Psychotic disorders are frequently preceded by depressive disorders, and it has been hypothesized that treatment of depression in youth may reduce risk for later psychosis. Using quasi-experimental methods, we estimated the causal relationship between the treatment of adolescent depression with selective serotonin reuptake inhibitors (SSRIs) and the risk of later psychosis.

We used data linkage from multiple national Finnish registries for all individuals (n = 697,289) born between 1987 and 1997 to identify depression diagnosed before age 18, cumulative SSRI treatment within three years of diagnosis, and diagnoses of non-affective psychotic disorders by end of follow-up (age 20–29). We used instrumental variable analyses, exploiting variability in prescribing across hospital districts to estimate causal effects. Analyses were conducted using two-stage least squares modelling. Sensitivity analyses examined effects stratified by confounders and effects of specific SSRIs.

Our final sample included 22,666 individuals diagnosed with depression in adolescence, of whom 60.2% (n = 13,650) had used SSRIs. 10.7% of adolescents with depression went on to be diagnosed with a non-affective psychotic disorder. SSRI treatment for adolescent depression was not associated with a reduced risk of developing a psychotic disorder (one-year β = 0.04,CI:−0.01 to 0.09; two-years β = 0.02,CI:−0.06 to 0.09; three-years β = −0.02,CI:−0.08 to 0.05).

Our quasi-experimental investigation does not support the hypothesis that treatment of adolescent depression reduces the subsequent risk of psychosis. Our findings question the assumption that treatment of common mental health disorders in youth may impact the risk of developing severe mental illnesses in adulthood.

The COVID-19 pandemic has impacted various aspects of daily life, leading to increased psychological symptoms and changes in alcohol use, yet little is known about their specific interactions, particularly early stages during the pandemic. We examined the relationship between psychological symptoms and alcohol-related behaviors associated with COVID-19, and determined whether associations shifted already early during the pandemic and whether changes in psychological symptoms from the pre- to during COVID-19 impacted changes in alcohol consumption.

Participants were young adults from a longitudinal cohort (N=435, age: 22–25) from two time points. We applied paired samples t-tests, correlation analyses, SHapley Additive exPlanations, and classification models to examine the multiple associations between psychological symptoms and alcohol use directly pre- and early during COVID-19.

We found significant associations between psychological symptoms and alcohol use pre- compared to during COVID-19. Anxiety was the strongest factor influencing alcohol use pre-pandemic, depression had the greatest impact during COVID-19. Changes in anxiety from pre- to during COVID-19 were the main factor associated with an increase in alcohol use, while changes in depression appeared to be most predictive for a decrease/persistence in alcohol use.

These findings suggest a shift in the association between psychological symptoms and alcohol use following COVID-19, as well as a differential impact of psychological symptoms, depending on their changes related to the pandemic. Changes in anxiety may contribute to riskier alcohol use behaviors following the pandemic, while depression appears to be one of the most critical factors influencing alcohol use during such crisis situations.

Common mental disorders (CMDs) are significant causes of work disability. Low socioeconomic status (SES) is a known risk factor for CMDs and work disability, one possible reason being poorer treatment adherence. We aimed to study the realization of pharmacological treatment and antidepressant adherence in patients with CMDs 3 years before and 3 years after being granted a disability pension (DP) and the role of SES in this. We also studied whether antidepressant adherence is associated with return to work (RTW) after a temporary DP.

Information on all persons granted a DP due to CMD between 2010 and 2012 in Finland (n = 12,388) was retrieved from national registers, which included medical, socioeconomic, and sociodemographic information of the subjects. We used the PRE2DUP method to estimate drug use periods and regression analyses to study associations between SES, taking medications, and RTW.

Prevalence of taking antidepressants increased towards the DP grant and decreased thereafter, but 14.6% of subjects did not take antidepressants or antipsychotics at all during the study period. Of SES factors, only income was positively associated with antidepressant adherence, lasting over a year. Antidepressant adherence was not associated with RTW.

An alarming result was the absence of recommended medication in fewer than every seventh patient estimated to be disabled due to pharmacologically treatable psychiatric disorders. Contrary to expectations, SES had only a minor predictive role in antidepressant adherence in this patient group. Contrary to taking antidepressants, rehabilitation was associated with RTW. The results adduced the importance of CMD treatment optimization regardless of SES.

