Sexual health survey recruitment approaches among patients in community health centers

Abstract

Introduction

Black and Latino communities are disproportionately affected by HIV and often underrepresented in sexual health research. It is crucial to understand their participation in research and gauge the impact of nonresponse bias in reported results. A sequential multimodal approach to recruitment could potentially increase the response rates and representativeness of the sample by increasing the likelihood of response from the participants. However, data on the use of various recruitment approaches (i.e., low-barrier strategies) for engaging patients from urban Federally Qualified Health Centers (FQHCs) in sexual health research is limited.

Methods

We analyzed survey recruitment data among patients prescribed PrEP at FQHCs serving largely Black and Latino/a communities to understand the impact of virtual recruitment. We used four sequential attempts over six months to recruit participants for an incentivized patient feedback survey on sexual health and PrEP services. We also examined demographic differences including age, sexual identity, gender identity, and race/ethnicity between responders and non-responders.

Results

We recruited participants from January to June 2022. The overall response rate was 20.8% (54/259). There was an incremental increase in responses based on number of strategies used: 66.7% (36/54) of participants were recruited by using one strategy (email only), 22.2% (12/54) were recruited by using two strategies (email and phone calls), and 11.1% (6/54) were recruited using three (email, phone calls and SMS) or four strategies (email, phone calls, SMS and an additional phone call). Responders had received a more recent PrEP prescription compared to non-responders; mean ±SD months between the most recent PrEP prescription and the first recruitment attempt date of the study was 15 ±17 months for responders and 27 ±24 months for non-responders. We observed no other significant differences in demographic characteristics between responders and non-responders.

Conclusion

Response rates to an online sexual health survey among urban FQHC patients were low using sequential multimodal virtual recruitment approaches, particularly among those without recent healthcare engagement. There were no demographic differences between responders and non-responders. Findings highlight the need for alternative innovative methods for research engagement and to reduce potential bias.

Introduction

Black and Latino communities are disproportionately affected by HIV in the United States, accounting for approximately 70% of new diagnosis although representing smaller proportions of the population. These disparities stem from intersecting structural and social barriers, including limited access to culturally responsive care, stigma, medical mistrust, and under-representation in sexual health research (1). These barriers underscore the importance of intentionally including Black and Latino populations in sexual health research to ensure interventions are equitable and effective. As such, patient experience surveys are an important tool in evaluating and improving healthcare provider performance, clinical care program delivery, patient engagement and retention (2).

In the past few years, response rates of patient experiences in primary care surveys has been negatively impacted, especially for racial and ethnic minorities (3). Ineffective recruitment approaches have often resulted in lower response rates and underrepresentation in research and quality improvement studies, contributing to bias in reported findings among these groups and potentially affecting quality of care (4). It is crucial to address underrepresentation in sexual health research to improve health equity and reduce HIV-related disparities. To address recruitment barriers among underrepresented and other hard-to-reach groups like patients with less recent care engagement, multifaceted recruitment approaches are needed. Traditional in-person clinic-based recruitment often yield low participation and incur high resource utilization, making it inefficient (5) (6).

Remote recruitment approaches (e.g., SMS, online, telephone) could be potentially more efficient in engaging minoritized patients for HIV-prevention or sexual health research (5). Previous studies have shown that using a multimodal approach to recruitment yields higher response rates than single-mode approaches (3), however, there is limited data on the impact of multimodal approach to recruitment of federally qualified health center (FQHC) patients in sexual health research. Sequential multimodal recruitment (3). Understanding potential differences between patients responding and not responding to studies on HIV-prevention is crucial to gauge non-response bias, generalizability, and inform quality improvement strategies. However, few studies, if any, have reported on such information, particularly among patients who often hold multiple minoritized statuses. Additionally, no data exists on the number of attempts used towards recruitment at urban federally qualified health centers (FQHCs) among those prescribed HIV pre-exposure prophylaxis (PrEP).

We therefore analyzed recruitment data from an online patient feedback survey on sexual health and PrEP care among patients prescribed PrEP from FQHCs serving largely Black and Latino/a communities to understand the impact of sequential multimodal remote recruitment attempts, explore differences between responders and non-responders, and assess the dose-response relationship between virtual recruitment attempts and response rates.

