Ethics statement

Ethical approval was obtained from the Yale University (#2000028354) and the American Samoa Department of Health (#00001249) Institutional Review Boards. Written informed consent was waived for adult participants. Adolescent participants provided written assent and one legal guardian provided written consent.

Study overview

This cross-cultural adaptation process was part of a larger qualitative study in American Samoa [13,20]. We took a community-partnered approach, collaborating closely with local clinicians, adolescents, young adults, and community leaders during all stages. When possible, we followed the Gjersing et al. [21] guidelines for cross-cultural adaption but modified these principles to better align with Samoan practices for consensus-making [22,23] and for feasibility, when needed. Adult recruitment occurred between October 19 2020 to February 1 2021. Survey recruitment occurred between June 8 2022 to August 16 2022; focus group recruitment occurred between June 8 2022 to June 1 2023. Fig 1 provides an overview of the adaption process.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Overview of the adaptation process.

https://doi.org/10.1371/journal.pmen.0000106.g001

Construct selection

Given our on-going research partnerships in American Samoa, instruments were selected based on local mental health needs, with the intention to inform an upcoming territorial school-based survey. To identify which constructs to measure, we analyzed results from a larger qualitative study that interviewed adult key informants (≥18 years) based in American Samoa between October 2020 and February 2021 (see Mew et al., 2023 for methodology [13]). Participants were asked open-ended questions on a broad range of topics (Table A in S1 Data), which included: “What are the most common mental health problems among adolescents in American Samoa, if any at all?” and “What are the signs that an adolescent in American Samoa is struggling with their mental health that someone else might notice?”. We used responses to these questions, as well as others that captured relevant information, in consultation with local clinical experts (including SLL and JM, among others), to arrive at the final list of mental health problems–or constructs–for which to select measurement instruments that would be most useful to inform local policy and practice.

Instrument selection

After selecting the list of constructs, we then identified publicly available measurement instruments. Given our intent that these tools could be used in both surveys and clinical practice, where possible, we sought to identify tools that would meet the following criteria: (1) adolescent self-reported (rather than clinician administered); (2) currently implemented in screening and clinical settings in American Samoa; (3) validated for the Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition (DSM-5); (4) sufficient validity among adolescent populations; and (5) administered in <15 minutes. To identify candidate instruments, EM systematically examined the literature and held discussions with collaborators to arrive at the list of candidate instruments for each construct. From there, NH, SRL, JM, and EM reviewed instrument candidates and selected the final instruments to adapt.

Expert committee

We assembled an expert committee of six members to guide the cross-cultural adaptation process. This included members with scientific expertise (NH; professor of epidemiology with expertise in Samoan health research and adapting and translating research measures for use in Samoan communities), local clinical expertise (JM, JA, and AG; Samoan mental health clinicians serving adolescent populations across in- and out-patient settings), clinical expertise in trauma and stressor-related conditions (SRL; clinical psychologist and trauma researcher) and one Samoan young adult (FI) with professional experience delivering mental health content to American Samoan adolescents. Samoan members were bilingual in Samoan and English. At the time of this project, this committee included all doctoral-level clinical psychologists (JM and JA) practicing in American Samoa.

Translation and adaptation process

The first stage of the Gjersing et al [21] process is to assess and modify instruments for conceptual equivalence (ensuring instrument domains and the emphasis placed on different domains is the same in both cultures), item equivalence (that the items are relevant and acceptable in both cultures), and semantic equivalence (the meaning of the items is the same in both cultures) [21,22,24]. Our team first modified the English wordings. Given the lack of data on Samoan conceptions of mental health problems, we leveraged secondary data from our qualitative study (as described above [13]) where adult participants explained their perspectives of the signs and symptoms of mental illness among Samoan adolescents, and to which we later broadly validated among adolescent focus groups (focus group methodology described in Mew et al., 2024 [20]). Using the qualitative data, EM modified English wordings to represent the described Samoan-specific conceptions and presentations of each mental health problem. Expert committee members (JM, FI, NH, and SRL) then made further modifications until consensus was reached.

