A scoping review was performed and reported using the Preferred Reporting Items for Systematic Reviews and Meta-Analyses extension for scoping reviews (PRISMA-ScR) guidelines, and the checklist is provided in Multimedia Appendix 1 [16]. The protocol can be accessed through the Open Science Framework website [17]. Since this study focused on previously published literature and thus did not involve direct contact with human participants, institutional review board approval was not required [18].

We performed a comprehensive search in July 2019 using the following databases: PubMed; MEDLINE, EMBASE, and PsycInfo via Ovid; Cochrane Library via Wiley; Scopus; and Web of Science Core Collection. A combination of natural language and controlled vocabulary was employed in accordance with Methodological Expectations of Cochrane Intervention Reviews guidelines [19]. A complete search strategy is available in Multimedia Appendix 2. To ensure that we did not overlook potentially relevant items, we also reviewed reference lists of related systematic and scoping reviews and included studies. Results were compiled and deduplicated in EndNote (version X.9; Clarivate Analytics).

Papers employing strategies according to the stated definitions of social media and traditional methods were included. Papers also needed to be a clinical trial and describe at least one of the following outcomes: number of participants recruited or enrolled, cost of recruitment, or length of recruitment. Papers that contained no original data (editorials, letters to editors, opinions, conference proceedings, comments, systematic reviews), that recruited health care professionals as participants, or that did not report outcomes of interest were excluded.

Papers were screened in 2 stages: (1) title and abstract screening and (2) full-text screening using the aforementioned inclusion and exclusion criteria. Two independent screeners reviewed papers at both the title and abstract stage and the full-text stage (ID, CB, and TAB). We resolved discrepancies through discussion and consensus or through the intervention of a third party when necessary. Study screening was facilitated using Rayyan (Qatar Computing Research Institute).

Two independent reviewers charted data using standardized collection forms created in REDCap (Vanderbilt University). Again, discrepancies were resolved by discussion to achieve consensus. Data charted included study characteristics (goal of intervention, disease, mode of intervention), recruitment methods (social media platforms, traditional methods, other online methods), participant characteristics (age, sex, ethnicity, geographic location, type of residential area), and recruitment outcomes (number of participants recruited via each method, cost of recruitment, length of recruitment). We encountered differences in the way researchers described recruitment versus enrollment outcomes across studies. As a result, we defined recruitment as the first contact between prospective participants and study staff and enrollment as when participants were enrolled in the study after they signed informed consent. We extracted both recruitment and enrollment data when available.

Our search strategy identified a total of 5177 papers. After removing duplicates (n=2806), the titles and abstracts of 2371 papers were screened against the exclusion criteria, resulting in 95 papers for full-text screening. From this, 64 papers did not meet inclusion criteria (Figure 1); the remaining 31 papers were included in this review and described in detail below.

Of the 31 papers, 29 reported on a single study and 2 included reports on 2 studies each. This review identified a total of 33 studies that reported using social media and traditional methods in recruitment to clinical trials. The 33 studies described interventions for treatment (n=19) and prevention (n=14). A total of 19 studies consisted of in-person visits only, 10 were online only, and 4 consisted of in-person and online participation. Study diseases or conditions focused on lifestyle and behavior (eg, weight loss, smoking), neurology, obstetrics/gynecology, and HIV. Most studies (32/33) recruited adults (18 years or older), with 5 studies focusing on older adults and seniors; 1 study recruited adolescents. Detailed characteristics of included studies are presented on Table 1.

Facebook was the most commonly used social media platform (31/33), followed by Twitter (6/33). The majority of studies (21/33) used a single social media platform and the remainder used two or more social media platforms, with one study reporting the use of 4 platforms. Facebook use occurred across all studies reporting the use of more than one social media platform.

Studies that reported using Facebook for their recruitment used a variety of Facebook features: Facebook ads, Facebook pages (including boosted posts on a page) [10,36], or sending a friend invite to prospective participants using a Facebook account [32]. One study used untargeted Facebook ads, then switched to targeted Facebook ads that showed ads only to participants who met the study criteria [22]. Several studies started using social media after other recruitment methods were deployed [28,31,38,44], while others used social media at the beginning of their recruitment period.

The most commonly used traditional method was print (32/33), followed by in-person venues (17/33) and referrals from health care professionals (16/33). A total of 23 studies used other online media, such as website ads (15/23), emails (11/23), and Craigslist (8/23).

A total of 17 out of 33 studies reported overall enrollment rates. Of these, 9 studies achieved or exceeded their enrollment target. Of these 9 studies, 8 reported the proportion of participants enrolled through social media, which ranged from 0% to 49%. One study that achieved its enrollment target reported that social media outperformed other recruitment methods [48].

About half of the studies (17/33) reported both recruitment and enrollment rates from social media, 3 studies reported recruitment rates only, 11 studies reported enrollment rates, and 2 did not report either (Table 2). The proportion of participants recruited and enrolled from social media varied greatly from study to study (Table 2) and across study types (Table 3). Studies with a high proportion (50% or greater) of participants enrolled from social media recruited adult participants. Studies recruiting seniors [26,35,43] and the 1 study involving adolescents [32] enrolled the majority of participants from traditional methods (Table 3).

Out of 20 studies that reported recruitment rates by method, 7 reported higher recruitment rates from social media than from any other methods. Similarly, out of 28 studies that reported enrollment rates by method, only 9 reported higher enrollment rates using social media than any other methods (Table 4). Among the 9 studies that reported the highest enrollment rates from social media, 6 studies were for treatment (3 lifestyle-related conditions, 3 neurological) and 3 were for prevention (2 HIV, 1 lifestyle). Of the 9 studies reporting enrollment, 5 involved an online intervention and the remainder used an in-person intervention. A total of 5 of these 9 studies used Facebook only, and the rest used more than one social media platform.

