Secondary Logo

Journal Logo

Epidemiology and Prevention

Participation and Retention of Youth With Perinatal HIV Infection in Mental Health Research Studies

The IMPAACT P1055 Psychiatric Comorbidity Study

Williams, Paige L. PhD*; Chernoff, Miriam PhD*; Angelidou, Konstantia MS*; Brouwers, Pim PhD; Kacanek, Deborah ScD*; Deygoo, Nagamah S. MS, CCRP; Nachman, Sharon MD§; Gadow, Kenneth D. PhD

Author Information
JAIDS Journal of Acquired Immune Deficiency Syndromes: July 1, 2013 - Volume 63 - Issue 3 - p 401-409
doi: 10.1097/QAI.0b013e318293ad53



Children perinatally infected with HIV (PHIV) commonly suffer from psychiatric symptoms, particularly depression and attention deficit hyperactivity disorder (ADHD).1–14 Obtaining accurate estimates of the prevalence of mental health problems among PHIV youth is necessary for informing clinicians, developing policies, and targeting interventions but requires representative samples. Participation bias may occur because of recruitment from populations that differ with respect to personal and family characteristics. In addition, PHIV youth with comorbid mental health problems (or their caregivers) may show differential desire to participate in and remain involved in research studies.

Psychiatric conditions influence rates of both initial participation and retention of adults in health research surveys and longitudinal studies.15–26 Depression, anxiety, and panic disorder are associated with higher risk of loss to follow-up (LTFU), particularly because of lost contact.15–17 In addition, comorbidity with multiple psychiatric diagnoses has been associated with higher risk of attrition than single diagnoses.16 However, in a study from the Netherlands, adults with diagnoses of depression and anxiety were more likely to be included in initial study screening, attributed to more frequent contact with primary care doctors.17 Similarly, Pfeiffer et al27 noted an “unhealthy volunteer” effect for studies of psychiatric outcomes, where adults with more severe depression were more likely to participate. Whereas over-inclusion combined with higher attrition of participants with psychiatric disorders may in some cases balance to yield reasonable prevalence estimates, both forms of selection bias may impair generalizability and lead to biased inference regarding determinants of psychopathology.24

Fewer studies have evaluated the association of psychiatric conditions with participation and retention in studies of youth,28–30 and none have specifically evaluated the relationship in PHIV youth. Most studies in youth have focused on identifying predictors of treatment dropout for psychotherapy and have shown higher dropout among youth with depression.30–34 In one of the few longitudinal research studies of adolescents, no association of baseline psychopathology was observed with attrition, but a significant association was observed for incident disorders during follow-up with difficulty in contacting participants. Several studies, although not addressing psychiatric disorders, have identified sociodemographic and program-related factors associated with LTFU of PHIV youth in both developed countries35 and in lower-resource settings.36,37 Because many of these factors (including gender, race, age, socioeconomic status, and parental education level) may in turn be associated with risk of psychiatric conditions in PHIV youth, assessment of the impact of psychiatric disorders on attrition requires adjustment for these shared risk factors.

The P1055 study was designed to evaluate the prevalence and incidence of psychiatric problems in PHIV youth and an uninfected comparison group and was one of the first nationally based prospective assessments of psychiatric problems in this clinical population; results of various aspects of the study have been reported.8,9,38–41 An uninfected comparison group was included to facilitate a better understanding of how chronic HIV disease influences mental health problems above and beyond home environment. However, during the design phase, there were concerns that participating sites would tend to enroll the most compliant patients (perhaps less likely to have psychiatric problems) and that retention might be lower among youth with psychiatric conditions at study entry, limiting the ability to evaluate persistence and resolution of psychiatric symptoms over time. In this article, we describe the procedures designed to enroll a representative study sample of PHIV youth and evaluate the influence of co-occurring psychiatric symptoms on retention adjusting for sociodemographic factors.


Participants and Study Design

The P1055 study was a prospective, multisite, 2-year observational study of psychiatric symptoms in 2 cohorts of youth (PHIV and uninfected Controls) conducted at 29 sites in the United States between 2005 and 2008 by the International Maternal Pediatric Adolescent AIDS Clinical Trials (IMPAACT) network. Youth were eligible if they were 6 to 17 years old at study entry and had lived with the same primary caregiver for at least 12 months to obtain accurate caregiver assessments of youth symptoms. Youth with intellectual disability (IQ < 70) were excluded owing to concerns associated with an extensive battery of self-assessment measures. Controls were uninfected youth meeting these same criteria and were either perinatally HIV exposed or living in a household with at least 1 HIV-infected member.

