Developmental Psychopathology and Ethnicity: I. The Young Adulthood Assessment of the Boricua Youth Study

Abstract

Objective:

Developmental psychopathology processes pertinent to underserved ethnically diverse youth may not always coincide with those relevant to youth from non-disadvantaged groups. This article reports on the young adulthood assessment (fourth wave; April/2013-August/2017) of the Boricua Youth Study (BYS), which includes two population-based samples of children of Puerto Rican background (N=2,491) aged 5 to 13 years (recruited in 2000), in the South Bronx (SBx), NYC and San Juan, Puerto Rico (PR).

Method:

Study procedures included intensive participant tracking, in-person interviews of young adults, and when possible, their parents. Study participation rates, measures and weights are described.

Results:

At BYS Wave 4 (on average 11.3 years since last wave of participation), we re-assessed 2,004 young adults (mean age=22.9, range 15–29 years; 51% women; retention rate adjusted for ineligibility=82.7%) and available parents (n=1,180). Non-participation was due to inability to locate/contact participants (8.6%); refusal (4.7%) and ineligible status (2.8%) due to cognitive impairment, incarceration or death. Among those originally from PR, 91% stayed in PR during young adulthood. Of those from the SBx, 52.4% remained in the area (85.8% within 100 miles). Most study measures had good internal consistency (Cronbach’s alpha 0.70 or above).

Conclusion:

Our results support the viability of retaining a population-based cohort of children from the same ethnic group across two contexts during a life stage when individuals are likely to move. Longitudinal samples that are generalizable to underserved populations can elucidate developmental processes of relevance for curtailing the risk of psychopathology in disadvantaged contexts.

Introduction

The period spanning childhood through young adulthood is critical for understanding the development of psychiatric disorders. Longitudinal, population-based studies that originate in childhood and follow children and their families into adulthood are the preferred way of generating knowledge about developmental psychopathology. These studies involve high levels of complexity and resources, but are ideally suited to inform intervention strategies. In the US and worldwide, such studies have failed to focus on relevant processes within relatively homogeneous racial/ethnic minority groups exposed to high levels of poverty, stressors, and adversities.^1–10 This limitation is difficult to address, as historically, due to factors such as lack of trust or a different perspective about the potential benefits of the research endeavor, racial/ethnic minority groups have been challenging to recruit and retain in research studies.⁶ Better representation and retention in population-based research studies could reveal, for example, relevant patterns of mobility and mortality among youth of racial/ethnic minority groups growing up in disadvantaged areas potentially related to development of psychopathology.

The present report focuses on the methods employed to carry out a fourth wave of longitudinal assessment of children of Puerto Rican ethnic background who were recruited and retained from childhood to young adulthood in the Boricua Youth Study (BYS). The BYS was named after the original indigenous name of the island of PR, Boriken, based on which, to the present day, individuals of Puerto Rican background are still referred to as “Boricuas”. The BYS brings three main innovations to the field of developmental psychopathology. First, it focuses on a homogeneous Latino subgroup (Puerto Ricans), specifically, the Latino subgroup with the highest prevalence of several psychiatric disorders in the US.¹¹

Second, with its two-site design, it is the only study starting in childhood that examined psychopathology in population-based samples of the same ethnic group living in two unique disadvantaged contexts – the South Bronx, New York (SBx) and the San Juan and Caguas Metropolitan Area, Puerto Rico (PR) – with common and unique risks. The SBx is the US congressional district with one of the highest proportions of children living below the poverty line,¹² similar to the situation of youth growing up in PR.¹³ In both contexts, about 60% of the children live in single-parent families (64% in the Bronx and 59% in PR). Youth in the SBx, however, growing up as part of an ethnic minority group, are likely to face ethnic discrimination throughout life, an experience likely related to psychopathology.^14,15 Rates of people living with HIV/AIDS were 10 times higher than the US average in the SBx, in 2017,¹⁶ 5 times more than the total PR population and over 2 times higher than those in San Juan, PR.^16–18 Rates of STI infection in the SBx ranged from 3 to 7 times higher than the rates for those living in PR in 2016.^17,19 For Puerto Ricans, not all citizenship rights have been applicable to those living in the island as a result of its history as a Spanish colony and current status as a Commonwealth of the United States. The healthcare system, for example, is financed by government programs at lower rates than those in the US mainland. This has created an unstable, many times insufficient health system, possibly contributing to high rates of chronic diseases (e.g. diabetes, hypertension and asthma).^20,21 This situation has been aggravated by the substantial economic decline experienced by Puerto Rico, now for more than one decade.²²

