Patient concerns regarding antidepressant drug–drug interactions: a retrospective analysis using data from a medicines call centre

Keywords: Antidepressive agents, call centre, drug information services, drug interactions, help-seeking behaviour, information-seeking behaviour, prescribing, primary care.

Antidepressant use has increased over the last two decades, with Australia and New Zealand among the top 10 users of antidepressants in the Organisation for Economic Co-operation and Development (OECD).¹ Antidepressants represent >5% of prescriptions written in Australian general practice,² and >29 million antidepressant prescriptions were dispensed on the Australian publicly subsidised Pharmaceutical Benefits Scheme in 2019–20.³ In addition, 12.6% of all New Zealanders were prescribed an antidepressant (16% of females, 9% of males) in 2015, an increase of 21% since 2008.⁴

Comorbidity and polypharmacy are common in antidepressant users.⁵ Previous studies have shown that older people, children and adolescents are more likely to experience drug–drug interactions with antidepressants.^6,7 However, few studies have determined drug–drug interaction prevalence across life stages. Antidepressant drug–drug interactions increase the likelihood of adverse drug events, which can result in increased morbidity and mortality, loss of drug efficacy, and misdiagnosis of symptoms.^8,9 Moreover, nearly half (46.5%) of drug–drug interaction-related adverse events are preventable.¹⁰ We aimed to identify, profile, and analyse drugs commonly used concomitantly with antidepressants and potential drug–drug interactions in antidepressant users who sought medicines information from a national medicines call centre.

Between September 2002 and June 2010, we conducted a retrospective observational study of consumer questions concerning antidepressant interactions, using data from National Prescribing Service (NPS) MedicineWise (formerly the NPS Medicines Line), operated by pharmacists at Mater Health Services, Brisbane, Queensland, Australia. Although there was a change in service provider after this time, the consistency of caller demographics and types of enquiries generating consumer concerns over the 8 years of service supports the current relevance of our study aims.¹¹ Few new antidepressants have emerged in the last decade, so the drugs in this study continue to have prominent use. Eight specific antidepressants comprised 84% of all antidepressants used in Australia between 2006 and 2018.¹²

For the first part of our analysis, we extracted all calls about antidepressant medicines from the database, based on their Anatomical Therapeutic Chemical (ATC) Level 3 (therapeutic class).¹³ Call characteristics were compared with their respective ‘rest of calls’ (Fig. 1). We explored antidepressant drug–drug interaction queries as this was a more highly ranked enquiry type, (22.9%) for antidepressants than the rest of the calls (13.8%, P < 0.001). A potential drug–drug interaction was defined as any concomitant drug use that could modify a person’s response, potentially affecting the therapeutic intent or causing undesired effects, based on pharmacokinetic (PK), pharmacodynamic (PD) and clinical evidence. We excluded entries with incomplete data or those that were not drug–drug related.

Fig. 1.  Study outline of the patients included in each step of the analysis and the number of interactions found. *People either calling for themselves or as a carer of an antidepressant user. CMI, consumer medicines information; DDI, drug–drug interaction; PK, pharmacokinetic; PD, pharmacodynamic.

For the second part of our analysis, we focused on antidepressant users (patients) who were at potential drug–drug interaction risk. Patient demographics, including patients’ age, gender, and their medication use profile, were extracted from calls where the patient was a caller ringing for themself (80.4%) or the data were provided by the caller (19.6%), a family member, carer or friend, where the patient was prescribed an antidepressant, had questions or concerns regarding potential interactions with their concomitant drug use (January 2007–June 2010), and where the database held detailed narratives from each call. All concomitant patients’ medicines (prescribed or self-medicated) were classified by their ATC therapeutic class (Fig. 1).¹³

Pharmacists answering the call categorised interaction queries into two patient groups: patients having an information gap or concern but not experiencing symptoms – the ‘no symptoms’ group; and patients presenting with symptoms of concern to themselves and to the pharmacist answering the call – the ‘worrying symptoms’ group. All calls classified as associated with ‘worrying symptoms’ were referred to patients' designated primary healthcare professional. We excluded calls solely requesting consumer medicines information and entries with incomplete data, including no description of their query. We compared patient characteristics to profile and differentiate between the two patient groups.

