This is an open-access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0/), 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.
When patients need health information to manage their personal health, they turn to both health professionals and other patients. Yet, we know little about how the information exchanged among patients (ie, patient expertise) contrasts with the information offered by health professionals (ie, clinician expertise). Understanding how patients’ experiential expertise contrasts with the medical expertise of health professionals is necessary to inform the design of peer-support tools that meet patients’ needs, particularly with the growing prevalence of largely unguided advice sharing through Internet-based social software.
The objective of our study was to enhance our understanding of patient expertise and to inform the design of peer-support tools. We compared the characteristics of patient expertise with that of clinician expertise for breast cancer.
Through a comparative content analysis of topics discussed and recommendations offered in Internet message boards and books, we contrasted the topic, form, and style of expertise shared in sources of patient expertise with sources of clinician expertise.
Patient expertise focused on strategies for coping with day-to-day personal health issues gained through trial and error of the lived experience; thus, it was predominately personal in topic. It offered a wealth of actionable advice that was frequently expressed through the narrative style of personal stories about managing responsibilities and activities associated with family, friends, work, and the home during illness. In contrast, clinician expertise was carried through a prescriptive style and focused on explicit facts and opinions that tied closely to the health care delivery system, biomedical research, and health professionals’ work. These differences were significant between sources of patient expertise and sources of clinician expertise in topic (
Patients offer other patients substantial expertise that differs significantly from the expertise offered by health professionals. Our findings suggest that experienced patients do not necessarily serve as “amateur doctors” who offer more accessible but less comprehensive or detailed medical information. Rather, they offer valuable personal information that clinicians cannot necessarily provide. The characteristics of patient expertise and the resulting design implications that we identified will help informaticians enhance the design of peer-support tools that will help meet patients’ diverse information needs.
In addition to the indispensable information received from health professionals, patients use information and advice offered by other patients to help them actively participate in their own health care and make informed personal health decisions. Although patients are best known for providing emotional support, they also offer other patients personal health guidance based on the expertise they have gained from managing similar health situations. We define
In contrast to other forms of social support, including
Patient expertise has been valued in varied and growing contexts. For example, personal knowledge, such as lifestyle, priorities, and experiences, is an important contribution patients make to shared decision making with health professionals [
In this work, we focus on patients sharing their expertise with one another. Breast cancer patients, for example, have expressed a strong need for experiential health information provided by peers [
Historically, patients who share similar health situations have helped one another to cope with illness by sharing their expertise through participation in patient-led support groups [
Patient expertise has continued to gain visibility as Web-based social software (eg, forums, social networking tools, blogs, and wikis) helps patients to readily exchange information and advice with others who are facing similar health situations [
Despite the growing prevalence of expertise sharing among patients on the Internet, disparate views about the characteristics of that expertise remain. For example, Meier et al [
Facilitating patient-expertise sharing with innovative technology will depend on a solid understanding of the fundamental characteristics of the expertise that patients share. For example, could we meet patients’ needs for information solely by enhancing communication between patients and health professionals? Alternatively, do patients also need help finding other patients who have had similar health experiences because clinicians have neither the time nor the expertise to meet all their needs? An important first step to answering these questions is to understand the role of both patient expertise and clinician expertise in meeting patients’ needs.
Our aim in this study was to enhance our understanding of patient expertise by assessing how it differs from clinician expertise. In the context of breast cancer, we conducted an in-depth and comparative content analysis [
Using an evolving coding scheme that was grounded in the data [
We analyzed sources of patient expertise and sources of clinician expertise from both online message boards and books. Message boards are common online resources for patients to seek advice from peers through online communities or to seek advice from health professionals through ask-the-expert forums. Books are traditional resources that patients also commonly turn to for advice, both authored by health professionals (ie, clinician expertise) and authored by peer cancer survivors (ie, patient expertise). Books are particularly important because they are one of the few written forms available to patients who have no Internet access. Although books offer a limited source of perspectives because of the short list of authors, they attempt to provide in-depth expertise from that perspective. In contrast, message boards bring insights into the kinds of expertise actively sought by patients from a breadth of perspectives. Although patients have available to them a spectrum of valuable resources, our combined analysis of message boards and books enabled us to capture a variety of expertise that is available to and sought by patients at both ends of that spectrum, both online and offline.
