Assessing the Success Rate and Costs of Common Recruitment Strategies for a Community-Based Environmental Health Study

ABSTRACT

Context:

Recruitment is a vital component of public health research to produce generalizable findings and analytical power for subgroup differences in health outcomes. Recruitment strategies in population-based health studies vary widely with limited information on response rates and expected costs.

Objectives:

To evaluate recruitment strategies to better inform future recruitment planning and practices based on response rates, implementation costs, and demographics of enrolled participants.

Design:

We cross-sectionally examined recruitment data from the Colorado Study on Community Outcomes from Per- and polyfluoroalkyl Substances (PFAS) Exposure (CO SCOPE). Passive strategies included announcements and word-of-mouth referrals; active strategies included in-person outreach via door-to-door canvassing and/or tabling at community events, and phone or email contact to participants from a previous study. We calculated response rates per population reached by each strategy and estimated costs of materials and staff time.

Setting and Participants:

In 2021–2023, we recruited and enrolled participants from 4 communities in El Paso County, Colorado. This site was part of a national multisite study of exposure to per- and polyfluoroalkyl substances in drinking water and associated health effects. We enrolled 916 adult participants (92% of our goal).

Results:

Overall participant response rate across all strategies was 0.5%. Response rates were lowest after announcements (0.1%), and highest from direct contact with previous study participants (32.3%). In-person outreach accrued the greatest number of enrolled participants but also at the highest estimated cost ($242/enrolled participant). Word-of-mouth referrals recruited the greatest proportions of young adults (aged 18–34) and people with a high school education or less. Ultimately, the enrolled study population over-represented older adults, women, and non-Hispanic White people compared with the target communities.

Conclusions:

Recruitment strategies differed substantially in response rates and cost effectiveness. Word-of-mouth referrals and in-person outreach were most successful in recruiting diverse populations.

Introduction

Epidemiologic research on the community health impacts resulting from environmental exposures and contamination events relies on successful recruitment of participants for estimating statistical associations between exposures and health outcomes. Achieving enrollment goals and recruiting a representative sample from a community improves the generalizability of findings and analytical power to examine subgroup differences in health outcomes^1,2 and can reduce potential biases such as nonresponse bias.^3,4

Although there exist prior data on the effectiveness of recruiting strategies in clinical trials, relatively few studies address successful recruitment for community-based studies. Factors reported to influence participation in community-based studies include age, family history, scheduling ease, and trusting relationship between participants and study teams.^5,6 Several articles have reported on approaches to overcome barriers to recruitment in community-based studies: Blumenthal et al⁷ conducted a study in Atlanta, Georgia and found face-to-face recruiting was most effective for African American women; Mansfield et al⁸ found word-of-mouth referrals and advertisements were effective for engaging Latino populations in the southern United States, and multiple studies have reported that recruitment by trusted community members increased enrollment success.^2,4,5,9 In addition, external partnerships (eg, medical professionals, churches, and community programs such as The Women, Infants, and Children [WIC] nutritional program), and/or community advisory panels have been known to aid in participant recruitment.^5,6,9,10

We assessed recruitment rates and cost effectiveness after implementation of 4 types of recruitment strategies in a community-based study in Colorado between 2021 and 2023, as part of the Agency for Toxic Substances and Disease Registry Multi-site study (MSS).¹¹ Our goal was to quantify the relative success of each recruitment strategy, estimate the implementation cost associated with each approach, and assess how each approach contributed to identifying and enrolling a demographically representative group of participants from our target communities.

Methods

Study design and population

We used recruitment data from the Colorado Study on Community Outcomes from Per- and polyfluoroalkyl Substances (PFAS) Exposure (CO SCOPE), the name used for the Colorado site of a federally funded multi-site study examining health effects from exposure to PFAS via drinking water in communities in the United States.^11,12 Participants were recruited from the communities of Fountain, Security, Stratmoor Hills, and Widefield, hereafter referred to as the FVC (Fountain Valley Communities), which have a combined population of 74 949.¹³ Eligibility criteria for CO SCOPE included adults and children aged 4 years and older who lived in the FVC during the period of known PFAS drinking water contamination (2005–2016) or were presumed to be exposed in utero via maternal residence in these communities.¹⁴