We then forward- and back-translated the modified instruments. We deviated from the Gjersing guidelines, which recommend three translators (two to translate independently, and one to synthesize the forward translations) and three back-translators (two to back-translate independently, and one to synthesize the back-translations). Instead, we followed an approach that leveraged knowledge from a team of six translators/back-translators, which worked in the Samoa-na-lua style, which is a group consensus-based approach, similar to a method previously used in the independent state of Samoa [23]. The items were first forward translated by a team of five professional translators at the Samoan Studies Institute at the American Samoa Community College. This team used a locally-developed process to align with Samoan conceptions of “soalaupule”–meaning, “consultation”, or the traditional Samoan inclusive decision-making process where decisions are based on the consensus of all [25]. Specifically, four translation professionals translated each item individually, and then this group reviewed the translations as a team, with each English item and its Samoan translation displayed on screen. During this process, the team would reconcile and merge translations into one single translation. To address discrepancies, more research was done and presented to show why a particular translator believed their chosen terms should be used. When consensus was reached, the Director of the Institute (OFM) reviewed the final translations and either approved the translation or returned it to the translators for adjustments. If adjustments were minimal, OFM would make final adjustments herself. The translators were bilingual (fluent in both English and Samoan) and bicultural (familiar with both Samoan and American cultures). We then took this synthesized version of translations and conducted one back-translation, conducted by a bilingual and bicultural public health researcher (JN) with experience in adolescent mental health research.

The expert committee then took an iterative process to further modify the instruments according to cultural relevance, acceptability, and conceptual equivalence. We circulated the instruments with the original wordings, our modifications to the English wordings, the professional translations, and the back-translation for expert committee review. From here, we held three online Zoom meetings between April to June 2022 to discuss areas flagged by team members and review each instrument item in English and Samoan. During this process, we continued to modify the English and Samoan wording of the items (drawing on the final forward- and back-translations). We then returned survey items back to the translation team as needed for further adjustments until consensus was reached.

Pretesting with adolescents through an online survey

We then pretested the tools with adolescents to assess and adapt the instruments for acceptability and operational equivalence (i.e., that the questionnaires can be used in the same way by its target population in both cultures) [22]. We did this with an online survey using Research Electronic Data Capture (REDCap) software [26] that presented the English and Samoan wordings of each instrument item side-by-side. Before implementation, we informally piloted this survey with two adults. Though the Gjersing guidelines recommend pretesting the instruments with 30 to 40 respondents, that was not feasible given resource restraints, so we instead developed an alternative approach that leveraged youth feedback through an online survey followed by a focus group. We administered the survey to six bilingual Samoan adolescents in September 2022 (n = 3 girls and n = 3 boys; average age of 16.4 years and an average of 14.2 years spent living in American Samoa; all fluent in Samoan). Participants were recruited from our broader qualitative project; specifically, they were handpicked from a pool of participants who were most engaged during a previous mental health focus group, as described elsewhere [20]. Due to technical difficulties, one participant was unable to finish all questions, which left n = 5 adolescents completing questions for most items.

The survey included each item from all included instruments. The objective of the survey was to gain feedback from adolescents as to whether they understood the wording and whether they would answer the questions honestly. For each prompt, we then asked participants to answer two questions on a Likert scale about comprehension and honesty, specifically: “Is this question easy to understand?” (with the options: “Strongly disagree (total nonsense)”, “Disagree”, “Neutral”, “Agree”, “Strongly agree (totally understandable)”) and “Would adolescents answer this question honestly?” (with the options: “Strongly disagree (totally dishonest)”, “Disagree”, “Neutral”, “Agree”, “Strongly agree (totally honest/truthful)”). When adolescents answered “Neutral”, “Disagree” or “Strongly Disagree”, they were then prompted with a qualitative text box asking for further information (specifically, either “Why is this question hard to understand?” or “Why might adolescents not be honest in answering this question?”). We then analyzed items that received at least one “disagree” or “strongly disagree” response in the surveys. These items were first reviewed by EM and NH, and additional members of the research team when needed, to identify whether further discussion was needed to prioritize items for focus group discussion (necessary given the time restraints in only having one session). Results of this pre-testing phase are presented in each of the Tables below, although no formal analysis was undertaken.