To enroll a representative study population, some studies have approved participant registration requests with a certain probability, sometimes within study-defined strata.42,43 However, this approach can be biased by neglecting hard to reach patients who rarely present at the clinic. In addition, clinical research sites find such designs difficult to implement because willing candidates may be prevented from enrollment. Instead, we developed a design based on random ordering of all potentially eligible participants.

Sites submitted a screening form for each potential participant reporting the age, gender, and race/ethnicity. This information was deemed by both the funding agency and the individual site Institutional Review Boards (IRBs) to be sufficiently limited that it did not require signed informed consent. Using these screening forms, the PHIV-screened youth were randomly ordered into blocks of 8 within each site, balanced by gender and age (<12, ≥12 years). The recruitment blocks allowed sites to contact and enroll multiple participants within a given period, and sites were allowed to enroll as many youth as they could manage, provided they completed blocks in the prespecified order. Thus, all screened youth within a block either had to be enrolled, identified as ineligible, or confirmed as unwilling to enroll before the site could proceed to the next block. For screenees who agreed to enroll, both the screening ID and the standard patient ID numbers were entered, allowing the study team to link information at the individual level. A similar process was implemented for the Control group. Use of randomly ordered recruitment blocks provides a representative sample regardless of how many blocks a site completed, in contrast to simply allowing sites to choose which participants they wanted to enroll.

Similarity between age and gender distributions between the PHIV and Control cohorts was designed by setting target accrual of 100 participants for each of the 4 age–gender groups within each cohort, such that balance would be achieved if the study fully accrued its target sample size (400 within each of the PHIV and Control cohorts).

Study Assessments

Participants were assessed at study enrollment and at 1 and 2 years after entry, with each visit taking 2–3 hours. Instruments were available in both English and Spanish. Caregivers reported their education level, household income, and biological relationship to the study participant. Information about family environment and the youth’s neuropsychological, school, and social functioning was collected via questionnaires. For PHIV youth, antiretroviral medication histories and HIV disease severity measures were obtained through chart reviews and face-to-face clinical interviews.

At each study visit, participating youth and their primary caregivers completed behavior rating scales (Youth Inventory-4 and Child and Adolescent Symptom Inventory-4R, respectively) based on the Diagnostic and Statistical Manual of Mental Disorders IV (DSM-IV).44,45 These rating scales include symptoms of common childhood psychiatric conditions, which we grouped into 4 categories of targeted disorders: ADHD (inattentive, hyperactive, or combined type), anxiety (generalized anxiety or separation anxiety disorders), disruptive behavior (conduct or oppositional defiant disorders), and depression (major depression or dysthymia). These previously validated rating scales do not require administration by licensed and/or trained psychologists, reducing their cost and increasing the ease of assessment, and have demonstrated high reliability and internal consistency for evaluation of child and adolescent psychiatric symptoms.46,47 Youth reported on their symptoms, and caregivers rated both their child’s symptoms and their own, as detailed previously.8,9 The presence of psychological conditions was determined by predefined cutoff scores (as defined by DSM-IV threshold criteria) for each condition. Severity scores also were computed for each condition. In addition, we constructed a composite category of “any disorder,” reflecting whether a participant met or exceeded threshold criteria for any of these targeted conditions or for a limited number of additional disorders (somatization, bipolar disorder, posttraumatic stress disorder, or social phobia).8,9,38–41 Younger children (<12 years) did not self-report ADHD or disruptive behavior symptoms owing to concerns about the validity of self-assessment of these symptoms by young children.46,47

This study was approved by an IRB at each participating clinical site. Written informed consent was obtained from primary caregivers of all participants, and assents were obtained from youth as allowed by local IRBs. Each participating site submitted an implementation plan for managing psychiatric referrals and unintended HIV disclosure, recruitment and retention, incentives, and quality control. Incentives varied by site owing to differences in local IRB requirements and included cash or vouchers to cover the cost of attending the study visit (eg, meals or parking) and small cash amounts, gift cards, or vouchers.