Depression is characterized by disturbed emotion processing, with aberrant neural and physiological responses to emotional stimuli. Here, we applied an emotion anticipation and processing paradigm to investigate brain neural and electrodermal reactivities in patients with depression compared with healthy controls.

The study included 42 patients (27 females) and 44 healthy controls (21 females). Subjects underwent functional magnetic resonance imaging with simultaneous measurement of electrodermal activity. During scanning, red or green color cues were presented, followed by pictures of negative or positive valence, respectively. Behavioral valence and arousal ratings of the picture stimuli were conducted after scanning. Anhedonia was assessed through a semi-structured interview in both subject groups.

Patients perceived positive pictures as less positive than controls did. Positive anticipation (i.e., green color cues) elicited stronger activations in the anterior cingulate cortex and the right insula in patients than in healthy controls, indicating salience network disturbances. An exploratory analysis of all regions in the Automated Anatomical Labeling Atlas 2 found significant differences in activity to positive anticipation between groups in several brain regions involved in cognition and emotion processing. Positive and negative anticipation elicited stronger electrodermal responses in healthy controls. However, electrodermal reactivity to negative pictures was higher in patients than in controls.

Ongoing depression affects emotion anticipation and processing at the behavioral, neural, and physiological levels. These findings contribute to increased understanding of the disorder.

Patients in forensic psychiatric care (FPC) are commonly treated with a wide range of psychotropic medications. There is, however, a lack of understanding regarding how pharmacological treatment and psychotropic polypharmacy are used throughout care.

This register-based cohort study included patients admitted to FPC in Sweden between 2009 and 2020. We estimated the prevalence of the use of major psychotropic medication, as well as psychotropic polypharmacy, at admission and discharge. We also examined the change in antipsychotic use after admission.

In total, 1962 patients were included. Antipsychotics were the most used psychotropic medication class, with 86.2% (95%CI: 84.5–87.8) of patients receiving at least one typical or atypical antipsychotic at admission. Changes in the antipsychotic regime were more common at the beginning of FPC, compared to later time points. Within the subgroup of patients discharged during the study period (n = 561), there was a reduction in the use of typical antipsychotics (admission: 34.9%; discharge: 26.6%) and hypnotics and sedatives (admission: 37.4%; discharge: 28.1%). Other major medication classes remained relatively stable. The prevalence of psychiatric polypharmacy at admission was 70.6% (95%CI: 68.5–72.7) and remained similar during care.

Our study documented a high prevalence of antipsychotic use and psychotropic polypharmacy through FPC. Further, a high level of off-label antipsychotic use and antipsychotic polypharmacy was observed. Stronger evidence regarding the effectiveness and safety of these treatment strategies is needed.

The coronavirus disease 2019 (COVID-19) pandemic has impacted global mental health, with individuals with severe mental illness (SMI) being particularly vulnerable. Research on changes in psychiatric symptoms during this pandemic has yielded inconsistent results, often due to individual heterogeneity and a limited focus on broader outcomes such as psychosocial functioning, societal and personal recovery, and quality of life (QoL). Furthermore, long-term effects remain underexplored. This longitudinal cohort study aimed to assess the COVID-19 pandemic’s impact on mental and psychosocial functioning, QoL, and recovery in individuals with SMI, and to explore individual and treatment characteristics associated with outcome changes.

Two cohorts were included, involving adults (≥18 years) diagnosed with DSM-5 disorders and experiencing long-term impairments. Participants received care between January 1, 2018 and December 31, 2023. Outcomes included the Health of the Nation Outcome Scales, the Manchester Short Assessment of Quality of Life, and the Individual Recovery Outcomes Counter. Changes were analyzed across five pandemic periods using linear mixed models.

Improvements in mental and psychosocial functioning, QoL, and recovery were observed over time, regardless of the COVID-19 pandemic period. However, progress was slower during the COVID-19 pandemic compared to pre-pandemic levels. No individual or treatment characteristics were significantly linked to changes in outcomes.

The findings suggest that the COVID-19 pandemic had a minimal negative impact on individuals with SMI. This may be due to the marginal negative effects of the pandemic on this population, or the mitigating role of stabilizing factors within the current Dutch care models.

