The survey population consisted of 115 outpatient psychotherapy clinicians associated with McLean Hospital in Belmont, Massachusetts, who completed a Web-based survey regarding their use of social and electronic media as part of their standard clinical care. The principal investigator emailed the survey to the 244 individuals included on the McLean Hospital master outpatient clinician mailing list. Emails were sent out in 2 rounds of email blasts between April 19, 2018, and June 6, 2018. The first round of emails went to 102 outpatient clinicians who work within formal hospital clinics, and the second went to 142 clinicians doing hospital-based private practice. For both rounds, reminder emails were sent 3, 5, and 8 days following the original email. Of the 244 email recipients, 13 email addresses were deemed undeliverable, resulting in 231 valid email addresses. With 115 total respondents out of 231 recipients, the overall response rate for this survey was 49.8%. The respondents spanned a range of professional disciplines and reflected the diversity of psychotherapy provided at McLean hospital. The email link to the survey specified that the purpose of the survey was to gain an understanding of whether and how clinicians were accessing electronic media in regular care. We did not collect demographic information on clinicians.

We were unable to identify an existing instrument that would adequately allow us to capture the breadth of information regarding clinician attitudes and behaviors around use of electronic media. Hence, we developed our own instrument designed by consensus among senior study investigators (KR, CB, and IV) to capture naturalistic use of electronic media in therapy. We built in items to facilitate stratification by qualification and years of experience. We elected to limit information about clinician demographics to minimize survey time and maximize response rate. We next built items to capture what types of electronic media platforms were used and in which patient populations this approach had been tried. Finally, we developed items to capture with greater detail the process of accessing this information, maintaining privacy, and its usefulness.

The survey, which is available in Multimedia Appendix 1, was created using Google Forms and consisted of 17 questions. First, respondents were asked to indicate their professional degree and number of years of clinical experience. Respondents were then asked to respond yes or no to whether they had viewed a patient’s social or electronic media as part of outpatient psychotherapy. If they responded no, they were asked whether or not they had considered viewing this content and the major factors or concerns they might consider in making the decision.

If they responded yes, they were prompted to answer 12 more questions aimed at characterizing their use of this content in psychotherapy. Clinicians were asked which platforms they accessed and how they accessed them—whether it was directly with the patient in the session, outside of the session with the patient’s permission, or indirectly via report from the patient or their loved ones. The survey also asked with which age demographic and clinical populations the clinicians used this approach, with how many patients they used it, and in approximately how many sessions per patient. Following these questions, clinicians were asked whether they believed access to this information helped improve the quality of care they delivered. Additional details were collected around the process of accessing this information—namely, which party had suggested access to this content and whether they had discussed issues around privacy. Finally, 2 free-response questions allowed respondents to share their reasons for and concerns about accessing this type of content in their clinical care.

The survey was distributed via a private Google Survey invitation to clinicians’ professional secure email addresses. Only those with the link were able to access it. Data from the survey were downloaded and saved on private Partners Healthcare servers. No identifying information was collected, unless the respondents chose to volunteer such information at the end of the survey; there was space for respondents to give their name and email address if they were interested in learning more about technology-related projects being conducted at the hospital. This question was removed from the dataset before data analysis to deidentify the responses.

All data analyses were performed using the SPSS 3 (IBM SPSS Statistics) statistical package. For quantitative items, descriptive statistics were generated, and then respondents were stratified by years of clinical experience and by professional degree, followed by a comparison of differences using chi-square tests. For questions in which respondents had the option to write free-text responses, these responses were consolidated into categories upon consensus by study staff and then recoded accordingly. As the volume of qualitative data was relatively low and we had distinct statements from each respondent, we elected not to use customized qualitative analysis software. Instead, we adapted a simple qualitative classification approach [12] and sorted qualitative statements into broad themes which this study investigators agreed to by consensus.

This project was undertaken as a quality improvement initiative at McLean Hospital and as such was not formally supervised by the institutional review board per their policies. The institutional review board provides a checklist [13] to determine whether a specific project qualifies as a quality improvement initiative. A total of 4 of the study investigators (KH, PM, PO, and IVV) assessed this study simultaneously and independently using this checklist, and all determined that the project met criteria to qualify as a QII. We also communicated this to the institutional review board who confirmed that if the study meant QII, formal supervision was not required. We utilized the existing mechanism for mass email communication with McLean clinicians to appropriately disseminate the survey, and we specified that responding to the survey was voluntary and part of an information gathering process.

Of the 115 clinicians who responded to the survey, 31 (27.0%) hold MDs as their professional degree, 35 (30.4%) hold PhDs, and 30 (26.1%) hold Licensed Clinical Social Worker (LCSW) degrees. Other degrees represented in smaller numbers include: Advanced Practice Registered Nursing degrees, MD/PhD, Licensed Mental Health Counselor degrees, and MA degrees. Moreover, 47 (40.9%) respondents have less than 10 years of clinical experience, 29 (25.2%) have 10 to 20 years of experience, and 39 (33.9%) have more than 20 years of experience.