Sources of patient expertise in our analysis included 3 online message boards that support correspondence among breast cancer patients, and 2 books written by cancer survivors. We selected the 3 patient message boards (message boards A, B, and C) based on public accessibility, high volume of use, longevity, and variation in formality (ie, varied levels of moderation and affiliation with health-related organizations). We selected the 2 patient books because they are autobiographical yet differ in style. The first patient book (book 1) is highly narrative in its compilation of experiences contributed by 10 breast cancer survivors. The second patient book (book 2) is an interactive guide written by a 6-year survivor of metastatic cancer who provides extensive strategies for staying organized and informed during the cancer experience.
Sources of clinician expertise included an ask-the-doctor message board that supports correspondence between breast cancer patients and health professionals. We selected this message board over clinical advice summaries or health professionals’ blogs to enable analysis of questions from patients and answers from health professionals. As an additional source of clinician expertise, we chose
Content sources
Source | Text pages | Content units | |
Patient expertise | Message board A | 174 | 50 |
Message board B | 316 | 50 | |
Message board C | 276 | 50 | |
Book 1: McCarthy and Loren, 1997 [ |
230 | 79 | |
Book 2: Willis, 2001 [ |
220 | 131 | |
Total | 1216 | 360 | |
Clinician expertise | Ask-the-doctor message board | 277 | 150 |
Book: Love and Lindsey, 2000 [ |
552 | 225 | |
Total | 829 | 375 |
Characteristics of message boards
Message board | ||||
A | B | C | Ask the doctor | |
Year of inception | 1998 | 1994 | 1998 | 2000 |
Moderation | Yes | No | No | Yes |
Affiliation with a health-related organization | Yes | Yes | No | Yes |
Total messages | 379 | 152 | 316 | 300 |
Mean messages/thread (range) | 8 (1–31) | 3 (1–10) | 8 (1–27) | 2 (1–2) |
Days’ worth of threads | 5 | 24 | 55 | 85 |
In phase 1, we analyzed content from the sources of patient expertise. Our unit of analysis for message boards was the
We collected archived threads from the patient message boards with posting dates starting in February 2006 until we obtained 50 content units from each board that met our inclusion criteria. Obtaining an equal number of content units from each patient message board required the collection and filtering of more threads from message board B (130 threads) than from message board A (66 threads) or message board C (81 threads). Common kinds of threads that we excluded from the analysis reflected exchanges of pure emotional support, technical support issues, threads labeled by correspondents as
Based on themes that emerged from our preliminary analysis of informational support exchanged in the patient message boards [
In
In
Codebook of (a) topics and (b) recommendations (percentages reflect proportions of content units from each type of content source)
Patient message boards | Patient books | Ask the doctor | Clinician book | ||||
|
|||||||
|
|||||||
Deciding on care teams, treatments, and procedures | 16 (11%) | 9 (4%) | 19 (13%) | 14 (6%) | |||
Understanding biomedical concepts and processes | 49 (33%) | 6 (3%) | 102 (68%) | 145 (65%) | |||
Managing interactions with health professionals | 2 (1%) | 15 (7%) | 17 (11%) | 1 (0.5%) | |||
