Psychology students need the ability to create informed consent and debriefing forms that adequately state the purpose of studies to the participants in an ethical manner. In your capstone class (your
ORIGINAL ARTICLE—ETHICS Truth and Consequences: Ethics, Confidentiality, and Disclosure in Adolescent Longitudinal Prevention Research CHRISTINE LOTHEN-KLINE, M.P.H, C.H.E.S., DONNA E. HOWARD, Dr.P.H., ELLEN K. HAMBURGER, M.D., KEVIN D. WORRELL, M.D., AND BRADLEY O. BOEKELOO, Ph.D., M.S.Purpose:To examine data from a natural experiment resulting from a change made in the confidentiality agreement midway through a randomized, longitudinal, controlled trial to prevent or delay adolescent alcohol use. The investigators explored the impact of the change on adolescents’ rates of participation and affirmative responses to a question on suicidal thoughts.
Methods:Adolescents 12–17 years old were adminis- tered a question on suicidal thoughts as part of a confi- dential exit interview after a general health examination with their primary care provider. After administration to 263 adolescents, the exit interview was made condition- ally confidential with the remaining 181 adolescents. The revised consent form and protocol stipulated that re- searchers would reveal to appropriate professionals and parents any adolescent indicating suicidal thoughts.
Prevalence estimates for the suicidal thoughts question and study participation rates were computed for condi- tions both before and after the change.
Results:Fewer adolescents responded affirmatively to the suicidal thoughts question when they were recruited using the revised (1%) than the original (8%) consentform and protocol (p .001). The revised confidentiality agreement did not affect participation rates.
Conclusions:Adolescents who assent to participate in research studies may be less likely to disclose personal information regarding suicidal thoughts if they know that their disclosure may result in a break in confidenti- ality. Specific guidelines are needed for conditional and unconditional confidentiality agreements to study men- tal health in adolescent longitudinal prevention research.
©Society for Adolescent Medicine, 2003 KEY WORDS :
Adolescents Confidentiality Disclosure Ethics Longitudinal research Mental health Prevention Randomized controlled trial Risk behavior Suicide Prospective research is necessary to investigate the effectiveness of interventions aimed at preventing risky behavior in adolescents. Yet, investigators en- gaged in behavioral prevention research with ado- lescents face many challenges. The challenges relate to the nature of the issues under study, the regula- tions and guidelines directing research with minors, and particularly concerns with confidentiality and disclosure. From the Department of Public and Community Health, College of Health and Human Performance, University of Maryland, College Park, Maryland (C.L.-K., D.E.H., B.O.B.); School of Medicine and Health Sciences, Departments of Medicine and Pediatrics, George Washington University, Washington, DC (E.K.H.); and Mid-Atlantic Permanente Medical Group, Rockville, Maryland (K.D.W.).
Address correspondence to: Bradley O. Boekeloo, Ph.D., M.S., Department of Public and Community Health, University of Maryland, 2387 Valley Drive, HHP Building, College Park, MD 20742. E-mail:
[email protected] Manuscript accepted March 10, 2003.JOURNAL OF ADOLESCENT HEALTH 2003;33:385–394 © Society for Adolescent Medicine, 20031054-139X/03/$–see front matter Published by Elsevier Inc., 360 Park Avenue South, New York, NY 10010 doi:10.1016/S1054-139X(03)00184-8 Many adolescent health problems are related to sensitive, even illicit, behaviors such as underage drinking, illegal drug use, smoking, and sexual ac- tivity. Longitudinal survey research on these behav- iors requires that adolescents disclose their private thoughts, intimate social interactions, and personal practices. Owing to the sensitive nature of this infor- mation, adolescents may be concerned that it will be disclosed to others, particularly parents, and that they could suffer consequences from such disclosure [1–4].
To enhance the likelihood that youth will honestly disclose information about personal behaviors, a promise of confidentiality may be required. Confi- dentiality may, however, be hard for the researcher to guarantee. If, during the course of data collection, adolescents reveal information that suggests they may be in harm’s way, researchers may be ethically and legally obligated to disclose this information in their efforts to assist individual adolescents. Guidelines and Laws Directing Research Current guidelines concerning issues of confidential- ity, disclosure, and informed consent, as applied to research with minors, are incomplete. Title 45 Code of Federal Regulations Part 46 (The Code) is gener- ally viewed as the accepted standard for the protec- tion of human subjects [5]. Although the Code does have a separate section focused on research involv- ing children, it does not differentiate between ado- lescents and young children [5]. The Code does not provide any guidance on the circumstances under which a researcher has an obligation to report infor- mation that suggests an adolescent may be at risk for harm or intervene to reduce risk [5–9]. Even when professional guidelines stipulate more directly the responsibility of the researcher to take action if a child appears in jeopardy, no guidance is given as to what constitutes“jeopardy”or an appropriate inter- vention [6,10]. Furthermore, when the intervention research occurs within a clinical setting, additional issues are raised regarding the responsibility for mandatory reporting by the academic researchers and/or clinical staff.