Third, to our knowledge, the cohort based in PR provides the only source of population-based developmental psychopathology data on Latino children living in a disadvantaged socio-economic context outside of the continental US and assessed over the course of almost two decades. Informed by a developmental psychopathology framework,²³ recently expanded to include the consideration of cultural influences,^24–26 the BYS is well-positioned to generate additional knowledge about the influence of culture on the development of psychopathology among ethnically diverse children as they enter young adulthood. The first three waves of the BYS focused on the development of disruptive behavior disorders (DBDs) and antisocial behaviors (ASB) among Puerto Rican children, ages 5–13 at baseline, living in the two contexts.²⁷ The results showed that over time, Puerto Rican children from the SBx started to present higher prevalence rates of DBDs, as well as higher levels of ASB and delinquency, as compared to those residing in PR.^27–29 While culturally-relevant risk (e.g., cultural stress or discrimination) and protective factors (e.g., familism) for ASB exerted similar effects in both contexts, risks were more frequently present in the SBx. The BYS has also provided information about other topics relevant to understanding developmental psychopathology within a racial/ethnic minority group. For example, we have shown how, in a homogenous subgroup of Latino youth, risk factors and outcomes may differ by social context,^30–32 or how racial/ethnic minority-related stressors, such as acculturation and cultural stress, may account for contextual differences in psychopathology.^33,34

With the fourth wave of assessment, the scope of the BYS’s contribution is broadened to include young adulthood, a critical developmental period when common psychiatric disorders start or increase substantially.^35,36 Young adulthood is progressively recognized as one of the most challenging developmental transition periods,³⁷ as youth begin to assume adult roles and responsibilities,⁸ and establish patterns of positive or risky behaviors that not only influence immediate functioning, but also promote or curtail optimal functioning and health later in life.

In what follows, we describe the methods used in the young adulthood assessment of the BYS. Besides summarizing information relevant for those specifically interested in the BYS, this paper also provides useful material about methodologies utilized to recruit young adults, retain disadvantaged youth and their parents, despite frequent relocation, over the course of almost two decades. Methodological strategies employed to ensure comparability across two disadvantaged contexts with their unique qualities are features that make the BYS a novel study.

Method

Sampling and Study Design

The methodology employed in the initial study has been described in detail previously.^27,28 In brief, the samples are multistage household probability samples representative of two target populations: the Metropolitan Area of San Juan and Caguas, Puerto Rico and the South Bronx, New York. The initial recruitment took place in 2000. In each household, up to 3 children of Puerto Rican background ages 5 to 13 years were included (3 children were randomly selected in the households that had more than 3 age-eligible children). Being of Puerto Rican background was defined as having at least one primary caretaker who self-identified as being of Puerto Rican background.^27,28 Three yearly waves of assessment (Waves 1–3, or W1-W3) were carried out from 2000 to 2004 (total N=2,491 at Wave 1; SBx=1,138; PR=1,353), including children and their primary caretaker. The fourth study wave (W4; from April 2013 to August 2017) targeted all BYS youth who participated in W1. A total of 2,004 BYS youth ages 15 to 29 participated in W4. In W4, whenever possible, a parent (N=1,180) was also interviewed. Youth at W4 were deemed ineligible for the study if, based on self- or family-report, they had cognitive or neurological impairments, were deceased (death records were also searched) or in prison for the entire duration of the data collection period.

Sampling and non-response weighting:

The original (W1) BYS sampling design includes sampling weights to provide estimates that represent the full population of children of Puerto Rican background ages 5 to 13 in the SBx and the target areas in PR. The BYS sampling weights adjust for the probability of household and individual selection resulting from the sampling design in each site (originally based on the 1990 US Census) adjusted for the sex and age distributions in the 2000 US Census.²⁷ Two sets of sampling weights are available: 1) the full BYS sampling weights (or the full target-population sampling weight) that differentially weight the samples from the two sites according to the full population sizes of the SBx and the target areas in PR. Thus, the sample from the SBx is up-weighted and the sample from PR is downweighted since the SBx has nearly 3 times the total population size of the target areas in PR, and 2) the site-specific sampling weights that simply rescale the full BYS sampling weights so that each sample is weighted proportionate to the nearly equal sample size in each site.