To obtain an objective picture of the likelihood of a drug–drug interaction, we used a drug–drug interaction database (YouScript)¹⁴ for primary interaction analysis. We also used Stockley’s Interactions Checker,¹⁵ the Australian Medicine Handbook,¹⁶ AccessPharmacy,¹⁷ and Natural Medicines¹⁸ as secondary resources when a medication was not listed on YouScript. Each call was analysed separately for potential pharmacokinetic and pharmacodynamic interactions. The significance of a potential pharmacokinetic interaction was expressed by severity grade and estimated percentage change in serum drug concentrations as predicted on YouScript: 0% = nil, ~20% = minimal, ~60% = minor, ~100% = major, and >100% = contraindicated. This pharmacokinetic drug–drug interaction scale estimates potential patient impact, taking into consideration the many factors that can influence the clinical impact of drug exposure including dose, therapeutic use, inter-patient and intra-patient variabilities in drug disposition.¹⁹ Pharmacodynamic interactions were classified by potential severity of an interaction-induced adverse event using the National Cancer Institute Common Terminology Criteria for Adverse Events (CTCAE):²⁰ no adverse event (due to a drug–drug interaction); mild: asymptomatic or mild symptoms – no intervention indicated; moderate – symptoms with minimal intervention indicated; severe – medically significant symptoms but not life- threatening – intervention indicated; life-threatening or disabling symptoms – drug use contraindicated. In cases where more than one drug–drug interaction was identified, the highest level was chosen. A team including pharmacists and a general practitioner (EP, GM, DP and MVD) assessed cases where the drug–drug interaction evidence was ambiguous or conflicting.

As data were originally collected for routine service without specific a priori research goals, this research was conducted and reported in accordance with the REporting of studies Conducted using Observational Routinely-collected health Data (RECORD) guidelines.²¹

All variables are categorical, and descriptive results are presented using numbers and percentages. Evidence for a difference between patients with worrying symptoms and no symptoms was investigated using Pearson’s chi-squared test of independence. Statistical analysis was performed using R Statistical Software (version 4.0.3; R CoreTeam 2020).

This study was approved by the Human Research Ethics Committee of Mater Health Services, Brisbane (LNR submission 2012-68).

Of 123 217 calls received from Australian consumers between September 2002 and June 2010, 18 724 (15.2%) involved questions about antidepressants, of which one in four (4292 calls, 22.9%) focused on potential interactions. After applying the inclusion criteria, we analysed 3899 calls for concomitant medicines by antidepressant class and summarised the 20 most common concomitant therapeutic classes. In rank order, antipsychotics, opioid analgesics, benzodiazepines, and complementary medicines constituted more than half of medicines taken concomitantly with antidepressants (Table 1).

Table 1.  Top 20 common concomitant medicines ranked by use for five major antidepressant classes, and all antidepressant classes.

Of 2011 drug–drug interaction calls with a recorded narrative, 1801 (89.6%) patients with interaction queries were categorised as presenting with no symptoms and 210 (10.4%) presenting with worrying symptoms (Table 2). Irrespective of symptom causality, a review of drug–drug interaction resources showed good correlation between consumer help-seeking behaviours for concerns about antidepressant interactions, with a potential drug–drug interaction identified in 81.0% (1628) of these individuals.

Table 2.  Characteristics of antidepressant users (patients) with queries about potential antidepressant interactions (January 2007–June 2010).

More than two in three people who called about potential drug–drug interactions were female. This proportion was similar whether the antidepressant user was experiencing symptoms or not (P = 0.30). Whether or not people were identified as having worrying symptoms was associated with age (P < 0.001, Table 2). Carers of children aged <15 years and older adults (aged 65–74 years) were more likely than other age groups to report experiencing worrying symptoms. The antidepressant drug use profiles were also similar across different antidepressant classes, except for complementary medicines commonly used for depression: for example, St. John’s wort, S-adenosyl-l-methionine (SAMe) (worrying symptoms 1.0% vs no symptoms 7.9%, P < 0.001). The most commonly implicated antidepressants were selective serotonin reuptake inhibitors (SSRI; n = 1069, 53.2%), serotonin noradrenaline reuptake inhibitors (SNRI; n = 347, 17.3%) and tricyclic antidepressants (TCA; n = 277, 13.8%), and depression-related complementary medicines (n = 142, 7.1%).

We identified 2513 interactions in the 2011 patients in our dataset (Table 3). In general, more potential PD interactions were recorded compared to potential PK interactions, but when comparing the distribution of the significance of interaction, PK interactions tended to have a higher significance of clinical impact than PD interactions, with 48.0% of patients experiencing a potential drug level change between 61 and >100%) (Table 3). We describe the frequency of potential PD effects in Table 3; risk of central nervous system (CNS) depression, increased anticholinergic effects, serotonin syndrome and suicidal thoughts were more commonly identified from patients’ concurrent medication use. We compiled a resource of key and common potential interactions that clinicians should be aware of before prescribing an antidepressant, derived from our data and for antidepressants marketed since 2010 (Table 4).

