The study was conducted through PatientsLikeMe® (PLM) [13], an online platform comprising many disease communities where patients with life-changing medical conditions are able to find other patients like themselves, learn more about their condition, and share information about their outcomes. PLM has over 10,000 registered members with T2DM and approximately 8000 reporting T2DM as their primary condition, many of whom also report significant comorbidities. Patients with T2DM joining the website are asked to share information about their disease through custom questionnaires that populate their profile.
Following institutional review-board approval, a mixed-methods research process, including a comprehensive literature review, qualitative one-on-one patient interviews (January 2014), and a quantitative survey (April–July 2014), was used to comprehensively assess patient preferences regarding self-management support and diabetes management programs. Informed consent was obtained from all participants.
An extensive literature review of program support and assistance for patients with T2DM was first conducted to identify areas of evidentiary gaps for further exploration in the qualitative patient interviews. The areas identified included understanding the context in which patients make decisions about support for their diabetes (family structure, sources of assistance available), what the most important symptoms and problems requiring management are, who gives what kind of support, and which support programs patients may have tried in the past. Patients’ descriptions of their own goals, the relative importance of different kinds of support and support sources, and their preferences helped to inform the content of a structured survey, which was subsequently administered to a larger group of patients with T2DM.
Qualitative interview process
A list was generated of PLM website users who reported T2DM in their patient profiles, had multiple log-in sessions including activity in diabetes forums, and were actively participating in discussions. Of the identified individuals, 44 were contacted, mainly by private message (using PLM email), and invited to participate in 1-h qualitative telephone interviews. Once 10 of the invitees had responded positively, no further invitations were sent out.
The respondents, half of whom were female, had a mean age of 57.5 years (range 34–78). Overall, 60.0 % (n = 6) were non-Hispanic white, and educational levels ranged from high-school diplomas to Master’s degrees.
Patients’ experiences with controlling blood glucose levels and managing symptoms and treatment regimens, and types of programs and support systems used were the key topics that guided the interviews. For a full description of the qualitative interview process, please see Additional file 1.
Qualitative interview findings
Respondents reported common comorbidities including hypertension, hypercholesterolemia, depressive disorders, and chronic obstructive pulmonary disease. The majority perceived their blood glucose levels to be under control, although they were concerned about their weight and activity levels. Of the T2DM support systems/programs that respondents reported having participated in, most were one-time programs in which all follow-up was initiated by a nurse, not by the patient themselves.
Respondents reported that much self-education was performed through reading both online and printed materials. Websites commonly consulted included those of the ADA, PLM, insurance companies, and medical-supplies companies, as well as general medical websites. This was in agreement with information given by respondents that they preferred their materials to be from a ‘serious’ source and to be accessible at the patient’s own convenience. A few female respondents strongly favored participation in small groups for encouragement, sharing, and friendship; however, none actually belonged to such a group.
When questioned about their expectations/hopes for management programs, a range of responses were given, including the desire to not have diabetes or to slow or stop disease progression, to decrease or stop medication use, and to avoid complications (e.g., neuropathy, foot ulcers/amputation, heart disease, vision problems, peripheral artery disease). Respondents reported being motivated by wanting to keep learning and to avoid the negative consequences of their disease. They reported being discouraged from participating in programs because of the difficulty in navigating certain health-care systems and from continued participation due to repetition of material.
When asked from whom they received help in the management of their condition and what form that help took, responses included spouse, adult children, other family members (who often also had diabetes themselves), and online friends. Support and encouragement from health-care professionals (HCPs) was mostly related to diet and medication; additional encouragement and praise on dietary decisions and/or weight loss were desired. ‘Being accountable to someone else’ was considered to be a positive thing.
Overall, and importantly, respondents perceived ‘programs’ as consisting of short-term education by HCPs, whereas ‘support’ was regarded as daily interactions with friends and family.
Concepts and themes regarding diabetes management and support strategies that emerged from the patients’ descriptions informed the creation of quantitative survey questions and response options that reflected the patients’ experiences. These themes included, but were not limited to, weight loss as an area of concern and as a goal for T2DM programming; the importance of modes of support, such as printed materials, in T2DM education; sources of support, such as spouses, partners, family, and friends, in effectively managing T2DM; and how personal expectations/hopes for T2DM can align with goals of T2DM support programming.
