Adverse drug reactions in geriatric psychiatry—retrospective cohort study of a 6-year period

Objective To investigate the frequency and characteristics of adverse drug reactions (ADRs) that occurred on the gerontopsychiatric ward of Hannover Medical School over a 6-year period. Design Retrospective monocentric cohort study. Results Six hundred thirty-four patient cases (mean age 76.6 ± 7.1 years; 67.2% female) were analysed. In total, 92 ADRs in 56 patient cases were registered in the study population. The overall ADR prevalence, the ADR prevalence upon hospital admission, and the ADR prevalence during hospitalisation were 8.8%, 6.3%, and 4.9%, respectively. The most frequent ADRs were extrapyramidal symptoms, alterations in blood pressure or heart rate, and electrolyte disturbances. Of note, two cases of asystole and one case of obstructive airway symptoms related to general anaesthesia in the context of electroconvulsive therapy (ECT) were detected. The presence of coronary heart disease was associated with an increased risk of ADR occurrence (odds ratio (OR) 2.92, 95% confidence interval (CI) 1.37–6.22), while the presence of dementia was associated with a decreased risk of ADR development (OR 0.45, 95% CI 0.23–0.89). Conclusions Type and prevalence of ADRs in the present study were largely in accordance with previous reports. By contrast, we did not observe a relationship between advanced age or female sex and ADR occurrence. We detected a risk signal for cardiopulmonary ADRs related to general anaesthesia in the context of ECT that warrants further investigation. Elderly psychiatric patients should be carefully screened for cardiopulmonary comorbidities before initiation of ECT. Supplementary Information The online version contains supplementary material available at 10.1007/s11845-023-03300-1.


Introduction
An adverse drug reaction (ADR) can be defined as "an appreciably harmful or unpleasant reaction, resulting from an intervention related to the use of a medicinal product, which predicts hazard from future administration and warrants prevention or specific treatment, or alteration of the dosage regimen, or withdrawal of the product" [1].
The European Commission estimates that approximately 5% of all ADRs lead to hospitalisation and that 5% of inpatients develop an ADR during their hospital stay.Moreover, ADRs constitute the fifth leading cause of death in hospital [2].The risk of ADR occurrence has been reported to increase significantly with advancing age due to age-associated multimorbidity and polypharmacy [3,4], inadequate prescribing [5], and insufficient monitoring of medications [6].In addition, age-related physiological changes in pharmacokinetics and pharmacodynamics are considered to increase the likelihood of ADR occurrence in elderly people [7].In the European Union, over 20% of citizens are 65 years of age or older, and in the next five decades, the proportion of this age group is expected to rise by one-third [8].A study that investigated mental disorders in the German population found that 20% of people aged 65-79 years suffer from a mental illness [9,10].
Extensive information about ADRs in the elderly general population is available.A meta-analysis by Oscanoa and colleagues concluded that nearly one in ten hospital admissions in older people was due to an ADR [11].Alhawassi and co-workers estimated that 11.5% of patients experienced an ADR during hospitalisation [12].In a recent systematic review and meta-analysis by Jennings et al., the prevalence of ADRs in hospitalised older adults was found to be even higher (16%), and five clinical presentations-fluid and electrolyte disturbances, gastrointestinal motility and defaecation disorders, renal disorders, hypotension/blood pressure dysregulation disorders/shock, and delirium-accounted for almost half of all ADRs [13].
By contrast, there is a paucity of data about the frequency and characteristics of ADRs in elderly patients suffering from psychiatric disorders.A large German registry study showed that people with severe mental illness suffer more frequently from somatic comorbidities compared to people without severe mental illness, leading to an elevated mortality rate and a loss of up to 12.3 life years [14].Furthermore, the use of psychotropic drugs is associated with an increased risk of many somatic disorders [15].
Consequently, elderly people suffering from psychiatric disorders must be considered a vulnerable patient population that is particularly susceptible to the occurrence of ADRs.In order to protect elderly psychiatric patients more effectively from the occurrence of ADRs and their sequelae, it is of paramount importance to dissect the precise causes of ADRs in this population.The aim of the present study was to investigate the frequency and characteristics of ADRs in a cohort of elderly psychiatric inpatients over a 6-year period.Based on the study results, appropriate ADR prevention strategies shall be devised in the future.

