A cross-sectional study of psychotropic drug use in the elderly: Consuming patterns, risk factors and potentially inappropriate use

Abstract

Objectives

The aims of the present study were: (1) to describe psychotropic drug consumption patterns in an outpatient population aged 65 years and older; (2) to determine the impact of a number of demographic and clinical factors on psychotropic consumption; and (3) to determine the ratio of potentially inappropriate psychotropic agents prescribed to the above population.

Methods

Cross-sectional, observational study of outpatients aged 65 years and older. Data on sociodemographic and clinical variables were collected. Psychotropic drugs were classified into three categories: anxiolytics-hypnotics, antidepressants, and antipsychotics. To determine the risk factors for psychotropic drug use among these patients, a multivariate logistic regression model was developed and subsequently validated using bootstrap resampling techniques. To identify the psychotropic drugs to be avoided, a review of treatments received by the patients was performed based on the 2015 version of the Beers criteria.

Results

The study included 225 outpatients of whom 30.7% were on psychotropic drugs for chronic treatment. The highest likelihood of psychotropic utilisation corresponded to the following profile: female, living in a nursing home, having two or more prescribing physicians, and having received six or more different diagnoses. According to Beers criteria, 51 patients (22.7% of the sample and 73.9% of patients on psychotropic drugs) had been prescribed at least one potentially inappropriate psychotropic drug.

Conclusion

Elderly patients commonly use psychotropic medications and are the most vulnerable to the adverse effects of these drugs. It is necessary to re-evaluate the pertinence and accuracy of these medical prescriptions.

Introduction

Nowadays, there is an increasing tendency toward the medicalisation of everyday life.¹ Thus, the transformation of the ordinary and self-limiting processes of life into medical problems has led to an augmentation of drug prescribing to the general population, including a striking increase in the number of psychotropic drugs prescribed to the elderly population.^2–7 Furthermore, with the prolongation of life expectancy, elderly people suffer from more comorbidities and, consequently, often need an increasing number of medications.

Sleep pattern disturbances, loneliness resulting from the loss of a spouse, and anxiety and feelings of sadness make the elderly a propitious social group for the prescribing of psychotropic medications.⁵ In addition, there is now an increased number of approved indications for several psychotropic drugs, as well as a more frequent off-label use of these drugs.⁸ Indeed, an off-label prescription prevalence up to 60% has been reported.⁹ All these factors have favoured an increment of psychotropic drug consumption in all age groups.

There has been a critical discussion over the long-term utilisation of psychiatric medication; recently the director of the Nordic Cochrane Centre published a paper stating that every year psychotropic drugs cause a large number of deaths all over the world, especially among older people, with minimal clinical benefit. Based on the available evidence, he suggested that the prescription of nearly all psychotropic agents could be stopped without causing harm to patients. In contrast, many psychiatrists and patients believe that the benefits of psychotropic drugs outweigh their potential harm, and that the question should be assessed on a case-by-case basis.¹⁰

To date, there have been few population surveys on the utilisation of psychotropic drugs in Spain. An increased prevalence of psychotropic use among the elderly population from 4.6% in 1993 to 20.5% in 2003 was found in the periodic National Health Survey.² According to data from the ESEMeD-España study, which was released in 2007, the prevalence of psychotropic utilisation was as high as 25.9% among those aged >65 years.⁴ Data from the majority of surveys indicate that anxiolytics and antidepressants are the most frequently used psychiatric drugs.^{4 6 7}

The prescription of psychotropic drugs in the elderly patient can be problematic because broad comorbidities and polypharmacy, which are very common in these patients, can give rise to a large number of drug interactions. These two factors, along with the characteristic pharmacokinetic and pharmacodynamic changes associated with advanced age, as well as the increased susceptibility to conditions related to tolerance and dependence, result in an increased risk of side effects, such as diminished cognition, excessive sedation, and a heightened risk of falls and fractures.^{11 12}

