Theory-driven computational psychiatry attempts to use cognitive models of computation to understand how mental health problems might relate to (or be caused by) changes in cognitive processes such as learning and decision-making. However, the potential applications and relevance of this approach are contingent on several (often implicit) assumptions, including that computational parameters 1) are recoverable, 2) are reliable over time, 3) reflect conceptually similar processes across different tasks, and 4) relate to symptoms. To illustrate and test these assumptions for a selection of commonly used tasks, we recruited a large online sample of participants (n=548), who completed seven mental health questionnaires and five tasks. A subset of n=115 was re-invited to complete the five tasks 14 days later. For each task, five models (including a null model, or model of no interest) were fit and the winning model was selected through Bayesian model comparison. The parameters from the winning models showed good recovery (mean r=0.853, sd=0.135), and good to excellent levels of test-retest reliability (mean ICC=0.635, sd=0.30). Parameters that were theoretically or mathematically similar to each other were, however, generally not related, with relationships only found within the class of ‘inverse temperature’-like parameters (significant r values ranged from 0.11 to 0.23): indicating that in general, parameters do not generalise across task contexts. Finally, only parameters from two of the five tasks (four-armed bandit and cognitive effort) related to symptoms, and these relationships were weak (maximum absolute r values ~0.10). To conclude, it seems that at least some of the implicit assumptions when advocating for the use of computational models in psychiatry are not met for these popular tasks and computational models. This limits the clinical and translational utility of this approach. Computational psychiatry researchers should carefully assess the assumptions and psychometric properties of their tasks and models as part of their early-phase development to ensure robustness of the field going forward.

Anhedonia, a core symptom of mood disorders such as depression, is marked by diminished pleasure and motivation for rewards. Traditional reinforcement learning (RL) tasks, like the 4-arm bandit (4AB), show limited sensitivity to reward-processing impairments associated with anhedonia. Here, we developed and validated a modified 3-arm bandit (3AB) task designed to reduce cognitive load while retaining sensitivity to reward and punishment learning. Following this validation, 1,000 participants were pre-screened with the Snaith-Hamilton Pleasure Scale (SHAPS), Dimensional Anhedonia Rating Scale (DARS), Generalized Anxiety Disorder Assessment (GAD), and Zung Self-Rating Depression Scale (ZUNG). Participants scoring above 2 on SHAPS and below 45 on DARS were classified as anhedonic, while those scoring 0 on SHAPS and above 55 on DARS were classified as non-anhedonic, resulting in 111 anhedonic and 95 non-anhedonic individuals who completed the 3AB task. Modelling results revealed no significant group differences in reward learning rate (p = 0.23), punishment learning rate (p = 0.37), reward sensitivity (p = 0.28), or punishment sensitivity (p = 0.46). Additional assessments of win-stay/lose-shift strategies and reaction times also showed no significant differences. Bayes Factor t-tests provided moderate -to-strong evidence for the null hypothesis, with BF01 values of 3.36, 5.14, 4.97, and 5.96 for each of the above parameters, respectively. These findings indicate either that anhedonia does not impair reward and punishment learning or that the 3AB task lacks the sensitivity to detect such differences. By simplifying the task structure while maintaining core learning mechanisms, the 3AB task provides a novel approach for studying reward processing, challenging the notion that anhedonia impairs reward sensitivity and indicating that this learning mechanism may remain intact.

When adolescents start a new school or arrive at university, they must navigate new campuses, new peers and new routines. Exploration is recognised as a key cognitive process supporting adolescent development and wellbeing but how real-world exploration relates to emotional adjustment during major life transitions remains unclear. To address this question, we tracked adolescents entering high school and university (N=64) over the first three months of transition using continuous smartphone geolocation and ecological momentary assessment (EMA). We quantified daily exploration as the dispersion of time in a day across locations (“roaming entropy”) and the number of newly visited locations (“novelty”) and assessed in-the-moment affect, autonomy and new social connections using EMA. On days when individuals visited more novel locations than usual, they reported higher positive affect, lower negative affect and more new social connections. Novelty was a stronger predictor of these outcomes than roaming entropy. Associations between exploration and positive affect were more pronounced in university students than in high school students, consistent with greater autonomy during this transition. Greater anxiety and depressive symptoms were not associated with lower exploration, but attenuated the positive associations between novelty, affect and perceived autonomy. These findings link theories and laboratory studies of adolescent exploration and mental health to naturalistic behaviour, showing that engagement with novel environments during major transitions is linked to positive emotional and social experiences. Importantly, these associations were weaker when depression and anxiety symptoms were greater. Real-world exploration may therefore represent a behavioural pathway supporting adjustment as young people navigate environmental change.

