Off-label use of antidepressants, antipsychotics, and mood-stabilizers in psychiatry.

Off-label drug prescribing in psychiatry is increasing. Many psychotropic drugs are approved for psychopathologic syndromes rather than based on international standard diagnostic classification systems which might facilitate the clinical decision for off-label prescriptions. The objective of this study was to analyze the prevalence and category of off-label use of psychotropic drugs. The study was conducted in 10 psychiatric hospitals in Germany over a period of 2 years. Prescription data of all patients were retrospectively analyzed after identification of antidepressants, antipsychotics, and mood-stabilizers, which were classified as off-label according to the German prescribing information and diagnostic classification according to ICD-10. In total, 53,909 patient cases (46% female) with a mean age of 46.8 (SD: 18) years were included in the study. 30.2% of the cases received at least one off-label prescription of a psychotropic drug during hospital stay. Off-label prevalence rates differed markedly between different diagnostic groups (ICD-10 F0/G3: 47%, F1: 33%, F2: 25%, F3: 21%, F4: 27%, F6: 46%, F7: 84%). The most often off-label prescribed drugs were quetiapine and mirtazapine for organic mental disorders (F0/G3), valproate and quetiapine in patients with disorders due to psychoactive substance use (F1), valproate in patients with psychotic disorders (F2), and risperidone and olanzapine in patients with affective disorders (F3). The prevalence rate of psychotropic off-label prescriptions is high if restricted to product description and ICD-10 diagnosis. Therefore, current psychiatric guidelines should drug-specifically issue this problem by defining psychiatric off-label indications based on a clear benefit-risk assessment.

Current use of anticholinergic medications in a large naturalistic sample of psychiatric patients.

Due to the high number of psychotropic drugs with anticholinergic potential, patients taking psychotropic drugs are at high risk for anticholinergic adverse drug reactions (ADRs). The aim of this study was to analyze the prevalence and type of pharmacodynamic anticholinergic drug-drug interactions in psychiatric patients. The retrospective longitudinal analysis used data from a large pharmacovigilance study conducted in ten German psychiatric hospitals. Anticholinergic burden of drugs was defined as "strong" or "moderate" based on current literature. Number and type of anticholinergic drugs were assessed. In total, 27,396 patient cases (45.6% female) with a mean age of 47.3 ± 18.3 years were included. 17.4% (n = 4760) of patients were ≥ 64 years. 35.4% of the patients received between one and four anticholinergic drugs simultaneously. A combination of drugs with anticholinergic potential was detected in 1738 cases (6.3%). Most prescribed drugs were promethazine (n = 2996), olanzapine (n = 2561), biperiden (n = 1074), and doxepin (n = 963). Patients receiving anticholinergic combinations were younger (45.7 vs. 47.4 years, p < 0.01) and had a longer inpatient stay (median 18 vs. 26.5 days, p < 0.001). The prevalence of anticholinergic drug use in psychiatry is high. Further efforts need to focus on reducing the rate of anticholinergics and inappropriate medication especially in the elderly. Anticholinergic ADRs can be prevented by avoiding high-risk drug combinations. Replacing tricyclic antidepressants and first-generation antihistamines with drugs with lower anticholinergic potential and avoiding biperiden could reduce 59.3% of anticholinergic drug application.

Pharmacodynamic Drug-Drug interactions of QT-prolonging drugs in hospitalized psychiatric patients.

At least 170 approved drugs are linked to QT prolongation, which can lead to serious adverse drug reactions (ADRs), such as Torsade de Pointes (TdP). The aim of this study was to analyze the prevalence and type of pharmacodynamic drug-drug interactions (DDIs) between QT-prolonging drugs in psychiatry. The present retrospective analysis used data from a large pharmacovigilance study, conducted in 10 psychiatric hospitals in Germany. Patients medication lists were screened for QT-prolonging drugs, classified according to the Arizona Center for Education and Research on Therapeutics (AZCERT). In total, 27,396 patient cases (46% female) with a mean (± standard deviation) age of 47 ± 18 years were included in the study. Altogether, 83% of the cases received at least one and up to eight QT-prolonging drugs at the same time. Combination of drugs with a known or possible risk for TdP (according to the AZCERT) was detected in 13,670 cases (50%). Most frequently prescribed psychotropic high-risk drugs (n = 48,995) were the antipsychotics pipamperone (n = 6202), quetiapine (n = 5718), prothipendyl (n = 4298), and risperidone (n = 4265). The replacement of high-risk drugs such as tricyclic antidepressants, levomepromazine, melperone, and promethazine with more tolerable drugs could avoid 11% of QT-prolonging drugs and increase the tolerability of psychopharmacological treatment. More than 80% of psychiatric patients receive at least one QT-prolonging drug during their hospital stay, and almost 50% of these drugs are combined in clinical practice. For the prevention of cardiac ADRs, the physician should evaluate the risk for QT prolongation for each drug and patient-specific risk factors before prescribing these drugs or drug combinations.