Focus group with adolescents

After collecting survey data, we recruited adolescent participants to participate in a focus group, whose aim was to modify the instruments to improve honesty and understandability, confirm understandability of certain item concepts, and discuss specific safety concerns. We conducted a two-hour focus group in June 2023 with bilingual Samoan youth (13–19 years) who lived in American Samoa. Our sampling frame maximized diversity across gender, as the survey results led us to believe that experiences would likely differ between boys and girls. Five adolescents, recruited through our team’s networks, attended the focus group (n = 3 girls, n = 2 boys; average age of 17.3 years and an average 16.3 years living in American Samoa); one of these adolescents participated in both the survey and focus group, providing us with a sample of n = 10 adolescents who provided pretesting feedback in this study. Participants were sent the list of tools to review prior to the focus group. All participants identified as being of Samoan ethnicity; some also identified that they were multi-ethnic. Other demographic characteristics are not described to protect confidentiality (necessary given the small and close-knit community).

The focus group was led by two experienced Samoan moderators (LH and JN) with public health training (interviewer positionalities in Table B in S1 Data). EM (project principal investigator) observed the session as it encouraged participants to explain cultural nuances in explicit terms that could be understood by an outsider and also contributed to the discussion as needed. Interview prompts were read primarily in English, but the general discussion used both English and Samoan languages. The interview guide (Table C in S1 Data) built on questions generated from instances where the sample communicated strong concerns about understandability and/or honesty for a particular item. As part of this focus group, we also asked high-level questions about promoting an environment for honesty and safety. Intentional efforts were taken to breakdown hierarchies and build trust and rapport between the adults and adolescents in the focus group, such as playing ice breaker games and reaffirming the importance of each participant’s perspectives and contributions to the work. Sections of the transcript in Samoan were translated into English by a Samoan colleague. EM was responsible for reading and reviewing the focus group transcript and the excerpts from the adult semi-structured interviews to generate themes.

Finalizing tools

Incorporating feedback from the adolescents, EM modified the instruments and sent them back to the expert committee for final signoff. TMC and OFM, experts on Samoan language and translation, also reviewed the final tools (presented in a downloadable format in S1–S4 Texts) before dissemination. We also sent the final tools to a Samoan research expert (KS) who has lived in American Samoa, the independent state of Samoa, and New Zealand, who confirmed that the instruments should be valid for use among Samoan adolescents across the Samoan diaspora.

Construct selection

Adult participants from the qualitative semi-structured interviews identified that suicidal thoughts and behaviors (including ‘cutting’ as well as suicidal attempts), depressive disorders, trauma- and stressor-related disorders, and anxiety disorders appear to be the most common mental health conditions among adolescents in American Samoa; those involved in providing clinical services also confirmed that depression, anxiety, and post-traumatic stress disorders are most commonly observed among adolescents in their practice. Traumatic events were described as being very common, particularly sexual assault and physical discipline. Participants described other mental health conditions, but noted that they were less common: bipolar disorder, neurodevelopmental disorders (e.g. attention-deficit/hyperactivity disorder), schizophrenia, and feeding and eating disorders (with an emphasis on over-eating). Based on these results, the research team determined that the most important mental health problems to quantify among adolescents in American Samoa would be: depression, anxiety, post-traumatic stress disorder (which would include potentially traumatic experiences), and suicidality. Substance use (alcohol, methamphetamines, marijuana, and nicotine) was also reported as common, usually as a coping mechanism. We did not adapt an instrument to measure substance use, as several single-item questions involving alcohol, methamphetamine, tobacco, and marijuana use have been regularly employed in internationally deployed school-based surveys in American Samoa and the independent state of Samoa [10,15] and were developed for cross-cultural applications. The authorship team also felt that these existing measures were likely valid without additional cross-cultural adaptation, and thus felt that our resources would be better suited to adapting other instruments.

