Capacity to consent to research among adolescent-parent dyads in Rakai, Uganda

Abstract

Objectives:

The capacity of adolescents–of different ages and in diverse settings–to comprehend risks, benefits, and other elements of informed consent is a critical but understudied area in research ethics. We assessed the cognitive capacity of early, middle, and late adolescents, and their parents or guardians to provide informed consent to a population-based cohort study.

Methods:

Eighty adolescent-parent/guardian dyads including 40 early (10–14 years), 20 middle (15–17 years), and 20 late (18–19 years) adolescents were recruited from the Rakai Community Cohort Study (RCCS), an open demographic cohort in Uganda. Participants were administered the MacArthur Competence Assessment Tool for Clinical Research (MacCAT-CR), a structured open-ended assessment; interviews were recorded and transcribed. Twenty transcripts were scored independently by two coders; the intraclass correlation coefficient (ICC) was 0.89. The remaining interviews were scored individually. We compared mean scores for early and middle/late adolescents using a one-sided t-test and score differences between parent/guardian and adolescent dyads using two-sided paired t-tests.

Results:

Early adolescents (mean score, 95% confidence interval (CI)) (28.8, 27.1–30.5) scored significantly lower (p<.01) than middle/late adolescents (32.4, 31.6–33.1). In paired dyad comparisons, we observed no statistically significant difference in scores between parents/guardians and middle/late adolescents (difference=−0.2, 95% CI=−1.0–0.6). We found a statistically significant difference in scores between parents/guardians and early adolescents (difference=3.0, 95% CI=1.2–4.8).

Conclusions:

Our findings support the practice of having middle and late adolescents provide independent informed consent for sexual and reproductive health studies. Early adolescents may benefit from supported decision-making approaches.

Background

Adolescence (10–19 years) is a critical period for health research and intervention, as the behaviors established during this period have life-long consequences (1,2). In addition, adolescents exhibit increased risk of adverse sexual and reproductive health (SRH) outcomes (1,3). Despite this increased vulnerability, adolescents are often excluded from biomedical and behavioral research, especially in low-and middle-income countries (LMICs) (4,5). Similarly, most studies assessing adolescents’ capacity to consent into research have been conducted in high income countries (HICs) (6). This is problematic as 90% of adolescents live in LMICs (7,8). In East and Southern Africa, research with adolescents is essential given their considerable risk of HIV infection (9).

Adolescence is a time of rapid physical, cognitive, and psychological change that begins in puberty and ends when the adolescent achieves adult status in society (10). During adolescence, wide differences are evident in cognition, social status, behaviors, and health risks. Despite this variability, there are biological, social, and cognitive similarities among adolescents by age. Adolescence is often conceptualized as three periods: early (10–14 years), middle (15–17 years), and late (18–19 years) (11,12). Current ethical practice involves obtaining both parental permission and adolescent assent for early and middle adolescents as the capacity of adolescents to independently provide informed consent is poorly understood.

Research on informed consent has conceptualized capacity to consent as understanding the research study, appreciating how it will affect one personally, reasoning and weighing risks/benefits, and making a voluntary choice (13). For research on sensitive topics like HIV, it is critical to understand the capacity of adolescents of different ages and in diverse settings to provide informed consent, as parental consent has been identified the biggest barrier to adolescent participation in HIV prevention research (14). A U.S. study of 15–22-year-old participants in a multi-site HIV prevention trial in the US found that only one of the 58 participants surveyed spoke with a parent or guardian prior to participation, and that individual was over 18 years (15). Overall, limited data are available on adolescents’ actual capacity to consent, particularly in LMICs (5,16). The MacArthur Competence Assessment Tool for Clinical Research (MacCAT-CR) (17) has been used in research to assess capacity to consent and has been adapted in several languages for use with adults in both HMIC and LMIC settings (18–21). It has the advantage of assessing all four domains of capacity, as well as being used in research with children and adolescents (21–24).

