Seventy percent of U. Twenty-four percent relied on those who had the same health condition. Many have now added the internet to their personal health toolbox, helping themselves and their loved ones better understand what might be ailing them.
This study was not designed to determine whether the internet has had a good or bad influence on health care. It measures the scope, but not the outcome, of this activity. Physician Ted Eytan, a Washington-based director of the Permanente Federation which is part of Kaiser Permanente, told USA Today that doctors need to listen to patients concerns, even if they got them from the Internet. Information about weight loss supplements suggested green coffee supplements might help.
On the other hand, while the site dubiously claims it has "10 easy, painless ways to lose weight," the page actually included some reasonable, if obvious, tips: walk more, hydrate, share restaurant meals.
I also found problems with how the site conveys the effectiveness and possible side effects of some prescription drugs. When I visited the page on weight loss pills , an advertisement on meal replacement shakes popped up, as did an ad for the drug Qsymia — which is among the six drugs featured in the article:. While the site's content is produced by a team of doctors and medical writers, the article failed to mention any basic information about the drug's effectiveness or how many people the drug was likely to help the number needed to treat, in medical parlance.
And some of the information was worryingly incomplete. For example, WebMD didn't note the serious side effects associated with the drug Contrave — it can cause severe, potentially fatal skin reactions and liver failure. But those were just my observations after spending a few hours on the site. In the absence of better evidence, I decided to get the views of independent doctors. The subscription-based website, used mainly by doctors to access summaries of the latest medical information, accepts no advertising money as part of its editorial policy and pursuit of independence.
Overall, the doctors I spoke to said they didn't find anything exceptionally egregious about WebMD. But they noted the lack of context around some of the site's medical advice, as well as a smattering of misinformation.
Ryan Connolly found "a few less-than-evidence-based medications listed Risperdal, Zyprexa. Vagus nerve stimulation , a medical treatment that involves delivering electrical impulses to the vagus nerve, was also listed — even though it's no longer considered evidence-based and is almost never done, he said. Meanwhile, one recently approved drug for depression, brexpiprazole, was left out.
Connolly's conclusion: WebMD's depression treatment information is not totally unreliable but is sloppy and incomplete. And it could easily give patients a skewed view of their treatment options. University of Michigan's Sandeep Vijan thought WebMD's cholesterol treatments page was "oversimplified" and "often phrased in an overly frightening way.
Again, Vijan noted a range in the quality of the site's information. Some of it "may be fine for an initial introduction for patients," he said. Within the group of doctors I surveyed, some spoke highly of the site. In particular, non-alphanumeric characters and extra spaces were removed, and location strings with a frequency less than 14 were removed.
Each location string was mapped to a location city, state, country using the Google Geocoding API [ 27 ]. We focus on US users, and hence we remove users from other countries. In Figure 2 , we show the distribution of users for each type of Web-based health-related social outlet, normalized by state population. To better understand these results, we created Table 6 [ 31 , 42 - 45 ], which shows the correlation across all states between the normalized by population number of users in various health-related social outlets and other societal measures see Multimedia Appendix 1 for more details.
Our fourth key finding is that users in areas with higher income and more access to health care are more likely to participate in Web-based health-related outlets, and particularly in Web forums and drug review sites, which are the primary social sites for health-related information sharing [ 10 ].
A reason could be that The writing level, as previously mentioned, is measured using a standard reading level formula that assigns a school grade to the given text. For example, when a person writes text at a 5 th grade reading level, it implies that their writing should be understood by people that have passed the 5 th grade.
Table 7 reports our results for writing level of health-related social outlet users. Next, we try to put these findings in perspective. Unfortunately, related work reports only on reading levels and not writing levels of the US population participating in social outlets. Thus, we compare our results in Table 7 to Figure 3 , which reports the reading level of the general US population [ 15 ].
Our fifth key finding is that the writing level in health-related social outlets Table 7 is generally lower than the reading level of the population Figure 3. The benefit of social interaction with respect to health empowerment has been demonstrated before [ 48 ]. Our results can help health care providers customize educational campaigns for different groups.
For example, white women should be informed to a larger extent on the possible misinformation spreading in health Web forums, since they participate much more.
However, this is not true for black ethnicity, who are not overrepresented in any health-related social outlet. Further, drug review websites and health Web forums are better to target females when advertising for their products than other health-related social outlets. In the age results section, we found that younger groups years old participate in large numbers in health forums, which may sound counterintuitive.
By analyzing posts for this age bracket, we found the most popular keywords are related to pregnancy such as birth control, ovulation, and miscarriage. On the other hand, their participation is lower for drug review websites. A possible explanation may be that often patients who talk about pregnancy are not taking any drugs, compared to other conditions like diabetes, where drugs are more common.