Therapeutic Drug Monitoring of Long-Acting Injectable Antipsychotic Drugs.

BACKGROUND: The use of therapeutic drug monitoring (TDM) to guide treatment with long-acting injectable (LAI) antipsychotics, which are increasingly prescribed, remains a matter of debate. The aim of this review was to provide a practical framework for the integration of TDM when switching from an oral formulation to the LAI counterpart, and in maintenance treatment. METHODS: The authors critically reviewed 3 types of data: (1) positron emission tomography data evaluating dopamine (D2/D3) receptor occupancy related to antipsychotic concentrations in serum or plasma; D2/D3 receptors are embraced as target sites in the brain for antipsychotic efficacy and tolerability, (2) pharmacokinetic studies evaluating the switch from oral to LAI antipsychotics, and (3) pharmacokinetic data for LAI formulations. Based on these data, indications for TDM and therapeutic reference ranges were considered for LAI antipsychotics. RESULTS: Antipsychotic concentrations in blood exhibited interindividual variability not only under oral but also under LAI formulations because these concentrations are affected by demographic characteristics such as age and sex, genetic peculiarities, and clinical variables, including comedications and comorbidities. Reported data combined with positron emission tomography evidence indicated a trend toward lower concentrations under LAI administration than under oral medications. However, the available evidence is insufficient to recommend LAI-specific therapeutic reference ranges. CONCLUSIONS: Although TDM evidence for newer LAI formulations is limited, this review suggests the use of TDM when switching an antipsychotic from oral to its LAI formulation. The application of TDM practice is more accurate for dose selection than the use of dose equivalents as it accounts more precisely for individual characteristics.

Potentially inappropriate medication in older psychiatric patients.

PURPOSE: Many psychotropic drugs are listed as potentially inappropriate medication (PIM) in the older population. Potentially inappropriate means that prescription of those drugs in older adults may cause significant harm. The objective of this study was to analyze the prevalence and sort of PIM prescribing in a naturalistic, real-world psychiatric setting. METHODS: The retrospective analysis gathered data from a large pharmacovigilance study, conducted at 10 psychiatric hospitals. Data from inpatients aged ≥ 65 years were included for the analysis. The number and sort of PIM, as defined by the German PRISCUS list, were controlled by analyzing the patients' medication profile. RESULTS: In total, 4760 patient cases (59.2% female) with a mean (mean ± standard deviation (SD)) age of 77.33 ± 7.77 years were included into the study. Altogether, 1615 cases (33.9%) received at least 1 PRISCUS-PIM per day (regular and as-needed medication included). The most frequently prescribed PRISCUS-PIM (n = 2144) were zopiclone > 3.75 mg/day (n = 310), lorazepam > 2 mg/day (n = 269), haloperidol > 2 mg/day (n = 252), and diazepam (n = 182). Cases with PRISCUS-PIM were younger (75.7 vs. 78.2 years, p < 0.001) and had a longer (26 vs. 22 days, p < 0.001) hospital length of stay. Replacing benzodiazepines and z-substances, haloperidol > 2 mg, tricyclic antidepressants, first generation antihistaminergic drugs, and clonidine by non-PIM could reduce 69.9% of PRISCUS-PIM-prescribing. CONCLUSIONS: The prevalence of PRISCUS-PIM is high in the hospitalized psychiatric setting. Rational deprescribing of inappropriate anticholinergics, benzodiazepines, and antipsychotics in the older population is a key component to reduce the risk of adverse drug reactions. More tolerable medications should be prescribed.

Polypharmacy and the risk of drug-drug interactions and potentially inappropriate medications in hospital psychiatry.