Mandatory Reporting Requirements Much of the discussion about mandatory reporting of research subjects centers around the issue of child abuse. Although there are certain categories of pro- fessionals, such as teachers, psychologists, clergy,and physicians, who are legally mandated to report disclosures of abuse or suicidal intent [11], the re- sponsibility of the researcher in similar situations is more ambiguous. Steinberg et al [11] point out that although all states have mandatory child abuse re- porting laws, no state specifically designates re- searchers as being among those listed as mandatory reporters. Some states specify, however, that people who do not have appropriate training in identifying and reporting child abuse cannot be required to report [11].
Despite this, many Human Subject Protection Committees interpret the state laws as applying to researchers and require mandatory reporting [11]. It has been argued that researchers are morally, if not legally, obligated to report suspected child abuse [11–14]. Steinberg et al note that specifically naming researchers as mandated reporters is in the best interest of not only the child but also the researcher, in that inclusion as a mandated reporter also serves to protect them from civil or criminal liability for their“good faith”efforts [11]. Yet, not all researchers may have the training to be able to identify children in jeopardy [15].
Researchers themselves vary in their interpreta- tion of whether, when and how, to intervene with adolescents at potential risk for harm. For example, one survey study explored how childhood depres- sion researchers had prepared for and responded to children in community samples whose nonanony- mous scores on a self-report depression inventory indicated that they might be severely depressed or suicidal [16]. One-fourth of the 16 researchers indi- cated that they did not anticipate identifying se- verely depressed or suicidal children, nor did they make any plans as to how they would respond if they had. Nineteen percent had anticipated the risk, but also did not have a plan, whereas 25% antici- pated the risk, but chose to take no action. Of the five (31%) respondents who anticipated the risk and made plans to respond, all planned to intervene if an adolescent met the criteria for being severely de- pressed, and three planned to if an adolescent ex- pressed suicidal ideation. Three of the five did not specify in the consent form who would be contacted if the criteria for intervening with an adolescent were met.
Other concerns surrounding the issue of manda- tory reporting include whether or not the research tools provide sufficient information with which to intervene. Fisher cautions investigators to be wary of the potential to either overestimate or underestimate the diagnostic validity of assessment tools used in 386LOTHEN-KLINE ET AL JOURNAL OF ADOLESCENT HEALTH Vol. 33, No. 5 research [6,17,18]. Scott-Jones points out that this is particularly critical when the assessment instrument has not been normed on the study population [15].
Others suggest that only researchers in a quasi- clinical relationship with their study participants should be required to take action to protect potential victims, arguing that the clinical setting is necessary to allow for adequate information with which to make a“reasonable judgment”of potential harm [19]. Steinberg et al point out that not all types of research provide adequate information to make clin- ical judgments. They suggest that“research involv- ing only questionnaires or telephone surveys”will not typically result in adequate information on which to base a judgment regarding abuse [11]. It has been recommended that researchers include more comprehensive evaluations in their research to allow for adequate information with which to make a judgment regarding a child’s current level of risk [11,14,20].
Another important issue is whether or not man- datory reporting laws actually decrease adolescents’ health risks. Gans and Brindis suggest that manda- tory reporting undermines research on the preva- lence, nature and etiology of many adolescent health problems [21]. There is evidence that adolescents do not seek needed health care when they fear their parents will find out the nature of their visit [1–4,22].
Several authors have explored the impact of requir- ing parental consent on adolescents’decision to be tested for HIV [23–25]. Meehan et al [24] found that after mandatory parental consent for HIV testing was abolished in Connecticut in 1992, the number of adolescents obtaining an HIV test increased by 44% in the 12 months after the change, compared with the 12 months preceding the change.
Confidentiality and Disclosure Unconditional confidentiality can be defined as the circumstance under which there are no exceptions to protecting the confidentiality of information dis- closed by participants in a research study. Condi- tional confidentiality suggests, however, that certain conditions may necessitate a break in confidentiality agreements between researchers and study partici- pants. These confidentiality requirements may affect participation rates and the veracity of adolescents’ responses. In survey research conducted among ad- olescents, issues of confidentiality are often side- stepped by the use of anonymous self-report ques- tionnaires that allow trend data to be generated on adolescent risk behavior. For example, the YouthRisk Behavior Survey (YRBS) [26] uses an anony- mous, cross-sectional survey methodology. This ap- proach achieves unconditional confidentiality, po- tentially minimizing adolescent concerns regarding confidentiality, while relieving the investigator of any obligation to respond to any disclosures that may signal that a youth is in trouble or at personal risk. The anonymity of disclosure makes it impossi- ble for the investigator to intervene, regardless of the nature of the information revealed. To evaluate the effectiveness of intervention trials, however, it is often necessary to track changes in participant be- haviors over time, and anonymity cannot be guaran- teed.