Instrument selection

To best capture the four identified constructs, we selected the Patient Health Questionnaire-9: Modified for Teens (PHQ-9M) [27] to measure depressive symptoms and provide a clinical measurement of suicide risk; the Generalized Anxiety Disorder-7 (GAD-7) [28] to measure anxiety symptoms; and the Child PTSD Symptom Scale for DSM-5 (CPSS-5) [29] along with its corresponding Trauma Screen to measure exposure to potentially traumatic events and post-traumatic stress symptoms. Of these instruments, the PHQ-9 (original version for adults) and GAD-7 were reported to be used in clinical practice among adolescents in American Samoa. As a supplement to the suicide risk screener in the PHQ-9M, and existing measures employed in the YRBS, we developed a new questionnaire to supplement knowledge gaps with respect to self-harming behavior and intentionality, suicide-related help-seeking behaviors, social networks, and more detailed questions on individual history of suicidal thoughts and attempts. Though we identified several eligible suicide screening instruments, most were developed to be clinician administered, and those that were self-administered were developed for use among adults, making them not well suited for this context. To develop the new questionnaire, we drew questions from the Avon Longitudinal Study of Parents and Children (ALSPAC) self-harm questionnaire [30], given that it was developed to assess a variety of suicidality questions for 16 year olds in a self-administered survey format. As a supplement, we drew questions from the Self-Injurious Thoughts and Behaviors Interview (SITBI) [31] for questions related to suicide attempts and suicidal thoughts. See Table D in S1 Data for our list of candidate instruments and rationale for the selection of each instrument for each construct.

Instrument adaptation

The majority of modifications to the English wordings came from the expert committee; the adolescent focus group only generated three minor modifications to the CPSS-5 and the CPSS-5 Trauma Screener. Before starting, given that adolescents in American Samoa speak both English and Samoan, and that our expert committee agreed that reading comprehension is limited for youth in both languages, the committee recommended that it would be important to present all instrument items in both English and Samoan, side-by-side. The committee thought this would help with understandability so that adolescents could piece together understanding based on both translations as needed. This would also then allow for increased generalizability to other Samoan populations (in both the independent state of Samoa–which is predominantly Samoan speaking–and to Samoans living in English-speaking countries, such as in the continental United States and New Zealand).

The pretesting survey included 86 survey prompts. Given that the results from the survey indicated repeated themes related to honesty that transcended one specific item, much of the adolescent focus group centered around development of administrative practices that would help improve honesty across all items. We present representative quotations from the survey and focus group throughout, but we do not present participant identifiers because the sample was so small. See Tables E, F, G, H, and I in S1 Data for a full presentation of results from the adolescent surveys and focus group decisions for each instrument.

One common recommendation from the expert committee that transcended all instruments was to present the Samoan translations using very simple language, given concerns regarding Samoan literacy. This was further validated during the focus group, as there were common concerns in understanding the Samoan translations and challenges in finding agreement for the Samoan translations. Despite this, members of the expert committee concurred that this issue is mainly due to the complexity of the Samoan language and that it would be challenging to achieve consensus in Samoan translations that everyone could follow, given the many ways to translate English text into Samoan based on the level of formality, among other factors. Given that this was the primary issue, for these instances, the focus group concluded that the tools were sufficiently understandable as long as each instrument would be presented with the English and Samoan versions side-by-side, especially since most adolescents would likely rely primarily on the English wordings. These suggestions are further supported by recent locally-conducted surveys in American Samoa, such as the 2000 Language Status survey that showed that Samoan is only used to converse 67% of the time at home and 38% at school [32], demonstrating the need for both English and Samoan translations.

PHQ-9M

Table 1 provides the cross-culturally adapted PHQ-9M for Samoan adolescents along with changes made from the original instrument. The adult expert committee modified the English wording for six of the thirteen PHQ-9M questions (46%; items 1, 2, 6, 8, 9, and 11 in Table 1). No changes were made to the instrument prompt. The majority of changes were to simplify language and improve comprehension. One exception to this was for item 2, as the expert team expanded the item to also include little interest in “talking to people” as the adult stakeholder interviews suggested a major symptom of depression among youth in American Samoa is social withdrawal–a noticeable deviation from Samoan social norms–and this was not included as a concept in the original instrument. The expert committee felt this change was essential to capture symptom presentation in a Samoan context (Table E in S1 Data). The expert committee also modified items 2 and 11 to include more locally relevant examples (Table E in S1 Data). The expert committee considered adding the concept of anger or aggression, as to build on the definitions of depression from the qualitative interviews, as adult participants described that Samoan adolescents often channel their depression into fights/aggression (as one participant stated: “it fits into the culture better”). The committee, however, did not make this change as to be most conservative in protecting psychometric validity.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Cross-culturally adapted Patient Health Questionnaire-9: Modified for Teens (PHQ-9M) for Samoan adolescents with changes from the original PHQ-9M and pretesting results for understandability and honesty among n = 6 adolescents.

