ERIC EJ917440: Recruitment of African American and Latino Adolescent Couples in Romantic Relationships: Lessons Learned

2011·0.15 MB·English

by  ERIC

Research Article
Recruitment of African American and latino Adolescent
Couples in Romantic Relationships: lessons learned
Angelic Rivera, Dana Watnick, and Laurie J. Bauman
AbstrAct
Background: There is considerable literature on effective engagement strategies for recruiting adolescents individually
for health research studies, but literature on recruiting adolescent couples is new and minimal. Purpose: This paper
describes the recruitment strategies used for Teen Connections (TC), a longitudinal study that recruited 139 mainly
African American and Latino adolescent couples in romantic relationships living in New York City. Methods: We
collected data in Microsoft Access and documented the date each recruitment strategy was implemented, date each
partner was enrolled, and amount of effort required to enroll participants. We identified individual and relationship
characteristics from each partner’s baseline survey. Results: We found that relationship type and characteristics, lan-
guage used in printed materials, parental consent, implementing a screener questionnaire and gender of partner had
implications for enrollment in TC. Discussion: Couples studies are highly demanding but achievable with dedicated
staff and access to a large number of youth. Translation to Health Education Practice: Research on sexual health
and risk often relies on individual reports of dyadic events. Adolescent couples’ studies may not be pursued because of
recruitment limitations, but they can provide invaluable insight into relationship dynamics, characteristics, etc. that
may help design better health education interventions, and should be pursued nonetheless.
Rivera A, Watnick D, Bauman LJ. Recruitment of African American and Latino adolescent couples in romantic relationships: lessons learned.
Am J Health Educ. 2011;42(1):30-40. This paper was submitted to the Journal on March 31, 2010, revised and accepted for publication on
August 22, 2010.
BACKgRoUND tant predictors of condom use among teens, upon which others can build.
Young people of color in America are but these have been neglected. We began There is considerable literature on ef-
disproportionately affected by human a five-year study of teenage couples called fective engagement strategies for recruiting
immunodeficiency virus and other sexu- Teen Connections (TC), which required adolescents individually for health research
ally transmitted infections.1-3 An extensive developing recruitment procedures that studies,8-13 but research on adolescent
body of research has documented sources were effective, efficient and ethically sound. couples is very new. To identify papers de-
of risk, patterns of risk behavior, predictors In this paper, we report our recruitment scribing recruitment of both members of
of condom use and persistence of unsafe experiences to provide a base of information adolescent couples, we searched PsychInfo,
sexual behavior.1, 4-7 This literature is based
predominantly on individual-level models
of behavior, and findings show that condom Angelic Rivera is a project director in the De- 1300 Morris Park Avenue VE6B23, Bronx, New
outcome and efficacy expectancies, inten- partment of Pediatrics, Albert Einstein College York 10461. Laurie J. Bauman is a professor of
tions to use condoms and perceptions of risk of Medicine/PIRC, 1300 Morris Park Avenue Pediatrics and director of the Preventive Inter-
are related to safer sex behavior. 5-7 We believe VE6B4, Bronx NY 10461; E-mail: Angelic. vention Research Center in the Department of
that couple-level factors such as love, trust, [email protected]. Dana Watnick is a Pediatrics, Albert Einstein College of Medicine/
future commitment, expectation of mo- project director in the Department of Pediat- PIRC, 1300 Morris Park Avenue VE6B25, Bronx,
nogamy and actual monogamy are impor- rics, Albert Einstein College of Medicine/PIRC, New York 10461.
30 American Journal of Health Education — January/February 2011, Volume 42, No. 1
Angelic Rivera, Dana Watnick, and Laurie J. Bauman
Medline, and Eric using key words like ado- Hospital in the bronx, New York. The sec- one-on-one qualitative interviews four to
lescents, teenagers, couples and recruitment ond step was to use responses from TLS to six months apart whether they remained
strategies. We did identify a few publications identify teens who met the eligibility for TC: together or not.
that addressed couple recruitment of young live in the bronx, NY; be in a heterosexual
Recruitment of Adolescents
adults.14-17 These studies provided helpful relationship for four or more weeks; have to the TLS Screener
insights into recruitment of older youth, engaged in vaginal, oral, or anal sex with
We recruited adolescents between the
but they did not address the special issues that partner in the past six months; read at
ages of 14-17 from a computerized database
involved in recruiting teens under age 18, least at a fifth grade-reading level, and be
used for medical record keeping and billing
such as the need for parental permission for able to read, write and speak English fluently.
at Montefiore Medical Center. Parents of
study participation; ethical concerns over Reading level was measured with the read-
youth identified through the database first
recruitment of adolescents named by teen- ing subtest of the Wide-Range Achievement received a letter in English and Spanish from
agers as their sexual partner; the frequency Test-3 (WRAT-3),19 administered by clinical the clinic where their child receives health
of being in a private relationship (family is interviewers when participants came to the care. The letter informed parents about the
unaware of the romantic relationship); the Center. At the completion of the reading TLS screener and of their child’s potential
inadvertent meaning that being invited into subtest, a raw score was calculated and con- eligibility for participation in a further study.