In the present study, the implementation of a randomized, controlled, longitudinal, prevention trial (RCT) on adolescent alcohol use provided an opportunity for a natural experiment to examine how changes in confidentiality agreements affected adolescents’reporting of suicidal thoughts. Specifi- cally, three questions were explored: (a) Did changes in the consent form from unconditional to condi- tional confidentiality result in a change in reporting of suicidal thoughts? (b) Did changes in the consent form from unconditional to conditional confidential- ity result in a change in adolescent participation rates? (c) How many adolescents continued to have suicidal thoughts that were unaddressed by parents and providers after the researchers implemented the unconditionally confidential follow-up telephone call? Methods Study Design and Sample The study used baseline data gathered from a 1-year longitudinal, randomized controlled effectiveness trial of a brief office-based intervention to prevent or delay adolescent alcohol use. All adolescents be- tween the ages of 12 and 17 years, who received a general health examination during the study period with any one of the 27 participating providers at five pediatric primary care practices in the Washington, DC metropolitan area, were eligible to participate in the study.
Survey Procedures and Measures Parents of eligible participants were sent a letter, signed by their primary care providers, encouraging participation in the study. Written informed consent from parents and assent from adolescents were ob- November 2003 ETHICS AND ADOLESCENT PREVENTION RESEARCH387 tained separately by a research assistant on the day of the adolescent’s primary care visit. At the time of consent, parents and adolescents were given a list of community substance use and other support services for teenagers. All parents were given a booklet on underage drinking. All of the adolescents completed interviewer-assisted surveys at the clinic before (pre- visit) and after (exit) their visits with their primary care providers. The interviewer did not see adoles- cents’responses to the survey questions, as adoles- cents responded to questions by circling an answer on an answer form. This baseline data collection was conducted in five primary care sites during the year 2000. The study was approved by Institutional Re- view Boards (IRBs) of two medical institutions with sites participating in the study and the principal investigators’academic research institution.
The exit survey included a battery of psychosocial measures, including a 10-question mood inventory.
The mood inventory was included to explore the relationship between adolescent risk-taking behavior and mood. One of the questions in the mood inven- tory (M16) addressed suicidal thoughts. The ques- tion read as follows:For the last 2 weeks, have you had thoughts that you would be better off dead or of hurting yourself in some way nearly everyday?The 10-question mood inventory, which was based on the Prime MD Mood Module [27], was selected because it was short and readily adapted to an assisted survey format as well as a telephone survey format that was the method of follow-up data collection.
Just before the visit with the provider, a research assistant administered one of three study protocols, depending on the random assignment of each ado- lescent. Adolescents in the usual care control group (Group I) listened to a Walkman radio for 15 min.
Intervention Groups II and III listened to a 15-min Walkman audio program on alcohol risk prevention, on which they were asked to respond to nine ques- tions assessing their potential risk in relationship to alcohol. For Group III only, youth responses were shared with their providers as a method of physician prompting for the visit. More information about theintervention and study design can be found else- where [28].
During the initial period of study data collection, and as the initial data were being entered into an electronic database, it was noted that 9% of adoles- cents answered“yes”to a question (M16) in the mood inventory suggesting suicidal thoughts. The research team consulted with the project co-investi- gators, both pediatricians at the two medical institu- tions with sites participating in the study, and it was agreed that the IRB-approved general protocol for adolescents at high risk for harm should be imple- mented. Also, the large number of adolescents re- sponding“yes”to this question indicated that this issue could not be treated as an infrequent isolated incident to be handled on a case-by-case basis based on the IRB-approved general protocol, but required special attention and a detailed, standardized proto- col for follow-up. A detailed telephone protocol based on the IRB-approved general protocol was developed, which specified actions to be taken for youth who responded“yes”to M16 (Table 1). Con- current with implementation of the detailed tele- phone protocol, the telephone protocol along with the most current prevalence data for affirmative responses to question M16 were sent to the three IRBs overseeing the study. Two of the IRBs approved the telephone protocol. The third IRB determined that the telephone protocol was insufficient and accordingly suspended the study at their sites, effec- tively suspending the entire study, until an accept- able protocol was put into place.
The third IRB committee suspended the study because it could not come to consensus as to what follow-up was warranted. The administrator and lawyer for this IRB committee made it clear to the principal investigator, however, that a protocol that included parental notification was necessary. After extensive consultation with national experts and review of existing guidelines and literature [5,7,29], a new follow-up protocol was developed. Ultimately, it was determined that most of the mental health experts supported breaking confidentiality for ado- Table 1.Original Actions Taken to Address Adolescents’Suicidal Thoughts Reported on Questionnaire Item M16 Actions Taken By Researchers With Each Adolescent Who Reported Suicidal Thoughts 1. The researchers telephoned each adolescent and reread question M16 to see if the adolescent still had suicidal thoughts.
2. Encouraged all adolescents to grant permission for researchers to contact his or her parent and provider.
3. Contacted the parent and provider of those who granted permission.
4. Provided information about a 24-hour suicide hotline and how to access mental health services within the adolescent’s health plan.