PURPOSE: The aim of this study was to analyze the epidemiology of polypharmacy in hospital psychiatry. Another aim was to investigate predictors of the number of drugs taken and the associated risks of drug-drug interactions and potentially inappropriate medications in the elderly. METHODS: Daily prescription data were obtained from a pharmacovigilance project sponsored by the Innovations Funds of the German Federal Joint Committee. RESULTS: The study included 47 071 inpatient hospital cases from eight different study centers. The mean number of different drugs during the entire stay was 6.1 (psychotropic drugs = 2.7; others = 3.4). The mean number of drugs per day was 3.8 (psychotropic drugs = 1.6; others = 2.2). One third of cases received at least five different drugs per day on average during their hospital stay (polypharmacy). Fifty-one percent of patients received more than one psychotropic drug simultaneously. Hospital cases with polypharmacy were 18 years older (p < 0.001), more likely to be female (52% vs. 40%, p < 0.001) and had more comorbidities (5 vs. 2, p < 0.001) than hospital cases without polypharmacy. The risks of drug-drug interactions (OR = 3.7; 95% CI = 3.5-3.9) and potentially inappropriate medication use in the elderly (OR = 2.2; CI = 1.9-2.5) substantially increased in patients that received polypharmacy. CONCLUSION: Polypharmacy is frequent in clinical care. The number of used drugs is a proven risk factor of adverse drug reactions due to drug-drug interactions and potentially inappropriate medication use in the elderly. The potential interactions and the specific pharmacokinetics and -dynamics of older patients should always be considered when multiple drugs are used.

Prevalence and sort of pharmacokinetic drug-drug interactions in hospitalized psychiatric patients.

Psychiatric patients are high-risk patients for the development of pharmacokinetic drug-drug interactions (DDIs), leading to highly variable (victim) drug serum concentrations. Avoiding and targeting high-risk drug combinations could reduce preventable adverse drug reactions (ADRs). Pharmacokinetic cytochrome P450 (CYP)-mediated DDIs are often predictable and, therefore, preventable. The retrospective, longitudinal analysis used informations from a large pharmacovigilance study (Optimization of pharmacological treatment in hospitalized psychiatric patients study, study number 01VSF16009, 01/2017), conducted in 10 psychiatric hospitals in Germany. Medication data were examined for the co-prescription of clinically relevant CYP inhibitors or inducers and substrates of these enzymes (victim drugs). In total, data from 27,396 patient cases (45.6% female) with a mean (mean ± standard deviation (SD)) age of 47.3 ± 18.3 years were available for analysis. CYP inhibitors or inducers were at least once prescribed in 14.4% (n = 3946) of the cases. The most frequently prescribed CYP inhibitors were melperone (n = 2504, 28.1%) and duloxetine (n = 1324, 14.9%). Overall, 51.0% of the cases taking melperone were combined with a victim drug (n = 1288). Carbamazepine was the most frequently prescribed CYP inducer (n = 733, 88.8%). Combinations with victim drugs were detected for 58% (n = 427) of cases on medication with carbamazepine. Finally, a DDI was detected in 43.6% of the cases in which a CYP inhibitor or inducer was prescribed. The frequency of CYP-mediated DDI is considerably high in the psychiatric setting. Physicians should be aware of the CYP inhibitory and inducing potential of psychotropic and internistic drugs (especially, melperone).

QTc Time Correlates with Amitriptyline and Venlafaxine Serum Levels in Elderly Psychiatric Inpatients.

INTRODUCTION: Many antidepressants cause QT prolongation but the classification of cardiac risk of these drugs varies markedly in different published lists. This retrospective study analyzed the correlation of QTc time with amitriptyline and venlafaxine serum level in elderly psychiatric inpatients. METHODS: Elderly inpatients aged≥65 years for whom venlafaxine or amitriptyline serum level had been measured were selected retrospectively from a therapeutic drug monitoring database and screened for an electrocardiogram measurement at the time of blood withdrawal. The correlation of amitriptyline or venlafaxine serum levels with QTc time was examined by using Pearson's correlation analysis. RESULTS: Amitriptyline serum levels (n=11) correlated significantly with QTc time (r=0.918, p<0.001, CI 95%). Venlafaxine serum levels (n=27) also correlated significantly with QTc time (r=0.382, p<0.05, CI 95%). DISCUSSION: Amitriptyline and venlafaxine induce QT prolongation depending on drug concentrations in blood. Its extent, however, is very low when drug serum levels are within the therapeutic range. Future pharmacokinetic studies that correlate drug serum level and QT time should classify the cardiac risk of drugs based on the grade of the regression line in relation to the therapeutic range.