Statistical Analyses

Participation rates were summarized overall, by age/gender strata and HIV status, and compared between cohorts using Fisher's exact test. Reasons for ineligibility and refusal to participate were compared using χ2 tests. Logistic regression models were used to evaluate sociodemographic risk factors for LTFU, which we defined as missing the 2-year study visit. Models were fit including all enrolled participants with a completed baseline visit and were separately fit to the subset of PHIV participants to evaluate the added contribution of HIV disease severity. Individual covariates with P < 0.15 were considered for inclusion in a multivariable model, which was reduced to those covariates with P < 0.15. This resulted in 2 separate core models, 1 for all participants and 1 among PHIV participants. We fit logistic regression models to evaluate both youth self-assessed and caregiver-assessed psychiatric conditions at baseline as risk factors for LTFU, both unadjusted and adjusted for the aforementioned core covariates. Primary emphasis was given to presence of psychiatric conditions based on DSM-IV symptom cutoffs, but we also evaluated the continuous severity scores of each psychiatric condition. In addition, we evaluated potential interactions of HIV status with both demographic and psychiatric conditions on risk for LTFU via inclusion of interaction terms in the logistic regression models. Sensitivity analyses were conducted to account for correlation of youth within both families and within clinical sites, via generalized estimating equations with an equicorrelation assumption.


Characteristics of Screened Candidate Participants

A total of 1328 PHIV and 986 Control youth were screened by participating sites (Table 1; Fig. 1), of whom 1319 and 962, respectively, were randomly ordered into blocks of 8 youth within each site. Not included were individuals who did not meet the age requirements or were screened after the announced deadline. Four screenees under age 6 were included with the stipulation that study visits occur after age 6. Among those included in the randomly ordered site lists, 863 screenees (485 PHIV and 378 Controls) were never contacted because of study closure. Of the remaining 1418 screenees, 256 were determined to be ineligible (174 PHIV and 82 Controls). The primary reason for ineligibility was that the youth had not lived with their caregiver for the past 12 months (101 screenees, or 39% of ineligible). Ineligibility was more often the result of intellectual disability in PHIV youth than in Controls (26% vs 9%), consistent with higher cognitive impairment observed in PHIV youth48 and also was more often attributable to death or relocation (16% vs 0%, respectively), whereas caregiver issues accounted for a similar percentage of ineligible PHIV and Control youth (40% and 38%, respectively).

Number of Candidate Participants Screened, Randomly Ordered, and Enrolled by HIV Infection Status and Age and Gender Strata
P1055 study design flowchart.

Participation Rates and Reasons for Refusal

Of the 1162 eligible youth who were approached, 582 participated (323 PHIV and 259 Controls, 49% and 52% of eligible, respectively) (Fig. 1). Over half of the sites (16 of 29, 55%) enrolled more than half of their eligible contacted youth, with 5 sites enrolling more than 90% of their eligible candidates. Among the 580 youth who refused to participate, the primary reasons included difficulties with the time commitment or travel to the site (n = 205, 35%), caregiver concerns about privacy or disclosure of HIV status (n = 123, 21%), active illness preventing participation (n = 10, 2%), and general unwillingness to participate (n = 242, 42%), with no overall difference in the distribution of reasons for refusal between PHIV and Controls.

Participation rates by youth and caregiver characteristics at screening are summarized in Table 2. Participation rates did not differ significantly by gender or by HIV status. Participants were slightly older on average than nonparticipants and participation rates were higher among black non-Hispanic and Hispanic youth than among white non-Hispanic youth (P = 0.01). The older age of participants versus nonparticipants was primarily observed among the PHIV screenees (median 13.1 vs 12.0 years, P < 0.001), whereas there was no age difference for Controls (median 11.2 vs 11.0 years, P = 0.40).

Comparison of Participation Rates by Screening Characteristics, Among 1162 Eligible Contacted Participants

PHIV youth and Controls who enrolled were comparable by gender, race, and ethnicity, as reported previously.9 However, PHIV youth were older on average than Controls, less likely to live with biological parents and more likely living in households with higher household income and caregiver education. At study entry, 83% of the PHIV youth were taking highly active antiretroviral therapy (HAART). At the baseline visit, both groups had a similar prevalence of psychiatric symptoms based on either youth self-report or caregiver assessment (61% with at least one condition), although PHIV youth had slightly higher rates of conduct disorder.8,9

Retention in Study at 2-Year Follow-up Visit

Among the 575 enrolled participants who completed a baseline visit, 506 (88%) attended the 1-year follow-up visit and 465 (81%) attended the 2-year follow-up visit, including 22 youth who missed the Year 1 visit but returned at Year 2 (Fig. 1). The retention rates over the 2-year study were significantly higher for PHIV participants than for Controls (84% vs 77%, P = 0.03). In adjusted logistic regression models for LTFU (Table 3), PHIV participants and those whose primary caregiver was their biological mother had reduced odds of LTFU [adjusted odds ratios (aORs) = 0.45, P = 0.002; aOR = 0.61, P = 0.06, respectively]. In contrast, increased odds of LTFU were observed for youth whose caregivers had at least a high school education (aOR = 2.40, P = 0.003) or higher annual income (aOR = 1.61, P = 0.07). Although Hispanic youth had reduced odds of LTFU in unadjusted models, this finding did not persist after adjustment for other sociodemographic factors. We observed no association of age with LTFU after adjustment for HIV status and no interaction between HIV status and any demographic risk factor.

