INTRODUCTION

The prevention of adverse drug reactions (ADRs) represents an essential goal of rational and safe pharmacotherapy, especially in geriatric patient populations.¹ In addition to potentially inappropriate medications (PIMs) for elderly patients, duplicate prescriptions (DPs)—among other factors—have been identified as a risk factor for the development of ADRs.² Of note, DPs are defined heterogeneously in the literature (Table 1), with—for example—the description of a DP as the ‘concurrent use of two drugs of the same class to treat the same condition’ falling noticeably short.² To exemplify this, the co-administration of two drugs that belong to different pharmacological classes—such as the combination of an angiotensin-converting enzyme inhibitor with an angiotensin-receptor blocker—can also be considered a DP.

It is striking that to date the concept of DPs in the scientific literature has been associated almost exclusively with detrimental effects for patients, ignoring the fact that, for example, the combined use of oral antidiabetic agents (e.g., the combination of metformin with a sodium–glucose cotransporter type 2 (SGLT2) inhibitor) in patients with type-2 diabetes mellitus can exert meaningful therapeutic benefits.³ The majority of existing DP definitions, including the frequently used Screening Tool of Older Persons' Prescriptions (STOPP) criteria,⁴ do not (sufficiently) differentiate between clinically appropriate and inappropriate DPs (Table 1). To overcome the currently prevailing inconsistency and largely absent consideration of DP appropriateness of DP definitions found in the scientific literature (Table 1), Heck and colleagues proposed a new categorisation of DPs that fundamentally distinguishes between appropriate duplicate prescriptions (ADPs) and potentially inappropriate duplicate prescriptions (PIDPs).⁵ ADPs represent rational and commonly used combination treatments (often as add-on therapy, combining treatment principles with synergistic mechanisms of action), as opposed to PIDPs. PIDPs are subdivided into three different grades; higher grades reflect an increasing degree of inappropriateness (Table 2). It must be stressed, however, that the degree of inappropriateness of PIDPs per se is not equivalent to clinical severity since it is conceivable that an ADR resulting from a grade-1 PIDP is clinically more severe than an ADR elicited by a grade-2 or grade-3 PIDP.

Elderly patients are at increased risk of developing ADRs due to altered organ functions and age-related changes in pharmacokinetic and pharmacodynamic characteristics.⁶ Furthermore, elderly patients are often exposed to multimorbidity-associated polypharmacy (i.e., the concomitant intake of ≥5 different drugs) and prescription of PIMs, further increasing the likelihood of developing ADRs.⁷ Gerontopsychiatric patient populations have been identified as high-risk groups for ADR occurrence in recent years.⁸ While the influence of inappropriate polypharmacy and PIM prescriptions has been investigated quite extensively in this context,^{9, 10} prescribing characteristics of DPs in gerontopsychiatric patients have not been analysed comprehensively to date.

Therefore, the present study sought to investigate the frequency and characteristics of DPs in a cohort of elderly patients treated on the gerontopsychiatric ward of a large university hospital in northern Germany on the basis of weekly medication reviews conducted by an interdisciplinary expert team. Particular emphasis was put on the differentiation between ADPs and PIDPs.

METHODS

RESULTS

PIDPs

One hundred and forty-six of 339 patients for whom medication chart reviews were performed (43.1%) were exposed to at least one PIDP. As demonstrated in Table 3, there were no statistically significant differences in terms of age or sex between patients with and without PIDPs.

With respect to PIDP grades, 39.5% (134/339) of the study population were exposed to at least one grade-1 PIDP, 5.0% (17/339) to at least one grade-2 PIDP, and 1.5% (5/339) to at least one grade-3 PIDP.

Overall, 185 PIDPs were detected in the study population. The majority of PIDPs were of grade 1 (88.1%; 163/185) (Fig. 1a). Sedatives were most frequently responsible for grade-1 PIDPs (58.3%; 95/163) (Fig. 1b), with lorazepam + pipamperone (49/95) being most commonly implicated (Table 4). Sedatives were also most frequently responsible for grade-2 PIDPs (76.5%; 13/17) (Fig. 1c), with melperone + pipamperone (11/13) being the predominant combination (Table 4). Of note, all grade-3 PIDPs were elicited by non-opioid analgesics (100%; 5/5). Etoricoxib taken both regularly and on demand (thus exceeding the approved maximum daily dose of 120 mg), and metamizole taken both regularly and on demand (thus exceeding the approved maximum daily dose of 4000 mg) accounted for 40% (2/5) and 60% (3/5) of grade-3 PIDPs, respectively (Table 4).