Managing information to collaborate with clinicians or understand biomedical issues | 2 (1%) | 3 (1%) | 6 (4%) | 5 (2%) | |||
|
|||||||
Managing life at home | 8 (5%) | 16 (8%) | 0 (0%) | 0 (0%) | |||
Managing work life | 3 (2%) | 14 (7%) | 1 (0.7%) | 1 (0.5%) | |||
Managing one’s emotional response to illness | 12 (8%) | 11 (5%) | 0 (0%) | 5 (2%) | |||
Managing interactions with social networks | 8 (5%) | 18 (9%) | 1 (0.7%) | 3 (1%) | |||
Managing personal tasks and projects | 16 (11%) | 86 (41%) | 1 (0.7%) | 19 (9%) | |||
Managing advocacy and volunteer work | 6 (4%) | 2 (1%) | 0 (0%) | 0 (0%) | |||
|
28 (19%) | 30 (14%) | 3 (2%) | 32 (14%) | |||
Total content units | 150 | 210 | 150 | 225 | |||
|
|||||||
|
|||||||
Prescriptive | 248 (23%) | 303 (14%) | 122 (35%) | 474 (13%) | |||
Narrative | 192 (18%) | 223 (10%) | 0 (0%) | 27 (1%) | |||
|
|||||||
Prescriptive | 159 (15%) | 419 (19%) | 225 (65%) | 1,620 (45%) | |||
Narrative | 204 (19%) | 264 (12%) | 0 (0%) | 133 (4%) | |||
|
|||||||
Prescriptive | 96 (9%) | 70 (3%) | 0 (0%) | 76 (2%) | |||
Narrative | 48 (4%) | 97 (4%) | 0 (0%) | 3 (<1%) | |||
|
|||||||
Books | 13 (1%) | 11 (1%) | 0 (0%) | 195 (6%) | |||
Contact information | 17 (2%) | 23 (1%) | 0 (0%) | 314 (9%) | |||
Magazines and magazine articles | 2 (<1%) | 7 (<1%) | 0 (0%) | 15 (<1%) | |||
Multimedia | 0 (0%) | 4 (<1%) | 0 (0%) | 140 (4%) | |||
News articles | 19 (2%) | 11 (1%) | 0 (0%) | 2 (<1%) | |||
Poems and quotes | 5 (<1%) | 24 (1%) | 0 (0%) | 0 (0%) | |||
Research articles and academic journals | 11 (1%) | 64 (3%) | 0 (0%) | 350 (10%) | |||
Templates | 0 (0%) | 115 (5%) | 0 (0%) | 4 (<1%) | |||
Webpages | 70 (6%) | 482 (22%) | 1 (<1%) | 118 (3%) | |||
Miscellaneous publications | 0 (0%) | 86 (4%) | 1 (<1%) | 102 (3%) | |||
Total recommendations | 1084 | 2203 | 349 | 3573 |
We used the codebook to test the reliability of our coding procedure using a 10% reliability sample of content units. Based on the number of contributing units, we randomly selected a set of content units from each content source for the reliability sample. An independent coder (CL) applied the codebook to the reliability sample. We calculated kappa scores to determine the level of intercoder agreement between codes applied to the reliability sample by AH (1 of the authors) and by CL. We applied linear weighting to our kappa calculations [
In phase 4, we compared the kinds of topics discussed and the types of recommendations offered in sources of patient expertise versus sources of clinician expertise. We compared the distribution of topics and recommendations across patient sources and across clinician sources. Then, we explored differences in the proportions of subtopics as well as the types and styles of recommendations among content sources. Finally, we used Pearson’s chi square statistic to assess differences in the frequencies of topics and recommendations between sources of patient expertise and sources of clinician expertise.
We thought deeply about ethical considerations and evolving guidelines for conducting Internet-based research [
We analyzed 735 content units across all sources. Patient sources contributed 360 content units and clinician sources contributed 375 content units. Each content unit was associated with 1 topic, but usually with many recommendations. Content units contained 7209 recommendations in total. Content units from patient sources contained 3287 recommendations and content units from clinician sources contained 3922 recommendations.