The COVID-19 pandemic increased the incidence and burden of eating disorders (EDs) globally. First Episode Rapid Early Intervention for EDs (FREED) is a nationally implemented early intervention service model for young people (16–25 years) with EDs in England. This study evaluates the longer-term impact of the pandemic on presentations and service provision in FREED.

Data from January 2019 to September 2024 were analyzed, including three services with pre-, pandemic, and post-pandemic data (Sample 1), and 62 services with post-pandemic data (Sample 2), 32 of which also contributed data during the pandemic. Linear mixed models examined referral numbers, duration of untreated ED, diagnostic mix, average body mass index for anorexia nervosa, and wait times.

In the three services with pre-, pandemic, and post-pandemic data, referrals remained significantly higher post-pandemic compared to the pre-pandemic period. Across all services, post-pandemic referrals declined compared to the pandemic period. Consistently, anorexia nervosa diagnoses decreased, and the duration of untreated ED at presentation increased post-pandemic.

The COVID-19 pandemic has had lasting impacts on ED service provision in England. Sustained investment and national support are essential to ensure FREED services continue to meet the needs of young people with recent-onset EDs and to reduce the duration of untreated ED.

Physical activity is a known protective factor against depression but physical activity competes with other time-consuming behaviors that may increase depression risk. This study investigates the association between time spent in various movement-related activities and incident major depression, with a particular focus on the effects of replacing TV-watching time with other activities. Additionally, we explored whether the impact of substituting TV-watching differs across age groups.

A population-based cohort study (Lifelines) with four-year follow-up, including 65,454 non-depressed adults (18+). Participants self-reported time spent in active commuting, leisure, sports, household, work or school physical-related activities, TV-watching, and sleep. Major depressive disorder was assessed using the Mini International Neuropsychiatric Interview. Compositional isotemporal data analysis was performed to analyze the effect of reallocating time in TV-watching with other activities adjusting for potential confounders. Interactions with age groups were also examined.

The incidence of major depressive disorder was 2.4%. Reallocating TV-watching time to any other physical activity or sleep reduced this risk in middle-aged adults. In older adults, only substituting TV-watching time with sports reduced the probability of becoming depressed. No significant reduction in probabilities for incident depression was found in younger adults.

Replacing TV-watching time with other activities, including sleep, may serve as a preventive strategy against depressive disorder in middle-aged adults, while only the substitution with sports seems beneficial for older adults. Future research should aim to identify other activities, particularly in younger adults, that may prevent depression.

Maternal alcohol consumption can adversely affect children’s development, but the impact of paternal drinking is less understood. We aimed to investigate whether maternal or paternal alcohol consumption during pregnancy affected children’s mental health and behavior.

A total of 2,013 parent–child triads from the European Longitudinal Study of Pregnancy and Childhood cohort were used. Data on alcohol consumption was obtained from questionnaires during pregnancy and after the child’s birth. Mental health and behavior of children were assessed with Strength and Difficulties Questionnaire (SDQ). The associations were tested using linear regression, adjusting for socio-demographic and psychosocial covariates.

Increased maternal alcohol consumption was associated with higher total SDQ scores at ages 7, 11, and 18 years old when the outcomes were reported by mothers, but only at 11 years when reported by children. We did not observe any dose–response relationship, and the effect size did not change during the follow-up. The effects were observed across various domains of SDQ: in the emotional symptoms subscale at age 11, in the conduct problems subscale at ages 7 and 11, and in the hyperactivity/inattention subscale at age 18. Paternal alcohol consumption was not associated with SDQ.

Maternal alcohol consumption during pregnancy is associated with long-term effects on children’s mental health and behavior, particularly when reported by mothers. No association was found between paternal alcohol consumption, suggesting that the results may stem from biological effects of alcohol or other factors beyond the direct exposure, potentially encompassing broader maternal psychosocial or behavioral characteristics.

Exposure to Adverse Childhood Experiences (ACEs) might increase the risk of suicide behaviors in the general adult population, while this association in individuals with affective disorders remains less characterized.

A comprehensive search was conducted in MEDLINE, PsycINFO, CINAHL, Web of Science, Scopus, and PubMed up to July 10th, 2024. Observational studies that compared the risk of suicide behaviors in individuals exposed and unexposed to ACEs were included. Pairwise random-effects meta-analyses were conducted, and the certainty of evidence was assessed with validated criteria.