a couple study imposes on the couple; and verted to a grade score using the age specific The letter told parents that, if they agree to
the limited ability to schedule participation norm tables of the WRAT-3. Teens who met the release of their child’s name, address,
in a research study around familial and other eligibility criteria were called “index” teens telephone number and date of birth, this
obligations (e.g., having to baby-sit a sib- and were invited to join TC with their part- information would be passed on to the prin-
ling, pick up a family member from school, ner. If more than one partner was eligible, cipal investigator and a TLS invitation would
household chores). This paper will describe we allowed index teens to choose the partner be mailed to their child. If parents preferred
the recruitment strategies we implemented they wanted to enroll. Less than 4% of index that their child’s contact information not
and their effects on couple enrollment. teens had more than one eligible partner that be released to the study, they were given
Retention strategies will be discussed in a met study criteria. two ways to opt-out: they could return the
separate manuscript. The Teen Connections study design com- enclosed postage paid postcard to the clinic,
bined qualitative and quantitative methods. or they could call a special number to request
pURpoSE The quantitative study was a longitudinal to opt out. Parents had three weeks from the
Teen Connections Project Description panel study of 139 adolescent couples (278 date of the clinic letter mailing to opt their
Teen Connections was designed to docu- individual partners) at baseline; adolescents child out of receiving a TLS invitation. Nei-
ment how different types, characteristics ranged in age from 14 to 20. Participants ther the parent nor teen at this point had to
and duration of sexual relationships were completed quantitative surveys every 3 make a decision about study participation.
related to condom use in early to middle months for 12 months regardless of whether The opt-out letter only requested permis-
adolescence. Relationship type was defined they remained a couple over time, a total of sion from the parent to allow the study to
based on previous qualitative research as 5 time points. We chose to follow up every send their child a TLS invitation packet that
“committed/serious,” “dating/seeing,” “new,” three months to measure changes in rela- provided more information about the study
or “not serious.”18 Relationship characteris- tionships, particularly break-ups. Data from and the project consent form.
tics we focused on were love, trust, future each partner were collected independently The database identified 7,593 adolescent
commitment, expectation of monogamy and confidentially. For the TC baseline patients that met age criteria, and clinics
and actual monogamy. survey both partners were encouraged to mailed their parents the study opt-out letter.
We engaged in a two-step recruitment come in to our Center together to assure Of the 7,593 letters mailed (Figure 1), 5,368
process for Teen Connections. The first step concurrency of baseline data collection, but (71%) were returned because the family had
was to recruit youth into a screening study they could come in separately as long as the moved and no forwarding address or phone
called the Teen Lifestyle Survey (TLS) a surveys were conducted within one week of number was provided and 151 (7%) parents
confidential, self-administered computer- each other. declined permission to have their child’s
assisted questionnaire. TLS is the single The TC qualitative study was a subgroup contact information shared with the study.
point of entry that we use at our Center to of 15 couples enrolled to participate in Neither of these household types was con-
recruit teens into several ongoing studies. in-depth individual (separate) interviews tacted again. Excluding households whose
We recruited teens aged 14-17 for the TLS (mean length 75 minutes) about their letters were returned and parental opt-outs,
who had received any medical care in the relationship. These couples were selected 2,074 parents gave passive consent for their
last 18 months from any of seven general based on the length and seriousness of child to receive a TLS study invitation.
pediatric clinics affiliated with Montefiore their relationship, and they completed four The mailed invitation explained that
American Journal of Health Education — January/February 2011, Volume 42, No. 1 31
Angelic Rivera, Dana Watnick, and Laurie J. Bauman
Figure 1. Adolescents identified by Computerized Clinic Database
Teens identified by
Computerized Database
N = 7,593
Parental permission granted Parental study opt-outs Study opt-out letters returned
N= 2,074 N=151 and no forwarding address or
contact phone number available
N= 5,368
Teens sent TLS invitation
and households contact No further No further
about the study contact by contact by
study study
Teens that completed TLS study refusals
TLS Screener N=767 (37%)
N= 1,307 (63%)
No further
contact by
study
youth were being invited to participate in commodate teens who attended school or child was not allowed to participate due to
the TLS, a confidential, computerized survey worked. The project consent form had to be being punished for misbehavior or poor
that takes about 90 minutes to complete and signed by both the teen and the teen’s par- grade report.