A strategy aimed at decreasing the risk of side effects involves the avoidance of those medications whose benefit-risk ratio is unfavourable. To this purpose, Beers criteria have been developed, with the aim of explicitly identifying those medications that can be potentially inappropriate for the elderly.¹³ Several studies have assessed psychotropic medication with these criteria and found a significant proportion of patients, ranging from 7–27%, were on a potentially inappropriate drug.^{14 15}

Summarising, the rise in the consumption of psychotropic medications is a matter of concern, not only from a clinical point of view, but also because of the high financial burden for healthcare systems. In Spain, the cost of drugs accounts for around 17% of the overall public healthcare expenditure,¹⁶ with a substantial amount spent on psychiatric medication. In fact, psychotropic agents rank third among the most frequently prescribed drugs in Spain, only surpassed by antihypertensive and anti-ulcer agents.¹⁷

The aims of the present study were: (1) to describe psychotropic drug consumption patterns in the outpatient population aged 65 years and older; (2) to determine the impact of a number of demographic and clinical factors on psychotropic consumption; and (3) to determine the ratio of potentially inappropriate psychotropic agents prescribed to the above population.

Methods

Study population

This was an observational, cross-sectional study, conducted in Spain in March 2016. The study population consisted of outpatients aged 65 years and older who had an appointment to see a specialist at any hospital department, except for oncology, selected by random sampling at two hospitals situated in Soria and Valladolid, Spain. In this way, we obtained a heterogeneous sample of the target ambulatory population, including psychiatry outpatients. A detailed explanation about the procedure for sample size calculation has been recently reported.¹⁸ The ultimate sample comprised 225 patients with valid data for all variables.

Variables

Only chronic prescriptions were included—that is, the drugs prescribed under the Electronic Prescription Programme (Medora®)¹⁹—within at least the previous 3 months. For the purposes of this study, psychotropic drugs were classified into three categories according to the anatomical therapeutic chemical (ATC) coding system: anxiolytics-hypnotics, antidepressants, and antipsychotics.²⁰ We decided to group anxiolytics and hypnotics into the same category because the percentage of patients on hypnotics was too small, and, in addition, these agents belong to an analogous pharmacological class. In the category ‘antidepressants’, a distinction was make between selective serotonin reuptake inhibitors (SSRIs), tricyclics and others antidepressants, while antipsychotics were divided into typical and atypical.

We collected data on the demographic and clinical variables that were regarded as being relevant from a clinical point of view. Data on the study variables were collected from patients’ electronic medical records and then entered in an anonymised database. Demographic variables included in the study were the following: age at the beginning of the study, sex, place of residence (urban vs rural), and institutionalisation (home vs institution). Clinical variables were: number of different prescribing doctors, polypharmacy (yes/no; defined as the continuous use of more than five drugs), and number of diagnoses according to the ninth revision of the International Classification of Diseases (ICD-9).²¹

Statistical analysis

A descriptive statistical analysis was performed. The quantitative variables were expressed as mean±SD, while categorical variables were expressed as frequencies and percentages.

For categorical variables, the comparisons between the group of patients on psychotropic drugs and the group of patients on non-psychotropic drugs were carried out by means of the Pearson χ² test.

Bootstrap resampling was used to select significant variables that were predictive of psychotropic use and to estimate the internal validity of the predictive logistic regression model. The odds ratios (ORs) and confidence intervals (95% CIs) were estimated. To measure the discriminating accuracy of the tests, the area under the curve and its 95% CI of the receiver-operating curve (ROC AUC) were scrutinised.

To identify the psychotropic medications to be avoided, a review of treatments received by the patients was performed based on the 2015 version of the Beers criteria,²² adapted for commercially available drugs in Spain.

Statistical analyses were performed using software packages IBM SPSS Statistics version 24 with a campus licence. Statistical significance was defined as p<0.05 for all tests. The study was approved by the ethics committees of the healthcare areas within which the work was undertaken (Soria and Valladolid, Spain).

