Replicability and reproducibility of predictive models for diagnosis of depression among young adults using Electronic Health Records

BACKGROUND

Recent advances in machine learning combined with the growing availability of digitized health records offer new opportunities for improving early diagnosis of depression. An emerging body of research shows that Electronic Health Records can be used to accurately predict cases of depression on the basis of individual's primary care records. The successes of these studies are undeniable, but there is a growing concern that their results may not be replicable, which could cast doubt on their clinical usefulness.

METHODS

To address this issue in the present paper, we set out to reproduce and replicate the work by Nichols et al. (2018), who trained predictive models of depression among young adults using Electronic Healthcare Records. Our contribution consists of three parts. First, we attempt to replicate the methodology used by the original authors, acquiring a more up-to-date set of primary health care records to the same specification and reproducing their data processing and analysis. Second, we test models presented in the original paper on our own data, thus providing out-of-sample prediction of the predictive models. Third, we extend past work by considering several novel machine-learning approaches in an attempt to improve the predictive accuracy achieved in the original work.

RESULTS

In summary, our results demonstrate that the work of Nichols et al. is largely reproducible and replicable. This was the case both for the replication of the original model and the out-of-sample replication applying NRCBM coefficients to our new EHRs data. Although alternative predictive models did not improve model performance over standard logistic regression, our results indicate that stepwise variable selection is not stable even in the case of large data sets.

CONCLUSION

We discuss the challenges associated with the research on mental health and Electronic Health Records, including the need to produce interpretable and robust models. We demonstrated some potential issues associated with the reliance on EHRs, including changes in the regulations and guidelines (such as the QOF guidelines in the UK) and reliance on visits to GP as a predictor of specific disorders.

Selective serotonin reuptake inhibitors and suicidality in children and young adults: analyses of pharmacovigilance databases

BACKGROUND

Since the warnings by the United States (US) and European regulatory authorities in 2004 and 2005 it had been discussed whether there is some link between selective serotonin reuptake inhibitors (SSRIs) and suicidality in the pediatric population. The aim of our study was to describe trends and patterns in spontaneous reporting data referring to suicidality in children, adolescents and young adults treated with SSRI after the warnings.

METHODS

Descriptive analyses of reports for 0-24 year olds referring to suicide/suicidal ideations, self-harms and overdoses with SSRIs reported as suspected submitted to the US (FAERS) and the European (EudraVigilance) adverse drug reaction databases until 2019 were performed. The causal relationship was assessed in accordance with the WHO criteria for the European reports. For Germany, prescription data for SSRIs were provided and reporting rates (number of reports/number of prescriptions) were calculated for the reports with possible causal relationship (so called "confirmed reports").

RESULTS

Since 2004, the number of reports referring to suicide/suicidal ideations, self-harm and overdoses increased steadily in the US and EU. However, only a slight increase was seen for the confirmed EU reports. After 2008, the proportion of reports informing about suicidal ideations increased, while the proportion of fatal suicide attempts decreased. Reporting rates were higher for females and adolescents (12-18 years).

CONCLUSIONS

Our results demonstrate the importance of further monitoring suicidality in 0-24 year olds treated with SSRI in order to recognize suicidality early avoiding fatal suicide attempts. The higher reporting rates for females and adolescents should be further investigated.

Value of a web-based pediatric drug information system to prevent serious adverse drug reactions in child and adolescent psychiatry

Psychotropic drugs are frequently prescribed 'off-label' to children and adolescents and carry the risk of serious adverse drug reactions (sADR). We examined the frequency of sADRs of psychotropic drugs in pediatric inpatients and explored their potential preventability through following the recommendations of a web-based pediatric drug information system (PDIS). The potential socio-economic impacts of using this online system is also addressed. Routine clinical data from all inpatients treated in a child and adolescent psychiatry department between January 2017 and December 2018 were retrospectively examined for the occurrence of sADRs as defined by the European Medicines Agency. The preventability of the sADRs was assessed based on the information of the PDIS. Furthermore, the expected prolongation of the hospital stay due to sADRs was calculated as well as the associated treatment costs. The study was supported by the Innovation Fund of the Joint Federal Committee, grant number 01NVF16021. In total, 1036 patients were screened of whom 658 (63.5%) received psychopharmacological treatment. In 53 (8.1%) of these patients 54 sADRs were documented, of which 37 sADRs were identified as potentially preventable through PDIS. Mitigating sADR through PDIS would likely have prevented prolonged hospital stays and conferred considerable savings for health insurance companies. PDIS provides systematic and evidence-based information about pediatric psychopharmacotherapy and helps to prevent prescribing errors. Therefore, PDIS is a useful tool to increase drug therapy safety in child and adolescent psychiatry. Further prospective studies are needed to confirm the results.