Table 4. Absolute and relative frequencies of potentially inappropriate duplicate prescriptions detected in the study population

PIDPs	n	%†
Grade-1 PIDPs	163	100
Antidepressants	5	3.1
Opipramol + venlafaxine	1	0.6
Paroxetine + trazodone	1	0.6
Sertraline + trazodone	3	1.8
Antipsychotics	19	11.7
Amisulpride + aripiprazole	2	1.2
Amisulpride + clozapine	1	0.6
Amisulpride + olanzapine	3	1.8
Amisulpride + risperidone	1	0.6
Aripiprazole + olanzapine	1	0.6
Clozapine + risperidone	1	0.6
Haloperidol + olanzapine	2	1.2
Haloperidol + quetiapine	2	1.2
Olanzapine + risperidone	1	0.6
Quetiapine + risperidone	5	3.1
Sedatives	95	58.3
Alprazolam + zopiclone	2	1.2
Pipamperone + valerian	6	3.7
Chlorprothixene + lorazepam	5	3.1
Diazepam + pipamperone	5	3.1
Lavender oil + pipamperone	2	1.2
Lorazepam + melperone	3	1.8
Lorazepam + pipamperone	49	30.1
Lorazepam + promethazine	1	0.6
Lorazepam + zopiclone	1	0.6
Melperone + oxazepam	2	1.2
Oxazepam + pipamperone	5	3.1
Pipamperone + low-dose quetiapine	5	3.1
Pipamperone + zolpidem	2	1.2
Pipamperone + zopiclone	7	4.3
Analgesics	38	23.3
Diclofenac + metamizole	6	3.7
Ibuprofen + metamizole	20	12.3
Ibuprofen + paracetamol	4	2.5
Metamizole + paracetamol	7	4.3
Paracetamol + tramadol	1	0.6
Anticoagulants	2	1.2
Edoxaban + tinzaparin	2	1.2
Laxatives	4	2.5
Lactulose + macrogol	4	2.5
Grade-2 PIDPs	17	100
Antidepressants	3	17.6
Citalopram + sertraline	1	5.9
Duloxetine + venlafaxine	2	11.8
Sedatives	13	76.5
Alprazolam + oxazepam	1	5.9
Lorazepam + oxazepam	1	5.9
Melperone + pipamperone	11	64.7
Diuretics	1	5.9
Furosemide + torasemide	1	5.9
Grade-3 PIDPs	5	100
Analgesics	5	100
Etoricoxib taken regularly and on demand (exceeding the approved maximum daily dose of 120 mg)	2	40
Metamizole taken regularly and on demand (exceeding the approved maximum daily dose of 4000 mg)	3	60

• ^† Percentages may not total 100 because of rounding.
• Abbreviation: PIDP, potentially inappropriate duplicate prescription.

DISCUSSION

The present pilot study sought to investigate the frequency and characteristics of DPs in a cohort of elderly psychiatric inpatients treated in a university hospital in Germany. The analysis was based on weekly medication chart reviews conducted by an interdisciplinary expert team. We detected at least one PIDP in 43.1% of all reviewed medication charts. Of note, 39.5%, 5.0%, and 1.5% of the study population were exposed to at least one grade-1, grade-2, and grade-3 PIDP, respectively. Sedatives accounted for the majority of grade-1 and grade-2 PIDPs, while non-opioid analgesics were responsible for all grade-3 PIDPs. Lorazepam + pipamperone and melperone + pipamperone represented the most frequent grade-1 and grade-2 PIDP combinations, respectively. Etoricoxib taken regularly + etoricoxib PRN and metamizole taken regularly + metamizole PRN constituted the grade-3 PIDP combinations. Of interest, patients with PIDPs took significantly more drugs than patients without PIDPs, and—accordingly—were significantly more often affected by polypharmacy.

Almost half of the study population displayed ADPs. This is an important finding because previous studies did not sufficiently discriminate between beneficial and potentially harmful DPs. Instead, other authors regarded DPs exclusively as deleterious to patients' health,^14-18 a notion that appears to be challenged by our study results.

When comparing our results with those of other studies we must stress that there is a plethora of different terminologies and definitions of DPs (Table 1; demonstrated by the use of quotation marks upon first appearance in the following text), which substantially limits comparability.

Zwietering et al. found a very low (between 0% and 1.9%) prevalence of ‘double medications’ in geriatric outpatients,¹⁵ suggesting that outpatients may be less affected by PIDPs than inpatients, and that psychiatric illness may pose a particular risk factor for the occurrence of PIDPs.

