Duration of New-Onset Depressive Symptoms During Medical Residency.

Importance: The implications of new-onset depressive symptoms during residency, particularly for first-year physicians (ie, interns), on the long-term mental health of physicians are unknown. Objective: To examine the association between and persistence of new-onset and long-term depressive symptoms among interns. Design, Setting, and Participants: The ongoing Intern Health Study (IHS) is a prospective annual cohort study that assesses the mental health of incoming US-based resident physicians. The IHS began in 2007, and a total of 105 residency programs have been represented in this national study. Interns enrolled sequentially in annual cohorts and completed follow-up surveys to screen for depression using the 9-item Patient Health Questionnaire-9 (PHQ-9) throughout and after medical training. The data were analyzed from May 2023 to March 2024. Exposure: A positive screening result for depression, defined as an elevated PHQ-9 score of 10 or greater (indicating moderate to severe depression) at 1 or more time points during the first postgraduate year of medical training (ie, the intern year). Main Outcomes and Measures: The main outcomes assessed were mean PHQ-9 scores (continuous) and proportions of physicians with an elevated PHQ-9 score (≥10; categorical or binary) at the time of the annual follow-up survey. To account for repeated measures over time, a linear mixed model was used to analyze mean PHQ-9 scores and a generalized estimating equation (GEE) was used to analyze the binary indicator for a PHQ-9 score of 10 or greater. Results: This study included 858 physicians with a PHQ-9 score of less than 10 before the start of their internship. Their mean (SD) age was 27.4 (9.0) years, and more than half (53.0% [95% CI, 48.5%-57.5%]) were women. Over the follow-up period, mean PHQ-9 scores did not return to the baseline level assessed before the start of the internship in either group (those with a positive depression screen as interns and those without). Among interns who screened positive for depression (PHQ-9 score ≥10) during their internship, mean PHQ-9 scores were significantly higher at both 5 years (4.7 [95% CI, 4.4-5.0] vs 2.8 [95% CI, 2.5-3.0]; P < .001) and 10 years (5.1 [95% CI, 4.5-5.7] vs 3.5 [95% CI, 3.0-4.0]; P < .001) of follow-up. Furthermore, interns with an elevated PHQ-9 score (≥10) demonstrated a higher likelihood of meeting this threshold during each year of follow-up. Conclusions and Relevance: In this cohort study of IHS participants, a positive depression screening result during the intern year had long-term implications for physicians, including having persistently higher mean PHQ-9 scores and a higher likelihood of meeting this threshold again. These findings underscore the pressing need to address the mental health of physicians who experience depressive symptoms during their training and to emphasize the importance of interventions to sustain the health of physicians throughout their careers.

Intensive longitudinal assessment following index trauma to predict development of PTSD using machine learning.

There are significant challenges to identifying which individuals require intervention following exposure to trauma, and a need for strategies to identify and provide individuals at risk for developing PTSD with timely interventions. The present study seeks to identify a minimal set of trauma-related symptoms, assessed during the weeks following traumatic exposure, that can accurately predict PTSD. Participants were 2185 adults (Mean age=36.4 years; 64% women; 50% Black) presenting for emergency care following traumatic exposure. Participants received a 'flash survey' with 6-8 varying symptoms (from a pool of 26 trauma symptoms) several times per week for eight weeks following the trauma exposure (each symptom assessed ∼6 times). Features (mean, sd, last, worst, peak-end scores) from the repeatedly assessed symptoms were included as candidate variables in a CART machine learning analysis to develop a pragmatic predictive algorithm. PTSD (PCL-5 ≥38) was present for 669 (31%) participants at the 8-week follow-up. A classification tree with three splits, based on mean scores of nervousness, rehashing, and fatigue, predicted PTSD with an Area Under the Curve of 0.836. Findings suggest feasibility for a 3-item assessment protocol, delivered once per week, following traumatic exposure to assess and potentially facilitate follow-up care for those at risk.

Measuring algorithmic bias to analyze the reliability of AI tools that predict depression risk using smartphone sensed-behavioral data.