A total of 20 studies reported itemized recruitment and enrollment costs. Of these, almost all studies reported the costs to place the recruitment ads in various media (19/20), and 6 reported the costs to develop these ads. Only 5 studies reported the costs of staff involved in participant recruitment and enrollment.

Of the 19 studies that reported cost per enrolled participant, only 4 [10,21,30,33] reported lower cost per enrolled participant using social media than any other methods. Recruitment methods used in these 4 studies included Facebook, Instagram, print, television, radio, in-person recruitment, and website ads. Detailed costs per enrolled participant by media type can be found in Table 5.

This scoping review examines literature describing the use of social media and traditional methods as well as other online media in clinical trial recruitment. We found that using social media resulted in the highest recruitment rates in 7 of 20 studies and the highest enrollment rate in 9 of 28 studies. We also found that social media resulted in the lowest cost per enrolled participant in 4 of 19 studies. However, the data reported about social media outcomes varied greatly across studies, obscuring our ability to evaluate if particular studies might benefit more from using social media.

Our review discovered a lack of consistency in defining and reporting recruitment and enrollment data across studies. Several studies seemed to use these terms interchangeably, and some failed to specify which data they reported (eg, when consent occurred). Furthermore, we noted inconsistencies resulting from the varied use of social media, even within a platform, and the opportunistic use of social media by several studies that started using social media only after traditional methods failed to achieve sufficient interest and enrollment. We also observed that some studies did not indicate how many participants were counted as being reached by more than one recruitment or enrollment method. Lastly, we observed inconsistencies in cost reporting, with some studies reporting itemized costs and others reporting aggregated costs. Collectively, these inconsistencies contributed to the variations in describing the outcomes of using social media and traditional methods to support clinical trial recruitment and enrollment. The inconsistencies in reporting also rendered comparison of data across the studies included in this scoping review impossible.

To address these inconsistencies, we recommend that future studies use the terms “recruitment” and “enrollment” in alignment with the definitions used by the Food and Drug Administration’s advisory committee [49]. The term “recruitment” should report on all interactions with potential study participants prior to obtaining informed consent, while the term “enrollment” should report only on the total number of individuals who provide informed consent. Studies should report recruitment using as the denominator the total number of participants reached and as numerators the number of study participants reached through each method. Enrollment rates should be calculated using as the denominator the total number of participants providing informed consent and as numerators the number of participants enrolled through each recruitment method. Additionally, screening that results in the exclusion of participants after they provide informed consent should be reported separately, enabling a more accurate count and cost of the individuals actively participating in the intervention. Using consistent definitions of key terms across research studies will allow researchers to improve the clarity and comparability of clinical trial management data. Establishing reporting standards will also allow researchers to adopt evidence-based recruitment strategies by choosing optimal recruitment methods to reduce cost and timelines [50].

To facilitate shared learning based on the experience of prior research, we further recommend that future studies report in detail their approaches to using each social media platform (eg, Facebook ad, Facebook page) and specify the number of participants recruited or enrolled from multiple methods, which will allow for more accurate response analyses and cost estimates. As the ability to capitalize on existing social media relationships may influence both the success of recruitment and enrollment as well as their subsequent costs, researchers should consider explicitly stating whether their social media recruitment strategies are related to a larger social media initiative.

When adding social media to existing recruitment methods, researchers should report both the incremental costs of social media and the personnel cost to develop and monitor posted content. Furthermore, researchers are encouraged to report trade-offs between cost and time when analyzing the effectiveness of each recruitment method. For example, social media may result in higher cost per enrolled participant but reduce the time to enroll participants. Although we recognize the complexity in analyzing and comparing cost in different time increments and the potential for historical trends to confound results, such reporting may provide deeper insights on different recruitment methods. Such analyses may help researchers, organizations, and corporations determine whether to build social media expertise directly into their research teams to optimize the full extent of social media capabilities or to engage social media experts as consultants. The costs associated with such arrangements should be identified and, ideally, reported.

The mixed results on the use of social media found in this study are similar to those found in a prior scoping review [12]. While the prior scoping review attributed the different success rates to the effort researchers put into the use of social media as a recruitment method, we found that the variations in the use of social media and the inconsistencies in the way outcomes were reported prevented comparisons across studies. Having explored the ways researchers used and reported the use of social media in clinical trial recruitment, this study advances recommendations for achieving data consistency in reporting, which would facilitate comparison across results.

Similar to the findings from a systematic mapping study conducted by Frampton et al [51], we found that the use of social media to recruit participants involved a variety of diseases and health conditions. We also failed to find many studies involving minority populations (eg, African American or lesbian, gay, bisexual, transgender, and queer populations). As a result, we are unable to provide any insights on the use of social media to recruit diverse populations when compared with traditional recruitment methods.

Our scoping review is limited in its examination of data from clinical trials that combined social media and traditional methods in their recruitment strategies because we cannot guarantee that the review identified all relevant studies. We also cannot claim this review is comprehensive, as research not organized or reported as a clinical trial may inform reasoning about incorporating social media into the research recruitment armamentarium. The relatively recent use of social media also renders highly speculative any conclusions drawn about its appropriateness in recruitment and enrollment methods for particular diseases and conditions. We also recognize a limitation in our ability to assess the quality of the recruitment strategy, the recruitment materials, or the enrollment process. However, we advance recommendations to improve consistency in reporting on recruitment and enrollment that we believe to be necessary precursors for comparative analysis.

The use of social media for clinical trial recruitment holds great promise. However, this scoping review identified continued inconsistency in reports on the use of social media for clinical trial recruitment and enrollment. We recommend that future studies incorporate the recommendations for collecting and reporting recruitment and enrollment data advanced here to facilitate comparison of study data and shared learning.