[Therapeutic drug monitoring in neuropsychopharmacology : Summary of the consensus guidelines 2017 of the TDM task force of the AGNP]. / Therapeutisches Drug-Monitoring in der Neuropsychopharmakologie : Zusammenfassung der Konsensusleitlinien 2017 der TDM-Arbeitsgruppe der AGNP.

Therapeutic drug monitoring (TDM) is the quantification and interpretation of drug concentrations in blood serum or plasma to optimize pharmacological therapy. TDM is an instrument with which the high interindividual variability of pharmacokinetics of patients can be identified and therefore enables a personalized pharmacotherapy. In September 2017 the TDM task force of the Working Group for Neuropsychopharmacology and Pharmacopsychiatry (AGNP) published an update of the consensus guidelines on TDM published in 2011. This article summarizes the essential statements for the clinical practice in psychiatry and neurology.

Comedication of Valproic Acid Is Associated With Increased Metabolism of Clozapine.

OBJECTIVES: Valproic acid and clozapine are drugs commonly used in the treatment of schizophrenic and schizoaffective disorders. Pharmacokinetic interactions of valproic acid with several drugs are well known, yet results concerning the interaction with clozapine are inconsistent. METHODS: Steady-state dose-corrected serum concentrations of clozapine and its main metabolite norclozapine were retrospectively analyzed in 45 patients receiving both clozapine and valproic acid. Controls were matched for sex, age, smoking, comedication, and inflammatory response. RESULTS: The group receiving comedication with valproic acid showed significantly lower median dose-corrected serum concentrations of norclozapine (0.44 [0.27-0.58] (ng/mL)/(mg/d) vs 0.78 [0.60-1.07] (ng/mL)/(mg/d)) as well as metabolite to parent compound ratios (0.40 [0.36-0.47] vs 0.71 [0.58-0.84]) by approximately 44%. Dose-corrected serum concentrations of clozapine were not significantly lower. The effect of valproic acid was independent of sex and smoking. CONCLUSIONS: Comedication with valproic acid accelerated metabolism of clozapine with predominant effects on the degradation of norclozapine. Therapeutic drug monitoring should be applied to guide individual patient responses upon initiation of comedication.

Depression and cognitive deficits as long-term consequences of thrombotic thrombocytopenic purpura.

BACKGROUND: Thrombotic thrombocytopenic purpura (TTP) is an acute life-threatening microangiopathy with a tendency of relapse characterized by consumptive thrombocytopenia, microangiopathic hemolytic anemia, and spontaneous von Willebrand factor-induced platelet clumping leading to microthrombi. The brain is frequently affected by microthrombi leading to neurologic abnormalities of varying severity. STUDY DESIGN AND METHODS: The aim of this observational cohort study was to investigate the prevalence of depression and cognitive deficits in 104 patients having survived acute TTP. TTP survivors were repeatedly assessed by means of different standardized questionnaires to evaluate depression (IDS-SR) and mental performance (FLei). We received answers of 104 individual TTP patients and 55 of them participated in both surveys. RESULTS: Seventy-one of the 104 responding TTP patients (68%) suffered from depression and the severity of depression was similar in both surveys performed 1 year apart. Furthermore, TTP patients had considerably lower cognitive performance than controls. There was no correlation between prevalence of depression and cognitive deficits and the number and the severity of acute episodes. Impairment of mental performance correlated with the severity of depression (rs  = 0.779). CONCLUSION: The prevalence of depression and cognitive deficits was significantly higher in TTP patients. Cognitive impairment seemed to be a consequence of depression, almost independently of number and severity of TTP episodes.

Melperone but not bisoprolol or metoprolol is a clinically relevant inhibitor of CYP2D6: evidence from a therapeutic drug monitoring survey.

