Prevalence and determinants of polypharmacy in Switzerland: data from the CoLaus study

The Colaus study (www.​colaus.​ch) is an ongoing prospective survey investigating the biological and genetic determinants of cardiovascular disease in the population of Lausanne, Switzerland. Detailed descriptions of the study design have been reported elsewhere [25]. A simple, non-stratified random sample of 19,830 subjects (corresponding to 35% of the source population) was drawn [25]. Inclusion criteria were: (a) written informed consent and (b) willingness to take part in the examination and to provide blood samples.

The baseline study was conducted between 2003 and 2006 and included 6733 participants, with a participation rate of 41%; the first follow-up was conducted between April 2009 and September 2012, five and a half year on average after the baseline and included 5064 participants (75.2%) [26].

The baseline evaluation included an interview, a physical exam, blood sampling and a set of questionnaires. All participants were interviewed by trained recruiters regarding personal and family history of cardiovascular disease and risk factors and medicines taken. The questionnaires can be obtained from the authors upon request. The same procedure (questionnaires, interview and physical examination) was applied at follow-up. The data from the first follow-up was used in this study.

This study was conducted according to the guidelines laid down in the Declaration of Helsinki and all procedures involving human subjects were approved by the Institutional Ethics Committee of the University of Lausanne (decision reference 33/09). Written informed consent was obtained from all participants.

CVD and medication status were assessed by questionnaire. Smoking status was defined as never, former (irrespective of the time since quitting) and current (irrespective of the amount smoked). Educational level was categorized as low (obligatory school or apprenticeship), medium (high school), or high (university degree). Marital status was dichotomized into living alone (single, divorced, widowed) and living with somebody (married or partnership). Country of birth was categorized into Switzerland and other [25]. Body weight and height were measured with participants standing without shoes in light indoor clothes. Body weight was measured in kilograms to the nearest 100 g using a Seca® scale, which was calibrated regularly. Height was measured to the nearest 5 mm using a Seca® height gauge [25]. Body mass index (BMI) was defined as weight/height2. Overweight was defined as 25 ≤ BMI < 30 kg/m2 and obesity as BMI ≥ 30 kg/m2^. Venous blood samples (50 ml) were drawn after an over-night fast, and most clinical chemistry assays were performed by the CHUV Clinical Laboratory on fresh blood samples. Measurements included blood lipids, liver markers, cytokines and adipokines.

Participants were asked to bring all their medicines, which were checked by the research assistants. Participants were asked if the medicines were prescribed by a doctor or obtained over the counter. The regular consumption of the medicines over the last six months was also queried. Posology was not taken into account, i.e. a patient with propranolol 3 × 40 mg a day was considered as taking a single drug.

Polypharmacy was defined as the regular use of five or more different pharmacologically active medicines, regardless if a medicine contained one or more components [21, 27]. Excessive polypharmacy was defined as the regular use of ten or more medicines [6]. Regular use was defined as a medicine taken regularly over the past six months. Only medicines considered as medically needed (i.e. prescribed by a doctor) were considered; hence, we excluded medicines obtained over the counter, alternative therapies such as plant extracts, dietary supplements and homeopathy.

Participants were excluded if they missed any information regarding individual, clinical or socio-economic data.

Statistical analyses were performed using Stata version 14.1 (Stata Corp, College Station, Texas, USA). Descriptive results were expressed as mean ± standard deviation (SD) for continuous variables or as number of participants (percentage) for categorical variables. For prevalences, exact 95% confidence intervals were also computed. Bivariate analysis was performed using Student’s t-test for continuous variables and chi-square test for categorical variables. Multivariate analysis was performed using logistic regression and results were expressed as multivariate-adjusted odds ratio (OR) and [95% confidence interval (CI)]. Statistical significance was considered for a two-sided test with p < 0.05. Two sensitivity analyses were performed: the first one included over the counter (OTC) drugs, and the second considered all active substances among prescribed and OTC drugs. As some drugs combine several pharmacologically different active substances, we identified the ATC codes corresponding to combinations of different active substances (Additional file 1 : Table S1), excluding combinations of vitamins and minerals (ATC codes A11A; A11C; A11D; A11E; A11G¸ A11J; A12AX; B03AD and B03AE). Categories were defined similarly to polypharmacy, i.e. 0–4, ≥5 and ≥10 active substances.

Of the 5064 participants at follow-up, 126 (2.5%) were excluded due to missing data for socio-economic characteristics or body mass index, leaving 4938 participants (97.5%) for the current analysis. The characteristics of the included and excluded participants are summarized in Additional file 1 : Table S2. Excluded participants lived more frequently alone, while no differences were found for the other individual and socio-economic characteristics.

