Impacts on practice

  • Careful therapeutic planning is essential for the management of patients with limited life expectancy in the context of original goals of treatment;

  • Clear guidelines are needed on discontinuing inappropriate medication in patients with limited life expectancy; and,

  • Evidence-based approaches to discontinuation of medication for patients with limited life expectancy may contribute to simpler, safer and more appropriate treatment regimens.

Introduction

Patients with limited life expectancy experience significant symptom burden associated with their disease [1]. These patients may require specialist care—including the initiation of pharmacotherapy—delivered by a palliative care team [2]. In addition to this care, patients also routinely use medication to treat long-term conditions or medication to prevent adverse events associated with these conditions. Consequently, the pharmacotherapeutic burden for these patients is high and polypharmacy common [3]. This can be problematic as the time until benefit of preventative medication (typically months to years) can extend beyond patient life expectancy, raising questions over the risk: benefit ratio; patients taking five or more medications per day are more likely to become non-adherent; [4] and, as these patients have changing pharmacodynamic and pharmacokinetic parameters, the risks of developing drug-related toxicities through drug–drug interactions is increased [5, 6]. In view of these issues, the rational use of medication in patients with incurable illness has recently been highlighted and discussed within the medical community [7, 8].

Medication use should, therefore, be evaluated regularly to align with achievable therapeutic outcomes considering limited life expectancy. However, there is a growing body of evidence suggesting inappropriate prescribing for patients with limited life expectancy [9]. Previous work has shown that patients admitted for hospice care were using many inappropriate medications in view of their life limiting diagnosis [10] but, in order to establish the true extent of inappropriate medication use among patients accessing specialist palliative care services, it is essential to include day care patients, as the majority of such services are accessed in this way [11].

In a cohort of day care patients attending a specialist palliative care centre this study seeks to (1) assess the prevalence of inappropriate medication use; (2) identify potential drug–drug interactions using electronic drug interaction software; and, (3) determine how many potential drug interactions could be prevented by discontinuing inappropriate medication.

Methods

Setting

A specialist tertiary care palliative care centre in Northern England serving a population of 330,000. Approximately, 2,000 patients are referred for specialist palliative care each year including inpatient services, home visits from specialist palliative care community nurses, lymphoedema clinics and a day care centre. Of the patients referred to the specialist palliative care centre, around 300 patients access the day care centre per year.

Study design

This was a prospective cohort study of all day care patients attending a specialist palliative care centre in the period September 2012 until January 2013.

Inclusion criteria

Patients were included in the study if they had evidence of taking at least one prescribed medication; the study assessed both medications prescribed acutely (e.g. a course of antibiotics) and chronically (e.g. antihypertensives). The study focused exclusively on medications prescribed by healthcare practitioners. We did not explore the use of over-the-counter medication.

Data collection

As part of the normal provision for day care patients, a paper copy of their electronic medical record is obtained from their GP each time they attend the palliative care centre to ensure the most up-to-date information is used. For each day care patient, demographic details, medical history and medication lists were extracted from the paper copy of the medical record. In terms of medication, we extracted information on the type of medication, formulation, indication, dose, and how long it had been prescribed.

Study outcomes

  1. 1.

    Prevalence of inappropriate medications in day care patients attending a specialist palliative care centre;

  2. 2.

    the number of potential drug–drug interactions; and,

  3. 3.

    the number of potential drug–drug interactions that could be prevented by discontinuing inappropriate medication.

Medication appropriateness

We used a modified Delphi process [12] to reach consensus about medication appropriateness based on the conceptual framework described by Holmes and colleagues for patients with limited life expectancy [13]. Our panel comprised ten clinical pharmacists and five palliative medicine consultants. All panel members were experienced practitioners in palliative care and were either working directly at the specialist palliative care centre or working in centres within the study region. Among those approached to be part of the consensus panel, 7 clinical pharmacists (mean years qualified 15 years; range 5–28) and 3 consultants in palliative medicine (mean years qualified 26 years; range 17–39) agreed to take part. Before the survey was undertaken, the panel received a recently published literature review regarding medication use in patients with limited life expectancy [9] and the Holmes and colleagues conceptual framework to guide their decision-making [13]. According to the framework, the following factors were considered when determining medication appropriateness: patients estimated life expectancy, time until benefit of treatment, goals of care and treatment targets.

Our modified Delphi process comprised of 3 stages:

  1. 1.