Distribution of (a) topics and (b) recommendations across content sources (percentages reflect proportions from individual sources)
Patient message boards | Patient books | Ask the |
Clinician book | |||||
A | B | C | 1 | 2 | ||||
|
||||||||
Medical | 25 (50%) | 12 (24%) | 32 (64%) | 12 (15%) | 21 (16%) | 144 (96%) | 165 (74%) | |
Personal | 18 (36%) | 22 (44%) | 13 (26%) | 58 (73%) | 89 (68%) | 3 (2%) | 28 (12%) | |
Both | 7 (14%) | 16 (32%) | 5 (10%) | 9 (12%) | 21 (16%) | 3 (2%) | 32 (14%) | |
Total content units | 50 | 50 | 50 | 79 | 131 | 150 | 225 | |
|
||||||||
Action strategies | 215 (39%) | 119 (47%) | 106 (38%) | 300 (36%) | 226 (17%) | 122 (35%) | 501 (14%) | |
Knowledge | 200 (36%) | 52 (21%) | 111 (40%) | 368 (44%) | 315 (23%) | 225 (64%) | 1753 (49%) | |
Perspectives | 86 (15%) | 33 (13%) | 25 (9%) | 121 (14%) | 46 (3%) | 0 (0%) | 79 (2%) | |
Information resources | 54 (10%) | 47 (19%) | 36 (13%) | 49 (6%) | 778 (57%) | 2 (1%) | 1240 (35%) | |
Total recommendations | 555 | 251 | 278 | 838 | 1365 | 349 | 3573 |
Next, we detail the kinds of topics and recommendations that emerged from our analysis across content units from all sources. The descriptive detail we provide on topics and recommendations corresponds to the codes making up our codebook (see
Most content units fell into 2 broad topic categories: discussion that was mostly
Topics that were medical in nature involved problems or concerns about constructs or processes that are tied closely to the health care delivery system, biomedical research, and health professionals’ work. Medical topics often reflected discussion that could stimulate an improved understanding of the medical domain or strategies to better fit one’s life to the health care delivery system. Common clusters of subtopics that fell in the medical category with representative examples include the following:
(a) Deciding on health care teams, treatments and procedures, and research trial enrollment
“Tackling the selection of our health care team”
Being in a “dilemma about reconstruction”
Dealing with competing recommendations from different doctors
Deciding between “radiation and tamoxifen or chemo and radiation”
Deciding whether to have a biopsy
Determining eligibility for participation in “genetic research
(b) Understanding biomedical concepts and processes, clinical treatments, procedures and tests, side effects, and biomedical research
Understanding “cancer staging” and other medical terminology
Determining whether a “bone scan” is a typical part of cancer care
Discussing a “pathology report question”
Uncovering how the diagnostic process typically flows
Understanding effects of Arimidex on cholesterol.
(c) Managing interactions with health care professionals
“Good care is also about communication”
Determining when to seek a second opinion
“What can I expect” for my upcoming appointment
Understanding considerations doctors make when recommending treatments.
(d) Managing information to collaborate with clinicians or to understand biomedical issues
Tracking medications, pain, or side effects to share with your health care provider
Preparing information for appointments
“I was supposed to take the reports to a general surgeon, but I wonder if this is necessary, since nothing was found?”
Discussing a research article on the effectiveness of Herceptin.
Topics that were personal in nature involved problems or concerns around constructs or processes that are tied closely to one’s personal life, including ongoing responsibilities and day-to-day activities associated with family, friends, the home, work, and health-related activities that occur outside of the health care delivery system. Personal topics often reflected discussion that could stimulate the development of practical strategies to fit health management into one’s ongoing life. Common clusters of subtopics that fell in the personal category with representative examples include the following:
(a) Managing life at home
Recovering from medical treatments and procedures: “What to expect following surgery”
Keeping up with family and household responsibilities: Sharing my experiences with hospice planning
Maintaining oversight of legal, financial, and insurance issues: how to “keep track of your medical expenses.”
(b) Managing life at work
Shifting your work load during treatment: “Worry about health, not your job performance”
Considering the impact of cancer on work prospects and insurance:
Interacting with coworkers, colleagues, or clients during treatment: “Maintaining a work persona”
Deciding whether to work during treatment: “Have any of you gone back to work during part of your chemo?”