asks questions about school, peers, family, ent; parents were not required to accompany Enrollment of Couples into Teen
sexual behavior practices, and drug and teens completing their survey. Data collec- Connections
alcohol use. Printed materials also informed tion was conducted at our Center which The TLS screening questionnaire was
parents and teens that after completion of is easily accessible by mass transportation. completed by 1,307 adolescents (Figure 2),
the TLS, teens might be eligible for other Adolescents who completed the TLS were 62% (N=811) females and 38% (N=496)
projects. Five business days after the TLS given a $25 monetary honorarium. males, mean age 15.9 years. After teens com-
invitation was mailed, study staff called teens Of the 2,074 youth who were mailed a pleted the TLS, the project recruiter reviewed
to verify that they had received the printed study invitation and contacted via phone, the quantitative data and identified 373
materials and to answer any questions they 1,307 (63%) teens completed the TLS. Of index teens who met study criteria for Teen
or their parents may have had. Adolescents the 767 study refusals, 549 parents and/or Connections: currently in a heterosexual
were given the option to come in any day teens gave scheduling conflicts as the reason relationship for four or more weeks, and
during a six-week period to complete TLS. (worked, school activities or familial obliga- had sex with that partner (vaginal, oral, or
Surveys were conducted during after school tions such as babysitting a younger sibling anal) within the past six months. The proj-
hours, Saturdays, or by appointment to ac- or child) and 218 parents reported that their ect recruiter met with eligible index teens
32 American Journal of Health Education — January/February 2011, Volume 42, No. 1
Angelic Rivera, Dana Watnick, and Laurie J. Bauman
immediately following TLS administration lation of the New York State Office of referred to a project called Teen Connec-
to explain the TC study and to make a pre- Children and Family Services, nor did we tions, a research study interested in learning
liminary determination of the eligibility of have any legal responsibility to report it to about different types of teen relationships,
the index teens’ partner. Partner eligibility authorities because it was illegal. However, and that the adolescent had been referred
criteria were age no more than four years we decided not to enroll couples who met by another teen to participate. The project
older than the index teen, lives in New York criteria for statutory rape, and we collected recruiter encouraged index teens to invite
City, at least a 5th grade reading level, and able data on exclusion criteria to minimize the the person he or she was referring into TC
to read, write, and speak English. likelihood that we would learn of it by not via telephone from the study office. Index
Reading level was measured with the collecting data on the age of partners on the partners were told that if their referral
reading subtest of the WRAT-3, 19 adminis- TLS screener, and only asking during in- partner consented to talking with study
tered by clinical interviewers when referral person recruitment if the partner he or she staff, the project recruiter would introduce
partners came to the Center. At the comple- was referring was less than four years older. herself, answer any of his/her study related
tion of the reading subtest, a raw score was Index teens with partner(s) who were four or questions, and request permission to send
calculated and converted to a grade score more years older were excluded from being printed materials to the household. Printed
using the age specific norm tables of the invited to enroll in Teen Connections. materials mailed to the partner’s home
WRAT-3. Of the 373 partners referred by because their referral partner was in- informed the partner that data will be col-
eligible index teens, 169 referral partners eligible, we excluded 169 index teens. The lected independently and confidentially at
were ineligible during the in-person recruit- remaining 204 index teens were told about our office, and that both teens will remain
ment: 90 (53%) were ineligible due to age, Teen Connections and our interest in learn- in the follow-up study even if they can no
69 (41%) lived outside of the five boroughs ing more about different types of adolescent longer come in together. Couple partners
of New York City and 10 (6%) self-reported relationships. Teens were also informed that were asked to come in within one week of
to the study recruiter that they were Spanish- in order to participate in TC, their referral each other to assure that the data were col-
only speakers and were unable to read, write, partner would have to agree to participate lected concurrently. Each partner was given
or speak English fluently. as well. TC initially was approved by the a $30 monetary honorarium for completing
Over half of the referral partners were Einstein Institutional Review board (IRb) the TC baseline survey. Of the 204 couples
deemed ineligible due to age. Partner age to recruit adolescents without requiring that were eligible to participate in the study
criteria were imposed based on the goal of parental consent. We believed that requir- as indicated by the TLS screener and after
the study (to study relationships in early and ing parental consent would increase study the in-person recruitment session, 18%
middle adolescence) and ethical consider- refusals by couples who were in private (N=37) dissolved their relationship before
ations. When enrolling youth in a study of relationships, which would increase the risk completing their TC baseline survey and
sexual relationships the issue of statutory of sample bias, with teens in public relation- were no longer eligible for the study. Of
rape must be addressed. In New York State ships more likely to participate. However, the remaining 167 eligible couples, 139
couples (83%) were enrolled, 26 couples
(NYS), statutory rape is defined as sexual this approach introduced different barri-
abuse with the degree of abuse (e.g., 1st, ers to recruitment and we changed study (16%) refused study participation, and 2
index teens (1%) reported that the referral
2nd, or 3rd degree) determined by the age of requirements to require parental consent.
partner was too ill to participate due to
the victim and the defendant. In NYS, an The rationale for changing from a waiver to
long-term hospitalization.
individual is deemed incapable of sexual a parental consent process will be discussed
consent when he or she is less than 17 years later in the manuscript.
METHoDS
of age.20 However, individuals are exempt When enrolling couples we used a
from prosecution for rape or criminal sexual brokering plus invitation strategy coupled Participants
acts under the following circumstances: (1) with a co-recruitment approach similar Of the 139 enrolled couples, the majority
if the victim is between 15 and 17 years of to those used with adult couples.16 Eligible of participants (73%) were partnered with
age and the partner is less than 21 years of index teens were given two information someone of the same race/ethnicity: there
age; and (2) if the victim is between 11 and packets, one for themselves and the other to were 67 Latino couples and 34 black couples.
