Although early psychosis intervention provides the best hope for a full functional and symptomatic recovery, the time between symptom onset and receiving appropriate care remains alarmingly high (
1). Results from the Recovery After Initial Schizophrenia Episode-Early Treatment Program (RAISE-ETP) trial (
2), which included 34 clinics in 21 states across the United States, suggest that the median duration of untreated psychosis is 74 weeks (
3). Considerable research has demonstrated that shortening the duration of untreated psychosis leads to better outcomes (
2,
4). Identifying and engaging youths with early-stage psychotic disorders in order to facilitate timely treatment initiation remains a major public health challenge.
Prior public outreach and educational initiatives have demonstrated that it is possible to reduce the duration of untreated psychosis through widespread informational campaigns and easily accessible services. The Treatment and Intervention in Psychosis (TIPS) project in Norway (
5,
6) and the Early Psychosis Intervention Programme (EPIP) project in Singapore (
7) both successfully utilized mass media to educate the public and relevant professionals regarding the benefits of early intervention and available services. In the United States, a similar early detection program that integrates professional outreach with an online educational and social marketing initiative is currently under way (
8). These strategies require extensive and costly resources and organization and focus primarily on improving public education and awareness. Limited efforts have focused on directly identifying, targeting, and engaging prospective patients.
One effective strategy for targeting and engaging specific populations online is through Google AdWords (
9), which enable advertisers to highlight advertisements by having them appear near Google search results. As opposed to other methods of digital outreach, Google AdWords provides a unique opportunity to present an intervention in the form of an online advertisement at the very moment that someone is engaging in information-seeking behavior through online search inquiry. This type of digital marketing strategy has been used to efficiently broaden the reach of traditional, community-based public health campaigns (
10). From a mental health perspective, recruitment campaigns using Google AdWords have demonstrated success in encouraging research participation (
11–
13) and have been shown to be less expensive and more effective than traditional outreach (
14,
15).
The Internet may prove to be a valuable resource for understanding and facilitating help seeking by patients with early psychosis and their families. Google search is currently the most widely used online search engine and manages more than three billion searches daily (
16). The Pew Research Center found that 72% of Internet users have searched online for health-related information within the past year (
17). Furthermore, the Internet has become a primary source for youths wishing to seek out mental health information (
18). This is particularly true for stigmatized illnesses such as schizophrenia (
19). In addition, the Internet appears to be a primary source for information gathering for both patients and their families while psychotic symptoms are emerging, and use of the Internet may represent one of the first proactive steps toward help seeking (
20). Furthermore, information gathered online has been shown to affect health-related decisions and help-seeking behavior (
21,
22).
Although advertisers have long used the Internet to target consumers directly beyond the capabilities of traditional media (
23), limited efforts have focused specifically on applying available technology to target and identify help-seeking individuals in real time and refer them to appropriate resources. We set out to explore how one could take advantage of existing digital advertisement technology in order to reach prospective patients. The specific objectives were to determine whether digital ads are a reasonable method to reach individuals who are seeking help for psychosis, to understand the types of digital advertising messages most likely to engage a user, and to understand how users engage with different forms of available online content.
Methods
Northwell Health’s Early Treatment Program (ETP) and Strong365, a not-for-profit organization, collaborated on a digital outreach campaign aimed at understanding the role that digital media can play in help seeking for early psychosis. Strong365 and its National Youth Leadership Board, a group of 16- to 29-year-old persons with personal experiences of psychosis recovery, served as the creative team responsible for crafting the messaging and Web and ad content and for managing the ad campaign performance and analytics.
Landing Page Design
A landing page was developed with the primary goal of encouraging help-seeking individuals in the New York City metro area to contact the ETP. [Further details about the layout of the landing page are available as an online supplement to this article.] In order to optimize visibility, the landing page focused on two core messages believed to be most relevant to help-seeking youths and their families: understanding early signs and symptoms of psychosis and finding local help.
The campaign landing page offered connection to the ETP through links to the ETP Web site, e-mail address, and phone number and the opportunity to “ask a question” by using an online form in which an ETP staff member could call, text, or e-mail an answer; examples of personal storytelling emphasizing the message that “you are not alone” via videos and blogs featuring participants sharing their stories of recovery through treatment; and the Prodromal Questionnaire–Brief (PQ-B), an evidence-based screening tool that visitors to the page could use to assess psychosis risk (
24).
The campaign landing page was integrated into a parent Web site (
www.strong365.org), offering visitors the opportunity to explore additional educational information about psychosis, treatment options, recovery stories, and resources. Content was informed by individuals with personal experiences of psychosis as well as mental health professionals with expertise treating early psychosis.
AdWords Campaign Design and Tracking
In order to provide the best opportunity to reach the intended audience, Google AdWords was utilized. Google AdWords allows advertisers to select keywords/phrases, create a linked advertisement, and allocate a budget and time frame. The ads appear in response to online searches that match selected keywords. Individuals who click on the ads are then directed to a dedicated landing page. Google AdWords are pay-per-click (the advertiser pays when a user interacts with ads), and ads can be demographically and geographically targeted. The campaign ran in New York City for 14 weeks between April 11 and July 18, 2016 with a total budget of $1,427.