The enrollment goal was 1000 adults and 300 children over a 26-month (July 2021 to September 2023) recruiting period.¹⁴ Individuals were ineligible if they had worked as a firefighter (including as part of military service) or in PFAS manufacturing facilities. Multiple members could enroll from the same household. Enrolled participants received $75 honorarium after completion of 2 questionnaires (∼45 minutes) and a clinic visit that included collection of urine, blood, and anthropometric measurements (∼30 minutes). Clinic visits were offered at a wide range of times to minimize potential work conflicts, with options available before standard work hours and on Saturdays. Recruitment material included residential eligibility criteria. In this analysis, recruitment of children was not included, as child participation depended on the permission of an adult parent or guardian.

Recruitment strategies used

We defined passive recruitment as activities that required the potential enrollee to initiate contact with the study team.^2,15 Passive recruitment methods included local newspaper articles and announcements, bi-lingual flyers posted at locales where people gathered, flyers included in water bills, postcards sent by US mail, door hangers placed during door-to-door canvassing, social media announcements/posts (paid announcements via Facebook¹⁶ and Instagram¹⁷ targeting adults in the FVC, and shareable posts via CO SCOPE's Facebook, Instagram, and Nextdoor¹⁸ accounts), paid announcements on YouTube,¹⁹ study website, school e-newsletters (sent via Peachjar²⁰ to each parent/guardian's email in participating schools at the primary and secondary level in the FVC), and word-of-mouth referrals from enrollees, communities, or CAP members.

We defined active recruitment as activities in which a study team member initiated contact^2,15 with either (1) previous local study enrollees by phone or email (the PFAS Assessment of Water and Residential Exposure [PFAS-AWARE] study conducted in 2018 with residential and non-smoking eligibility requirements²¹) or (2) potential new enrollees by door-to-door canvassing or events in the study catchment. Canvassing efforts began in neighborhoods across the entire FVC. As we neared the end of our recruitment period, we prioritized neighborhoods with fewer obstacles to staff (ie, canines, gates, or fences present) and those that had been most responsive to canvassing (ie, in-person or doorbell camera conversation about CO SCOPE). Events included tabling at local community sites including farmers markets, public libraries, a CO SCOPE–hosted movie event, and outside of prominent local businesses (ie, grocery and hardware stores). These active recruitment efforts are hereafter called “in-person outreach.”

Definitions of recruitment metrics

Metrics included reach (defined as the population encountering the recruitment strategy), response rate for recruited and enrolled adults (defined as the population completing the eligibility screening questionnaire (recruited) or all study activities (enrolled) divided by the reach), enrollment efficiency (defined as participants who completed enrollment as a proportion of recruited adults²²), and estimated implementation cost per enrolled participant. Demographic information was only collected from enrolled participants. Reach was not measured for flyers and word-of-mouth because of our inability to measure total views or conversations that would constitute a denominator.

Recruited adults were asked how they heard about the study during the eligibility screening questionnaire. Responses were coded by 2 study team members working independently and any discrepancies were resolved through discussion. If a respondent reported more than one strategy was effective, we selected the strategy that was implemented closest in time to the date of that participant's eligibility screening questionnaire.

A unique household was defined as a residence with one enrolled participant, and nonunique households included residences with 2 or more enrolled participants. “Enrolled with children” was defined as an adult participant who disclosed they were a parent or guardian to an enrolled child participant <18 years of age.

Data analysis

We calculated the number and percentage of enrolled adult participants who were recruited by each strategy. We examined the demographic characteristics of participants enrolled by each strategy to compare the successes of different strategies in reaching diverse populations and performed χ² tests to identify statistically significant differences in participant characteristics between the strategies. In addition, we compared the characteristics of the enrolled population with the age, race, and sex distribution of the FVC and of the United States in 2022 (the mid-point of our recruitment period) to understand generalizability of study enrollment. We calculated the reach, response rate, and enrollment efficiency for each recruitment strategy. Finally, we calculated the costs incurred in implementing each recruitment strategy. For passive recruiting strategies, these expenses included printing, emailing/mailing costs, and fees for posting/publishing announcements and the associated personnel time for coordination and distribution of study materials. For active recruiting, we estimated the number of study team hours spent contacting individuals who had participated in prior studies via phone and/or email, door-to-door canvassing, or tabling at events. We then multiplied staff hours by an average wage of $25/hour. When contacting prior study enrollees, staff were estimated to make 15 calls/h. Staff time for in-person outreach was estimated based on the duration of scheduled time in the field for 2 staff members.