Methods

This study was conducted in the Bronx, New York, a region with high HIV and STI burden and a federally designated Ending the Epidemic priority jurisdiction (7). The Bronx is comprised of over 90% Black and Latino identifying individuals and amongst the most socio-economically disadvantaged counties in New York and the U.S. We identified 259 patients (aged ≥18 years) prescribed PrEP between January 2015 and December 2021 in nine Bronx FQHCs using a validated algorithm (8). We extracted phone numbers and email addresses from electronic health records (EHR) as these were the only contact details available to the research team, along with demographic information including age, race, gender identity, sexual orientation, insurance status, and the most recent PrEP prescription date. We used four remote recruitment attempts (email, phone call, SMS, and a combination of phone call and SMS – described further below) based on our previous experiences from web-based recruitment approaches, in sequential order from January to June 2022 to recruit patients for a web-based survey on sexual health and PrEP support services. Participants were offered a $25 electronic gift card upon survey completion. Up to three additional recruitment attempts were used for reaching individuals not responding to the first recruitment attempt. All recruitment attempts indicated the brief nature of the survey (less than 10 minutes), the incentive, and that findings would help inform care improvements.

Email (First attempt):

We sent emails using an institutional email with an institutional logo, Principal Investigator contact information, and an individualized survey link to all eligible patients (N = 259). Only non-responders received up to three reminder emails to complete the survey over a period of seven days to provide survey response without being overwhelmed.

Phone call (Second attempt):

Patients not responding to emails (223), received a phone call; if reached and verbally consenting to participate, the survey link was immediately sent by SMS or email according to their preference. Only non-responders received a voicemail followed by one additional phone call for recruitment.

Text Message/SMS (Third attempt):

To those not responding to phone calls (211), we sent an SMS that included a study flyer and an individualized link to the patients. Only non-responders to the SMS received one additional SMS reminder to complete the survey.

Phone Call + SMS (Fourth attempt):

Finally, patients not responding to the previous three attempts (207) received a final round of a combination of phone call followed by SMS. We left a brief voicemail to non-responders and indicated they could also text or call us back or click on the link sent to complete the survey.

We calculated the frequencies of total responders from each attempt and reported them as response rates. We explored differences between responders and non-responders by demographic characteristics and recency of PrEP prescription using Chi-Square or Independent t-tests. We also reported the dose response relationship of attempts, and responses received. The study was approved by Albert Einstein College of Medicine Institutional Review Board.

Results

Of the 259 patients prescribed PrEP, 31.3% were cisgender male, 22.8% gay, 48.5% Hispanic, and 67.9% aged <40 years (Table). Overall, 54 patients (21%) responded to the survey (Table). Response rate by e-mail (first attempt) was 14% (36/259), by phone call (second attempt) was 5% (12/223), by SMS (third attempt) was 2% (4/211) and, by a combination of phone call and SMS (fourth attempt) was 1% (2/207). Of the total respondents, 67% (36/54) were recruited with the first recruitment attempt, 22% (12/54) were recruited with the second attempt, and an additional 11% (6/54) were enrolled using third or fourth recruitment attempts. There was a significant difference in recency of PrEP prescription with responders having received a more recent PrEP prescription than non-responders; mean months (SD) between last prescription and enrollment date was 15 (17) months for responders and 27 (24) months for non-responders (P <.001). There were no other significant differences between the responder groups (Enrolled vs. Non-Enrolled) by observed demographic characteristics (Table).

Table.