More than 2,000 keywords/phrases were selected and organized across 15 thematic ad groups (
Table 1). [Further details about the keywords are available in the
online supplement.] In order to identify keywords, experts at the ETP were consulted, and feedback was obtained from the Strong365 National Youth Leadership Board. Within each thematic ad group, multiple ads were crafted that would appear as search results when a relevant keyword was typed into Google search. [Examples of ads are available in the
online supplement.] General terms such as “psychosis,” as well as more specific terms such as “hearing voices” and “mind control,” were included in order to target both individuals searching for information for themselves and persons searching on behalf of a loved one. Keywords related to bipolar disorder were included in order to cast a wide net, given the occasional diagnostic uncertainty at the onset of illness (
25).
A total of 154 unique ads were crafted, each comprising a headline, description line, and call to action. The headline thematically related to the user’s search and drove ad impressions (an impression was counted each time the ad was shown on a search result page). The description line offered supporting information, and the call to action encouraged searchers to perform a desired action (take a quiz, click on the ad, find support, etc.). Google AdWords’ algorithm selects ads based on a user’s search, the ad’s relevance to that search, and competition from other advertisers. The more a keyword was searched by users, the more frequently relevant campaign ads were shown.
Measures and Data Analysis
Google analytics were used to record Web site engagement (time spent on the Web site, number of pages visited, and activity on the Web site, including watching the video and completing the psychosis screener). Google analytics was also used to collect demographic information and information about interests of users who interacted with the ads (
26,
27). Help-seeking actions were recorded by ETP staff based on e-mails and phone calls received. The Google AdWords platform recorded the number of impressions per ad group, keywords used within each theme, number of clicks linked to the Web site, and cost per click.
Results
Ad Performance
Table 2 presents the number of impressions, clicks, click-through rate, and cost per click by ad theme. The ads appeared 191,313 times and were clicked on 4,350 times, reaching a click-through rate (impressions/clicks) of 2.3%, at an average cost of $.33 per click, and $25 per person contacting the ETP (total cost of the campaign/number of individuals contacting the ETP). Compared with the Health and Medical Industry average (
www.wordstream.com), the campaign had an above-average click-through rate (2.3% versus 1.8%) at a favorable cost per click ($.33 versus $3.17) and cost per desired outcome or action ($25 versus $126.29). The top five ad groups/themes in terms of number of clicks were bipolar disorder signs and symptoms, mental illness test/psychosis test, delusions/suspicion, schizophrenia signs and symptoms, and psychosis signs and symptoms.
Whereas the headlines drove impressions, supporting text affected the number of clicks. [Further details about the number of impressions, clicks, and click-through rate by description line are available in the online supplement.] The most clicked ads had a description line that read, “You’re Not Alone. We Can Help” (93% of clicks). The next two most popular description lines were “Hard to Tell What Is Real?” (3% of clicks) and “Getting Help Early Matters” (1% of clicks). The campaign’s top clicked call-to-action lines included: “Take an Online Quiz/ Find Support” (55% of clicks) and “Learn the Signs/Find Support” (38% of clicks). Thematically, ads related to bipolar disorder had the most ad clicks over the campaign period, with 3,234 clicks (74% of total campaign clicks).
User Background
According to Google analytics, among recorded sessions of those who clicked on ads (N=3,257), 74% (N=2,410) were female and 26% (N=847) were male. Age distribution was as follows: 18–24 (N=393), 25–34 (N=499), 35–44 (N=596), 45–54 (N=746), 55–64 (N=654), and ≥65 (N=369). Eighty percent (N=2,606) clicked ads through mobile devices, 10% (N=326) through tablets, and 9% (N=293) through desktop computers. The top three interests of those who clicked ads were movies, news (entertainment and celebrity), and television. The peak time of day for ad clicks was between 5:00 and 6:00 p.m.
User Engagement
The median time visitors spent on the landing page (session duration) was 51 seconds (interquartile range=74 seconds; mean=115 seconds). The median number of pages viewed on the Web site (not including links to other relevant Web sites) per session was 1.29. Fifty-six percent (N=2,436) of users ended their sessions on the landing page, and 44% (N=1,918) clicked further into the Strong365 site. Landing page visitors engaged in a range of help-seeking actions (
Table 3).
Fifteen percent (N=671) of landing page visitors completed an online psychosis screener, and 453 individuals (10%) scored positive. All were encouraged to reach out to the ETP in order to learn more. Only 1% (N=57) directly contacted the ETP. In the 14 weeks prior to campaign initiation, the ETP received on average of 2.5 community referrals weekly. During the campaign, the ETP received on average of 5.5 community referrals weekly. Among those who did contact the ETP, none were enrolled. They were either lost to follow-up or were ineligible based on symptoms (not experiencing a primary psychotic disorder) or symptom duration (more than 2 years) and were referred elsewhere.