15 years of age and the partner is less than be given to their partner. In order to avoid Of the mixed couples, 28 were black/La-
18 years of age or less than 4 years older inadvertent disclosure of the relationship, tino and 10 were Asian, white, or “other.”
than the victim.20 We consulted with police neither packet stated that TC was a study for Participant age varied by gender with boys
and professional ethicists and discussed sexually active couples in romantic relation- averaging one year older than girls (17.8 yrs,
options with the funding agency. Statutory ships. The printed materials were also sent SD=1.3 vs. 16.8 yrs, SD=1.1), respectively.
rape is not a mandated reporting offense to the household to be reviewed by parents Less than one-third of individuals (N=70)
per the child abuse or maltreatment regu- and teens. It stated that the teen was being reported living with both biological parents:
American Journal of Health Education — January/February 2011, Volume 42, No. 1 33
Angelic Rivera, Dana Watnick, and Laurie J. Bauman
Figure 2. Adolescents Who Completed TlS Screener
TLS Screened Participants
N = 1307
Screened TC Eligible from the TLS Screened Ineligible for TC
N = 373 (29%) from the TLS
N = 934 (71%)
Teen Connections Enrollment Outcome
Sexually Inexperienced
N = 528 (57%)
No relationship reported
Eligible TC Enrollees Ineligible TC Enrollees
N = 222 (24%)
N = 167 (45%) N = 206 (55%)
Enrolled Couples < 5th grade reading level
N = 139 (83%) N = 122 (13%)
Study Refusals
N = 26 (16%) In-person Recruitment *Couple Relationship with a
N = 169 Broke-up same sex partner
Long-term N = 37 N = 14 (1%)
Partner did not meet
hospitalization
age criteria (e.g., older
N = 2 (1%)
Clinical Score on YSR
partner)
N = 48 (5%)
N = 90 (53%)
NYC non-resident
N = 69 (41%)
Partner didn’t speak,
write, or read English
fluently
N = 10 (4%)
* These couples were initially eligible for study participation, however reported a break-up occurring between initial screener eligibility and study enrollment,
therefore becoming ineligible.
44% (N = 122) lived with their mother relationship: 95 out of 139 couples (68%) couples agreed that their relationship was
only; 27% (N = 65) lived with neither agreed they were in a serious/ committed public. We defined a public relationship as
biological parent. relationship; nine couples agreed that they one where “most” or “all” of the participant’s
Length of relationship was reported by were dating or seeing one another; and seven friends knew about their relationship and
boys as one and a half months longer on couples agreed they were in a relationship at least one parent also knew. When we
average than girls reported (1.3 yrs, SD = that was not serious. Eighty-five percent of examined how important the relationship
1.2 vs. 1.1 yrs, SD = 0.9). Within couples, girls and 84% of boys reported that their was to girls and boys, 91% of girls and 87%
111 (80%) agreed about the status of their relationship was public, and 76% of the of boys reported it being very or extremely
34 American Journal of Health Education — January/February 2011, Volume 42, No. 1
Angelic Rivera, Dana Watnick, and Laurie J. Bauman
important to them, and 83% of the couples pre-study focus groups and a pilot). Packets only 13% (N = 167) of teens were identified
were in agreement that it was very or ex- included informational flyers and a formal as eligible for Teen Connections. During the
tremely important. letter on Center letterhead. Passive recruit- TC enrollment phase, the recruiter averaged
ment strategies also included sending email 11 calls per eligible couple. On average, two
Data Procedures
or text messaging reminders to partners to calls were made to each partner for recruit-
We systematically studied how various
encourage them to come in to complete their ment purposes the additional calls were cou-
recruitment strategies affected enrollment
baseline interview. ple requested. We found that most couples
of adolescent couples. We collected data in
Active recruitment is defined by person- utilize the recruiter as a broker between the
Microsoft Access and documented the date
to-person contact, either by telephone or partners to schedule when they would come
each new enrollment strategy was imple-
in person. Telephone contact was most to the Center to complete the survey and to
mented. The relationship type (“serious/
common and most important and usually assist when a survey appointment had to
committed,” “dating/seeing,” “new,” or “not
involved calling participants at home or on be rescheduled. Over 1,830 calls were con-
serious”) identified by the couple from their
a cell phone to answer questions they or ducted to enroll the sample of 139 couples.
surveys, the date each partner was enrolled,
their parents had: to explain the importance These calls were made in conjunction with
and the amount of effort required to enroll
of the study, to assure the confidentiality of mailings that included information packets,
participants (mailings, calls, length of time
from youth identification to enrollment). We the data collection and to assist partners to interview reminder flyers, thank-you cards
also documented the questions and com- schedule their surveys. and address verification postcards.