Demographic characteristics by enrollment status and recruitment attempts

	Total Enrolled	Number of Virtual Recruitment Attempts to Enrollment			Not Enrolled
Characteristic		1*	2**	3 or 4***		P value
	n (%)	n (%)	n (%)	n (%)	n (%)
TOTAL (N =259)	54	36	12	6	205
Age (Years)						0.435
18–29	15 (27.8)	11 (30.6)	3 (25)	1 (16.7)	54 (26.3)
30–39	28 (51.9)	19 (52.8)	5 (41.7)	4 (66.7)	79 (38.5)
40–49	7 (12.9)	4 (11.1)	3 (25)	1 (16.7)	38 (18.5)
50 and older	4 (7.4)	2 (5.6)	1 (8.3)	0	34 (16.6)
Gender Identity						0.342
Cisgender Female	17 (31.5)	11 (30.6)	3 (25)	3 (50)	32 (15.6)
Cisgender Male	11 (20.4)	9 (25)	2 (16.7)	0	70 (34.1)
Transgender Female	1 (1.9)	0	1 (8.3)	0	8 (3.9)
Transgender Male	0	0	0	0	2 (1)
Unknown	25 (46.3)	16 (44.4)	6 (50)	3 (50)	93 (45.4)
Sexual Identity						0.959
Gay	11 (20.4)	9 (25)	2 (16.7)	0	48 (22.7)
Bisexual	1 (1.9)	1 (2.8)	0	0	3 (1.4)
Pansexual	0	0	0	0	1 (0.5)
Heterosexual	15 (27.8)	9 (25)	3 (25)	3 (50)	45 (21.3)
Unknown	27 (50)	17 (47.2)	7 (58.3)	3 (50)	114 (54)
Race/Ethnicity						0.358
Black/African American	19 (35.2)	14 (38.9)	1 (8.3)	4 (66.7)	57 (27.8)
Hispanic/Hispanic White	23 (42.6)	13 (36.1)	8 (66.7)	2 (33.3)	89 (43.4)
Hispanic Black	3 (5.6)	3 (8.3)	0	0	10 (4.9)
Asian	1 (1.9)	0	1 (8.3)	0	4 (2)
Non-Hispanic White	2 (3.7)	1 (2.8)	1 (8.3)	0	10(4.9)
Multiracial	1 (1.9)	1 (2.8)	0	0	1 (0.5)
Unknown	4 (7.4)	3 (8.3)	1 (8.3)	0	7(3.4)
Other	1 (1.9)	1 (2.8)	0	0	27 (13.2)
Mean number of months since last PrEP prescription (Mean [SD] in months)	15 (17)	-	-	-	27 (24)	<0.001

^*First attempt – Email only

^**Second attempt – Phone call only

^***Third or Fourth attempt – SMS only or Phone call along with SMS

Discussion

Among patients prescribed PrEP at urban FQHCs, only 21% responded to and completed the survey despite using sequential multimodal remote recruitment approaches and offering incentives. This response rate was lower than that reported by other studies using online recruitment of populations affected by HIV (9) (10), suggesting ineffectiveness of multimodal online recruitment approaches to engage urban FQHC patients in sexual health prevention research or increasing sample representativeness. The observed low response rates could have been influenced by diverse factors including the sensitive survey topic or perceived irrelevance (i.e., sexual health), competition for digital attention, mistrust of digital platforms, socioeconomic conditions (e.g., unstable internet access), digital or other literacy, and lack of organizational trust (11). Combining virtual and in-person modes of recruitment, the use of trusted messengers, and patient involvement in designing the study may yield better response rates by overcoming some of these barriers and merit further research (e.g., qualitative inquiry to better understand barriers and solutions to improving participation in sexual health studies).

Our study found no differences in demographic characteristics assessed between the study groups. The non-responders had older dates of last PrEP prescription than responders, likely indicating a less recent sexual healthcare encounter. This observation suggests that there are gaps in engagement with PrEP care that may impact PrEP continuation for non-responding patients, suggesting the need for innovative strategies to enhance PrEP care retention in urban FQHCs. The low response rate and gaps in PrEP care engagement throws light on the impact of recruitment attempts on response rate, potentially leading to underrepresentation and is pivotal in gauging non-response bias. While a majority of respondents were enrolled using the first recruitment attempt (email), it took multiple (2 or 3) additional attempts to recruit another third of the respondents. This is consistent with other studies demonstrating that using multiple attempts and diverse recruitment strategies can increase survey response rates (12) (13).

A limitation of this study includes our inability to compare relative effectiveness of the remote recruitment approaches (e.g., telephone vs SMS vs email) given the non-randomized design. Despite attention to recruitment design and survey wording, the online survey likely could not overcome digital and reading literacy barriers among some potential participants, which merits investigation of how literacy impacts response rates. However, concurrent use of multiple approaches likely represents real-world practice and increases the chances of a patient’s preferred mode being offered to them, further impacting the response rates.

Our findings demonstrate the impact of a sequential multimodal recruitment approach for recruiting urban FQHC patients for sexual health research. These results underscore the need for innovative recruitment methods that leverage technology, patient partnerships, and hybrid virtual and in-person outreach approaches to enhance representation and reduce reporting bias, especially for those without recent care engagement.