Peer support based on Mental Health First Aid for medical residents in Lyon, France: protocol for a non-randomised controlled mixed-methods study Authors: Guillaume Bailly, Guillaume Barbalat, Juliette Danckaert, Nicolas Franck, Florence Guillotin, Arthur Poncin, Ludivine Nohales, Emmanuel Poulet, Audrey Maillet, Isabelle Célérier, Killian L’Helgouarc’h, Gilles Rode, Philippe Paparel, Bach-Nga Pham, Edouard Leaune Published: 2026 ABSTRACT AbstractIntroductionMedical residents frequently experience anxiety, depression, burnout and suicidal ideation. Existing support strategies have shown limited benefit. The Impact de la Pair-Aidance Étudiante sur la Santé Mentale des internes de Médecine de première année (PAESSM) study will evaluate whether embedding Mental Health First Aid (MHFA)-trained peer supporters within a first-year residency cohort is associated with improved mental well-being and reduced psychological distress over 12 months.Methods and analysisPAESSM is a non-randomised controlled mixed-methods quasi-experimental study in Lyon, France involving the two Lyon medical schools. Two contemporaneous cohorts will be followed for 12 months: first-year residents exposed to a structured peer-support programme and second-year residents receiving usual support resources without a structured peer-support network. About 430 residents are eligible in each cohort, with anticipated enrolment of approximately 260 per group. In the intervention cohort, around 10% of first-year residents will complete certified MHFA training and act as peer supporters. Online questionnaires will be administered at baseline and at 3, 6, 9 and 12 months. The primary outcome is 12-month mental well-being measured by the Warwick-Edinburgh Mental Well-Being Scale. Secondary outcomes include anxiety and depression symptoms, suicidal ideation, job strain and help-seeking behaviours. The primary analysis will use longitudinal targeted maximum likelihood estimation adjusted for prespecified baseline covariates and informative censoring. Sensitivity analyses will include mixed-effects models and complete-case analyses. A qualitative sub-study will explore peer supporters’ experiences and implementation processes.Ethics and disseminationEthical approval was obtained from the Comité de Protection des Personnes Nord-Ouest IV (Ref: 2025-A01549-40). In accordance with French legislation applicable to minimal-risk research, formal written informed consent was waived and participants will provide documented non-opposition before inclusion. Findings will be disseminated through peer-reviewed publications, conference presentations and communication with institutional stakeholders and residents.Trial registration numberNCT07161284. STRENGTHS AND LIMITATIONS OF THIS STUDY This study evaluates a Mental Health First Aid-based peer-support programme in a real-world residency setting using a prospective, non-randomised, controlled, mixed-methods design. This study evaluates a Mental Health First Aid-based peer-support programme in a real-world residency setting using a prospective, non-randomised, controlled, mixed-methods design. Repeated assessments over 12 months will allow longitudinal evaluation of mental well-being and psychological distress. Repeated assessments over 12 months will allow longitudinal evaluation of mental well-being and psychological distress. The mixed-methods approach will provide both outcome data and contextual information on implementation, acceptability and mechanisms. The mixed-methods approach will provide both outcome data and contextual information on implementation, acceptability and mechanisms. A contemporaneous comparison cohort from the same residency training setting in Lyon should limit institutional confounding, but residual confounding cannot be excluded because the groups differ by year of training. A contemporaneous comparison cohort from the same residency training setting in Lyon should limit institutional confounding, but residual confounding cannot be excluded because the groups differ by year of training. The non-randomised design conducted within a single residency training setting in Lyon, together with self-reported measures, may limit causal interpretation, external validity and susceptibility to reporting and attrition bias. The non-randomised design conducted within a single residency training setting in Lyon, together with self-reported measures, may limit causal interpretation, external validity and susceptibility to reporting and attrition bias. Background Medical residents experience high levels of psychological distress, including depression, anxiety, burnout and suicidal ideation. Meta-analyses have reported substantial rates of depression and psychological distress among medical trainees, with the prevalence of depression estimated at around 27% in medical students.1–6 This vulnerability appears early, as incoming medical students already show higher levels of distress than their peers in the general population.4 French data are consistent with these findings. A 2022 nationwide survey of more than 3200 medical trainees reported high levels of anxiety, depression, burnout and nearly 20% of suicidal ideation.5 At the local level, the BASIL study conducted within the Lyon University Hospital network mirrored these national trends and confirmed a high prevalence of distress among residents in this setting.7 Several structural and cultural factors contribute to this burden. Residents face heavy workloads, long working hours and the combined pressure of clinical duties and academic expectations.3 8–10 At the same time, training in mental health self-care or peer support is rarely included in medical curricula. Stigma also remains a major