The present study analyzed the risk of ADRs within the inpatient psychiatric setting according to age (i.e., < 65 and ≥ 65 years of age). While the overall risk for ADRs did not differ between the two age groups, older patients were at higher risk for certain ADRs such as delirium, ataxia, certain types of EPS (e.g., parkinsonism, Pisa-/metronome syndrome), cardiovascular symptoms, and falls. Other ADRs such as suicidality, acute dystonia, akathisia, liver dysfunction, weight gain, sexual dysfunction, and hyperprolactinemia/galactorrhea were more common in younger patients. Older patients treated with ADDs—especially SSRIs and SNRIs—, low potency FGAs, and lithium had a higher risk of ADRs than younger patients, while younger patients treated with SGAs had a higher risk of ADRs than older patients treated with these drugs. Further, we found that ADRs in older patients were more likely to involve multiple drugs.

Older age is a well-described risk factor for ADRs [23]. While risk factors, such as pre-existing organ damage, are more common in patients ≥ 65 years of age, as found in this study, we were unable to detect a higher rate of ADRs in older adults. However, in this study, ADRs in older patients were more likely to be classified as life-threatening and required specialized care (Table 5). In addition, we found that older patients who experienced ADRs had an 6.4-fold increased risk of a fatal outcome. This significantly higher mortality greatly exceeds the findings of Dubrall et al., who examined ADRs reported to the German Federal Institute for Drugs and Medical Devices and found that ADR-related mortality was 3 times higher among patients ≥ 65 years of age than in those < 65 years of age [24]. The inpatient setting of our study, suggesting patients are more severely ill, may be one reason for this.

Moreover, we found that ADRs were more likely to affect patients with certain diagnoses. Younger patients with schizophrenia or acute mania had a significantly higher risk for ADRs than older patients with these diagnoses (Table 2). An explanation for this may be in the way these patients are treated: Kleimann et al. previously described remarkably high rates of polypsychopharmacy—defined as the intake of ≥ 4 psychotropic drugs—in patients with acute mania, which declined with higher age [25]. Further, Zolk et al. found that older schizophrenic patients were generally treated with lower doses of APDs [26]. Both aspects (i.e., lower doses, less polypsychopharmacy) may in turn reduce the risk of ADRs in these two diagnostic subgroups.

We found lower median doses in older patients for all drugs with the exception of venlafaxine (no difference between age groups or between patients with and without ADRs), sertraline (lowest dose in older patients with ADRs), and pipamperone (highest median dose in older patients with ADRs; Table 4). Compared to other ADDs, venlafaxine has well-characterized dose-dependent efficacy. Increasing venlafaxine dose to 150 mg yields benefits [27], which appears to be the target dose in both age groups. However, this seems to come at the expense of tolerability, as ADRs are apparently to be expected at this dose. Pipamerone, on the other hand, is voluptously used in geriatric patients [28] presumingly under the assumption that its use for these patients—even at high doses—is safe. Our data suggest, that this is not the case and higher pipamerone doses increase the risk for subsequent ADRs.

In general, we found a higher ADR risk in older patients treated with low potency ADRs than in younger patients (Fig. 3A, Table 3), especially prothipendyl, melperone, chlorprothixene, levomepromazine, and promethazine (Fig. 3B, Table 3). While promethazine’s, chlorprothixene’s, and levomepromazine’s anticholinergic effects (see below) surely contribute to this, melperone and prothipendyl lack this specific effect. Prothipendyl, a frequently used drug in Austrian nursing homes despite its classification as a potentially inappropriate drug, is not available in most European countries to due a higher risk of EPS [4, 29, 30]. A previous AMSP study additionally found a higher risk of cardiovascular ADRs under treatment with prothipendyl, even as a single imputation ADR [31]. Most ADRs imputed a low-potency FGA alongside other drugs, suggesting that pharmacokinetic and additive pharmacodynamic effects are the leading cause for ADRs in this drug group, while the capacity of low-potency FGAs to induce ADRs on their own was relatively low (suppl. Table 6). However, bearing in mind that this study presents relative risks, the absolute risk of ADRs unter treatment with most low-potency FGAs is comparable to or even lower than the ADR risk of other drugs with sedating properties, such as mirtazapine, olanzapine, and trazodone (Fig. 3B, suppl. Table 5).

