Background

Psychiatric disorders are among the most leading causes of burden globally and the number of patients with mental illnesses has been continuously increasing [1]. According to complex neurobiological aetiologies of several mental disorders, pharmacological treatments are commonly used and novel treatments, targeting on multidimensional of the pathophysiology, have been developed [2]. However, some patients who are unresponsive to standard treatment tend to be prescribed for multiple classes of psychotropic medication to reduce their symptoms [3]. Thus, the higher risks of psychotropic drug-drug interaction (DDI) are inevitable.

Apart from various psychiatric symptoms requiring different psychopharmacological actions, longer life expectancy, followed by further medical comorbidities, also warrants polypharmacy and increases risks of psychotropic DDIs [4]. Numerous previous studies have shown high prevalence of DDIs in various settings, either clinical or non-clinical, including nursing home (81.1%), community pharmacies (26.0–49.6%), inpatient and outpatient department of the hospital (22.3%) [5–7]. Despite the observed association between DDIs and clinical outcomes such as hospitalisation or hospital visits, only a few patients clinically manifested complications from DDIs [8, 9]. Nowadays, healthcare technology systems have been implemented to detect DDIs, but too frequent, false and unnecessary alarms can have negative effects on both providers’ exhaustion and patients’ safety [10]. Therefore, the multidisciplinary team should consider an appropriate instrument to identify clinically significant DDIs under the balance between alarm fatigue and patient safety.

Electronic databases used for DDI identification are widely used and varied in their sensitivity and comprehensiveness and those with sufficient information include Drugs.com®, Lexicomp®, and Epocrates® [11]. All mentioned databases have been used in the clinical practices and several previous studies have been carried out to investigate their DDIs identification utility in different countries, settings, and clinical populations [12–14]. However, identification utility of each database is diverse in terms of approved psychotropic medications or prescription patterns in individual countries [15]. Therefore, the drug list in each database, the number, and clinical significance of DDIs detected should be considered when selecting the appropriate database that balances the utilisation of efficient healthcare technology systems, as mentioned earlier. Regarding all aforementioned differences, studies examining databases for psychotropic DDI identification are indispensable.

Up to our knowledge, there have been no studies evaluating psychotropic DDIs from these three commonly used electronic databases (Drugs.com®, Lexicomp®, and Epocrates®) in the context of Thai hospital university setting. In addition to being frequently used in many healthcare systems, these three databases are comprehensive and suggest the management in case DDIs are detected. Some studies from Thailand also investigated DDI identification utility from two databases (Micromedex® and Drugs.com®) or three databases (Micromedex®, Drugs.com®, and Liverpool HIV interactions), but they did not identify psychotropic DDIs, and were not done in the real clinical situations [16, 17].

Objectives

Our study primarily aimed to explore the prevalence and associated factors of psychotropic DDIs during a one-year observation in the psychiatric inpatient department of one Thai medical school. Secondly, we aimed to investigate psychotropic DDI identification utility from three databases e.g., Drugs.com®, Lexicomp®, and Epocrates®, based on the context of psychiatric hospitalisation in the quaternary hospital.

Methods

This study was a retrospective cohort design. We extracted data from the electronic health records of all patients hospitalised during 2020 in the psychiatric inpatient unit at one university hospital in Bangkok, Thailand. The records included all pharmacological prescriptions which were recorded and followed from the first day of admission till the date of discharge or the end of our study. The psychotropic DDI was identified by three different databases. We analysed the association between psychotropic DDI and clinical outcomes, including recorded adverse drug events, length of hospitalisation, and readmission. The informed consent was not obtained because we collected data from service records and report anonymously. All patients with age over than 18 years old were included, thus neither sample size calculation nor sampling was done in this study. This study was granted the ethical approval from the Institutional Review Board of Faculty of Medicine, Chulalongkorn University.

Results

Throughout the study, there were 149 psychiatric admissions from 128 patients. Most patients were female (n = 75, 58.6%), diagnosed with schizophrenia spectrum disorders (n = 42, 32.8%), and averagely at 39.4 ± 18.8 years of age (min–max = 19–86 years old). All 1,357 prescriptions during the study period were assessed. In total, 176 individual agents of either psychotropic or other medications were prescribed. Three most frequently prescribed psychotropic agents were clonazepam (n = 109, 8.0%), quetiapine (n = 82, 6.0%), and haloperidol (n = 64, 4.7%). Seven patients (5.5%) reported history of adverse drug reactions and sixteen patients (12.5%) were readmitted within a year. The demographic, clinical, and prescription data of all patients are shown in Table 1.