For W4, we developed non-response propensity weights using standard methods that incorporate known information from participants collected at earlier waves.³⁸ Specifically, we performed a logistic regression of responder status at W4 (yes/no) as the outcome predicted by several W1 demographic, psychopathology, and contextual variables to obtain the propensity of being a W4 responder. Following recent recommendations,³⁹ this logistic regression incorporated the BYS sampling weights. To maximize prediction, all chosen predictors were included in the logistic regression regardless of statistical significance. The newly developed W4 weights were then calculated as the product of the BYS sampling weight times the inverse of the propensity of being a W4 responder. This new W4 weight adjusts for non-response so that estimates from W4 responders are representative of the full BYS population. Checks were performed to ensure there were no outlying weights, and that the largest weight was no greater than 10 times the size of the smallest weight so that no single participant would contribute disproportionately to the estimates.⁴⁰

Procedures

Study procedures for W4 were approved by the Institutional Review Boards (IRB) at the New York State Psychiatric Institute, the University of Puerto Rico Medical School, Cambridge Health Alliance, and Massachusetts General Hospital. Parents and youth 18 years of age and older in the mainland US, and 21 years of age and older in PR, signed informed consent. Youth younger than these ages signed assent forms and their parents signed informed consent.

Tracking of Study Sample and Cohort Maintenance:

Since the earlier waves 1–3, cohort maintenance required constant updates of participants’ contact information through relatives, reliance on detailed contact information from at least three re-contacts (or three people that would always know the whereabouts of each participant), yearly holiday cards to maintain contact, and constant tracking of address changes. Beginning 9 years after completing W3 data collection, intensive tracking procedures were used to locate participants from both sites for participation in BYS W4. Greater resources were allocated to recruitment of youths rather than their parents because of their new role as main informants in W4. Tracking procedures varied depending on the level of response of participants. Initially, participants were contacted through mail, followed by phone calls, and then home visits at different days and times. Every attempt to contact a participant was documented. Field supervisors reviewed the contact attempt history and developed new strategies tailored to each participant. For difficult to locate participants, a paid locate-and-search tool was also used. For participants who had moved outside of the recruitment site, interviews were carried out in-person or over the phone (the latter did not include questions assessing suicide due to an IRB requirement).

Interview Procedures and Quality Control:

Computerized parent and young adult interviews were available in both English and Spanish and were conducted in the participant’s preferred language, usually in the participant’s home. The equivalence of the translated instruments was safeguarded using procedures that have been described elsewhere.^27,41,42 The research team was flexible and always tried to accommodate participants’ schedules and preferences; therefore, interviews were also conducted after hours (early evening) and on weekends (particularly in the SBx site), in parents’ or other relatives’ homes, offices, or clinics, as long as a private area was available. Interviewers were lay persons who were intensively trained, culturally competent, and bilingual or fluent in Spanish. Most had an Associate’s or Bachelor’s degree. Every interview in both sites was audio-recorded for quality control if the participant consented to being audio recorded. The first two interviews of each interviewer were listened to in their entirety by a field supervisor and 15% of subsequent interviews were spot checked at random. Ongoing quality control and data cleaning processes were employed throughout the study.

Measures

Measurement in BYS W4 consisted mostly of self-reported psychiatric and risk/protective factor measures, collection of urine and saliva for assessing HIV and sexually transmitted infections (STIs), as well as in-depth qualitative interviews and neighborhood observations (to be described in future manuscripts). Whenever possible, measures used in the first three waves were included in W4 with the aim of comparability across waves. Also, measures indexing developmentally appropriate processes pertinent to late adolescence and young adulthood were added.

Data Analysis

Retention rates were calculated based on the number of participants who completed interviews at W4 in relation to 1) all W1 participants except those identified as ineligible at W4 (described below), 2) all W1 participants, and 3) all participants in the last wave of assessment (W3) (Table 1). Non-participation in BYS W4 was categorized into five types: (a) Refusal; (b) Unable to contact/locate; (c) Contacted and unable to interview; (d) Incomplete; (e) Identified as ineligible (Table 2).

Table 1.

Participants (Young Adults and Parents) Who Completed Interviews for Wave 4 of the Boricua Youth Study, by Site