Of the 115 respondents, 71 (61.7%) indicated that they had viewed at least one patient’s social or electronic media as part of psychotherapy. The remaining 44 (38.3%) indicated that they had never viewed a patient’s electronic media. A total of 1 respondent who indicated having viewed a patient’s media did not answer all the subsequent questions regarding this content. Therefore, some of the follow-up questions for clinicians who reported yes to viewing media had 71 total responses and some had only 70 responses.

Of the 44 respondents who have not viewed patient’s media, 10 (23%) said that they have considered incorporating this content into therapy sessions.

Table 1 summarizes the breakdown of patients’ social and electronic media platforms that clinicians viewed. A total of 60 of 70 respondents to this question (86%) viewed SMS text messages and 56 (80%) viewed emails, making these the most frequently accessed platforms. The use of either SMS text messages or email was significantly greater than the use of any of the other electronic media platforms (χ²₁=24.1, N=115, P<.001).

A summary of media viewing according to respondents’ clinical experience is shown in Table 2. A total of 33 of the 47 respondents (70%) with less than 10 years of experience and 20 of the 29 respondents (69%) with 10 to 20 years of experience stated they had viewed a patient's social or electronic media; meanwhile, 18 of the 39 respondents (46%) with more than 20 years of experience reported having done so. The relationship between years of experience and viewing rates was significant (χ²₂=6.1, N=115, P=.048). We also performed similar analyses stratifying clinicians by their professional degree but did not note any significant differences between groups or even nonstatistically significant trends.

Of the 70 respondents to this question who have viewed patients’ electronic or social media, 62 (89%) reported having viewed this content directly in the patients’ presence, and 46 (66%) indicated having gathered information about the media content through patient self-report, as demonstrated in Table 3.

Clinicians reported viewing social or electronic media with the following patient age groups: 20 of 70 (29%) adolescents, 46 (66%) young adults, 46 (66%) adults, and 5 (7%) older adults.

Moreover, clinicians indicated accessing this content most commonly with patients diagnosed with depression or anxiety; 42 of 70 (60%) of yes respondents stated that they viewed media with patients diagnosed with either or both conditions, followed by 32 (46%) with borderline personality disorder. Each of the remaining diagnoses yielded less than 35% of yes responses, as indicated in Tables 4 and 5.

A total of 9 of 70 (13%) clinicians who viewed their patient’s electronic or social media reported doing so only 1 time, whereas 40 (57%) clinicians reported doing so very infrequently. A total of 18 of 70 (26%) reported viewing media content in roughly every 5 to 10 sessions per patient, and 2 (3%) reported viewing it every 2 to 3 sessions. Finally, 1 (1%) reported viewing this content in every session.

Moreover, 12 of 70 (17%) clinicians who viewed their patient’s electronic or social media reported doing so with just 1 to 2 patients, 26 (37%) reported doing so with 3 to 5 patients, 14 (20%) reported doing so with 6 to 10 patients, and 18 (26%) reported doing so with more than 10 patients.

A total of 52 of 70 (74%) clinicians reported that it was the patient’s idea to incorporate social and electronic media into psychotherapy, whereas 10 (14%) clinicians reported that it was their own idea. A total of 7 (10%) indicated that it was a mutual suggestion, 1 (1%) reported that it was a family member’s idea, and 1 (1%) indicated that multiple parties suggested it.

We incorporated a single item asking clinicians to rate the extent to which they were able to provide more effective treatment in part because of accessing their patients’ electronic or social media. A total of 17 of 70 (24%) clinicians reported noting significant improvement in the level of care, whereas 30 (42%) reported seeing moderate improvement, 18 (25%) reported slight improvement, and 6 (8%) reported no improvement in their ability to deliver effective care. There were no differences on this measure when we stratified clinicians by years of experience or highest level of training.

We inquired whether clinicians had discussed privacy with patients and, if so, who initiated the conversation. A total of 44 of 71 (62%) clinicians who had accessed patients’ social and electronic media indicated that they discussed issues of privacy with the patients. We found that clinicians with greater experience were significantly more likely (χ²₄=13.2, N=115, P=.01) to bring up privacy concerns with their patients; 16 of the 18 (89%) clinicians with more than 20 years of experience who viewed this content reported having a conversation around privacy, whereas 12 of 20 (60%) of those with between 10 and 20 years of experience and 16 of 33 (48%) of those with less than 10 years of experience who viewed this content reported doing so.

The majority of clinicians who discussed privacy initiated the conversations themselves; only 2 of 71 (3%) clinicians reported that their patients raised privacy concerns regarding sharing this type of content.

Of the 71 respondents who indicated having accessed this content, 63 (89%) provided free-text explanations of their motivations for accessing patients’ media as part of clinical care. Using a qualitative analytic approach described in the Methods section, we categorized those responses by consensus into the following 5 general thematic categories, listed below in order of frequency. If comments fell into multiple categories, they were counted under each relevant category. Therefore, the percentages add to 117, rather than 100.

Of the 71 respondents who indicated having accessed this content, 48 (68%) provided free-text explanations of their concerns regarding accessing this content. On the basis of our qualitative analytic approach described in the Methods section, responses fell into 7 general thematic categories, listed below in order of frequency. In this case, all responses fell into just 1 category, and thus, were counted once.