Results

The study included 225 outpatients (mean age 78.0±7.2 years). Female patients represented 52.0% of the study sample. Mean drug consumption was 5.5±3.1. Polypharmacy (>5 drugs) was seen in 46.7% of the patients. Concerning psychotropic medication, 69 (ie, 30.7% of the sample) patients were on psychotropic drugs for chronic treatment. As shown in table 1, 73.9% of patients taking psychotropic drugs were on anxiolytics/hypnotics, 47.8% on antidepressants, 18.8% on antipsychotics, and 44.9% took more than one psychotropic class.

Table 1.

Prevalence of use of the different types of psychotropic medications (N=69)

Medications*	n (%)
Anxiolytics/hypnotics	51 (73.9)
Lorazepam	16 (23.2)
Alprazolam	8 (11.6)
Lormetazepam	8 (11.6)
Bromazepam	6 (8.7)
Diazepam	6 (8.7)
Others	11 (15.9)
Antidepressants	33 (47.8)
SSRIs	22 (31.9)
Tricyclics	4 (5.8)
Others	9 (13.0)
Antipsychotics	13 (18.8)
Typical	8 (11.6)
Atypical	6 (8.7)
N° of psychotropic drugs
1	38 (55.1)
≥2	31 (44.9)

Values are given as number of patients (%).

*The percentage exceeds 100% because some patients consume more than one type of psychotropic medication.

SSRIs, selective serotonin reuptake inhibitors;

Of the 73.9% of patients on anxiolytics/hypnotics, 23.2% were on lorazepam. With regard to antidepressants, SSRIs were the most frequently used subclass (31.9% of patients on psychotropic drugs), while only 5.8% was on tricyclics and 13.0% took other antidepressants. In the group of patients on antipsychotics, 11.6% and 8.7% took typical and atypical antipsychotics, respectively.

Table 2 shows the relationship between psychotropic medication consumption and patients’ demographic and clinical characteristics. Psychotropic consumption was significantly higher among females (37.6%) as compared with males (23.1%), though these figures were slightly different when comparing each class of psychotropic agents (figure 1). In addition, we noted statistically significant differences in the variables of institutionalisation, number of prescribing doctors, number of diagnoses, and polypharmacy (yes/no). While psychotropic utilisation seemed to increase with age, we did not find any statistically significant differences with respect to this variable. Likewise, we did not note any differences for the variable place of residence (urban vs rural).

Table 2.

Relationship between psychotropic medication consumption and patients’ demographic and clinical characteristics

	All patients	Patients receiving psychotropic drugs
		N (%)	95% CI	P value*
Age (years)				0.128
65–69	39	8 (20.5)	10.8 to 35.5
70–74	44	9 (20.5)	11.2 to 34.5
75–79	47	15 (31.9)	20.4 to 46.2
80–84	51	20 (39.2)	27.0 to 52.9
85+	44	17 (38.6)	25.7 to 53.4
Sex				0.019
Male	108	25 (23.1)	16.2 to 31.9
Female	117	44 (37.6)	29.4 to 46.6
Place of residence				0.796
Urban	104	31 (29.8)	21.9 to 39.2
Rural	121	38 (31.4)	23.8 to 40.1
Institutionalisation				<0.001
Home	203	54 (26.6)	21.0 to 33.1
Institution	22	15 (68.2)	47.3 to 83.6
Number of different prescribers				<0.001
1	145	30 (20.7)	14.9 to 28.0
2+	80	39 (48.8)	38.1 to 59.5
Number of diagnoses				<0.001
1–5	87	13 (14.9)	8.9 to 23.9
6+	138	56 (40.6)	32.7 to 48.9
Polypharmacy				0.001
No	120	25 (20.8)	14.5 to 28.9
Yes	105	44 (41.9)	32.9 to 51.5
Total	225	69 (30.7)	25.0 to 37.0

*Pearson χ² tests were used to evaluate the differences between psychotropic drug users.

Polypharmacy, >5 drugs.

Figure 1.