(c) Managing the emotional response to cancer
Coping with anxiety, anger, depression, and fear
“Finding ways to cope with the emotional roller coaster”
Managing the “fear of recurrence”
“Humor is a necessary healing component.”
(d) Managing interactions with one’s social network
“What to tell your children”
The “Fears of our loved ones”
Getting help to find others with a similar diagnosis
“Letting our partners know what we expected and needed.”
(e) Managing personal tasks and projects
Managing lifestyle and self-care, including diet, exercise, and meditation: the “Dixie cup method” to organize medications and supplements; dealing with scalp pain while losing one’s hair; seeking a good “self-massage video”
Focusing on spirituality and hobbies
Managing personal health information: “Identifying and utilizing information resources.”
(f) Managing advocacy and volunteer work
Sharing notices about upcoming cancer-related fundraisers
Reaching out to others: “Breast cancer has helped us discover our mission and taught us that we can make a difference.”
We placed content units that shared medical and personal topics fairly equally into the overlapping category
(a) Understanding biological concepts and processes AND Managing interactions with one’s social network
The risk of developing breast cancer is higher for women who have family history of cancer...Telling our mothers about our diagnosis and anticipating their responses were a source of major concern and anxiety for all of us.
(b) Managing interactions with health care professionals AND Managing personal tasks and projects
After all of your treatments are completed...write down how you feel in general terms about once a month. Not only will it assist you in communicating with your doctor but it will also give you a barometer by which to measure your progress.
(c) Deciding on treatments and procedures AND Managing work life
Schedule your chemotherapy right before the weekend so that it interferes with work as little as possible.
Recommendations offered across content units fell into 4 major types: action strategies (1589/7209 recommendations, 22%), knowledge (3024/7209 recommendations, 42%), perspectives (390/7209 recommendations, 5%), and information resources (2206/7209 recommendations, 31%). Whereas recommendations in the form of action strategies offered procedural knowledge through suggested tasks (ie, “things to do”), recommendations in the form of knowledge offered declarative knowledge through facts and opinions (ie, “things to know”). Perspectives recommended attitudes or belief systems (ie, ways of believing or approaching situations), and information resources recommended tangible artifacts (ie, “things to obtain and use”). We describe each type of recommendation in greater detail below.
During our analysis we also recognized style differences between the recommendations; some action strategies, knowledge, and perspectives were direct, or
Action strategies are recommended tasks to deal with a personal health issue. This procedural knowledge about specific and actionable tasks can contribute toward solving a problem
Recommended knowledge refers to informative facts and opinions that one can learn to deal with a personal health issue. Unlike action strategies that represent tasks, recommended knowledge reflects declarative descriptions of concepts or ideas a person comes to understand
Perspectives are recommended belief systems, attitudes, or philosophies that drive an overarching approach for dealing with a personal health issue, such as putting one’s faith in God or acting as a strong advocate for oneself. In contrast to action strategies and recommended knowledge, perspectives reflect high-level and generalized beliefs, values, or attitudes toward an overarching experience
Information resources are recommendations for obtaining and using specific tools and tangible items to deal with a personal health issue. A diverse range of information resources were recommended (see
Templates, which included structured lists, tables, and worksheets for correspondents to personalize by filling them in, were an unexpected type of information resource. Templates reflect an embodiment of expertise that offer scaffolding to organize thoughts or actions surrounding personal health issues, such as tracking one’s health status, side effects, medical expenditures, or day-to-day events, recording research evidence on treatments, preparing for medical procedures, or assessing one’s personal level of resilience, pain, or nutrition. In contrast to clinician-oriented templates (eg, drain logs for patients to record and communicate postsurgery recovery to clinicians) that draw upon professional expertise, templates created by patients draw upon patients’ personal health experiences. For example, patient book 2 and the clinician book both offered templates that outline considerations for choosing a clinician or care team. Both templates suggested assessing clinicians’ communication style, their involvement with clinical studies, and the ability to tape-record visits. However, the patient book also reflected the patient’s experience by recommending consideration of clinicians’ personal character, professional reputation, availability, and payment options. In contrast, the clinician book reflected the clinician’s experience by recommending consideration of clinicians’ explanations for clinical tests, their interactions around complimentary and alternative treatments, and whether they are threatened when patients bring in information from the media to discuss.