Data management and statistical analyses were performed using SAS version 9.4.¹¹ Cost calculations were conducted in Microsoft Excel.²³

Results

Recruitment rates and enrollment success

A total of 916 adults enrolled in CO SCOPE, which was 91.6% of the adult enrollment goal. The overall response rate across all recruitment strategies was 0.5%. Figure 1 illustrates passive (n = 2) and active (n = 2) recruitment strategies used to recruit (n = 1332) and enroll (n = 916) adult participants. The most common reasons for potential participant loss between initial recruitment and study enrollment were a stated lack of interest or we were unable to recontact the individual (n = 295).

FIGURE 1.

Flow of Adult Enrollment Through Passive and Active Recruitment Strategies. *Recruited Adults Completed the Study's Eligibility Screening Questionnaire. †Enrolled Participants Completed the Eligibility Screening Questionnaire, Study Questionnaires, and In-Person Clinic Visit. Recruited Adults Who Did Not Enroll Consisted of a Total of 416 Recruited Adult With Reasons of Ineligibility (Announcements n = 15, Word-of-Mouth Referrals n = 7, Previous Enrollees n = 4, In-Person Outreach n = 11), No Longer Interested/Unable to Contact (Announcements n = 94, Word-of-Mouth Referrals n = 66, Previous Enrollees n = 13, In-Person Outreach n = 122), Did Not Have Time to Participate in the Study (Announcements n = 3, Word-of-Mouth Referrals n = 11, Previous Enrollees n = 2, In-Person Outreach n = 20), Did Not Complete Questionnaire (Announcements n = 2, Word-of-Mouth Referrals n = 5), Did Not Complete Blood Draw (Announcements n = 3, Word-of-Mouth Referrals n = 8, Previous Enrollees n = 1, In-Person Outreach n = 13), or Other (Announcements n = 5, Word-of-Mouth Referrals n = 4, Previous Enrollees n = 1, In-Person Outreach n = 6).

Of the 1332 recruited adults, the majority (58%) were recruited via passive strategies. Similarly, among those who enrolled in the study, 60% were recruited by passive strategies (Table 1).

TABLE 1.

Comparison of Demographic Characteristics of Enrolled Adult Participants in CO SCOPE (2021–2023) by Recruitment Strategy

	Passive Recruitment		Active Recruitment
	Announcementsa	Word-of-Mouth Referralsb	Prior Study Enrolleesc	In-Person Outreachd	Total	P e
		n (%)
Total enrollment	288 (31.4)	261 (28.5)	75 (8.2)	292 (31.9)	916 (100)
Age group						<.01
18–34	20 (6.9)	55 (21.1)	1 (1.3)	32 (11.0)	108 (11.8)
35–44	39 (13.5)	41 (15.7)	8 (10.7)	46 (15.8)	134 (14.6)
45–64	109 (37.9)	100 (38.3)	36 (48.0)	107 (36.6)	352 (38.4)
≥65	120 (41.7)	65 (24.9)	30 (40.0)	107 (36.6)	322 (35.2)
Sex						.61
Male	104 (36.1)	100 (38.3)	23 (30.7)	112 (38.4)	339 (37.0)
Race						.58
White	237 (82.3)	208 (79.7)	67 (89.3)	240 (82.2)	752 (82.1)
Otherf	43 (14.9)	42 (16.1)	8 (8.0)	52 (13.7)	131 (14.3)
Unknown	8 (2.78)	11 (4.2)	2 (2.7)	12 (4.1)	33 (3.6)
Hispanic/Latino						.65
Yes	32 (11.1)	38 (14.6)	10 (13.3)	35 (12.0)	115 (12.6)
Unique householdsg						<.01
Yes	240 (83.3)	124 (47.5)	62 (82.7)	252 (86.3)	678 (74.0)
Enrolled with childrenh						.03
Yes	41 (14.2)	40 (15.3)	6 (8.0)	52 (17.8)	139 (15.2)
Education level						.04
HS graduate/GED or less	49 (17.0)	71 (27.2)	13 (17.3)	70 (24.0)	203 (22.2)
Some university/college	105 (36.5)	66 (25.3)	23 (30.7)	81 (27.7)	275 (30.0)
Technical/trade school	21 (7.3)	15 (5.8)	2 (2.7)	21 (7.2)	59 (6.4)
≥University/college	113 (39.2)	109 (41.8)	37 (49.3)	120 (41.1)	379 (41.4)
Household income						.65
<$25K	39 (13.5)	40 (15.3)	6 (8.0)	43 (14.7)	128 (14.0)
$25–69 999	127 (44.1)	108 (41.4)	36 (48.0)	122 (41.8)	393 (42.9)
$70–149 999	100 (34.7)	90 (34.5)	22 (29.3)	103 (35.3)	315 (34.4)
≥$150 000	14 (4.9)	18 (6.9)	7 (9.3)	14 (4.8)	53 (5.8)
Missing/unknown	8 (2.8)	5 (1.9)	4 (5.3)	10 (3.4)	27 (2.9)