Psychotropic drug use by gender.

Table 3 shows the adjusted and non-adjusted estimation for prediction of psychotropic medication use. Based on the variables that were selected by means of the adjusted model, the highest likelihood of psychotropic medication utilisation corresponded to the following profile: female (OR 2.29, 95% CI 1.18 to 4.42), living in a nursing home (OR 6.69, 95% CI 2.33 to 19.18), having two or more prescribing physicians (OR 3.04, 95% CI 1.59 to 5.80), and having received six or more different diagnoses (OR 2.78, 95% CI 1.34 to 5.77). The ROC AUC for the model-building sample was 0.77 (95% CI 0.71 to 0.83).

Table 3.

Predictors of psychotropic consumption derived from the multivariate logistic regression

	Crude OR* (95% CI)	P value	Adjusted OR† (95% CI)	P value
Age (years)
65–69	Reference
70–74	1.00 (0.34 to 2.90)	0.995
75–79	1.82 (0.67 to 4.89)	0.237
80–84	2.50 (0.96 to 6.52)	0.061
85+	2.44 (0.91 to 6.54)	0.076
Sex
Male	Reference		Reference
Female	2.00 (1.12 to 3.59)	0.020	2.29 (1.18 to 4.42)	0.014
Place of residence
Urban	Reference
Rural	1.08 (0.61 to 1.91)	0.796
Institutionalisation
Home	Reference		Reference
Institution	5.91 (2.29 to 15.28)	<0.001	6.69 (2.33 to 19.18)	<0.001
Number of different prescribers
1	Reference		Reference
2+	3.65 (2.01 to 6.61)	<0.001	3.04 (1.59 to 5.80)	0.001
Number of diagnoses
1–5	Reference		Reference
6+	3.89 (1.97 to 7.68)	<0.001	2.78 (1.34 to 5.77)	0.006
Polypharmacy
No	Reference
Yes	2.74 (1.52 to 4.93)	0.001

*Crude OR by univariate logistic regression.

†Adjusted OR by multivariate logistic regression. Includes only variables retained after bootstrap selection.

Polypharmacy, >5 drugs.

According to Beers criteria, 51 patients had been prescribed at least one potentially inappropriate psychotropic drug, independently of diagnosis, which accounted for 22.7% of the sample and 73.9% of patients on psychotropic drugs. Furthermore, 23.5% (n=12) of these patients had been prescribed two potentially inappropriate psychotropic medications. As shown in table 4, the most frequently involved psychotropic class was that of anxiolytics/hypnotics, because long-, intermediate- and short-acting benzodiazepines are considered to be potentially inappropriate drugs, according to the 2015 update of Beers criteria.

Table 4.

Use of potentially inappropriate psychotropic medications* (N=51)

Medications	n (%)
Antidepressants	16 (31.4)
Amitriptyline, paroxetine
Antipsychotics, first and second generation†	9 (17.6)
Chlorpromazine, haloperidol, aripiprazole
Asenapine, olanzapine, paliperidone
Quetiapine, risperidone, zuclopenthixol
Sulpiride, tiapride
Benzodiazepines: long-, intermediate-, and short-acting‡	33 (64.7)
Diazepam, clorazepate, lorazepam, midazolam
Alprazolam, bromazepam, lormetazepam
Non-benzodiazepine hypnotics	5 (9.8)
Zolpidem, zopiclone

*As determined by the 2015 Beers list.

†Except for treating schizophrenia or bipolar disorder.

‡Except for treating seizure disorders, rapid-eye-movement sleep disorders, benzodiazepine withdrawal, ethanol withdrawal, severe generalised anxiety disorder, and periprocedural anaesthesia.