This survey addresses the naturalistic use of patients’ electronic and social media by mental health clinicians. With a 49.8% (115/231) response rate, the survey results provide useful insights into the current practices among a diverse group of therapists. We noted that the majority of outpatient clinicians surveyed (115/71, 61.7%) reported having viewed at least one patient’s electronic or social media as part of care, with email and SMS text messaging emerging as the most frequently accessed platforms by far. We also found that clinicians who had been in practice for fewer years accessed this information more frequently than more experienced clinicians, recognizing that this is confounded by age cohort of the clinicians as well. This may reflect a greater proficiency among more junior clinicians in understanding how technology may be intrinsic to daily life [14,15]. Conversely, this may also reflect a lesser-perceived need for concrete collateral information among more experienced therapists, who may practice with a better-established frame [16].

Another prominent finding was that only 3% of clinicians who reported viewing social or electronic media indicated that any of their patients had voiced concerns around privacy. This was a surprisingly low number, which may reflect patient confidence in allowing therapists to access this information, or a hesitancy to disagree with a therapist’s suggestion because of power dynamics. It may also be related to our finding that the majority of clinicians accessed these data during sessions, in the patients’ presence. Accessing such personal information face to-face and giving patients control over the ability to share this information may have helped foster a sense of the privacy of the session extending to electronic data as well [17]. The ability to access collateral information, whether from family members [18] or through patient writings or other forms of expressive therapy [19], has long been a tradition in psychotherapy. Accessing electronic communications represents an extension of this tradition, incorporating contemporary means of communication and leveraging technology that may enable extraction on deeper behavioral signals from natural language.

Overall, our finding that a high percentage of clinicians are accessing electronic media and doing so with patients present—and in over 74% of cases, at patients’ suggestion—indicates that the process of incorporating this information into therapy is a common and organic process across various provider types (MD, PhD, LCSW). Clinicians reported accessing more private data, such as email and SMS text messaging, to a greater extent than social media, which may indicate clinicians’ perceptions that these more personal platforms contain more relevant behavioral signals. Our finding that 92% of clinicians reported that accessing electronic or social media improved their ability to provide effective treatment points to a role for increased attention to electronic media in care. Therefore, there is an imperative to explore whether this perception of benefit holds true in observational and experimental studies, which quantify potential benefit in a controlled manner and to understand the mechanisms underlying any observed benefit. Future studies are also needed to establish best practice standards that guide the appropriate access and use of this type of content across different providers.

Our findings should be interpreted within the context of the limitations of this survey. Although our respondents spanned a range of degrees and years of experience, we only surveyed clinicians affiliated with a single psychiatric institution in the Boston region, which may impact the generalizability of the findings. Furthermore, the patient population served by McLean Hospital therapists may not be representative of the population at large, which in turn may impact the representativeness of the sample with regard to use of electronic media as well.

In addition, our survey was designed to be brief, anonymous, and easy to complete to maximize response rates, rather than to be comprehensive. Consequently, we did not obtain details on provider demographics. We also did not ask clinicians to share the breakdown of diagnoses or age groups that they see in their outpatient practices, so results regarding media viewing by these variables may be skewed by the makeup of the overall patient population seen by clinicians in our sample. Collection of these data would be important in follow-up work. Although we do not believe this is a common occurrence, we did not specifically inquire whether clinicians accessed patients’ data without patients’ permission.

We acknowledge that our survey did not address all of the ways in which clinicians may be utilizing electronic platforms to obtain clinical information about a patient’s wellbeing. Specifically, we did not ask about the use of apps to track patients’ mood and behavior over time, for example, through daily mood or activity logs on one’s mobile phone. Apps of this type are emerging within psychotherapy, and clinicians may utilize these behavioral tracking methods in addition to or in lieu of the informal methods of accessing and discussing patients’ social and electronic media content that we laid out in the survey.

Finally, limitations associated with surveys, including responder bias and recall bias, apply to this work as well. As we did not track provider demographics, and the survey was anonymized, we are not able to compare responders with nonresponders and draw insights into the effects of responder bias. We recommend that future work, including from our team, study this issue to gain a better understanding of predictors of using electronic media in therapy.

Despite these limitations, our work provides an early indication that therapists in practice are incorporating information from digital platforms into the care process. Although evidence regarding the impact of this information is limited, our findings identified 2 primary purposes for accessing the information—to monitor and address a specific target behavior and to provide feedback on patients’ electronic communications and behaviors. This finding suggests that the additional information serves as a valuable tool in the therapeutic process beyond just collecting collateral information.

Future work on this topic should focus on replicating our initial findings in larger, more diverse clinician populations. More data are also needed to determine, in a more specific manner, exactly how and when electronic media may enhance or hinder the therapy process. Issues around privacy and confidentiality of this information also merit thoughtful discussion. This, in turn, may help develop a more systematic approach toward optimally utilizing electronic and social media in the augmentation of the therapeutic process in mental health care settings.