Abbreviation: CO SCOPE, Colorado Study on Community Outcomes from Per- and polyfluoroalkyl Substances (PFAS) exposure.

^aStudy information sent to or seen by community members through traditional and electronic mailings, newspaper announcements, flyers, or internet via social media, YouTube, and our study website.

^bEnrolled after a recommendation from another CO SCOPE participant or community member.

^cPeople involved in previous health research studies and who indicated interest in participating in future studies.

^dEnrolled after contact with study team members door-to-door canvassing or tabling at local events.

^eP-values < .05 were considered significant.

^fRace for “Other” includes Black or African American, American Indian or Alaska Native, Asian, and Native Hawaiian or Other Pacific Islander.

^gIf unique household = no, then this participant enrolled with at least one other adult from the same household.

^hIf enrolled with children = yes, then this participant also consented for at least one of their children to participate.

Success of recruitment strategies in enrolling diverse populations

We noted significant differences in enrolled participant characteristics across recruitment strategies for age groups (P < .01), unique households (P < .01), enrolled with children (P = .03), and education level (P = .04). Announcements were more effective in enrolling older adults (aged 45+) and less effective in enrolling people with a high school educational attainment or less (Table 1). Word-of-mouth referrals were the most successful strategy for enrolling younger adults (aged 18–34) and people with lower educational attainment. Recontacting prior study enrollees was also highly effective in enrolling older adults (aged 45+). In-person outreach was most effective in bringing in unique households and participants who enrolled with children.

Compared with the demographic characteristics of the FVC and US populations during the same period, CO SCOPE's enrolled population over-represented people who are aged 65 and older, White, and female (Table 2). People of Hispanic or Latino ethnicity were notably under-represented (12.6% in this study compared with 26.6% and 19.1% in the FVC and US population, respectively), as were people who identified as multiracial (3.5% in this study compared with 14.1% and 12.5% in the FVC and US population, respectively).

TABLE 2.

Comparison of Demographic Characteristics of CO SCOPE Adult Participants, Fountain Valley Community, and US Population

	CO SCOPE (2021–2023) n = 916 n (%)	Fountain Valley Communitiesa (2022) n = 54 807 n (%)b	United States (2022) n = 260 961 960 n (%)b
Age group
18–64	592 (64.6)	46 860 (85.5)	203 029 405 (77.8)
≥65	324 (35.4)	7947 (14.5)	57 933 555 (22.2)
Sex
Female	577 (63.0)	27 568 (50.3)	131 524 828 (50.4)
Male	339 (37.0)	27 239 (49.7)	129 437 132 (49.6)
Self-reported race
White	752 (82.1)	38 255 (69.8)	158 925 834 (60.9)
Black or African American	52 (5.7)	4330 (7.9)	31 837 359 (12.2)
American Indian or Alaskan Native	14 (1.5)	493 (0.9)	2 609 620 (1.0)
Asian	24 (2.6)	1261 (2.3)	15 396 756 (5.9)
Native Hawaiian or Other Pacific Islander	9 (1.0)	329 (0.6)	521 924 (0.2)
Multiracial	32 (3.5)	7728 (14.1)	32 620 245 (12.5)
Unknown	33 (3.6)	—	—
Self-reported Hispanic/Latino
Yes	115 (12.6)	14 579 (26.6)	49 843 734 (19.1)

Abbreviation: CO SCOPE, Colorado Study on Community Outcomes from Per- and polyfluoroalkyl Substances (PFAS) exposure.