Discussion

In the present study, we found a psychotropic consumption prevalence in outpatients aged 65 and older of 30.7%. This figure is higher than that reported in earlier Spanish and other European studies,2–4 6 7 but lower than that found by Kuopio for patients older than 75 years in Finland (37%).²³

There are considerable differences in consumption prevalence across European countries. In the ESEMeD Project, the authors found a psychotropic consumption prevalence in patients aged >65 years of 18.8%, with Holland and Germany being the countries with the lowest prevalence rates, while France presented the highest (more than three times higher than in Holland), and Belgium, Italy and Spain were found to have intermediate prevalence rates.⁷ In another study, the authors observed notably higher prevalence rates in France and Italy compared with Germany and the UK.⁶

These striking differences may be due to multiple factors, including differences in the prevalence of mental disorders, legal regulation for psychotropic prescription, and differences in design/methods in each study. On the other hand, the existence of programmes to review psychotropic drug use, like the English National Dementia Strategy (a government plan to reduce the use of antipsychotic medication in people with dementia, with pharmacist involvement), should be also taken into account.²⁴

In keeping with earlier studies, we found that anxiolytics/hypnotics were the most frequently prescribed psychotropic agents, followed by antidepressants and antipsychotics.4 6 7 23

In Spain, the lengthy, and often inappropriate, use of anxiolytics or hypnotics reportedly constitutes a major public health concern.⁴ It should be stressed that some frequently used psychotropic drugs, like benzodiazepines, entail the risk of dependence with continuous use, as well as other more controversial risks, such as cognitive deterioration in the elderly.¹² In addition, the use of antipsychotics in people with dementia can cause significant harm, including all-cause mortality, cerebrovascular adverse events, and falls and fractures, with only a limited benefit in treating behavioural symptoms.²²

In our study, 44.9% of patients on psychotropic drugs used concomitantly two or more drugs, which represents a prevalence higher than that reported in earlier studies (about 30%).²³ There are few data supporting the use of two or more psychotropic medications of the same or different class; however, this is a very widespread clinical practice, which results in both increased adverse effect risk and healthcare costs.

In this study, the factors best predicting psychotropic use were sex, institutionalisation, number of prescribing physicians, and number of diagnoses.

The influence of the patient’s sex in psychotropic consumption is a usual finding in the literature.2–7 15 23 In our study, female patients presented a 2.3-fold higher likelihood of being prescribed a psychotropic drug compared with their male counterparts. Multiple hypotheses had been put forward to explain why female patients are more frequently exposed to these drugs. Thus, some authors indicate that greater job instability plays an important role, while other authors put special emphasis on women’s greater willingness to express mental symptoms and seek medical help as compared with men.^{2 7}

In an additional analysis, like Simoni-Wastila,²⁵ we analysed psychotropic consumption by sex in the different therapeutic groups, and found that the differences in psychotropic medication were due to the higher use of antidepressants and anxiolytics/hypnotics among female patients, while consumption of antipsychotics was similar in both sexes.

In line with other studies,²⁶ we found that the highest psychotropic consumption odds ratio corresponded to the factor ‘institutionalisation’ (OR 6.69). The psychosocial impairment, the coexistence of more than one disease or condition, and patients’ functional dependence are some of the factors which are characteristic of institutionalised patients. These factors can explain the frequent use of psychotropic medication in institutionalised patients. Indeed, it has been reported that 98% of patients living in a nursing home used at least one central nervous acting medicine as compared with 65% of community-dwelling older patients.²⁷

Having more than one prescribing doctor is also a potential risk factor for the use of psychotropic drugs, which concurs with findings of earlier studies.²⁸ In fact, in our study, the risk was multiplied by three when the patient had two or more prescribing doctors.

As was the case in other studies,2 3 5 6 in our study psychotropic drug use prevalence was higher in patients with health problems. Suffering from certain kinds of diseases/conditions during lengthy periods of time can result in depressive episodes and cause a negative self-perception of the patient's own health state, which, in turn, is associated with taking psychoactive substances.

Psychotropic medications are rarely prescribed as a single agent, and they are most often prescribed in combination with drugs belonging to many pharmacological classes.^{3 23} In fact, we found that polymedicated patients took more psychotropic drugs than their non-polymedicated counterparts. However, after adjusting for the rest of the variables, the multivariate analysis showed that polypharmacy did not predict psychotropic use.