Prevalence of any drug use, polypharmacy and excessive polypharmacy are summarized in Table 1. Almost six out of ten (59.7%) participants reported taking at least one drug; one out of nine (11.8%) was with polypharmacy, and slightly over 1 % (1.4%) was on excessive polypharmacy (Table 1). Cardiovascular drugs were the most frequent, being prescribed to over one third (37.3%) of the participants; psychiatric drugs ranked second highest, being prescribed to one sixth (15.8%) of the participants. Among cardiovascular drugs, the two most frequently prescribed categories were antihypertensive and hypolipidemic drugs (Table 1).

The distribution of the number of drugs consumed according to age groups is indicated in Fig. 1; the older the participant, the more drugs he/she consumed, with a considerable increase in the prevalence of polypharmacy.

The bivariate and multivariate analyses of the individual and socioeconomic determinants of polypharmacy are summarized in Table 2. Participants with polypharmacy were older, had a higher BMI, were less well educated, and were more frequently former smokers. These findings were further confirmed by multivariate analysis including all determinants simultaneously (Table 2). Conversely, no association was found between polypharmacy and gender, marital status or country of birth (Table 2). Similar findings were obtained for participants reporting excessive polypharmacy, although the association with smoking was no longer significant (Additional file 1 : Table S3).

When OTC drugs were considered, the prevalence rates of polypharmacy and excessive polypharmacy were 14.7% and 1.8%, respectively. The results of the multivariate analysis are provided in Additional file 1 : Table S4 and Additional file 1 : Table S5. Similar associations were found as for the original analysis; further, men had a lower likelihood of being on polypharmacy (Additional file 1 : Table S4), while no gender differences were found for excessive polypharmacy (Additional file 1 : Table S5).

When all active substances were considered (including active substances in OTC drugs), the prevalence rate of taking ≥5 active substances/day was 16.9% and of taking ≥10 active substances/day was 2.3%. The results of the multivariate analysis using active substances are provided in Additional file 1 : Table S6 and Additional file 1 : Table S7. Results were similar to the previous sensitivity analysis; further, living in a couple was associated with a lower likelihood of taking ≥10 active substances/day (Additional file 1 : Table S7).

The overall prevalence of polypharmacy was 11.8%, and it increased considerably with age, from 2.9% for age group 40–49 to 25.5% for age group 65–81. Comparison with the literature is difficult as there are differences on how data were collected, in the age groups and, definitions of polypharmacy [29]. In Switzerland, a study based on claims from the largest health insurance reported a prevalence of polypharmacy of 16.7% among adults, and of 41.2% in individuals aged ≥65 years [12]. The prevalence was based on health claims, and only subjects who asked for reimbursement of at least one drug were included. The other Swiss study was based on patients followed in a university primary care setting and reported prevalence rates ranging between 20.8% for age group 50–54 and 54.6% for age group 75–80 [28]. A Scottish study based on electronic data from pharmacy claims reported an overall prevalence of polypharmacy of 22.1% in individuals aged ≥20 years [3], while an Irish study using the same methodology reported a prevalence of 21.5% in individuals aged ≥20 years; 30.2% for age group [45–64] and 60.4% in individuals aged ≥65 years [11]. A possible explanation for these higher values is that individuals consuming “occasional” drugs such as antibiotics or anti-histaminics were also included. Similarly, in another Scottish study, the prevalence of polypharmacy was 36% for age group [60–70]; this higher prevalence could be due to a different definition of polypharmacy, i.e. ≥4 instead of ≥5 drugs [13]. Overall, our results indicate that prevalence of polypharmacy is common in the Swiss mid-aged population, and that almost one quarter of individuals aged 65–81 years are with polypharmacy.

Increased age was associated with increased polypharmacy and excessive polypharmacy, a finding consistently reported in the literature [1, 2, 11‐13]. This reflects the increase in the number of pathologies requiring therapy with ageing, or the difficulty to stop treatment once it has been initiated, leading to cumulative prescriptions. Still, the fact that four out of ten participants aged 65–81 were with polypharmacy stresses the need for the optimization of the prescriptions, such as using for example of the START/STOPP criteria [30], the adoption of a patient-centered rather than a disease-centered approach, and an effective physician-patient communication [31].

Increased body mass index was associated with increased polypharmacy and excessive polypharmacy rates, a finding also in agreement with the literature [4, 32]. A likely explanation is the wide array of comorbidities associated with obesity, namely diabetes, hypertension, dyslipidemia and arthrosis, which frequently require multiple treatments [33].

Participants with a lower education level presented higher rates of polypharmacy. Such association has already been reported, but mostly among elderly subjects [6, 10, 16]. An explanation is that subjects with a lower education level tend to present higher multimorbidity rates [8, 9], possibly due to adverse socio-economic conditions or to less interest for preventive measures. Still, our results suggest that even among middle aged subjects, a low educational level leads to increased rates of polypharmacy.