Few studies have assessed adolescent capacity to provide informed consent for research (21–24); and these studies have yielded similar results. In a hospital-based study of capacity to consent to research among early, middle, and late adolescents in the Netherlands, even early adolescents demonstrated adult-type capacity to consent (21). In a US community-based study, young people (14–21 years) had capacity scores that were similar to those of adults (24). Another US study adapted the MacCAT-CR for online use and similarly found adult-level capacity to consent to HIV pre-exposure prophylaxis research among middle-late adolescent men who have sex with men recruited through social media (22). Most of these studies used a simulated (22,24) rather than an actual consent process (21), all were conducted in HIC, and none assessed parents’ capacity to consent.

The current study, conducted in southcentral Uganda, compared the capacity of parents and adolescents of different ages to comprehend risks, benefits, and other elements of informed consent and therefore to provide informed consent. Specifically, we assessed and compared the cognitive capacity of dyads of early, middle, and late adolescents and their parents/guardians to provide informed consent for the Rakai Community Cohort Study (RCCS) using the MacCAT-CR tailored for adolescent research.

Methods

Study design and participants

The study was conducted in southcentral Uganda, one of the regions hardest hit by HIV/AIDS (25). In the region, the Rakai Health Sciences Program (RHSP) has focused on HIV intervention and observational studies using the RCCS - an open community-based epidemiological cohort study in southcentral Uganda. The current study was nested within the RCCS. Since 1994, the RCCS has surveyed adults and adolescents 15–49 years of age approximately every 18-months (25). Each RCCS survey round includes a household census, where the head of household provides information on all household members and on household assets. Eligible household members then complete interviewer-administered demographic and behavioral questionnaires and are offered HIV testing and counseling. Participants in the RCCS provide informed consent at each round. For unemancipated minors (generally under age 18), both written minor assent and parental/guardian permission are obtained; 18+ year-olds and emancipated minors provide their own informed consent. Currently, adolescents <15 years are not eligible to participate in the RCCS but are enumerated in the household census. This research was approved by IRBs at Columbia University and John Hopkins University, and the Research Ethics Committee at the Uganda Viral Research Institute. This study follows both US federal regulations (26) and Uganda National Council for Science and Technology (UNCST) regulations (27).

Participants for the present study were recruited as parent-adolescent dyads as a purposive sample from the RCCS during the time that the survey was being conducted. We aimed to enroll 40 early (age 10–14 years), 20 middle (age 15–17 years), and 20 late (18–19 years) adolescents, divided by gender within each age group, and their parents/guardians (N=80 dyads). Eligible parents/guardians and adolescents aged 15–19 years had participated in RCCS previously or were participating in the RCCS data collection at the time. For this study of adolescent capacity, research staff first contacted parents/guardians, invited them to participate and obtained permission to recruit the eligible adolescent in their household. Parents and older adolescents provided informed consent; parental permission and informed assent were obtained for early and middle adolescents. Because early adolescents do not participate in the RCCS, they were identified on household census forms and recruited through their consenting parents. Fieldwork was conducted between January 2020 and October 2021.

The MacArthur Competence Assessment Tool for Clinical Research (MacCAT-CR)

The MacCAT-CR consists of structured, open-ended questions to assess four areas of competence to consent: 1) Understanding of the purpose, procedures, and associated risks and benefits of participation; 2) Appreciation of one’s situation and the effects of participating or not; 3) Reasoning for a decision about participation based on expected consequences; and 4) Expressing a choice about research participation (28). Adaptations to the MacCAT-CR for use with adolescents have been described elsewhere (24). The published MacCAT-CR, originally designed for clinical trials, includes 13 items on Understanding, three for Appreciation, four for Reasoning, and one for Choice with total scores ranging from 0–42 (28,29). For the current study we tailored the MacCAT-CR to the RCCS, an epidemiologic study, and included 10 items to assess Understanding, 3 for Appreciation, 4 for Reasoning, and 1 for Choice, with possible total scores ranging from 0–36. We modified questions regarding understanding in the current study to assess participants’ understanding of the longitudinal nature of the study as well as the biological and sensitive sexual partnership information modules of the RCCS survey instrument. The interview guide was translated from English into Luganda to accommodate Luganda-only speaking participants. We then piloted the translated guide and further tailored questions and translated text.

MacCAT-CR interviews were conducted in Luganda by trained researchers from the Rakai Health Sciences Program’s Social and Behavioral Sciences (SBS) team. The interview was conducted after the participant completed the RCCS procedures, except for the early adolescents, who participated in a simulated RCCS consent procedure because they were not yet eligible for the RCCS.

