Please cite this paper as:

Eysenbach G, Diepgen TL: Responses to unsolicited patient e-mail requests for medical advice on the World Wide Web. JAMA. 1998;280:1333-1335.

Originally published in the Journal of the American Medical Association (JAMA).

Responses to unsolicited patient e-mail requests for medical advice on the World Wide Web.

Gunther Eysenbach, Thomas L. Diepgen

Abstract

Context – The Internet is increasingly used by consumers to seek health and medical information, but online medical advice has not been explored systematically.

Objective - To explore the attitude of physicians and other medical Internet information providers toward unsolicited e-mails from patients and their reaction to a simulated acute medical problem described in such an e-mail.

Design E-mails in December 1997 and January 1998 to Web sites from a fictitious patient describing an acute dermatological problem. Follow-up questionnaire survey to the same sites.

Setting – World Wide Web.

Subjects - 58 physicians and "webmasters"

Main Outcome Measures - Response rate and types of responses

Results - 29 (50%) responded to the pseudo-patient request; 9 of the responders (31%) refused to give advice without having seen the lesion, 27 (93%) recommended the patient see a physician and 17 (59%) explicitly mentioned the "correct" diagnosis in their reply. In the questionnaire survey, 8 (28%) of the 29 responders stated that they tended not to answer any patient emails, 7 (24%) stated that they usually reply with a standard email and 7 (24%) stated that they answer each request individually.

Conclusions –Consumers and patients seek health information via the Internet, but there is no accepted policy within the medical profession on how to handle unsolicited e-mails regarding medical problems. Even physicians who might be capable of providing such information may be unwilling to do so. Thus, patients may find this approach of little practical use and might receive information that could delay appropriate treatment.

Keywords: patient-physician relationship, Internet, communication, email, e-mail in medicine, quality control

Introduction

There has been explosive growth in recent years of the Internet and World Wide Web as tools for seeking and communicating health and medical information [1], with increasing numbers of physicians and health care institutions maintaining Web sites. Physicians or medical information providers who run such sites and post e-mail addresses likely will receive unsolicited e-mails from unknown patients asking for medical information and/or advice. On our bilingual Web site in Germany, patients frequently cross geographic borders to seek such information (about 75% of unsolicited patient emails we receive come from abroad, mostly the US [2]). As our experience likely is not unique, this situation raises two key issues: how do such sites react to these unsolicited e-mails, and how should they, if at all?

The American Medical Informatics Association recently published guidelines for physician-patient e-mails in the context of an established relationship. But apart from some anecdotal reports about "clinical encounters in cyberspace" [2-6], to our knowledge there are no studies in the literature investigating the question of unsolicited e-mails. Our objective was to determine how other medical information providers react to unsolicited e-mails sent by patients, to assess the quality of responses given by physicians, to explore how unsolicited emails sent by patients are perceived by doctors and medical "webmasters" (individuals who maintain Web pages), and to determine whether there is any consensus on how best to deal with this near-anonymous form of medical communication.

Methods

We wished to mimic the process by which a patient with a dermatological concern might seek medical advice on the Web and so simulated the search strategy a patient might use to find information on his condition. We searched the popular Alta Vista (http://altavista.digital.com) Web index using the query "+blisters +skin +pain +dermatology". The first 100 Web pages were viewed and all information providers were selected who publish some patient information and/or have dermatologists who could answer the e-mail. We found 12 distinct sites: 4 dermatology societies or organizations, 2 commercial sites, 2 universities, 1 dermatologist in private practice, 1 clinical psychologist self-described as specializing in skin diseases, 1 publisher, and 1 hospital. 3 of the emails extracted from these sites clearly belonged to a medical doctor.

We then visited all academic dermatology Web sites published in a list compiled by Thomas Ray of the University of Iowa Hospitals and Clinics (http://tray.dermatology.uiowa.edu/AcadDept.html), except those in Japanese. As a result, we found 45 Web sites with suitable e-mail addresses.

This process yielded a total of 57 sites that would likely be approached by patients with dermatological problems. In both approaches, we visited the websites and looked for an e-mail address belonging to the department, to a physician, or to the webmaster.