In older adults, ADRs often present as nonspecific geriatric syndromes such as falls, delirium [32], decreased mobility, cognitive decline [33], and incontinence [12] possibly making them more difficult to accurately identify as drug-induced phenomena [32, 33]. Additionally, cognitive impairment may reduce the patient’s ability to adequately express any drug-related discomfort [34], perhaps explaining why dementia has even been found to decrease the risk of ADRs [13]. This is likely to have significantly contributed to the under-reporting of ADRs in cognitively impaired patients in the present study and emphasizes the importance of careful clinical monitoring, the collection of baseline parameters, and the explicit assessment of drug-related symptoms. Moreover, comorbidities may mask ADRs or be sufficient in themselves to explain a particular symptom. Additionally, certain symptoms, such as severe edema, may appear more alarming, when they occur in younger patients because they are unusual for this age group and are more likely to lack an alternate explanation other than drug-induced. Finally, older patients with conditions such as schizophrenia have most likely been treated with psychotropic drugs for an extended period of time, reducing the likelihood of ADRs that generally emerge early in treatment, whereas long-standing ADRs become more challenging to recognize.

In the present study, we identified age-related differences in the risk of various ADRs. A selection of findings will be discussed in detail below.

Delirium and central anticholinergic effects

Unsurprisingly, we found that older patients had a higher risk of drug-induced delirium (Fig. 1B, Table 3), consistent with the observations reported by Greil et al. [14]. Up to 39% of deliriums in older hospitalized patients are associated with drug use. One of the most concerning drug properties in this regard is a high affinity for antimuscarinic acetylcholine receptors [35]. Central anticholinergic properties may be particularly harmful in older adults, which is why drugs such as amitriptyline, biperiden, and olanzapine are not generally recommended for older patients [4, 5], especially when multiple drugs with anticholinergic properties are combined [34]. Using AMSP data, Friedrich et al. previously found that APDs and ADDs with potent anticholinergic properties, such as TCAs and several SGAs (e.g., clozapine, olanzapine), have a higher propensity to cause drug-induced delirium in psychiatric inpatients. Clozapine and amitriptyline were the psychotropic drugs most frequently associated with drug-induced delirium, and most cases of drug-induced delirium were caused by multiple drugs [18]. In the present study, older patients treated with several drugs with strong anticholinergic properties, such as amitriptyline, trimipramine, levomepromazine, promethazine, chlorprothixene, and biperiden, did indeed have a significantly higher risk of experiencing ADRs than younger patients (Fig. 2B, suppl. Table 5). In both age groups, drug-induced delirium in the present study was often the effect of multiple drugs, indicating that this ADR typically results from pharmacodynamic drug-drug interactions. However, older patients were also at higher risk of experiencing delirium imputing a single drug (Fig. 2A, suppl. Table 6 and 7).

Suicidality and serotonergic ADRs

Early warnings by the US Food and Drug Administration in 2004 highlighted the risk of suicidality associated with the use of SSRIs in children, adolescents, and young adults under 25, primarily during the initial stages of treatment [36]. While this rare effect has also been noted in patients above the age of 25 [37, 38], the risk significantly declines with age [37]. This is consistent with our findings indicating only a single instance of this ADR in older patients (Table 3). Although the exact mechanisms of drug-induced suicidality remain elusive, serotonergic activation induced by SSRIs and SNRIs is proposed to significantly contribute [39]. Notably, our study found older patients had an overall higher risk for serotonergic ADRs (including serotonin syndrome; Fig. 1B, Table 3), especially as a multiple imputation ADR (Fig. 2B, suppl. Table 4). Serotonin syndrome, though rare, preferentially affects high-risk patients, i.e., critically ill patients and those with polypharmacy, and often goes unrecognized [40].