AI tools intend to transform mental healthcare by providing remote estimates of depression risk using behavioral data collected by sensors embedded in smartphones. While these tools accurately predict elevated symptoms in small, homogenous populations, recent studies show that these tools are less accurate in larger, more diverse populations. In this work, we show that accuracy is reduced because sensed-behaviors are unreliable predictors of depression across individuals; specifically the sensed-behaviors that predict depression risk are inconsistent across demographic and socioeconomic subgroups. We first identified subgroups where a developed AI tool underperformed by measuring algorithmic bias, where subgroups with depression were incorrectly predicted to be at lower risk than healthier subgroups. We then found inconsistencies between sensed-behaviors predictive of depression across these subgroups. Our findings suggest that researchers developing AI tools predicting mental health from behavior should think critically about the generalizability of these tools, and consider tailored solutions for targeted populations.

Measuring algorithmic bias to analyze the reliability of AI tools that predict depression risk using smartphone sensed-behavioral data.

AI tools intend to transform mental healthcare by providing remote estimates of depression risk using behavioral data collected by sensors embedded in smartphones. While these tools accurately predict elevated depression symptoms in small, homogenous populations, recent studies show that these tools are less accurate in larger, more diverse populations. In this work, we show that accuracy is reduced because sensed-behaviors are unreliable predictors of depression across individuals: sensed-behaviors that predict depression risk are inconsistent across demographic and socioeconomic subgroups. We first identified subgroups where a developed AI tool underperformed by measuring algorithmic bias, where subgroups with depression were incorrectly predicted to be at lower risk than healthier subgroups. We then found inconsistencies between sensed-behaviors predictive of depression across these subgroups. Our findings suggest that researchers developing AI tools predicting mental health from sensed-behaviors should think critically about the generalizability of these tools, and consider tailored solutions for targeted populations.

Trends in Sexual Harassment Prevalence and Recognition During Intern Year.

This cohort study uses Internal Health Study and Sexual Experiences Questionnaire data to assess changes in sexual harassment prevalence and recognition among training physicians.

Multi-ancestry genome-wide association study of major depression aids locus discovery, fine mapping, gene prioritization and causal inference.

Most genome-wide association studies (GWAS) of major depression (MD) have been conducted in samples of European ancestry. Here we report a multi-ancestry GWAS of MD, adding data from 21 cohorts with 88,316 MD cases and 902,757 controls to previously reported data. This analysis used a range of measures to define MD and included samples of African (36% of effective sample size), East Asian (26%) and South Asian (6%) ancestry and Hispanic/Latin American participants (32%). The multi-ancestry GWAS identified 53 significantly associated novel loci. For loci from GWAS in European ancestry samples, fewer than expected were transferable to other ancestry groups. Fine mapping benefited from additional sample diversity. A transcriptome-wide association study identified 205 significantly associated novel genes. These findings suggest that, for MD, increasing ancestral and global diversity in genetic studies may be particularly important to ensure discovery of core genes and inform about transferability of findings.

Unraveling the interplay of circadian rhythm and sleep deprivation on mood: A Real-World Study on first-year physicians.

The interplay between circadian rhythms, time awake, and mood remains poorly understood in the real-world. Individuals in high-stress occupations with irregular schedules or nighttime shifts are particularly vulnerable to depression and other mood disorders. Advances in wearable technology have provided the opportunity to study these interactions outside of a controlled laboratory environment. Here, we examine the effects of circadian rhythms and time awake on mood in first-year physicians using wearables. Continuous heart rate, step count, sleep data, and daily mood scores were collected from 2,602 medical interns across 168,311 days of Fitbit data. Circadian time and time awake were extracted from minute-by-minute wearable heart rate and motion measurements. Linear mixed modeling determined the relationship between mood, circadian rhythm, and time awake. In this cohort, mood was modulated by circadian timekeeping (p<0.001). Furthermore, we show that increasing time awake both deteriorates mood (p<0.001) and amplifies mood's circadian rhythm nonlinearly. These findings demonstrate the contributions of both circadian rhythms and sleep deprivation to underlying mood and show how these factors can be studied in real-world settings using Fitbits. They underscore the promising opportunity to harness wearables in deploying chronotherapies for psychiatric illness.