To administer the MacCAT-CR, the interviewer read each section of the RCCS consent form and followed this disclosure with specific questions designed to assess the domains of competence (e.g., “What is the purpose of this study?”). Interviews were conducted in a private location and took one hour on average. Interviews were audio-recorded and then transcribed and translated into English by the SBS team.

To score MacCAT-CR responses, we created an a priori rubric based on the RCCS consent form. Using the verbatim transcripts that had been translated from Luganda to English, each of the 18 items was scored from 0 to 2: 2 points for a complete answer, 1 for a partially correct or vague answer, and 0 for an incorrect answer or non-response. To refine the rubric and ensure inter-rater reliability, the team scored the first 14 interviews as a group and discussed the results. Examples from those 14 interviews were included in the rubric for future scoring. Using the finalized rubric, two team members scored 20 additional interviews independently; intraclass correlation coefficient (ICC) was 0.89. The two scorers then individually scored the remainder of the transcripts and had weekly meetings with the entire team to resolve any questions or challenges.

Other Measures

Demographic data were obtained from three sources: an interview intake form prior to the MacCAT-CR questionnaire, the RCCS census, and the RCCS questionnaire. Age, gender, and schooling enrolment status were obtained from all participants prior to their completing the MacCAT-CR interview. Household SES, a composite measure derived using principal component analysis of nine household assets, was obtained from the household census of the current RCCS round (30). Additional demographic data were extracted from the RCCS questionnaire for participants 15 years and older and included highest educational level attended, ever married, currently married, and occupation. Early adolescents were ineligible to participate in the RCCS and were assigned SES based on the household census data.

Data Analysis

We examined participant characteristics by age group and used scatter plots, boxplots, and summary statistics to examine the distribution of total MacCAT-CR score and sub-scores by age group. To gain insight into challenging domains, we identified items where 50% or more of participants in an age group scored low (0 or 1). We selected representative quotes from the transcripts to illustrate difficult items.

Main analyses tested the hypotheses that (1) the mean MacCAT-CR score among middle/late adolescents would be higher than the mean score of early adolescents, using a one-sided two sample t-test, and (2) in paired analyses, there would be significant differences in MacCAT-CR scores between parents/guardians and early adolescents but not between parents/guardians and middle/late adolescents, using two-sided paired t-tests, separately among the early adolescent dyads and middle/late adolescent dyads. Analyses were conducted for total scores and for each sub-scale score separately, using non-parametric tests – Mann-Whitney for group differences and Wilcoxon signed rank for paired differences.

Based on literature which suggested that middle to late adolescents have similar cognitive capacity to adults (21–24), we combined the middle and late adolescents to increase statistical power. For the first hypothesis—that the mean MacCAT-CR score among middle/late adolescents would be higher than the mean score of early adolescents by at least 3 units—we used the following parameters to calculate sample size for a one-sided two-sample t-test assuming equal variances: a detectable difference of 3 which is the smallest difference in MacCAT-CR of clinical interest; a pooled standard deviation of 5; 80% power; and a 5% level of significance. The resulting sample size was 40 middle/late adolescents and 40 early adolescents. Next, we performed a power analysis to ensure that this sample size was sufficient for the second hypothesis—that there would be significant differences in MacCAT-CR scores between parents/guardians and early adolescents. With 40 dyads of early adolescents and their guardians, there was at least 95% power at a 5% level of significance to detect a mean difference of 3 units in a paired t-test assuming a standard deviation of 3.75 for the difference in MacCAT-CR scores for each dyad and a correlation of 0.5 of the early adolescent and guardian scores. Mean MacCAT-CR scores, standard deviations, and other values used in sample size calculation and power analysis were informed by prior work from Ott and colleagues (23).

In secondary analyses, we examined whether demographic characteristics predicted total score within age groups. Separate multivariable models were conducted for adolescents, and then for parents/guardians. Models for adolescents included age, gender, SES, and number of RCCS rounds their parent had participated in. Models for parent/guardians included age, gender, SES, number of RCCS rounds they had participated in, as well as education level.