Rates of psychotropic medicine prescribing in paediatric populations in Australian general practice from 2000-2016

General practitioner (GP) prescribing of psychotropic medicines to paediatric patients is increasing across countries, sparking the need for additional research into this field. We examined prescribing rates, GP and patient characteristics and indications associated with prescribing psychotropic medicines to paediatric patients in Australian general practice, using data from the Bettering the Evaluation and Care of Health (BEACH) program. We extracted all encounters with children aged 3 to 17 from 2000 to 2016. Psychotropic medicines were defined as those in the ATC codes N05 (Psycholeptics) and N06 (Psychoanaleptics). Of the 144,397 encounters, GPs prescribed 1829 psychotropic medicines to paediatric patients at an average rate of 1.16 prescriptions per 100 encounters (95% confidence interval 1.09-1.23). We found that the rate of psychotropic medicines prescribed to paediatric patients in Australian general practice increased. Patients who were adolescent, female, socio-economically disadvantaged or from an English-speaking background were significantly more likely to be prescribed a psychotropic medicine. GP practices in remote or regional areas and Australian graduate GPs were more likely to prescribe psychotropic medicines to paediatric patients. Depression, attention deficit hyperactivity disorder, anxiety and autism were the most common psychiatric indications managed with psychotropic medicines. Antidepressants, psychostimulants, benzodiazepines, antipsychotics and other psychotropic medicines were prescribed, signifying a high rate of off-label use. Sertraline was the most common psychotropic medicine prescribed, followed by fluoxetine and methylphenidate. Future studies involving data from other prescribers, e.g. paediatricians and psychiatrists, and studies linking prescribed medicines to their indications may widen our understanding of psychotropic medicine prescribing in Australian paediatric patients.

Antidepressant drug use and regional prescribing patterns in Germany: results from a large population-based study

OBJECTIVE

To consider antidepressant prescribing on a population level with a focus on regional prescribing patterns in Germany.

BASIC METHODS

Based on data from about 70 million individuals covered by all statutory health insurance funds in Germany in 2010, the prevalence of antidepressant use (overall, for drug classes and individual drugs) was calculated stratified by age and sex. Regional analyses were performed on a state and also on a district level.

MAIN RESULTS

The study population comprised 68 427 464 (female: 53.0%) persons, of which 5 052 293 (7.4%) were prescribed at least one antidepressant. The most frequently prescribed drug class was tricyclic antidepressants whereas on a substance level citalopram was most commonly used. Antidepressant prescribing was lowest in children and adolescents (0.2%) and most common in persons aged 70 years and older (13.4%). Women more often received antidepressants than men (9.7% vs. 4.8%). Prevalence of antidepressant use varied between 8.7% (Saarland) and 6.3% (Saxony-Anhalt) and was generally highest in the southwestern and lowest in the eastern states. Accordingly, districts with the highest prevalence were located in the southwestern states.

PRINCIPAL CONCLUSIONS

Antidepressant use in Germany varied considerably by age and sex and also on a state and district level.

Global prevalence of antidepressant drug utilization in the community: protocol for a systematic review

INTRODUCTION

Antidepressant drugs are the most frequently prescribed medication for mental disorders. They are also used off-label and for non-psychiatric indications. Prescriptions of antidepressants have increased in the last decades, but no systematic review exists on the extent of their use in the community.

METHODS AND ANALYSIS

We will conduct a systematic review to estimate the prevalence of antidepressant use in the community. We will search for studies published from 1 January 2010 in the Embase and MEDLINE databases using a combination of controlled vocabulary and keywords adjusted for each database without any language restriction. The main inclusion criterion is the presence of prevalence data of antidepressant utilization. Thus, we will include all studies with a descriptive observational design reporting the prevalence of antidepressant use in the community. Study selection (by title/abstract and full-text screening) and data extraction for included studies will be independently conducted by pairs of reviewers. We will then synthesize the data on the prevalence of antidepressant use in individuals living in the community. If possible, we will perform a meta-analysis to generate prevalence-pooled estimates. If the data allows it, we will conduct subgroup analyses by antidepressant class, age, sex, country and other sociodemographic categories. We will evaluate the risk of bias for each included study through a quality assessment using the Joanna Briggs Institute Critical Appraisal tool: Checklist for Studies Reporting Prevalence Data. DistillerSR software will be used for the management of this review.