Predictors of LTFU at 2 Years Based on Logistic Regression Models, Including 575 Eligible Youth With Completed Baseline Visit

Our study included 457 families, of which 91 (20%) enrolled 2 or more children and 45 (10%) included at least 1 PHIV and 1 Control youth. However, even after adjusting for within family correlation in LTFU, there remained a significant decrease in LTFU for PHIV as compared with Control youth (aOR = 0.49, 95% confidence interval: 0.32 to 0.75, P = 0.001). Effects of other sociodemographic predictors shown in Table 3 remained similar. Sensitivity analyses accounting for clustering of youth within research sites also yielded very similar estimated associations.

In models adjusting for sociodemographic covariates, youth with at least one self-assessed psychiatric condition compared with those without had a 56% increased odds of LTFU (P = 0.05, Table 3). The rate of LTFU for Controls was fairly similar for those with and without a psychiatric condition (20.9% vs 24.1%), whereas the percent LTFU was twice as high for PHIV youth with versus without psychiatric conditions (23.9% vs 10.9%, P-value for interaction test = 0.011). Although no individual condition was a significant predictor of LTFU, estimated odds ratios were consistently >1. No association with LTFU was observed for any psychiatric condition by caregiver assessment (Table 3). Additional analyses evaluating associations with symptom severity scores indicated greater odds of LTFU for youth with higher self-assessed severity of ADHD or disruptive behavior symptoms (see Table S1, Supplemental Digital Content,

Within the subset of PHIV youth, associations of demographic measures with LTFU were similar to those for the overall study population (Table 4). In addition, PHIV youth with higher viral load at entry had almost twice the odds of LTFU (aOR = 1.90, P = 0.08), whereas those receiving HAART at entry had reduced odds (aOR = 0.53, P = 0.08). After adjustment for sociodemographic and HIV disease severity measures, youth with any self-assessed psychiatric condition had over 3-fold higher odds of LTFU (aOR = 3.11, P = 0.001). The psychiatric condition associated with the highest odds of LTFU was anxiety disorder (aOR = 2.77, P = 0.003). However, in analyses based on severity scores, youth with higher self-assessed symptom severity had significantly higher odds of LTFU for each of the 4 targeted conditions (see Table S2, Supplemental Digital Content, In evaluating the association of caregiver-assessed psychiatric conditions with LTFU, we observed marginally increased odds of LTFU for youth with at least one condition (aOR = 1.90, P = 0.07) and for all other specific conditions except anxiety (Table 4). Higher caregiver-assessed severity scores for disruptive behavior were also associated with LTFU (see Table S2, Supplemental Digital Content,

Predictors of LTFU at 2 Years Based on Logistic Regression Models Among 319 Perinatally HIV-Infected Youth


As PHIV youth are living longer owing to advances in treatment and clinical management, improving our understanding of psychiatric and neurocognitive outcomes and their treatment options has developed into a top research priority. However, observational studies and clinical trials face challenges in enrolling representative samples of participants with HIV and those living in families affected by HIV. These difficulties are exacerbated by both declining participation in health research in the United States owing in part to decreases in volunteerism and a lack of trust in scientific research, particularly in pediatric and underserved populations.49–54

In our study, 50% of eligible youth agreed to participate. This rate is slightly higher than the 42% enrollment rate observed among the HIV-infected pregnant women enrolling into the US-based IMPAACT P1025 perinatal cohort study42 but is slightly lower than a single-visit study of metabolic abnormalities in youth with HIV, in which 60% of screened eligible youth participated.43 We observed similar rates of participation by HIV status, in contrast to the PACTS-HOPE study that reported a much higher enrollment rate in HIV-infected youth than in HIV-exposed uninfected controls (89% vs 22%).49 Pediatric health research studies in non-HIV settings have reported higher response rates for completion of questionnaires but lower response rates for studies that involve a clinical examination or collection of biological samples.51 Our study required no physical examination or collection of biological samples but required completion of an extensive battery of psychiatric questionnaires at the clinical sites, which may have led to decreased participation rates.

After exclusion of ineligible and nonapproached candidates, we observed the highest participation rates among Hispanic families, similar to findings of Freedman et al.49 The P1025 study of HIV-infected pregnant women also observed a higher enrollment rate among Hispanics than non-Hispanic whites (28% vs 20%) and an even lower rate among non-Hispanic blacks (17%).42 In our study, Hispanic participants also had lower odds of LTFU, but this finding did not persist in adjusted models, possibly because of the correlation of Hispanic ethnicity with other demographic measures.