Although sources of patient expertise and sources of clinician expertise contained content units that spanned both medical and personal topics, the proportions of content units falling under each topic (ie, medical, personal, or both medical and personal) differed significantly between those sources (χ2
2[N = 735] = 233.4,
Although sources of patient expertise showed a high proportion of personal topics relative to clinician sources, the degree to which personal topics were discussed varied across individual books and message boards. For example, 58 out of 79 content units (73%) from the patient book 1 contained personal topics, whereas only 13 out of 50 content units (26%) from patient message board C did so (see
The most common medical topic discussed across all sources was “understanding biomedical concepts and processes,” making up 49 out of 150 content units (33%) discussed in patient message boards and 6 out of 210 content units (3%) in patient books, as well as 102 out of 150 content units (68%) discussed in the ask-the-doctor message board and 145 out of 225 content units (65%) in the clinician book (see
The most common personal topic discussed across all sources was “managing personal tasks and projects,” making up 16 out of 150 content units (11%) among patient message boards and 86 out of 210 content units (41%) in patient books, as well as 1 out of 150 content units (<1%) in the ask-the-doctor message board and 19 out of 225 content units (9%) in the clinician book (see
Although content units from both patient and clinician sources offered recommendations falling under all 4 types (action strategies, knowledge, perspectives, and information resources), the proportions of those types differed significantly between patient and clinician sources (χ2
3[N = 7209] = 482.1,
When we compared message boards alone and books alone, we also found significant differences in the types of recommendations offered between patient message boards and the ask-the-doctor message board (χ2
3[N = 1435] = 153.5,
When we delved further into the styles used to express recommendations, we found that action strategies, knowledge, and perspectives were frequently expressed implicitly through personal narratives in sources of patient expertise compared with the prescriptive style that was common to sources of clinician expertise. This difference in style was significant for action strategies (χ2
1[N = 1589] = 281.4,
Although sources of patient expertise and sources of clinician expertise offered similar proportions of information resources on average, the types of information resources that were most commonly exchanged differed between those sources. For example, sources of patient expertise offered more webpages, poems, quotes, and news articles, but fewer books, contact information, and academic journals or research articles than sources of clinician expertise (see
Results from this analysis show that patient expertise differs significantly from clinician expertise in topic (medical, personal, or both), type of recommendation (action strategies, knowledge, perspectives, and information resources), and style of recommendation (narrative vs prescriptive). Sources of clinician expertise were predominately medical in topic, knowledge-oriented in type, and prescriptive in style, whereas sources of patient expertise contained more personal topics that were carried through narrative-style action strategies and perspectives. These findings suggest that patients, by sharing their expertise about personal health, meet an important information need unmet by clinician sources. Our findings extend prior analyses of patient interaction with supportive evidence that differentiates patients’ experiential knowledge about personal health from the medical expertise of professionals. This contribution enhances our understanding about the fundamental nature of patient expertise and guides the design of peer-support tools that facilitate patient-expertise sharing.