ments of youth during in-person recruit- Enrollment Implications for Serious and
RESUlTS
ment sessions to identify areas of concern Non-Serious Relationships
for individual partners and couples, and to Described below are logistical field chal- We found that relationship type and
better understand which printed materi- lenges and engagement strategies that were characteristics had many implications for
als were effective and which needed to be implemented to recruit couples during the enrollment into the TC study. According to
revised to minimize misconceptions about enrollment phase of the TC study. in-person recruitment notes and participant
study requirements. Recruitment Efforts and Staffing Needs self-classification of relationship type, here
Recruitment Strategies The researchers have extensive experi- we classify couples as either “serious” or
Although relatively few adolescents met ence recruiting children and adolescents not serious. Teens in participant-defined
eligibility criteria, our tailored recruitment for several clinical trials, but enrollment of serious relationships frequently called or
strategies allowed us to successfully enroll teen couples was more challenging than an- text messaged one another, saw each other
83% of eligible couples. We used both pas- ticipated. It took 25 months to screen 1,307 several times a week, felt comfortable with
sive and active recruitment strategies with teens for the TLS and enroll 139 couples into the referral partner’s participation and knew
both index and referral partners. Study staff Teen Connections. the other’s schedule. In serious couples, part-
only contacted index and referral partners Our study protocol stipulated for TLS ners expressed less concern with issues of
who provided their respective contact infor- and TC that no more than four calls per confidentiality; the index teens were able to
mation directly to staff. Passive recruitment household should be made without response contact the referral partner more often from
is defined as activities that do not involve from the teen or parent. Recruiters, however, the office; referral partners were more likely
in-person conversation and focus mainly provided participants with as many survey to provide contact information directly to
on written materials that are clear, appeal- reminders or project inquiry call backs as study staff, and index teens were more read-
ing to the target-group and powerful. We requested. On average, participants invited ily able to provide the study packet to their
needed to rely on printed material at several to complete the TLS were contacted six partner within one or two days.
strategic points: (1) clinic letters mailed to times by telephone prior to completing Conversely, adolescents in non-serious
patient homes that requested permission the TLS data collection. On average, four relationships (friends with benefits, not
for the clinic to share adolescent patient out of the six calls were either participant- that serious or dating/seeing one another)
names and contact information with the TLS requested reminder calls or calls from staff to were more likely to spontaneously express
study team; (2) the TLS invitation packet reschedule a missed survey appointment. We concern over data confidentiality and re-
that was mailed to index teens; and (3) the made a total of 7,854 calls to youth, and an ferral partner participation. These index
Teen Connections study invitation packet additional 3,413 calls to households that had youth often acknowledged to the recruiter
used to enroll the partner. Written materi- a no answer, wrong or disconnected number that they felt it was unlikely that the referral
als always prominently displayed the study for a total of 11, 267 calls conducted during partner would participate since they rarely
logo, were colorful and nonthreatening and the TLS screener phase of the study. engaged in planned events together. In many
were vetted by youth (teen participants in After the completion of the TLS screener, instances, these index youth were unaware
American Journal of Health Education — January/February 2011, Volume 42, No. 1 35
Angelic Rivera, Dana Watnick, and Laurie J. Bauman
of the referral teen’s contact information sexual partners in serious relationships. We see how the study was described to their
making it unlikely that the index teen could omitted the word “partner” and replaced it parents and be reassured that we did not
contact the referral partner from the study with the terms youth themselves used, i.e., disclose that they were in a sexual relation-
office. Thus, the project recruiter seldom had “the person you are referring” (the referral ship. Due to IRb restrictions, the research
an opportunity to speak directly with these partner) and the “the person who referred staff was not allowed to contact the refer-
referral partners and request permission to you” (the index teen). ral partner directly, even if the index teen
send printed materials to the household. Following the implementation of the volunteered his or her contact information,
Moreover, due to the couple’s limited in- revised version of our printed materials, the unless the partner was 18 or older. The first
person interaction it was often not feasible recruiter noted a significant decrease among contact with the project had to be made by
for the index youth to provide the study teens asking if the study was trying to recruit the referral teen.
packet to an eligible partner. sexual partners in serious relationships. Of After the referral partner called in and
Omission of the Word “Partner” from All the 73 couples recruited with the revised provided his or her contact information to
Printed Materials version of the information packet, only one the study team directly, the study recruiter
When we created the first information couple questioned if the study was looking to mailed the opt-out letter to the referral
packet for Teen Connections we had two recruit adolescents in serious relationships. partner’s parents; parents had three weeks
goals: (1) to create materials that were easy to Also, no teens reported concerns or confu- from date of mailing to decide whether they
understand, and (2) to prevent inadvertent sion with the terms we used to replace the denied permission for staff to mail a study
disclosure of the teen’s sexual and romantic word “partner” in the materials. invitation packet to their teen. Parents also
relationship. The printed materials were Parental Consent: Waiver vs. Active had the opportunity to call the TC office if
written to emphasize that the research study Protocol? they had questions about the study. Neither
was interested in learning more about dif- the parent nor teen at this point had to make
Traditionally, research conducted with
ferent types of adolescent relationships from a decision about study participation. The
adolescent minors requires parental con-
both partners in a relationship. opt-out letter only requested permission
sent for participation. However, required
During recruitment efforts with this from the parent to allow the study to send
parental consent can lead to smaller samples
first version, we found that most index their child a TC invitation packet. For par-
as well as sample bias.21-25 We reasoned that
teens expressed trepidation about the word ents who did not want their child to receive
requiring parental consent for couple re-
“partner.” This apprehension was expressed a study invitation, they could opt their child
cruitment could double and compound this
equally by boys and girls, but for different out by telephone or by returning a postage
bias. Therefore, we requested and obtained
reasons. boys thought the word “partner” paid postcard. Parents who did not object to
approval from the Einstein IRb to proceed
signified a person you shared a serious, mo- their child receiving a TC study invitation
without parental consent, with the stipula-
nogamous relationship with. Male partners needed to do nothing.
tion that parents of both the index and refer-
questioned their eligibility for the study The second step of the waiver protocol,
ral teen would be notified about the study.
when they considered their relationship to which was initiated unless the referral parent
To comply, we designed a two-step consent
be new or non-serious. Moreover, males denied permission, was that each member
waiver protocol.
were concerned that their partner would of the couple was mailed a TC study invita-
misconstrue their invitation to join the Waiver Protocol tion packet and a project consent form. The
study, that is, that a partner would misper- The first step of the waiver protocol invitation defined available date and time
ceive the invitation as an exclamation that required that index teens (those deemed windows when partners could come in to
the relationship was serious. eligible for Teen Connections) give a letter our Center to complete baseline surveys.
