Digital Trajectories to Care in First-Episode Psychosis

Participants ages 15–35 were recruited from Northwell Health’s inpatient and outpatient psychiatric departments. Youths with psychotic disorders were recruited between March 2013 and September 2017 from the Early Treatment Program, Northwell Health’s early psychosis intervention clinic, as well as from eight collaborating early psychosis programs located in Alabama, Alberta, California, Florida, Georgia, Missouri, Ontario, and Oregon. Eligibility criteria included an FEP, a nonpsychotic mood or anxiety disorder, or attenuated psychotic symptoms identified within the past 24 months, along with a suspected IQ >70. Healthy individuals were also recruited as a control group. Given our focus on DUP reduction, here we present data pertaining to participants with early psychosis, and we plan on presenting data from other cohorts in subsequent publications. The study was approved by Northwell Health’s Institutional Review Board and locally at participating sites. Consent was obtained from adults and from legal guardians of minors. Assent was obtained from minors.

Participants were interviewed with the Pathways to Care for Psychosis Questionnaire (PCP-Q). This questionnaire was developed by three of the authors (MLB, CUC, and JMK), with additional input from experts in early psychosis. The PCP-Q consists of 81 open-ended and multiple-choice questions designed to retrospectively explore trajectories to care emphasizing online resources used to obtain information about emerging symptoms and inform the decision to seek care. [A copy of the PCP-Q is available in an online supplement to this report.] In addition, questions adapted from items on the Topography of Psychotic Episode (9) were included to extract DUP and mode of symptom onset (acute versus insidious). All interviews were conducted by three members of the coordinating institution (authors MLB, AFR, and KF), either in person or via two-way video.

Most questions were closed ended, asking participants to pick one response from a list of available options, to provide a date (month/year), or to provide a number (quantity/frequency). Participants were occasionally encouraged to choose multiple responses from a list of available options (for example, “rank top 5 social media sites used”). Some questions offered the opportunity to elaborate by listing an “other” option. Responses to open-ended questions (for example, “Which changes did you first notice?”) were initially assigned to one of several categories (e.g., preoccupations, perceptual disturbances, or sleep dysfunction) by two of the authors separately and again together to achieve consensus. Results are presented as descriptive statistics. Frequencies and percentages are reported for categorical variables. Means and SDs are used for normally distributed continuous variables; medians and interquartile ranges are used for nonnormally distributed variables. Histograms, q-q plots, and the Shapiro-Wilk test were used to determine whether the continuous variables were normally distributed. Differences in sex, age, and race-ethnicity were examined by using chi-square analysis for categorical variables and independent-samples analysis of variance (ANOVA) or Kruskal-Wallis test (nonparametric ANOVA) for continuous variables. No significant differences were found in sex, age, and race-ethnicity except for greater duration of social media use in the older cohort.

Altogether, 190 patients and 79 control group participants were interviewed. Of these, 112 had a diagnosis of a primary psychotic disorder (Table 1). The mean±SD age of the 112 patients was 22.5, and 68% were male. All participants reported either current or past social media use. Ninety percent of patients reported current regular use, and they reported checking social media 9.0 times daily and spending an average of 1.8 hours daily online. The 112 participants reported using social media on average for the past 7.6 years, predating diagnosis by approximately five years. Facebook was listed as the most popular site (86%), followed by Instagram (62%). The most common reasons cited for social media use (N=109) included communication with others (N=89, 82%), followed by sharing information (N=32, 29%) and gathering information (N=29, 27%).

Participants reported first noticing preoccupations and delusions (76%), followed by changes in sleep patterns (55%). Stress was the most frequently reported perceived contributor to emerging symptoms (39%), followed by illicit drug or alcohol use (32%). Most participants (66%) believed that early symptoms would resolve on their own without intervention, whereas 36% considered that changes might get worse. Participants reported waiting a mean of 36.6±102.4 weeks before disclosing their symptoms to anyone.

The Internet was the most widely used resource for participants seeking mental health information. Approximately 62% reported using online search engines to understand early changes. This was followed by seeking information from medical professionals (42%), family (24%), and friends (12%). Although the Internet was listed among the most helpful resources by some participants (N=22, 20%), it was also listed among the least helpful resources by others (N=9, 8%). Participants (N=104) were most interested in understanding reasons for early changes (N=50, 48%), followed by how to manage symptoms (N=35, 34%).

Altogether, 74% of participants reported noticing changes in their social media habits while symptoms were emerging. Approximately 24% (N=27) of the entire sample reported that others expressed concern or noticed changes in their social media activity. A minority of participants (19%) explicitly shared their experiences with others via social media. Only 6% (N=7) stated that social media activity affected their trajectory to care.

Most participants reported a desire to obtain digital resources and online interventions. Fifty-nine percent stated that they would have felt comfortable if they had been proactively identified and engaged by a mental health professional via the Internet or social media while symptoms were emerging. Seventy-six percent responded favorably to the possibility of receiving online mental health help or support via the Internet. Eighty-one percent reported a willingness to share their online activity data with researchers.