^aFountain Valley Communities include residents of Fountain, Security, Stratmoor Hills, and Widefield, Colorado.

^bNumbers and percentages are based on adult populations only (FVC: 54 807; US: 260 961 960). Total populations of adults and children for Fountain Valley Communities is 74 959, and the United States is 333 287 562 (United States Census Bureau, 2024).

Cost effectiveness of different strategies

Tabling was among the costliest of our recruitment strategies, with an estimated implementation cost per enrolled participant of $292, more than 6 times the estimated per-enrolled participant costs of distributing flyers (Table 3). Announcements had a large population reach among FVC residents but a low response rate, and a relatively modest implementation cost estimate of $110 per enrolled participant, on average. Among the types of announcements used, study postcards produced the highest response rate and largest number of enrolled participants, whereas social media, public school e-newsletters, and door hangers had large population reaches but brought in low response rates and consequently low numbers of enrolled participants. Multiple cycles of recruitment efforts were implemented using flyers included in water bills (n = 3), study postcards (n = 2), and public school e-newsletters (n = 5) with trends of lower participant response rates after the initial cycle. Word-of-mouth referrals brought in 27.2% of recruited adults and 28.5% of enrolled adults and had a high enrollment efficiency at no implementation cost to the study. The response rate for word-of-mouth referrals is not calculated because the denominator (population reached) is unknown. The strategy of recontacting prior study enrollees had a finite population reach but yielded a high response rate for both recruited and enrolled adults (41.4% and 32.3%, respectively) and high enrollment efficiency at a very low implementation cost. In-person outreach had a modest response rate (2.2%) and a fairly high implementation cost of $242 per enrolled participant. However, we devoted much of our effort to in-person outreach, and these strategies brought in 34.8% of recruited adults and 31.9% of enrolled adults. Comparing the 2 in-person outreach strategies of canvassing vs tabling, canvassing brought in a greater total yield of recruited and enrolled participants while tabling had a somewhat higher enrollment efficiency. We spent a greater overall amount of money and time on canvassing efforts than we did on any other single recruitment strategy (Table 3).

TABLE 3.

Comparison of the Estimated Population Reached, Response Rates, and Implementation Cost by Recruitment Strategy

	Population Reacheda n	Recruited Adult Response n (Rate %)b	Enrolled Participant Response n (Rate %)c	Enrollment Efficiency %d	Estimated Implementation Cost $	Estimated Implementation Cost per Enrolled Participant $
Announcements	231 708	410 (0.2)	288 (0.1)	70.2	31 723	110
Flyers	—	40 (−)	33 (−)	82.5	1560	47
Door hangere	13 750	27 (0.2)	20 (0.1)	74.1	5241	262
Internet/social media	∼162 800	23 (0.1)	14 (0.1)	60.9	3441	246
Public school e-newsletterf	33 402	25 (<0.1)	17 (<0.1)	68.0	4475	263
Newspaper adsg	∼2500	33 (1.3)	22 (0.9)	66.7	3877	176
Study postcard	11 856	173 (1.5)	114 (1.0)	65.9	9334	82
Water bill	7400	89 (1.2)	68 (0.9)	76.4	3795	56
Word-of-mouth referrals	–	362 (−)	261 (−)	72.3	0	0
Prior study enrolleesh	232	96 (41.4)	75 (32.3)	78.1	1160	15
In-person outreach	20 991	464 (2.2)	292 (1.4)	63.1	70 785	242
Canvassingi	15 178	323 (2.1)	195 (1.3)	60.4	41 263	212
Tabling/eventj	5813	141 (2.4)	97 (1.7)	68.8	28 363	292

^aFlyers and word-of-mouth referrals do not have a population reached because of the inability to measure total views or conversations. Internet/social media population reached was measured from how many people saw the ad. Newspaper ad reach was calculated based on how many people received the newspaper. Population reach for study postcard and water bill was determined by the number of mailers sent on the first contact attempt.

^bRecruited response rate percentage = (recruited adults/population reached) × 100. Recruited adults completed the eligibility screening questionnaire. Response rates were cumulative from all contact attempts (eg, study postcards were sent twice, and flyers were sent in water bills 3 times).

