This is an open-access article distributed under the terms of the Creative Commons Attribution License (https://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work, first published in the Journal of Medical Internet Research, is properly cited. The complete bibliographic information, a link to the original publication on http://www.jmir.org/, as well as this copyright and license information must be included.
Respecting patient privacy and confidentiality is critical for doctor-patient relationships and public trust in medical professionals. The frequency of potentially identifiable disclosures online during periods of active engagement is unknown.
The objective of this study was to quantify potentially identifiable content shared on social media by physicians and other health care providers using the hashtag #ShareAStoryInOneTweet.
We accessed and searched Twitter’s API using Symplur software for tweets that included the hashtag #ShareAStoryInOneTweet. We identified 1206 tweets by doctors, nurses, and other health professionals out of 43,374 tweets shared in May 2018. Tweet content was evaluated in January 2019 to determine the incidence of instances where names or potentially identifiable information about patients were shared; content analysis of tweets in which information about others had been disclosed was performed. The study also evaluated whether participants raised concerns about privacy breaches and estimated the frequency of deleted tweets. The study used dual, blinded coding for a 10% sample to estimate intercoder reliability using Cohen κ statistic for identifying the potential identifiability of tweet content.
Health care professionals (n=656) disclosing information about others included 486 doctors (74.1%) and 98 nurses (14.9%). Health care professionals sharing stories about patient care disclosed the time frame in 95 tweets (95/754, 12.6%) and included patient names in 15 tweets (15/754, 2.0%). It is estimated that friends or families could likely identify the clinical scenario described in 242 of the 754 tweets (32.1%). Among 348 tweets about potentially living patients, it was estimated that 162 (46.6%) were likely identifiable by patients. Intercoder reliability in rating the potential identifiability demonstrated 86.8% agreement, with a Cohen κ of 0.8 suggesting substantial agreement. We also identified 78 out of 754 tweets (6.5%) that had been deleted on the website but were still viewable in the analytics software data set.
During periods of active sharing online, nurses, physicians, and other health professionals may sometimes share more information than patients or families might expect. More study is needed to determine whether similar events arise frequently and to understand how to best ensure that patients’ rights are adequately respected.
Physicians, nurses, and other health professionals remain among the most trusted professionals in the United States because of their commitment to the well-being of others; they are a trusted source of health information and guidance [
Fulfilling physicians’ obligations to protect the well-being and privacy of their patients is complicated in the age of the internet. Internet culture is very different from that of the medical profession, creating potential ethical problems with boundaries and privacy that did not exist when physicians interacted exclusively offline. In order to maintain the trust of the public and that of individual patients, physicians increasingly need to understand the limits and risks of disclosure of certain types of information online. Although concerns about unprofessional medical student and resident behavior online have been articulated before [
Social media usage has become popular among medical professionals. A survey in 2014 by QuantiaMD [
An article in July 2018 [
I delivered a baby very underweight, weighing two pounds. They said he did not have a chance. I remained with him for a couple of days. Nine years later, he played his first football game last week.
Hashtags can make online content searchable and discoverable online, regardless of time since publication [
Early research on health professionals using Weblogs [
Despite the importance of previous work [
The results of this study are likely to be of interest to those compiling guidelines for the use of social media by medical professionals.
The overall research aim of this study was to develop a better understanding of the content shared with the hashtag #ShareAStoryInOneTweet.
The objectives of the study were to (1) identify unique tweets sent by doctors and other health care providers using the hashtag, (2) to develop an understanding of the characteristics of the doctors and health care providers using the hashtag, and (3) to categorize tweets into themes and identify the frequency of instances in which patients could be identified by themselves or by their family.