	Wave 4 (2013–2017)
	Both Sites	SBx at W1	PR at W1
Number of Wave 4 participants
Young adult	2004	921	1083
Parent	1181	490	691
Parent or young adulta	2182	1003	1179
Number of Child participants at Wave 1	2491	1138	1353
Number of Parent participants at Wave 1	1643	721	922
Ineligible Young Adults at Wave 4b	69	29	40
Ineligible Parents at Wave 4b	46	38	8
Retention rate adjusted for ineligibilityc, % of Wave 1 participants excluding those identified as ineligible at Wave 4
Young adults	82.74	83.05	82.48
Parents	73.95	71.74	75.60
Retention rate, % of Wave 1 participants
Young adults	80.45	80.93	80.04
Parents	71.88	67.96	74.95
Retention rate, % of Wave 3
Young adult (N both sites: 2177. n Bronx: 968. n PR: 1209)	92.05	95.14	89.58
Parents (N both sites: 1435. n Bronx: 610. n PR: 825)	82.23	80.33	83.76
Time since last assessment, years
Mean (SD)	11.33 (1.36)	11.29 (1.44)	11.37 (1.29)
Median	11.11	10.98	11.19
Range	(8.93, 16.58)	(8.96, 16.58)	(8.93, 15.58)
Young Adult Age
Mean (SD)	22.88 (2.85)	22.91 (2.86)	22.86 (2.84)
Median	22.96	23.02	22.91
Range	(15.72, 29.46)	(15.72, 29.18)	(16.88, 29.46)
Young adult gender (%)
Female	1023 (51.05)	462 (50.16)	561 (51.80)
Male	981 (48.95)	459 (49.84)	522 (48.20)

Young Adult refers to youths aged 16–29

^aRefers to the number of young adults with either a parent or a young adult interview.

^bThose who are deceased, incarcerated, or have a serious cognitive disability (see Table 2)

^cRetention rate adjusted for ineligibility is calculated as, e.g. 82.78% = 2004/(2491–70)

Table 2.

Reasons for Non-participation Among Young Adults and Parents in the Boricua Youth Study, Wave 4

	Young Adults at W4						Parents at W4
	Both Sites W1 (N=2491)		SBx at W1 (n=1138)		PR at W1 (n=1353)		Both Sites W1 (N=1643)		SBx at W1 (n=721)		PR at W1 (n=922)
	n	%a	n	%a	n	%a	n	%a	n	%a	n	%a
Refused	118	4.7	45	4.0	73	5.4	64	3.9	27	3.7	37	4.0
Unable to contact/locate	214	8.6	85	7.5	129	9.5	226	13.8	78	10.8	148	16.1
Contacted but unable to interview	74	3.0	52	4.6	22	1.6	126	7.7	86	11.9	40	4.3
Incomplete	11	0.4	5	0.4	6	0.4	<5	--	<5	--	<5	--
Ineligible	69	2.8	29	2.5	40	3.0	46	2.8	38	5.3	8	0.9
Cognitively impaired	11	0.4	6	0.5	5	0.4	<5	--	<5	--	0	0.0
Incarcerated	21	0.8	10	0.9	11	0.8	0	0.0	0	0.0	0	0.0
Deceased	37	1.5	13	1.1	24	1.8	45	2.7	37	5.1	8	0.9
Total (excluding ineligibles)	417	17.2	187	16.9	230	17.5	419	26.2	193	28.3	226	24.7
Total (including ineligibles)	486	19.5	216	19.0	270	20.0	465	28.3	231	32.0	234	25.4

Note. Refusal: participant or parent of minor expressed that they did not want to participate

Unable to contact/locate: participant was never spoken to. In this case, the study team may have spoken to the parent or relative, but never connected with the participant

Contacted and unable to interview: at some point participant was reached, but an interview was never conducted.

Incomplete: participant’s interview started but was never completed

Identified as ineligible: participant who was cognitively impaired or otherwise unable to be interviewed because of impairment per parent information; deceased; or in jail/prison and not released during W4 data collection period.

^aPercentages are of total at W1 and are not adjusted for sampling weights.

Participants were also described according to W4 interview method and location (Table 3). Addresses of place of residence of all study participants at W4 (exact addresses for those who completed the W4 interviews or most likely addresses for those who did not complete W4 interview but were located) were determined and geocoded using ArcMap 10.4.1.⁴³ Geographic location of W4 participants is described by study recruitment site at W1 (SBx or PR). Because a number of participants had moved during the intervening years, the boundaries for interviewing were not restricted to the original recruitment boundaries. Participants were classified as residing in the SBx, NYC (according to BYS W1 definition), anywhere on the island of PR, or in any other area within the US, the latter further subdivided by how distant they were from the SBx (within 100 miles; 100–300 miles and more than 300 miles). Participants were also classified by US region (Mid-West/West, South and Northeast), specific area of concentration (Bronx, New York City and Tri-state area) and the US states where most of the sample was concentrated.

Table 3.