Girls, in contrast, associated the word to the referral partner. The letter explained Teens were provided a four week window to
“partner” with “sexual partner” and were the study briefly; if the referral partner was decide whether to participate in TC. Enroll-
more concerned about inadvertent disclo- interested in learning more about the study, ment in Teen Connections was considered
sure of their sexual relationship to parents. he or she had to contact our office and pro- complete when both teens completed their
Moreover, teens in our sample rarely used the vide his/her parents’ contact information. baseline data collection.
terms partner, main partner, casual partner, Then, we used that contact information to Of the 94 couples eligible for enroll-
or new partner when referring to one other. send the referral partner’s parents a study ment using the waiver consent protocol, 70
When speaking about the other they used opt-out letter, which requested permission couples (75%) were enrolled into TC, two
terms such as “the person I am bringing to to send their child an invitation to partici- couples were unable to participate due to
TC” or “the person who told me about TC.” pate in the Teen Connections study. Index the referral partner’s long-term hospitaliza-
We revised the written materials to avoid im- and referral participants received a copy tion, and 22 couples (23%) refused study
plying to youth that we were only recruiting of the same opt-out letter so they could participation. The most common reason for
36 American Journal of Health Education — January/February 2011, Volume 42, No. 1
Angelic Rivera, Dana Watnick, and Laurie J. Bauman
refusal was the referral partner’s reluctance 69 couples (95%) enrolled in the study and and the couple enrollment was quicker (10
to call our office. four couples (5%) refused study participa- days vs. 7 days, respectively).
In an effort to further understand dif- tion. The four refusals were due to one or In addition to increased enrollment rates
ficulties with our two-step procedure, our both members in the relationship report- for couples in public relationships, we saw
project staff conducted brief phone inter- ing scheduling difficulties due to school or a similar increase in other types of couples
views with enrolled participants as well as work obligations. when we changed our recruitment protocol
refusers. Of the 70 couples enrolled with We were concerned that requiring from using a parental waiver to requiring pa-
the waiver protocol, fifty-eight (83%) index parental consent would increase study rental consent. (Table 1) According to index
partners reported that having to encourage refusals by couples who were in private partner reports, couples that reported being
their partner to call our office to provide relationships, leading to a sample bias with “in love” or in a “serious/committed rela-
their contact information as stressful due to teens in public relationships more likely to tionship” enrolled at significantly increased
the numerous reminders required of them. participate. This was not the case. For those rates when recruited using the parental con-
Over three-quarters of enrolled referral teens in private relationships, there was no sent protocol. Couples that were not in love
partners reported concern that study staff difference in enrollment rates if they were or serious/committed continued to enroll
would give unsolicited advice about their recruited via the waiver or the parental con- at similar rates regardless of recruitment
relationship and/or that having to make sent process (65% vs. 63%, respectively), protocol. Gender, length of relationship and
the first contact was inconvenient. We also and for those in public relationships there age of index participant were not associ-
asked enrolled couples for suggestions on was an increase in enrollment, from 80% ated with increase in enrollment when the
how we could simplify the recruitment to 96% (P<0.01). (Table 1) The process was recruitment protocol changed. blacks were
process. Eighty-four percent (N = 59) of reportedly less awkward and challenging the only demographic group that showed a
the waiver-enrolled couples reported that
having a process similar to a school permis-
sion slip, that a parent would sign for their Table 1. Couple Enrollment by index partner Characteristics
child to attend a field trip, would be less of
Waiver Parental
an inconvenience. Similarly, we asked the 22
Protocol Consent Protocol
index teens who refused study participation
(% Enrolled) (% Enrolled)
with the waiver protocol for suggestions on
how we could make the recruitment process In Love with Referral partner
more engaging. Of these refusals, 19 (86%) Yes* 81.70% 93.90%
index teens reported that eliminating the No 60% 77.80%
step where partners were required to call
Public relationship
our Center would make the process easier
(at least 1 parent knows)
and less threatening for their referral part-
Yes** 79.70% 95.50%
ner. based on the feedback we received, we No 64.70% 62.50%
piloted a new one-step recruitment protocol
Serious/Committed Relationship
that required parental consent.