Results

Participants

The mean age of early adolescents, middle/late adolescents, and parent/guardians in our sample was 12, 17, and 43 years, respectively (Table 1). Almost all early adolescents were currently in school, as were 58% of middle/late adolescents. Most parent/guardians (54%) had attended primary school, with some having attended secondary and post-secondary school. Middle/late adolescents had participated in an average of 1.1 RCCS rounds (minimum=1, maximum=2), while parent/guardians had participated in an average of 8.8 rounds (minimum=1, maximum=18).

Table 1.

Participant Characteristics by Age Group

	Age Group
	Adolescents 10–14	Adolescents 15–19	Parent/guardian
N	40	40	80
Age (Mean (SD))	12.43 (1.30)	17.30 (1.18)	42.45 (8.98)
Gender = Male (N (%))	20 (50.0)	20 (50.0)	25 (31.2)
SES (Mean (SD))	−0.07 (1.19)	0.15 (1.26)	0.04 (1.20)
SES Category (N (%))
Lowest	12 (30.0)	10 (25.0)	21 (26.2)
Low Middle	6 (15.0)	6 (15.0)	12 (15.0)
High Middle	5 (12.5)	9 (22.5)	15 (18.8)
Highest	13 (32.5)	15 (37.5)	27 (33.8)
Missing	4 (10.0)	0 (0.0)	5 (6.2)
Current Student Status Among Adolescents (N (%))
In School	39 (97.5)	23 (57.5)
Out of School	1 (2.5)	17 (42.5)
Highest Education Attended Among Parents (N (%))
None			1 (1.2)
Primary			42 (52.5)
Secondary			16 (20.0)
Post-Secondary			16 (20.0)
Missing			5 (6.2)
Average # Rounds in RCCS (Mean (SD))	0.0 (0.0)	1.1 (0.3)	8.8 (5.1)
Marriage Status (N (%))
Currently Married		2 (5.0)	52 (65.0)
Never Married		37 (92.5)	7 (8.8)
Previously Married		1 (2.5)	16 (20.0)
Missing		0 (0.0)	5 (6.2)
Primary Occupation (N (%))
Admin/Teaching		0 (0.0)	2 (2.5)
Agriculture/Housework		6 (15.0)	44 (55.0)
Business		7 (17.5)	19 (23.8)
Other		3 (7.5)	8 (10.0)
Student		23 (57.5)	2 (2.5)
Missing		1 (2.5)	5 (6.2)

MacCAT-CR scores and hypothesis testing

Early adolescents demonstrated a wider distribution of total MacCAT-CR scores compared to middle/late adolescents and parents/guardians (Figure 1 and Table 2). The mean (Median, SD) total MacCAT-CR score was 28.8 (30, 5.2) in early adolescents, 32.4 (32, 2.3) in middle/late adolescents, and 32.0 (32.8, 3.0) in parents/guardians (Table 2). Figure 1 demonstrates that while most early adolescents had scores similar to middle/late adolescents and parents, the early adolescent distribution had a longer tail, with a higher proportion of early adolescents scoring very low (15% or n=6 early adolescents scored below 25, whereas only 4% or n = 3 parents and no middle/late adolescents scored below 25).

Figure 1.

Possible total score range was 0 to 36.

Table 2. MacCAT-CR Scores and Score Differences by Age Group.

Score difference is calculated as (Parent Score) – (Adolescent Score). A positive score indicates that the parent scored higher than the adolescent. A negative score indicates that the adolescent scored higher than the parent.

	Mean Score (95% CI)				Mean Score Difference (95% CI)
Scale	Young Adolescents	Middle/Late Adolescents	Parents/Guardians	P-value, Young vs Middle/Late Adolescents	Parents/Guardians vs Young Adolescents	Parents/Guardians vs Middle/Late Adolescents
Total (Range 0–36)	28.8 (27.1, 30.5)	32.4 (31.6, 33.1)	32.0 (31.3, 32.6)	<0.01 1	3.0 (1.2, 4.8) 2	−0.2 (−1.0, 0.6)2
Understanding (Range 0–20)	16.6 (15.8, 17.4)	18.7 (18.3, 19.0)	18.2 (17.8, 18.6)	<0.01 3	1.0 (0.0, 2.0) 4	0.1 (−0.4, 0.6)4
Appreciation (Range 0–6)	4.4 (3.9, 4.8)	4.6 (4.3, 4.9)	4.6 (4.3, 4.8)	0.343	0.6 (0.1, 1.0) 4	−0.4 (−0.8, 0.0)4
Reasoning (Range 0–8)	5.9 (5.3, 6.6)	7.0 (6.6, 7.3)	7.2 (6.9, 7.5)	<0.01 3	1.3 (0.6, 2.0) 4	0.2 (−0.2, 0.6)4
Choice (Range 0–2)	1.9 (1.8, 2.0)	2.0 (2.0, 2.0)	2.0 (2.0, 2.0)	<0.01 3	0.1 (0.0, 0.2) 4	0.0 (−0.1, 0.0)4