Only 1 out of 149 (99.3%) admissions had no psychotropic DDI during hospitalisation. In total, 956 DDI patterns with 2,825 pairs of psychotropic DDIs were identified based on all databases. According to Drugs.com®, Lexicomp®, and Epocrates®, DDIs were identified 2,500 times, 2,269 times, and 2,265 times, respectively. There were 26 and 17 serious psychotropic DDIs identified by Lexicomp® (X category) and Epocrates® (Contraindicated category); meanwhile, 254 DDIs were classified as Major category from Drugs.com®. Three most frequent DDIs were events of co-administrated clonazepam with quetiapine (n = 56), risperidone (n = 36), and valproic acid and derivatives (n = 36). Table 2 displays the top 20 frequent agents causing psychotropic DDI and pairs of psychotropic DDI, as well as their severity regarding different databases. The serious DDIs found from our study are shown in Table 3.

The Spearman’s correlation between numbers of identified psychotropic DDI and associated variables is demonstrated in Table 4. There was a very strong positive correlation between numbers of total medications prescribed during admission and numbers of psychotropic DDI (Spearman’s rho = 0.827, P-value < 0.001). Meanwhile, a moderate positive correlation between numbers of DDI and readmission within 1 year (Spearman’s rho = 0.391, P-value < 0.001), length of hospitalisation (Spearman’s rho = 0.374, P-value < 0.001), and having underlying diseases (Spearman’s rho = 0.331, P-value < 0.001) was found. Our results also suggested a weak positive correlation between age and numbers of psychotropic DDI (Spearman’s rho = 0.286, P-value = 0.002). According to prescription profiles, both antihypertensive and antiepileptic agents were strongly correlated with DDI (Spearman’s rho = 0.468, P-value < 0.001 and Spearman’s rho = 0.416, P-value < 0.001). We also found a weak correlation between DDIs and either benzodiazepines (Spearman’s rho = 0.220, P-value = 0.016) or other supportive and psychiatric agents (Spearman’s rho = 0.277, P-value = 0.002 and Spearman’s rho = 0.272, P-value = 0.003).

Table 5 demonstrates the kappa index of agreement strength of psychotropic DDI severity classification among three databases. According to Fleiss’ kappa index, the overall kappa was 0.162 (95% CI 0.14–0.19, P-value < 0.001). Meanwhile kappa index of Drugs.com® – Lexicomp® was 0.04 (P-value = 0.07), Drugs.com® – Epocrates® was 0.21 (P-value < 0.001), and Lexicomp® – Epocrates® was 0.24 (P-value < 0.001).

Discussion

The psychotropic DDI occurred in nearly all psychiatric admissions. Interestingly, the only admission without psychotropic DDI was a patient admitted for drug dependence rehabilitation and prescribed for vitamin supplements and supportive therapies. Our findings confirmed the high prevalence of psychotropic DDIs detected by electronic databases in the psychiatric inpatient settings [21, 22].

However, the median of number of medications prescribed during hospitalisation was 9 and this limited its generalisability with patients in the psychiatric outpatient department, where the number of medications prescribed was comparatively lower [23]. Due to the nature of quaternary hospital services, patients in our institute often had more complicated medical comorbidities and then were at higher risk for psychotropic DDIs.

Among all patterns of psychotropic DDI pairs, the top-three psychotropic agents causing DDIs (i.e., quetiapine, clonazepam, and trazodone) were used to remedy sleep disturbances. Apart from its indication for both psychotic and affective disorders, quetiapine has been widely prescribed for sleep problems, even it is an off-label use, because of its lower addictive qualities compared to benzodiazepines [24]. Similarly, either clonazepam or trazodone are known for their sleep disorders management, especially for trazodone, in older patients or patients with depression [25, 26]. Yet, several adverse reactions from these medications alone or their DDIs could be found. There were only 6 out of 149 admissions that clinical adverse reactions were recorded. Most of them were explained by an idiosyncratic agent except one patient with schizophrenia and prescribed olanzapine and clonazepam. Additionally administrated trazodone, the patient had developed retrograde amnesia afterwards, which could be caused by DDI of trazodone and such medications. During psychiatric hospitalisation, implementation of insomnia psychosocial interventions will therefore be advantageous in lowering both sedative agents prescription and incidence of psychotropic DDIs [27].

Correlation between higher numbers of psychotropic DDIs and several associated factors was found. It was not surprise that patients with greater numbers of medications prescribed during hospitalisation would be strongly correlated with more events of psychotropic DDIs. Weak correlation with ageing and presence of medical comorbidities was also observed as these groups were at risk of polypharmacy [28]. Because numbers of psychotropic DDI events were cumulatively recorded, correlation with DDIs was found in patients with longer hospitalisation or readmission.