Kinoshita and colleagues reported an 8.8% prevalence of ‘duplicative medications’ among insurants of a corporate health insurance society.¹⁶ Of note, their study population was considerably younger than ours (mean age (± standard deviation) 33.4 (±17.5) years¹⁶ vs median age (IQR) 73 years (68–82 years)), which makes it very difficult to compare our study results with Kinoshita's. Furthermore, Kinoshita et al. analysed health insurance data,¹⁶ while we strictly focused on clinical data. Astonishingly, whereas antibiotics represented a common cause of duplicative medications in the Kinoshita et al. study,¹⁶ they were not implicated in any PIDPs in our study, a discrepancy that might be explained by the fact that all DPs of antibiotics in our study were classified as ADPs, while Kinoshita et al. did not distinguish between appropriate and inappropriate DPs.¹⁶

According to Heinze and co-workers, 13–15% of the Austrian population taking antihypertensive, antilipidemic, or antidiabetic agents were affected by ‘double medications’.¹⁴ However, their applied methodology did not allow for a discrimination between appropriate and inappropriate double medications. Besides, detection of grade-3 PIDPs was not possible with Heinze et al.'s definition of double medications.¹⁴

In a cross-sectional study of potentially inappropriate prescribing in older patients admitted to a psychiatric hospital in The Netherlands, benzodiazepines, antipsychotics, and antidepressants most frequently accounted for ‘any regular duplicate drug class prescription’.¹⁰ The definition used by the authors (which was based on STOPP criteria) resembled our definition of grade-2 PIDPs, albeit without differentiating between appropriate and potentially inappropriate DPs. Therefore, it remains elusive whether, for example, the (clinically indicated) co-administration of two platelet aggregation inhibitors (e.g., low-dose acetylsalicylic acid plus a P2Y₁₂ receptor antagonist) subsequent to coronary stent implantation¹⁹ would have been assessed as a ‘regular duplicate drug class prescription’. This again demonstrates the advantage of our DP categorisation (which does differentiate between ADPs and PIDPs) compared to other DP definitions (Table 1).

Analgesics accounted for 23.3% of grade-1 PIDPs in our study, with ibuprofen + metamizole being the predominant combination. Similarly, in a retrospective cohort study among multimorbid elderly outpatients, diclofenac + metamizole were most frequently responsible for ‘DPs’ equivalent to grade-1 PIDPs.¹⁸ Grade-2 PIDPs due to analgesics were not observed in our study; however, all grade-3 PIDPs were elicited by analgesics. In the abovementioned study,¹⁸ two DPs equivalent to grade-3 PIDPs were detected, both elicited by diclofenac + diclofenac. In our study, by contrast, etoricoxib + etoricoxib and metamizole + metamizole were responsible for grade-3 PIDPs.

Kovac and colleagues investigated ‘dual use’ of analgesics and demonstrated a correlation between dual non-steroidal anti-inflammatory drug (NSAID) use and worse physical health-related quality of life, albeit in a highly selected patient population (their study population consisted exclusively of NSAID users).¹⁷ According to the DP categorisation applied in our study,⁵ dual NSAID use as defined by Kovac et al.¹⁷ is equivalent to grade-2 or grade-3 PIDPs. Although grade-2 PIDPs due to analgesics were not observed in our study, and although Kovac et al. analysed the effects of dual use in only one medication class (i.e., NSAIDs), their results suggest that PIDPs might in fact be harmful to patients.

A strength of our study is the large number of medication chart reviews which were conducted by an interdisciplinary expert team comprising specialists from psychiatry, internal medicine, geriatrics, and clinical pharmacology. We are convinced that utilisation of the DP categorisation by Heck et al.⁵ minimised subjective bias and abrogated the blending of clinically beneficial and detrimental DPs, which was a major limitation of previous works (Table 1). On the other hand, the relatively short duration of our pilot study, as well as its retrospective and monocentric design clearly represent limitations. Our study was carried out at a single university hospital; therefore, the results may not be generalisable to other settings in which elderly psychiatric patients are treated and cared for.

The most important limitation of our study is the lack of information about the number of cases in which PIDPs actually resulted in adverse events. Such an evaluation was not feasible due to the retrospective character of our study, which did not allow for a precise and unbiased ADR causality assessment. Future prospective investigations, however, should evaluate the clinical significance of PIDPs, for example, by systematically analysing PIDP-associated ADRs immediately upon occurrence with established causality assessment tools.^{20, 21}

Even though the clinical implications of PIDPs are not fully established to date, we recommend that physicians who treat elderly psychiatric patients pay special attention to PIDPs, especially PIDPs elicited by sedatives. Termination of PIDPs due to sedatives may not only reduce the risk of cognitive impairment and the risk of falls, but may also save healthcare expenditures caused by unnecessary drugs in an ageing society.

AUTHOR CONTRIBUTIONS

MSW, SS, BK, and JH conceptualised the study. MSW, SS, FK, BK, and JH analysed the study data. BK provided statistical advice. BK and JH conducted the statistical analyses. BK devised the tables. JH created the figure. All authors contributed to the interpretation of the study results. MSW, SS, AG, and JH wrote the first draft of the manuscript. All authors commented on previous versions of the manuscript. All authors read and approved the final version of the manuscript. The corresponding author (JH) attests that all listed authors meet authorship criteria and that no persons meeting the criteria have been omitted.

DISCLOSURE

The authors declare that there are no conflicts of interest. There are no funders to report for this submission.