Quantitative survey process
Survey development
Based on the results from the qualitative interviews and literature review, a survey consisting of a maximum of 90 questions (accounting for branching and variable questions asked only in response to specific answers to prior questions) and including both closed (Likert-scale, multiple-choice, matrix, and numerical) and open (free-form text) response formats was created [see Additional file 2]. Patient-reported interview data were used to frame and develop questions for use in the quantitative survey. For example, descriptions of diabetes support programs obtained in response to questions from Part 3 of the qualitative survey [see Additional file 1] were used to construct the support program types mentioned in Sections IV, V, and VI of the quantitative survey [see Additional file 2] as well as to provide lists of commonly mentioned response options.
In the survey, respondents were asked to confirm the demographic information previously collected from their profiles by completing a basic demographics panel prior to addressing the main survey questions. Demographic data concerning sex, date of birth, ethnicity, race, educational level, and health-insurance type were collected. Location information, such as country and state, were re-collected from respondents at this point as well. In addition to the demographic covariates, the main survey was composed of seven domains which were framed from the patient perspective:
-
(1)
Overall quality of life – to understand all areas of health of the patient
-
(2)
Diabetes goals – to focus on what patients want from their disease control
-
(3)
Current and past programs – to assess program participation and preferences/dislikes
-
(4)
Preferences for self-support – to understand the types, sources, and formats of support that patients want when managing their diabetes
-
(5)
Support network – to give an insight into who (or what) comprises the patient’s current support network
-
(6)
Treatments and complications – to provide data on what patients have experienced
-
(7)
Comorbidities – to provide data on patient-identified additional conditions
An appended sub-survey queried patients on the following: their satisfaction with their health-care; difficulties managing symptoms; glycated hemoglobin A1c (A1C), low-density lipoprotein, and blood pressure measurements; use of oral steroids; weight-loss goals; health literacy; and self-reported treatment adherence.
Branching within the survey was incorporated to address program characteristics, information covered, program length, and outcomes for respondents who participated in a program. Approximately 20 of the 90 questions were in free-form text format, which queried survey respondents on what ‘other’ types of programs, information, or diabetes support they participated in, wanted, or preferred.
Study population
The online survey was fielded during April to July 2014 to active PLM participants reporting T2DM as a condition on their profile, who also reported residing in the USA or did not specify their location and who were aged ≥ 18 years. The initial pool of invitations was limited to the most active patients (those who had logged in to the site in the past 90 days). A second pool of invitations was created for participants who had lower activity (at least one log-in during the past year).
Survey fielding
Patients fulfilling study inclusion criteria were emailed an invitation to participate in a custom survey. The survey fielding was conducted in two waves, a pilot survey and a full survey. The pilot survey was fielded to assess the interpretability of questions by a sample of 100 members with T2DM, and remained open for 2 weeks. The full survey was fielded to the larger T2DM population and remained open for a period of approximately 12 weeks. For both the pilot and full surveys, the invitation to participate appeared as a private message when a T2DM patient logged in to the site. The invitation included a link directing the patient to where the survey could be completed online. Users who did not complete surveys within 3 days were sent an email reminder to participate in the survey.
Statistical analyses
The population for the main analysis consisted of all confirmed US-based patient respondents who fulfilled the inclusion criteria. As there were no changes to the survey, responses obtained during the pilot survey were merged with full survey results prior to data analysis. To provide additional context on the PLM T2DM community, survey respondents were compared with nonrespondents with respect to their demographic characteristics, any listed (additional) conditions, and their PLM website participation as a preface to the main analyses. Before analyzing survey results, available profile data (such as location, primary condition, and zip code) were matched to survey responses by a unique user ID number. Zip codes were then grouped according to area type (urban, suburban, rural) using US Census data.
Descriptive statistics were used to describe patient characteristics by demographic covariates of interest, and summary statistics were used to tabulate frequencies and relative percents, such as patient use of different self-management support systems. χ
2 statistics were used for categorical variables, and two-sample t-tests were used to compare groups for continuous covariates. Trend tests were used to test two-category predictor variables on ordinal-ordered outcomes, such as program participation and satisfaction with one’s health-care. Agreement between binary paired variables, such as past and current program participation, was assessed with the kappa statistic. Only P-values of ≤ 0.05 were reported, and all tests were two-tailed. Free-text or ‘other’ text response options were analyzed by summarizing key concepts by ‘theme’ and enumerating the mention of each theme. Quantitative analyses were conducted with SAS, Version 9.4 (Cary, NC, USA).
Ethics, consent, and permissions
This study was approved by the Western Institutional Review Board on December 31, 2013. Materials sent to study participants such as invitation messages, together with the interview guide, research information, and participant consent were reviewed and approved by the Review Board. Using online means, the nature of the study was explained to potential participants prior to participation, and informed consent was obtained for participation in the study via an online informed consent form. Verbal consent was obtained for the audio recording of verbal interviews.