5. Strongly encouraged adolescents to talk to their parent and/or provider. 388LOTHEN-KLINE ET AL JOURNAL OF ADOLESCENT HEALTH Vol. 33, No. 5 lescents who had responded“yes”to question M16.
Therefore, the new protocol stipulated that a consult- ing psychologist would follow up by phone with all adolescents who had responded“yes”to question M16. The psychologist would assess the adolescents’ current suicidal risk (i.e., thoughts, plan, access, and intent) and inform parents and providers of the adolescents’suicidal thoughts (Table 2). The new protocol also required that the consent form for future participants be rewritten to include a state- ment in the confidentiality clause stating that if the adolescent revealed suicidal thoughts, then his or her provider and parents would be informed. The re- vised consent form read as follows: This study is not about sexual abuse and researchers will not ask about sexual abuse. Nevertheless, if an adolescent tells them about sexual or other abuse, researchers are required by law to report it to the appropriate officials to protect the adolescent. The privacy of the information will be strictly maintained at all times.If teens tell research- ers they are being abused or are suicidal, however, researchers must tell appropriate authorities so that help can be provided. Researchers must tell health authorities and parents if they learn an adolescent is suicidal(the information added to the revised con- sent form is in bold text).
The 10-question mood inventory was moved to the previsit survey. After asking the suicidal ideation question (M16) during the previsit survey, the re- searcher stated:“If you answered‘yes’to question M16, please tell me now so that you and I can talk to your doctor.”After the adolescent completed the survey, the researcher then reviewed the adoles- cent’s answers and notified the provider, before the health examination, if an adolescent responded“yes” to question M16. This enabled the provider to assess the adolescent and take appropriate action, if needed, during the scheduled visit. After a period of months, these protocol changes were approved by allthree IRBs, the study was reinstated, and the ap- proved protocols were implemented.
Data Analysis Prevalence estimates for the suicidal thoughts ques- tion (M16) and study participation rates were com- puted for conditions both before and after the con- sent form was changed. Both Fisher’s Exact Test and Pearson Chi-square analysis were performed to eval- uate differences between conditional and uncondi- tional confidentiality. All statistical analyses were performed using SPSS statistical software. Results Demographic characteristics of participants in the randomized controlled trial overall and by wave are represented in Table 3. The sample was comprised of slightly more females than males. Roughly 60% of the participants were between 14 and 17 years of age, and approximately 80% were African-American. A higher percentage of participants in Wave #2 of the study were African-Americans, 85.6% in Wave #2 versus 75.3% in Wave #1 were African-American (p .008). Providers at one study site no longer wanted to participate after the IRB suspended the project.
Study sites were represented differently in Wave #1 and Wave #2 (p .000), with most participants coming from just two sites in Wave #2.
The percentage of adolescents who responded “yes”to question M16 under circumstances of un- conditional (Wave #1) and conditional (Wave #2) consent regarding suicidal thoughts are presented in Table 4. Compared with circumstances of uncondi- tional confidentiality, significantly fewer adolescents responded“yes”to question M16 when the consent form explicitly stated that this response would ne- Table 2.Follow-up Actions Taken, as Approved by IRBs, to Address Adolescents’Suicidal Thoughts Reported on Questionnaire Item M16 Follow-up Action Taken By a Certified Clinical Psychologist With Each Adolescent Who Reported Suicidal Thoughts 1. A certified clinical psychologist telephoned adolescents, assessed their current suicidal risk (i.e., thoughts, plan, access, and intent), and told them that their parent/guardian and provider would be informed of their answer to M16.
2. The clinical psychologist informed the parent and provider about the adolescent’s response to M16 and made suggestions for follow-up.
3. A consent clarification letter indicating the change from unconditional to conditional confidentiality was sent to each family that consented using the original consent form before more follow-up data were collected from adolescents.
4. Question M16 was moved from the exit survey to the previsit survey for future participants and reviewed before the adolescent’s general health check-ups so that immediate follow-up with the provider could be arranged.
5. The consent agreement for future participants was changed to indicate conditional confidentiality regarding suicidal thoughts. November 2003 ETHICS AND ADOLESCENT PREVENTION RESEARCH389 cessitate breaking the confidentiality agreement. As all but two affirmative responses to question M16 were among African-American adolescents and the proportion of African-Americans increased from Wave #1 to Wave #2, the decrease in affirmative responses between waves was not owing to ethnic differences between waves. An affirmative response to M16 was not associated with site during Wave #1 (p .371). The two sites contributing 99% of the study participants during Wave #2 had a total of 10 (9.5%) affirmative responses to M16 during Wave #1 and 2 (1.1%) affirmative responses during Wave #2.
This drop in the number of M16 cases between Waves was significant (p .0000). Hence, the reduc- tion in affirmative responses between Waves was not owing to some sites with high rates of affirmative response during Wave #1 not being represented in Wave #2.
Examination of the participation rates under cir- cumstances of unconditional and conditional confi- dentiality, that is, before and after the consent form was revised to state that confidentiality would be broken if an adolescent responded“yes”to question M16, revealed nonsignificant differences (Table 4).Under circumstances of unconditional confidential- ity the participation rate was 47.6%, compared with 53.4% for conditional confidentiality.