Other currently recognised factors whose implication we could not find in our study are age and place of residence (urban/rural). Many studies have pointed out that the use of psychotropic drugs increases with age.^{4–7 23} However, while the prevalence of both physical and mental diseases rises with age, doctors are generally cautious in prescribing psychotropic medications to older patients because of the increased risk of adverse effects. As far as the place of residence is concerned, several authors have reported a lower consumption of psychotropic medications in rural areas.^{4 5 15} This is likely to be due to lifestyle-related factors, since it is known that urban lifestyle is more stressful than that found in rural communities.

An obvious drawback in our study is that all the data were collected exclusively by reviewing the patients’ medical records; consequently, we could not include variables that have been shown to be predictive of psychotropic utilisation in earlier studies, such as employment status, educational level or marital status.2–5 7 23

The use of bootstrapping as an internal validation technique was a strength of this study. The bootstrap technique is the preferred approach for internal validation of prediction models, which has been demonstrated by the consistency of our results with previously published data.^{18 29}

Concerning inappropriate medication prescribing, it is said that a drug should be avoided when the risk associated with its use outweighs its potential benefits. Our study disclosed a high prevalence of patients on potentially inappropriate psychotropic medication—22.7% of the sample and 73.9% of the patients on this kind of medication—which is strikingly higher than the prevalence rates of 7.1% and 37.9%, respectively, found in an earlier study.¹⁵ These substantial differences may be due in part to the fact that we included institutionalised patients (ie, patients living in a nursing home) who were more likely to have been inappropriately prescribed antipsychotics for behavioural symptoms of dementia. However, it is our belief that these differences are mainly due to the different versions of Beers criteria used in each one of the studies. In ours, we used the 2015 version, which includes a criteria expansion,²² while in the other study the authors used the 1997 version.¹³

Despite the existence since 1991 of explicit criteria to identify potentially inappropriate drugs, elderly patients continue to be prescribed this type of medication all too frequently. Some explanatory factors are doctors’ lack of knowledge, the absence of consensus as to the best criteria to determine whether a medication is inappropriate, the high number of concomitant prescribing doctors, and the pressure exerted by patients themselves to get their psychotropic medications. Moreover, there are a number of difficulties in deprescribing these medicines, including potential withdrawal reactions, and also general barriers to deprescribing that include physician inertia and patients' attitude surrounding the benefits and lack of harm of their medications.³⁰

Conclusions

We found that almost one third of the patients in our sample were on psychotropic drugs for chronic treatment. Of note, female patients living in a nursing home, with more than one prescribing doctor and worse health state, are the most exposed to this kind of medication. Our study also disclosed a high prevalence of patients on potentially inappropriate psychotropic medication.

In view of these results, it is necessary to develop good clinical practice protocols, in order to minimise the adverse events related to these drugs. These protocols should give priority to the utilisation of non-pharmacological therapeutic measures as an initial treatment option, start treatment with a dose as low as possible, maintain the pharmacological treatment as short as possible, and check the beneficial and detrimental effects of medication on a regular basis. These protocols should also remind doctors managing older patients to avoid drugs that are considered inappropriate for the elderly.

What this paper adds.

What is already known on this subject

• Psychotropic drugs tend to be overprescribed among elderly patients.

• Patients in this age group are most susceptible to suffering from adverse reactions due to psychotropic medications.

• Thus, it is important to analyse the use of psychotropic drugs in the elderly.

What this study adds

• Almost one third of the patients of our sample were on psychotropic drugs for chronic treatment; the factors best predicting psychotropic use were sex, institutionalisation, number of prescribing physicians, and number of diagnoses.

• Our study also disclosed a high prevalence of patients on potentially inappropriate psychotropic medication.

• There is a need to evaluate psychotropic drug prescribing patterns, in order to minimise the adverse events related to these drugs.