Differences in the characteristics of patient expertise and clinician expertise support the notion that patients and health professionals possess different domains of health expertise [
In addition to clinician expertise obtained from health professionals, patients are finding new ways to reach out to other patients to exchange complementary personal health advice based on their own experiences through collaborative tools on the Internet [
Patient expertise-sharing tools are technologies that bring patients together to interact and exchange their personal health knowledge. Enhancements to tools that patients already use to exchange personal health information, such as health-related social software [
Design efforts to facilitate patient-expertise sharing can offer patients opportunities to interact with these collaborative technologies in ways that extend beyond the traditional, text-based message boards of the past. For example, participatory design work illustrates patients’ strong desire for online collaboration and networking tools, such as Facebook [
Design opportunities to facilitate patient-expertise sharing
Type of support | Design feature |
Collaboratively managing shared resources | Common space to share and interact with varied resources |
User-generated tags and folksonomies that are meaningful to patients | |
Methods for rating and recommending tailored resources | |
Narrative and template formats for sharing experiences and expertise | |
Locating patient expertise | Detailed user profiles that illustrate areas of experience and expertise |
Methods for people finding and matchmaking | |
Analytic tools for identifying topics of interest from user contributions | |
Safeguarding against misinformation | Features that preserve natural safeguarding strategies in a public context |
Change log to provide audit trail of corrections to content | |
Vetting features to note affirmation, rebuttal, or reference sources |
Designers should focus on developing common spaces for patients to manage the multitude of information resources they share together. The wide range of information resources (eg, webpages, books, articles, and multimedia) that patients exchange suggests the need for tools that enable patients to work together to create, annotate, store, share, and reuse content across a diverse range of formats and topics. Patients need help managing this full range of resources they recommend to and garner from one another. Collaborative features of social software, such as user-generated tags to organize content shared through a wiki, have the potential to support this need for collaboratively managing shared resources. For example, Weiss and Lorenzi [
Collaborative recommendation systems like these help users share their expertise by rating resources and benefit from each other’s views, opinions, and experiences through collaborative filtering [
Given the range of medical and personal topics discussed among patients in our analysis, medically oriented resources (eg, medical dictionaries and patient information summaries) would certainly make up a valuable component of collaboratively managed collections of patient resources. However, the prominence of personal topics (eg, tracking medical expenses, working during treatment, what to tell your children, and selecting wigs) suggests that a fundamental component of such collections must include nonclinical resources as well. These resources should provide advice on personal topics related to work, family, the home, and social relationships in the context of illness. For example, one of the threads we analyzed consisted of dozens of suggestions from patients on considerations to make when writing an “end of life memoir” for family members (eg, your favorite books, family heirlooms, hobbies you enjoy, and world travels). Other examples include discussions about favorite “juicer recipes” and “experiences with sick-leave policies.” The breadth of these personal topics could link to a full range of relevant information resources from multiple domains (eg, medicine, law, social work, art, cooking, community resources, and finance). Users could tag and annotate these resources collaboratively in ways that capture important contextual ties to their specific experiences and facilitate later reuse by other users [
A medical library model [
Our findings provide additional insights for supporting collaboratively managed collections of resources. The common style of personal stories used to express patient expertise (see also [
Designers should focus on developing tools that help patients find and connect with other patients who have specific kinds of experiences or expertise. As health-related use of social software grows [
Features of social networking tools, such as user profiles, can help bring users’ expertise to the surface, enabling a targeted search for patients with specific health and personal characteristics [
We could also leverage the solid foundation of expertise-sharing research conducted in other settings to make progress in this important design direction of supporting the locating of patient expertise. For example, when confronted with an unfamiliar problem, people in professional work settings locate needed expertise by identifying potential sources (eg, other people and artifacts) and selecting specific sources to approach for help [
Sally seeks advice about whether to work during chemotherapy. She wants to locate a patient who has already dealt with this situation (eg, “I want to find another mother of school-aged children who worked throughout chemotherapy”). She enters age, gender, and condition into a directory search service offered by a social networking tool for cancer patients. Unfortunately, she is overwhelmed by the large number of user profiles the system returns, which she must now manually review to find another user with the specific characteristics she is looking for. In particular, Sally needs awareness of not only the health condition and demographics of other users, but details about their specific knowledge and health-related experiences to answer questions, such as “Does this person have the experience I am interested in? If so, how recently? What is their experience level?”