barrier: fear of being perceived as weak, incompetent or unfit for practice may discourage help-seeking, despite the documented consequences of untreated distress, including substance use, professional misconduct, attrition and suicidality.9 10 Traditional support systems, such as counselling services, helplines and general wellness initiatives, have not clearly reduced burnout or improved mental health outcomes among medical trainees. By contrast, peer-support approaches and Mental Health First Aid (MHFA) training have shown promising results in educational settings, including among medical students.11–16 MHFA trains participants to recognise signs of psychological distress, provide initial support and guide peers towards appropriate professional resources. Two studies have specifically evaluated MHFA training among medical students, in the UK and Australia, using online and face-to-face formats, respectively.13 16 Bond et al reported that MHFA training improved mental health first aid intentions, confidence in providing help, mental health literacy and stigmatising attitudes towards mental health problems.16 Davies et al similarly found that students assigned to the intervention group reported a greater number of actual mental health first aid actions at follow-up.13 Participants also reported positive views of the training and greater confidence in supporting someone in distress. Beyond the direct support provided to individuals experiencing difficulties, the presence of trained peer supporters within a residency cohort may exert a broader preventive effect at group level. By normalising conversations about mental health, increasing awareness of available support resources, and reducing the sense of isolation, peer supporters may help create a more supportive social environment, including for residents who do not actively seek help. Such an approach may contribute to reducing stigma around psychological vulnerability, facilitating earlier identification of distress and promoting timely access to care. On this basis, the Impact de la Pair-Aidance Étudiante sur la Santé Mentale des internes de Médecine de première année (PAESSM) study was developed to evaluate whether embedding trained peer supporters within a first-year residency cohort improves mental well-being and reduces psychological distress among medical residents. This protocol describes a non-randomised, controlled, mixed-methods quasi-experimental study evaluating this peer-support intervention in a real-world residency setting. Methods Study setting In France, medical education consists of 6 years of undergraduate training followed by 4–6 years of residency training, depending on specialty. At the end of undergraduate training, students are ranked according to a national examination and then choose both their specialty and training location.17 In France, residency training is organised within regional subdivisions. The present study is conducted in Lyon, across its two medical schools (Lyon Est and Lyon Sud). Study design The PAESSM study is a non-randomised, controlled, mixed-methods quasi-experimental study conducted in Lyon, France, across the two medical schools in Lyon (Lyon Est and Lyon Sud) and the associated hospital training setting, including the Hospices Civils de Lyon. Two cohorts of residents will be followed over the same 12-month period, from November 2025 to October 2026: a first-year cohort exposed to the intervention and a second-year comparison cohort not exposed to a structured peer-support programme during their first year of residency (figure 1). The intervention cohort comprises first-year residents starting residency in Lyon during the 2025–2026 academic year, across all specialties. The comparison cohort comprises second-year residents training in Lyon during the same academic year. These second-year residents completed their first year of residency without exposure to a structured peer-support programme. Second-year residents were selected as the comparison group because they provide the closest available contemporaneous comparator within the same institutional context. Both cohorts train within the same residency training setting in Lyon and hospital network, are exposed to similar organisational structures and formal support resources, and are assessed over the same calendar period. This choice is intended to limit institutional, geographical and organisational confounding. Evidence from the literature also suggests that year of residency is not, in itself, a major determinant of psychological distress or burnout.18 19 However, second-year residents are not identical to first-year residents, particularly with regard to the transition into residency and accumulated clinical experience. Baseline differences between cohorts will therefore be explicitly described and taken into account in the analyses. Variables to be compared at baseline will include age, sex, specialty, Warwick-Edinburgh Mental Well-being Scale (WEMWBS) total score, Hospital Anxiety and Depression Scale (HADS) anxiety and depression subscale scores, suicidal ideation items, Job Content Questionnaire (JCQ) dimensions and self-reported prior access to psychological support. Randomisation was not considered feasible because the intervention was implemented at cohort level within routine residency organisation. Once trained peer supporters were embedded within the incoming first-year cohort, contamination between individually randomised participants would have been highly likely. For this reason, a controlled quasi-experimental design using a contemporaneous comparison cohort was considered the most appropriate and pragmatic approach. Due to the structural nature of the intervention and