Usually, only one e-mail from each site was chosen. I If we found several addresses on the site, we selected the e-mail address in the following order of preference: 1) General e-mail address of the department 2) The physician responsible for the site; 3) E-mail address of the head of the department; 4) E-mail address of any other physician in the department; 5) E-mail address of the webmaster (if he or she did not have a medical degree).

The clinical problem

Having made our selections, we sent an e-mail to all addresses between December 1997 and January 1998 with the following text, without disclosing the fact that the "patient" was fictitious and using a neutral company email sender address. No Institutional Review Board approval of this study was required by our institution because no actual patients were involved. This approach was considered to be ethically appropriate as an assessment of quality in lieu of other means of testing the quality of information on the Internet. After the study was completed, we e-mailed the sites that responded to inform them that the "patient" query they had received was in fact fictitious and sent out as part of a research project that would report their responses without identifying them.

Hello,

I am a 55 year old male and have a sudden skin problem.

During the last 4 days I had a little fever and headache. There was also a burning pain and tingling and extreme sensitivity in one area of the skin of the chest. Since yesterday suddenly multiple fluid filled painful red blisters appeared on the same skin area on the chest, on a broad streak of reddened skin. The skin eruptions are very painful and my wife says the groups of blisters look a lot like chickenpox, though they are only on that girdle-like skin area.

I am on Sandimmune since I had a kidney-transplant some time ago.

What is this? Is it dangerous? What is the therapy? Do I have to see a doctor?

 

Please reply!

Thank you very much,

 

Peter

 

With this description, we tried to suggest a Herpes zoster infection, caused by Varicella-Zoster-Virus reactivation due to immunosuppressive treatment (Cyclosporin). In this case, early treatment with Acyclovir is essential to prevent severe and possibly deadly complications [7-10].

Questionnaire survey

About two weeks after we sent the "patient" e-mail, we distributed a 9-item questionnaire via e-mail to the same physicians and webmasters, without disclosing the fact that the previous e-mail has been a part of our study, and using a different e-mail sender address. The questionnaire inquired as to how many unsolicited e-mails the site received per week, the policy how these e-mails were handled, and questions regarding personal views on the topic. Individuals were asked to return the questionnaire by e-mail. If we received no reply, we resent the questionnaire after 5 and again after 9 weeks.

Results
Demographics and backgrounds of our responders

The simulated patient email was sent to all 58 addresses; 2 bounced back due to technical reasons. The subsequent questionnaire was sent to 53 of the 58 information providers (we excluded the 2 non-working e-mail addresses and 3 information providers from the USA, Canada and Germany that actively solicit patient e-mails, since we wanted to study unsolicited e-mails).

Our sample consisted of 22 sites from the USA (10 replied to the pseudo-patient request and 7 returned the questionnaire), in Japan (3 and 6, respectively), 7 in Germany (3 and 4), 2 in Canada (2 and 1), 2 in the UK (1 and 0), 2 in Spain (2 and 2), and 1 each in Austria (0 and 0), Brazil (0 and 0), Chile (1 and 1), Egypt (1 and 1), Finland (1 and 0), Hungary (1 and 1), Israel (1 and 1), Italy 0 and 1), South Korea (1 and 1), New Zealand (1 and 1), Sweden (0 and 1), Switzerland (0 and 0), Singapore (0 and 0), Taiwan (0 and 0) and Turkey (1 and 1). As a result of our search strategy, 41 individuals contacted were physicians (26 and 25), 37 of whom were dermatologists in university departments (22 and 23), including at least 11 department heads (6 and 7). Only 6 (1 and 4) were clearly non-medically trained webmasters. In 11 cases (department addresses and/or no reply to the patient email and/or to the questionnaire) the function and qualification of the person who would read the email remained unclear (2 and 0).

Responses to the fictitious patient e-mail

Twenty-nine (50%) of the individuals who were sent the simulated patient e-mail replied, 26 of whom reported themselves to be physicians. All but 2 of the respondents urged the "patient" to see a physician; the 2 who did not refused to give any advice at all. Seven of those who advised the patient to see a doctor refused to give additional advice. Of the remaining 20 respondents, 18 mentioned a diagnosis and 17 specifically mentioned Herpes zoster (the other mentioned Stevens-Johnson syndrome or toxic epidermal necrolysis as possible diagnoses). Thirteen, including the provider who did not mention H zoster, expressed the diagnosis with caution. Five respondents explained possible causes and 9 pointed out possible complications. Finally, 5 (all of whom had mentioned the diagnosis of H zoster) provided specific advice regarding therapy, which consisted of acyclovir, valaciclovir, and famcyclovir). The usual response time was 1-2 days.