Generalised anxiety disorder (GAD) is common and increasingly recognised in primary care. Although antidepressants and psychological therapies are first-line treatments, many patients have residual anxiety and limited access to therapy. Evidence for effective next-step pharmacological options for treatment resistant anxiety remains limited. This paper describes the protocol for the PETRA trial, which will evaluate the clinical and cost-effectiveness of adding pregabalin to antidepressant treatment, compared with placebo, for people with GAD who have not responded or partially responded to antidepressant treatment. PETRA is a multicentre, individually randomised, double-blind, placebo-controlled superiority trial conducted in UK primary care. Participants are recruited by the study team from approximately 150 General Practices and randomised in a 1:1 ratio, using minimisation, to either pregabalin or a matching placebo for 26 weeks (followed by a tapering period where their dosage is reduced over approximately 4 weeks). Sample size is 498. Eligible participants are adults aged 18–74 years, who have been taking antidepressant medication for at least 8 weeks, received treatment with at least one other antidepressant before their current antidepressant and meet ICD-11 criteria for GAD and score ≥ 12 on the revised clinical interview schedule (CIS-R) total score. Follow-up assessments are at 3, 6, 12, 26 and 30 weeks. Our primary outcome measure will be anxiety symptoms measured with GAD-7 at 12 weeks (continuous score). Secondary outcomes are anxiety (GAD-7) at other time points, depressive and panic symptoms, suicidal thoughts, self-rated global improvement, adherence to study medication, serious adverse events, adverse effects, alcohol consumption and benzodiazepine use, quality of life and resources and costs used. The 30-week assessment will investigate symptoms during the withdrawal from pregabalin. Exploratory analyses will include cognitive tasks. A cost-effectiveness analysis and a nested qualitative study will evaluate the implementation and intervention acceptability. The trial findings will inform primary care prescribing practice by providing an accurate and generalisable estimate of the clinical and cost-effectiveness of prescribing pregabalin to individuals with generalised anxiety who have not responded or only partially responded to antidepressant treatment. Controlled Trials ISRCTN Registry, ISRCTN 16993990, registered on 19/09/2023. First participant enrolled in January 2024.

Anhedonia, a transdiagnostic symptom marked by diminished reward sensitivity, is often linked to impairments in reinforcement learning (RL). Standard tasks (e.g., the 4-arm bandit) can place substantial demands on participants and may blur valuation with other processes. We therefore adapted a three-arm bandit (3AB) task from Seymour et al. (2012), incorporating design features intended to lessen task demands (fewer options; denser feedback) while enabling separate estimation of reward and punishment learning rates and sensitivities. In an online sample pre-screened for anhedonia (N = 206; 111 anhedonic, 95 non-anhedonic), hierarchical Bayesian modelling using a four-parameter specification showed no credible group differences in reward learning rate, punishment learning rate, reward sensitivity, or punishment sensitivity; Bayes factors favoured the null (BF01 = 3.36–5.96). Model-agnostic win-stay/lose-shift strategies likewise showed no group differences (Welch’s tests, all p > .05). Posterior predictive checks indicated above-chance choice prediction: the model’s highest-probability action matched participants’ actual choices on 59.6% of trials (chance = 33%). Parameter recovery was excellent for valuation parameters (r = 0.96–0.97) and acceptable for learning rates (r = 0.67–0.85). Simulations generated from fitted parameters preserved individual-difference structure, with high correlations between observed and simulated win-stay (r = 0.89 anhedonic; 0.86 non-anhedonic) and moderate correlations for lose-shift (r = 0.62; 0.67), alongside small systematic mean-level biases (simulated win-stay lower by 3.5–4.9 percentage points; simulated lose-shift higher by 12.8–13.2 points). Model comparison showed that lapse-augmented variants achieved marginally better predictive fit, but group comparisons under both lapse models yielded overlapping posteriors with 95% HDIs including zero for all learning, sensitivity, and lapse parameters, indicating that the null findings were robust to inclusion of lapse terms. Non-anhedonic participants also responded more slowly on average than anhedonic participants, which we treat as exploratory. Together, these results suggest that in this 3AB task, anhedonia is not reliably associated with differences in core RL parameters or simple choice strategies, while providing a detailed characterisation of model performance and limitations in an online setting.