the use of distinct academic cohorts (first-year vs second-year residents), blinding of participants and investigators to group allocation is not possible. However, the risk of assessment bias is minimised as all quantitative outcomes are collected via standardised, self-reported online questionnaires. Blinding of the statistical analysis will not be performed because the analysis relies on characterising and adjusting for baseline differences related to year of residency. Flowchart of the Impact de la Pair-Aidance Étudiante sur la Santé Mentale des internes de Médecine de première année study design. HADS, Hospital Anxiety and Depression Scale; JCQ, Job Content Questionnaire; M, months; MHFA, Mental Health First Aid; TMLE, Targeted Maximum Likelihood Estimation; WEMWBS, Warwick-Edinburgh Mental Well-being Scale. Participants and inclusion criteria All residents aged 18 years or older enrolled in first- or second-year residency training in Lyon, across the two medical schools (Lyon Est and Lyon Sud), during the 2025–2026 academic year will be eligible. Eligibility is broad and does not depend on specialty, sex or previous mental health history. Participation is voluntary and follows the French non-opposition procedure after provision of study information applicable to minimal-risk research. Recruitment will occur in two phases. Peer supporters will first be recruited during the November 2025 orientation session. The wider resident cohort will then be recruited from mid-November to mid-December 2025 through study presentations and electronic registration on a secure platform. Reminder messages will be sent during the enrolment period. Participants may withdraw at any time without consequences for their training or access to support services. Intervention: peer-support programme The intervention consists of embedding a structured network of peer supporters within the first-year resident cohort. Before the start of clinical duties, 43 first-year volunteers, representing approximately 10% of the incoming cohort, will complete a standardised 2-day MHFA training certified by PSSM France, the French national licensee for Mental Health First Aid International (figure 2). Recruitment of peer supporters will aim to ensure representation across specialties, with recruitment targets established proportionally to the size of each residency specialty. This strategy is intended to provide equitable access to trained peer supporters across clinical disciplines. The distribution of peer supporters across specialties will be documented as part of the process evaluation. Delivered by an expert MHFA trainer, this curriculum provides core competencies for recognising signs of psychological distress, approaching peers and listening non-judgementally, identifying potential crises including suicidal ideation, and orienting peers towards appropriate mental health resources. After training, peer supporters will receive a curated list of institutional and community mental health resources and will begin their role as active sentinels within their cohort. Their role goes beyond passive availability. They are expected to adopt a proactive stance of vigilance and early detection, approaching peers who appear withdrawn or distressed rather than waiting for an explicit request for help. They will not provide psychotherapy or medical management, but rather an initial, confidential and approachable form of support, helping colleagues articulate their difficulties, feel reassured and access professional care when needed. To support the sustainability and emotional safety of this role, peer supporters will participate in quarterly supervision sessions led by certified MHFA instructors or mental health professionals. These structured debriefings will provide space to discuss difficult situations, share experiences, clarify role boundaries and reduce the risk of helper fatigue or emotional overload. Attendance at supervision sessions will be documented as part of the intervention process evaluation. The peer-support initiative is explicitly designed as an adjunct to usual institutional resources rather than a substitute. Residents in both cohorts will retain full access to existing support services, including occupational health consultations, university counselling, institutional psychological support and emergency psychiatric pathways. Regular reminders of these services will be disseminated across both cohorts to ensure that no participant is deprived of standard care. Trajectory of a peer support: from selection to evaluation. MHFA, Mental Health First Aid; PSSM, Premiers Secours en Santé Mentale. Study procedures and data collection Data will be collected electronically using a secure General Data Protection Regulation (GDPR)-compliant eCRF platform. Participants will complete online questionnaires at baseline (M0) and at 3, 6, 9 and 12 months. For all participants, M0 is defined as the date of completion of the baseline questionnaire during the enrolment window (mid-November to mid-December 2025). All participants therefore share the same baseline assessment window. Follow-up assessments are scheduled at M3 (January 2026), M6 (April 2026), M9 (July 2026) and M12 (October 2026), each within a ±15-day window. The M12 follow-up will extend into the second year of residency for participants in the intervention cohort, a contextual factor that will be accounted for in the analysis. The WEMWBS, HADS, Columbia-Suicide Severity Rating Scale (C-SSRS) items, JCQ and help-seeking questionnaire are all administered at all five time points (M0, M3, M6, M9 and M12). Mental well-being will be assessed using the WEMWBS. This 14-item scale covers functional and feeling aspects of mental