Returning Individual Research Results from Digital Phenotyping in Psychiatry.

Psychiatry is rapidly adopting digital phenotyping and artificial intelligence/machine learning tools to study mental illness based on tracking participants' locations, online activity, phone and text message usage, heart rate, sleep, physical activity, and more. Existing ethical frameworks for return of individual research results (IRRs) are inadequate to guide researchers for when, if, and how to return this unprecedented number of potentially sensitive results about each participant's real-world behavior. To address this gap, we convened an interdisciplinary expert working group, supported by a National Institute of Mental Health grant. Building on established guidelines and the emerging norm of returning results in participant-centered research, we present a novel framework specific to the ethical, legal, and social implications of returning IRRs in digital phenotyping research. Our framework offers researchers, clinicians, and Institutional Review Boards (IRBs) urgently needed guidance, and the principles developed here in the context of psychiatry will be readily adaptable to other therapeutic areas.

Unspoken Truths: Mental Health Among Academic Surgeons.

OBJECTIVE: To characterize the current state of mental health within the surgical workforce in the United States. BACKGROUND: Mental illness and suicide is a growing concern in the medical community; however, the current state is largely unknown. METHODS: Cross-sectional survey of the academic surgery community assessing mental health, medical error, and suicidal ideation. The odds of suicidal ideation adjusting for sex, prior mental health diagnosis, and validated scales screening for depression, anxiety, post-traumatic stress disorder (PTSD), and alcohol use disorder were assessed. RESULTS: Of 622 participating medical students, trainees, and surgeons (estimated response rate=11.4%-14.0%), 26.1% (141/539) reported a previous mental health diagnosis. In all, 15.9% (83/523) of respondents screened positive for current depression, 18.4% (98/533) for anxiety, 11.0% (56/510) for alcohol use disorder, and 17.3% (36/208) for PTSD. Medical error was associated with depression (30.7% vs. 13.3%, P <0.001), anxiety (31.6% vs. 16.2%, P =0.001), PTSD (12.8% vs. 5.6%, P =0.018), and hazardous alcohol consumption (18.7% vs. 9.7%, P =0.022). Overall, 13.2% (73/551) of respondents reported suicidal ideation in the past year and 9.6% (51/533) in the past 2 weeks. On adjusted analysis, a previous history of a mental health disorder (aOR: 1.97, 95% CI: 1.04-3.65, P =0.033) and screening positive for depression (aOR: 4.30, 95% CI: 2.21-8.29, P <0.001) or PTSD (aOR: 3.93, 95% CI: 1.61-9.44, P =0.002) were associated with increased odds of suicidal ideation over the past 12 months. CONCLUSIONS: Nearly 1 in 7 respondents reported suicidal ideation in the past year. Mental illness and suicidal ideation are significant problems among the surgical workforce in the United States.

Variation by Institution in Sexual Harassment Experiences Among US Medical Interns.

This cross-sectional study investigates possible institutional and specialty variations in experiences of sexual harassment among US medical interns.

The impact of COVID-19 on a college freshman sample reveals genetic and nongenetic forms of susceptibility and resilience to stress.

Using a longitudinal approach, we sought to define the interplay between genetic and nongenetic factors in shaping vulnerability or resilience to COVID-19 pandemic stress, as indexed by the emergence of symptoms of depression and/or anxiety. University of Michigan freshmen were characterized at baseline using multiple psychological instruments. Subjects were genotyped, and a polygenic risk score for depression (MDD-PRS) was calculated. Daily physical activity and sleep were captured. Subjects were sampled at multiple time points throughout the freshman year on clinical rating scales, including GAD-7 and PHQ-9 for anxiety and depression, respectively. Two cohorts (2019 to 2021) were compared to a pre-COVID-19 cohort to assess the impact of the pandemic. Across cohorts, 26 to 40% of freshmen developed symptoms of anxiety or depression (N = 331). Depression symptoms significantly increased in the pandemic years and became more chronic, especially in females. Physical activity was reduced, and sleep was increased by the pandemic, and this correlated with the emergence of mood symptoms. While low MDD-PRS predicted lower risk for depression during a typical freshman year, this genetic advantage vanished during the pandemic. Indeed, females with lower genetic risk accounted for the majority of the pandemic-induced rise in depression. We developed a model that explained approximately half of the variance in follow-up depression scores based on psychological trait and state characteristics at baseline and contributed to resilience in genetically vulnerable subjects. We discuss the concept of multiple types of resilience, and the interplay between genetic, sex, and psychological factors in shaping the affective response to different types of stressors.