Smoking was associated with a higher prevalence of polypharmacy, a finding in accordance with another study [4]. Possible explanations include the higher prevalence of pulmonary, cancer and psychiatric disease among smokers, leading to the use of more drugs. Unfortunately, due the scope of the CoLaus study, detailed information regarding those diseases is not available and it would be of interest that this analysis be replicated.

Cardiovascular drugs were the most prescribed therapeutic class, a finding in agreement with the literature [2, 11, 34]. Antihypertensives ranked first, closely followed by statins. Possible explanations include the relatively high prevalence of cardiovascular risk factors in this population [35, 36] and the existence of guidelines regarding cardiovascular risk factor management [37‐39]. Conversely, the high prevalence of psychiatric drugs (namely anxiolytics, hypnotics and sedatives) and of analgesics raises some concerns, as these drugs are not supposed to be prescribed on a long term basis.

Considering only the number of drugs prescribed might underestimate the prevalence of polypharmacy, as subjects can acquire other products over the counter. Considering also non-prescribed drugs increased the prevalence of polypharmacy from 11.8% to 14.7%, but had no effect on its determinants.

Further, as drugs might contain combinations of active substances, simply counting the number of drugs (including OTC) might still underestimate the prevalence rate of polypharmacy. Indeed, when all active substances were considered, the prevalence of participants taking ≥5 active substances was 5.1% higher than the rate based on drugs (0.9% higher if at least 10 active substances). Again, no significant changes were found regarding the main determinants of taking ≥5 active substances. Overall, our results suggest that the definition of polypharmacy (only prescribed drugs, prescribed + OTC or active substances) considerably impacts the prevalence rates but does not influence significantly its determinants. Studies assessing both polypharmacy and active substances in the general population are scarce and it would be of interest that our results be replicated in other settings.

With the ageing of the population, the number of individuals with polypharmacy will increase. This will require general practitioners, hospital medical staff and pharmacists to be increasingly attentive to such a condition, in order to prevent over-prescription and the occurrence of drug-drug interactions and adverse drug reactions. In the forthcoming years, drug prescriptors and dispensers will be required to optimize prescriptions, a difficult task where the pros of adding an extra drug and the cons related to its possible adverse effects will have to be carefully balanced. Hence, it would be of interest that strategies aimed at optimizing polypharmacy are provided to health professionals either at the pre or at the postgraduate level. For instance, in Switzerland, there are initiatives aimed at reducing inappropriate polypharmacy that provide medication reconciliation to ambulatory patients [40]. Indeed, such strategies have been shown to reduce polypharmacy and its potential drug-drug interactions [41] and to be cost-saving [42]. Similarly, the use of medicines combining several drugs may tackle the treatment burden and improve adherence [43].

This study has several strengths. Firstly, it was based on a large mid-aged population-based sample, allowing to estimate the prevalence of polypharmacy at the population level including non-prescribed (OTC) medicines, a condition that studies based on pharmacy claims cannot perform in Switzerland [3, 12]. Secondly, it clearly assessed drugs taken on a regular basis, precluding a possible overestimation bias due to the occasional consumption of drugs. Finally, and contrary to most studies [1‐3, 6, 10‐12], several definitions for polypharmacy were applied, allowing a wider comparison with the existing literature [22].

This study has also some limitations. First, participation rate was low (41%), but in line with other epidemiological studies [44]. Thus, a recruitment bias cannot be excluded, the healthiest participants being selected, which would underestimate the prevalence rates of polypharmacy and excessive polypharmacy. Still, the distribution of age groups in our sample was comparable to the source population and there was no difference in gender distribution between the source population and the CoLaus participants (not shown). Further, our results provide a conservative estimate for the prevalence rates of polypharmacy and excessive polypharmacy, and the fact that 25.6% of participants aged 65–81 were with polypharmacy (and 2.7% on excessive polypharmacy) is already concerning. Secondly, it was not possible to assess all comorbidities in our participants; hence, it was not possible to assess if (excessive) polypharmacy was due to increased number of comorbidities. Third, due to legal constraints, it was not possible to cross-check the information provided by the participants with data from medical or pharmaceutical electronic records. No specific training regarding drug collection was provided to the research assistants. Complete medication reconciliation or tools such as the Swiss polymedication check or the brown bags’ method are difficult to apply in large samples due to economic and human resources issues [40, 45, 46]. However, recent studies suggest that self-reported information on medication use closely relates with pharmacy records [47]; hence, memory bias might be small and the impact on prevalence rates might be reduced.