Cytochrome P450 enzymes (CYP) can be inhibited or induced by drugs, resulting in clinically significant drug-drug interactions that can cause unanticipated adverse reactions or therapeutic failures. The objective of the study was to analyze the in vivo inhibitory potential of the beta-blockers bisoprolol and metoprolol as well as the low-potency antipsychotic melperone on CYP2D6. By utilizing a large therapeutic drug monitoring database of 2874 samples, data from patients who had been treated with venlafaxine (VEN) either without (control group) or with a concomitant medication with bisoprolol, metoprolol or melperone were evaluated retrospectively to study the CYP2D6-catalyzed O-demethylation to O-desmethylvenlafaxine (ODVEN). Dose-adjusted serum levels (C/D) of VEN and ODVEN as well as the metabolic ratios (ODVEN/VEN) were computed for the four groups and compared using Kruskal-Wallis test. In total, 381 patients could be included for analysis. No significant difference was found in the median C/D (VEN), C/D (ODVEN) or C/D of the active moiety (VEN + ODVEN) in either the metoprolol (N = 103) or bisoprolol group (N = 101), compared to the control group (N = 108). In contrast, a significantly higher median C/D (VEN) (0.79 ng/ml/mg, range 0.13-5.73 ng/ml/mg) (P < 0.01) was found in the melperone group (N = 69), compared to the control group (0.46 ng/ml/mg, range 0.02-7.39 ng/ml/mg). A significant decrease (P < 0.01) was solely found in the median metabolic ratios of ODVEN/VEN between the melperone group (0.90, range 0.14-15.15), compared to the control group (2.39, range 0.06-15.31). The results of this study provided evidence that melperone but not bisoprolol or metoprolol has a clinically relevant inhibitory potential on CYP2D6.

Therapeutic drug monitoring for patients with Alzheimer dementia to improve treatment with donepezil.

BACKGROUND: Aiming to verify that therapeutic drug monitoring has the potential to optimize treatment with acetylcholine esterase inhibitors of patients with Alzheimer dementia, this study investigated whether serum concentrations of donepezil are associated with clinical improvement. METHODS: Clinical improvement was measured using the clinical global impression (CGI) scale, and donepezil concentrations were measured in serum by a high-performance liquid chromatographic method with spectrophotometric detection. RESULTS: In total, 206 serum samples from 106 patients (49.5% female) were retrospectively available for analysis. Patients included were treated under everyday conditions. Their mean ± SD age was 72 ± 9 years, daily doses of donepezil were 5 and 10 mg, and their mean ± SD serum concentrations were 23 ± 9 and 47 ± 18 ng/mL, respectively. Serum concentrations correlated significantly (P < 0.001) with CGI scores (Pearson's correlation coefficient r = 0.511, P < 0.01). In patients who were "very much improved," according to their CGI score, the mean serum concentration was 66 ± 20 ng/mL and thus significantly higher (P < 0.01) than in patients with "minimal improvement" (29 ± 12 ng/mL). Receiver operating characteristics analysis suggests that donepezil serum concentrations of at least 50 ng/mL may be recommended for maximal clinical benefit. CONCLUSIONS: Because donepezil serum concentrations were highly variable between individual patients and the majority of patients exhibited concentrations that were below 50 ng/mL at therapeutic doses of 5 and 10 mg/d, it can be concluded that therapeutic drug monitoring may be used to enhance the effectiveness of donepezil treatment.

Side effects related to potentially inappropriate medications in elderly psychiatric patients under everyday pharmacotherapy.

PURPOSE: Potentially inappropriate medication (PIM) is suggested to give rise to adverse drug events. To study this suggestion for elderly psychiatric patients, an observational analysis related prescription of PRISCUS PIMs and drug-induced side effects in old aged (≥65 years) psychiatric inpatients and outpatients under conditions of everyday pharmacotherapy. METHODS: Request forms from a therapeutic drug monitoring (TDM) survey and medical files were screened for medication to identify PIMs of the PRISCUS list and assessed using the Udvalg for Kliniske Undersøgelser (UKU) side effect rating scale. RESULTS: From 914 TDM request forms, data were available for 168 patients (64.3 % female). Patients (mean ± SD age 73.0 ± 5.5 years) received by mean 6.4 ± 3.9 drugs per day. More than half of them (53.0 %, n = 89) had at least one PIM, inpatients 0.9 ± 0.8 and outpatients 0.5 ± 0.7. Predominant PIMs were hypnotic drugs (69 %) in inpatients and antipsychotic drugs (35.6 %) in outpatients. The number of PIMs correlated with the total number of drugs administered per day (Spearman correlation coefficient 0.225, p < 0.01, CI 95 %). Side effects were documented for 106 patients (63 %). Severity of side effects did not correlate significantly (p > 0.05) with number of PIMs. However, only 6 of 77 patients who took no PRISCUS PIMs but 2 of 3 patients who took 3 PRISCUS PIMs exhibited severe side effects. CONCLUSIONS: Though the prevalence for PIMs and side effects was high in old aged psychiatric inpatients and outpatients, PIMs could not be identified as major determinants of overall unwanted side effects. Nevertheless, prescription of PIMs should be minimized, especially of hypnotic drugs, to improve safety.