Bold font indicates statistical significance at the 0.05 level

¹Two-sample t-test

²Paired t-test

³Mann-Whitney test

⁴Wilcoxon signed rank test

In a test of mean differences, middle/late adolescents had significantly higher total MacCAT-CR scores than early adolescents (difference = 3.6, p<0.01). In paired analyses, we observed no difference in total capacity scores for parents/guardians and middle/late adolescents (mean difference=−0.2, p=0.59, Table 2). However, there was a statistically significant difference in total capacity scores between parents/guardians and early adolescents (mean difference=3.0, p<0.01, Table 2). Statistical significance and conclusions remained the same when using nonparametric statistical tests (Mann-Whitney and Wilcoxon signed rank test).

Subscores for understanding, reasoning, and choice were significantly higher among middle/late adolescents compared to early adolescents, whereas the appreciation subscore was not meaningfully different between middle/late and early adolescents. We found no significant differences in subscores between parents/guardians and middle/late adolescents. All subscores were significantly different between parents/guardians and early adolescents.

Challenging questions

Certain items stood out as being especially challenging for one group or all groups (Figure 2 and Appendix). Within the domain of Understanding, the majority of late and middle adolescents demonstrated that they understood the purpose of the research (item U2), while over 70% of early adolescents did not. Interview data revealed that early adolescents tended not to communicate an understanding that the RCCS interview would include sensitive topics such as personal questions regarding HIV and reproductive health (U2). When asked about the risks of the study, more than 55% of early adolescents failed to demonstrate that they understood that participating in the research included the potential risk of loss of confidentiality and/or experiencing discomfort with sensitive topics (U6). In contrast, more than 75% of middle adolescents and nearly all late adolescents understood these potential risks of participating.

Figure 2.

U indicates understanding subscale, A indicates appreciation subscale, R indicates reasoning subscale, and C indicates choice subscale.

A low score is considered a 0 or 1. Questions that 50% or more of respondents in an age group scored low were considered challenging.

The MacCAT-CR question that asked participants to offer comparative statements regarding their choice to participate (R1) was also challenging for early adolescents. The majority (65%) made no comparative statements or stated a comparison but gave no differences (e.g., “It would be better if I participate”, but did not say why). In contrast, only 45% of middle and 35% of late adolescents and parent/guardians earned a low score on this question.

Questions that were difficult for all groups (Figure 2 and Appendix) included community benefit (U9), withdrawal from the study (A2), personal vs. societal benefit (A3) (particularly for middle adolescents), and effects on daily life (R3) (particularly for early adolescents).

In multivariable analyses examining the predictors of total scores, the only factor associated with MacCAT-CR score among adolescents was participant age, which showed a 0.63 unit increase in score for a one-year increase in age (p<0.01). Other characteristics including gender, SES, and parent rounds in the RCCS were non-significant. Among parents/guardians, there were no significant factors associated with total score.

Discussion

Adolescents, particularly adolescents in LMICs, have been underrepresented in biomedical and behavioral research (4,5). Ethical inclusion of adolescents in research - following the principles of justice, beneficence, and respect for persons - is essential if adolescents are to receive the full benefits from research (31,32). The challenges of conducting adolescent research are complex, but not unachievable. Establishing the capacity of adolescents to make informed consent is an important step in promoting the ethical inclusion of adolescents in research (33).

The findings of this study support the cognitive capacity of adolescents in Uganda to provide informed assent and consent for HIV and SRH studies. Notably, MacCAT-CR scores for middle and late adolescents were similar to those of their parents/guardians, supporting the practice of allowing adolescents aged 15 and older to provide their own informed consent or assent for SRH studies. Although early adolescents had a lower mean score, this hides wider variation in their scores and the fact that most early adolescents scored in a similar range as older adolescents and parents. These findings suggest that some early adolescents might benefit from increased decision-making support, such as an ombudsman or teach-back methods (34,35).

