Development and validation of a machine learning model for prediction of type 2 diabetes in patients with mental illness.

BACKGROUND: Type 2 diabetes (T2D) is approximately twice as common among individuals with mental illness compared with the background population, but may be prevented by early intervention on lifestyle, diet, or pharmacologically. Such prevention relies on identification of those at elevated risk (prediction). The aim of this study was to develop and validate a machine learning model for prediction of T2D among patients with mental illness. METHODS: The study was based on routine clinical data from electronic health records from the psychiatric services of the Central Denmark Region. A total of 74,880 patients with 1.59 million psychiatric service contacts were included in the analyses. We created 1343 potential predictors from 51 source variables, covering patient-level information on demographics, diagnoses, pharmacological treatment, and laboratory results. T2D was operationalised as HbA1c ≥48 mmol/mol, fasting plasma glucose ≥7.0 mmol/mol, oral glucose tolerance test ≥11.1 mmol/mol or random plasma glucose ≥11.1 mmol/mol. Two machine learning models (XGBoost and regularised logistic regression) were trained to predict T2D based on 85% of the included contacts. The predictive performance of the best performing model was tested on the remaining 15% of the contacts. RESULTS: The XGBoost model detected patients at high risk 2.7 years before T2D, achieving an area under the receiver operating characteristic curve of 0.84. Of the 996 patients developing T2D in the test set, the model issued at least one positive prediction for 305 (31%). CONCLUSION: A machine learning model can accurately predict development of T2D among patients with mental illness based on routine clinical data from electronic health records. A decision support system based on such a model may inform measures to prevent development of T2D in this high-risk population.

Stability of diagnostic coding of psychiatric outpatient visits across the transition from the second to the third version of the Danish National Patient Registry.

OBJECTIVE: In Denmark, data on hospital contacts are reported to the Danish National Patient Registry (DNPR). The ICD-10 main diagnoses from the DNPR are often used as proxies for mental disorders in psychiatric research. With the transition from the second version of the DNPR (DNPR2) to the third (DNPR3) in February-March 2019, the way main diagnoses are coded in relation to outpatient treatment changed substantially. Specifically, in the DNPR2, each outpatient treatment course was labelled with only one main diagnosis. In the DNPR3, however, each visit during an outpatient treatment course is labelled with a main diagnosis. We assessed whether this change led to a break in the diagnostic time-series represented by the DNPR, which would pose a threat to the research relying on this source. METHODS: All main diagnoses from outpatients attending the Psychiatric Services of the Central Denmark Region from 2013 to 2021 (n = 100,501 unique patients) were included in the analyses. The stability of the DNPR diagnostic time-series at the ICD-10 subchapter level was examined by comparing means across the transition from the DNPR2 to the DNPR3. RESULTS: While the proportion of psychiatric outpatients with diagnoses from some ICD-10 subchapters changed statistically significantly from the DNPR2 to the DNPR3, the changes were small in absolute terms (e.g., +0.6% for F2-psychotic disorders and +0.6% for F3-mood disorders). CONCLUSION: The change from the DNPR2 to the DNPR3 is unlikely to pose a substantial threat to the validity of most psychiatric research at the diagnostic subchapter level.

Positive predictive value of a register-based algorithm using the Danish National Registries to identify suicidal events.

PURPOSE: It is not possible to fully assess intention of self-harm and suicidal events using information from administrative databases. We conducted a validation study of intention of suicide attempts/self-harm contacts identified by a commonly applied Danish register-based algorithm (DK-algorithm) based on hospital discharge diagnosis and emergency room contacts. METHODS: Of all 101 530 people identified with an incident suicide attempt/self-harm contact at Danish hospitals between 1995 and 2012 using the DK-algorithm, we selected a random sample of 475 people. We validated the DK-algorithm against medical records applying the definitions and terminology of the Columbia Classification Algorithm of Suicide Assessment of suicidal events, nonsuicidal events, and indeterminate or potentially suicidal events. We calculated positive predictive values (PPVs) of the DK-algorithm to identify suicidal events overall, by gender, age groups, and calendar time. RESULTS: We retrieved medical records for 357 (75%) people. The PPV of the DK-algorithm to identify suicidal events was 51.5% (95% CI: 46.4-56.7) overall, 42.7% (95% CI: 35.2-50.5) in males, and 58.5% (95% CI: 51.6-65.1) in females. The PPV varied further across age groups and calendar time. After excluding cases identified via the DK-algorithm by unspecific codes of intoxications and injury, the PPV improved slightly (56.8% [95% CI: 50.0-63.4]). CONCLUSIONS: The DK-algorithm can reliably identify self-harm with suicidal intention in 52% of the identified cases of suicide attempts/self-harm. The PPVs could be used for quantitative bias analysis and implemented as weights in future studies to estimate the proportion of suicidal events among cases identified via the DK-algorithm.