Results of the initial follow-up contact by the research team, with the 21 youth who had responded “yes”to the question about suicidal thoughts (M16), are presented in Table 5. One of the adolescents had moved away from home and the parent would not provide the researchers with follow-up contact infor- mation. Seven of the adolescents, after being reread M16, emphatically stated that they no longer had suicidal thoughts and three of these said their“yes” answer on the questionnaire was in error. Two of the adolescents were evasive about whether they actu- ally had suicidal thoughts. All adolescents were asked directly whether they intended to take action on suicidal thoughts and none of the adolescents indicated that they did. Four said they had told their parent about their thoughts and two others said their parents knew about their thoughts. All adolescents were asked if they would grant permission for their parents and providers to be informed. Verbal con- sent for the researcher to notify parents and health care providers was granted by two and three youths, Table 3.Participant Characteristics Total Participants N (%)Wave pValue #1 Unconditional Confidentiality N (%) #2 Conditional Confidentiality N (%) Gender Male 199 (44.8) 112 (42.6) 87 (48.1) Female 245 (55.2) 151 (57.4) 94 (51.9) .254 Ethnicity African-American 353 (79.5) 198 (75.3) 155 (85.6) Non–African-American a 91 (20.5) 65 (24.7) 26 (14.4) .008 Age (yrs) 12–13 169 (38.1) 95 (36.1) 74 (40.9) 14–17 275 (61.9) 168 (63.9) 107 (59.1) .310 Site A 1 6 (1.4) 6 (2.3) 0 (0.0) B 2 235 (52.9) 87 (33.1) 148 (81.8) C 3 88 (19.8) 87 (33.1) 1 (0.6) D 4 65 (14.6) 65 (24.7) 0 (0.0) E 5 50 (11.3) 18 (6.8) 32 (17.7) .000 aNon–African-American includes Asian, Hispanic, Native American, Caucasian/White and other. Table 4.Percent of Adolescents Participating in the Study and Reporting Suicidal Thoughts by Type of Confidentiality in the Consent Form Confidentiality AgreementEligibles NParticipants N (%)pValue a Suicide Thoughts N (%)pValue b Unconditional 553 263/553 (47.6) .091 21/263 (8.0) .001 Conditional 339 181/339 (53.4) 2/181 (1.1) aPearson Chi-Square.bFisher’s Exact Test. 390LOTHEN-KLINE ET AL JOURNAL OF ADOLESCENT HEALTH Vol. 33, No. 5 respectively, during this initial follow-up. The re- searchers informed the two parents and three pro- viders of the adolescents’answers to M16.
After IRB approval of a new protocol, the consult- ing clinical psychologist followed up with the re- maining 19 adolescents for whom both parents and providers were not contacted by the researcher at initial contact (Table 5). Of these, two had persisting suicidal thoughts and both stated that they did not intend to act on these thoughts. One of the 19 adolescents said she had told her provider about the suicidal thoughts, eight said they had shared their suicidal thoughts with parents. One adolescent ver- bally expressed resistance to parent and provider notification of her suicidal thoughts but the psychol- ogist was unable to obtain any information to pre- vent contacting the parents and providers about the thoughts. One teen had turned 18 since the initial contact with the researcher and did not have suicidal thoughts when the psychologist conducted the fol- low-up assessment. For this teen, only the provider was notified, according to the adolescent’s wishes.
Hence, the psychologist informed all 18 providers and 18 of the remaining 19 parents who had not previously been informed during the initial contact by the researcher.
In summary, 21 adolescents reported suicidal thoughts on a questionnaire. After researchers con- ducted follow-up telephone calls to adolescents, the researchers could not rule out persisting suicidal thoughts for 14 adolescents based on their ability to contact and communicate with the adolescents, and they received adolescents’permission to notify two parents and three providers. At the second follow-up telephone call by the psychologist, most of the ado- lescents denied persisting suicidal thoughts. The two adolescents that reported continued suicidal thoughts had not told a parent or provider about their thoughts. Discussion This longitudinal alcohol prevention research project allowed for a natural experiment to study the impact on adolescents’reports of suicidal thoughts when the confidentiality agreement regarding such thoughts was changed midway through the project. Results suggested that conditional confidentiality agree- ments appear to have a censuring effect on adoles- cent reporting of suicidal thoughts. After being in- formed that their parent and provider would be notified if they reported suicidal thoughts, adoles- cents may have reconsidered whether their thoughts were serious enough to warrant answering the ques- tion affirmatively. Alternatively, adolescents who would have responded affirmatively may have cho- sen not to because they did not want their response to be disclosed to their provider and/or their parent.
Assurances of unconditional confidentiality from physicians have been shown to increase the number of adolescents willing to disclose sensitive health information about sexuality, substance abuse, and mental health, and increase the number willing to seek future health care when compared with assur- ances of conditional confidentiality [22,30,31]. Ado- lescents given confidentiality assurances from pro- viders also reported increased honesty of reporting and likelihood of future visits [30,31]. Overall, this study and the literature seem to support the premise that disclosure of sensitive information increases when confidentiality is emphasized [30–33].