Table 3.  Pharmacokinetic and pharmacodynamic interactions analysis.

Table 4.  Common drug–drug interactions for antidepressant medicines or class by potential outcomes and action needed.

Our real-world medicines call-centre data demonstrated help-seekers who take antidepressants have drug–drug interaction concerns about concomitantly used medicines or lifestyle products including complementary medicines.²² There is a bidirectional relationship between mental health disorders and comorbid chronic physical conditions.^23–25 The guidelines from the National Institute for Health and Care Excellence indicate that one in five (20%) patients with a chronic physical condition also experience depression;²³ whereas World Mental Health surveys from the World Health Organization reported that almost three in four (72%) patients with a major depressive disorder also have a chronic physical condition, increasing the probability of a clinically relevant adverse medication event relating to a drug–drug interaction.²⁶

In our study, CNS-active drugs of dependence; for example, benzodiazepines, opioids, and alcohol, were commonly used concomitantly with antidepressants, consistent with previous research.²⁷ Patients with depression and a substance use disorder; for example, in the form of dependence, were more likely to seek help, but their needs were often unmet due to barriers accessing mental health care.²⁸ It is therefore important to closely monitor patients concurrently using antidepressants and CNS-active drugs to address perceived barriers to mental health care.

Short-term suicidal ideation was reported as a potential concern. There is a correlation between antidepressant dose and suicidal ideation risk;²⁹ a high antidepressant concentration in the blood may be associated with increased suicidality. This highlights the importance of identifying and monitoring for a potential PK drug–drug interaction involving antidepressants.³⁰ If a patient is prescribed a treatment that reduces antidepressant metabolism and/or elimination, it will manifest as an increased antidepressant dose, thereby raising suicidal ideation risk. In contrast, if the new treatment induces antidepressant clearance, symptoms of depression may return, or the patient may experience withdrawal effects. Suicidal ideation generated by antidepressant use continues to court controversy.³¹ Our results suggest that monitoring patient progress during a severe mental health disorder may be suboptimal, and that appropriate, regular, counselling remains essential to ensure the patient comprehends their symptoms and how they can be appropriately managed. For many, this includes reassurance that these symptoms will be temporary, but that they need to seek help should they persist.

Worrying symptoms related to antidepressant drug–drug interactions were more likely to be reported by carers of children aged <15 years and older adults (65–74 years). This would be expected in an older age group who use multiple medicines, as polypharmacy is a predictor of adverse drug reactions,³² but it is less well recognised that children may have a similar rate of experiencing worrying symptoms or adverse drug events.³³ Children and adolescents are increasingly being prescribed antidepressants, with at least 101 174 Australians aged 0–17 years (1.8%) having an antidepressant dispensed between July 2017 to June 2018.³¹ Over an 11- year period (2008–18), Australian antidepressant dispensing (0–27 years of age, per capita) increased 66% and suicide rates (0–24 years) increased 49%.³¹ It is unclear whether suicidality is correlated with antidepressant use or the limited efficacy of antidepressants in suicidal children and adolescents.³⁴ This issue warrants early identification and monitoring for PK drug–drug interactions involving antidepressants, with patients reassured that most symptoms will be temporary, but to seek help should they persist.³⁰

In our study, people using a complementary medicine for depression, for example St John’s wort, were more likely to seek information in the absence of worrying symptoms. This parallels our previous medicines call centre research demonstrating general consumer concerns about complementary medicine risks, where 34% of calls asked about possible interactions with complementary medicines versus only 13% for conventional medicines.³⁵ Mainstream and social media have also increased consumer awareness of the dangers of ‘mixing medicines’. As complementary medicines are often self-initiated without input from health practitioners, it places the onus on the consumer to seek information about potential interactions. This is supported by a study indicating that complementary medicine consumers consult more clinical resources than patients who are on conventional medicines.³⁶ Widespread use of complementary medicines carries increased drug–drug interaction risk.^35,37 This is an opportunity for prescribers to counsel patients that ‘natural’ is not synonymous with ‘safe’, and highlights the importance of gathering information regarding complementary medicine use, especially when first prescribing antidepressants.