^cEnrolled response rate percentage = (enrolled adults/population reached) × 100. Enrolled adults completing all the study activities.

^dEnrollment efficiency percent = (enrolled adults/recruited adults) × 100.

^eCost from printing. Does not include staff time to distribute because of distribution occurring during canvassing efforts when there was no answer at the resident's door.

^fNumber of local schools participating in the parent e-newsletter program varied across the recruitment period.

^gTwo articles and 13 weekly ads run in the local newspaper.

^hCost determined by study team hours for recruitment outbound calls or emails at $25/h and completing an average of 15 calls/h.

ⁱCanvassing cost determined by hours spent in the field ($25/h per person in teams of 2, $39 900) and flyers to hand out (36% of total flyers printed, $1362).

^jTabling/event cost determined by event registration fees ($470), movie license fee ($930), and printing flyers to hand out (36% of total flyers printed, $1362), in addition to hours spent in the field ($25/h per person with 2 people in the field at once, $25 600).

Details of canvassing efforts

In-person outreach required substantial hands-on effort and implementation cost for participant recruitment. Because of safety and time limitations, we canvassed approximately 15 000 households (58.7% of total estimated households) in the FVC. In 22.9% of the households canvassed, members of the household conversed (in-person or doorbell camera) about CO SCOPE with the study team (Figure 2). However, adults agreed to be contacted for study participation in only 2.4% of total households canvassed.

FIGURE 2.

Flow of Household Study Participation Interest from Door-to-Door Canvassing Efforts. *Fountain Valley Communities Include Fountain, Security, Stratmoor Hills, and Widefield, Colorado. ^†Household Conversed Included Conversations With the Study Team via In-Person or Doorbell Camera About CO SCOPE (Colorado Study on Community Outcomes from Per- and Polyfluoroalkyl Substances [PFAS] Exposure). A Door That Was Open and Immediately Closed Was Not Considered as a Household That Conversed With the Study Team. ^‡Percentages Are Based on the Number of Households Canvassed.

Discussion

We achieved over 90% of our adult enrollment goal by using both active and passive recruiting strategies for enrolling participants in a large community-based environmental health study. Passive recruitment strategies, including various announcement types, resulted in extensive participant reach but low response rates, consistent with previous reports.^24,25 Nonetheless, passive methods were cost- and time-efficient per participant enrolled. Word-of-mouth referrals had greater enrollment of young adults, or adults of lower educational attainment, but was less effective enrolling participants aged 65 or older. Active recruitment strategies to reengage previous study participants were most successful, in proportion, in recruiting older adults, and had higher response rates compared with other approaches (albeit word-of-mouth referrals and flyers within announcements did not have a measurable reach to calculate a response rate). In-person outreach was, on average, more costly than other strategies implemented per participant enrolled. Overall, the enrolled population over-represented participants who were older and who identified as non-Hispanic White.

Announcements, collectively, had the second lowest enrollment efficiencies after word-of-mouth referrals and prior study enrollees. However, looking more granularly within announcements, flyers and water bills had notably high enrollment efficiencies. It is possible the community is more receptive to announcements pertaining to water and PFAS because of their awareness of the historical contamination of their drinking water. Announcements via social media had limited impact on recruitment evidenced by low response rates and enrollments. Other studies support the theoretical utility of social media, such as Facebook, with targeted ads for location, age, and gender,²⁶ but in practice, social media advertisements, as reported by other investigators, have often shown low response rates for eligible participants.^26–28 We observed evidence of saturation of potential participants evidenced by diminishing response rates with each subsequent contact attempt using postcard mailings, flyers in water bills, emails to parents at schools, and newspaper announcements. For future studies, we recognize that the utility of newspaper advertisements will likely decline because of downsizing and closures of local newspapers.²⁹ For example, the local newspaper in our study catchment closed during our recruitment period.

Word-of-mouth referrals were the most effective strategy in recruiting younger adults and those with lower educational attainment. Word-of-mouth referrals had almost no implementation cost and the second highest enrollment efficiency after recontacting prior study enrollees. Prior literature also supports our observation that use of personal networks, including word-of-mouth referrals and contacts by community organizations, may facilitate the recruitment of participants with lower educational attainment.³⁰ Our partnership with a Community Advisory Panel, composed of a diverse panel of engaged community members including local governmental officials, business owners, school employees and other residents, likely facilitated early onset of word-of-mouth referrals for our study. The long recruitment period (26 months) allowed ample opportunity for word-of-mouth referrals to spread through personal networks as participants shared their experiences within social networks.