Yes* 84.50% 96.60%
Parental Consent Protocol No 63.60% 75%
In the parental consent protocol, the
Gender
referral partner received a study invitation
Male 61.50% 87.50%
both from the index partner and by mail
Female 83.10% 94.10%
from the study team. The packet included
detailed study information and a consent Ethnicity
Latino 79.60% 90.00%
form that required a parent’s signature. This
Black* 70.00% 95.70%
protocol eliminated the need for referral
partners to call our office, eliminated the Length of Relationship 1.07 yrs (.92) 1.26 (1.15)
study opt-out letter being mailed to par-
Age of Index Partner 16.79 (.85) 16.57 (.82)
ents, and eliminated the waiting period for

passive consent from parents. It did add the
*P < 0.05; **P <0.01
requirement of parental consent for referral
All independent variables were significance tested using chi-squares except for ‘Length of Relationship’
partners’ participation. Of the 73 couples
and ‘Age of Index Partner,’ for which we used t-tests.
recruited with the new one step protocol,
American Journal of Health Education — January/February 2011, Volume 42, No. 1 37
Angelic Rivera, Dana Watnick, and Laurie J. Bauman
significant increase in enrollment when we was an advantage to the study’s overall designated time frame, and to document all
changed protocols (from 70% to 96%). enrollment target. We found through TLS recruitment efforts throughout the study.
that boys reported fewer relationships that Second, research studies recruiting adoles-
Fictional Partners
qualified them to be in Teen Connections cent couples need to understand how words
We considered but rejected the possibil-
than girls. Further, eligible couples where the may carry unknown and/or unintended
ity of advertising for adolescent couples to
boys were the index partners were less likely meanings for participants in romantic rela-
enroll in the study. In addition to the bias
introduced by a volunteer sample, we feared to enroll in the study than couples where the tionships. In particular, the word “partner,”
that some adolescents would “recruit” a girls were the index partners (74% vs. 88%; a common term used in many measures of
fictional partner in order to participate and P<0.05). Of the 139 couples enrolled, 37 sexual risk behavior, has multiple meanings
receive the $30 monetary honorarium. We (27%) couples were enrolled using a male for youth, different for boys and girls, and
preferred a procedure that would use the index partner. Despite there being fewer may cause a recruitment bias. Third, when
Teen Lifestyle Survey to identify eligible male index partners to begin with, and a recruiting adolescent couples, studies should
youth to avoid this since partner data could lower rate of enrollment when the male avoid burdening the index respondent with
be reviewed and couple status verified prior was the index partner, including males did responsibilities for recruiting their partner,
to the teens being invited into Teen Con- increase the overall rate of couple enrollment such as passing on study materials or making
nections. The use of the screener to verify into the study. phone call reminders. Half-way through our
couple status was a necessary procedure. Potential for Inter-Partner Violence study we recognized that our waiver protocol
Of teens who initially screened eligible for A main concern in implementing this posed several recruitment challenges because
TC, 8% (N=17), tried to bring in a fictional study of young couples was assuring the of the burden placed on the index partner.
partner, usually a friend, cousin, or class- confidentiality and the safety of each partner. In effect, index teens were placed in the role
mate. Without the screener it would have Several studies16, 17, 26 have raised concerns of broker between the project recruiter and
been nearly impossible to identify fictional that measures of condom use can raise is- the referral partner, and many teens reported
partners. When these incidents occurred we sues of infidelity or lack of trust that might that the process was overwhelming for them.
spoke with the index teen to assess whether cause tension in the relationship, possibly Fourth, studies recruiting adolescent couples
it was intentional or a misunderstanding. threaten the stability of the relationship, should consider using a screener question-
In all 17 cases, index teens reported that the or of more concern, cause conflict between naire to verify the legitimacy of the dyadic
monetary incentive motivated them to try to partners leading to potential intimate part- relationship since it was our experience that
remain in the study regardless of change in ner violence. We addressed these concerns a monetary incentive did encourage some
eligibility status; typical status changes were seriously by taking steps to protect the safety teens to create a fictional partner. Fifth, a
break-ups or referral partners who did not and privacy of each partner during couple couples study may prove to be less burden-
want to participate. enrollment: (1) each partner was given the some to teens that are in serious relation-
Implications of Recruiting Couples opportunity to interact with study staff ships compared to those in non-serious
through Male Partners privately to express any concerns; (2) during relationships. It was our experience that
Almost all previous research of adult every data collection point each partner was partners in serious couples are more likely
heterosexual couples started with a female placed in a separate data collection room to to engage in planned events and socialize
as the index partner and a male partner re- avoid either partner seeing or reading the with one another, while teens in non-serious
cruited through her. We considered whether other’s responses; and (3) survey data or relationships often had no expectation of
to recruit couples solely using girls as index other personal information was never shared calling or seeing each other. Thus, studies
partners or to permit both boys and girls between partners. At the time of this publi- working with adolescent couples need to
to be identified as index partners. We con- cation, no adverse events were reported. consider flexible and fluid approaches (e.g.