Ethics approval
This study was approved by the Ethics Committee of Hannover Medical School (No. 9496_BO_K_2020) and was conducted in accordance with the Declaration of Helsinki (1964) and its later amendments (current version from 2013).

Study design and study site
The study was designed as a retrospective monocentric cohort study of a 6-year period (01 January 2014 to 31 December 2019).The study was carried out at the gerontopsychiatric ward (27-bed facility) of the Department of Psychiatry, Social Psychiatry and Psychotherapy of Hannover Medical School (a university hospital in northern Germany).

Eligibility criteria
Patients were eligible for enrolment in the study (a) if they were ≥ 65 years of age, (b) if they were treated on the gerontopsychiatric ward of the Department of Psychiatry, Social Psychiatry and Psychotherapy of Hannover Medical School between 01 January 2014 and 31 December 2019, and (c) if they or their legal representative had provided written informed consent for the use of patient-related data for research purposes.

Patient selection and data acquisition
The study population was identified by the Enterprise Clinical Research Data Warehouse (ECRDW) [16] using structured medical data such as demographic characteristics, diagnoses, basic information on medications, and laboratory values.The ECRDW is a medical data repository which is maintained by Hannover Medical School (MHH) Information Technology and which comprises data from over 2.2 million patients.As one of the largest medical data repositories worldwide, the ECRDW has contributed to more than 300 clinical research projects of various medical disciplines (e.g.refs.[17][18][19]).Details on patient medication-such as dosages, mode of administration, and frequency of administration-were additionally extracted from unstructured medical reports by means of natural language processing [20].

Identification, assessment, and validation of adverse drug reactions
Patient records (including hospital discharge letters) of all patients in the study population were manually screened for ADRs by the joint first authors (JH, NN), and identified ADRs were discussed with the senior author (AG), taking into account details on medication.All ADRs were subsequently validated by an interdisciplinary expert panel comprising specialists in internal medicine (BK), neurology (MF), clinical pharmacology (DOS), and psychiatry (MSW, SDG, SS, SB, HF).

Drug interaction checks
Drug interaction checks were conducted with the electronic drug information system Arzneimittel-Informations-Dienste (AiD) Klinik ® (Dosing GmbH, Heidelberg, Germany).

Assessment of potentially inappropriate medications for elderly patients
Potentially inappropriate medications for elderly patients (PIMs) were assessed with the aid of the PRISCUS list (priscus (Latin), ancient, venerable) [21] and the FORTA (Fit fOR The Aged) classification system (version 2021) [22,23], both of which apply to people ≥ 65 years of age.
The PRISCUS list, which is specifically tailored to the German pharmaceutical market, tabulates a total of 83 PIMs and provides examples of suitable pharmacological alternatives for these PIMs [21].Drugs used in the study population were categorised as "PRISCUS-listed" (i.e.PIMs), "Not PRISCUS-listed" (i.e.non-PIMs), or "No assignment possible".The PRISCUS list classifies certain drugs (such as haloperidol, olanzapine, and zopiclone) as PIMs only if their cumulative daily dosages exceed specific thresholds (e.g. 2 mg, 10 mg, and 3.75 mg for haloperidol, olanzapine, and zopiclone, respectively).If the daily dosage of one of these drugs could not be retrieved from the patient records, the drug was categorised as "No assignment possible".
The FORTA classification system (version 2021) [22,23] was applied to assess the drugs used in the study population based on their main therapeutic indications.Drugs were categorised as A (i.e.indispensable drugs in the pharmacological treatment of elderly people), B (i.e.drugs with proven or obvious efficacy in elderly people), C (i.e.drugs with questionable efficacy-safety profiles in elderly people), or D (i.e.drugs that should be avoided in elderly people) [24].If a drug was not listed in the FORTA classification system, it was classified as "Not labelled".