Our findings regarding middle and late adolescents’ capacity to consent are consistent with research in the US and the Netherlands (22–24). Our study has direct relevance for HIV treatment and prevention research, the concept of mature minor, and for minor consent laws in clinical care in both LMICs and HICs that allow adolescent independent access to care without parental permission. The notion of mature minor, developing from English common law, is that a minor is capable of giving informed consent when they achieve sufficient decision-making skills to enable them to understand what is proposed (36). Minor consent laws, while not the same as the idea of a mature minor, cover access to clinical care including HIV treatment and prevention, STD treatment, pregnancy prevention, mental healthcare (31,32,37). In the research realm, US regulations on research (45 CFR 46) (26) directly reference minor consent laws and draw on the mature minor doctrine in both the definition of children and the provision that allows an IRB to waive parent permission (31). The UNCST (27) also addresses mature minor, minor consent laws, age of majority, and emancipation status. The definition of children in the US regulations is “persons who have not attained the legal age for consent to treatments or procedures involved in the research, under the applicable law of the jurisdiction in which the research will be conducted.” This definition references US state laws on age of majority, mature minor status, age to consent for general medical care, emancipation, and minor consent for specific services, such as HIV and STI treatment (31). Requiring parent/guardian permission in research where adolescents are receiving healthcare independently under a minor consent law is highly problematic because it means that the minor adolescents in the study would have fewer protections (in the form of confidentiality) than those receiving clinical care (31,32,38–40).

Likewise, requiring parent/guardian permission in research may discourage disclosure of sensitive information on sexual initiation or HIV status and may discourage adolescent participation in research (38–43). Adolescent concerns about confidentiality are highly relevant in HIV and SRH research, where adolescents are asked to share personal and sensitive information and in which adolescents may be assented into research with their parents’ permission, potentially disclosing not just sexual behaviors, but gender identity and sexual orientation (22). A growing body of research has informed guidelines to ethically include adolescents in global health research (31,32,44). Our study aims to strengthen these guidelines through research to better understand adolescent decision-making capacity to consent into research.

Capacity with early adolescents is less well established and not as consistent. In a 1982 study, Weithorn found 14-year-olds did not differ from adults in competence to make informed treatment decisions (45). A children’s hospital-based study from the Netherlands found that children as young as 11 years demonstrated capacity to consent to clinical research; however, this group was likely familiar with the medical procedures in the studies (21). Our study found that early adolescents in a community-based sample generally understood the actual biomedical procedures and risks (i.e., finger prick for HIV-testing). Those with low scores often omitted mention of the interview procedures and potential risks such as discussing sensitive topics or loss of confidentiality. Clearly, additional research is needed to understand nuances of the capacity of early adolescents.

Strengths and Limitations

Strengths of our study include the recruitment of adolescent and parent/guardian dyads from a community-based sample in an LMIC. This allowed us to compare MacCAT-CR scores between each adolescent and their parent/guardian, effectively controlling for potentially confounding factors such as socio-economic status, religion, or other regional or cultural factors that may influence adolescent cognitive capacity to consent (21,24). In addition, the RCCS itself is one of the longest running biobehavioral studies in East and Southern Africa and findings may have direct relevance for future cohort inclusion criteria as well as national and regional policy.

One limitation of the study was that no early adolescents had prior experience participating in the RCCS, while nearly all middle and late adolescents and parent/guardian had previously participated in the RCCS. Therefore, the experience of participating may have given these groups an advantage in responding to the MacCAT-CR as compared to early adolescents, who had never participated in the RCCS. In a related interview focused on attitudes about bioethics, we found that adolescents became aware about RCCS from an early age, typically from an older family member, most often a parent or guardian. The RCCS also recruits in communities after community mobilization activities, such as village loudspeakers and, the RCCS individual visit is done after the household census visit. Thus, while early adolescents had not directly participated in the RCCS, early adolescents were likely aware of RCCS activities, but not necessarily all of the RCCS interview processes, risks and benefits (46). Finally, the MacCAT-CR was not designed to provide an absolute criterion for consent competence. Future research is needed to determine how MacCAT-CR scores align with a formal capacity assessment interview by a trained clinician.