    All panel members received a structured questionnaire by e-mail containing a list of medications, alongside the specific indication for each one and were required to rank each one as inappropriate or appropriate (n = 247 medications).

  2. 2.

    Each panel member was then asked to review their choices based upon an anonymous summary of other panel members’ responses.

  3. 3.

    A face-to-face meeting was then held with panel members to discuss the medications ranked as inappropriate from step 2. During the meeting, the patient’s life limiting illness, expected remaining life expectancy, dosage of the medication and co-morbidities were considered to help determine the final decision on whether to rank the medication as inappropriate.

During the process no patient identifiable details were provided to panel members. Consensus for medication inappropriateness was defined according to agreement of seven of the ten respondents in line with previous Delphi processes [14].

Potential drug–drug interactions

For each patient, medications were manually entered into the drug interaction software, Proscript [15], which identified potential drug interactions and categorised them as either clinically significant or not. Clinically significant interactions were then manually sub-classified independently by two clinical pharmacists (AT and AH) as moderately significant or severe, based upon previous literature for categorising drug interactions that considered: if the drug interaction is likely to result in hospitalisation; if it is reversible or irreversible; and, if any treatment would be required to manage the drug interaction [16]. Stockley’s drug interactions [17] and the electronic summary of product characteristics (SPCs) [18] were the data sources used. If agreement was not reached by AT and AH, a third clinical pharmacist (IA) was asked to review the data and make a final decision.

Statistical analysis

This was a descriptive study. We calculated the means, standard deviations, or percentages for outcomes as appropriate using Microsoft Excel. We used forest plots to graphically represent selected outcomes using the metaphor package in the statistical programme, R.

Ethical issues

This study was certified for ethical approval by the research team (HN) in accordance with University of Sunderland Ethics Committee. The research team was advised that National Health Service (NHS) ethics approval was not required and the work was registered with the Trust as a baseline audit. All patient data used in this study were handled and processed in accordance with NHS best practice and Caldicott recommendations.

Results

Cohort characteristics

During the study period, 132 day patients accessed the specialist palliative care centre; all patients met the inclusion criteria. The mean age was 70 years (range 26–94) and 68 (52 %) were male. The most common life limiting illness was cancer (108 patients, 82 %), then, end-stage chronic obstructive pulmonary disease (COPD) (11 patients, 8 %), followed by end-stage congestive heart failure (8 patients, 6 %) and Parkinson’s disease (5 patients, 4 %).

Prevalence of inappropriate medications

The total number of medications prescribed for the cohort was 1,532 (mean per patient, 12; range 1–21); the average number of prescribed medications was 10 (cancer patients), 10 (Parkinson’s Disease patients), 13 (COPD patients) and 14 (end-stage heart failure patients). Of the 1,532 medicines assessed for appropriateness, 238 (16 %) were considered to be inappropriate in the context of limited life expectancy; 92 (70 %) patients were considered to be taking at least one inappropriate medication: 30 patients (22 %) were taking one inappropriate medication; 21 patients (16 %) were taking two; 14 patients (11 %) were taking three; 14 (11 %) patients were taking four; and, 13 patients (10 %) were taking more than four. Statins were the most common ‘inappropriate’ therapeutic group (35 patients, 27 %) followed by mineral supplements (32 patients, 24 %) and aspirin when used for antiplatelet therapy (27 patients, 20 %), Table 1. The average number of ‘inappropriate’ medications ranged from 1 (patients with Parkinson’s disease) and 3 (patients with heart failure), Fig. 1. The most common ‘inappropriate’ medications by disease state were statins in cancer (30 patients, 28 %), calcium supplements in end-stage congestive heart failure (3 patients, 38 %), aspirin in end-stage COPD (5 patients, 45 %) and statins in Parkinson’s disease (2 patients, 40 %).

Table 1 Inappropriate medication identified in day care patients
Full size table
Fig. 1
figure 1

Mean and confidence intervals of inappropriate medications per patient according to life limiting illness

Full size image

Potential drug–drug interactions

The drug interaction software identified 267 potential drug interactions, categorising 112 as clinically significant and 155 as not clinically significant. Among those categorised as significant, 92 were further sub-classified as moderate (Table 2) while 20 were considered severe (Table 3); all severe drug interactions had the potential to result in hospitalisation, irreversible harm or death. At least one potential drug interaction was identified in 85 patients (64 %). The average number of potential drug interactions observed was between 1 (COPD patients) and 4 (heart failure patients), Fig. 2.