Because all information about the published content was publicly accessible, Lowell General Hospital approved this study as institutional review board exempt. To evaluate content in the #ShareAStoryInOneTweet phenomenon, Symplur Signals (Symplur LLC), a proprietary health care–focused database and analytics program collecting data on Twitter using its Enterprise application program interface, was utilized [
Unique tweets were reviewed by reading text provided within the data set and then evaluating the URL and each account’s public profile on Twitter’s website as of March 2019. Tweets from students misclassified as health care providers (1.3%) (eg, those listing “future doctor” in profile), from blocked accounts (0.3%), or with no relevant content posted with the hashtag (1.2%) were excluded, leaving a total of 1172 tweets shared by physicians, nurses, or other health professionals (
Characteristics of the health professionals sharing these tweets was examined, using information publicly available in their online profiles, including profession, gender, and country. Physicians were also categorized by specialty as described in their profiles or as unknown if not stated. More detailed content analysis focused upon the tweets in which the health care professional shared the illness or clinical experience of someone other than themselves.
The study also evaluated tweets commenting upon the hashtag-related phenomenon or recommending participation to others. The study analyzed tweets individually rather than as content threads.
Tweets by doctors and other health care providers during #ShareAStoryInOneTweet in May 2018. HCP: health care providers.
To assess the magnitude of hashtag use, the study evaluated total number of tweets. We also evaluated the number of total participants, focusing upon physician, nurse, and other health care professionals. We calculated tweets per hour, number of tweets, and use of images both in aggregate and by health care stakeholder categories. In order to capture hourly tweet activity rates, we restricted the time frame to the first two weeks starting May 4th to focus on the viral period. We evaluated the potential reach of the tweets using the software’s definition of
For each tweet, we coded several measures: the tweet author’s role in the other person’s clinical care; whether the patient died or was dying; whether the author helped save the other person’s life; inclusion of patient name; inclusion of a clinical image; and inclusion of a specific age. We categorized the time frame of the event described within a tweet as within the past year, 1-2 years ago, 2-5 years ago, >5 years, or unknown.
Whether either a patient or the patient's family or friends would be able to identify the clinical scenario described in each tweet was categorized broadly in response to codebook question “Could patient or family potentially identify the clinical situation?” as
One author (MK) assessed all tweets; a second author (WA) coded a 10% sample independently. Intercoder reliability and percentage agreement were assessed using ReCal [
Because the tweets could be discoverable in malpractice or tort suits, we also analyzed whether the author made comments with a negative opinion about the patient or family, or if the author acknowledged that a medical error occurred. We also assessed whether information was shared about vulnerable patients, as defined by the US Department of Health and Human Services [
We separately evaluated tweets commenting upon the hashtag to determine whether the authors had a favorable or unfavorable opinion of the viral sharing, or if they invited others to share stories or to participate. We also identified whether these tweets expressed any concern about privacy breaches.
Overall activity and frequencies for stakeholder participation using Symplur. Frequencies, median, and mean endpoints for content analysis were calculated using Excel (for Mac 2011 version 14.7.2, Microsoft Inc). Cohen κ was used to measure interrater reliability [
For May 2018, we identified 31,690 individuals who posted tweets with the hashtag, with a potential of 106.5 million views; 1725 (5.3%) individuals self-identifying as doctors and 861 (2.6%) as other health care providers shared tweets. At its peak, activity showed 1274 tweets per hour (
Tweets per hour including the hashtag #ShareAStoryInOneTweet in May 2018.
The characteristics of the health professionals sharing tweets with disclosures about others are presented in
The majority (659/754, 87.4%) involved the sharing of stories about direct patient care, rather than the author’s role as a family caregiver or in another role (
Nearly half of the tweets (337/754, 44.7%) described a clinical scenario involving death or dying. Comments disclosing medical errors (6/754, 0.8%) or expressing a negative opinion about the patient or family were rare (4/754, 0.5%). Agreement between coders was 86.8%, and intercoder reliability Cohen κ=0.8 suggested substantial agreement [
Characteristics of doctors and health care professionals sharing tweets disclosing information about others.