Extrapyramidal symptoms

Affecting about 1 in 5 patients [41], EPS are one of the most significant ADRs of treatment with APDs and a major concern in older patients [4, 13]. When systematically analyzed, drug-induced movement disorders occur more frequently in older adults [42]. However, distinguishing new-onset drug-induced EPS from pre-existing movement disorders may pose a challenge [42], increasing the risk that they remain unnoticed [43]. This may in part explain why the overall incidence of EPS in this study did not differ between older and younger patients (Fig. 1A, Table 3). The study by Greil et al. using Swiss AMSP data indicated that the risk for EPS decreases with age, though with limited statistical significance (p < 0.05) [14]. However, in examining a much larger patient collective, we found that individual types of EPS showed age-related effects, which was not previously considered by Greil et al. [14]. The risk for some types of movement disorders, such as acute dystonia and akathisia, was higher in younger patients in the present study (Fig. 1B, Table 3). Indeed, apart from male sex, younger age is a well-known risk factor for acute dystonia [44]. The likelihood of developing akathisia and acute dystonia increases with the use of high doses and rapid titration strategies and is highest when antipsychotic treatment is first initiated, all of which may affect younger patients more often [44, 45].

On the other hand, this study found that the risk for other types of EPS was higher in older patients (i.e., parkinsonism, Pisa/metronome syndrome; Fig. 1B, Table 3) and EPS among older patients were more likely to result in death (4 out of 5 fatal EPS cases; Table 6). Parkinsonism occurs in up to 50% of older patients treated with APDs and up to 67% of those with dementia [46] and often occurs even when APDs are used at lower than usual doses [42]. An earlier analysis of severe parkinsonism using AMSP data found that pre-existing organic brain damage (such as dementia) is a relevant risk factor for APD-induced parkinsonism and that high-potency FGAs expectably have an expectably higher risk than SGAs or low-potency FGAs [47]. Further, both older age and organic brain damage are known risk factors for Pisa syndrome and the related condition, metronome syndrome, which can emerge either acutely or after prolonged exposure to APDs [48, 49].

Seizures

In the present study, younger patients had a significantly higher risk for drug-induced seizures than older patients (Fig. 1B, Table 3). Druschky et al., who analyzed the occurrence of APD-induced seizures within the AMSP database over a slightly shorter time period (i.e., 1993 to 2015) found that young men with schizophrenia were most at risk for this ADR. The by far highest risk of seizures was found for clozapine, with a comparatively low rate for risperidone [50], an APD with high use among older patients [26]. ADD-associated seizures are rare, but seem to particulary be associated with the use of TCAs and tend to affect younger men and patients suffering from schizophrenia [51]. While certain risk factors for drug-induced seizures, such as higher doses [52], are presumably more prevalent in younger patients, other significant risk factors, such as somatic comorbidities [53], pre-existing brain damage, and EEG abnormalities [52], are more common in older adults. But again, the co-occurrence of these risk factors may make it more difficult to definitively attribute a seizure to drug use, resulting in an only “possible” probability rating for the involvement of a psychotropic drug among older patients.

Liver dysfunction

Our results suggest that the risk for elevated transaminases is higher in younger patients (Fig. 1). Liver injury associated with APDs is most often associated with olanzapine, followed by perazine and clozapine [16], therefore providing one explanation for the overall higher risk of ADRs under olanzapine and clozapine in younger patients (Fig. 2A, suppl. Table 5). Drug-induced liver injury caused by ADDs most commonly implicate mianserine and agomelatine [54]. Greil et al. also suggested a higher risk of psychotropic-drug induced liver dysfunction in younger patients, though the effect was statistically weak (p < 0.05) [14]. Whether age is a susceptibility factor for drug-induced liver injury appears to be drug-specific. However, persistent liver injury is appears to be more common with higher age [55].

Sexual dysfunction, galactorrhea/hyperprolactinemia, and weight gain

The present study found only one case of an ADR presenting with genital dysfunction in older patients, making it one of the main ADRs that are significantly more common in younger patients (Fig. 1A, Table 3). In general, the prevalence of sexual dysfunction increases with age [56] and in the presence of comorbidities such as hypertension, diabetes, and benign prostatic hyperplasia [57], making it more difficult to identify drug-induced effects. Moreover, this often shame-filled ADR is significantly under-reported [58], especially among older patients [59].