The Dosing of Mobile-Based Just-in-Time Adaptive Self-Management Prompts for Caregivers: Preliminary Findings From a Pilot Microrandomized Study.

BACKGROUND: Caregivers of people with chronic illnesses often face negative stress-related health outcomes and are unavailable for traditional face-to-face interventions due to the intensity and constraints of their caregiver role. Just-in-time adaptive interventions (JITAIs) have emerged as a design framework that is particularly suited for interventional mobile health studies that deliver in-the-moment prompts that aim to promote healthy behavioral and psychological changes while minimizing user burden and expense. While JITAIs have the potential to improve caregivers' health-related quality of life (HRQOL), their effectiveness for caregivers remains poorly understood. OBJECTIVE: The primary objective of this study is to evaluate the dose-response relationship of a fully automated JITAI-based self-management intervention involving personalized mobile app notifications targeted at decreasing the level of caregiver strain, anxiety, and depression. The secondary objective is to investigate whether the effectiveness of this mobile health intervention was moderated by the caregiver group. We also explored whether the effectiveness of this intervention was moderated by (1) previous HRQOL measures, (2) the number of weeks in the study, (3) step count, and (4) minutes of sleep. METHODS: We examined 36 caregivers from 3 disease groups (10 from spinal cord injury, 11 from Huntington disease, and 25 from allogeneic hematopoietic cell transplantation) in the intervention arm of a larger randomized controlled trial (subjects in the other arm received no prompts from the mobile app) designed to examine the acceptability and feasibility of this intensive type of trial design. A series of multivariate linear models implementing a weighted and centered least squares estimator were used to assess the JITAI efficacy and effect. RESULTS: We found preliminary support for a positive dose-response relationship between the number of administered JITAI messages and JITAI efficacy in improving caregiver strain, anxiety, and depression; while most of these associations did not meet conventional levels of significance, there was a significant association between high-frequency JITAI and caregiver strain. Specifically, administering 5-6 messages per week as opposed to no messages resulted in a significant decrease in the HRQOL score of caregiver strain with an estimate of -6.31 (95% CI -11.76 to -0.12; P=.046). In addition, we found that the caregiver groups and the participants' levels of depression in the previous week moderated JITAI efficacy. CONCLUSIONS: This study provides preliminary evidence to support the effectiveness of the self-management JITAI and offers practical guidance for designing future personalized JITAI strategies for diverse caregiver groups. TRIAL REGISTRATION: ClinicalTrials.gov NCT04556591; https://clinicaltrials.gov/ct2/show/NCT04556591.

Use of mobile technology to identify behavioral mechanisms linked to mental health outcomes in Kenya: protocol for development and validation of a predictive model.

OBJECTIVE: This study proposes to identify and validate weighted sensor stream signatures that predict near-term risk of a major depressive episode and future mood among healthcare workers in Kenya. APPROACH: The study will deploy a mobile application (app) platform and use novel data science analytic approaches (Artificial Intelligence and Machine Learning) to identifying predictors of mental health disorders among 500 randomly sampled healthcare workers from five healthcare facilities in Nairobi, Kenya. EXPECTATION: This study will lay the basis for creating agile and scalable systems for rapid diagnostics that could inform precise interventions for mitigating depression and ensure a healthy, resilient healthcare workforce to develop sustainable economic growth in Kenya, East Africa, and ultimately neighboring countries in sub-Saharan Africa. This protocol paper provides an opportunity to share the planned study implementation methods and approaches. CONCLUSION: A mobile technology platform that is scalable and can be used to understand and improve mental health outcomes is of critical importance.