ETHICS AND DISSEMINATION

Ethical approval is not required for this review as it will not directly involve human or animal subjects. The findings of our systematic review will be disseminated through publications in peer-reviewed journals, the Qualaxia Network (https://qualaxia.org), presentations at international conferences on mental health and pharmacoepidemiology, as well as general public events.

PROSPERO REGISTRATION NUMBER

CRD42021247423.

Dispensation Patterns of Glucose-Lowering Drugs in Newly Diagnosed Type 2 Diabetes: Routine Data Analysis of Insurance Claims in Germany

AIMS

To describe dispensation patterns of glucose-lowering drugs in newly diagnosed type 2 diabetes in Germany.

MATERIALS AND METHODS

Based on claims data from four statutory health insurances (German Pharmacoepidemiological Research Database,>25 million insurants), all individuals with newly diagnosed type 2 diabetes were identified. Eligible patients had a first diagnosis for type 2 diabetes between January 2012 and December 2016. We analyzed the dispensation patterns of first-line glucose-lowering therapies initiated in the year after diabetes diagnosis and patterns of second-line therapies dispensed one year after first-line treatment.

RESULTS

A total of 356,647 individuals with newly diagnosed type 2 diabetes were included (average age [SD]: 63.5 [13.4] years; 49.3% males). Of the 31.6% of individuals who were pharmacologically treated in the year after diagnosis, metformin monotherapy was most frequently dispensed (73.1%), followed by dual therapy of metformin and dipeptidyl peptidase-4 inhibitors (DPP-4is) (6.4%), and monotherapy with DPP-4is (2.9%). From 2012 through 2016, sulfonylurea dispensations were reduced by more than 50%. Dispensations for combination therapies with DPP-4is increased up to 10.6%. Glucagon-like peptide-1 receptor agonists and sodium-glucose co-transporter-2 inhibitors contributed to 2% of all treatments. After a median of 5 months, 20.0% of individuals on pharmacological therapy initiated second-line glucose-lowering treatment.

CONCLUSIONS

Data from German statutory health insurances (2012 to 2016) showed that most individuals with newly diagnosed type 2 diabetes were dispensed metformin monotherapy in line with diabetes care guidelines. A substantial decrease in the use of sulfonylureas was observed after the introduction of DPP-4i and GLP-1 receptor agonists.

Adverse Events Related to Off-Label Drugs Using Spontaneous Adverse Event Reporting Systems

Purpose

The purpose of this study was to investigate the adverse events (AEs) related to the use of off-label drugs.

Materials and Methods

A cross-sectional study was carried out using available data pertaining to off-label drug were sourced from U.S. FDA spontaneous adverse drug reaction reporting database (FAERS) and Korea Adverse Event Reporting System database (KIDS-KD) for the years 2014 to 2018. The number and frequencies of AE cases were calculated. Disproportionality was analyzed using the proportional reporting ratio (PRR), the reporting odds ratio (ROR), the information component (IC), and the empirical Bayes geometric mean (EBGM) methods.

Results

The reported AEs associated with off-label drug use were more common among older patients compared with younger patients. Gastric nonspecific symptoms and therapeutic procedure (4.16–4.57%) and haemorrage term (4.16–5.29%) were the most common AE symptoms and antithrombotic agents and immunosuppressants were the drugs most commonly reported to cause AEs in FAERS. Secondary term events (43.45–48.62%) including inappropriate schedule of drug administration and medication error were the most common AEs, and immunosuppressants and antipsychotics were the most common AE-related drugs from KIDS-KD. The numbers of reported AEs in new drug categories such as other antineoplastic agents trended to increase from 2014 to 2018 in both datasets.

Conclusion

The numbers of reported AEs with off-label drug increased annually. AEs associated with off-label drugs may have a significant impact on older patients. Healthcare experts should be concerned about prescriptions of off-label drugs, especially anticoagulants and newly developed drugs such as immunosuppressants and antineoplastic agents.