https://doi.org/10.1371/journal.pmen.0000106.t001

Pretesting demonstrated that this instrument would be sufficiently understandable and elicit sufficiently honest answers (Table 1). For 11 of the 13 items (85%) and for the question prompt, no adolescent raised concerns in understandability (reporting either “disagree” or “strongly disagree” to “Is this question easy to understand?”). There was unanimous agreement (reporting either ‘agree’ or ‘strongly agree’) that eight of the 13 items (62%) were understandable and there were only two items (15%) where at least one adolescent disagreed (reported ‘disagree’ or ‘strongly disagree’) (items 4 and 8; Table 1). The focus group discussed why these two items were unclear and offered suggested wording changes to improve understandability (this included suggestions to change the wording of item 4 to include “eating disorders”). Ultimately, the group reached consensus to leave these items unchanged as the majority of the adolescents in the group preferred the original wording (Table E in S1 Data).

For seven of the 13 items (54%), at least one adolescent reported that they disagreed (reporting answers of “disagree” or “strongly disagree”) with the question “Would adolescents answer this question honestly?” (items 1, 2, 6, 7, 8, 12, and 13). Respondents provided several qualitative reasonings (see Table J in S1 Data); examples included: “they don’t want the truth to come out”, “they’re going to lie so that they are positive about themselves but in reality they’re not”, “they feel as they can’t quite express their feelings to anyone or even a stranger”, and “they don’t want to be seen as weak or vulnerable” (Table J in S1 Data). Given a general theme across items, the focus group discussed general approaches to improving honesty rather than changing the wording for any specific item (see Development of Administration Instructions section).

GAD-7

Table 2 provides the cross-culturally adapted GAD-7 for Samoan adolescents along with changes made from the original instrument. The adult expert committee modified the English wording for five of the eight questions (63%; items 1, 3, 6, 7, and 8 in Table 2). No changes were made to the instrument prompt. All changes were to simplify the language for understandability or to add in contextually- and locally-relevant examples (see Table F in S1 Data for detailed reasoning). No changes were made to the instructions or Likert Scale answers.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Cross-culturally adapted Generalized Anxiety Disorder-7 (GAD-7) for Samoan adolescents with changes from the original GAD-7 and pretesting results for understandability and honesty among n = 5 adolescents.

https://doi.org/10.1371/journal.pmen.0000106.t002

Pretesting with adolescents demonstrated that this instrument would be sufficiently understandable and would elicit sufficiently honest answers (Table 2). No participants raised concerns about understandability; seven of the nine items (78%) had all participants either “agree” or “strongly agree” that the item was understandable. Based on this, the adolescents made no further changes to either the English items or Samoan translations. For honesty, no participant reported that they “disagreed” or “strongly disagreed” with the question “Would adolescents answer this question honestly?”. Respondents, did however, report three qualitative reasons for items 5, 6, and 7 (see Table F in S1 Data and Table J in S1 Data). Examples include: “they don’t want to be mean and selfish” and “lack to confidence to speak up”. Similar to the PHQ-9M, we used the focus group to decide on general strategies to improve honesty, rather editing individual items. Table 3 provides the cross-culturally adapted Trauma Screen for Samoan adolescents along with the changes made from the original instrument. The adult expert committee modified the English wording for ten of the 20 question items (50%; items 1, 2, 4, 5, 6, 7, 8, 13, 15, and 16 in Table 3) and to both question prompts. Changes to the English wording of items were to accommodate the local context where some traumatic events were considered so normative that they might not be considered traumatic; for example, the committee changed item 2 to be “seriously injured” by a dog, since minor dog bites are common locally and not considered traumatic enough to be classified as a potentially traumatic event. Similarly, the committee changed the language to “serious injury” from sports to again not confuse it with minor injuries in this context. Likewise, for items 4–7, we replaced “slapped” with “severely physically hurt” to differentiate between the light slapping that remains a very common form of communicating irritation or frustration in this context [20] from intentional discipline/punishment. Given that this could potentially change the construct, we validated understanding of what “severely physically hurt” would mean for adolescents in the focus group, and this was interpreted as bruises on the body and “whatever wound or the injury you have actually gets you to the hospital. Like you need to be treated right away”. The remaining changes made to the English wording of items were to either simplify language for understandability or to add in contextually- and locally-relevant examples (see Table G in S1 Data for reasoning).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. Cross-culturally adapted Trauma Screener (as part of the CPSS-5) for Samoan adolescents with changes from the original Trauma Screener and pretesting results for understandability and honesty among n = 5 adolescents.