Posttraumatic Stress Disorder Symptoms Among First-Year Resident Physicians Working Before and During the COVID-19 Pandemic.

This cohort study investigates differences in posttraumatic stress disorder (PTSD) symptoms among first-year resident physicians training before and during the first wave of the COVID-19 pandemic (March to June 2020).

The reliability and validity of the TBI-CareQOL system in four diverse caregiver groups.

PURPOSE: Establishing the psychometric reliability and validity of new measures is an ongoing process. More work is needed in to confirm the clinical utility of the TBI-CareQOL measurement development system in both an independent cohort of caregivers of traumatic brain injury (TBI), as well as in additional caregiver groups. METHODS: An independent cohort of caregivers of people with TBI (n = 139), as well as three new diverse caregiver cohorts (n = 19 caregivers of persons with spinal cord injury, n = 21 caregivers for persons with Huntington disease, and n = 30 caregivers for persons with cancer), completed 11 TBI-CareQOL measures (caregiver strain; caregiver-specific anxiety; anxiety; depression; anger; self-efficacy; positive affect and well-being; perceived stress; satisfaction with social roles and activities; fatigue; sleep-related impairment), as well as two additional measures to examine convergent and discriminant validity (PROMIS Global Health; the Caregiver Appraisal Scale). RESULTS: Findings support the internal consistency reliability (all alphas > 0.70 with the vast majority being > 0.80 across the different cohorts) of the TBI-CareQOL measures. All measures were free of ceiling effects, and the vast majority were also free of floor effects. Convergent validity was supported by moderate to high correlations between the TBI-CareQOL and related measures, while discriminant validity was supported by low correlations between the TBI-CareQOL measures and unrelated constructs. CONCLUSION: Findings indicate that the TBI-CareQOL measures have clinical utility in caregivers of people with TBI, as well as in other caregiver groups. As such, these measures should be considered as important outcome measures for clinical trials aiming to improve caregiver outcomes.

Barriers to Disclosure of Disability and Request for Accommodations Among First-Year Resident Physicians in the US.

Importance: Ensuring access to accommodations is critical for resident physicians and their patients. Studies show that a large proportion of medical trainees with disabilities do not request needed accommodations; however, drivers of nonrequests are unknown. Objective: To assess the frequency of accommodation requests among first-year resident physicians (ie, interns) with disabilities and to identify possible drivers of nonrequest for needed accommodations. Design, Setting, and Participants: As part of the Intern Health Study, a longitudinal cohort study of first-year resident physicians, residents at 86 surgical and nonsurgical residency programs in 64 US institutions provided demographic and training characteristics 2 months prior to matriculation (April-May 2021). At the end of their intern year (June 2022), participants completed a new survey with questions about disability-related information, including disability status, disability type, whether they received accommodations, and if not, reasons for nonaccommodation. Poststratification and attrition weights were used to estimate the frequency of accommodation requests and reasons for not requesting accommodations. Interns reporting at least 1 disability were included in the analysis. Main Outcomes and Measures: Prevalence of reported disabilities, residency specialties distribution, frequency of accommodation requests, and reasons for nonaccommodation among resident physicians with disabilities. Results: Among the 1486 resident physicians who completed the baseline survey, 799 (53.8%) replied to the disability questions. Of those, 94 interns (11.8%; weighted number, 173 [11.9%]) reported at least 1 disability and were included in the present study (weighted numbers, 91 [52.6%] men, 82 [47.4%] women, mean [SD] age, 28.6 [3.0] years). Among interns with reported disability and need for accommodations (83 of 173 [48.0%]), more than half (42 [50.6%]) did not request them. The most frequently reported reasons for not requesting needed accommodations were fear of stigma or bias (25 [59.5%]), lack of a clear institutional process for requesting accommodations (10 [23.8%]), and lack of documentation (5 [11.9%]). Conclusions and Relevance: Program directors should investigate cultural and structural factors within their programs that contribute to an environment where residents do not feel safe or supported in disclosing disability and requesting accommodation and review their disability policies for clarity.