An International Comparison of the Information in the Regulatory-Approved Drug Labeling and Prescribing Guidelines for Pediatric Depression

Objectives: To determine the differences in information between prescribing guidelines and drug labeling, as well as to compare the approval of psychotropic medicines for major depression in pediatric patients ("pediatric depression") across countries. Methods: The recommendations of The Maudsley Prescribing Guidelines in Psychiatry (MPGP) for the treatment of pediatric depression (<18 years) were compared against the regulatory-approved drug-labeling documents from the United Kingdom, Australia, New Zealand, Canada, and the United States. The use of medicines outside of their regulatory approval is defined as off-label use, so differences between the drug labeling and MPGP were characterized according to unapproved age, indication, dosage, or route of administration. Information in the drug labeling was also compared across countries. Results: MPGP provides recommendations for 6 medicines for the treatment of pediatric depression, for which, 30 drug labeling were retrieved. Three of 30 drug labeling were consistent with MPGP recommendations (fluoxetine in the United Kingdom, fluoxetine and escitalopram in the United States). Differences in information between MPGP and the drug labeling were identified in 26 of 30 drug labeling analyzed, most often due to age (24/26) followed by indication (2/26). No differences pertaining to dosage or route of administration information were identified. The number of approved psychotropic medicines varied across the studied countries and we found cross-country discrepancies in information in the drug labeling. Conclusion: Significant differences in information exists between MPGP and the drug labeling for psychotropic medicines for pediatric depression, due to unapproved ages or indications. Additionally, approval information in the drug labeling are not consistent across countries. Further research into reasons for variability and impact on practice may be warranted.

[Patterns of antidepressant dispensing among children and adolescents between 2015 and 2017 in the French region of Midi-Pyrenees]

BACKGROUND

In France, consumption of antidepressant drugs by children and teenagers has been on the rise, even though recommendations for use are limited due to their association with serious adverse reactions. The objective of this study was to describe the patterns of antidepressant drug dispensing among children and adolescents in the French region of Midi-Pyrenees.

METHOD

The data regarding the Midi-Pyrenees region were extracted from the French Health Insurance Database (SNDS) using their ATC codes. The reimbursements for antidepressants, benzodiazepines and neuroleptics dispensed to patients from 6 to 17 years old between January 2015 and June 2017 were analyzed, after which data on the antidepressants themselves were selected. The population was divided into sub-groups according to age (children : 6 to 11 years old, teenagers: 12 to 17 years old).

RESULTS

During the 30 months analyzed, 12,783 antidepressants were dispensed to 3506 patients. The antidepressants were primarily issued (90%) to teenagers. In terms of prevalence, 24.7% of the teenagers had amitriptyline delivered at least once, while 31.2% of them received sertraline at least once. Regarding total amount of antidepressant issuances, sertraline was first in both the child (26.9%) and the teenage (40.7%) populations. Benzodiazepine with an antidepressant was issued to 35.1% of the children and teenagers.

CONCLUSIONS

Amitriptyline was the most widely dispensed antidepressant among children, and sertraline among teenagers. However, fluoxetine is recommended as the first-line treatment for depression affecting this population. A psychotropic drug prescription assistance website such as psychotropes.fr addressed to general practitioners might improve the implementation of recommendations and guidelines.

How often are antidepressants prescribed off-label among older adults in Germany? A claims data analysis

AIM

To estimate the extent of off-label prescribing of antidepressants in older adults and to characterize patients with off-label vs on-label prescriptions of antidepressants using a large German health claims database.

METHODS

Using data from the German Pharmacoepidemiological Research Database (GePaRD), we conducted a cross-sectional study in adults aged 65 years or older with a dispensation of an antidepressant between 1 January 2009 and 31 December 2015 after a period of 365 days without such a dispensation. We assessed the overall and annual proportion of off-label prescriptions of antidepressants by class and individual substance.

RESULTS

Among 263 276 incident users of antidepressants, the proportion of off-label prescribing was 43.6% (95% CI 43.4-43.8%) with little variation between 2009 and 2015 (42.2-44.4%). The proportion of off-label use was higher in men (49%) than women (41%). While the proportion of off-label prescriptions was highest for tri- and tetracyclic antidepressants with 56.2% (amitriptyline 54.6%, maximum 65.9% for trimipramine), it amounted to 41.8% for selective serotonin reuptake inhibitors (citalopram 41.6%, maximum 46.0% for escitalopram) and was 51.2% for mirtazapine. Indicators of overall morbidity were similar in both groups, eg, pain was coded in 72% of off-label users vs 77% of on-label users (insomnia 20% vs 24%).