The Internet and social media may offer opportunities to improve pathways to care through innovative identification, outreach, and engagement strategies. Our data reinforced several relevant findings. Online resources, including Google, Facebook, and Twitter, were an important part of daily life for participants with early psychosis. Online activity often predated illness onset and continued throughout the DUP, which offers the prospect of earlier intervention. Importantly, participants with early psychosis expressed positive attitudes toward Internet- and social media–based intervention efforts.

Online search engines have become a primary source of mental health information for the general population (10). For many participants, results of online search queries represented a primary source of information during symptom emergence. However, despite the ease and popularity of search engines as a vehicle for exploring early experiences, some participants described the Internet as being among the least helpful resources owing to the vast amount of misinformation available. Most participants attributed early changes to increased levels of stress, and some incorrectly anticipated resolution of symptoms without intervention, possibly contributing to treatment delay. The impact of inaccurate information gathered online supporting participants’ misconceptions about mental illness remains to be fully appreciated.

Some participants chose to obtain information or support via social media by self-disclosing symptoms to their social network. This finding is in line with existing work suggesting that social media is increasingly being used in the general population as an opportunity to share personal health information, including mental health–related information (11). Although social media is not an anonymous setting, it is transforming the ways by which people communicate and can offer opportunities to seek social support and feedback from individuals online through a personalized and selective social platform. Despite the popularity of social media, few participants reported that social media activity affected their pathway to care, suggesting that the unique potential of this channel has yet to be fully developed.

At the time of symptom emergence, participants appeared to be mostly interested in obtaining information to understand the cause of their experiences. Few participants were searching for traditional referral information to begin psychiatric intervention. This finding is consistent with prior work suggesting that many who utilize the Internet as a mental health resource have yet to connect with treatment (5) and may have little, if any, desire to do so. This result highlights an important discrepancy in DUP reduction efforts. For prospective FEP patients, accessing care may not be a priority. Mental health clinicians may need to refocus efforts on providing digital tools that meet the needs of their target audience. These tools should be designed to facilitate a smooth transition to care once a prospective patient is ready to take that step. Unfortunately, utilization and engagement with existing digital mental health tools are poor (12), and new and innovative solutions need to be explored.

Harnessing social media and Internet-based technologies provides opportunities to improve the ways we identify, engage, and meaningfully interact with individuals experiencing emerging psychosis. Mental health clinicians have an opportunity to contribute to the online universe by developing effective mental health Web sites and social media pages. These resources must be designed to compete with the abundance of inappropriate, stigmatizing, and misleading Web sites. At the same time, there is a growing body of literature that describes use of personalized online digital data in search engines and social media platforms to automatically identify individuals who may be experiencing mental illness (13,14). Advertisers mine online data to identify customers and tailor advertisements that have the highest likelihood of influencing behavior. Similar efforts to apply available technology to identify, engage, and facilitate treatment initiation have begun in psychiatry (15). Nearly all participants in our study were engaging in online and social media activity during symptom emergence, prior to receiving care, which offers the prospect of earlier detection and intervention online. As efforts to successfully identify and effectively interact with individuals with emerging psychosis improve, these tools may become increasingly relevant in targeted DUP reduction initiatives.

The study had noteworthy limitations. First, interviews were retrospective and limited to participant recall. Although participants had been given a diagnosis within the past 24 months, accurate recall of use of and response to online resources can be challenging. Furthermore, the degree of insight varied, and discrepancies were occasionally noted between recollected accounts from participants and medical records. In addition, at the time of the interview, all participants were receiving care, which may have affected their memory of events. Finally, differences may have existed in variables, including speed of symptom onset, first identified psychotic symptom, access to technology, education, and race-ethnicity, which may affect pathways to care and which need to be further delineated.

Information gathered online appears to play a role in help seeking and may represent one of the first proactive steps toward treatment initiation. As technology companies, including Facebook and Google, expand efforts to improve mental health resources for their users, clinicians must take an active role in studying and refining these tools. Mental health clinicians should focus on mapping out the precise digital trajectories from online search query to receipt of mental health services and explore points of potential intervention. Furthermore, clinicians need a better understanding of how individuals interact with and respond to online resources throughout the trajectory to care in order to develop successful interventions. Finally, stakeholders should focus on improving the array of online resources available to individuals with psychosis, providing them with the tools they seem to be most interested in obtaining at different stages of illness development. These tools should be designed to provide information and support as well as encourage and facilitate treatment initiation at the right time.

The authors thank the following individuals for their assistance with participant recruitment: Felix Amirfathi, Marzena Boczulak, Joy Brasfield, Georgia Carstensen, Hazel Doctor, Nandan Katiyar, Carrie Kriz, Catherine Marshall, Dielle Miranda, Christina Plagiannakos, Sheila Pugh, Sarah Sprague, and David White.