Enhanced features that make specific and detailed health experiences of users more prominent could make Sally’s work much easier and tailored to her needs. For example, it was common for correspondents in the message boards we analyzed to preface their thread postings with detailed descriptions of their health experiences (eg, “I’m a 4-year survivor...”). Such details could be combined with a larger range of medical and nonmedical characteristics [
Designers should also focus on features that preserve and encourage self-correction, self-monitoring, and other natural safeguarding strategies already used by patients online. Some might fear that enhancing informatics support for patient-expertise sharing could lead to the spread of mistaken, misinterpreted, outdated, incomplete, or otherwise poor-quality information. Although the potential for medical misinformation certainly exists, studies have examined patient interactions in online health communities and found minimal levels of medical misinformation [
We did not assess the accuracy of information exchanged in the patient message boards that we analyzed, but our observations of message board correspondents were consistent with previous research on strategies used to actively safeguard against the potential for misinformation, such as self-correction [
Our observations point to the importance of preserving functionality that encourages patients’ natural misinformation-safeguarding strategies, such as vetting features within a public context, as health-related social software evolves to support patient-expertise sharing. In particular, our observations suggest support for audit trails that make content changes explicit and easy to log, and simple vetting features for noting affirmation or rebuttal (eg, thumps up/down) and for referencing source material.
The characteristics of patient expertise we present are derived from a deep exploration of content from selected message boards and books in the context of cancer. The codebook resulting from our analysis is necessarily shaped by diverse interests and viewpoints of book authors and message board correspondents from the sources we analyzed. For example, expertise captured from a book written by a single author might not be as diverse or transferable as the expertise of multiple voices from a book coauthored by several people. Thus, our findings might not capture the breadth of expertise across the wide array of resources available to patients. For example, message boards reflect patients’ information needs through discussion that is initiated by patients’ own questions or offers of support. In comparison, books written by cancer survivors could provide a less direct reflection of authors’ and publishers’ perceptions about patients’ needs. Furthermore, individual content sources vary in their predominance of personal topics. Despite these differences, we found a strikingly similar distinction between the patient expertise in both books and message boards and the expertise in clinician sources. While our findings illustrate the volume of patient discussion beyond the medical realm of personal health, additional research is needed to tease out issues of misinformation and deeper distinctions within medical topics discussed.
Although our in-depth effort was scoped to small diverse samples, the work yielded rich descriptions that provide a solid basis for understanding patient expertise as a critical facet within the breadth of patients’ information needs. Details of our codebook point to a range of information needs related to the personal side of health and contribute to a holistic view of the patient. Given the experiential nature of patient expertise, it is plausible that the characteristics we ascribe to this specialized form of knowledge are also reflected by the experiential knowledge that people develop from personally managing other health situations, such as diabetes, heart disease, or pregnancy. Although aspects of patient expertise we identified are specific to cancer, other aspects could be widespread and shared by patients with other conditions [
Future work could also explore how our design implications play out within patients’ expanding space of social participation on the Internet [
Our findings demonstrate that patient expertise differs significantly from the expertise of clinicians in topic, type, and style. Neither increasing the amount of time that patients spend with health care providers nor training patients with medical knowledge to become amateur doctors appears sufficient to meet the needs for patient expertise. Instead, we offer alternatives in the form of design directions for facilitating patient-expertise sharing with health-related social software. Patients provide other patients with a unique and valued information resource that complements expertise provided by health professionals. Patients deserve informatics support that can fill the breadth of their health information needs by facilitating this patient-expertise sharing.
This work was supported by the National Institutes of Health (NIH) grant R01 LM009143. We thank Ching-Ping Lin for her assistance with inter-coder reliability testing, John Gennari and Richard Boyce for their assistance in preparing this manuscript, and David McDonald, Huong Nguyen, William Jones, and the iMed research group at the University of Washington for their invaluable discussions and feedback throughout this project. We thank MedHelp International (http://www.medhelp.org) for their support.
None declared
Unified Medical Language System