Statistical analyses
Quantitative variables are shown as means ± standard deviations or as medians with interquartile ranges (IQRs).For categorical variables, absolute and relative frequencies are reported.To investigate associations of sex and comorbidities with ADR occurrence, odds ratios (ORs) with corresponding 95% confidence intervals (CIs) were calculated.
As the nature of our study was exploratory and hypothesis generating, no adjustments for multiple testing were carried out.Age distributions of patients with and without ADRs were tested for normality by utilisation of the Shapiro-Wilk test and by inspection of the histograms and Q-Q plots.Since the age distributions did not follow a normal distribution, median ages of patients with and without ADRs were compared with a two-sided Mann-Whitney U test.A P value < 0.05 was considered statistically significant.All statistical analyses were conducted with IBM ® SPSS ® Statistics for Windows, version 28 (Armonk, New York, USA).

Study population
Overall, 634 patient cases involving 481 individual patients were identified by ECRDW and were manually screened for ADRs.The higher number of patient cases as compared to the number of individual patients is explained by returners.As ADRs may have occurred during every hospital stay of a returning patient, each case was evaluated separately.Therefore, the following statistical analyses refer to n = 634 as denominator unless stated otherwise.The mean age of the study population was 76.6 ± 7.1 years (median age 76 years, IQR 71-81 years, range 65-99 years), and more than twothirds (67.2%; 426/634) of the patients were female.Overall, the study population was characterised by a high burden of both somatic and psychiatric diseases (Table 1).The most prevalent somatic and psychiatric diseases were arterial hypertension and depression, affecting 54.3% (344/634) and 39.1% (248/634) of the study population, respectively.
Remarkably, in 6.3% (40/634) of patient cases, ADRs were identified immediately upon hospital admission.Of note, in nearly half of patient cases with ADRs detected upon hospital admission (42.5%; 17/40), the detected ADRs were responsible for or contributed to patients' hospitalisations.The prevalence of ADRs that occurred in hospital was 4.9% (31/634).

Drug-drug interactions
To investigate a possible link between drug-drug interactions (DDIs) and ADR occurrence, the medication of patients affected by (an) ADR(s) were scrutinised for potential DDIs by utilisation of AiDKlinik ® .DDIs were likely involved in the occurrence of 29 of 92 ADRs (31.5%).In 20 of these 29 ADRs (69.0%),DDIs were elicited exclusively by combinations of psychotropic drugs (Fig. 1E).

Potentially inappropriate medications for elderly patients
PIM prescriptions have been recognised as a significant contributor to ADR development in elderly people [25][26][27].
To investigate the role of PIMs in ADR development in the present study, drugs (n = 150) with a suspected relationship to ADRs were assessed for PIM prescriptions with the aid of the PRISCUS list and the FORTA classification system.Thirteen of the 150 suspected agents (8.7%) were designated as PIMs according to the PRISCUS list: amitriptyline (n = 3), clozapine (n = 3), digoxin (n = 2), doxepin (n = 1), haloperidol (at a daily dose of more than 2 mg; n = 1), olanzapine (at a daily dose of more than 10 mg; n = 1), trimipramine (n = 1), and zopiclone (at a daily dose of more than 3.75 mg; n = 1).Sixty-seven (44.7%) and 21 (14.0%) of the 150 suspected agents were designated as category C drugs and category D drugs according to the FORTA classification system, respectively (Fig. 1D).The most frequent category C and category D drugs with a suspected relationship to ADRs were risperidone (n = 11) and aripiprazole (n = 6), respectively (Table 4).