Characteristics | Value (N=656), n (%) | ||
|
|
||
|
Female | 384 (58.5) | |
|
Male | 266 (40.5) | |
|
Unknown | 6 (0.9) | |
|
|
||
|
|
|
|
|
|
Anesthesia | 22 (4.5) |
|
|
Cardiology | 25 (5.1) |
|
|
Critical care | 12 (2.5) |
|
|
Emergency medicine | 77 (15.8) |
|
|
Family medicine-general practitioner | 48 (9.9) |
|
|
Gastroenterology | 6 (1.2) |
|
|
Hematology-oncology | 39 (8.0) |
|
|
Hospitalist | 5 (1.0) |
|
|
Infectious disease | 5 (1.0) |
|
|
Internal medicine | 12 (2.5) |
|
|
Neurosurgery | 5 (1.0) |
|
|
Obstetrics and gynecology | 13 (2.7) |
|
|
Palliative care | 18 (3.7) |
|
|
Pathology | 9 (1.9) |
|
|
Pediatrics | 28 (5.8) |
|
|
Psychiatry | 7 (1.4) |
|
|
Pulmonary medicine | 7 (1.4) |
|
|
Radiation oncology/clinical oncology | 10 (2.1) |
|
|
Radiology | 6 (1.2) |
|
|
Surgery | 19 (3.9) |
|
|
Trauma surgery | 20 (4.1) |
|
|
Unknown | 36 (7.4) |
|
|
Other | 57 (11.7) |
|
|
|
|
|
|
Nurse, not otherwise specified | 66 (67.3) |
|
|
Critical care | 16 (16.3) |
|
|
Emergency medicine | 8 (8.2) |
|
|
Other | 8 (8.2) |
|
Nurse practitioner | 12 (1.8) | |
|
Paramedic | 18 (2.7) | |
|
Pharmacist | 7 (1.1) | |
|
Physical therapy | 10 (1.5) | |
|
Psychologist | 5 (0.8) | |
|
Social worker | 7 (1.1) | |
|
Speech therapy | 5 (0.8) | |
|
Other | 8 (1.2) | |
|
|
||
|
Australia | 11 (1.7) | |
|
Canada | 99 (15.1) | |
|
India | 7 (1.1) | |
|
Ireland | 16 (2.4) | |
|
Saudi Arabia | 7 (1.1) | |
|
South Africa | 4 (0.6) | |
|
Spain | 4 (0.6) | |
|
United Kingdom | 82 (12.5) | |
|
United States | 347 (52.9) | |
|
Unknown | 55 (8.3) | |
|
Other | 24 (3.7) |
Content characteristics tweets with disclosures about others.
Content characteristic | Value (N=754), n (%) | |||
|
|
|||
|
Health care professional | 669 (88.7) | ||
|
Patient | 0 (0.0) | ||
|
Caregiver | 42 (5.6) | ||
|
Other | 43 (5.7) | ||
|
|
|||
|
Within past year | 5 (0.7) | ||
|
1-2 years | 6 (0.8) | ||
|
2-5 years | 5 (0.7) | ||
|
> 5 years | 79 (10.5) | ||
|
Unknown/not described | 659 (87.4) | ||
|
|
|||
|
Author involved in patient care | 635 (84.2) | ||
|
Dying patient or patient death | 337 (44.7) | ||
|
Saving a patient’s life | 131 (17.4) | ||
|
Include a clinical image | 11 (1.5) | ||
|
Include patient name | 15 (2.0) | ||
|
Provide specific patient age | 163 (21.6) | ||
|
Express negative opinion of patient or family | 4 (0.5) | ||
|
Mention medical error | 6 (0.8) | ||
|
|
|||
|
Can family or friends identify situation described? | 242 (32.1) | ||
|
|
|
||
|
|
All tweets | 183 (24.3) | |
|
|
Potentially living patients (n=348) | 162 (46.6) | |
Vulnerable population | 152 (20.2) |
Of 187 tweets actively part of the conversations without disclosures, 173 made some commentary: 6 tweets (3.2%) raised concerns about privacy or identifiable information in the tweets with disclosures, 1 (0.6%) tweet involved another critical comment, and 167 (96.5%) tweets were neutral or favorable. Of 187 tweets, 42 tweets (22.5%) invited others to read the hashtag’s stream or contribute to it.