This situation has potential implications for the individual adolescent as well as for research on sensitive health topics. Adolescents who may have suicidal thoughts, and may even want assistance from a responsible adult, may not report their sui- cidal thoughts under conditions of conditional con- fidentiality because they do not want their parent to be told. Adolescents who do disclose their suicidal Table 5.Summary of Telephone Contacts With Adolescents Who Reported Suicidal Thoughts on a Questionnaire Researcher ObservationTelephone Follow-up Time 1 (n 21) a Time 2 (n 19) b Adolescent denied persisting suicidal thoughts 7 17 Adolescent stated intent to take action on suicidal thoughts 0 0 Adolescent stated their parent/guardian knew about their suicidal thoughts 6 8 Adolescent stated they told their provider about their suicidal thoughts Not asked 1 aOne of the 21 adolescents had moved away from home and the parent would not provide contact information. The researchers had not located this adolescent before the study was suspended and, therefore, contact was only made at Time 2.
bTwo adolescents contacted at Time 1 were not re-contacted at Time 2 because they gave permission for their parents and providers to be informed at Time 1. November 2003 ETHICS AND ADOLESCENT PREVENTION RESEARCH391 thoughts under condition of conditional confidenti- ality are probably different from adolescents who choose to withhold under this condition. If adoles- cents are not responding honestly to this question on suicidal thoughts (M16), because of concerns that their provider and/or parent will be informed of their response, then the information collected on this particular issue will not be valid, affecting research conclusions and recommendations, and ultimately jeopardizing future research. At the very least, these findings suggest that researchers reporting studies on adolescent risk behavior need to disclose whether they used conditional or unconditional confidential- ity, to aid others with the interpretation of their results.
The impact of a change in the consent form from unconditional to conditional confidentiality, on the combined parent-adolescent study participation rates, was also examined. No significant difference was found. Provisions for breaching confidentiality agreements, when specifically defined in the consent form, did not appear to affect the combined parent- adolescent willingness to participate in the study.
These findings appear to contradict a previous study that explored parent-youth concordance on severity ratings for 14 risks and on the effect of four reporting options (taking no action, providing a referral to the teen, telling a parent, or informing a school coun- selor) on their decisions to participate in adolescent risk research [34]. Although there was high concor- dance on ratings of severity, there was less agree- ment on consent decisions. Over 90% of parents indicated that they would consent to allow their teen to participate in a research study if parental notifica- tion was required for reports of suicidal thoughts and child abuse. In contrast, only two-thirds of youth stated they would consent under those conditions.
Roughly 70% of youth said they would participate in a study that provided referrals to available resources for teens who reported suicidal thoughts or abuse, whereas less than half of parents agreed to such consent conditions.
This unplanned research situation also allowed for an exploration of how adolescents who reported suicidal thoughts responded to researchers who con- tacted them about their disclosure. Even among adolescents who confirmed a history of persistent suicidal thoughts, most adolescents tried to convince the researchers that no follow-up with parents and providers was needed. At the second telephone follow-up with a clinical psychologist a couple of months later, it was learned that many of these adolescents had not gone on to tell a parent and mostdenied continued suicidal thoughts. These findings suggest that most adolescents resisted a break in their confidentiality that is counter to what would be expected based on the Fisher et al study [6]. In that study, when hypothetical scenarios were presented to urban adolescents who were asked to rate whether they felt confidentiality should be broken, most ad- olescents felt that suicidal risk warranted a break in confidentiality.
There are several explanations for these discor- dant findings. First, the M16 question, asked as part of the mood inventory, may not reliably predict suicidality but a mood state less threatening to adolescents. Although youth may report recurring thoughts of suicide, they may not perceive these as serious symptoms of emotional instability or crisis, but part of the normal mood swings that accompany their adolescent years. Second, responses to hypo- thetical situations may not accurately represent how youth would respond when they are the ones having suicidal thoughts. Thirdly, adolescents actually expe- riencing suicidal thoughts may feel embarrassed or concerned about the burden they would place on their providers and parents. Additionally, these youth represented a subset who self-reported sui- cidal thoughts; thus they did not represent a general population of youth.
The decision to break confidentiality was a diffi- cult one for the researchers. One of the most pressing concerns was that the adolescents chose to partici- pate in the study based on a“good faith”agreement with the research team. This agreement did not state that adolescents’responses would be reported if they answered“yes”to the question on suicidal thoughts (M16). The researchers were also concerned about the validity of the instrument in assessing suicidal risk. The question (M16) was part of a larger, more comprehensive instrument designed to screen for mood disorders, not specifically suicidal risk. In addition, the instrument had not been normed on adolescents and was being used on an exploratory basis to determine if mood was altered in relation- ship to changes in the frequency of alcohol use.
Project researchers felt that a breach of confidential- ity was only warranted in cases of clear danger to the adolescent.