Discussion
While ADRs have been studied extensively in the older general population [28], data about the frequency and characteristics of ADRs in elderly patients suffering from mental illness are limited.The present study aimed at diminishing this knowledge gap by providing a detailed analysis of ADRs that occurred over a 6-year period in a population of elderly psychiatric inpatients treated on the gerontopsychiatric ward of a university hospital in northern Germany.We detected an overall ADR prevalence of 8.8%.The ADR prevalence upon hospital admission and the ADR prevalence during hospitalisation were calculated as 6.3% and 4.9%, respectively.
Whereas the ADR prevalence upon hospital admission in the present study (6.3%) was largely in accordance with figures put forward by Bouvy and colleagues (median 3.6%, mean 4.6%) [29], the in-hospital ADR prevalence in our study (4.9%) was markedly lower than in systematic reviews and meta-analyses by Alhawassi et al. (11.5%) [12] and Jennings et al. (16%) [13].However, it must be considered that there was a substantial degree of heterogeneity between the studies included in these systematic reviews/metaanalyses, leading to wide 95% CIs (0-27.7% in the Alhawassi et al. systematic review [12]; 12-22% in the Jennings et al. meta-analysis [13]).In addition, the systematic reviews and meta-analyses by Alhawassi et al. [12] and Jennings et al. [13] included data from studies across different medical disciplines, yielding a much broader spectrum of ADRs as compared to our study, which only enrolled elderly psychiatric inpatients.
Another explanation for the comparatively lower inhospital ADR prevalence in our study might be that older psychiatric patients are more acquainted with the (side) effects of psychotropic drugs and that they do not communicate clinical symptoms as readily to their treating physicians as their younger counterparts [30].Moreover, Greil and colleagues suggested that physicians may avoid medications with a history of intolerance in older patients, while younger patients may receive certain drugs for the first time and may therefore develop ADRs more frequently compared to older patients [30].Besides, older psychiatric patients often display a variety of somatic complaints, making it more difficult to establish a link between a reported symptom and a specific medication [30].In summary, these factors may have led to the comparatively low in-hospital ADR prevalence in the present study.
Extrapyramidal symptoms, alterations in blood pressure or heart rate, and electrolyte disturbances represented the three most frequent types of ADRs in our study.Taken together, these three ADR groups accounted for approximately half (46.7%) of all ADRs registered in the study population.These ADRs represent well-known side effects of psychotropic medications.Antipsychotics, especially first-generation agents, are notorious for causing extrapyramidal symptoms [31][32][33].Moreover, antipsychotics-but also tricyclic antidepressants-are known to elicit alterations in blood pressure and/or heart rate [34], especially if co-prescribed with other drugs acting on the cardiovascular system, e.g.antihypertensive agents [35].Selective serotonin reuptake inhibitors and selective serotonin-norepinephrine reuptake inhibitors may evoke electrolyte disturbances such as hyponatraemia, especially when co-administered with thiazide diuretics [35].
Even though the majority (62.0%) of the drugs that were likely related to ADR occurrence in our study were classified as psychopharmaceuticals, a considerable share (i.e.38.0%) were non-psychotropic drugs.This demonstrates that medications for the treatment of somatic diseases must always be taken into consideration alongside psychotropic drugs when the risk of ADRs is evaluated in elderly psychiatric patients.
Drug-drug interactions were implicated in the development of 31.5% of the ADRs in our study population, most notably interactions between psychotropic drugs.This finding is in accordance with previous reports that suggested DDIs as a risk factor for ADR occurrence [12] and that described a high DDI potential of psychotropic drugs [35].
