Prediction models of suicide and non-fatal suicide attempt after discharge from a psychiatric inpatient stay: A machine learning approach on nationwide Danish registers.

INTRODUCTION: To develop machine learning models capable of predicting suicide and non-fatal suicide attempt as separate outcomes in the first 30 days after discharge from a psychiatric inpatient stay. METHODS: Prospective cohort study using nationwide Danish registry data. We included individuals who were 18 years or older, and all discharges from psychiatric hospitalizations in Denmark from 1995 to 2018. We trained predictive models using 10-fold cross validation on 80% of the data and did testing on the remaining 20%. RESULTS: The best model for predicting non-fatal suicide attempt was an ensemble of predictions from gradient boosting (XGBoost) and categorical boosting (catBoost). The ROC-AUC for predicting suicide attempt was 0.85 (95% CI: 0.84-0.85). At a risk threshold of 4.36%, positive predictive value (PPV) was 11.0% and sensitivity was 47.2%. The best model for predicting suicide was an ensemble of predictions from random forest, XGBoost and catBoost. For suicide, the ROC-AUC was 0.71 (95% CI: 0.70-0.73). At a risk threshold of 0.15%, PPV was 0.34% and sensitivity was 56.0%. The most contributing predictors differed when predicting suicide and suicide attempt, indicating that separate models are needed. The ensemble model was fair across sex and age, and more so than the penalized logistic regression model. CONCLUSIONS: We achieved good performance for predicting suicide attempts and demonstrated a clinical application of ensemble models. Our results indicate a difference in predictive performance for models predicting suicide and suicide attempt, respectively. Thus, we recommend that suicide and suicide attempt are treated as two separate endpoints, in particular for clinical application. We demonstrated that the ensemble model is fairer across sex and age compared with a penalized logistic regression, and therefore we recommend the use of well-tested ensembles despite a more complex explainability.

Vagotomy and the risk of mental disorders: A nationwide population-based study.

OBJECTIVE: To investigate vagotomy, the severance of the vagus nerve, and its association with mental disorders, as gut-brain communication partly mediated by the vagus nerve have been suggested as a risk factor. METHODS: Nationwide population-based Danish register study of all individuals alive and living in Denmark during the study period 1977-2016 and who had a hospital contact for ulcer with or without vagotomy. Follow-up was until any diagnosis of mental disorders requiring hospital contact, emigration, death, or end of follow-up on December 31, 2016, whichever came first. Data were analyzed using survival analysis and adjusted for sex, age, calendar year, ulcer type, and Charlson comorbidity index score. RESULTS: During the study period, 113,086 individuals had a hospital contact for ulcer. Of these, 5,408 were exposed to vagotomy where 375 (6.9%) subsequently developed a mental disorder. Vagotomy overall was not associated with mental disorders (HR: 1.10; 95%CI: 0.99-1.23), compared to individuals with ulcer not exposed to vagotomy. However, truncal vagotomy was associated with an increased HR of 1.22 (95%CI: 1.06-1.41) for mental disorders, whereas highly selective vagotomy was not associated with mental disorders (HR: 0.98; 95%CI: 0.84-1.15). Truncal vagotomy was also associated with higher risk of mental disorders when compared to highly selective vagotomy (p = 0.034). CONCLUSIONS: Overall, vagotomy did not increase the risk of mental disorders; however, truncal vagotomy specifically was associated with a small risk increase in mental disorders, whereas no association was found for highly selective vagotomy. Thus, the vagus nerve does not seem to have a major impact on the development of mental disorders.

Efficacy and safety of anti-inflammatory agents in treatment of psychotic disorders - A comprehensive systematic review and meta-analysis.