Active recruitment involving prior study enrollees was a unique recruitment opportunity for CO SCOPE as this community has been involved in prior health-related research studies. Contacting prior study enrollees provided a recruitment strategy that was efficient, inexpensive, and often successful. This group of individuals were knowledgeable regarding exposure, and many were familiar with CO SCOPE's principal investigators, likely increasing trust and confidence. This unusual opportunity is finite and rarely available. In addition, participants involved in multiple research studies may not be representative of the larger population. Participants in CO SCOPE who had enrolled in previous PFAS-related studies in the area were more likely to be older adults and people of higher educational attainment.²¹ These characteristics mirror enrollment trends in other studies; other investigators have noted that individuals with greater educational attainment showed greater willingness to participate in studies^7,9 and willingness to return questionnaires via mail^3,31 compared with their counterparts. In a review article on child participation in research studies, which can be extrapolated to family dynamics within a community, Robinson et al³² observed that higher parental educational attainment likely influences work schedule control and flexibility, increased interest and comprehension of study goals, and belief in the altruistic value of research. Studies of recruitment can help to identify populations for which specific, and perhaps more expensive, strategies are required.

Strengths

This study was able to assess various recruitment strategies because of the relatively high recruitment target, a budget that allowed for implementation of a range of recruitment strategies, and a 2-year duration of recruitment. These features allowed the study team to implement multiple methods providing FVC residents ample opportunities to encounter at least one form of study recruitment. Moreover, using multiple recruitment strategies within the same catchment area allowed us to compare the effectiveness of these strategies in specific subgroups within the target population.

Limitations

We used multiple recruitment strategies, at times simultaneously, so participants may have had multiple contact points with the study. This could have decreased our ability to accurately estimate the efficacy of individual strategies. Prior studies have found that multiple contacts using different methods of recruitment can lead to increased enrollment^5,24 and diversity.³⁰ We did not prospectively record to which recruitment strategies each participant was exposed but rather relied on participant's recall of recruitment approaches encountered. Other studies may wish to prospectively ask questions that more precisely delineate what recruitment method or methods were the ultimate motivational and deciding factors for participation.

Overall cost estimates were based on hours worked in the field, printed materials, and postage costs. Our estimates did not include CO SCOPE project development, outreach equipment, rental office space, or the salary of 1.4 full time effort (FTE) staff devoted to recruitment coordination and project management. These costs appear to drastically differ between studies based on study personnel expertise, local salaries, availability of software to manage in-person outreach efforts, information needed for postcards or canvassing endeavors (ie, residential addresses), establishment of local networks for tabling event opportunities, and other variables unique to a study and locale.

Finally, demographic characteristics were only captured from enrolled adults who were willing to share this information. There were 416 recruited adults who did not ultimately enroll because of lack of interest or our inability to recontact them. It remains unclear whether a study with differing participant requirements (eg, no clinic visit, blood draw, or a smaller time commitment) could have increased interest in study participation and, therefore, increase enrollment efficiency, allowing for a more complete understanding of the success of recruitment strategies with various demographic groups.

Conclusions

Community-based studies rely on successful representation of the target population and a sufficient number of enrolled participants for statistical power, validity, and generalizability of research findings.² Here, we provide an example and comparison of multiple recruitment strategies implemented in a Colorado-based study, highlighting the role of both active and passive recruitment strategies to reach enrollment goals. Our results may enable future studies to develop tailored recruitment plans that balance priorities, including cost effectiveness and representativeness, of the enrolled sample.

Implications for Policy & Practice

• ■ Our study demonstrates the utility of deploying multiple recruitment strategies to better achieve a diverse and generalizable population and overcome low recruitment response rates.

• ■ Recruitment strategies that allowed in-person conversation (ie, word-of-mouth referrals and in-person outreach) were most effective in recruiting young adults and adults of lower educational attainment with the caveat of in-person outreach costing the most per enrolled participant.

• ■ Results may inform future researchers as they plan community-based recruitment.