well-being. Items are scored on a 5-point Likert scale, yielding a total score ranging from 14 to 70. The WEMWBS has shown good psychometric properties.20 Its validity has been confirmed in French populations and used in large-scale online mental health surveys.21 22 Symptoms of anxiety and depression will be assessed using the HADS. This 14-item scale comprises two 7-item subscales for anxiety and depression. Each item is scored from 0 to 3, producing anxiety and depression subscale scores ranging from 0 to 21.23 Suicidal ideation will be monitored using a three-item screener adapted from the C-SSRS, assessing wish to be dead, non-specific active suicidal thoughts and active suicidal ideation with any method.24 Job strain will be measured using a 27-item adaptation of the JCQ, which assesses psychological demands, decision latitude and social support.25 In addition, participants will be asked whether they sought support since the previous assessment. In the intervention group, this will include self-reported use of peer-support contacts. Automated reminders and personalised follow-up will be used to reduce attrition across repeated assessments. Safety procedure for suicidal ideation Because the study includes repeated assessment of suicidal ideation, a predefined safety procedure has been established. If a participant endorses any suicidal ideation item, an immediate on-screen alert will be displayed within the questionnaire. This alert will encourage the participant to seek help promptly and will provide contact details for available support resources, including emergency psychiatric services, national crisis helplines, occupational health services, university support services and local mental healthcare pathways available to residents. Participants will also be informed that they may contact the principal investigator, who is a psychiatrist, if they need help identifying an appropriate care pathway. This optional contact is available to participants in both groups, is initiated by the participant, and is not considered part of the study intervention. The peer-support intervention is intended as an adjunct to, and not a substitute for, professional mental healthcare. Participants in both groups will retain access throughout the study to usual institutional and emergency support services, and reminders of available mental health resources will be provided during follow-up. Given the minimal-risk observational design and the confidentiality framework of the study, questionnaire responses will not automatically trigger direct individual clinical outreach by the research team. This limitation will be explicitly stated in the study information materials, together with clear instructions that, in the event of acute distress or immediate danger, participants should contact emergency services or other appropriate professional support without delay. Contact with the principal investigator is intended solely to facilitate orientation towards appropriate care and does not constitute an emergency response pathway or a standardised component of the intervention. Outcomes The primary endpoint is mental well-being at 12 months, measured by the WEMWBS total score.20 Secondary quantitative endpoints include trajectories of anxiety, depression, suicidal ideation and job strain over follow-up, as well as help-seeking behaviour. Prespecified subgroup analyses by sex and specialty will be considered exploratory. Qualitative outcomes will focus on the lived experience of peer supporters, including their perception of the role, challenges encountered during the intervention, perceived usefulness of supervision and the perceived impact of the role on their own professional identity and well-being. Sample size considerations This study uses a pragmatic whole-cohort design. No formal a priori sample size calculation was performed because the intervention was implemented at the level of the whole incoming residency cohort, and the number of eligible participants was therefore determined by the study setting rather than by recruitment to a predefined target based on an assumed effect size. Based on cohort sizes observed during the previous academic year, approximately 430 first-year residents and 430 second-year residents are expected to be eligible. Assuming a participation rate of around 60%, we expect to recruit approximately 260 participants in each group. This sample size was considered feasible within the study setting and appropriate for estimating between-group differences and longitudinal trends in mental well-being in a real-world implementation context. However, the study may have limited precision to detect small effects, and subgroup analyses should be interpreted as exploratory. Qualitative sub-study: design and data collection This study uses an embedded mixed-methods design, in which the qualitative component is integrated within the quantitative evaluation to better understand implementation processes, mechanisms of action and contextual factors that may influence the observed outcomes. In parallel with the quantitative evaluation, we will conduct a qualitative study exploring the experiences of peer supporters. This component is intended not only to describe their lived experience, but also to clarify how the intervention was implemented in practice, which mechanisms were perceived as helpful or challenging, and which contextual conditions may have shaped its effects. A purposive sample of 15–25 peer supporters will be selected to capture diversity in specialty, gender and training background.26 Recruitment will continue until thematic saturation is reached.27 28 Interviews will be conducted during the final months of the intervention (months 11–12), allowing