A total of 41 studies from 17 countries, comprising 19,588 participants, were analyzed. The main findings indicated a significant association between ACEs and suicidal behaviors, with an odds ratio (OR) of 1.98 (95% confidence interval [CI] 1.74–2.26), and a “highly suggestive” strength of association. This was consistent across diagnostic subgroups (i.e., Major Depressive Disorders, Bipolar Disorders, and mixed diagnoses). The association was confirmed for any ACE, with sexual abuse being the most frequently reported and showing the highest risk (OR 2.24; 95% CI 1.90–2.64), for suicidal ideation (OR 2.16; 95% CI 1.42–3.29), and for suicide attempts (OR 1.95; 95% CI 1.70–2.25), while death by suicide and non-suicidal self-injury were underreported. Meta-regression analyses did not suggest potential moderators, though underreporting was noted.

This meta-analysis shows that exposure to ACEs nearly doubles the risk of suicide behaviors in individuals with affective disorders, warranting the targeted clinical, research, and policy measures to timely address this global mental health issue.

The neurobiological basis of suicidal behaviour remains poorly understood. However, emerging evidence suggests that inflammation and vascular homeostasis factors may play a role in its pathophysiology. Childhood trauma, through immune system dysfunction and increased risk of suicidal behaviours, might influence these associations. This study examined the relationships between immune-inflammatory and vascular homeostasis-related markers and their interaction with childhood trauma in relation to a history of suicide attempts in individuals with depression.

A total of 328 patients with major depression were recruited: 166 with a history of suicide attempts and 162 without. Using multivariate binary logistic regression models adjusted for cofounders, we examined the associations between childhood trauma, levels of platelet-related immune markers (serotonin, MCP-1, TSP-1, TSP-2, PDGF-AB, PDGF-BB), and suicide attempt history. Independent associations between PDGF-BB, childhood trauma, and suicide attempts were further assessed using interaction models. Stratified sensitivity analyses based on childhood trauma history were also conducted.

Childhood trauma consistently emerged as associated with suicide attempts across all models. Among the measured biomarkers, higher TSP-2 levels were associated with a suicide attempt history, independent of childhood trauma. Meanwhile, while PDGF-BB alone was not directly linked to suicide attempt history, the interaction analysis revealed that individuals with lower PDGF-BB levels and more severe childhood trauma were more likely to have attempted suicide.

TSP-2 and PDGF-BB are potential biomarkers linked to suicide attempts, with distinct roles in the interplay between biological processes and early-life adversity. These insights can inform the biomarker-informed development of tailored prevention and treatment strategies.

The use of coercive measures is an increasingly debated aspect of psychiatric treatment. Considering the multitude of negative effects, patients, clinicians, and ethicists alike have called for a more cautious application of coercion. It therefore remains important to investigate which organizational characteristics have the potential to facilitate efficient coercion reduction. The same holds true for the efficient reduction of symptom severity during inpatient treatment.

The current study compared 22 Swiss psychiatric clinics treating 45,095 cases regarding their relative efficiency in treating cases without coercion given their staff resources. To this end, we applied a Data Envelopment Analysis to clinical routine data. We focused specifically on inefficiencies attributable to management factors independent of the clinics’ total staff numbers. We further compared the clinics’ relative efficiencies regarding changes of self-reports and third-person reports of symptom severity during inpatient stays.

Efficiency scores suggest that on average, the clinics could improve the percentage of cases treated without coercion by 9% and the changes of symptom severity by 34% (for third-person ratings) or 18% (for self-reports) while keeping staff numbers constant. An analysis of specific coercion types revealed that the potential for efficiency improvements via management was highest for movement restrictions. We found no effect of clinic size on efficiency scores regarding any of the outcome measures.

Our results underline the importance of management factors beyond staff resources (e.g., staff trainings or changes in ward structure and treatment concepts) for the efficient reduction of coercion and psychiatric symptoms during inpatient stays.

Improving media adherence to World Health Organization (WHO) guidelines is crucial for preventing suicidal behaviors in the general population. However, there is currently no valid, rapid, and effective method to evaluate the adherence to these guidelines.