BACKGROUND:Motivational dysfunction is a core feature of depression and can have debilitating effects on everyday function. However, it is unclear which cognitive processes underlie impaired motivation and whether impairments persist following remission. Decision-making concerning exerting effort to obtain rewards offers a promising framework for understanding motivation, especially when examined with computational tools. METHODS:Effort-based decision-making was assessed using the Apple Gathering Task, where participants decide whether to exert effort via a grip-force device to obtain varying levels of reward; effort levels were individually calibrated and varied parametrically. We present a comprehensive computational analysis of decision-making, initially validating our model in healthy volunteers (N = 67), before applying it in a case-control study including current (N = 41) and remitted (N = 46) unmedicated depressed individuals and healthy volunteers with (N = 36) and without (N = 57) a family history of depression. RESULTS:Four fundamental computational mechanisms that drive patterns of effort-based decisions, which replicated across samples, were identified: overall bias to accept effort challenges; reward sensitivity; and linear and quadratic effort sensitivity. Traditional model-agnostic analyses showed that both depressed groups showed lower willingness to exert effort. In contrast with previous findings, computational analysis revealed that this difference was primarily driven by lower effort-acceptance bias, but not altered effort or reward sensitivity. CONCLUSIONS:This work provides insight into the computational mechanisms underlying motivational dysfunction in depression. Lower willingness to exert effort could represent a trait-like factor contributing to symptoms and a fruitful target for treatment and prevention.

Childhood maltreatment is an established risk factor for poor mental health, yet not all maltreatment-exposed children develop psychiatric symptoms. This highlights the need for individualised metrics to detect those at heightened risk. Previous efforts to develop such metrics have largely focused on maltreatment severity and aggregate demographic risk. Although neurocognitive alterations in domains like social cognition and reward processing have been linked to maltreatment, most research has examined cross-sectional group-level differences in isolated domains. This proof-of-concept study examines whether indexing combined neurocognitive vulnerability can help predict which maltreatment-exposed children are most at risk of future symptom increases. A sample (mean age = 12.8) of maltreatment-exposed children (MT = 85) and propensity-score matched (PSM) non-maltreated peers (NMT = 90) were assessed at baseline using five neurocognitive tasks spanning various domains of social cognition and reward processing. A combined neurocognitive vulnerability metric was created, and symptom changes were measured 1.5 years later for a subset of individuals (MT = 57, NMT = 47). Combined neurocognitive vulnerability predicted future mental health symptom increases in the maltreatment-exposed group, even after controlling for baseline symptoms and clinical status. Maltreatment-exposed children in the high neurocognitive vulnerability category had a 92.1% adjusted predicted probability of experiencing worsening symptoms over time. Furthermore, we found that children in the maltreatment-exposed group were more likely to fall into medium and high neurocognitive vulnerability categories than non-maltreated peers. These preliminary findings underscore the potential of using combined neurocognitive vulnerability metrics to identify maltreatment-exposed children at the increased risk for future mental health problems, supporting its role in prevention strategies.

Willingness to exert effort for a given goal is dependent on the magnitude of the potential rewards and effort costs of an action. Such effort-based decision making is an essential component of motivation, in which the dopaminergic system plays a key role. Depression in Parkinson's disease (PD) is common, disabling and has poor outcomes. Motivational symptoms such as apathy and anhedonia are prominent in PD depression and related to dopaminergic loss. We hypothesized that dopamine-dependent disruption in effort-based decision making contributes to depression in PD. In the present study, an effort-based decision-making task was administered to 62 patients with PD, with and without depression, ON and OFF their dopaminergic medication across two sessions, as well as to 34 patients with depression and 29 matched controls on a single occasion. During the task, on each trial, participants decided whether to accept or reject offers of different levels of monetary reward in return for exerting varying levels of physical effort via grip force, measured using individually calibrated dynamometers. The primary outcome variable was choice (accept/decline offer), analysed using both logistic mixed-effects modelling and a computational model which dissected the individual contributions of reward and effort on depression and dopamine state in PD. We found PD depression was characterized by lower acceptance of offers, driven by markedly lower incentivization by reward (reward sensitivity), compared to all other groups. Within-subjects analysis of the effect of dopamine medication revealed that, although dopamine treatment improves reward sensitivity in non-depressed PD patients, this therapeutic effect is not present in PD patients with depression. These findings suggest that disrupted effort-based decision making, unresponsive to dopamine, contributes to PD depression. This highlights reward sensitivity as a key mechanism and treatment target for PD depression that potentially requires non-dopaminergic therapies.

