The prevalence and determinants of polypharmacy at age 69: a British birth cohort study

Polypharmacy is growing in the general population, with over half of those over age 65 in the UK taking more than three prescribed medications [1]. Between 1995 and 2010 the proportion of individuals prescribed five or more medications in a large-scale Scottish cohort study rose from 11.4 to 20.8% of all adults [2]. As individual lifespans increase, more chronic diseases accumulate over time; and a growing catalogue of pharmacological treatments for these diseases results in higher numbers of prescriptions [3]. In the MRC National Survey for Health and Development (NSHD), we previously reported that by their early sixties the majority of men and women have at least two clinical disorders requiring monitoring or treatment [4]; and by their late sixties, one in five have three or more doctor-diagnosed disorders [5]. While an increase in therapeutic options for chronic disease is welcome, often the complex interplay between these medicines is less well accounted for. Adverse drug reactions are estimated to account for around 6.5% of all hospital admissions in the UK, the majority of which are avoidable [6]. The risk of a significant interaction between pharmaceutical therapies rises along with the number of prescribed medications [7]. It is for this reason that polypharmacy has been linked with increased risk of falls, reduced functional status and cognition, and higher all-cause mortality in later life [8‐10].

Prior research has identified that female gender [11, 12], lower socioeconomic position and lower education [13, 14] increases the risk of polypharmacy. These groups are also at significantly increased risk of multi-morbidity, the key driver of polypharmacy [11, 15]. In a large scale analysis of Scottish primary care health records, those in more economically deprived areas had an earlier onset of multi-morbidity, and a higher proportion of chronic obstructive pulmonary disease (COPD), painful disorders and depression [16]. Those on treatments for specific disorders, such as cardiovascular syndromes, are far more likely to be on multiple therapeutic agents [3, 11], perhaps as a result of more established evidence bases for the use of multi-drug treatment regimens in this field. These drugs vary in their side-effects and risk of interaction with other pharmaceutical agents [17]. Cardiovascular medications in particular are most commonly associated with adverse drug reactions in older adults, and this is the most frequently prescribed British National Formulary (BNF) category of medication in the general population [2, 18]. Men, and individuals from lower socioeconomic groups, are more likely to suffer from cardiovascular disorders [4, 19] and consequently cardiovascular polypharmacy [20, 21].

Within studies on its determinants, polypharmacy is commonly defined by an arbitrary cut-off of greater than a set number of total medications [22, 23]. Given both the prominence of cardiovascular medication prescription in older adults, and the increased side effect profile they carry, it is worth distinguishing between cardiological and non-cardiological polypharmacy and investigating the effects of gender and socioeconomic variation, independent of disease burden, on composition of polypharmacy. This understanding of the genesis of cardiovascular polypharmacy might help us better identify at-risk groups who may benefit from targeted interventions to reduce potential harm caused by adverse drug events.

We use the oldest of the British birth cohort studies to: a) describe medication use, general polypharmacy, cardiological polypharmacy and non-cardiological polypharmacy at age 69; b) assess the change in medication use across the seventh decade of life; and c) investigate how gender, education and adult socioeconomic position are associated with these types of polypharmacy. We hypothesised that being more educationally or socioeconomically disadvantaged would be associated with polypharmacy even after taking account of disease burden.

The MRC National Survey for Health and Development (NSHD), has followed 5362 individuals (2547 women) since their birth in England, Scotland & Wales in a single week of March 1946, so far to age 70 [5, 24]. The most recent data collection was conducted between 2014 and 2015, when study members were aged 68–69 years. Prior assessment of study member responses confirmed the NSHD as representative of this population at the age 60–64 [25]; since then, additional losses to follow-up other than death have remained very low [5]. At 68–69 years, following a postal questionnaire, study members still alive and with a known current address in mainland Britain (n = 2698) were invited to have a home visit at age 69; 2149 (79.7%) visits were completed. Invitations were not sent to those who had died (n = 995), who were living abroad (n = 583), who restrict participation to postal questionnaires (n = 22), had previously withdrawn from the study (n = 632) or had been lost to follow-up (n = 432) [5].

There were 2122 (98.7% of participants, of whom 51% were female) with known medication data at age 69 (Table 1). Almost a third (31.2%) had no educational qualifications although almost three quarters had been in non-manual occupations. A fifth (20%) reported three or more doctor diagnosed diseases and another fifth (20.3%) reported two.

At age 69, total polypharmacy was present in over a fifth of individuals in a nationally representative British cohort, and over half of this sample received at least one cardiovascular prescription. The prevalence of medication use increased with age across the seventh decade, with the greatest increase seen in the volume of cardiovascular prescriptions, particularly in those prescribed between two to four cardiological medications. Disease burden was a major predictor of all types of polypharmacy. Those with higher education were less likely to be prescribed polypharmacy, even when taking this burden into account, along with gender and social class. The associations with education were stronger than those with social class, and all types of polypharmacy were mediated by level of education. Gender differences were more varied; no differences in total polypharmacy were apparent at age 69, yet women were less likely than men to have cardiovascular prescriptions, and more likely to have non-cardiovascular medications.