Astonishingly, we observed marked discrepancies between the PRISCUS list and the FORTA classification Fig. 1 Analysis of drugs with a suspected relationship to adverse drug reactions in the study population.A Differentiation of drugs (n = 150) with a suspected relationship to ADRs into psychotropic and nonpsychotropic substances.B Relatedness of ADRs (n = 92) to psychotropic and non-psychotropic drugs.C PRISCUS listing of drugs (n = 150) with a suspected relationship to ADRs ("PRISCUS-listed" refers to drugs that are considered potentially inappropriate medications for elderly patients according to the PRISCUS list).D FORTA labelling of drugs (n = 150) with a suspected relationship to ADRs (Label A: indispensable drugs in the pharmacological treatment of elderly people; Label B: drugs with proven or obvious efficacy in elderly people; Label C: drugs with questionable efficacy-safety profiles in elderly people; Label D: drugs that should be avoided in elderly people; "Not labelled": drug not mentioned in the FORTA classification system).E Characterisation of drug-drug interactions that were likely related to the occurrence of 29 ADRs.F Adjustment of drugs (n = 150) with a suspected relationship to ADRs after the ADR had occurred.ADR denotes adverse drug reaction, FORTA Fit fOR The Aged ◂ system when evaluating the 150 drugs that were likely related to ADR occurrence in terms of their general appropriateness for elderly patients.Whereas 8.7% of the drugs with a suspected relationship to ADR occurrence were designated as PIMs according to the PRISCUS list [21], 58.7% were classified as unsuitable for elderly people according to the FORTA classification system (i.e.FORTA categories C plus D) [22,23].This finding suggests that the FORTA classification system has a higher sensitivity in the detection of potentially inappropriate prescribing for elderly psychiatric inpatients compared to the PRISCUS list.This divergence may be explained by the fact that the PRISCUS list has remained unaltered since its inception in 2010, whereas the FORTA classification system has been updated at regular intervals, its latest version dating from 2021.Tools for medication reviews, especially tools for the assessment of potentially inappropriate prescribing, need to be kept up to date in order to allow for reliable therapeutic conclusions to be drawn.It must be annotated that the PRISCUS list is currently being revised.A release date of the PRISCUS list 2.0, however, has not yet been officially announced [36].
Only one in five drugs likely related to ADR occurrence was discontinued after the ADR had occurred, as opposed to nearly 30% of the drugs with a suspected relationship to ADRs which were left unchanged after the event.This demonstrates that in geriatric psychiatry therapeutic alternatives are often sparse or even lacking altogether.Consequently, ADRs of limited clinical severity must sometimes be tolerated by both patient and treating physician(s) in order to safeguard a sufficiently effective pharmacotherapy.To abstain from pharmacotherapy often is not a feasible option in geriatric psychiatry.Fortunately, in the present study, more than every third drug with a suspected relationship to ADR occurrence could be replaced with an adequate pharmacological alternative.
Female sex has repeatedly been reported as a risk factor for ADR occurrence [12], whereas in the present study, we did not observe such an association.By contrast, the presence of coronary heart disease was associated with an increased risk of ADR occurrence.This finding is in accordance with previous reports which described cardiovascular disease as a risk factor for ADRs [12].However, other cardiovascular diseases such as arterial hypertension, chronic heart failure, and atrial fibrillation were not associated with an increased risk of ADR occurrence in the present study.In contrast to previous accounts, which cited dementia as a risk factor for ADR development [12], the presence of dementia was associated with a reduced risk of ADR occurrence in our study.We hypothesise that the reduced ability of patients suffering from dementia to verbalise and to communicate physical complaints to their treating physicians may have led to an underreporting of ADRs as compared to non-demented patients.However, it must be underscored that no adjustments for multiple testing were conducted for the ORs in our study (see also the paragraph "Statistical analyses" in the "Methods" section); therefore, these findings must be interpreted with circumspection and should be validated in a future prospective study.
Of particular clinical importance, the majority of patient cases with ADRs were not identified during inpatient treatment but were already present upon admission to the hospital (6.3%; 40/634).Of these, nearly half (42.5%; 17/40) were responsible for patients' hospital admissions.We speculate that the high ADR detection rate immediately upon hospital admission may be explained by the high priority of pharmacotherapy safety and pharmacovigilance at our institution.Hannover Medical School was the first German university hospital to establish a professorship for drug safety in 2011 [37].The pharmacological expertise at Hannover Medical School is integrated in and coordinated by the Centre for Drug Safety (Zentrum für Arzneimittelsicherheit, ZAS), an interdisciplinary and multiprofessional platform which comprises