https://doi.org/10.1371/journal.pmen.0000106.t003

Pretesting did, however, identify several items that might be challenging for adolescents to understand (Table 3). No participants raised concerns about understandability for the three instrument prompts. Though 17 of the items (out of 20; 85%) raised no concerns about understandability, there were four items (17%) that received at least one response of “disagree” or “strongly disagree” to whether the item is sufficiently understandable (items 3, 5, 11, and 14), though no qualitative explanations were provided. Based on this, we discussed items 3, 11, and 14 in the focus group; we also discussed item 5 to validate interpretation (see last paragraph). Themes that arose were that the Samoan was too advanced to understand, but the group agreed that this would be fine if presented alongside the English translation. Adolescents also stated for item 14, that the concept of “war” was unclear, and questioned whether it would also apply to inter-village conflicts and disputes. Based on these discussions, in collaboration with our expert team, we modified the English and Samoan wordings to define war (meaning, armed conflict that could result in death, as this is something that would be potentially traumatic in a way to elicit post-traumatic stress symptoms).

For honesty, there were two items (10%) that received at least one response of “disagree” or “strongly disagree” to whether the item is sufficiently understandable (items 3 and 8). Five qualitative explanations were provided (Table J in S1 Data) and these responses centered around fear of exposing secrets, most notably sexual abuse, such as: “still trying to gain the courage to speak and share their [sic] story” and “I think too many people would feel embarrassed or insecure if they talk about how they were sexually harassed”. One adolescent expressed fear of repercussions from exposing domestic violence at home, such as: “still the chances of them getting something worse or bad back”. Based on this, similar to the PHQ-9M and GAD-7, we made effort in the focus group to discuss how to create a safe environment for adolescents to feel comfortable sharing stigmatised and sensitive family topics.

CPSS-5

Table 4 provides the cross-culturally adapted CPSS-5 for Samoan adolescents along with the changes made from the original instrument. The adult expert committee modified the English wording for three of the 30 question items (10%; items 1, 4, 15 in Table 4) and one of the two instrument prompts. All changes made to the English wording were to either simplify the language or to add in contextually- and locally-relevant examples (see Table H in S1 Data for detailed reasoning).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 4. Cross-culturally adapted Child PTSD Symptom Scale for DSM-5 (CPSS-5) for Samoan adolescents with changes from the original CPSS-5 and pretesting results for understandability and honesty among n = 5 adolescents.

https://doi.org/10.1371/journal.pmen.0000106.t004

Pretesting identified two items (items 21 and 26) that might be challenging for adolescents to understand (Table 4). Though for 28 of the items (out of 30; 93%), no participant raised concerns about understandability, there were two items (7%) that received at least one response of “disagree” or “strongly disagree” to whether the item is sufficiently understandable (items 21 and 26), though no qualitative explanations were provided. Based on this, we discussed items 21 and 26 in the focus group, which revealed a small grammar edit to the item 26 translation, but otherwise there was consensus to leave the items unchanged.

For pretesting whether adolescents would report honest answers, no participants reported disagreement for 29 items (out of 30; 97%). Ony one item (3%) received at least one response of “disagree” or “strongly disagree” to whether adolescents would report honest answers (item 16). Two qualitative responses were provided with further reasoning (Table J in S1 Data), one of which involved fear of sharing negative information about their families: “I feel like most teens would lie so that others would think that they have a healthy relationship with their families”. This issue was discussed further in the focus group.