This comparative effectiveness study (January–August 2024) evaluated the ability of two artificial intelligence (AI) models (Claude Opus 3 and GPT-4O) to assess the adherence of media reports to WHO suicide-reporting guidelines. A total of 120 suicide-related articles (40 in English, 40 in Hebrew, and 40 in French) published within the past 5 years were sourced from prominent newspapers. Six trained human raters (two per language) independently evaluated articles based on a WHO guideline-based questionnaire addressing aspects, such as prominence, sensationalism, and prevention. The same articles were also processed using AI models. Intraclass correlation coefficients (ICCs) and Spearman correlations were calculated to assess agreement between human raters and AI models.

Overall adherence to WHO guidelines was ~50% across all languages. Both AI models demonstrated strong agreement with human raters, with GPT-4O showing the highest agreement (ICC = 0.793 [0.702; 0.855]). The combined evaluations of GPT-4O and Claude Opus 3 yielded the highest reliability (ICC = 0.812 [0.731; 0.869]).

AI models can replicate human judgment in evaluating media adherence to WHO guidelines. However, they have limitations and should be used alongside human oversight. These findings may suggest that AI tools have the potential to enhance and promote responsible reporting practices among journalists and, thus, may support suicide prevention efforts globally.

Heterogeneous symptoms in major depression contribute to unsuccessful antidepressant treatment, termed treatment-resistant depression (TRD). Psychometric network modeling conceptualizes depression as interplay of symptoms with potential benefits for treatment; however, a knowledge gap exists regarding networks in TRD.

Symptoms from 1,385 depressed patients, assessed by the Montgomery-Åsberg-depression rating scale (MADRS) as part of the “TRD-III” cohort of the multinational research consortium “Group for the Studies of Resistant Depression,” were used for Gaussian graphical network modeling. Networks were estimated for two timepoints, pretreatment and posttreatment, after the establishment of outcomes response, non-response, and TRD. Applying the network-comparison test, edge weights, and symptom centrality was assessed by bootstrapping. Applying the network-comparison test, outcome groups were compared cross-sectionally and longitudinally regarding the networks’ global strength, invariance, and centrality.

Pretreatment networks did not differ in global strength, but outcome groups showed distinct symptom connections. For both response and TRD, global strength was reduced posttreatment, leading to significant differences between each pair of networks posttreatment. Sadness, lassitude, inability-to-feel, and pessimistic thoughts ranked most centrally in unfavorable outcomes, while reduced-appetite and suicidal thoughts were more densely connected in response. Connections between central symptoms increased in strength following unsuccessful treatment, particularly regarding links involving pessimistic thoughts in TRD.

Treatment reduced global network strength across outcome groups. However, distinct symptom networks were found in patients showing response to treatment, non-response, and TRD. More easily targetable symptoms such as reduced-appetite were central to networks in patients with response, while pessimistic thoughts may be a key symptom upholding disease burden in TRD.

Nowadays, both researchers and clinicians alike have to deal with increasingly larger datasets, specifically also in the context of mental health data. Sophisticated tools for dataset visualization of information from various item-based instruments, such as questionnaire data or data from digital applications or clinical documentations, are still lacking, specifically for an integration at multiple levels and for use in both data organization and appropriate construction for its valid use in subsequent analyses.

Here, we introduce ItemComplex, a Python-based framework for ex-post visualization of large datasets. The method exploits the comprehensive recognition of instrument alignments and the identification of new content networks and graphs based on item similarities and shared versus differential conceptual bases within and across data and studies.

The ItemComplex framework was evaluated using four existing large datasets from four different cohort studies and demonstrated successful data visualization across multi-item instruments within and across studies. ItemComplex enables researchers and clinicians to navigate through big datasets reliably, informatively, and quickly. Moreover, it facilitates the extraction of new insights into construct representations and concept identifications within the data.

The ItemComplex app is an efficient tool in the field of big data management and analysis addressing the growing complexity of modern datasets to harness the potential hidden within these extensive collections of information. It is also easily adjustable for individual datasets and user preferences, both in the research and clinical field.

Due to limited inpatient care resources and high healthcare expenditures, understanding factors that affect lengths of stay (LOS) is highly relevant. We aimed to investigate associations between metabolic disturbances and LOS in a psychiatric hospital and to identify other clinical and sociodemographic LOS predictors.