The value of drug and metabolite concentration in blood as a biomarker of psychopharmacological therapy.

Desirable and undesirable effects of a drug are related to its concentration at various sites of actions. For many psychotropic drugs, it has been shown that drug concentration in brain correlates with concentration in blood. The latter is also an available estimate of clearance and bioavailability. Its monitoring enables identification of multiple factors that have an impact on clinical outcomes, especially uncertain compliance and pharmacokinetic peculiarities. For this review we analysed for antidepressants if drug concentration in blood can be used as biomarker for psychopharmacological treatment. Systematic review of the literature revealed for new and old antidepressant drugs that drug and metabolite concentrations in blood are measures of the pharmacokinetic phenotype and related differentially to occupancy of primary target structures, therapeutic effects and unwanted anticholinergic, cardiac and other side effects. Drug concentration in blood can therefore be used as biomarker in clinical practice to guide psychopharmacological treatment with established antidepressant drugs. Monitoring of drug concentration is suitable to improve efficacy and safety of the pharmacotherapy, especially in elderly patients who require complex pharmacological therapies.

Systematic review and meta-analysis on the therapeutic reference range for escitalopram: Blood concentrations, clinical effects and serotonin transporter occupancy.

Introduction: A titration within a certain therapeutic reference range presupposes a relationship between the blood concentration and the therapeutic effect of a drug. However, this has not been systematically investigated for escitalopram. Furthermore, the recommended reference range disagrees with mean steady state concentrations (11-21 ng/ml) that are expected under the approved dose range (10-20 mg/day). This work systematically investigated the relationships between escitalopram dose, blood levels, clinical effects, and serotonin transporter occupancy. Methods: Following our previously published methodology, relevant articles were systematically searched and reviewed for escitalopram. Results: Of 1,032 articles screened, a total of 30 studies met the eligibility criteria. The included studies investigated escitalopram blood levels in relationship to clinical effects (9 studies) or moderating factors on escitalopram metabolism (12 studies) or serotonin transporter occupancy (9 studies). Overall, the evidence for an escitalopram concentration/effect relationship is low (level C). Conclusion: Based on our findings, we propose a target range of 20-40 ng/ml for antidepressant efficacy of escitalopram. In maintenance treatment, therapeutic response is expected, when titrating patients above the lower limit. The lower concentration threshold is strongly supported by findings from neuroimaging studies. The upper limit for escitalopram's reference range rather reflects a therapeutic maximum than a tolerability threshold, since the incidence of side effects in general is low. Concentrations above 40 ng/ml should not necessarily result in dose reductions in case of good clinical efficacy and tolerability. Dose-related escitalopram concentrations in different trials were more than twice the expected concentrations from guideline reports. Systematic review registration: [https://www.crd.york.ac.uk/PROSPERO/display_record.php?RecordID=215873], identifier [CRD42020215873].

Predicting the risk of drug-drug interactions in psychiatric hospitals: a retrospective longitudinal pharmacovigilance study.

OBJECTIVES: The aim was to use routine data available at a patient's admission to the hospital to predict polypharmacy and drug-drug interactions (DDI) and to evaluate the prediction performance with regard to its usefulness to support the efficient management of benefits and risks of drug prescriptions. DESIGN: Retrospective, longitudinal study. SETTING: We used data from a large multicentred pharmacovigilance project carried out in eight psychiatric hospitals in Hesse, Germany. PARTICIPANTS: Inpatient episodes consecutively discharged between 1 October 2017 and 30 September 2018 (year 1) or 1 January 2019 and 31 December 2019 (year 2). OUTCOME MEASURES: The proportion of rightly classified hospital episodes. METHODS: We used gradient boosting to predict respective outcomes. We tested the performance of our final models in unseen patients from another calendar year and separated the study sites used for training from the study sites used for performance testing. RESULTS: A total of 53 909 episodes were included in the study. The models' performance, as measured by the area under the receiver operating characteristic, was 'excellent' (0.83) and 'acceptable' (0.72) compared with common benchmarks for the prediction of polypharmacy and DDI, respectively. Both models were substantially better than a naive prediction based solely on basic diagnostic grouping. CONCLUSION: This study has shown that polypharmacy and DDI can be predicted from routine data at patient admission. These predictions could support an efficient management of benefits and risks of hospital prescriptions, for instance by including pharmaceutical supervision early after admission for patients at risk before pharmacological treatment is established.