OBJECTIVE: Antipsychotic effects of immunomodulating drugs have been suggested; however, a thorough, comprehensive meta-analysis on the effect and safety of anti-inflammatory add-on treatment on psychotic disorders is lacking. METHOD: Multiple databases were searched up until February 2020. Only double-blinded, randomized, placebo-controlled clinical trials (RCTs) were included. Primary outcomes were change in total psychopathology and adverse events. Secondary outcomes included, amongst others, positive and negative symptoms, general psychopathology and cognitive domains. We performed random-effects meta-analyses estimating mean differences (MD) and standardized mean differences (SMD) for effect sizes. RESULTS: Seventy RCTs (N = 4104) were included, investigating either primarily anti-inflammatory drugs, i.e. drugs developed for immunomodulation, such as NSAIDs, minocycline and monoclonal antibodies (k = 15), or drugs with potential anti-inflammatory properties (k = 55), e.g. neurosteroids, N-acetyl cysteine, estrogens, fatty acids, statins, and glitazones. Antipsychotics plus anti-inflammatory treatment, compared to antipsychotics plus placebo, was associated with a PANSS scale MD improvement of -4.57 (95%CI = -5.93 to -3.20) points, corresponding to a SMD effect size of -0.29 (95%CI = -0.40 to -0.19). Trials on schizophrenia (MD = -6.80; 95%CI, -9.08 to -4.52) showed greater improvement (p < 0.01) than trials also including other psychotic disorders. However, primarily anti-inflammatory drugs (MD = 4.00; 95%CI = -7.19 to -0.80) were not superior (p = 0.69) to potential anti-inflammatory drugs (MD = 4.71; 95%CI = -6.26 to -3.17). Furthermore, meta-regression found that smaller studies showed significantly larger effect sizes than the larger studies (p = 0.0085), and only 2 studies had low risk of bias on all domains. Small but significant effects were found on negative symptoms (MD = -1.29), positive symptoms (MD = -0.53), general psychopathology (MD = -1.50) and working memory (SMD = 0.21). No differences were found regarding adverse events, but only 26 studies reported hereon. CONCLUSIONS: Anti-inflammatory add-on treatment to antipsychotics showed improvement of psychotic disorders; however, no superiority was found in primarily anti-inflammatory drugs, raising the question of the mechanism behind the effect, and treatment effect might be overestimated due to the large number of small studies.

Infections of the central nervous system as a risk factor for mental disorders and cognitive impairment: A nationwide register-based study.

BACKGROUND: CNS infections have been suggested as risk factors for cognitive decline and mental disorders; however, large-scale studies have been lacking regarding types and agents of CNS infections. METHODS: We utilized the unique personal registration number to create a cohort of 1,709,867 individuals born 1977-2010. CNS infection was exposure and data were analysed with 1) cox regression analyses estimating hazard ratios (HR) for developing mental disorders and 2) binomial regression estimating relative risk (RR) for completion of 9th grade including average grade score in a sub-cohort born 1988-1998. RESULTS: CNS infection increased the risk for developing mental disorders with a HR of 1.34 (95% CI 1.27-1.42). The highest risk observed was within the first 6 months after the CNS infection with a HR of 26.98 (95% CI 21.19-34.35). Viral CNS infections (HR 1.47, 95% CI 1.35-1.61) conferred a higher risk (p < 0.001) than bacterial (HR 1.24, 95% CI 1.15-1.35). Encephalitis (HR 1.64, 95% CI 1.41-1.90) conferred a higher risk (p < 0.001) than meningitis (HR 1.26, 95% CI 1.18-1.35). The risk was highest for organic mental disorders (HR 6.50, 95% CI 5.11-8.28) and disorders of intellectual development (HR 3.56, 95% CI 2.94-4.31), with a HR of 19.19 (95% CI 7.46-49.35) for profound disorder of intellectual development (IQ < 20). Furthermore, CNS infection decreased the RR of completing 9th grade of mandatory schooling (RR 0.89, 95% CI 0.88-0.91) and lowered average grade score for completers (p < 0.001). CONCLUSIONS: CNS infections increased the risk for mental disorders and decreased the likelihood of completing 9th grade, indicating long-term consequences of CNS infections.

Immune Cell Alterations in Psychotic Disorders: A Comprehensive Systematic Review and Meta-Analysis.