Questionnaire results

Twenty-nine individuals (55%) replied to the follow-up questionnaire, including 17 who also previously replied to the fictitious patient e-mail. The results of the questionnaire survey are shown in Table 1 and Table 2. We further asked how many unsolicited patient e-mails these sites receive per week. Responses ranged between 0 and 50, with a mean value of 4.4 (STD 9.47) and a median of 1e-mail per week.

Except for two cases, we noted no obvious discordance between actual behavior in the fictitious patient experiment and the policy stated in the questionnaire. The questionnaire results corresponded well with the results of the experiment, indicating that about two-thirds of doctors answer unsolicited e-mail requests.

Discussion

The study found a striking lack of consensus among medical information providers on the theoretical and practical handling of unsolicited patient emails and their judgment of this topic. About one-third of those who replied explicitly refused to answer patient requests individually, arguing that it would be impossible to make a diagnosis via email without an examination and/or arguing with lack of resources and/or mandate to reply to these kind of inquiries. The remaining twothirds attempted to help individually, with differing depth of information provided, including 5 in which detailed treatment advice was.

Our small sample size precludes stratifying answers by country or other characteristics. Further, the results not be generalizable to other specialties or sites. Also, the response-rate to the "patient" e-mail and the information provided likely depends on the clinical contents of the e-mail (eg, the urgency and clarity of the description). Another clinical problem might produce different results.

Opportunities and concerns related to e-mail medical advice

There likely are a number of reasons for why patients may turn to the Internet for medical information and advice, rather than (or in addition to) asking their own doctors [2]. However, our study suggests that patients approaching unknown physicians to request e-mail advice will not guarantee that such information will be accurate, timely or appropriate.

First, only half of the doctors/webmasters responded to our fictitious patient e-mail even though the problem being described clearly was a medical emergency; in chronic medical problems, which constitute the majority of unsolicited emails [2], the response rate may be even lower.

Second, even among those who replied, the response time often was long (up to 10 days). For a real immunosuppressed patient experiencing herpes zoster, waiting 10 days for advice could have been a deadly delay.

Doctors who do answer unsolicited patient e-mail requests seem to have several reasons for being cautious. Aside from the problem that the sheer amount of e-mail received (in our survey, up to 50 per week) can hardly be handled if patients overuse easy and anonymous access to medically qualified personnel, the danger of misdiagnosing is clear without access to a complete patient history and physical examination

The legal consequences of providing incorrect, incomplete, or inappropriate advice under these circumstances are unclear. Although our study suggests that making a diagnosis via e-mail is possible in principle and no clear misinformation was given, one responding doctor considered Stevens-Johnson syndrome and toxic epidermal necrolysis, which could have created a considerable degree of patient anxiety. It is interesting to note that the reason for taking these diagnoses into account apparently was that this physician’s site focused primarily on dealing with these severe skin reactions. Thus, e-mail answers may be influenced by the personal interests and emphasis of those publishing information on the Web.

Confidentiality and security issues also are crucial. Unencrypted e-mails can be read by third parties as easily as postcards, so a patient who sends out e-mail to individuals unknown to him or her can never be sure that a physician really is behind the e-mail address published on the Web. Physicians also could also be misled, for example by commercial entities that might use fictitious patient e-mails as a subtle means of praising their own product or discrediting those of a competitor.

In addition, both parties can never be sure whether e-mails actually reached their receivers, as most Internet-based e-mail systems do not provide confirmation that a message was delivered. Even if the software returns notification of receipt, it cannot assure that the message was actually read and understood [11]. Again, patients waiting for answers from physicians who don’t read their e-mails may waste precious time.