Interview Method and Location of Young Adults the Boricua Youth Study, Wave 4

	Wave 4 Completers (N=2004)						Wave 1 Participants (N=2491)
	Both sites		South Bronx at Wave 1		Puerto Rico at Wave 1		Both sites		South Bronx at Wave 1		Puerto Rico at Wave 1
	N	%	N	%	N	%	N	%	N	%	N	%
Interview method
In person	1803	89.97	840	91.21	963	88.92
Telephone	201	10.03	81	8.79	120	11.08
Participant location
South Bronxa	484	24.37	483	52.44	<5	0.09	524	21.04	522	45.87	<5	0.15
Puerto Rico	992	49.95	7	0.76	985	90.95	1083	43.48	8	0.70	1075	79.45
Other areas in the US	528		431		97
Within 100 miles of SBxb	315	15.86	308	33.44	7	0.65	356	14.29	340	29.88	16	1.18
100–300 miles from SBxb	53	2.67	28	3.04	25	2.31	91	3.65	34	2.99	57	4.21
More than 300 miles from SBxc	160	8.06	95	10.31	65	6.00	237	9.51	113	9.93	124	9.16
Location Unknown							200	8.03	121	10.63	79	5.84
Location in US mainland
Bronx	144	7.25	141	15.31	<5	0.28	156	6.26	153	13.44	<5	0.22
Other NYC Boroughs	93	4.68	93	10.10	0	0.00	98	3.93	98	8.61	0	0.00
Tri-state area (outside NYC)	92	4.63	78	8.47	14	1.29	127	5.10	92	8.08	35	2.59
Region in US mainland
Mid-West/West	35	1.76	18	1.95	17	1.57	52	2.09	19	1.67	33	2.44
South	126	6.34	79	8.58	47	4.34	185	7.43	97	8.52	88	6.50
Northeast	848	42.70	816	88.60	32	2.95	968	38.86	892	78.38	76	5.62
Top states in US mainland
New York	773	38.92	761	82.63	12	1.11	843	33.84	823	72.32	20	1.48
Florida	98	4.93	66	7.17	32	2.95	142	5.70	79	6.94	63	4.66
Pennsylvania	21	1.06	17	1.85	4	0.37	29	1.16	21	1.85	8	0.59
Massachusetts	10	0.50	3	0.33	7	0.65	28	1.12	5	0.44	23	1.70

^aSouth Bronx, NY according to Boricua Youth Study Wave 1 definition

^bNot in South Bronx

^cNot in Puerto Rico

BYS W4 study measures were summarized by category assessed and youth or parent construct measured (Table 4). Table S1, available online, includes Cronbach’s alpha (when appropriate) for study measures, a complete list of measures available across all BYS waves, the age for which they are available, and the informant who provided the information. Characteristics of participants and their parents utilized to create W4 non-responder weights were compared between W4 responders and non-responders (Table S2, available online).

Table 4.