We found a moderate correlation between DDIs and two medication classes: antiepileptic and antihypertensive agents. Antiepileptic agents are known for their frequent DDIs with other psychotropic medications under various mechanisms, especially through pharmacodynamic interaction [29]. Many psychotropic medications are related with orthostatic hypotension, so psychotropic DDIs will be identified when co-prescribing with antihypertensive agents [30]. Also, additive effects on sedation could be caused by co-administration of benzodiazepines and hypnotics [31]. To prevent adverse psychotropic DDIs, clinicians should be considerate prior to prescribing such groups of medications.

Our findings found the discrepancy between numbers of identified psychotropic DDIs and reports of clinical adverse reactions, which was consistent with previous studies [32]. The severity of the identified DDIs may not always be high enough to cause a clinically significant adverse reaction. Additionally, the identified DDIs might be managed appropriately by healthcare providers such as adjusting dosages. It also emphasised the importance of DDI detection alert system that needed both clinical and informational integration [33].

According to the number of detected psychotropic DDIs, Drugs.com® was a comparatively more sensitive instrument. Compared to Epocrates® and Lexicomp®, more than 130 times of psychotropic DDIs were detected from Drugs.com® but none of clinically significant events was shown. The agreement strength among three databases’ severity classification was slight and Lexicomp® – Epocrates® had the strongest agreement. Even though both Lexicomp® and Epocrates® comprise 4 severity classifications, Lexicomp® database provide most extensive list of medications for our study as it included some medications approved in Thailand but disapproved in other countries, such as domperidone or flunarizine. Information regarding drug interaction with herbal medicine was also given, and we considered, from three examined databases, that Lexicomp® held the most suitability for its psychotropic DDI identification utility in the context of our country.

Interestingly, identified DDIs classified as major severity were found in DDIs caused by both non-psychotropic drugs (gemfibrozil-simvastatin and omeprazole-rilpivirine) or by psychotropic and non-psychotropic drug interaction (flunarizine and orphenadrine). None of patients receiving mentioned pairs of DDIs reported any clinical adverse reaction. Meanwhile, psychotropic DDIs from dual psychotropic agents were much more common and classified as less severe categories. In psychiatric practices, several identified pairs of psychotropic DDIs were broadly prescribed, such as antipsychotic agents and trihexyphenidyl, as observed in our study [23]. Despite the fact that they were less clinically significant, co-administration of such medications should be carefully considered by clinicians in terms of benefits and risks of negative consequences. Furthermore, considerate clinical judgment should be highly concerned when co-prescribing psychotropic agents with other non-psychotropic medications, particularly flunarizine and orphenadrine, as found in our study.

Our study investigated psychotropic DDIs occurring within one-year long in the clinical setting, so it reflected the real situation of psychotropic DDIs in the psychiatric inpatient department of the quaternary hospital where several non-psychotropic agents were co-administrated. All examined databases are commonly used and comparative studies of them are still lacking. Some limitations should be addressed. We retrospectively extracted data from the psychiatric record which some information or associated factors might be missed or not enrolled in case of unrecorded. Some prescriptions with severe DDI might be warned by the hospital and such prescriptions were prohibited. Psychiatric diagnoses were mainly schizophrenia spectrum and major depressive disorders, so psychotropic medications out of these two indications were less prescribed and analysed in our study. The rate of clinical adverse reactions was low and the association with DDIs could not be well identified. As mentioned earlier, our study was conducted in the quaternary hospital, so the results might not be generalised to other types of healthcare settings or to populations with different characteristics. However, implications of the results still applied to non-mental hospitals where patients were supposed to be prescribed both medication classes, reported in our study.

Further studies should be conducted to develop updated databases for psychotropic DDI identification in specific populations. The comparison of DDIs among databases in terms of drug mechanisms would enable professionals to select the most suitable database and be helpful for clinicians’ practical implications. Apart from theoretical interactions, clinical significance should be focused and appropriately alerted to prevent alarm fatigue. However, it is challenging to integrate both clinical and informational databases as psychotropic DDIs are tremendously common, and their serious interactions are also difficult to be predicted.

Conclusion

Almost all patients hospitalised in the psychiatric inpatient unit had at least one psychotropic DDI, but clinical adverse consequences were rarely reported. Among three databases, interactions detected by Drugs.com® were greatest in number, whereas Lexicomp® provided the broadest list of medications. Development of such databases based on both theoretical and clinical conceptions should be focused to balance safety of patients and weariness of healthcare providers.