The Internet is becoming a primary source for obtaining medical information. This growing trend may have serious implications for the medical field. As patients increasingly regard the Internet as an essential tool for obtaining health-related information, questions have been raised regarding the quality of medical information available on the Internet.1 Studies have shown that health-related sites often present inaccurate, inconsistent, and outdated information that may have a negative impact on health care decisions made by patients.2
According to the US Census Bureau, 71.7% of American households report having access to the Internet.3 Of those who have access to Internet, approximately 72% have sought health information online over the last year.4 Among people older than age 65 years living in the United States, there has been a growing trend toward using the Internet, from 14% in 2000 to almost 60% in 2013, according to the Pew Research Internet Project.5 Most medical websites are viewed for information on diseases and treatment options.6 Since most patients want to be informed about treatment options, as well as risks and benefits for each treatment, access to credible information is essential for proper decision-making.7
To assess the quality of information on the Internet, we used DISCERN, a standardized questionnaire to aid consumers in judging Internet content.8 The DISCERN instrument, available at www.discern.org.uk, was designed by an expert group in the United Kingdom. First, an expert panel developed and tested the instrument, and then health care providers and self-help group members tested it further.8,9 The questionnaire had been found to have good interrater reliability, regardless of use by health professionals or consumers.8-10
More than 53,000 shoulder arthroplasties are performed in the United States annually, and the number is growing, with the main goal of pain relief from glenohumeral degenerative joint disease.11,12 The Internet has become a quasi–second opinion for patients trying to participate in their care. Given the prevalence of shoulder-related surgeries, it is critical to analyze and become familiar with the quality of information that patients read online in order to direct them to nonbiased, all-inclusive websites. In this study, we provide a summary assessment and comparison of the quality of online information pertaining to shoulder replacement, using medical (total shoulder replacement) and nontechnical (shoulder replacement) search terms.
Methods
Websites were identified using 3 search engines (Google, Yahoo, and Bing) and 2 search terms, shoulder replacement (SR) and total shoulder arthroplasty (TSA), on January 17, 2014. These 3 search engines were used because 77% of health care–related information online searches begin through a search engine (Google, Bing, Yahoo); only 13% begin at a health care–specialized website.4 These search terms were used after consulting with orthopedic residents and attending physicians in a focus group regarding the terminology used with patients. The first 30 websites in each search engine were identified consecutively and evaluated for category and quality of information using the DISCERN instrument.
A total of 180 websites (90 per search term) were reviewed. Each website was evaluated independently by 3 medical students. In the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) flow diagram, we recorded how websites were identified, screened, and included (Figure 1).13 Websites that were duplicated within each search term and those that were inaccessible were used to determine the total number of noncommercial versus commercial websites, but were excluded from the final analysis. The first part of the analysis involved determining the type of website (eg, commercial vs noncommercial) based upon the html endings. All .com endings were classified as commercial websites; noncommercial included .gov, .org, .edu, and .net endings. Next, each website was categorized based on the target audience. Websites were grouped into health professional–oriented information, patient-oriented, advertisement, or “other.” These classifications were based on those described in previous works.14,15 The “other” category included images, YouTube videos, another search engine, and open forums, which were also excluded from the final analysis because they were not easily evaluable with the DISCERN instrument. Websites were considered health professional–oriented if they included journal articles, scholarly articles, and/or rehabilitation protocols. Patient-directed websites clearly stated the information was directed to patients or provided a general overview. Advertisement included sites that displayed ads or products for sale. Websites were evaluated for quality using the DISCERN instrument (Figure 2).
DISCERN has 3 subdivision scores: the reliable score (composed of the first 8 questions), the treatment options (the next 7 questions), and 1 final question that addresses the overall quality of the website and is rated independently of the first 15 questions. DISCERN uses 2 scales, a binary scale anchored on both extremes with the number 1 equaling complete absence of the criteria being measured, and the number 5 at the upper extreme, representing completeness of the quality being assessed. In between 1 and 5 is a partial ordinal scale measuring from 2 to 4, which indicates the information is present to some extent but not complete. The ordinal scale allows ranking of the criteria being assessed. Summarizing values from each of the 2 scales poses some concern: the scale is not a true binary scale because of the ordinal scale of the middle numbers (2-4), and as such, is not amenable to being an interval scale to calculate arithmetic means. To summarize the values from the 2 scales, we calculated the harmonic mean, the arithmetic mean, the geometric mean, and the median. The means were empirically compared with the median, and we used the harmonic mean to summarize scale values because it was the best approximation of the medians.