Participants were 121 pregnant women attending a new pregnancy intake at an outpatient clinic that is part of a large health system in the Midwest. Eligibility criteria included being 18 years or older of age, understanding spoken English, and being pregnant with the intention to carry the pregnancy to term.

Data collection began in September 2018 and concluded in May 2019. An RA was present at the clinic on 2 half days per week; during this time, willing participants were introduced to the RA by the nurse who was completing the intake with the patient. At all other times, clinic staff gave patients a flyer describing the study and provided a website (via QR code, along with a unique login ID) through which patients could enroll in and complete the study. The flyer was provided by intake nurses at the end of the appointment, and participants used their own device to complete the study.

Participants who enrolled in the study completed a series of screening questions regarding substance use before and during pregnancy, as well as questions related to demographics, general health, and technology access. Those recruited by the RA were given a tablet to complete the study at the clinic following their intake appointment. Those enrolling in the study via the flyer used their own device and completed screening at a time and place convenient for them. In addition, those who screened positive for any substance were offered within the app to participate in a subsequent extended assessment (duration of 10 to 20 minutes) with separate consent. The app would link participants to the next screening if they agreed to participate.

All procedures for the study were approved by both the university (#085518B3A) and health system (#12267) institutional review boards. The app used for data collection read aloud the consent form that explained to participants that the study has 2 parts. Electronic information sheets were used for the study. Participants agreed to participate in the study in the computerized questionnaire by clicking a box and then answering the questions. There were no physical copies of the consent form. Part I included content regarding broad health behaviors such as nutrition and sleep, as well as brief questions on smoking, alcohol, and marijuana use during the month before they became pregnant. For part II, patients who screened positive for smoking, alcohol, or marijuana use in the month before becoming pregnant were invited to complete a 15- to 20-minute survey asking additional questions about risk factors. This assessment included more sensitive information regarding substance use, traumatic experiences, partner violence, depression, and anxiety. Participants who completed part I received a US $10 Target gift card, and those who were eligible and completed part II received an additional US $20 Target gift card. No identifying information was collected until after participants completed the assessment items. Once participants completed the portions of the study they were eligible for, they were linked (within the survey) to a separate survey, where they entered their email or phone number to receive their gift cards and a copy of the consent form. The data participants gave in order to send the gift cards were kept in a separate password-protected spreadsheet from the rest of the data and were destroyed once the study was complete. Participants who completed the survey in the clinic received the consent form and gift card directly from the RA.

All participants were asked to complete 47 items regarding alcohol, marijuana, and tobacco use before pregnancy and during the past month, as well as questions about pregnancy and general health. These items included, but are not limited to, the following:

Chi-square analyses compared differences between in-person– and flyer-based recruitment as well as differences in disclosure on the T-ACE, WIDUS, and each item of NIDA Quick Screen. Chi-square analyses used all available screening information from each participant. However, participants with missing items were dropped from that specific analysis. Two individuals had missing data for the NIDA Quick Screen binge drinking and tobacco questions. One person had missing data for the NIDA Quick Screen prescription drug and illegal drug use questions. There were no missing data for the T-ACE or WIDUS.

Study participants were primarily Black and African American (92/121, 76%) and had a mean age of 27.7 (SD 4.9) years (Table 1). Approximately half (66/121, 54.5%) of the participants had completed some education beyond high school.

Between-group differences in positivity rates were examined for 2 validated screening tools, the WIDUS and the T-ACE. A total of 20 (31%) out of 64 participants recruited through flyers screened positive for drug use risk on the WIDUS versus 9 (16%) out of 57 participants recruited at the clinic (χ²₁=4.0; P=.05). Additionally, a total of 24 (38%) out of 64 women recruited through flyers screened positive for alcohol risk on the T-ACE versus 11 (19%) out of 57 participants recruited at the clinic (χ²₁=4.9; P=.03; Table 2).

Chi-square analyses compared the disclosure of substance use (on the NIDA Quick Screen) for participants recruited in person versus via the flyer. Questions included the frequency of binge drinking (4 or more drinks per day), tobacco use, prescription drugs for nonmedical purposes, and illegal drugs in the past month. Each NIDA Quick Screen item was treated as dichotomous reflecting either any or no reported use (Figure 1). There were no significant differences across groups in binge drinking (χ²₁=0.1; P=.72), tobacco use (χ²₁=0.3; P=.58), prescription drug use for nonmedical reasons (χ²₁=2.0; P=.26), or illegal drugs (χ²₁=2.0; P=.15).