The primary strengths of this study are the nature of the NSHD being representative of the general population, and having age homogeneity. The measures available allowed accurate capture of a potentially at-risk group for polypharmacy and patterns of sociodemographic and health related factors and disease burden on pharmaceutical prescriptions to be estimated. In addition, the longitudinal data allow for description of change in prescribed medications. One limitation is that medication data relied on self-reports, albeit collected by research nurses, and making use of prescription lists. Evidence suggests that our method should provide robust medication data, as self-reported measures alone correlate well with pharmacy prescription records; identifying most regular medications and only occasionally missing ‘as required’ and non-oral medications, such as transdermal patches [28]. The accuracy of self-reported diagnosed disease varies dependent on condition severity [29], though most major diseases have fairly high accuracy using this measure. Prior work on NSHD showed self-reported diabetes diagnoses were comparable with GP records in 95% of cases [30]. Additionally, capping our measure at 3+ diseases reduces the potential for variance at higher levels of disease burden. Similarly, our measure of disease severity, defined as whether long-term illness limits activity, is likely to detect most severe diseases within the sample population – although it may be influenced by participant perception of expected function. No data were available to account for the primary general practitioner’s (GPs) prescribing habits, which holds significant associations with major polypharmacy [31]. Given the geographical spread of study members across England, Scotland and Wales, a high variation in GP prescribing patterns is likely to exist, but this is unlikely to alter results. Also of note, data used here were collected at two time points, over 5 years apart. Participants may have taken numerous prescriptions of a shorter duration between these time points. This is suggested, for example, by the low numbers of anti-infective prescriptions among the cohort. However, given that it is more important to understand chronicity of medication use in polypharmacy, this is unlikely to affect our findings.

The overall prevalence of polypharmacy within the study sample correlates well with that of a large scale health records study in Scotland [2], with similar proportions of individuals taking over five medications, and a high prevalence of cardiological medications. The increase in the number of medications taken by participants between the ages of 60–64 and 69 supports prior evidence suggesting that medication use increases with age [2, 11, 32]; of note here is the considerable increase in cardiovascular prescriptions revealed by this study. Epidemiological evidence from a Swedish cohort study suggested that women were 50% more likely to be medicated at any age, though the prevalence of polypharmacy (over five medications) was roughly equivalent for both genders by age seventy [11]. A large increase in male polypharmacy between the ages of 60 and 69 has previously been noted in large population databases, and appears to have been replicated in NSHD [33]. Total polypharmacy in this study follows this trend; however, gender appears to play a role in the composition of that polypharmacy, with men prescribed more cardiovascular drugs than women, in keeping with prior studies on cardiological polypharmacy [20, 21], even after additionally adjusting for disease burden in the current study.

Our findings for multi-morbidity, represented here by the number of diagnosed diseases, supports prior evidence that this is the major driver of polypharmacy [3, 11, 15], although again, this appears to have a greater impact on the number of cardiovascular rather than non-cardiovascular medications. Disease severity however was more strongly associated with non-cardiological polypharmacy, possibly accounting for the greater breadth of non-cardiological diseases in the general population leading to more variation in their severity; with more severe diseases requiring more medication to treat. Low levels of education have previously been highlighted as a predictor of polypharmacy [13, 32, 34] but only some studies have controlled for multi-morbidity [14, 34]. In our study, these associations persisted even when adjusting for disease burden, and were found to be more pronounced for cardiovascular medications than for non-cardiovascular medications. Cardiovascular disorders are strongly associated with negative health behaviours, such as smoking and reduced exercise [35], and these negative health behaviours are in turn associated with lower levels of education and social class [36, 37]. The resulting disorders such as myocardial infarction, ischaemic heart disease and heart failure, commonly warrant multiple pharmacological agents for treatment [27]; leading to a high volume of cardiovascular polypharmacy in this group. Yet this may not entirely explain this phenomenon, as we found that these associations with education also existed for non-cardiological polypharmacy, and persisted when adjusting for disease severity.

The reasons for the inverse associations between education and polypharmacy are therefore potentially multifaceted, and could indicate an additional disparity in interactions with healthcare services. Individuals with higher levels of education may be more likely to view encounters with medical professionals as a shared responsibility, and to question treatment options [38, 39]. Therefore less educated individuals may accept or be offered pharmacological treatment more readily than those who have spent time exploring side-effect profiles and alternative therapies. Indeed, prior analysis of GP data suggests that more money is spent on prescriptions for individuals of lower socioeconomic position [40]. Further research on consultation practice and prescriptions would be warranted to explore this trend.

When defining polypharmacy by quantity of drugs alone, key differences in the composition of the medications involved are neglected. Men and those with lower levels of education take more cardiovascular drugs, which have a higher risk of potentially adverse drug-drug interactions, and may have later life consequences such as increased mortality or reduced functional outcomes. Future research on polypharmacy outcomes should aim to account for variations in both the prevalence and composition of polypharmacy by gender and education as well as disease burden. While further work is required to unpick these associations, they highlight the potential for targeted medication reviews in high risk populations.