The researchers were also concerned that by inter- vening and disclosing the responses to question M16 to parents, they may inadvertently increase rather than decrease the adolescent’s risk of harm. A thor- ough psychological and home assessment was thought to be beyond the researchers’purview. Dur- ing the initial follow-up, some of the youth did 392LOTHEN-KLINE ET AL JOURNAL OF ADOLESCENT HEALTH Vol. 33, No. 5 indicate that it was their home situation that had triggered their emotional stress. Researchers were concerned about the potential contexts in which confidentiality could be broken, as well as who was the person best equipped to evaluate intended and unintended consequences of disclosure and to con- vey the information to providers and parents.
A final concern was related to maintaining study integrity. The breach of adolescents’confidentiality, and the subsequent follow-up with providers and parents, constituted an additional, unintended inter- vention effect. Secondary intervention effects and the extent to which they might compromise or confound the examination of planned program effects from the original RCT was a serious consideration.
The ultimate decision to break confidentiality was made after consultation with numerous mental health researchers and clinicians. The prevailing sen- timent was that the risk of not breaking confidenti- ality was greater than the risk of breaking confiden- tiality.
Implications The results of this study highlight several important considerations when conducting longitudinal pre- vention research on sensitive health topics with adolescents. It is critical to have a well-informed process when selecting questions on adolescents’ mental and emotional states for inclusion in a sur- vey. This is particularly so if these questions are not central to the study aims, as their inclusion may require the researchers to intervene with adolescents who select certain responses and this intervention may affect the validity of the central study. Although mood is a relevant domain for prevention research, prevention researchers must be sensitized to poten- tial expectations regarding follow-up with adoles- cents who indicate potentially life-threatening mood disorders. Researchers must carefully weigh the risks and benefits to adolescent health and the study design when including mood measurements, partic- ularly if conditional confidentiality is required. Fac- tors to consider include: how to optimize full disclo- sure by adolescents; what specifically, if anything, constitutes“imminent harm”requiring intervention by the researchers; what would a harm reduction protocol consist of for adolescents requiring inter- vention; and how would this protocol, as covered in the confidentiality agreement, affect adolescent dis- closure? Adolescent mental health experts should be involved, beginning with the study development, in this deliberation to clarify the issues and to helpformulate a clear plan for how researchers will respond if adolescents do report emotional distress.
This plan must be clearly stated in the consent forms for participants and their parents.
The results of this study suggest that many ado- lescents will not report suicidal thoughts when told that the researchers will share information with their providers and parents. Given that researchers may not be able to guarantee unconditional confidential- ity to adolescents in relationship to issues such as suicidal thoughts, longitudinal prevention research on suicidal thoughts may not be feasible unless the information has already been shared, such as in cases where adolescents are already in ongoing clinical care. Also, prevention research that attempts to in- clude mood indicators may only be feasible if ques- tions related to severe mood disorders such as sui- cidal thoughts are avoided.
Finally, this case illustrates that risk-benefit anal- yses currently used by IRBs are subjective. Although all three of the IRBs overseeing this research project reviewed the same initial protocol for adolescents who responded“yes”to M16, the three IRBs did not agree on how to proceed. Summary Adolescents who assent to participate in research studies may be less likely to self-disclose personal information that may suggest mental distress if they know such disclosures will necessitate a breach in their confidentiality agreement. Unless researchers can guarantee adolescents full privacy for any self- disclosures, they may be unable to answer important research questions regarding adolescents’mental health. These issues underscore the broader need for guidelines that more specifically address the ambi- guities surrounding the issue of imminent harm as it relates to adolescent preventive health research.
The authors thank Judith Jerry, Kathleen O’Connor, and Wilhel- mena Lee-Ougo for their assistance in this research. This research was supported by a grant (Grant #: RO1AA1225701) from the National Institute on Alcohol Abuse and Alcoholism (Bradley Boekeloo, Principal Investigator).
References 1. Reddy DM, Fleming R, Swain C. Effect of mandatory parental notification on adolescent girls’use of sexual health care services. JAMA 2002;288:710–4.
2. Klein J, Wilson K, McNulty M, et al. Access to medical care for adolescents: Results from the 1997 Commonwealth Fund Survey of the Health of Adolescent Girls. J Adolesc Health 1999;25:120–30. November 2003 ETHICS AND ADOLESCENT PREVENTION RESEARCH393 3. Ford CA, Millstein SG, Halpern-Felsher BL, Irwin CE, Jr.
Influence of physician confidentiality assurances on adoles- cents’willingness to disclose information and seek future health care. JAMA 1997;278:1029–34.
4. Cheng TL, Savageau JA, Sattler AL, DeWitt TG. Confidential- ity in health care. JAMA 1993;269:1404–7.
5. Department of Health and Human Services, National Insti- tutes of Health. Protection of Human Subjects. 45 Code of Federal Regulations (CFR) Part 46 1991.