This retrospective study describes a physician-initiated event sharing health-related stories and information on Twitter by quantifying the global participation of health care professionals and the type of content shared. The tweeted stories became widely shared, attracting media attention and disseminating the information widely. Almost none (either explicitly or appear to) confirm consent to share information publicly on the popular social network. Nurses, physicians, and other health professionals commenting using the hashtag were more likely to express support for the event and encourage others to participate than they were to raise concerns about patient privacy breaches. However, recent research suggests that 12% of patients may have less trust in physicians describing patient stories on social media, even if shared respectfully [
The study showed a relatively high incidence of sharing stories including details that might make them potentially identifiable to patients themselves or to families and friends in a setting that involved a large number of health care professionals. This finding highlights a lack of awareness about the privacy issues intrinsically connected to interactions on social media. Early in the use of social media, most US state medical boards received at least one report of an episode of online professionalism violations for disciplinary action, including violations of patient confidentiality [
The findings of this study differ from those of prior studies [
Most research evaluating online disclosures focus on the privacy paradox, in which people value their privacy but still share their own information. Surveys indicate people may value short-term social rewards of self-disclosure online more than long-term privacy concerns [
Based upon the temporal pattern of sharing, this hashtag-related event may be less similar to narrative medicine and writing and more similar to a brief episode of social contagion, in which viral sharing of content or emotions online may occur and involve more than simple, conscious risk–reward tradeoffs [
Another contributing factor may be a knowledge gap for physicians and other health care professionals on how to behave online. While many recognize the importance of online professionalism, curricula for use in formal medical education are only beginning to emerge and remain uncommon [
This study had several limitations. First, this study examined a very specific event that may occur during very active periods of online engagement but likely overestimated the general incidence of online behaviors that could, in some cases, constitute violations of medical professionalism. Future research could analyze a broader collection of social media posts by medical professionals. Second, the study could not assess the number of people actually seeing these tweets; only the number of likes was measured, and the potential reach was estimated. Third, the study only analyzed tweets from accounts that the software identified as health care professionals. The evaluation of all tweets in the cohort confirms the software rarely misclassified nonprofessionals into this group, but the study did not evaluate any other participants in the event to determine if the study could identify more participating health care professionals not categorized as doctors or nurses by the software, which could decrease or increase the incidence of potential privacy breaches. Fourth, by analyzing only tweets with the hashtag, the study potentially underestimated the frequency of others expressing concern about patient privacy. Fifth, given the brevity inherent to the medium of Twitter, it is possible that some authors did indeed have formal documentation of patients’ consent to share their stories but that there was insufficient room to include due to character limits in each post. Finally, the assessment of identifiability in this study may differ from those in other studies, and we cannot exclude the possibility that some physicians and nurses tweeting what seemed to be identifiable stories consciously changed important details to deidentify. It was beyond the scope of this study to confirm whether any harm occurred.
Despite these limitations, the findings of this study clearly show that internet-based sharing raises potential pitfalls for medical professionalism. The internet provides nurses, physicians, and other professionals the opportunity to help or harm others on a global scale. Although internet culture may favor maximizing transparency, it can also pose the risk of directly contradicting health professionals’ fiduciary duty: first, do no harm, including harm that may be inflicted by what we say.
The study identified a high incidence of potential privacy breaches online. More research is essential to confirm the findings of this study and determine how to ensure physicians, nurses, and other professionals adapt their behavior to maintain medical professionalism in the digital age. Our results suggest that some who were using the hashtag may not have appreciated that the information being shared might breach patients’ privacy We recommend greater specification of professional ethical standards in this context along with evidence-based training in ethical digital communications skills for the undergraduate, graduate, and continuing medical education.
WA and TGG have nothing to disclose. MSK reports common stock ownership in Dr. Reddy’s Laboratories, Healthcare Services Group, Mazor Robotics, and US Physical Therapy. RJ has stock options as compensation for an advisory board role in Equity Quotient, a company that evaluates culture in health care companies. RJ has received personal fees from Amgen and Vizient and grants for unrelated work from the National Institutes of Health, the Doris Duke Foundation, the Greenwall Foundation, the Komen Foundation, and Blue Cross Blue Shield of Michigan for the Michigan Radiation Oncology Quality Consortium. RJ has a contract to conduct an investigator initiated study with Genentech. RJ has served as an expert witness for Sherinian and Hasso and Dressman Benzinger LaVelle. RJ is an uncompensated founding member of TIME’S UP Healthcare and a member of the Board of Directors of ASCO.