Analysis of Cases of Harm Associated With Use of Health Information on the Internet

Context There is concern about the potential harm associated with the use of poor quality health information on the Internet. To date, there have been no systematic attempts to examine reported cases of such harm.

Methods We conducted a systematic review of the peer-reviewed literature, to evaluate the number and characteristics of reported cases of harm associated with the use of health information obtained on the Internet. Using a refined strategy, we searched MEDLINE (from 1966 to February 2001), CINAHL (from 1982 to March 2001), HealthStar (from 1975 to December 2000), PsycINFO (from 1967 to March 2001), and EMBASE (from 1980 to March 2001). This was complemented with searches of reference lists. Two authors separately reviewed the abstracts to identify articles that describe at least 1 case of harm associated with the use of health information found on the Internet. Articles of any format and in any language deemed possibly relevant by either researcher were obtained and reviewed by both researchers.

Results The search yielded 1512 abstracts. Of these 186 papers were reviewed in full text. Of these, 3 articles satisfied the selection criteria. One article described 2 cases in which improper Internet searches led to emotional harm. The second article described dogs being poisoned because of misinformation obtained on the Internet. The third article described hepatorenal failure in an oncology patient who obtained misinformation about the use of medication on the Internet.

Conclusions Despite the popularity of publications warning of the potential harm associated with using health information from the Internet, our search found few reported cases of harm. This may be due to an actual low risk for harm associated with the use of information available on the Internet, to underreporting of cases, or to bias.

The Internet has spawned a plethora of publications warning of potential harm, particularly regarding the inappropriate use of the Internet by patients and families. Potential risks can emanate from the use of irrelevant or inaccurate information^1,2 or from misunderstanding relevant and valid information.³ In these cases, harm can be physical (eg, due to inappropriate treatments, adverse effects, or untreated disease), emotional (eg, from false hope or anxiety regarding unfounded diagnostic, prognostic, or therapeutic information), or financial (eg, expenses associated with unnecessary second opinions and purchase of inappropriate services or products).

Our objective was to conduct a systematic review to examine the number and characteristics of cases of harm associated with the use of health information available on the Internet.

A refined strategy that combined 2 sets of terms was used. One set related to the Internet and the other related to harm (Online Table available in PDF format ). The databases searched with this strategy included MEDLINE (from 1966 to February 2001), CINAHL (from 1982 to March 2001), HealthStar (from 1975 to December 2000), PsycINFO (from 1967 to March 2001), and EMBASE (from 1980 to March 2001). Searches were executed in April 2001 and were complemented with a review of the reference lists of relevant articles.

All citations were evaluated for eligibility independently by 2 of the authors (A.G.C., M.V.-K.). Disagreements were resolved by consensus or by the third author (A.R.J.). A citation was included if it described at least 1 case of harm associated with the use of health-related information found on the Internet. All publication types, in any language, were included. Harm was defined as adverse events or bad outcomes, either physical or emotional, that occurred after the use of materials, medications, and/or information obtained from a Web site. Publications were excluded if they described harm associated with use of e-mail or if they described cases of intentional harm.

The same authors who participated in study selection extracted data, from each article, on name of the first author, publication type, publication year and journal; number of individuals studied, age, health status; type of information obtained from the Internet; type, severity, duration, management, and outcome of harm; and suggested mechanism of harm. The data extracted were summarized in tables and described qualitatively. Meta-analysis of a small number of case reports was deemed inappropriate.

The search strategy yielded 1512 citations of indexed journal articles. Of these, 186 full articles were retrieved, of which only 3 articles met the inclusion criteria. Of the articles excluded, 11 described intentional self-inflicted harm (the list is available on request).

One of the included articles described 2 cases of emotional distress after inappropriate Internet search strategies led to irrelevant information.⁴ The first case described a pregnant woman and her partner, who after being informed that the fetus had borderline cerebral ventriculomegaly searched the Internet for information on hydrocephalus, a word they remembered from their medical visit. The information they found was related to aqueduct stenosis. Based on the irrelevant information they found on the Internet, the family made arrangements to have a postnatal shunt placed by a pediatric neurosurgeon. The family required several counseling sessions to deal with the anxiety caused by their failed attempts to find information. The second case described a patient who presented for termination of a pregnancy. She had been told that her fetus had a 7-mm nuchal fold, and she subsequently performed an Internet search using the wrong keywords. She found information on fetal cystic hygroma. Believing this was the fetal diagnosis, she sought termination of the pregnancy. The authors did not include outcome information for this case.