Table 2 Moderately significant potential drug interactions identified in hospice day patients
Full size table
Table 3 Severe potential drug interactions identified in hospice day patients
Full size table
Fig. 2
figure 2

Mean and confidence intervals of potential drug interactions according to life limiting illness

Full size image

Drug–drug interactions prevented

Discontinuation of inappropriate medication would reduce the mean number of medications to 10 per patient and prevent 57 interactions considered not clinically significant plus 33 moderate and 12 severe drug interactions. Among patients identified as being subject to a potential drug interaction, discontinuing inappropriate medication would prevent at least one drug interaction in 46 patients, while 21 of these patients would have all potential drug interactions prevented.

Severe drug interactions that could be prevented by discontinuing inappropriate medication include:

  • Simvastatin (>20 mg daily) and amlodipine,

  • Haloperidol and quinine sulphate,

  • Lisinopril and spironolactone,

  • Ibuprofen and aspirin (<300 mg),

  • Clopidogrel and warfarin,

  • Simvastatin and fenofibrate,

  • Candesartan and spironolactone,

  • Digoxin and bendroflumethiazide.

Discussion

Our results show that the pharmacotherapeutic burden is high and polypharmacy is common amongst our cohort of patients accessing specialist palliative day care services. The majority of patients were prescribed at least one inappropriate medication, contributing to potential drug–drug interactions and increasing the risk of patients developing drug-related toxicies. Discontinuation of inappropriate medication would reduce the pharmacotherapeutic burden amongst this patient group, decrease potential drug–drug interactions and, minimise the probability of developing drug-related toxicities.

Our findings build upon our previous work that demonstrated patients with advanced lung cancer take many inappropriate medications—some of which can potentially interact with medication resulting in negative outcomes for patients [19]. Several other studies have also reported inappropriate medication use amongst patients with limited life expectancy [20, 21]. These studies, however, have focused primarily upon patients with advanced cancer; for example, Fede and colleagues accessed the medication history of 87 patients with terminal cancer and concluded that 21 were taking at least one unnecessary medication in view of their life expectancy [21]. Our data suggest that other patient groups with limited life expectancy also use inappropriate medication. Indeed, given the mean number of inappropriate medications and potential drug–drug interactions were highest in patients with end-stage heart failure, a larger study examining appropriate medication use in this patient group is warranted.

Previous work has demonstrated that the prevalence of potential drug–drug interactions is high among cancer patients [16, 22]. Our work supports these findings and shows potential drug–drug interactions are common amongst other patients with life limiting illness. We acknowledge that many of the potential drug–drug interactions identified in this study are based on drug combinations frequently used in the management of long-term conditions (e.g. angiotensin II receptor antagonists and spironolactone in heart failure) or are routinely encountered in palliative care (e.g. a strong opioid and cyclizine) but we believe that, due to the unique and dynamic pharmacokinetic parameters of our patient cohort (e.g. declining renal function), the risk of developing toxicity from drug–drug interactions is heightened. As such, we believe that potential drug–drug interactions should always be considered in the decision making as part of the wider prescribing process—especially in patients with limited life expectancy. The two most commonly identified severe potential drug interactions in our study were simvastatin (>20 mg) and amlodipine; and, digoxin and furosemide. The interaction between simvastatin and amlodipine increases the peak concentration (Cmax) and the area under the curve (AUC) of simvastatin and, consequently, increases the risk of developing myopathy and rhabdomyolysis [23]; recent recommendations suggest to limit the dose of simvastatin to 20 mg/day when co-prescribed with amlodipine [24]. The interaction identified between digoxin and furosemide is, in comparison to the simvastatin and amlodipine interaction, more established. Indeed, furosemide can cause hypokalamia, which, in turn, increases the toxicity of digoxin. The mechanism of action is still being debated but, it is believed that furosemide exacerbates the loss of potassium ions from cardiac cells and, as digoxin inhibits the sodium–potassium ATP-ase in cardiac tissue, it increases the activity of digoxin [25]. This drug interaction is well documented and, consequently, patients are often given a reduced dose of digoxin to account for the effect of furosemide (or indeed any other potassium-depleting diuretic) [26]. However, as patients with limited life expectancy have constantly changing pharmacokinetic and pharmacodynamic parameters—with particular emphasis on declining renal function [27]—it is very difficult to accurately and continually account for these changes when calculating doses for such patients. As digoxin is predominantly excreted unchanged by the renal system, its use should always be closely monitored in patients with limited life expectancy—especially if used in combination with a potassium depleting diuretic such as furosemide.