Several mental health conditions seen in older people are associated with impaired emotion regulation. Heart rate variability (HRV) may be an index of emotion regulation capacity, but it is unclear whether and how aging influences this association. Early neurodegenerative processes, such as Alzheimer's disease-related reduction of locus coeruleus (LC) integrity, may play a role, as LC modulates both HRV and self-regulatory networks. We pre-registered a cross-sectional study to investigate the relationship between measures of emotion regulation, HRV and LC structural MRI integrity in a lifespan sample of cognitively normal healthy adults (n = 678, aged 18-88 years, 51% female), recruited between 2010 and 2012 as part of the Cambridge Centre for Ageing and Neuroscience (Cam-CAN) cohort. We hypothesized that age-related differences in the HRV-emotion regulation relationship could be attributed to reduced LC integrity in older versus younger adults. Exploratory analyses incorporated alternative and novel measures of emotionality and LC rostro-caudal functional connectivity gradients from more recent Cam-CAN studies. In contrast to younger adults, we found an inverse relationship between resting HRV and measures of emotion regulation performance in older adults. There was no evidence that LC integrity influenced this relationship. A more 'old-like' LC rostro-caudal functional connectivity gradient, but not LC signal intensity, was related to lower HRV and worse reappraisal outcomes. We identify complexity in the association between HRV and emotion regulation with age and gaps in understanding of the relationship between different measures of LC integrity. Future studies should explore compensatory mechanisms underlying age-related differences in autonomic and emotion regulation.

Pavlovian biases are patterns of behaviour that involve approaching stimuli associated with reward and avoiding those associated with punishment (regardless of whether this is actually optimal behaviour). Pavlovian biases are classically described as fixed and automatic, yet some studies indicate that their influence on behaviour can vary both over time and with task demands. While these results hint that people may in fact have some control over their Pavlovian biases, direct behavioural evidence is still lacking. Here, we tested a week-long cognitive bias training programme, in a preregistered, double-blind and sham-controlled design (N = 800 healthy adults). We found that the training led to significantly reduced Pavlovian biases (particularly avoidance bias) at follow-up. To our knowledge this is the first demonstration that people can learn to overcome their Pavlovian biases, and suggests greater flexibility in the way that Pavlovian biases affect cognition than had previously been appreciated.

Depression consists of heterogeneous symptoms that can occur in hundreds of possible combinations. However, intervention studies commonly operationalize depression as a homogeneous condition. Here we adopt a symptom-level approach to test the effects of the selective serotonin reuptake inhibitor sertraline on depression and anxiety symptoms and to test their associations. Using data from the PANDA randomized controlled trial, we use network models to estimate the effects of sertraline at different time points (contemporaneous networks at 2, 6 and 12 weeks) and across time (temporally lagged networks). Results show that sertraline has beneficial effects on core depression and anxiety symptoms as early as after 2 weeks of treatment, counteracted by detrimental effects on somatic symptoms of depression. This intricate pattern of treatment effects is typically masked when measuring depression on a single dimension. Focusing on individual symptoms of depression and anxiety may shed light on the nature, effectiveness and timing of antidepressant action. This research adopts a symptom-level approach to evaluate the effects of sertraline on depression and anxiety using network models from the PANDA trial, revealing early benefits for core symptoms while highlighting adverse impacts on somatic manifestations over time.

Optimal reward learning requires individuals to adjust their learning rates - the extent to which new information replaces old. Learning rates should be higher in volatile environments, where new information is more salient, and lower in stable environments, where the longer-term history of outcomes is more predictive. It is not known, however, whether this adjustment in learning rates changes with age or is associated with better mental health and social functioning. We administered a child-friendly probabilistic reinforcement learning task with both fixed and fluctuating reward schedules to 121 participants aged 8-16 years. Adjustment of learning rates across childhood and adolescence to suit the levels of uncertainty in the environment did not differ by age, nor was it associated with better mental health and social functioning. Instead we found that learning rates for worse-than-expected outcomes generally decreased with age, temperature increased with age and higher learning rates, specifically during positive stable environments, were associated with greater self-reported prosocial behaviour. Our results highlight the exaggerated impact of negative feedback on children and suggest an increase in exploratory behaviour between childhood and adolescence.