Of the 121 participants recruited overall, 57 were recruited directly by the RA, and 64 responded to the flyer. Nurses handed out 377 flyers resulting in 64 participants, representing 17% of those given the flyer. In contrast, of 109 patients completing a new pregnancy intake when the RA was in the clinic, 63 (57.8%) were introduced to the RA by the intake nurse, and 57 (91%) of the 63 agreed to participate and completed the screener (57/109, 52.3% of all available patients). Notably, flyer recruitment showed a stable increase over the course of the study (Table 2). This increase occurred following the introduction of a new approach, in which the RA and project coordinator began attending monthly staff meetings and updating nurses on the study progress, bringing in snacks, and building upon the relationships with the intake nurses in the clinic. This change in approach started in January, with an increase in nurse engagement and enthusiasm for the study happening in the next few months, resulting in an increase in flyer recruitment because many more were handed out. During the winter months of February and March, there were fewer intakes overall because of the weather. This decrease corresponds with an expected decrease in in-person recruitment during those months.

Of the 57 participants who were recruited in the clinic and used the tablet provided by the RA, 56 (98%) reported owning a smartphone, and 55 (96%) reported having an SMS text message plan on their phone. In total, 44 (77%) of these participants said that they would be willing to receive SMS text messages as part of a research study, and these participants also said that they would be willing to participate in additional surveys or programs if sent a link on their phone.

This study was set up to obtain substance use risk levels for women attending prenatal care at a large Midwestern hospital’s outpatient clinic, compare 2 different recruitment methods to examine which had higher recruitment rates and disclosure rates, and document participants’ access and comfort using smartphones and SMS text messaging for study participation. Recruitment via flyers distributed by health care staff was less efficient than when those same staff introduced patients to an on-site RA (57/109, 52.3% vs 64/377, 17% enrollment). However, participants in the flyer group were more likely to report substance use risk than those in the on-site RA group (20/64, 31% vs 9/57, 16% for the WIDUS and 24/64, 38% vs 11/57, 19% for the T-ACE). Most study participants owned a smartphone (56/57, 98%) and had an SMS text message package on their phone (55/57, 96%). Additionally, of the 121 participants, 94 (77.7%) were willing to receive SMS text messages or a link to further study participation on their devices.

Despite the lower overall enrollment compared to the on-site RA, the flyer approach requires less effort for medical staff and removes the need for a full-time RA at each study clinic. The flyer approach was also associated with greater disclosure on some measures of substance use. As is often the case, maintaining regular communication with clinic staff was particularly important in the flyer-based recruitment approach. These findings suggest that eligibility determination for substance use studies may be more successful and more representative (because of the wider possible reach with the same level of staffing) when using electronic screening with flyers rather than relying on full-time staff in the clinic. Flyer-driven recruitment appears to be a practical approach, given the high levels of access to technology among the pregnant urban participants and their willingness to use their personal devices for research. These latter findings are consistent with national survey data suggesting that smartphone ownership rates are high [27], and research suggesting that low-income patients are willing to use their own smartphone to participate in research [28].

The sample size, homogeneity of the sample, and preliminary nature of this research all contribute to clear limits in the generalizability of these findings. In addition, our sample size limited the ability to understand what variables may contribute to higher disclosure within the flyer recruitment group (ie, maternal age, parity, past substance use, or socioeconomic status).

Using electronic methods for eligibility determination appears to facilitate disclosure and, thus, recruitment efficiency. Although flyer-based approaches are less efficient than in-person recruitment with an on-site RA, they may also facilitate disclosure and can allow cost-effective recruitment at multiple sites. Even low-income patients in perinatal settings are very likely to own a smartphone and be willing to use their own device to participate in research. This method can allow for larger study samples by decreasing the amount of money needed to support full-time RAs in each recruitment site. Instead, 1 RA could be used across multiple sites, which can free up funds for a larger number of site locations. This can allow for a wider variety of participants across the country and could be more translatable and easier to replicate or continue once funding ends.