The P1055 study successfully retained more than 80% of enrolled participants over the 2-year follow-up study period, which compares favorably with other studies of youth with HIV infection.35–37 High retention was achieved in part by requiring sites to develop a retention approach as part of their site implementation plan. Most sites attempted to maintain engagement with participants through frequent contact between study visits (eg, telephone, mailings) along with financial incentives or vouchers for meals and transportation costs. However, the retention rate for Controls was lower, which may be partially attributable to their avoidance of the potential stigma associated with being seen at an HIV clinic55 and by the fact that most did not receive routine care at the clinical sites.35

Although retention rates were high overall, youth with a self-assessed psychiatric condition were less likely to remain on study for the 2-year follow-up. This difference in retention rates was more evident in the PHIV youth than in the Controls, and among PHIV youth, there were also similar associations of youth- and caregiver-assessed psychiatric conditions with LTFU. Studies of drug and alcohol treatment programs among adolescents also have observed lower retention rates for those with ADHD, whereas inconsistent patterns of retention have been reported among those with conduct disorder.56,57 The observed association of lower retention among PHIV youth with psychiatric conditions at entry is consistent with findings in adults15–17 and has several implications. First, it may lead to underestimates in the prevalence of psychiatric conditions at later ages and may similarly bias estimates of persistent or resolving psychiatric diagnoses over time. Second, it may introduce bias into analyses of the determinants of psychopathology in PHIV youth, particularly for risk factors that influence both psychiatric status and attrition. Approaches to adjust for selective attrition such as use of inverse-probability-weighting methods, imputation of missing psychiatric status outcomes based on measured covariates, and sensitivity analyses under various assumptions regarding differences in associations between dropouts versus retained youth may be desirable if the rates of attrition are high.22,24,29

Other characteristics associated with lower retention in this study included higher socioeconomic status, in contrast to previous studies that have generally observed both lower participation rates and lower retention rates for those of lower socioeconomic status.35–37,50,53 Within the limited range of our study, we observed no differences by age in LTFU after adjustment for HIV infection status, whereas previous studies have found higher rates of LTFU among older adolescents.35,51 We observed higher retention among female participants, consistent with other epidemiologic research studies.50 Among PHIV youth, we observed marginally higher rates of LTFU among those with greater HIV disease severity and lower LTFU among those receiving HAART, as noted in other studies.35 These findings suggest that youth who return regularly to clinics to obtain antiretroviral therapy medications may find study participation more convenient or perceive more personal benefit for attending research visits. Alternatively, clinicians may be more willing to initiate HAART in PHIV youth perceived to have good compliance with both treatment and scheduled clinic visits.

Our study design neglected to take into account household structure, and thus siblings were occasionally in separate blocks and could not be enrolled at the same date. Future studies should consider random ordering in clusters by household to enhance enrollment of multiple family members. Our study strengths included being able to evaluate participation based on characteristics of individual participants. Finally, our study demonstrated the feasibility of maintaining relatively high retention rates over the 2-year follow-up period. Future studies may require creative approaches for engaging the HIV-affected community in research efforts and in promoting retention efforts, particularly for youth with psychiatric conditions.


The authors thank Kimberly Hudgens for her operational support of this study and Janice Hodge for data management. In addition to their contributions to the manuscript, the authors acknowledge the help given by Nagamah Sandra Deygoo in representing the site research staff on the protocol team and Vinnie Di Paolo for representing the community affected by HIV.

The following institutions and individuals participated in IMPAACT P1055:

Children’s Hospital Boston: Sandra Burchett; UCLA-Los Angeles AIDS Consortium: Karin Nielsen, Nicole Falgout, Joseph Geffen, Jaime Deville; Long Beach Memorial Medical Center: Audra Deveikis; UCLA Medical Center: Margaret Keller; University of Maryland Medical Center: Vicki Tepper; Chicago Children’s CRS: Ram Yogev; UCSF Pediatric AIDS CRS: Diane Wara; UCSD Maternal Child and Adolescent HIV CRS: Stephen Spector, Lisa Stangl, Mary Caffery, Rolando Viani; Duke University Medical Center: Kreema Whitfield, Sunita Patil, Joan Wilson, and Mary Jo Hassett; New York University School of Medicine: Sandra Deygoo, William Borkowsky, Sulachni Chandwani, and Mona Rigaud; Jacobi Medical Center: Andrew Wiznia; University of Washington Children’s Hospital, Seattle: Lisa Frenkel; USF Tampa: Patricia Emmanuel, Jorge Lujan Zilberman, Carina Rodriguez, and Carolyn Graisber; Mt Sinai School of Medicine: Roberto Posada and Mary Dolan; San Juan City Hospital: Midnela Acevedo-Flores, Lourdes Angeli, Milagros Gonzalez, and Dalila Guzman; Yale University School of Medicine: Warren Andiman, Leslie Hurst, and Anne Murphy; SUNY Upstate Medical University: Leonard Weiner; SUNY Stony Brook: Denise Ferraro, Michele Kelly, and Lorraine Rubino; Howard University: Sohail Rana; University of Southern California: Suad Kapetanovic; University of Florida Jacksonville: Mobeen Rathore, Ayesha Mirza, Kathleen Thoma, and Chas Griggs; University of Colorado: Robin McEvoy, Emily Barr, Suzanne Paul, and Patricia Michalek; South Florida Center for Diagnostic Care: Ana Puga; St Jude: Patricia Garvie; Children’s Hospital of Philadelphia: Richard Rutstein; St Christopher’s Hospital for Children: Roberta LaGuerre; Bronx-Lebanon Hospital: Murli Purswani; Metropolitan Hospital Center: Mahrukh Bamji; WNE Maternal Pediatric Adolescent AIDS: Katherine Luzuriaga.