BACKGROUND: A comprehensive meta-analysis on the composition of circulating immune cells from both the myeloid and the lymphoid lines including specialized subsets in blood and cerebrospinal fluid (CSF) of patients with psychotic disorders compared with healthy control participants has been lacking. METHODS: Multiple databases (PubMed, EMBASE, Cochrane Library, Web of Science, ClinicalTrials.gov, and PsycINFO) were searched for eligible studies up until October 18, 2022. All studies investigating circulating immune cells in the blood and CSF from patients with psychotic disorders (ICD-10: F20 and F22-29) compared with healthy control participants were included. RESULTS: A total of 86 studies were included in the meta-analysis. In the blood, the following categories of immune cells were elevated: leukocyte count (31 studies, standardized mean difference [SMD] = 0.35; 95% CI, 0.24 to 0.46), granulocyte count (4 studies, SMD = 0.57; 95% CI, 0.12 to 1.01), neutrophil granulocyte count (21 studies, SMD = 0.32; 95% CI, 0.11 to 0.54), monocyte count (23 studies, SMD = 0.40; 95% CI, 0.23 to 0.56), and B lymphocyte count (10 studies, SMD = 0.26; 95% CI, 0.04 to 0.48). Additionally, the neutrophil/lymphocyte ratio (23 studies, SMD = 0.40; 95% CI, 0.19 to 0.60), the monocyte/lymphocyte ratio (9 studies, SMD = 0.31; 95% CI, 0.04 to 0.57), and the platelet/lymphocyte ratio (10 studies, SMD = 0.23; 95% CI, 0.03 to 0.43) were elevated. The CSF cell count showed a similar tendency but was not significantly elevated (3 studies, SMD = 0.14; 95% CI, -0.04 to 0.32). CONCLUSIONS: The results indicate a broad activation of the immune system in psychotic disorders, with cells from both the myeloid and the lymphoid line being elevated. However, CSF analyses were lacking in most of the studies, and many studies were hampered by insufficient adjustment for confounding factors such as body mass index and smoking.

Effects of intraspecific variation in reproductive traits, pectoral fin use and burst swimming on metabolic rates and swimming performance in the Trinidadian guppy (Poecilia reticulata).

There is considerable intraspecific variation in metabolic rates and locomotor performance in aquatic ectothermic vertebrates; however, the mechanistic basis remains poorly understood. Using pregnant Trinidadian guppies (Poecilia reticulata), a live-bearing teleost, we examined the effects of reproductive traits, pectoral fin use and burst-assisted swimming on swimming metabolic rate, standard metabolic rate (O2std) and prolonged swimming performance (Ucrit). Reproductive traits included reproductive allocation and pregnancy stage, the former defined as the mass of the reproductive tissues divided by the total body mass. Results showed that the metabolic rate increased curvilinearly with swimming speed. The slope of the relationship was used as an index of swimming cost. There was no evidence that reproductive traits correlated with swimming cost, O2std or Ucrit. In contrast, data revealed strong effects of pectoral fin use on swimming cost and Ucrit. Poecilia reticulata employed body-caudal fin (BCF) swimming at all tested swimming speeds; however, fish with a high simultaneous use of the pectoral fins exhibited increased swimming cost and decreased Ucrit. These data indicated that combining BCF swimming and pectoral fin movement over a wide speed range, presumably to support swimming stability and control, is an inefficient swimming behaviour. Finally, transition to burst-assisted swimming was associated with an increase in aerobic metabolic rate. Our study highlights factors other than swimming speed that affect swimming cost and suggests that intraspecific diversity in biomechanical performance, such as pectoral fin use, is an important source of variation in both locomotor cost and maximal performance.

Long-term neurological outcome after COVID-19 using all SARS-CoV-2 test results and hospitalisations in Denmark with 22-month follow-up.

Hospitalisation with COVID-19 is associated with an increased risk of neurological sequelae; however, representative nationwide studies comparing to other infections with similar severity and also including milder SARS-CoV-2 infections have been lacking. Using the nationwide Danish registers including all SARS-CoV-2 PCR test results and hospitalisations between March 1, 2020, and December 31, 2021, we estimate the risk of any first neurological disorder diagnosed in inpatient, outpatient, or emergency room settings. We show that positive tests increase the rate of neurological disorders by a hazard ratio of 1.96 (95% confidence interval: 1.88-2.05) compared to individuals not tested and by a hazard ratio of 1.11 (95% confidence interval: 1.07-1.16) compared to individuals with negative tests only. However, there is no evidence that the risk of neurological disorders is higher for individuals who test positive compared to non-COVID-19 infections treated with anti-infective medication. The risk of neurological disorders is increased after COVID-19-hospitalisation compared to no COVID-19 hospital admission; however, these risks are comparable to hospitalisation with other respiratory infections (P value 0.328). In conclusion, COVID-19 is associated with an increased risk of neurological disorders, but no more than that observed after other infections of similar severity.

COVID-19 and Risk for Mental Disorders Among Adults in Denmark.