CONCLUSION

Our study suggests a high prevalence of off-label antidepressant use among older adults in Germany, which was not restricted to certain classes of antidepressants or individual antidepressants. Given the unclear risk-benefit ratio, studies investigating the safety of off-label use among older adults for individual antidepressants are urgently needed.

Individual Antidepressants and the Risk of Fractures in Older Adults: A New User Active Comparator Study

Objective

To determine the risk of hip–pelvis and other non-vertebral fractures in older adults using antidepressants (ADs).

Methods

We conducted a case–control study nested in a cohort of new users of ADs aged ≥65 years without prior hip–pelvis or other non-vertebral fractures, identified in the German Pharmacoepidemiological Research Database (GePaRD) during 2005–2014. Cases were patients first hospitalized for hip–pelvis or other non-vertebral fractures. Up to 100 controls per case were selected using incidence density sampling. AD use was ascertained at index date (ID) based on the supply of last dispensing. Adjusted odds ratios (aORs) and 95% confidence intervals (CIs) were estimated using conditional logistic regression with current users of mirtazapine as reference (active comparator).

Results

A total of 39,853 cases of hip–pelvis fracture (80% women, median age 81 years) and 31,577 cases of other fractures (84% women, median age 79 years) were matched to >3 million controls. For hip–pelvis fracture, aORs in current users were about 1.3 with little variation between individual ADs, ranging from 1.33 for citalopram (95% CI 1.27–1.39) to 1.28 for amitriptyline (1.21–1.35). For other fractures, the aORs were highest in current users of citalopram (1.50; 1.42–1.58) and duloxetine (1.54; 1.39–1.71) and lowest for amitriptyline (1.18; 1.11–1.26) and trimipramine (1.16; 1.03–1.29). For all examined ADs, the aORs were higher for other fractures than for hip–pelvis fracture.

Conclusion

The risk of fractures varies between ADs, but for most agents is higher than the risk for mirtazapine. When treating older adults with ADs, prescribers should carefully consider the risk profile of individual ADs regarding fractures, which are a major health problem in this population.

Pharmacoepidemiological research for the development and evaluation of drugs in pediatrics

New regulations have come into force in Europe and the US establishing the pediatric development as an integral part of the early development of medicinal products. Parallel to the advances in pediatric clinical research, it became obvious that all available sources and research tools to gather valuable information for the safe and efficacious prescription of medicines in children should be used. Real-life, pharmacoepidemiological studies provide information that contribute to the better knowledge of drug utilization, effects and safety in the pediatric population and thereby, a better prescribing in children. In this paper, we suggest some possible applications, provide examples of impact of pharmacoepidemiological and pharmacovigilance studies and expose future perspectives in pediatric pharmacoepidemiology.

Antidepressant prescription patterns and CNS polypharmacy with antidepressants among children, adolescents, and young adults: a population-based study in Sweden

This study examines trends in antidepressant drug dispensations among young people aged 0-24 years in Sweden during the period 2006-2013, as well as prescription patterns and central nervous system (CNS) polypharmacy with antidepressants. Using linkage of Swedish national registers, we identified all Swedish residents aged 0-24 years that collected at least one antidepressant prescription (here defined as antidepressant users) between 1 January 2006 and 31 December 2013 (n = 174,237), and categorized them as children (0-11 years), adolescents (12-17 years), and young adults (18-24 years). Prevalence of antidepressant dispensation rose from 1.4 to 2.1% between 2006 and 2013, with the greatest relative increase in adolescents [by 97.8% in males (from 0.6 to 1.3%) and by 86.3% in females (from 1.1 to 2.1%)]. Most individuals across age categories were prescribed selective serotonin reuptake inhibitors, received their prescriptions from psychiatric specialist care, and had treatment periods of over 12 months. Prevalence of CNS polypharmacy (dispensation of other CNS drug classes in addition to antidepressants) increased across age categories, with an overall increase in prevalence from 52.4% in 2006 to 62.1% in 2013. Children experienced the largest increase in polypharmacy of three or more psychotropic drug classes (4.4-10.1%). Anxiolytics, hypnotics, and sedatives comprised the most common additional CNS drug class among persons who were prescribed antidepressants. These findings show that the dispensation of antidepressants among the young is prevalent and growing in Sweden. The substantial degree of CNS polypharmacy in young patients receiving antidepressants requires careful monitoring and further research into potential benefits and harms.