1. Brown LK, Lourie KJ, Pao M. Children and adolescents living with HIV and AIDS: a review. J Child Psychol Psychiatry. 2000;41:81–96.
2. Jeremy RJ, Kim S, Nozyce M, et al.. Neuropsychological functioning and viral load in stable antiretroviral therapy-experienced HIV-infected children. Pediatrics. 2005;115:380–387.
3. Mellins CA, Brackis-Cott E, Dolezal C, et al.. Psychiatric disorders in youth with perinatally acquired human immunodeficiency virus infection. Pediatr Infect Dis J. 2006;25:432–437.
4. Nozyce ML, Lee SS, Wiznia A, et al.. A behavioral and cognitive profile of clinically stable HIV-infected children. Pediatrics. 2006;117:763–770.
5. Scharko AM. DSM psychiatric disorders in the context of pediatric HIV/AIDS. AIDS Care. 2006;18:441–445.
6. Donenberg GR, Pao M. Youths and HIV/AIDS: psychiatry’s role in a changing epidemic. J Am Acad Child Adolesc Psychiatry. 2005;44:728–747.
7. Mellins CA, Smith R, O’Driscoll P, et al.. High rates of behavioral problems in perinatally HIV-infected children are not linked to HIV disease. Pediatrics. 2003;111:384–393.
8. Gadow KD, Chernoff M, Williams PL, et al.. Co-occurring psychiatric symptoms in children infected with HIV and peer comparison sample. J Dev Behav Pediatr. 2010;31:116–128.
9. Chernoff M, Nachman S, Williams P, et al.; IMPAACT P1055 Study Team. Mental health treatment patterns in perinatally HIV-infected youth and controls. Pediatrics. 2009;124:627–636.
10. Mellins CA, Tassiopoulos K, Malee K, et al.; Pediatric HIV/AIDS Cohort Study. Behavioral health risks in perinatally HIV-exposed youth: co-occurrence of sexual and drug use behavior, mental health problems, and non-adherence to antiretroviral treatment. AIDS Patient Care STDS. 2011;25:413–422.
11. Malee KM, Tassiopoulos K, Huo Y, et al.; Pediatric HIV/AIDS Cohort Study Team. Mental health functioning among children and adolescents with perinatal HIV infection or exposure. AIDS Care. 2011;23:1533–1544.
12. Lam PK, Naar-King S, Wright K. Social support and disclosure as predictors of mental health in HIV-positive youth. AIDS Patient Care STDS. 2007;21:20–29.
13. Lee SJ, Detels R, Rotheram-Borus MJ, et al.. The effect of social support on mental and behavioral outcomes among adolescents with parents with HIV/AIDS. Am J Public Health. 2007;97:1820–1826.
14. Bauman LJ, Silver EJ, Draimin BH, et al.. Children of mothers with HIV/AIDS: unmet needs for mental health services. Pediatrics. 2007;120:e1141–e1147.
15. Lamers F, Hoogendoorn AW, Smit JH, et al.. Sociodemographic and psychiatric determinants of attrition in the Netherlands Study of Depression and Anxiety (NESDA). Compr Psychiatry. 2012;53:63–70.
16. de Graaf R, Bijl RV, Smit F, et al.. Psychiatric and sociodemographic predictors of attrition in a longitudinal study: the Netherlands Mental Health Survey and Incidence Study (NEMESIS). Am J Epidemiol. 2000;152:1039–1047.
17. Van Der Veen W, van der Meer K, Penninx BW. Screening for depression and anxiety: correlates of non-response and cohort attrition in the Netherlands Study of Depression and Anxiety (NESDA). Int J Methods Psychiatr Res. 2009;18:229–239.
18. Eaton WW, Anthony JC, Tepper S, et al.. Psychopathology and attrition in the epidemiologic catchment area surveys. Am J Epidemiol. 1992;135:1051–1059.
19. Bellón JA, de Dios Luna JD, Moreno B, et al.. Psychosocial and sociodemographic predictors of attrition in a longitudinal study of major depression in primary care: the predictD-Spain Study. J Epidemiol Community Health. 2010;64:874–884.
20. Gustavson K, von Soest T, Karevold E, et al.. Attrition and generalizability in longitudinal studies: findings from a 15-year population-based study and a Monte Carlo simulation study. BMC Public Health. 2012;12:918.
21. Torvik FA, Rognmo K, Tambs K. Alcohol an mental distress as predictors of non-response in a general population health survey: the HUNT study. Soc Psychiatry Psychiatr Epidemiol. 2012;47:805–816.
22. Glymour MM, Chêne G, Tzourio C, et al.. Brain MRI markers and dropout in a longitudinal study of cognitive aging: the Three City Dijon Study. Neurology. 2012;79:1340–1348.
23. Matthews FE, Chatfield M, Freeman C, et al.. Attrition and bias in the MRC cognitive function and ageing study: an epidemiological investigation. BMC Public Health. 2004;4:12.
24. Weuve J, Tchetgen Tchetgen EJ, Glymour MM, et al.. Accounting for bias due to selective attrition: the sample of smoking and cognitive decline. Epidemiology. 2012;23:119–128.
25. Badawi MA, Eaton WW, Weimer LG, et al.. Psychopathology and attrition in the Baltimore ECA 15-year follow-up 1981-1996. Soc Psychiatry Psychiatr Epidemiol. 1999;34:91–98.
26. Farmer ME, Locke BZ, Liu IY, et al.. Depressive symptoms and attrition: the NHANES I epidemiologic follow-up study. Int J Methods Psychiatr Res. 1994;4:19–27.
27. Pfeiffer PN, Szymanski B, Dhawan N, et al.. Is there an “unhealthy volunteer effect” in mental health outpatient research? Psychiatr Res. 2010;176:224–228.
28. Allott K, Chanen A, Yuen HP. Attrition bias in longitudinal research involving adolescent psychiatric outpatients. J Nerv Ment Dis. 2006;194:958–961.
29. Eerola M, Huurre T, Aro H. The problem of attrition in a Finnish longitudinal survey on depression. Eur J Epidemiol. 2005;20:113–120.
30. Gonzalez A, Weersing VR, Warnick EM, et al.. Predictors off treatment attrition among an outpatient clinic sample of youths with clinically significant anxiety. Adm Policy Ment Health. 2011;38:356–367.