We also attempt to explain the disparities in the participation in health-related social outlets based on socioeconomic factors through the state-level participation distributions. Our results in Table 6 show that less access to physicians does not lead to higher participation in health-related social outlets as one would expect. In contrast, it seems that the participation in such outlets is correlated with the access to health care and the average income. The weak but positive correlation between income and participation to health Web forums and drug review sites may be partially attributed to the higher Internet usage of the more affluent groups, as shown in Table 6.
Another possible explanation is that lower income or uninsured persons are more likely to be part of a community with health care disparities [ 50 ]. On the other hand, we found that the content in health-related social outlets is easy to understand for almost all users, given the low writing level.
That is, the well-known health literacy issue, which is more severe in low-income and lower education populations [ 5 ], does not seem to apply to Web-based health-related social outlets. Of course, the low writing level does not address the issue of language, as many low income and low education users in the United States do not speak English at home [ 53 ].
Our ethnicity and gender classifiers are not perfect, as shown in Multimedia Appendix 1 , and thus introduce an error into our analyses. Another limitation is the informal writing style of social media posts, as our writing level method uses the average sentence length, which expects that posts are properly punctuated. We addressed this limitation to some degree by only considering sentences of a reasonable length less than 30 words. Estimating writing level could have been improved by considering other features like typos or spelling mistakes.
Further, it would be useful to measure the quality of the posted information, in addition to just the writing level. This is a challenging issue, which we leave as future work. Since all the attributes are reported by users, there is inevitably self-selection bias. In particular, gender, age, and location are not mandatory in any site. For instance, older people may choose not to report their age. Moreover, choosing to report the real names or posting profile pictures could also create self-selection bias in our gender and ethnicity classifiers.
There may also be various types or degrees of bias across different outlets. For instance, WebMD users may use their real name less frequently than Twitter users. This in turn may bias the study results, especially for ethnicity where we depend completely on the classifier results. We studied user demographics in Web-based health-related social outlets, which we split into three different types: social networks, drug review websites, and health Web forums.
The distributions of the demographic attributes—gender, age, ethnicity, location, and writing level—have been analyzed for each source type and compared with relevant baseline user distributions like Internet and general social outlets participation. The results reveal interesting and often unexpected disparities with respect to all demographic attributes. Any opinions, findings and conclusions or recommendations expressed in this material are those of the authors and do not necessarily reflect the views of the National Science Foundation NSF.
Online social outlets summary, health keywords, classifiers evaluation, and data coverage. Authors' Contributions: All authors contributed substantially to this work. They designed and performed the analysis and approved the final version of this manuscript. Conflicts of Interest: Conflicts of Interest: None declared. National Center for Biotechnology Information , U.
J Med Internet Res. Published online Aug 6. Reviewed by Gang Luo and Monica Tremblay. Author information Article notes Copyright and License information Disclaimer. Corresponding author. Corresponding Author: Shouq A Sadah ude. This article has been cited by other articles in PMC. Abstract Background The rapid spread of Web-based social media in recent years has impacted how patients share health-related information. Objective Our aim was to study the demographics of users who participate in health-related Web-based social outlets to identify possible links to health care disparities.
Conclusions We identified interesting and actionable disparities in the participation of various demographic groups to various types of health-related social outlets. Is he a full-time guy? How much time is being spent on the review? Smith counters that each story is reviewed by a board-certified physician. When patients perform their own research on complex illnesses, they're often misinformed.
Perskin said one out of every 20 of his patients suffering from extreme fatigue, visit sites and conclude that they have lupus. Perskin says nearly all of them are incorrect and only one in patients with those symptoms suffer from lupus. Nonetheless, he says in many cases doing the research educates consumers, makes them more informed and ask better questions. Another consumer risk is thinking that the Internet will immediately solve their problems and instead "missing an opportunity to get a diagnosis when it matters in time," Perskin says.
A patient with severe chest pains should be calling , not doing an extensive Internet search that can contribute to a delay in getting assistance and the possibility of death. Arnold Wald, a professor at the Wisconsin School of Medicine and specialist in gastroenterology, says patients can easily fall into traps if they rely solely on the web. He says many patients use websites in an "unfiltered" way, act on its advice, without consulting a physician.
Web MD recommends that "You should always seek the advice of your physician or other qualified healthcare provider. Wald says patients need to view these medical sites skeptically and not take everything posted on the site as fact or gospel. For example, he says the headline of an August 30, Web MD health article, "Linzess Relieves Constipation, Pain of Irritable Bowel Syndrome" is misleading and contradicts the info in the article Wald is quoted in it.
The article clearly states that only one in five patients that use Linzess returns to having normal bowel movements. Even though the Web MD article was reviewed by physician Louise Chang, Wald says it doesn't indicate her specialty she's an internist.
0コメント