use its modified version with a binary scale of “yes” versus “no” or “NA” (ie, no score) after Melloul et al [15]. This is because the answer “partly yes” is, in our opinion, too subjective. Furthermore, there is evidence that associates this answer with low dependability in the assessment of website content [22].

The data were independently assessed by three investigators and divergent results were defined by consensus. The obtained data were entered into a Web-based platform built on the open source content management system Drupal (version 7) [23], which guaranteed a standardized and complete data entry. According to the origin of the information, the 118 websites were categorized into 8 groups: (1) academic center, (2) encyclopedia, (3) hospital, (4) industry, (5) news service (the press), (6) practitioner, (7) professional society, and (8) patient group. Another classification was performed regarding the country of origin of the websites: (1) Australia, (2) Azerbaijan, (3) Canada, (4) France, (5) Germany, (6) New Zealand, (7) Singapore, (8) United Kingdom, and (9) United States.

To assess these risks, items 9 and 10 were applied (Table 1). Item 9 evaluates the description of qualitative risks and side effects or complications of surgical interventions (eg, if the risk of postoperative complications is mentioned on the website). Item 10 assesses the description of the quantitative risks of surgical techniques.

Proportions derived from nominal variables were compared with Fisher or chi-square tests and continuous variables were compared with the Student t test or 1-way analysis of variance (ANOVA) test, where applicable. All P values were 2-sided and considered statistically significant when P<.05. According to the 36 items of the expanded EQIP tool, all 118 websites were scored from 0 to 36. Each item was given equal weight of importance. The 75th quartile was arbitrarily used as a cut-off point to differentiate high-score websites from low-score ones, and we dichotomized the obtained EQIP score as previously performed by Melloul [15]. Statistical analysis was performed with SPSS version 22 for Mac (IBM Corp).

After screening 500 eligible websites, 118 websites were included for qualitative and quantitative analysis with the expanded EQIP tool. The criteria for noninclusion were duplicates and noneligible Web contents.

More than two-thirds (75.4%, 88/118) of all websites originated from the United States, followed by the United Kingdom (14.4%, 16/118). Canada was represented in 3.4% (4/118). Additionally, 23.6% (21/89) of the 89 US websites were rated as high-score websites, which made 65.6% of all (n=32) high-score websites (Figure 1).

Fifty-three websites (44.9%, 53/118) were developed by professional societies, which thus represent the most frequent source of information on Dupuytren disease. Practitioners were the source number 2 with 26 websites (22%, 26/118; Figure 2 ).

The median website score obtained from the EQIP tool was 16 points (interquartile range, IQR: 13-19). The lowest score of 8 points was achieved by one website and the highest score of 28 points by two websites. None of the screened websites provided information on all 36 items of the modified EQIP tool. When the source of medical patient information was concerned, there was no statistically significant difference between scores obtained by different website developers (Figure 3).

Websites above the 75th percentile (with the score of 19 or more) were defined as high-score websites, in contrast to low-score websites (obtaining 18 points or less). A high score was achieved by 32 websites (27.1%, 32/118) and a low score by 86 websites (72.9%, 86/118) (Figure 4 ).

We defined a top rated website with a score above the 95th percentile (Table 2 ). The top rated websites came from the United Kingdom (n=2) and from the United States (n=4). The highest score reported was 28, ex aequo from a British professional society and from an American professional society.

More than two-thirds (68.6%, 81/118) of the websites screened were published in 2013 in contrast to 37 websites published from 1990 to 2012. Within the passing time, the EQIP-based quality of the newly introduced websites did not increase significantly, as shown in Figure 5.

The most important findings of the study were, first, that the overall quality of patient information on Dupuytren disease evaluated with a validated tool was poor. Second, the source of medical patient information did not influence the scores obtained by the websites. Third, none of the screened websites provided information on all 36 items of the modified EQIP tool, and the high-score websites represented only a quarter of the screened websites. Finally, the quality of the newly developed websites did not increase with passing time.

The Internet presents a global, easily accessible, and unlimited source of any kind of information, and medical issues is one of the most searched topics. It is also an uncontrolled space, allowing anyone to put any kind of information out there, and also that of unknown accuracy. This may expose patients to the risk of getting wrong information and impact their further therapeutic decisions. These concerns led various authors to investigate the accuracy of the medical information for patients in different medical disciplines. [15,16,24-26]

The systematic evaluation of the quality of Internet information on Dupuytren disease is sporadically present in the literature in contrast to the information on other common hand pathologies.

Sproule et al [19] conducted in 2003 an evaluation of 172 websites containing medical information on 3 common hand pathologies such as Dupuytren disease, carpal tunnel syndrome, and trigger finger. The published patient information was evaluated for completeness and accuracy using a scoring system developed by the authors. The findings of that study in terms of those two evaluation criteria showed substantial shortcomings in most websites. In contrast to the methodology of our study, Sproule et al did not use a validated evaluation scoring system.

Almost a decade later, Kelly et al [20] performed an Internet search of “Dupuytren’s disease” using the most popular search engines. The identified websites were scored using the DISCERN scoring system [27] and the Journal of American Medical Association (JAMA) benchmark criteria [28]. Compared with other common hand pathologies examined in that study, the quality of the Internet information on Dupuytren disease measured by DISCERN and JAMA criteria was better, but nevertheless, the study revealed a small number of websites that could be recommended to patients to support their decision making in the therapeutic process. The used evaluation tool—the DISCERN instrument—was developed by an expert panel and comprises 16 criteria for judging the quality of written consumer health information on treatment choices. Although the instrument requires some subjectivity for rating certain criteria, its developers claim it to be reliable and valid [27], and this could be verified by other authors. [29-31] In contrast to the EQIP instrument, the DISCERN evaluates information on treatment choices but does not evaluate readability or design aspects of the written materials. In our opinion, the EQIP is a more comprehensive and practical tool to evaluate the large, constantly growing volume of patient information produced within the health service. It helps also to make decisions about the urgency of any revisions that are needed to be made to written information in order to prioritize limited resources and minimize costs [11].

This study shows that private institutions did not provide less quality of information in comparison with academic nonprofit oriented website developers. Since the market of hand surgery, especially in the private setting, is consumer-oriented and strongly relies on marketing and advertising tools in an increasing crowded field of providers, physicians tend to advertise their services with complete patient information. This tempts the physician to take marketing action of selling his “products” and to influence the patient’s interest. However, economic issues should never yield to medical responsibilities and ethics.

This study has some limitations. First, due to the assumption that English is spoken as the first or second language in most developed countries, only websites developed in English were included; therefore, the quality of websites published in other languages remains unknown. The same can refer to the selection of search engines. Second, this work was done according to the statistical popularity of the search engines [21]; nevertheless, the use of other search engines could have revealed other interesting websites. Third, the Internet is a highly dynamic and constantly growing medium, and an evaluation of 118 websites at one point of time can represent only a snapshot of the information provided on the Web. Finally, there were limitations in the assessment instrument itself. The modified EQIP tool and its scoring system was not designed to assess websites referring specifically to Dupuytren disease but rather to assess patient information regarding any kind of medical treatment, which could have led to interpretation bias.

The evaluation of the present Web-based patient information on Dupuytren disease using a validated tool revealed substantial shortcomings and lacked standardization of its quality. The health care providers are the first to blame for this condition because in their obligation to provide a patient with an accurate and complete information, they did not stay up to date and recognize the potentials and hazards of this continuously growing medium—the Internet.