Patients admitted to one of the units of the general psychiatric or psychogeriatric departments between January 1, 2007 and December 31, 2020, were included. Metabolic disturbances (i.e., the metabolic syndrome and its five components) were defined using the International Diabetes Federation definition. Cox frailty regression models with time-varying coefficients were used to investigate the association between metabolic disturbances and LOS. Hazard ratios (HR) >1 and HR < 1 indicated the relative likelihood of shorter and extended LOS, respectively.

A total of 7,771 patients for 16,959 hospital stays throughout 14 years of follow-up were included. Central obesity (HR = 0.82; 95% confidence interval [CI] = [0.76–0.89]), hyperglycemia (HR = 0.83; 95% CI = [0.78–0.89]), hypertriglyceridemia (HR = 0.87; 95% CI = [0.80–0.92]), and the metabolic syndrome (HR = 0.76; 95% CI = [0.70–0.82]) were associated with an increased risk of extended LOS in the psychiatric hospital, while underweight (HR = 1.30, 95% CI = [1.09–1.56]) and HDL hypocholesterolemia (HR = 1.10, 95% CI = [1.03–1.18]) were associated with a higher likelihood of shorter LOS. In first-episode psychosis patients, hypertriglyceridemia (HR = 0.82; 95% CI = [0.67–0.99]) and hypertension (HR = 0.76, 95% CI = [0.58–0.99]) were associated with extended LOS when considering all stays, while no association was found when considering the first stay per patient.

Future studies should determine whether better metabolic monitoring and treatment of metabolic disturbances can contribute to reducing LOS.

Neurodevelopmental conditions are crucial risk factors for catatonia in pediatric and adult populations. Recent case reports and studies have identified an increasing number of genetic abnormalities likely contributing to catatonia. Catatonia associated with genetic abnormalities is challenging in terms of identification, chronicity, and resistance to treatment. In addition, understanding these genetic abnormalities through identifying rare single nucleotide and copy number variants may offer valuable insights into the underlying pathophysiology.

We conducted a systematic review of all genetic abnormalities reported with catatonia and performed a gene-set enrichment analysis. Our systematic literature search for relevant articles published through July 15, 2024, using combinations of “catatonia,” “catatonic syndrome,” “genetic,” and “genes” in PubMed, yielded 317 articles. Of these, 94 were included, covering 374 cases of catatonia and 78 distinct genetic abnormalities.

This review discusses the clinical presentation of catatonia for each genetic disorder, the treatment strategies, and the putative underlying mechanisms.

The review highlights that catatonia underpinned by genetic abnormalities presents specific clinical and treatment-response features. Therefore, we propose genetic testing guidelines for catatonia and advocate for systematically investigating catatonia in several genetic diseases. Regarding the pathophysiology of catatonia, the gene ontology of biological processes reveals significant enrichment of variants in synaptic and post-synaptic regulatory genes, particularly within GABAergic neurons, reinforcing the implication of the excitatory/inhibitory imbalance. Finally, genetic variants are enriched in microglial cells, highlighting the role of brain inflammation in triggering catatonia. This comprehensive insight could pave the way for more effective management strategies for this condition.

Clinical high risk for psychosis (CHR) is often managed with antipsychotic medications, but their effects on neurocognitive performance and clinical outcomes remain insufficiently explored. This study investigates the association between aripiprazole and olanzapine use and cognitive and clinical outcomes in CHR individuals, compared to those receiving no antipsychotic treatment.

A retrospective analysis was conducted on 127 participants from the Shanghai At Risk for Psychosis (SHARP) cohort, categorized into three groups: aripiprazole, olanzapine, and no antipsychotic treatment. Neurocognitive performance was evaluated using the MATRICS Consensus Cognitive Battery (MCCB), while clinical symptoms were assessed through the Structured Interview for Prodromal Syndromes (SIPS) at baseline, 8 weeks, and one year.

The non-medicated group demonstrated greater improvements in cognitive performance, clinical symptoms, and functional outcomes compared to the medicated groups. Among the antipsychotic groups, aripiprazole was associated with better visual learning outcomes than olanzapine. Improvements in neurocognition correlated significantly with clinical symptom relief and overall functional gains at follow-up assessments.

These findings suggest potential associations between antipsychotic use and cognitive outcomes in CHR populations while recognizing that observed differences may reflect baseline illness severity rather than medication effects alone. Aripiprazole may offer specific advantages over olanzapine, underscoring the importance of individualized risk-benefit evaluations in treatment planning. Randomized controlled trials are needed to establish causality.