Importance: Psychiatric outcomes after COVID-19 have been of high concern during the pandemic; however, studies on a nationwide level are lacking. Objective: To estimate the risk of mental disorders and use of psychotropic medication among individuals with COVID-19 compared with individuals not tested, individuals with SARS-CoV-2-negative test results, and those hospitalized for non-COVID-19 infections. Design, Setting, and Participants: This nationwide cohort study used Danish registries to identify all individuals who were alive, 18 years or older, and residing in Denmark between January 1 and March 1, 2020 (N = 4 152 792), excluding individuals with a mental disorder history (n = 616 546), with follow-up until December 31, 2021. Exposures: Results of SARS-CoV-2 polymerase chain reaction (PCR) testing (negative, positive, and never tested) and COVID-19 hospitalization. Main Outcomes and Measures: Risk of new-onset mental disorders (International Statistical Classification of Diseases and Related Health Problems, Tenth Revision, codes F00-F99) and redeemed psychotropic medication (Anatomical Therapeutic Chemical classification codes N05-N06) was estimated through survival analysis using a Cox proportional hazards model, with a hierarchical time-varying exposure, reporting hazard rate ratios (HRR) with 95% CIs. All outcomes were adjusted for age, sex, parental history of mental illness, Charlson Comorbidity Index, educational level, income, and job status. Results: A total of 526 749 individuals had positive test results for SARS-CoV-2 (50.2% men; mean [SD] age, 41.18 [17.06] years), while 3 124 933 had negative test results (50.6% women; mean [SD] age, 49.36 [19.00] years), and 501 110 had no tests performed (54.6% men; mean [SD] age, 60.71 [19.78] years). Follow-up time was 1.83 years for 93.4% of the population. The risk of mental disorders was increased in individuals with positive (HRR, 1.24 [95% CI, 1.17-1.31]) and negative (HRR, 1.42 [95% CI, 1.38-1.46]) test results for SARS-CoV-2 compared with those never tested. Compared with individuals with negative test results, the risk of new-onset mental disorders in SARS-CoV-2-positive individuals was lower in the group aged 18 to 29 years (HRR, 0.75 [95% CI, 0.69-0.81]), whereas individuals 70 years or older had an increased risk (HRR, 1.25 [95% CI, 1.05-1.50]). A similar pattern was seen regarding psychotropic medication use, with a decreased risk in the group aged 18 to 29 years (HRR, 0.81 [95% CI, 0.76-0.85]) and elevated risk in those 70 years or older (HRR, 1.57 [95% CI, 1.45-1.70]). The risk for new-onset mental disorders was substantially elevated in hospitalized patients with COVID-19 compared with the general population (HRR, 2.54 [95% CI, 2.06-3.14]); however, no significant difference in risk was seen when compared with hospitalization for non-COVID-19 respiratory tract infections (HRR, 1.03 [95% CI, 0.82-1.29]). Conclusion and Relevance: In this Danish nationwide cohort study, overall risk of new-onset mental disorders in SARS-CoV-2-positive individuals did not exceed the risk among individuals with negative test results (except for those aged ≥70 years). However, when hospitalized, patients with COVID-19 had markedly increased risk compared with the general population, but comparable to risk among patients hospitalized for non-COVID-19 infections. Future studies should include even longer follow-up time and preferentially include immunological biomarkers to further investigate the impact of infection severity on postinfectious mental disorder sequelae.

Neuropsychiatric and Cognitive Outcomes in Patients 6 Months After COVID-19 Requiring Hospitalization Compared With Matched Control Patients Hospitalized for Non-COVID-19 Illness.