Conclusion

Inclusion of adolescents and protection from research should not be viewed as opposing elements, but instead as compatible goals that can be achieved with attention to adolescent development, an understanding of research regulations and bioethics principles, and attention to the social context of an adolescent’s life (32). Researchers must ensure adolescent protection and minimize research-related risks, uphold their autonomy and right to participate in research, while avoiding confidentiality violations (16,21,32,47).

Our findings support the cognitive capacity of middle and late adolescents to provide independent informed consent for SRH and HIV studies in rural Uganda; lower scores for early adolescents suggest they could benefit from supported decision-making approaches.

List of abbreviations

CI

confidence interval

HIC

high income countries

HMIC

high & middle income countries

ICC

intraclass correlation coefficient

LMICs

low-and middle-income countries

MacCAT-CR

the MacArthur Competence Assessment Tool for Clinical Research

RCCS

Rakai Community Cohort Study

SBS

Social and Behavioral Sciences

SD

standard deviation

SRH

sexual and reproductive health

SES

socioeconomic status

UNCST

Uganda National Council for Science and Technology

Appendix

ITEM & QUESTION(S)	SCORE	QUOTE & EXPLANATION
Understanding U2: Interview on Sensitive Topics
What sorts of things will people have to do if they agree to be in the study? 0= Does not mention the interview/survey 1= Says interview, but unclear about topics 2= Says interview will be on sensitive topics, or lists these topics (e.g. HIV, reproductive health).	2	“When we enter in the tent, the health care worker begins interviewing you. She asks you that are you married? if you are married, then you tell her She asks you that do you have children and then you tell her. If you have three children, you tell her that I have three children. She asks you that how many sexual partners have you had for the past year. Still you tell her. From the year, then she asks you about a month and then week”. -44 year old female parent Explanation: Participant clearly stated that interview includes asking about sensitive topics
	1	“The questions being asked to the participant.” -12 year old male adolescent Explanation: Participant said interview but did not mention that it will be on sensitive topics
	0	“I do not know about those. I do not remember them.” -13 year old female adolescent Explanation: Participant is unable to answer the question
Understanding U6: Loss of Confidentiality & Sensitive Topics
What are some risks about potentially participating in this study? 0= Does not mention loss of confidentiality with survey or HIV test, nor mention discomfort or sensitive topics 1= Vague answer, cannot provide adequate detail 2= Mentions potential loss of confidentiality and/or discomfort with sensitive topics	2	“If the results are kept well they cannot be known to anyone however, if they are not kept well and people know them it may affect my life” -12 year old male adolescent Explanation: The participant mentions confidentiality and how a breach would affect their life
	1	“When the person decides not to share the results with other people” -14 year old male adolescent Explanation: Participant shows some understanding of privacy but answer is vague and they are unable to elaborate further
	0	“Because as a participant, you must have wanted to get involved and consented to participate in the project study and if you want to know your health status, you stay in touch with that project” -37 year old female parent Explanation: Participant believes that getting involved in the RCCS negates any discomfort or loss of confidentiality
Appreciation A2: Reasons for Recruitment
Do you believe you have been asked to participate in this study for your personal benefit or for the benefit of society or both? 0= Believes they were personally recruited for personal benefit only 1= vague answer, unsure of answer 2= Understands that they were invited because they met inclusion criteria, and that while they personally may benefit, the ultimate purpose is science OR – says both and connects their personal benefits to the larger research study	2	“Not for personal benefit. This is a program for the project [RHSP]. They want to see a way forward in research. I am expecting feedback on how research has reached. Like the blood sample drawn from us.” -45 year old male parent Explanation: Participant stated that they are recruited for research.
	1	“It is for both to benefit. Benefiting from the research. I can know my HIV status, and how I can protect myself from HIV” -19 year old male adolescent Explanation: Participant states both but only describes how they benefit from the research
	0	“For personal benefit. I get access to free medical care” -14 year old female adolescent Explanation: Participant believes they were recruited just to receive medical care and not for a research study
Appreciation A3: Personal Outcomes