Discussion and Conclusions
Several important lessons were learned from our Google AdWords pilot project. Digital ads appear to be a reasonable and cost-effective method to reach individuals who are searching for behavioral health information online. During the 14-week campaign, approximately $1,500 was spent and 4,350 individuals clicked on the ads. Many users took additional help-seeking steps, including obtaining psychosis-specific information (N=1,918, 44%), completing a psychosis self-screener (N=671, 15%), and contacting the ETP (N=57, 1%). Engaging individuals by targeting those who are actively searching online is likely more efficient, direct, and cost-effective than previously employed forms of generalized education and recruitment through newspapers, radio, cinema, and social media (
7,
28).
Although it is unclear if the individuals engaging with the campaign were experiencing psychosis, the relatively high rate of self-screener completion and positive scoring for psychotic experiences suggest that the target population was reached. In addition, the campaigns relatively high click-through rate and favorable cost per action compared with the Health and Medical Industry average suggest that the ads were relevant to search queries. Prior work has shown that youths in the early stages of illness spend a significant amount of time online prior to making clinical contact (
20). This method of outreach and engagement may, therefore, be a potential adjunct to traditional forms of outreach and may be capable of reaching individuals who might not otherwise come into contact with traditional referral sources. Furthermore, given the geographical targeting, digital outreach strategies can be implemented locally by mental health clinics with minimal associated costs.
These results suggest that messaging content can affect the efficacy of digital outreach. The types of digital advertising messages most likely to engage a user had a description line that read “You’re Not Alone. We Can Help.” In previous work, concerns arose regarding the poor quality and stigmatizing nature of available online content related to psychosis and potential negative effects on treatment initiation (
21). Given the robust response to the “not alone/help and support” messaging, reminding people that they are not alone and offering hope might be an important foundational message for future campaigns.
Users appear to engage differently with online content depending on the content’s form. Many who interacted with the ads seemed to be interested in obtaining information, based on the percentage who clicked deeper into the site for educational material. In line with prior work (
20), this finding suggests that individuals searching online for behavioral health information might be more interested in education as opposed to treatment options. This is consistent with increasing demand for digital forms of mental health support (
29). A study using 7 Cups of Tea (7 Cups), an online support platform pairing users with trained listeners, demonstrated that users seek online support for a variety of psychological difficulties and are satisfied with the support they receive (
30). Furthermore, an adaptation of 7 Cups as an adjunct treatment for youths with early psychosis revealed that patients were interested in having an online outlet for emotions and socialization (
31). These data imply that individuals searching online might be more willing to engage with remote-based tools rather than with traditional health care. Digital supports may serve as stepping stones that help to facilitate the transition from online inquiry to formal clinical care.
Despite encouragement to reach out to the ETP that was given to all landing page visitors, including the 10% who scored positively on the psychosis screener, only 1% directly contacted the ETP. Although it is unclear whether the same individuals who contacted the ETP were among the cohort who scored positively on the screener, this gap highlights the complex step between obtaining online information and reaching out for help, which should be further explored.
Limitations
There are several noteworthy limitations to this study. Although Google analytics is capable of providing insight into the demographic characteristics of searchers, it is impossible to know precisely who they are and why they are searching. It is unclear if ads were interacting with a prospective patient, family member, concerned friend, or someone else entirely. However, based on the searches conducted and interaction with Web site content, one can assume that these are individuals using the Internet to search for information related to psychosis. Another limitation is the competition for specific keywords, notably on behalf of pharmaceutical companies. Certain keywords, such as schizophrenia and bipolar disorder, are already linked to drug company ads, which directly affected the number of times the campaign ads were served and where they were displayed on the search results page. Finally, Google does not allow advertisers to directly target teenagers. Given the goal of early identification and engagement, we have begun to explore targeting youths through social media platforms such as Facebook, Instagram, and Twitter, which allow for advertisement to this demographic group. In addition, we plan to link to secondary school, university, and community college Web sites.
Future Directions
Campaign results reinforce the need to better understand the population of users who are searching for psychosis-specific information online. In order to do so, researchers will need to employ a variety of methods, including running campaigns with limited keywords targeted to a very specific population (parents, for example), deeper analysis of the characteristics of users who enter the Web site, and deeper analysis of engagement by users once they are on the Web site. Future campaigns should also consider exploring various types of advertisements (video, for example), as well as different messaging content and images, to determine which ads most effectively engage specific populations. In addition, the dominance of ad impressions and clicks related to bipolar disorder suggests that future campaigns should consider how best to educate and engage individuals with psychotic symptoms more broadly.
Campaign results also reinforce the need to understand the many steps from online search inquiry to clinical treatment. Given the campaign’s high levels of engagement among users once they are on the site, there is an opportunity to explore different forms of online engagement tools, rather than focusing specifically on immediate referral to specialty care. It is possible that the leap between searching for information online and making contact with a clinical team is too big for many individuals as an initial step. Intermediate steps toward treatment in the form of digital engagement tools would also allow researchers to learn more about what users are seeking and why, as well as the barriers they face in considering treatment for psychosis.