Humans often attach notions of value to hearing the voices of specific loved ones, yet there is sparse scientific evidence supporting these claims. We present three experiments-two behavioural and one neuroimaging functional magnetic resonance imaging (fMRI)-that tested whether personally-valued voices engage reward-motivated behaviour and associated brain responses. Using novel voice incentive delay tasks, we show that listeners respond faster in anticipation of hearing the speaking voice of their music idol than when anticipating an unfamiliar voice or a pure tone (Experiment 1). A second behavioural experiment indicated that familiarity alone was insufficient to engage stronger reward-motivated behaviour in comparison with an unfamiliar identity (Experiment 2). These behavioural patterns were further reflected in an fMRI experiment, where the idol voice condition most strongly engaged brain regions associated with reward processing while responses to other familiar and unfamiliar voice conditions were often equivalent (Experiment 3). Taken together, these studies provide evidence that voices can be effective rewards, in particular when they are associated with intense parasocial interest. Future research should determine whether these findings generalize to personally known individuals.

Background:Diminished capacity for maintaining positive affect (PA) has been identified in subthreshold depression (StD). While recent studies have explored affective dynamics among StD, the relationship between early emotional processing impairments and the capacity to prolong PA remains uncertain. Furthermore, it is unclear how brain connectivity patterns observed in StD are associated with PA maintenance. Methods:The experimental procedure comprised a baseline rs-fMRI scan, followed by a PA-inducing movie viewing task, and three further rs-fMRI sessions. Participants provided PA ratings following each session. PA maintenance was quantified through the slope of mood change between each session after movie viewing. We performed a dynamic functional connectivity analysis on movie viewing data, as well as a series of static functional connectivity (FC), analyses on data of all rs-fMRI sessions from 25 StD and 25 healthy controls (HC). Correlations between brain-related measures and slope of mood change were calculated. Results:Individuals with StD exhibited reduced capacity in sustaining PA, reflected in a decrease in PA in the early maintenance stage. StD also had a lower number of transitions between four brain states during movie viewing, which was related to subsequent impairment in sustaining PA. In addition, StD had weaker static FC between left inferior frontal gyrus and right middle occipital gyrus during the first resting-state session following movie viewing, which in turn was related to a steeper decline in PA. Conclusions:These results highlight the brain features driving PA dysregulation in StD and provide a potential avenue for the development of future interventions.

Depression in Parkinson disease (PD) is common, is disabling and responds poorly to standard antidepressants. Motivational symptoms of depression are particularly prevalent in PD and emerge with loss of dopaminergic innervation of the striatum. Optimizing dopaminergic treatment for PD can improve depressive symptoms. However, the differential effect of antiparkinsonian medication on symptom dimensions of depression is not known. Using data from a large (n = 412) longitudinal study of patients with newly diagnosed PD followed over 5 years, we investigated whether there are dissociable effects of dopaminergic medications on different depression symptom dimensions in PD. Previously validated 'motivation' and 'depression' dimensions were derived from the 15-item geriatric depression scale. Dopaminergic neurodegeneration was measured using repeated striatal dopamine transporter imaging. We identified dissociable associations between dopaminergic medications and different dimensions of depression in PD. Dopamine agonists were shown to be effective for treatment of motivational symptoms of depression. In contrast, monoamine oxidase-B inhibitors improved both depressive and motivation symptoms, albeit the latter effect is attenuated in patients with more severe striatal dopaminergic neurodegeneration.