Is it possible you may not benefit or even be harmed from this study? Why or how? 0= Does not know, vague answer, unsure 1= Mentions harm OR benefit 2= Is able to explain how they may or may not personally benefit, AND may even be harmed from participating in the research (e.g. receiving health care, loss of confidentiality, bruising from blood draw)	2	“Researchers are experienced and know what they are doing so I donť expect to get any harm. In addition, after the interview, I get my HIV test results. I might feel bad if someone else gets to know my personal information” -16 year old male adolescent Explanation: Participant stated knowing HIV status as a personal benefit, and loss of confidentiality as a potential harm.
	1	“No iťs impossible for you not to benefit first of all you will get to know your health status and to me that is very beneficial” -42 year old male parent Explanation: Only personal benefit to participate is mentioned. Participant did not explain any potential harm
	0	“No iťs impossible not to benefit” -10 year old female adolescent Explanation: Parenticipant gives a vague answer, beliving they will always benefit, but cannot elaborate further
Expression C1: Able to express a choice
Do you want to participate in the RCCS? Why or why not? Whose decision would it be to participate in the RCCS? (For adolescents) Is participating your choice, your parenťs choice, or both of your choices? 0= Does not state a choice to participate or not participate 1= Uncertain, state more than one choice, says it is someone else's decision 2= Clearly states a choice	2	“Musawo [Lugandan word for healthcare worker] I wanted to know my status whether I am sick or whether I am okay. Yes musawo, I made the decision by myself because at first, I thought about it until when I finally made the decision” -44 year old female parent Explanation: Participant clearly stated a choice
	2	“I decided for myself after my father and mother had allowed me because they would not take me without my consent. I had to decide for myself” -18 year old male adolescent Explanation: Participant describes how they came to a decision and also discussed it with their parents.
	1	“Uhhu…, it was all over sudden, I was just told that you are now going to participate in research. I was not present, and my mother registered my name, [during RCCS census exercise]” “I also decided to join RCCS. The decision for my participation involved both my mother and father”. -37 year old female guardian Explanation: Participant did not clearly express their own choice to participate. They mentioned that the decision also involved their parents
	0	No participants received a zero on this question
Reasoning R1: Comparative
For participants 15+ years old : You decided to participate in the RCCS. Why is participating in RCCS a better decision for you than not participating in RCCS? For partiicpants 10 – 14 years old : You said that you wanted/did not want to participate in the RCCS. Why is/is not participating in RCCS a better decision for you than not participating in RCCS? 0= Makes no comparative statements 1= States a comparison, but gives no differences 2= Offers at least 1 statement in the form of a comparison of two options, with the comparison stating at least one specific	2	“You know when I participate in the RCCS research, I can know my HIV status. I can know whether I have syphilis or not and know how I can be helped. However, if I have not participated in the RCCS research, I cannot know my status and I cannot be helped to get medicine” -47 year old female guardian Explanation: Participant clearly stated a comparative statement. Identifying why participating in RCCS is a better decision for her than not participating in RCCS
	1	“It is good because I get free treatment and I also learn a lot” -56 year old female parent Explanation: Participant states the positive. but does not give differences. Answer is also vague - “learn a lot”
	0	[Participant stays silent] “I find the question difficult” -10 year old female adolescent Explanation: Participant was unable to answer the question
Reasoning R3: Mapping to Daily Life
How might these risks and benefits affect your daily life? 0= Gives no reasonable everyday consequences, even with encouragement 1= States only 1 reasonable possible effect of participating on their everyday life, relationships, or social activities 2= States at least 2 reasonable possible effects on their everyday life, relationships, or social activities	2	“When they test me for HIV and results show that I am negative, I continue to protect myself against HIV… When I swell the vein or get bruises, I might experience pain in the hand which might cause failure to do daily activities very well like lifting heavy things.” -16 year old male adolescent
	1	“They help me to be healthy and not to worrya lot. You may have issues that are worrying you, but if you get counselling all your worries will disappear. When you are not worried your heart will not be affected. Every time you are worried,your heart will be affected. You may even find yourself losing weight. Personally I know that worry is very dangerous it can even lead to death -44 year old female parent
	0	[Participant stays silent] -13 year old male adolescent Explanation: Participant was unable to answer the question, even when asked multiple probes