We found that potential PD interactions were more prevalent than potential PK interactions, with CNS depression and anticholinergic effects commonly reported as potential symptoms. Clinicians may also underestimate the extent to which less overt drug–drug interaction effects such as daytime somnolence, dizziness, or constipation may contribute to treatment discontinuation or poor adherence.³⁸ When comparing the distribution of interaction severity, a higher proportion of people with no symptoms (49.8%) than people with worrying symptoms (33.6%) were estimated to have a percentage PK drug change between 61 and >100%. However, intrinsic patient characteristics and a drug’s therapeutic window would contribute to interaction risk.³⁹ In contrast, the severity of PD interaction between patients with and without worrying symptoms were similar. This suggests that the clinical impact of PK and PD interactions cannot be directly compared, with current interaction checkers being poor predictors of patients likely to experience actual drug–drug interaction symptoms. They do, however, serve as a flag to explore individual patient characteristics and drug dose as contributors to drug–drug interaction risk. Clearly, patients who are taking antidepressants, with their inherent diverse pharmacological profiles, together with other drugs, are at increased drug–drug interaction risk that can negatively impact treatment outcomes. This highlights the value of a trusted healthcare professional with drug–drug interaction expertise assessing the drug–drug interaction risk of individual patients.

We did not identify any predictor that would indicate which patient group was more likely to experience symptomatic antidepressant drug–drug interactions. Nine in ten patients or carers in our study commonly sought help for their drug–drug interaction concerns despite a lack of worrying symptoms. This may suggest antidepressant users overestimate their medication risk. Maladaptive risk perception is a cornerstone of cognitive models of anxiety disorders where worried individuals generally overestimate negative outcomes,⁴⁰ and emotional reactions to perceived risk or uncertainty often drive behaviour such as help-seeking.⁴¹ As four in five patients had at least one identified drug–drug interaction of potential clinical significance, this validates healthcare professionals being receptive to patients’ drug–drug interaction concerns.

The main strength of this study was that we used real-world, routinely collected health service data and demonstrated remarkable consistency in consumer safety concerns, particularly about antidepressant drug–drug interactions, over an 8-year period. Consistent with information behaviour theories, callers used the medicines call centre to seek information in response to uncertainty associated with worrying symptoms or multiple interpretations of information (inadequate, conflicting, or overload).^11,35

We note two main limitations. First, although the data analysed were collected a decade ago, and have not captured some of the newer antidepressants, the individual antidepressants reported in our study continue to be widely used. A recent analysis comparing antidepressant use in Australia and Sweden demonstrated that SSRIs and SNRIs remain the most commonly used antidepressants over a 13-year period (2006–18).¹² We developed a guide (Table 4) for primary healthcare professionals to identify and monitor common antidepressant drug–drug interactions, their potential outcomes and actions needed. It includes newer antidepressants approved since 2010.

Second, our study findings might not represent all antidepressant users, but rather those concerned enough to contact a medicines call centre. A previous longitudinal analysis of over 125 000 calls to the same medicines call centre found call demographics were remarkably consistent. Callers originated from all Australian states and territories, and importantly, when caller location was grouped by the Accessibility/Remoteness Index of Australia (ARIA),⁴² those from ‘rural and remote’ areas approached the relative population frequency.¹¹ This provides reassurance that a medicines call centre model has utility for the Australian public, particularly those with less access to health services. Moreover, in a study of 2348 calls to a Finnish national consumer medicines call centre surveying use of the different medicines information sources, people with mental disorders were more frequent users of telephone- and internet-based medicines information sources and patient information leaflets than people without mental disorders.⁴³ Call centre popularity is cited as primarily due to their low consumer cost, ease of access, immediacy of help, provision of caller anonymity and protection from stigma.^44,45 Furthermore, some patient characteristics such as ethnicity were not recorded by the medicines call centre, although previous research has identified it can impact the prevalence of antidepressant use.⁴⁶ Finally, the study design precluded factors such as adherence and under-reporting being explored and thus, requires future research to address these factors.

In conclusion, antidepressant users who information seek commonly take combinations of medicines and lifestyle drugs, and often have drug–drug interaction safety concerns. Prescribers may underestimate the risk of excessive sedation, anticholinergic symptoms, serotonin syndrome, and suicidal ideation in people using antidepressants together with other drugs. Primary healthcare professionals, especially general practitioners and pharmacists, are ideally placed to identify clinically relevant drug–drug interactions when a patient commences an antidepressant or if there is a medicine change. They can proactively address any potential drug–drug interaction risk or patient interaction concerns to encourage shared decision-making.

The data that support this study were obtained through a shared agreement between NPS MedicineWise and Mater Health Services, Brisbane by permission/licence. Data will be shared upon reasonable request to the corresponding author with permission from both parties.

The authors declare no conflicts of interest.

This study did not receive any specific funding.

TM, GM, MvD, SH, DP and EP contributed to the study design. EP and HL contributed to data collection and management. EP, DP, GM and TM contributed to analysis. All authors contributed to interpretation of results and the final manuscript.