Importance: Studies on polygenic risk for psychiatric traits commonly use a disorder-level approach to phenotyping, implicitly considering disorders as homogeneous constructs; however, symptom heterogeneity is ubiquitous, with many possible combinations of symptoms falling under the same disorder umbrella. Focusing on individual symptoms may shed light on the role of polygenic risk in psychopathology. Objective: To determine whether polygenic scores are associated with all symptoms of psychiatric disorders or with a subset of indicators and whether polygenic scores are associated with comorbid phenotypes via specific sets of relevant symptoms. Design, setting, and participants: Data from 2 population-based cohort studies were used in this cross-sectional study. Data from children in the Avon Longitudinal Study of Parents and Children (ALSPAC) were included in the primary analysis, and data from children in the Twins Early Development Study (TEDS) were included in confirmatory analyses. Data analysis was conducted from October 2021 to January 2024. Pregnant women based in the Southwest of England due to deliver in 1991 to 1992 were recruited in ALSPAC. Twins born in 1994 to 1996 were recruited in TEDS from population-based records. Participants with available genetic data and whose mothers completed the Short Mood and Feelings Questionnaire and the Strength and Difficulties Questionnaire when children were 11 years of age were included. Main outcomes and measures: Psychopathology relevant symptoms, such as hyperactivity, prosociality, depression, anxiety, and peer and conduct problems at age 11 years. Psychological networks were constructed including individual symptoms and polygenic scores for depression, anxiety, attention-deficit/hyperactivity disorder (ADHD), body mass index (BMI), and educational attainment in ALSPAC. Following a preregistered confirmatory analysis, network models were cross-validated in TEDS. Results: Included were 5521 participants from ALSPAC (mean [SD] age, 11.8 [0.14] years; 2777 [50.3%] female) and 4625 participants from TEDS (mean [SD] age, 11.27 [0.69] years; 2460 [53.2%] female). Polygenic scores were preferentially associated with restricted subsets of core symptoms and indirectly associated with other, more distal symptoms of psychopathology (network edges ranged between r = -0.074 and r = 0.073). Psychiatric polygenic scores were associated with specific cross-disorder symptoms, and nonpsychiatric polygenic scores were associated with a variety of indicators across disorders, suggesting a potential contribution of nonpsychiatric traits to comorbidity. For example, the polygenic score for ADHD was associated with a core ADHD symptom, being easily distracted (r = 0.07), and the polygenic score for BMI was associated with symptoms across disorders, including being bullied (r = 0.053) and not thinking things out (r = 0.041). Conclusions and relevance: Genetic associations observed at the disorder level may hide symptom-level heterogeneity. A symptom-level approach may enable a better understanding of the role of polygenic risk in shaping psychopathology and comorbidity.

BackgroundAdolescents are susceptible to mental illness and have experienced substantial disruption owing to the COVID-19 pandemic. The digital environment is increasingly important in the context of a pandemic when in-person social connection is restricted. ObjectiveThis study aims to estimate whether depression and anxiety had worsened compared with the prepandemic period and examine potential associations with sociodemographic characteristics and behavioral factors, particularly digital behaviors. MethodsWe analyzed cross-sectional and longitudinal data from a large, representative Greater London adolescent cohort study: the Study of Cognition, Adolescents and Mobile Phones (SCAMP). Participants completed surveys at T1 between November 2016 and July 2018 (N=4978; aged 13 to 15 years) and at T2 between July 2020 and June 2021 (N=1328; aged 16 to 18 years). Depression and anxiety were measured using the Patient Health Questionnaire and Generalized Anxiety Disorder scale, respectively. Information on the duration of total mobile phone use, social network site use, and video gaming was also collected using questionnaires. Multivariable logistic regression was used to assess the cross-sectional and longitudinal associations of sociodemographic characteristics, digital technology use, and sleep duration with clinically significant depression and anxiety. ResultsThe proportion of adolescents who had clinical depression and anxiety significantly increased at T2 (depression: 140/421, 33.3%; anxiety: 125/425, 29.4%) compared with the proportion of adolescents at T1 (depression: 57/421, 13.5%; anxiety: 58/425, 13.6%; P for 2-proportion z test <.001 for both depression and anxiety). Depression and anxiety levels were similar between the summer holiday, school opening, and school closures. Female participants had higher odds of new incident depression (odds ratio [OR] 2.5, 95% CI 1.5-4.18) and anxiety (OR 2.11, 95% CI 1.23-3.61) at T2. A high level of total mobile phone use at T1 was associated with developing depression at T2 (OR 1.89, 95% CI 1.02-3.49). Social network site use was associated with depression and anxiety cross-sectionally at T1 and T2 but did not appear to be associated with developing depression or anxiety longitudinally. Insufficient sleep at T1 was associated with developing depression at T2 (OR 2.26, 95% CI 1.31-3.91). ConclusionsThe mental health of this large sample of adolescents from London deteriorated during the pandemic without noticeable variations relating to public health measures. The deterioration was exacerbated in girls, those with preexisting high total mobile phone use, and those with preexisting disrupted sleep. Our findings suggest the necessity for allocating resources to address these modifiable factors and target high-risk groups.