Effect of the GLP-1 receptor agonist semaglutide on metabolic disturbances in clozapine-treated or olanzapine-treated patients with a schizophrenia spectrum disorder: study protocol of a placebo-controlled, randomised clinical trial (SemaPsychiatry).

INTRODUCTION: Clozapine and olanzapine are some of the most effective antipsychotics, but both are associated with weight gain and relevant metabolic disturbances, including pre-diabetes and diabetes. Non-pharmacological/behavioural interventions have had limited effects counteracting these adverse effects. Semaglutide, a glucagon-like peptide 1 receptor agonist, is approved for the treatment of type 2 diabetes and obesity. We will investigate the long-term effects of add-on treatment with semaglutide once a week versus placebo once a week on the metabolic status in pre-diabetic (glycated haemoglobin A1c (HbA1c) 35-47 mmol/mol (5.4%-6.4%) and diabetic (HbA1c 48-57 mmol/mol (6.5%-7.4%)) patients diagnosed with a schizophrenia spectrum disorder who initiated clozapine or olanzapine treatment within the last 60 months. METHODS AND ANALYSIS: This is a 26-week, double-blinded, randomised, placebo-controlled trial. Altogether, 104 patients diagnosed with a schizophrenia spectrum disorder, aged 18-65 years, with pre-diabetes or diabetes will be randomised to injections of 1.0 mg semaglutide once a week or placebo for 26 weeks. The primary endpoint is change from baseline in HbA1c. Secondary endpoints include changes in body weight, hip and waist circumference and plasma levels of insulin, glucagon, glucose, and C-peptide, insulin sensitivity, beta cell function, hepatic function, fibrosis-4 score, lipid profile, incretin hormones, bone markers, body composition, bone density, proteomic analyses and oxidative stress markers. Together with alcohol, tobacco and drug use, potential effects on the reward value of a sweet-fat stimulus, psychopathology, level of activity and quality of life will also be assessed. ETHICS AND DISSEMINATION: This study is approved by the Danish Medicines Agency and the regional scientific ethics committee of the Capital Region of Denmark (committee C, #H-20019008) and will be carried out in accordance with International Council for Harmonisation Good Clinical Practice guidelines and the Helsinki Declaration. The results will be disseminated through peer-review publications and conference presentations. TRIAL REGISTRATION NUMBER: NCT04892199.

Validation of ratings on the six-item Positive and Negative Syndrome Scale obtained via the Simplified Negative and Positive Symptoms Interview among outpatients with schizophrenia.

BACKGROUND: The six-item Positive and Negative Syndrome Scale (PANSS-6) is a measure of the severity of core symptoms of schizophrenia, which can be administered via the brief Simplified Negative and Positive Symptoms Interview (SNAPSI). A recent study has confirmed the validity of PANSS-6 ratings as derived by SNAPSI (PANSS-6SNAPSI) among inpatients with schizophrenia. AIMS: We aimed to test the validity of PANSS-6SNAPSI among outpatients with schizophrenia using PANSS-6 ratings extracted from the 30-item PANSS-30 as derived by the Structured Clinical Interview for the Positive and Negative Syndrome Scale (PANSS-6SCI-PANSS) as a gold standard reference. METHODS: PANSS-6SNAPSI and PANSS-6SCI-PANSS ratings were obtained at two time points by independent raters with established inter-rater reliability. Agreement between PANSS-6SNAPSI and PANSS-6SCI-PANSS ratings was estimated via intra-class coefficients (ICCs) and responsiveness over time was quantified using Spearman's rank correlation coefficients. Post hoc "leave-one-out" analyses were carried out, in which each rater in turn was excluded from the ICC calculations. RESULTS: Seventy-three outpatients with schizophrenia participated in the study (mean age: 38.3 years; 56% males). The ICC for PANSS-6SNAPSI versus PANSS-6SCI-PANSS was 0.67 [95%CI = 0.56-0.76] and the Spearman's rank correlation coefficient for responsiveness was 0.40 (p = 0.004). When data from a specific outlying rater were excluded, the ICC for PANSS-6SNAPSI versus PANSS-6SCI-PANSS was 0.75 [95% CI = 0.63-0.83] and the Spearman's rank correlation coefficient for responsiveness was 0.55 (p = 0.018). CONCLUSIONS: We found PANSS-6SNAPSI ratings to have acceptable clinical validity, suggesting that PANSS-6SNAPSI can be used for both inpatients and outpatients with schizophrenia.