Deliberate self-harm questionnaire

Table 5 outlines the suicide questionnaire developed for Samoan adolescents. The majority (67%) of questions were copied verbatim from the ALSPAC questionnaire (12 questions and the instrument prompt; items 1, 2, 6–8, 10–13, 16–18); three items (17%; items 3–5) were copied from the SITBI and one question was adapted from it (item 9; Table I in S1 Data). We developed two new questions to capture history of suicide attempts (items 14 and 15), as we were interested in developing instruments that could measure frequency of attempts and age of first attempt; a measure that was not included in existing instruments. See Table I in S1 Data for detailed breakdown of question adaptations.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 5. Cross-culturally developed deliberate self-harm questionnaire for Samoan adolescents (based on the ALSPAC and SITBI questionnaires) with pretesting results for understandability and honesty among n = 5 adolescents.

https://doi.org/10.1371/journal.pmen.0000106.t005

Pretesting with adolescents identified four items that might be challenging to understand (Table 5). Although no participants raised concerns regarding understandability for 14 of the items (out of 18; 78%), four items (22%) received at least one response of “disagree” or “strongly disagree” as to whether the item was sufficiently understandable (items 6, 7, 8, and 10). Based on this, we discussed items 6, 7, 8, and 10 in the focus group (Table I in S1 Data); adolescents expressed that the Samoan is harder to understand than the English, but after attempts to make it clearer, the group decided that no further changes would need to be made.

For honesty, no participants disagreed that adolescents would respond honestly for 11 of the items (out of 18; 61%). Six items (22%) received at least one response of “disagree” or “strongly disagree” to whether adolescents would be honest (items 6–10, 12). Specifically, the questions related to social networks, suicidal thoughts, suicide attempts, and help-seeking behaviors had complete agreement that adolescents would be honest; but nearly all of the questions related to self-harming behaviors (6/7; 86%) had at least one adolescent report concerns about honesty. Only one qualitative reason was provided (Table J in S1 Data): “Scared of expressing themselves”. Based on this, we took the same approach as other instruments and centered focus group discussions around how to make adolescents feel safe being honest.

Development of administrative instructions

Adolescents in the focus group corroborated the pretesting survey data, expressing that adolescents would be less honest for a variety of sensitive questions concerning sexual trauma, physical punishment, violence at home, and communicating stigmatized feelings such as sadness, irritation, and shame. Adolescents provided suggestions to administer the survey in a way to promote honesty. Two major recommendations arose: (1) build a safe environment to make the adolescent feel comfortable; and (2) explain and communicate intentions behind the questionnaire. To better promote adolescent comfort, we developed instructions for survey administrators to implement prior to delivering the questionnaires (Table 6; also in S1–S4 Texts).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 6. Evidence-based administrative instructions to promote honesty developed for the PHQ-9M, GAD-7, CPSS-5, Trauma Screen, and Deliberate Self-Harm Questionnaire for Samoan adolescents.

https://doi.org/10.1371/journal.pmen.0000106.t006

Participants recommended administering the questionnaire in a private space: “maybe he or she will feel comfortable telling only one person, like [translation: should be one person to ask the question in private]”. Participants also recommended that the administrator should attempt to build trust before administering the survey, as: “maybe before giving out the surveys, maybe try to earn their trust first, like try to get to them [to] feel safe to give out those kinds of personal information”. This could include the adult administrator asking the adolescent questions about their interests, employing a non-judgmental, warm, and present demeanour, and actively listening. Some adolescent participants gave specific demographic examples for administrators who would naturally elicit more trust, including someone who is an outsider, such as ‘not the teacher’ or someone who identifies as Fa’afine (meaning, a person who identifies as non-binary) “[because] they make us very feel very comfortable”. Participants also stated the importance of building trust. Participants also recommended that administrators clearly communicate that the information gathered would be confidential: “maybe kind of telling them you guys will keep that as a secret [and] not telling anyone about what [their] answer is”. This included clearly explaining what circumstances or situations would require sharing disclosed information with their families.

Participants also recommended clearly communicating the purpose of the questionnaire so that adolescents understand why it is important to answer honestly, as: “translation: understand and know what the main purpose of the survey [is]”. Adolescents also recommended that administrators communicate that the purpose of the questionnaire is not to get them or someone they know in trouble. This was especially applicable for the questions concerning sexual trauma, as there was concern that the goal of the questionnaires would be to reprimand someone in the adolescent’s network. Participants also shared that the mode of delivery (paper versus electronic) would not make a difference in the level of honesty.