6. Fisher CB, Higgins-D’Alessandro A, Rau JB, et al. Referring and reporting research participants at risk: Views from urban adolescents. Child Dev 1996;67:2086–100.
7. Santelli J, Roseneld WS, DuRant RH, et al. Guidelines for adolescent health research. J Adolesc Health 1995;17:264–9.
8. Scarr S. Ethical problems in research on risky behaviors and risky populations. Ethics Behav 1994;4:147–55.
9. American Psychological Association (APA). APA’s ethical principles of psychologists and code of conduct. Am Psychol 1992;47:597–611.
10. Society for Research in Child Development (SRCD). Ethical Standards for Research with Children. Directory of Research in Child Development 1990–1991. Available at: http://www.
srcd.org/about.html. Accessed June 2, 2003.
11. Steinberg AM, Pynoos RS, Goenjian AK, et al. Are researchers bound by child abuse reporting laws? Child Abuse Negl 1999;23:771–7.
12. Hoagwood K. The certificate of confidentiality at the National Institute of Mental Health. Discretionary considerations in its applicability in research on child and adolescent mental dis- orders. Ethics Behav 1994;4:123–31.
13. Melton GB. Ethical and legal issues in research and interven- tion. J Adolesc Health Care 1989;10:36S–44S.
14. Kinard EM. Ethical issues in research with abused children.
Child Abuse Negl 1985;9:301–11.
15. Scott-Jones D. Ethical issues in reporting and referring in research with low-income minority children. Ethics Behav 1994;4:97–108.
16. Burbach DJ, Farha JG, Thorpe JS. Assessing depression in community samples of children using self-report inventories:
Ethical considerations. J Abnorm Child Psychol 1986;14:579– 89.
17. Fisher CB. Reporting and referring research participants:
Ethical challenges for investigators studying children and youth. Ethics Behav 1994;4:87–95.
18. Fisher CB. Integrating science and ethics in research with high-risk children and youth. Soc Policy Rep 1993;7:1–27.
19. Appelbaum PS, Rosenbaum A. Tarasoff and the researcher:
Does the duty to protect apply in the research setting? Am Psychol 1989;44:885–94.20. Urquiza AJ. Retrospective methodology in family violence research: Our duty to report past abuse. J Interpers Violence 1991;6:119–26.
21. Gans JE, Brindis CD. Choice of research setting in understand- ing adolescent health problems. J Adolesc Health 1995;17:306– 13.
22. Boekeloo BO, Schamus LA, Cheng TL, Simmens SJ. Young adolescents’comfort with discussion about sexual problems with their physician. Arch Pediatr Adolesc Med 1996;150:
1146–52.
23. Jackson S, Hafemeister TL. Impact of parental consent and notification policies on the decisions of adolescents to be tested for HIV. J Adolesc Health 2001;29:81–93.
24. Meehan TM, Hansen H, Klein WC. The impact of parental consent on the HIV testing of minors. Am J Public Health 1997;87:1338–41.
25. Adams W. But do you have to tell my parents? The dilemma for minors seeking HIV-testing and treatment. John Marshall Law Rev 1994;27:493–512.
26. Grunbaum JA, Kann L, Kinchen SA, et al. Youth Risk Behavior Surveillance—United States 2001. MMWR 2002;51(SS04):1–64.
27. Spitzer RL, Williams JB, Kroenke K, et al. PRIME-MD— Primary Care Evaluation of Mental Disorders: Clinical Evalu- ation Guide. Roerig and Pratt Pharmaceuticals, Division of Pfizer Inc., 1993.
28. Boekeloo BO, Bobbin MB, Lee WI, et al. Effect of patient priming and primary care provider prompting on adolescent- provider communication about alcohol. Arch Pediatr Adolesc Med 2003;157:433–9..
29. Pearson JL, Stanley B, King C, Fisher C. Issues to consider in intervention research with persons at high risk for suicidality.
National Institutes of Health. Available at: http://www.nimh.
nih.gov/research/highrisksuicide.cfm. Accessed February 16, 2003.
30. Ford CA, Halpern-Felsher BL, Irwin CE Jr, Millstein SG.
Assuring confidentiality: Its effect on adolescent patients’ beliefs, expectations, and intentions. J Adolesc Health 1997;20:
137.
31. Ford CA, Millstein SG. Delivery of confidentiality assurances to adolescents by primary care physicians. J Adolesc Health 1996;18:112.
32. Ginsburg KR, Slap GB, Cnaan A, et al. Adolescents’percep- tions of factors affecting their decisions to seek health care.
JAMA 1995;273:1913–8.
33. Kobocow B, McGuire JM, Blau BI. The influence of confiden- tiality conditions on self-disclosure of early adolescents. Prof Psychol Res Pr 1983;14:435–43.
34. O’Sullivan C, Fisher CB. The effect of confidentiality and reporting procedures on parent-child agreement to participate in adolescent risk research. Appl Dev Sci 1997;1:187–99. 394LOTHEN-KLINE ET AL JOURNAL OF ADOLESCENT HEALTH Vol. 33, No. 5