Description of Study Measures Available at Wave 4

Category	Youth Construct (Measure)
Demographics	Demographicsa,b,c, Pregnancies, Children, and Parentingb, Government Benefitsb
Psychiatric Disorders/Problems	GAD (Composite International Diagnostic Interviewa,b), MDD (Composite International Diagnostic Interviewb), Distress (K-10b), Overall Mental Health, Posttraumatic Stress (Posttraumatic Stress Disorder Checklistb), Tobacco, Alcohol, and Illegal Substance Use Disorder (Composite International Diagnostic Interviewb), Substance Use Frequency (Youth Risk Behavior Surveyb), Callous Unemotionala,b, Oppositional Defiant Disorder (Diagnostic Interview Schedule for Children-IVb), Conduct Disorder (Diagnostic Interview Schedule for Children-IVb)
Psychosocial Factors	Bullyingb, Perceived Unfair Treatment (Experiences of Discrimination Scaleb), Relationshipsb, Sex Communicationb, Sex Communication Efficacyb, Sexual Behavior (AIDS-Risk Behavior Assessment)b,c, Sexual Orientationb,c, Access to Resources, Relationships, Controlb, Civic Participationb, Family Functioning (Family APGARb,c), Marital Disharmonyb,c, Mentoringb, Peer Delinquencyb, Peer Relationshipsb,c, Social Positionb, Social Supportb,c
Neighborhood Characteristics	Characteristicsa,b,c, Collective Efficacyb,c, Discrimination (adapted from the Everyday Discrimination Scalea,b), Environmental Hazardsa,b, Movinga,b
Cultural	Acculturation (Cultural Life Style Inventory Bidirectional Scaleb,c), Cultural Stress (Hispanic Stress Inventoryb,c), Ethnic Identityb, Familismb,c, John Henryism Scaleb, Migrationb, BYS, Minority Statusd, Minority Status Stress Scaleb, Skin Toneb
Trauma and Stress	Chronic Stressb, (Perceived Stress Scaleb), Stressful Life Eventsb,c, Exposure to Violenceb,c
Parenting/Abuse	Acceptance/Warmth (adapted from Hudson’s Index of Parental Attitudea,c), Abuse and Neglect (Parent-Child Conflict Tactic Scaleb,c), Parent-Child Relationshipb,c, Sexual Abuse (Sexual Victimization Scaleb,c)
Physical Health	Chronic Conditionsb,c, HIV Testingb, Overall Physical Healthb,c, Sexually Transmitted Infectionsb
Other	Delinquent Behavior (Elliott Self-Reported Delinquency Scale and adaptationb,c), Resilienceb, Sensation Seekingb,c, Mental Health Service Usea,b,c, Food Insufficiencyb, Heightened Vigilanceb, Home Physical Environment (Home Scalec), Religiositb,c
Category	Parent Construct (Measure)
Psychiatric Disorders/Problems	Antisocial Personality (DSM-IV and Family History Epidemiologica,c,BYS), Anxiety Symptoms (Generalized Anxiety Disorder 7 item scalea), Depressive Symptoms (Patient Health Questionnaire-9a,c), Substance Abuse (Family History Epidemiologica,c)
Cultural	Acculturation (Cultural Life Style Inventory Bidirectional Scalea,c), Cultural Stress (Hispanic Stress Inventorya,c), Ethnic Identitya
Other	Demographicsa,b,c, BYS, (Brief Trauma Questionnaira, Functioning (Short Form Health Survey - 12 item version (SF-12)a), Government Benefitsa, Impairment (World Health Organization - Disability Assessment Scale (WHO-DAS II)a), Suicidality Attempta,c, Social Positiona, Social Supporta,c

Note. Measure associated with construct in parentheses; if no parenthetical measure, construct name represents measure used.

^adenotes a parent-reported measure.

^bdenotes a youth-reported measure.

^cConstruct available in Waves 1–4. Reference cited represents the measure used at Wave 4. When developmentally appropriate, different measures were used at waves 1–3 (in childhood and early adolescence) and wave 4 (in young adulthood). See Table S1, available online, for more information.

^dDerived from residence at baseline

Results

As displayed in Table 1, at BYS W4, the retention rate adjusted for ineligibility was 82.7%, even though some participants had not been interviewed in over 16 years (mean interval between last wave of assessment and W4=11.33 years (SD=1.36), range=8.9 to 16.6 years).

At BYS W4, young adults were on average 22.9 years old (SD=2.9). Only 4.4% (n=88) of the W4 sample were younger than 18 years at the time of the assessment, with 10 participants as young as 15 or 16, and the oldest being age 29 (n<5), therefore justifying referring to the BYS W4 participants as young adults. Figure S1 (available online) displays the number of observations at each specific age, combining data from Waves 1 through 4 such that participants can contribute at up to 4 different ages.

Table 2 describes the reasons for non-participation in BYS W4 by young adults and parents. Except for 19.5% of the sample (486 young adults), most BYS young adults were interviewed. The main reason for non-participation in W4 was not being able to locate participants (214 young adults and 226 parents). Only 4.7% of BYS participants actively refused to participate in W4 and 11 young adult and 3 parent interviews were started, but not completed (0.4%). Among young adults, 29 BYS participants in the SBx and 40 in PR were considered ineligible. The main reason for ineligibility was being deceased [n=13; 1.15% (95%CI= 0.51–1.78)] in the SBx and [n=24, 1.73% (95%CI= 0.92–2.53)] in PR, after weighting by site-specific weights. These differences were not statistically significant (95%CI= 0.7–3.2).

More women than men participated at W4 in contrast to W1 (W4 sample is 51.1% women and 48.9% men, compared to 48.8% and 51.2%, respectively, at W1). Reasons for higher non-participation of men compared to women included being ineligible (4.4% vs. 1.1%), unable to be located (9.6% vs. 5.9%) and refusal to participate (5.7% vs. 3.6%).

In Table 3 and Figure 1, the geographic distribution of young adult participants at W4 is summarized for W4 completers and for all those who did not complete but whose residence was known (as the location of some W4 non-participants was known based on self, parental or relative report). Despite the mobility of the cohort, most interviews were completed in-person at participants’ homes, and 10% were completed over the telephone.

Figure 1.