Because the Web is a global, unregulated medium, additional cross-border issues arise. Language problems on both sides may create misunderstandings and patients may presume that the standards of medical training in other countries are comparable to their own region, which may not always be the case. Doctors who give e-mail advice have to take into account different ethical, cultural or economic backgrounds of patients and variations in health care settings and delivery systems, which might preclude the availability of certain therapeutic or diagnostic procedures [12]. The international nature of the Web also poses licensing issues, raising the question whether a doctor is allowed to counsel "out-of-area" patients at all.

Although guidelines recently have been proposed for clinical use of e-mail with patients, these suggested protocols apply within an established patient-physician relationship [11]. We therefore propose that additional guidelines be developed to advise physicians on how to handle unsolicited e-mails from patients. The expected growth of the Internet and its potential applications in health communication make the need for such guidelines especially pressing. In the absence of outcome data for patients using online medical advice, such guidelines should address clinical and ethical issues as well as legal ramifications and concerns regarding confidentiality.

Until such guidelines are forthcoming, we suggest that medical institutions that maintain Web sites develop standardized policies for handling the unsolicited patient e-mail they can expect to receive. We recommend such policies provide, at a minimum, for the posting of a disclaimer clearly indicating that unsolicited patient e-mails may not be answered and are not a substitute for obtaining medical advice in person from a health professional.

Acknowledgment

This work is supported in part by a grant of the German Research Net Association (DFN-Verein), Berlin, and the German Research Ministry (BMBF), Bonn, grant number TK 598 - VA/I3.

References

1. Deering CM, Richardson J, Wares CG: Medical information: from Surgeon General to superhighway. Ann Trop Med Parasitol 1995 Dec;89(6):579-91

2. Eysenbach G, Diepgen TL: Patients looking for information on the Internet and seeking tele-advice: Their motivation, expectations and misconceptions as expressed in electronic mails sent to doctors. Arch Dermatol. In press 1998

3. Neill R A, Mainous AG 3rd, Clark J R, Hagen MD. The utility of electronic mail as a medium for patient-physician communication. Arch Fam Med 1994; 3(3): 268-71

4. Fridsma DB, Ford P, Altman R. A survey of patient access to electronic mail: attitudes, barriers and opportunities. Proc Annu Symp Comput Appl Med Care 1994:15-9

5. Sellu D: Clinical encounters in cyberspace. BMJ 1996; 312 (7022): 49

6. Widman L E, Tong D A. Requests for Medical Advice From Patients and Families to Health Care Providers Who Publish on the World Wide Web. Arch Intern Med 1997; 157: 209-212

7. Balfour HH Jr, Bean B, Laskin OL, et al. Acyclovir halts progression of herpes zoster in immunocompromised patients. N Engl J Med 1983 Jun 16;308(24):1448-1453

8. Balfour HH Jr. Current management of varicella zoster virus infections. J Med Virol 1993; Suppl 1:74-81

9. Shepp DH, Dandliker PS, Meyers JD. Treatment of varicella-zoster virus infection in severely immunocompromised patients. A randomized comparison of acyclovir and vidarabine. N Engl J Med 1986 Jan 23;314(4):208-212

10. Meyers JD, Wade JC, Shepp DH, Newton B. Acyclovir treatment of varicella-zoster virus infection in the compromised host. Transplantation 1984 Jun;37(6):571-574

11. Kane B, Sands DZ for the AMIA Task Force to Develop Guidelines for Clinical Use of Electronic Mail and the World Wide Web: Guidelines for clinical use of electronic mail with patients (White Paper). JAMIA 1998;5:104-111

12. Coiera E. The Internet's challenge to health care provision - A free market in information will conflict with a controlled market in health care [editorial] BMJ 1996; 312: 3-4

Table 1. Medical information providers on the WWW - How they handle (and state to handle) unsolicited patient emails

Table 2. Responses on different issues of unsolicited patient emails according to questionnaire survey among medical information providers (n=29)

Corresponding Author:

Gunther Eysenbach, MD

Unit for Medical Informatics, Epidemiology & Public Health

University Hospital Erlangen, Dept. of Dermatology (Chairman: Prof. Dr. med. Schuler)

Hartmannstr. 14

91052 Erlangen

Germany

Email: ey@yi.com

Ph. ++49-9131-85 27 27

Fax ++49-9131-85 38 50