31. Armbruster P, Kazdin AE. Attrition in child psychotherapy. Adv Clin Child Psychol. 1994;16:81–108.
32. Kazdin AE, Mazurick JL, Bass D. Risk for attrition in treatment of antisocial children and families. J Clin Child Psychol. 1993;22:2–16.
33. Pelkonen M, Marttunen M, Liappala P, et al.. Factors associated with early dropout from adolescent psychiatric outpatient treatment. J Am Acad Child Adoles Psychiatr. 2000;39:329–336.
34. Vanable PA, Carey MP, Carey KB, et al.. Predictors of participation and attrition in a health promotion study involving psychiatric outpatients. J Consult Clin Psychol. 2002;70:362–368.
35. Williams PL, Van Dyke R, Eagle M, et al.; PACTG 219C Team. Association of site-specific and participant-specific factors with retention of children in a long-term pediatric HIV cohort study. Am J Epidemiol. 2008;167:1375–1386.
36. Leroy V, Malateste K, Rabie H, et al.; International IeDEA Pediatric Working Group. Outcomes of antiretroviral therapy in children in Asia and Africa: a comparative analysis of the IeDEA Pediatric Multiregional Collaboration. J Acquir Immune Defic Syndr. 2013;62:208–219.
37. Fenner L, Brinkhof MWG, Keiser O, et al.; International epidemiologic Databases to Evaluate AIDS in Southern Africa (IeDEA-SA). Early mortality and loss to follow-up in HIV-infected children starting antiretroviral therapy in Southern Africa. J Acquir Immune Defic Syndr. 2010;54:524–532.
38. Serchuck L, Williams PL, Nachman S, et al.; IMPAACT P1055 Team. Prevalence of pain and association with psychiatric symptom severity in perinatally HIV-infected children as compared to controls living in HIV-affected households. AIDS Care. 2010;22:640–648.
39. Williams PL, Leister E, Chernoff M, et al.. Substance use and its association with psychiatric symptoms in perinatally HIV-infected and HIV-affected adolescents. AIDS Behav. 2010;14:1072–1082.
40. Nachman S, Chernoff M, Williams P, et al.. HIV disease severity and psychiatric symptomatology in perinatally HIV-infected youth. Arch Pediatr Adolesc Med. 2012;166:528–535.
41. Gadow KD, Angelidou K, Chernoff M, et al. Longitudinal study of emerging mental health concerns in youth perinatally infected with HIV and peer comparisons. J Dev Behav Pediatr. 2012;33:456–468.
42. Brogly S, Read JS, Shapiro D, et al.. Participation of HIV-infected pregnant women in research in the United States. AIDS Res Hum Retroviruses. 2007;23:51–53.
43. Aldrovandi GM, Lindsey JC, Jacobson DL, et al.; Pediatric AIDS Clinical Trials Group P1045 Team. Morphologic and metabolic abnormalities in vertically HIV-infected children and youth. AIDS. 2009;23:661–672.
44. Gadow K, Sprafkin J. Child and Adolescent Symptom Inventory-4R. Stony Brook, NY: Checkmate Plus; 2005.
45. Gadow KD, Sprafkin J. Youth's Inventory-4 manual. Stony Brook, NY: Checkmate Plus; 1999.
46. Gadow KD, Sprafkin J, Carlson GA, et al.. A DSM-IV-referenced, adolescent self-report rating scale. J Am Acad Child Adolesc Psychiatry. 2002;41:671–679.
47. Sprafkin J, Gadow KD, Salisbury H, et al.. Further evidence of reliability and validity of the Child Symptom Inventory-4: parent checklist in clinically referred boys. J Clin Child Adolesc Psychol. 2002;31:513–524.
48. Smith R, Chernoff M, Williams P, et al.; Pediatric HIV/AIDS Cohort Study. Impact of HIV severity on cognitive and adaptive functioning during childhood and adolescence. Pediatr Infect Dis J. 2012;31:592–598.
49. Freedman D, Koenig LJ, Wiener J, et al.. Challenges to re-enrolling perinatally HIV-infected and HIV-exposed but uninfected children into a prospective cohort study: strategies for locating and recruiting hard-to-reach families. Paediatr Perinat Epidemiol. 2006;20:338–347.
50. Galea S, Tracy M. Participation rates in epidemiologic studies. Ann Epidemiol. 2007;17:642–653.
51. Liese AD, Liu L, Davis C, et al.. Participation in pediatric epidemiologic research: the SEARCH for Diabetes in Youth Study experience. Contemp Clin Trials. 2008;29:829–836.
52. Claudio L, Stingone JA. Improving sampling and response rates in children’s health research through participatory methods. J School Health. 2008;78:445–451.
53. Ejiogu N, Norbeck JH, Mason MA, et al.. Recruitment and retention strategies for minority or poor clinical research participants: lessons from the Healthy Aging in Neighborhoods of Diversity Across the Life Span Study. Gerontologist. 2011;51:533–545.
54. Rudolph AE, Crawford ND, Latkin C, et al.. Individual, study, and neighborhood level characteristics associated with peer recruitment of young illicit drug users in New York City: optimizing respondent driven sampling. Soc Sci Med. 2011;73:1097–1104.
55. Gifford AL, Cunningham WE, Heslin KC, et al.; HIV Costs and Services Utilization Study Consortium. Participation in research and access to experimental treatments in HIV-infected patients. N Engl J Med. 2002;346:1373–1382.
56. Wise B, Cuffe S, Fischer T. Dual diagnosis and successful participation of adolescents in substance abuse treatment. J Subst Abuse Treat. 2001;21:161–165.
57. Pagey B, Deering D, Sellman D. Retention of adolescents with substance dependence and coexisting mental health disorders in outpatient alcohol and drug group therapy. Int J Ment Health Nurs. 2010;19:437–444.

HIV; psychiatric disorders; ADHD; antiretroviral treatment; retention

Supplemental Digital Content

© 2013 Lippincott Williams & Wilkins, Inc.