Geographic Distribution of Boricua Youth Study Participants Who Completed a Wave 4 Interview (N=2,004) or Whose Residence Was Known

Note: BYS = Boricua Youth Study; W4 = fourth study wave.

SBx participants were more likely to move out of the original study site than those living in PR (OR=3.7, 95%CI=3.1, 4.5; p<.0001): 91.0% (95%CI= 89.2–92.7) of the PR cohort W4 completers did not move out of PR, whereas 47.6% (95%CI= 44.3–50.8) of the SBx cohort W4 completers moved out of their original recruitment site. However, of those, 33.3% (95%CI= 30.3–36.4) of the SBx sample was still located within 100 miles of the original recruitment site (mostly in other areas of the Bronx, some in other NYC Boroughs or in the Tri-State Area). Of the SBx cohort, 88.6% (95%CI= 86.5–90.7) were still in the Northeast region, while most of the PR cohort participants who moved to the US were in the South. Of note, 98 participants or almost 5% (95%CI= 3.9–5.8) of the full cohort moved to Florida. Interestingly, less than 10 participants from the PR cohort moved to the SBx, or moved from the SBx to PR. For approximately 8% (95%CI= 7.0–9.1) of W1 participants (about 11% (95%CI= 8.8–12.4) in the SBx and 6% (95%CI= 4.6–7.1) in PR, their location was completely unknown.

Table 4 summarizes the measures available in W4. Internal consistencies of dimensional constructs were calculated and ranged from 0.57 for the John Henryism Scale to 0.96 for youth post-traumatic stress (see Table S1; available online).

Table S2 (available online) compares W4 responders and non-responders on a series of variables that were utilized to create W4 non-response weights. Using variables collected at W1 to characterize differences, non-responders at W4 were more likely to be men, come from households with incomes below the federal poverty guidelines and receive welfare assistance, have non-married parents, have an externalizing disorder, have been exposed to violence, and have health and school performance worse than others in the same age group. In addition, non-responders at W4 had lower W1 levels of maternal education, maternal warmth, and parental monitoring, and higher levels of cultural stress and familism, antisocial behaviors, and neighborhood problems.

Discussion

The Boricua Youth Study builds on a strong tradition of population-based longitudinal psychopathology investigations, including US studies such as the Great Smoky Mountains Study,³ and the Pittsburgh Youth Study,⁵ and other non-US studies such as the Dunedin Study,⁴⁴ and the Christchurch Health and Development Study.⁴⁵ The BYS extends knowledge from other studies by following up – over almost 20 years – children from one specific ethnic minority group (Puerto Ricans) in a disadvantaged area in the US (47% of children living in poverty)¹² and the same ethnic group in their context of origin, where they are the majority population but also experience high levels of disadvantage. In 2015, 58% of children in Puerto Rico were living in poverty,¹³ a percentage comparable to other low-income countries worldwide for which comprehensive long-term developmental psychopathology information has just started to be generated.^46,47

The effort required to launch and successfully execute population-based longitudinal studies, like the BYS, cannot be underestimated. Increasing attention has been drawn toward conducting similar studies in low-and-middle income countries so that relevant social determinants of mental disorders can be addressed.⁴⁸ The approaches underlying the successful implementation described in this report can be summarized in three main recommendations:

1. Have clear, highly relevant research goals that are meaningful to the study population (e.g., generating knowledge about risk and protective factors for children’s wellbeing).

2. Establish flexible and multi-faceted recruitment, sample maintenance, and survey methods that allow intensive, face-to-face, after hours, and long-distance, multiple contacts with study participants, and that convey, with actions, the importance of study participants.

3. Secure and assemble adequate and reliable funding streams in order to enroll and track a large cohort, including hard-to-reach individuals.

The prospective, population-based sampling design of the study, coupled with high retention rates, allows the examination of the extent to which two different contexts expose youth to different risk and protective factors for psychiatric disorders. The BYS has focused on Puerto Rican youth mostly born and raised in two different settings. This has allowed for an in-depth examination of longer-term (second generation) acculturation, which is relevant given that early effects of the culture of origin may be protective for certain groups (a phenomenon labeled, for Latino individuals, The Latino Paradox⁴⁹). Risk of developing a range of health,⁵⁰ and mental health problems may increase with higher acculturation to the US. This may be true for specific Latino sub-groups, but not necessarily for Puerto Ricans,⁵¹ and possibly not for other groups with more prolonged and intensive contact with a host culture. The patterns of mobility identified among young adults who have been part of the BYS since childhood holds promise to also improve understanding about motivators of moving to a different context and their possible bidirectional relationship with psychopathology.