Importance: Prolonged neuropsychiatric and cognitive symptoms are increasingly reported in patients after COVID-19, but studies with well-matched controls are lacking. Objective: To investigate cognitive impairment, neuropsychiatric diagnoses, and symptoms in survivors of COVID-19 compared with patients hospitalized for non-COVID-19 illness. Design, Setting, and Participants: This prospective case-control study from a tertiary referral hospital in Copenhagen, Denmark, conducted between July 2020 and July 2021, followed up hospitalized COVID-19 survivors and control patients hospitalized for non-COVID-19 illness, matched for age, sex, and intensive care unit (ICU) status 6 months after symptom onset. Exposures: Hospitalization for COVID-19. Main Outcomes and Measures: Participants were investigated with the Mini-International Neuropsychiatric Interview, the Montreal Cognitive Assessment (MoCA), neurologic examination, and a semi-structured interview for subjective symptoms. Primary outcomes were total MoCA score and new onset of International Statistical Classification of Diseases and Related Health Problems, Tenth Revision (ICD-10) psychiatric diagnoses. Secondary outcomes included specific psychiatric diagnoses, subjective symptoms, and neurologic examination results. All outcomes were adjusted for age, sex, ICU admission, admission length, and days of follow-up. Secondary outcomes were adjusted for multiple testing. Results: A total of 85 COVID-19 survivors (36 [42%] women; mean [SD] age 56.8 [14] years) after hospitalization and 61 matched control patients with non-COVID-19 illness (27 [44%] women, mean age 59.4 years [SD, 13]) were enrolled. Cognitive status measured by total geometric mean MoCA scores at 6-month follow-up was lower (P = .01) among COVID-19 survivors (26.7; 95% CI, 26.2-27.1) than control patients (27.5; 95% CI, 27.0-27.9). The cognitive status improved substantially (P = .004), from 19.2 (95% CI, 15.2-23.2) at discharge to 26.1 (95% CI, 23.1-29.1) for 15 patients with COVID-19 with MoCA evaluations from hospital discharge. A total of 16 of 85 patients with COVID-19 (19%) and 12 of 61 control patients (20%) had a new-onset psychiatric diagnosis at 6-month follow-up, which was not significantly different (odds ratio, 0.93; 95% CI, 0.39-2.27; P = .87). In fully adjusted models, secondary outcomes were not significantly different, except anosmia, which was more common after COVID-19 (odds ratio, 4.56; 95% CI, 1.52-17.42; P = .006); but no longer when adjusting for multiple testing. Conclusions and Relevance: In this prospective case-control study, cognitive status at 6 months was worse among survivors of COVID-19, but the overall burden of neuropsychiatric and neurologic signs and symptoms among survivors of COVID-19 requiring hospitalization was comparable with the burden observed among matched survivors hospitalized for non-COVID-19 causes.

Development and validation of multivariable prediction models of remission, recovery, and quality of life outcomes in people with first episode psychosis: a machine learning approach.

BACKGROUND: Outcomes for people with first-episode psychosis are highly heterogeneous. Few reliable validated methods are available to predict the outcome for individual patients in the first clinical contact. In this study, we aimed to build multivariable prediction models of 1-year remission and recovery outcomes using baseline clinical variables in people with first-episode psychosis. METHODS: In this machine learning approach, we applied supervised machine learning, using regularised regression and nested leave-one-site-out cross-validation, to baseline clinical data from the English Evaluating the Development and Impact of Early Intervention Services (EDEN) study (n=1027), to develop and internally validate prediction models at 1-year follow-up. We assessed four binary outcomes that were recorded at 1 year: symptom remission, social recovery, vocational recovery, and quality of life (QoL). We externally validated the prediction models by selecting from the top predictor variables identified in the internal validation models the variables shared with the external validation datasets comprised of two Scottish longitudinal cohort studies (n=162) and the OPUS trial, a randomised controlled trial of specialised assertive intervention versus standard treatment (n=578). FINDINGS: The performance of prediction models was robust for the four 1-year outcomes of symptom remission (area under the receiver operating characteristic curve [AUC] 0·703, 95% CI 0·664-0·742), social recovery (0·731, 0·697-0·765), vocational recovery (0·736, 0·702-0·771), and QoL (0·704, 0·667-0·742; p<0·0001 for all outcomes), on internal validation. We externally validated the outcomes of symptom remission (AUC 0·680, 95% CI 0·587-0·773), vocational recovery (0·867, 0·805-0·930), and QoL (0·679, 0·522-0·836) in the Scottish datasets, and symptom remission (0·616, 0·553-0·679), social recovery (0·573, 0·504-0·643), vocational recovery (0·660, 0·610-0·710), and QoL (0·556, 0·481-0·631) in the OPUS dataset. INTERPRETATION: In our machine learning analysis, we showed that prediction models can reliably and prospectively identify poor remission and recovery outcomes at 1 year for patients with first-episode psychosis using baseline clinical variables at first clinical contact. FUNDING: Lundbeck Foundation.