Similarly, symptomatic prolactin elevation and events of galactorrhea were significantly less common in older patients (Fig. 1B, Table 3), as was also found in the earlier publication of Greil et al. [14]. Amenorrhea, a possible symptom of hyperprolactinemia, is expected to occur only in premenopausal women, but breast tissue growth, galactorrhea, or sexual dysfunction may still affect older adults [60]. The clinical implications of elevated prolactin in older adults may also be less apparent. For example, (chronic) hyperprolactinemia is associated with osteoporosis [61] and certain types of breast cancer [62], both of which are complex conditions difficult to causally attribute to drug use. Previous studies suggest the prevalence of hyperprolactinemia is indeed higher in premenopausal (53–65.6%) than postmenopausal women (32–45.1%) [63, 64]. However, the risk detected in the present study is significantly lower, as only severe case with acute symptoms are included. Apart from amisulpride, risperidone has a particularly high propensity to cause hyperprolactinema [17]. Risperidone is also one of the most commonly used APDs in older patients [26], suggesting this ADR should be more common. However, the dose-dependency of hyperprolactinemia [65] may mitigante this effect, as older patients—even those with ADRs—were treated with lower median doses of risperidone (Table 4).

Weight gain is often a primary concern in patients treated with psychotropic drugs Consistent with the findings of Greil et al. [14] as well as other authors [66, 67], our study found that the risk for psychotropic drug-induced weight gain was significantly higher in younger patients (Fig. 1B, Table 3). Using AMSP data, Schneider et al. previously reported that olanzapine, quetiapine, risperidone, mirtazapine, and valproate were among the drugs most often associated with psychotropic drug-induced weight gain [15]. The higher propensity of these four drugs to cause this ADR may explain their higher ADR risk in younger patients in this study. Additionally, younger patients in the present study were treated with higher median doses of quetiapine, valproate, and risperidone, contributing to the risk of weight gain, which appears to have dose-dependent effects [68].

Cardiovascular adverse reactions

Cardiovascular ADRs are a major concern in older patients. The relevance of this ADR type is underlined by the 3 fatal cases of cardiovascular ADRs among older patients detected in the present study (Table 6). Because of their affinity for α1-adrengic receptors, APDs can cause hypotension [69]. In fact, a recent meta-analysis found that APDs, along with α-blockers and sodium–glucose-cotransporter (SGLT)-2 inhibitors, were the most common drug classes associated with orthostatic hypotension [70]. The risk of hypotension further increases when psychotropic drugs are used in combination with antihypertensive drugs, such as diuretics or β-blockers, or other psychotropic drugs [31, 71, 72]. Our study found that cardiovascular ADRs in older adults often imputed multiple drugs, whereas cardiovascular ADRs in younger patients often imputed a single drug (Fig. 2A). This is most likely due to a lower utilization of antihypertensive drugs in younger patients.

Hyponatremia

Among the ADRs examined in this study, hyponatremia was one of the ADRs with the highest risk (i.e., 3.7-fold) for older compared to younger patients (Fig. 1B, Table 3), also explaining the higher risk of ADRs in older patients treated with SSRIs, and even more so with SNRIs (Fig. 2A, suppl. Table 2). Using AMSP data, Seifert et al. previously described that older patients, particularly women ≥ 65 years of age treated with SNRIs and other potentially hyponatremia-inducing drugs, were the most vulnerable patient group for this ADR [19]. Among psychotropic drugs, SSRIs and SNRIs are best known for their propensity to cause this ADR, especially at the beginning of treatment and, therefore, even at lower doses [19]. This potentially explains the lowest sertraline dose in older patients with ADRs (Table 4). The risk for hyponatremia increases when SSRI and SNRI are used in combination with angiotensin-converting enzyme (ACE) inhibitors, thiazide and thiazide-like diuretics, and proton pump inhibitors [