specialists from clinical pharmacology, microbiology, hospital epidemiology, transfusion medicine, pharmacy, clinical chemistry, psychiatry, and general practice, as well as the office of the Drug Commissioner [38].Through ZAS, regular training programmes such as lectures, seminars, and clinical rotations/internships as well as continuing medical education activities are offered to both physicians and medical students.Furthermore, the trinational (Germany, Austria, and Switzerland) pharmacovigilance programme Drug Safety in Psychiatry (Arzneimittelsicherheit in der Psychiatrie, AMSP), which aims at improving drug safety in psychiatric inpatients, has its headquarters at and coordinates its activities from Hannover Medical School [39].Besides other activities, AMSP organises regular case conferences on a regional and supraregional level, where (suspected) ADR cases are being discussed by experts in the field.
Electroconvulsive therapy (ECT) is generally considered an effective and well-tolerated therapeutic modality in the treatment of older patients with major depression and other psychiatric disorders [34,40,41].However, we registered three ADRs (two asystoles and one case of obstructive airway symptoms) related to general anaesthesia in the context of ECT.This risk signal warrants further investigation.We suggest a diligent benefit-risk evaluation and a thorough screening for cardiopulmonary comorbidities (e.g. by electrocardiography and chest X-ray) in elderly psychiatric patients prior to the initiation of ECT.
Limitations of the present study mainly arise from its monocentric and retrospective design.However, due to the long observation period (i.e. 6 years) and the size of the study population (n = 634), we think that our study results are overall representative of elderly psychiatric patients treated in German university hospitals.The ADRs analysed in this study were retrieved from patient records, in particular from hospital discharge letters, an approach that has been utilised in similar form in previous studies [42,43].It may be assumed that only ADRs with a certain severity were deemed as sufficiently clinically relevant by the treating physician(s) to be mentioned and described in further detail in the hospital discharge letter and hence to be communicated to the patient's general practitioner or ambulatory psychiatrist/neurologist.A certain degree of underreporting of ADRs, especially ADRs of mild severity, must therefore be assumed in this study.Notwithstanding, the ADR documentation via hospital discharge letters adequately reflects treating physicians' approach to, assessment and documentation of ADRs during clinical routine.
The primary objective of our study, i.e. the estimation of the frequency and characterisation of ADRs that occurred on the gerontopsychiatric ward of our institution over the course of 6 years, was achieved.However, our study was non-interventional.The effectiveness of interventions to optimise pharmacotherapy of elderly people in order to reduce ADRs was shown by Gray and colleagues in a meta-analysis of 13 randomised controlled trials.Participants in the intervention group were 21% less likely to experience an ADR compared to those in the control group (OR = 0.79; 95% CI = 0.62-0.99).The benefit was even more pronounced for the reduction of serious ADRs (OR = 0.64, 95% CI = 0.42-0.98)[44].A similar intervention-i.e. an interdisciplinary ward round with medication review conducted by a team of specialists in psychiatry, internal medicine, geriatrics, neurology, and clinical pharmacology-has already been tested at our institution in a pilot study [45].This pilot study, however, did not include a control group.Hence, in order to investigate if the in-hospital ADR rate can be reduced by this intervention, a randomised controlled study is currently being devised at our institution.

Table 1
Characteristics of the study population (n = 634) a Patients could have more than one diagnosis b ICD-10 F32, F33 c ICD-10 F31

Table 2
Absolute and relative frequencies of adverse drug reactions (n = 92) detected in the study population

Table 3
Absolute and relative frequencies of drugs (n = 150) with a suspected relationship to adverse drug reactions.For a listing of the individual substances please refer to Supplementary Table1ADR denotes adverse drug reaction a Percentages may not total 100 due to rounding

Table 4
Absolute and relative frequencies of FORTA category C drugs (i.e.drugs with questionable efficacy-safety profiles in elderly people) and FORTA category D drugs (i.e.drugs that should be avoided in elderly people) with a suspected relationship to adverse drug reactions in the study population ADR denotes adverse drug reaction, FORTA Fit fOR The Aged a Percentages may not total 100 due to rounding