ducted a pilot with 16 adolescent couples, Lessons Learned extending enrollment periods, providing
where 8 of the couples were recruited using We have identified eight lessons learned materials in-person and via mail, collecting
a male index partner and the other 8 were that were pivotal to recruitment efforts. First, data separately for each partner) so that
recruited through a female. Regardless of studies of youth in general, and of young enrollment includes couples who socialize
index partner gender, couple recruitment couples in particular, must not underes- infrequently. While we were successful in
was equally successful and without incident. timate the staffing required to handle the enrolling 83% of couples eligible for Teen
based on this pilot recruitment experience, intensive day-to-day logistical field needs Connections, screened participants who
we decided to proceed with couple recruit- of the study. In our experience a small, tight reported their relationship as “not serious”
ment through either partner. team of trusted staff was essential to respond were more likely to refuse study participa-
The flexibility of recruiting couples to time-sensitive correspondence within 1-2 tion than teens engaging in relationships
through either the male or female partner days, to distribute study materials within a they described as serious and important. We
38 American Journal of Health Education — January/February 2011, Volume 42, No. 1
Angelic Rivera, Dana Watnick, and Laurie J. Bauman
believe refusals from teens in non-serious cent couples. First, parents are gatekeepers their recruitment experiences and strate-
relationships stem from the limited interac- to their child’s participation. We often see gies. Studies should track empirically how
tion these couples have. Thus, the ability to parental consent as a burden or barrier to changes in recruitment approaches affected
talk to the partner to provide an invitation to study participation and several studies have recruitment rates. by identifying the kinds
the study and schedule a date to come in to reported biased recruitment attributable to of teens that are hard to reach, flagging par-
complete an interview was not feasible and this requirement. 21-25 However, we believe ticipants with unstable living arrangements,
posed the most challenge during recruit- that parental consent can be compatible and calculating the number of contact points
ment efforts. Sixth, for studies recruiting with study recruitment and shouldn’t be required for study completion, we can better
adolescent couples, incorporating parental ruled out. Second, it is important for the understand which recruitment efforts are
signature may not interfere with participa- study staff to be the primary recruiters. most effective and efficient.
tion. However, special consideration needs As a general principle, the study should
to be given to ensure the following are not be responsible for recruitment rather than TRANSlATioN To HEAlTH
reported on the consent form: (1) disclos- burdening participants with recruiting their EDUCATioN pRACTiCE
ing it is a couples study, (2) disclosing the partners. Third, access to a very large num- Adolescent sexual behavior is rightly
study is recruiting sexual partners, and (3) ber of youth is needed to recruit couples in considered to be of vast importance due to
referring to the members in the relationship romantic relationships. There are multiple the risk of contracting sexually transmitted
as partners. Seventh, adolescent couples can challenges to recruiting an individual teen diseases. This preoccupation on adolescent
be recruited through male partners and did such as logistics, study eligibility, and the sexual behavior and its risks has ironically
increase the overall rate of couple enrollment role of their parent. Recruiting teen dyads lead to the neglect of an essential ingredient
into the study. Eighth, interpersonal violence more than doubles these challenges because to the understanding of this very topic, the
is an issue that requires special attention in of each partner’s study availability as well nature of adolescent romantic relation-
couple studies. In terms of data collection as inter-partner dynamics that may further ships in which most of this sexual behavior
procedures, one approach that was effective impede study participation. occurs.27-28 The study of individuals in a
for ensuring confidentiality and safety of relationship and the study of the relation-
Limitations
both participants was to use separate data ship as a dyad require two different levels
There are several limitations to acknowl-
collection rooms. This eliminated the pos- of analysis with the study of dyads involving
edge. First, Teen Connections did not have an
sibility of inadvertent disclosure of survey characteristics that are inherently relational,
intervention arm, thus the study requirements
responses from one partner to another, embedded in a relational context that has its
for couples were not as demanding as inter-
and the potential conflict or violence that own trajectory. 28-29 Little information about
vention protocols, but also we did not have
could ensue. adolescent relationships has been collected
a program to offer youth as an incentive to
from both members of the dyad rather than
enroll. Second, our sample consisted mainly
DiSCUSSioN a single partner. However, individual percep-
of African American and Latino youth living
Recruitment of adolescent couples can tions often are not adequate to evaluate the
in New York City and our findings may not
be perplexing even for the most experi- functioning of the couple because there may
be generalizable to youth of other ethnicities
enced researchers. To date, we found no be discrepancies between partner reports.
or those living in other geographical regions.
publications that address recruitment of Relationships are influenced by the interac-
Third, over two-thirds of our sample reported
adolescent couples where both partners are tion of the partners’ characteristics, their
being in a serious/committed relationship,
recruited and one or both are under the age feelings toward each other, their history
and our procedures may not be as appropri-
of 18. Recruiting teen couples in romantic and their expectations.28 Thus, methodolo-
ate or efficient for recruiting those in newly
relationships is feasible but demanding. gies that address subjectivity, examine how
formed or non-serious relationships. Finally,
Teen Connections revised its recruitment individual-level and dyadic-level variables
although our opt-out letters were available in
protocol two times and printed new ma- interface, and develop strategies for inter-
Spanish and English, our TC parental consent
terials several times in a two-year period. preting discordant information are an im-
forms were only written in English. For those
Strategies that were successful at recruiting portant next step for theory and research on
participants recruited through the parental
adolescents individually did not meet with adolescent relationships. Moreover, further
consent process, we may have limited the
success when implemented with adolescent study is needed to develop theoretical and
number of eligible teens because their parents
couples. Moreover, the response time in re- methodological approaches for adolescents
were Spanish-only speakers and, therefore,
vising an ineffective protocol was as critical that address the formation, progression
could not give consent.
as the new protocol developed. and dissolution of individual relationships
We have identified several principles Future Recommendations and the changing of relationships over time
from our experience in recruiting adoles- We encourage researchers to share as adolescents move from early to middle
American Journal of Health Education — January/February 2011, Volume 42, No. 1 39