Two articles described physical harm. In the first report, the authors described phenol poisoning in 3 dogs. Their owner had used erroneous information about the use of phenol for heartworm. The 3 dogs required intensive care intervention, and all 3 recovered.⁵

The only reported case of unintentional physical harm in humans involved a 55-year-old man with squamous-cell carcinoma of the sinus who found information on the Internet that promoted the use of hydrazine sulphate for cancer treatment. After self-medicating for 4 months with hydrazine that he had obtained from an alternative medicine Web site, he presented with hepatorenal failure. He died a week later. The autopsy findings suggested an adverse reaction from use of hydrazine sulfate, which is an unregulated alternative remedy for cancer marketed on the Internet.⁶

The Internet's capacity for harm is likely to be equal to or exceeded by its capacity for providing good and useful health information to users in a relatively inexpensive and timely manner. Better data on both harms and benefits associated with the use of the Internet around health-related decisions are needed as is more valid information about the accessibility of such information by members of the public.⁷

The dearth of information on harm associated with the Internet is unfortunate but not unexpected. It has been known for a while that peer-reviewed literature provides a gross underestimate of harm associated with health-related interventions because authors tend to concentrate on efficacy and effectiveness rather than on harm. In this case, the lack of information about harm related to the Internet may be explained by several reasons. First, it is possible, albeit unlikely, that a low risk for harm associated with the use of health information from the Internet exists. Finding health-related information is different from using it. Patients may also be more discerning and Internet savvy than what experts may be willing to credit them for evaluating for themselves the quality of the information that they find before acting on this information.

Second, it is possible and plausible that many cases of harm associated with inappropriate use of health information on the Internet have occurred but remain unreported. People harmed after using information found on the Internet may feel a sense of guilt or intimidation at having ventured outside the confines of the relationship with their caregivers.

Third, it is possible that health care professionals do not think about the role that the Internet may play in relation to adverse events experienced by their patients. They may also feel intimidated by assertive patients and become reluctant to ask about their Internet use.

Fourth, health care professionals may be aware of harm resulting from their patients' inappropriate use of information from the Internet but may not be submitting these cases for publication. Fifth, these cases may be published but only as a subsidiary feature of other stories, hindering attempts to find them. Lastly, it is possible that more cases of harm have been submitted for publication but were not accepted by journal editors. These kinds of case reports may not be considered newsworthy, or alternatively, they fail to succeed when competing with other articles for the limited space available in most indexed journals. If the latter were correct; however, it would be difficult to explain the abundance of opinion pieces about potential, not actual, harm associated with the use of the Internet. All these issues, obviously, are likely to apply to any other source of health information.

However, gaining a valid assessment of the potential for harm associated with the Internet and its relative magnitude in relation to other media will not be an easy task. Strategies that may appear simple, such as the inclusion of questions about Internet use (or other sources of information, for that matter) during history taking, may prove difficult to implement and monitor. Making the public more comfortable disclosing their use of information from the Internet in cases of harm may be as difficult as getting health professionals to acknowledge their errors. Journal editors may not have enough interest or incentives to invest resources on efforts to increase reporting cases of harm associated with Internet use. Even if not interested, they may consider declaring a moratorium on more opinion pieces about such harm.

The Internet itself may prove to be the most effective source of information about harm associated with its use and the most efficient vehicle to disseminate it. We hope that a systematic review of information available on the Internet be conducted soon. Recently DAERI (Database of Adverse Events Related to the Internet), a European initiative, has been launched to promote the systematic collection of case studies on possible harm caused by the Internet.⁸ We encourage readers to support this project. It is efforts like this that may provide the only efficient and effective way to determine whether the issue of harm associated with the Internet is just much ado about nothing or a serious matter for concern.