The retention rate for BYS participants is comparable to other successful longitudinal epidemiologic studies of psychopathology.^36,52–54 What is unique is obtaining such retention rate in a 15-year long cohort study of a group of individuals that is traditionally excluded from key institutions (e.g. health and mental health care, higher education) due to a lack of access to services. In that sense, the BYS provides a window to better understand developmental processes relevant to often excluded and socially deprived youth. However, non-participation was related to socio-economic disadvantage and risk factors for psychopathology (Table S2, available online). Aiming for high retention rates with a population-based design is key to understanding developmental psychopathology processes that are generalizable to children at high risk as a result of living in disadvantaged contexts. These findings can inform interventions and public policy for hard-to-reach populations.

The mortality rates observed in the BYS cohort are likely related to contextual factors. PR has a high homicide rate (approximately 20:100,000), 4 times higher than the average US rate and comparable to other low-income areas worldwide. The high crime rate in the SBx (998 violent crimes per 100,000 people) is 2 times higher than the national average.⁵⁵ It should be noted however, that the BYS mortality rates are comparable to those of the Dunedin Study (children born in the 1970s) and Great Smoky Mountains Study (children born in the 1980s), in which mortality rates by age 26 were 1.74% and 2.11%, respectively.^36,54 Considering the mortality rate in the US has steadily declined since the 1970s until 2010 when it stabilized, the mortality rate in the BYS (children born in the 1990s) might actually be higher in comparison to these other cohort studies.⁵⁶

Finally, we identified considerable geographic relocation/movement (mobility) in the cohort, particularly among those originally residing in the SBx. The fact that more than half of the SBx cohort has moved from its original recruitment location over the life of the study may be related to attempts to escape some of the risk factors present in the area, plus the intensive gentrification with substantial increases in rent prices which may be driving pre-existing inhabitants out of the area.⁵⁷ Historically, young adults ages 18 to 24 migrate at higher rates than other age groups.⁵⁸ Interestingly, however, the same level of relocation was not present in the PR cohort, indicating that Puerto Rican youth from the SBx and PR may experience very different mobility patterns. Also, surprisingly, virtually no participants relocated from PR to the SBx, which had been a preferred migration destination in the US for Puerto Ricans since the 1970s.⁵⁹ SBx cohort participants, despite moving at much higher rates, rarely moved to PR. Implications for developmental psychopathology of such shifts in migration patterns over the course of a few decades deserve to be understood, particularly in a globalized world where movement across national boundaries will in all likelihood continue to be frequent. The study was carried out prior to the devastating hurricane that affected Puerto Rico on September 20, 2017 which has resulted in a significant migration from the island to the mainland. What impact this event may have had on the mobility of PR participants, in addition to health outcomes among our participants is yet to be determined.

Despite the many strengths of the BYS study, there are several limitations. The study design permits the examination of how social, cultural, neighborhood, familial, contextual and individual characteristics are related to the development of psychopathology. However, genetic and biological markers are only starting to be assessed in Wave 5. Genetically-based vulnerabilities likely interplay with environmental factors to determine both onset and persistence of most psychiatric disorders.⁶⁰ Furthermore, what in one sense is a major asset, the fact that only one ethnic group was studied may limit generalizability of findings. Nevertheless, there is wide agreement that reaching conclusions about heterogeneous groups (e.g. Latinos of different national origins) with varied levels of risk for health and mental health problems is not appropriate.⁶¹ By focusing on a homogeneous Latino subgroup with a very high risk of developing substance abuse and mental disorders,¹¹ the BYS has generated – and will continue to generate – information of high scientific quality about those who are most in need.

Uniquely among existing cohorts, the BYS is able to investigate the risk and protective processes within a specific racial/ethnic minority subgroup, known to be at high risk for future psychopathology,¹¹ including a wide range of risk and protective factors early in the life-course.^34,62,63 In this sense, this cohort expands the realm of research questions we have been able to investigate so far in the area of developmental psychopathology. Specifically, the BYS permits one to ascertain whether commonly observed developmental psychopathology processes, previously established in mostly non-Latino White samples, are also present in a specific racial/ethnic subgroup (both when this group is in a socio-cultural context where it is a racial/ethnic minority and where it is the majority). Furthermore, through the detailed longitudinal measurement of parental and youth social risks and cultural experiences (such as acculturation, cultural stress and familism), the BYS is uniquely informative about the role of risk and protective factors of specific relevance to understanding the development of psychopathology in a racial/ethnic minority group and a in cross-cultural context.