Exploring the Relationship Between Privacy and Utility in Mobile Health: Algorithm Development and Validation via Simulations of Federated Learning, Differential Privacy, and External Attacks.

BACKGROUND: Although evidence supporting the feasibility of large-scale mobile health (mHealth) systems continues to grow, privacy protection remains an important implementation challenge. The potential scale of publicly available mHealth applications and the sensitive nature of the data involved will inevitably attract unwanted attention from adversarial actors seeking to compromise user privacy. Although privacy-preserving technologies such as federated learning (FL) and differential privacy (DP) offer strong theoretical guarantees, it is not clear how such technologies actually perform under real-world conditions. OBJECTIVE: Using data from the University of Michigan Intern Health Study (IHS), we assessed the privacy protection capabilities of FL and DP against the trade-offs in the associated model's accuracy and training time. Using a simulated external attack on a target mHealth system, we aimed to measure the effectiveness of such an attack under various levels of privacy protection on the target system and measure the costs to the target system's performance associated with the chosen levels of privacy protection. METHODS: A neural network classifier that attempts to predict IHS participant daily mood ecological momentary assessment score from sensor data served as our target system. An external attacker attempted to identify participants whose average mood ecological momentary assessment score is lower than the global average. The attack followed techniques in the literature, given the relevant assumptions about the abilities of the attacker. For measuring attack effectiveness, we collected attack success metrics (area under the curve [AUC], positive predictive value, and sensitivity), and for measuring privacy costs, we calculated the target model training time and measured the model utility metrics. Both sets of metrics are reported under varying degrees of privacy protection on the target. RESULTS: We found that FL alone does not provide adequate protection against the privacy attack proposed above, where the attacker's AUC in determining which participants exhibit lower than average mood is over 0.90 in the worst-case scenario. However, under the highest level of DP tested in this study, the attacker's AUC fell to approximately 0.59 with only a 10% point decrease in the target's R2 and a 43% increase in model training time. Attack positive predictive value and sensitivity followed similar trends. Finally, we showed that participants in the IHS most likely to require strong privacy protection are also most at risk from this particular privacy attack and subsequently stand to benefit the most from these privacy-preserving technologies. CONCLUSIONS: Our results demonstrated both the necessity of proactive privacy protection research and the feasibility of the current FL and DP methods implemented in a real mHealth scenario. Our simulation methods characterized the privacy-utility trade-off in our mHealth setup using highly interpretable metrics, providing a framework for future research into privacy-preserving technologies in data-driven health and medical applications.

Peak-end bias in retrospective recall of depressive symptoms on the PHQ-9.

Mental health care is built around patient recall and report of clinical symptoms. However, memories of events and experiences rely on cognitive heuristics that influence our recall. The peak-end bias, which refers to the tendency for the most intense and proximate aspects of an experience to disproportionately influence our memory, has been understudied in the context of mental health symptoms and may unduly influence self-reported symptoms, even in the context of standardized assessments. To determine whether the peak-end bias applies to the report of depressive symptoms on the standardized Patient Health Questionnaire-9 (PHQ-9) assessment, we compared two scores from daily mood assessments collected over a 2-week period from 4,322 medical interns (56% women; 60% non-Hispanic White). The peak-end-mood score, which averaged the single lowest and most recent mood scores over 2 weeks had a significantly stronger correlation with the PHQ-9 than the mean-mood score, which averaged all mood scores during the 2 weeks. Likelihood ratio tests and fit statistics provided further support that the peak-end-mood score was a significantly better predictor of depression than the mean-mood score. Results were consistent when limiting the sample to those with mild-to-severe depressive symptoms, and when only examining the two primary mood items as the dependent variable. These findings provide evidence for a modest peak-end recall bias for mood and depressive symptoms. There may be benefits to implementing intermittent assessment strategies to support clinical decision-making. (PsycInfo Database Record (c) 2023 APA, all rights reserved).