participants to reflect on their experience over the academic year. Interviews will be carried out by a researcher trained in qualitative methods and not involved in resident supervision, in order to preserve neutrality and promote psychological safety. Interviews will last 30–60 min and will be conducted face to face whenever possible, or via secure videoconferencing if necessary. A semi-structured interview guide will explore participants’ perceptions of the MHFA training, role enactment, supervision, role boundaries, emotional burden, facilitators and barriers to peer support, perceived effects on peers, and the perceived impact of the role on their own professional identity and well-being (see online supplemental file 1). Interviews will be audio-recorded with consent, transcribed verbatim and pseudonymised before analysis. Statistical analysis plan Baseline characteristics will be described separately for first-year and second-year residents, including demographic characteristics, specialty distribution, baseline mental health measures and missing baseline values. Continuous variables will be summarised using means and SD, and categorical variables using frequencies and percentages. Primary analysis The primary analysis will use longitudinal targeted maximum likelihood estimation, implemented with the lmtp package in R. Repeated measurements collected at baseline and at months 3, 6, 9 and 12 will be used to estimate the adjusted between-group difference in mental well-being at 12 months while accounting for prespecified baseline covariates and informative censoring. All analyses will use a two-sided significance threshold of 5%. If extreme values suggest limited support for certain covariate patterns, an alternative causal estimand based on an intervention score may be used, as prespecified, to relax the positivity assumption. In the intervention cohort, peer supporters will initially be excluded from the primary analysis because they are exposed to the intervention in a qualitatively different way, combining MHFA training, active supportive responsibilities and supervision. Including them in the main intervention group could therefore conflate the effects of receiving peer support with those of providing it. In a prespecified sensitivity analysis, peer supporters will be pooled with the rest of the first-year cohort. Secondary and complementary analyses Secondary and sensitivity analyses will be performed to explore longitudinal trajectories and assess the robustness of the main findings. Difference-in-differences and event-study approaches, combined with a propensity score model, will be used to examine changes over time between groups. In these models, the effect of interest will be the time-by-group interaction. As a complementary longitudinal analysis, linear mixed-effects models will also be fitted, including fixed effects for group, time and group-by-time interaction, with a random intercept for each participant. Secondary outcome analyses Secondary quantitative outcomes, including HADS anxiety and depression scores, suicidal ideation items and JCQ dimensions, will be analysed using analogous longitudinal modelling strategies. Binary outcomes will be analysed using logistic regression models. Prespecified subgroup analyses by sex and specialty will be considered exploratory. As an additional complementary analysis within the intervention cohort, we will examine the association between self-reported exposure to peer-support contacts (collected at each follow-up assessment) and primary and secondary outcomes, to account for variation in actual exposure to peer support within the intervention group. Missing data Participants who complete the baseline questionnaire will form the analysis population. Missing baseline and follow-up data will be handled using multiple imputation, and informative censoring during follow-up will be addressed by weighting each participant according to their estimated probability of remaining under follow-up. Complete-case analyses will be performed as sensitivity analyses. Qualitative analysis and mixed-methods integration Qualitative analysis will follow Braun and Clarke’s thematic analysis approach, supported by NVivo software for data management.29 Two researchers will independently code transcripts and iteratively refine codes, categories and themes through constant comparison. The analysis will aim to develop a conceptual understanding of how peer supporters navigated their role, what facilitated or hindered their perceived effectiveness, and how the experience affected them personally and professionally. Methodological rigour will be supported using Lincoln and Guba’s criteria, including credibility through triangulation and multidisciplinary discussion, dependability through documentation of analytic decisions, confirmability through independent coding and transferability through detailed description of participants and context. Integration of the quantitative and qualitative components will occur at the interpretation stage. Quantitative findings will be used to assess whether the intervention is associated with differences in well-being and psychological distress over time, whereas qualitative findings will be used to explain how the intervention was experienced, how it may have operated, and which implementation barriers or facilitators may help interpret the quantitative results. This joint interpretation will provide a more comprehensive assessment of effectiveness, acceptability and implementation than either component alone. Patient and public involvement Medical residents, through representatives of the Intersyndicale Nationale des Internes, were actively involved in the design of the