To minimise inappropriate prescribing and polypharmacy, a number of tools have been developed to assist clinicans in their decision-making [28]. For example, the Beers criteria [29], the Medication Appropriateness Index [30] and the Screening Tool of Older Persons’ Potentially Inappropriate (STOPP) criteria [31] are all used in clinical practice to identify inappropriate medication, with a view to minimising polypharmacy. One limitation of these criteria is that they focus entirely on elderly patients and not necessarily those who have a limited life expectancy. This is problematic for several reasons; firstly, not every patient with limited life expectancy is elderly; this was observed in our study with several patients <65 years old (the youngest patient in our cohort was 26 years old). Secondly, many medications commonly used in a palliative care setting to treat acute symptoms associated with the life limiting illness are considered inappropriate according to these criteria e.g. lorazepam, frequently used to treat anxiety and breathlessness, is, according to the Beers criteria, inappropriate. In view of these limitations, Holmes and colleagues have developed a conceptual framework that is specific to patients with limited life expectancy [13]. This framework was successfully employed in this study but, is highly conceptual and does not necessarily lend itself to application within a busy clinical environment. Further guidance is thus required to assist prescribers with their decision-making for reviewing the medications of patients with limited life expectancy.

Within our cohort of patients, statins were the most commonly prescribed medications considered inappropriate. Statins are indicated for primary and secondary prevention of cardio- and cerebrovascular events and are used extensively throughout the world. Indeed, their efficacy in reducing cardiovascular events and mortality after an acute coronary syndrome, as well as the reduction of major cardiovascular events in people with established risk factors is well documented [32–34]. The time until benefit of the statins is variable depending on type and dose, but ranges from 6 months to 2 years for prevention of cardiovascular events and approximately 2–3 years for the prevention of cerebrovascular events [35]. Similarly, other common medication identified as inappropriate in this project, such as aspirin and calcium supplements, also have time until benefit of several years [36, 37]. Previous studies have explored statin use in limited life expectancy and have shown that, despite having questionable clinical benefit, they continue to be prescribed [38–40]. For example, Pearson and colleagues explored statin use among cancer patients and showed that more than 30 % of patients who died were dispensed statins within 30 days of death—adding unnecessary therapeutic burden to patients [40]; our results appear to support these data.

The reasons for the high use of inappropriate medications among patients with limited life expectancy are unclear and there is a dearth of studies exploring the qualitative aspects of these challenges. One plausable explanation is that there are no clear guidelines available for reviewing and discontinuing medications in this group. It is not clear who should instigate a medication review or where and when is the most appropriate setting to undertake it. Recent work suggests that GPs would welcome training in shared care decision-making in relation to discontinuing inappropriate medication for elderly patients [41]. Interestingly, the same study showed that GPs perceive stopping preventative medication as being more difficult when compared with medication used to treat symptomatic conditions. One small-scale study demonstrated that hospice patients do not object to having medications discontinued provided the reasons for doing so are properly explained [10]. It is possible that the difficulties perceived by healthcare professionals in regard of speaking to patients with limited life expectancy may act as a barrier toward discontinuing medication. A robust qualitative study exploring patient, carer and prescriber experiences of medication use in limited life expectancy is warranted; this may identify challenges associated with medication review and discontinuation.

Limitations

While we believe our work is robust and has important implications in the review and discontinuation of inappropriate medication in patients with limited life expectancy, we acknowledge that a limitation of this work is that the majority of our patient cohort were cancer patients, with only a minority having other life-limiting illnesses, such as end-stage heart failure. In addition, only patients from one specialist palliative care centre were accessed. Generalisation of this work to other centres in the UK and more widely should, therefore, be made carefully. We also acknowledge that the drug interaction software used throughout the study has not been validated in the literature. We do, however, believe Proscript is robust in terms of predicting the sensitivity and specificity of potential drug–drug interactions, as it is routinely used in clinical practice throughout the UK; all predicted drug–drug interactions were also independently checked with two experienced clinical pharmacists.

Conclusions

Patients who access specialist palliative day care services take many inappropriate medication for the treatment or prevention of long-term conditions. These medications not only increase the pharmacotherapeutic burden for the patient but they also contribute to potential drug–drug interactions, which can increase the risk of patients developing drug-related toxicies. These patients should have their medication reviewed in the context of their original therapeutic goals.