study. They contributed to identifying the research question, shaping the peer-support intervention and ensuring that the study addressed relevant needs of the target population. Residents will not be involved in recruitment or study conduct, but the results will be disseminated to them through appropriate institutional and professional channels. Discussion This PAESSM protocol evaluates a structured MHFA-based peer-support intervention for medical residents using a prospective, non-randomised controlled mixed-methods design. It will assess the intervention in a real-world residency setting, where evidence on effective preventive strategies for this population remains limited.30 31 Peer-support programmes for clinicians have also been described more broadly, although such approaches have been less frequently evaluated in residency training.32 A key methodological issue is the use of second-year residents as the comparison group for first-year residents exposed to the intervention. This choice does not eliminate the possibility of residual confounding, as first-year residents face the specific challenge of entering clinical practice. However, second-year residents were selected because they provide the closest available contemporaneous comparison under the same institutional conditions: they train in the same residency training setting in Lyon and hospital system, are exposed to the same duty structures and formal support resources, and represent usual residency conditions in Lyon without a structured peer-support programme. Zhou et al, in a meta-analysis of 48 studies, found no significant differences in burnout or stress according to year of training.18 Navinés et al similarly reported no clear association between burnout and residency year during the COVID-19 period.19 French national data are consistent with these findings: Rolland et al found no significant between-year differences in anxiety, depression, suicidal ideation or burnout dimensions among more than 3200 residents.5 Taken together, these data support the use of second-year residents as a pragmatic contemporaneous comparator, although any remaining baseline differences will need to be accounted for analytically. The 12-month primary endpoint was selected to reflect the organisation of French residency, which is structured around 6-month rotations. This timeframe should allow assessment across at least two different clinical settings and capture both the cumulative effects of training-related stressors and the progressive implementation of the peer-support network. In addition, the prospective design may provide useful longitudinal data on the evolution of mental health during the first year of residency, a period that remains insufficiently documented in France, where most available studies are cross-sectional.5 7 The mixed-methods design should also help interpret the findings. Quantitative analyses will assess the association between the intervention and changes in well-being and psychological distress, while the qualitative component will explore how peer supporters experience their role and how the intervention operates in practice. This may help clarify implementation barriers and facilitators and support interpretation of the quantitative findings. If the intervention is associated with improved well-being or reduced psychological distress, the findings could support broader implementation of structured peer-support programmes in residency. If no clear effect is observed, the study will still provide useful information on how such interventions function in practice and on the extent to which peer support alone can address distress in a training environment shaped by substantial structural stressors. Ethics and dissemination Ethical approval was obtained from the Comité de Protection des Personnes Nord-Ouest IV (Ref: 2025-A01549-40). In accordance with French legislation applicable to minimal-risk research, formal written informed consent was waived. After receiving detailed study information, participants will provide documented non-opposition before inclusion. Participation is voluntary, and participants may withdraw at any time without any consequences for their training or access to usual support services. The study will be conducted in accordance with the Declaration of Helsinki and applicable data protection regulations, including the GDPR. Because the study includes repeated assessment of suicidal ideation, a predefined safety procedure has been established: participants reporting suicidal ideation will receive an immediate on-screen message providing appropriate crisis and mental health support contacts, and may contact the principal investigator for help identifying an appropriate care pathway. Study findings will be disseminated through peer-reviewed publications and presentations at national and international conferences. Results will also be shared in aggregate form with institutional stakeholders, residency programme directors and medical residents to inform the future implementation of peer-support interventions in medical training. Any important protocol modifications will be communicated to the research ethics committee and trial registry, and reflected in protocol updates when applicable. This study protocol describes a prospective mixed-methods evaluation of an MHFA-based peer-support intervention embedded within a residency cohort. By combining longitudinal quantitative data with qualitative analysis, the study aims to generate evidence on the intervention’s implementation and its association with residents’ well-being and psychological distress. The findings may inform future mental health support strategies in residency training. Supplementary material