The concept of appropriate prescribing

Many words and terms have been coined to describe standards that should be achieved in prescribing, including ‘safe’, ‘rational’, ‘optimal’ and ‘appropriate.’

The concept of ‘appropriate prescribing’, and its early definition, dates from the early 1970s.Reference Parish¹² Over time its usage has developed, especially in relation to prescribing for older people, where there is evidence of greater heterogeneity than observed in the rest of the population.Reference Somel, Khaitovich, Bahn, Paabo and Lachmann^13, Reference Nelson and Dannefer¹⁴ According to one definition of appropriate prescribing, patients would ‘gain maximum benefit from medication’ but not ‘suffer unnecessarily from illness caused by excessive, inappropriate or inadequate consumption of medicines’.¹⁵ Therefore, conversely, a prescription where risks outweigh the potential health benefits could be considered inappropriate. It has also been suggested that the term ‘appropriate prescribing’ should be expanded to cover misuse, overuse and underuse of treatments.Reference Jackson, Mangoni and Batty¹⁶

It is also important to consider whether we are discussing appropriateness from a patient-benefit perspective or from a criteria-based perspective. Sometimes what may seem, at first sight, to be inappropriate may well be the most appropriate and acceptable option for a patient, and what may be perceived as appropriate by a patient may not be by the patient's physician. The 2012 Beers Criteria (discussed in detail in the Explicit Prescribing Indicators section) update goes some way to addressing the difficulty in shaping criteria that allow more individualized clinical care with the creation of a new third category of medications that should be used with caution.¹⁷ They recognize that plausible reasons exist why the use of particular drugs in certain individuals may be appropriate, although substantial potential for misuse or harm exists, meriting an extra level of caution.

We must question whether parameters such as safety, benefit and economics should always override the patient's wishes and beliefs. This is illustrated in an account by Barber, who describes the experience of a terminally ill man who requested a particular brand of hypnotic, which was the only one he had found to be effective.Reference Barber¹⁸ The drug the patient wanted was on a list of medications that, although licensed, are not generally prescribable on the NHS and the prevailing opinion was that temazepam was as effective, cheaper and readily available. Barber was the chief pharmacist and was asked for an opinion by a clinician. The request was refused although Barber said that it ‘felt’ wrong.

Some years later Barber published the anecdote with the assertion that his original judgement was wrong and that patient preference should, on occasion, prevail. Despite these complexities, tools have been designed to identify inappropriate prescribing, such as the Beers Criteria already mentioned, and these will be discussed in more depth later in this article. However, the authors of the 2012 Beers criteria update explicitly state that criteria should not be applied punitively or be viewed as superseding clinical judgement or an individual patient's values, recognizing that prescribing decisions are not always clear cut, and that physicians must consider multiple factors.¹⁷ It is important to recognize that the clinician–patient relationship has, over time, become more of a partnership and the concept of ‘shared decision-making’ (SDM) has become increasingly important. A systematic review of 11 randomized controlled trials (RCTs) of SDM concluded that it is particularly useful for long-term decisions, especially in the context of chronic illness, and can be an effective method of reaching treatment agreement.Reference Joosten, DeFuentes-Merillas, de Weert, Sensky, van der Staak and de Jong¹⁹

Prescribing difficulties

Over a century ago the physician and pioneer Ignatz Leo Nascher introduced the term ‘geriatrics’ into medicine, proposed that a specialist approach was required when prescribing for older people, and warned of the need for caution when using drugs in combination.Reference Nascher²⁰ Because of the difficulties and complexity associated with balancing safety and quality of prescribing with appropriate treatment of all co-morbidities in older people, a more specialized approach is needed. Unfortunately, it is not generally feasible for a geriatrician to assess all older patients. Therefore, the development of systems and indicators to guide individual prescribers is of paramount importance and we will come on to discuss developments in these areas.

Lack of evidence from clinical research trials

The best clinical evidence for treatment of medical disorders comes from the results of good quality RCTs. Unfortunately, there has been gross under-representation of the older population in clinical trials. There is a general paucity of high-quality studies in elderly patients. For example, although 60% of patients suffering from cancer are above 65 years of age, they make up only 22–36% of participants in trials for cancer treatment.Reference Lewis, Kilgore and Goldman^21, Reference Hutchins, Unger, Crowley, Coltman and Albain²² Commonly reported barriers to enrolment relate to the opportunity to participate in trials due to protocol design and exclusion criteria based on co-morbid conditions or chronological age.Reference Ford, Howerton and Lai²³ Although older men and women develop complications from myocardial infarction more commonly than the younger population,Reference Rich, Bosner, Chung, Shen and McKenzie²⁴ 40% of major randomized trials of coronary artery disease between 1991 and 2000 had explicit exclusions on the basis of age.Reference Lee, Alexander, Hammill, Pasquali and Peterson²⁵ Although there have been advances in the treatment of heart failure, fewer than one-third of older patients suffering from the condition met the enrolment criteria for heart failure trials.Reference Masoudi, Havranek and Wolfe²⁶ Despite these examples, evidence from phase II cancer trials suggests that older patients can be recruited successfully and are tolerant of treatments.Reference Giovanazzi-Bannon, Rademaker, Lai and Benson²⁷ This is of ever-increasing importance given the growing burden of disease within the older population. With reference to the future of cancer care, it is anticipated that, in the United States between 2010 and 2030, there will be a 67% increase in cancer incidence for older adults, compared with an 11% increase for younger adults. The percentage of cancers diagnosed in older adults will increase from 61 to 70% during this period.Reference Smith, Smith, Hurria, Hortobagyi and Buchholz²⁸

As a result of the dearth of RCT evidence, most groups developing ‘potentially inappropriate medication’ (PIM) lists for older patients have had to use a Delphi process to reach consensus agreement. Furthermore, evidence has also commonly come from observational studies with no interventional component, making it difficult to establish true ADE causality.

It is clear that there is a need for well-designed pragmatic RCTs to establish whether there is clinical benefit from the application of screening tools for PIMs.

Some progress has been made, although extensive further work is required to redress the imbalance of representation in clinical trials. In the United States, recognition of the under-representation of over-65-year-olds led to a change in Medicare policy, in 2000, to include coverage of routine patient care costs of clinical trials. This had a positive impact on older patient enrolment to cancer clinical trials, with an increase from 31% between 1997 and 2000, to 38% between 2001 and 2003.Reference Unger, Coltman and Crowley²⁹ With regard to thrombolysis treatment for acute ischaemic stroke, the recent publication of the Third International Stroke Trial (IST-3) has gone some way to addressing the previous under-representation of older adults in all previous trials. Prior to IST-3 fewer than 80 patients over the age of 80 had been recruited to trials of intravenous thrombolysis. In IST-3, 1617 (53%) of patients recruited were older than 80 years of age.³⁰

There are signs of a growing collective will to address age discrimination in clinical trials. As part of an American Geriatric Society Meeting, European Union and American Geriatric Societies met with regulatory agencies. The conclusion of a roundtable discussion was that geriatric societies and regulatory agencies must work together to propose new definitions, study designs and technologies aimed at improving evaluation of drugs in older people with multiple co-morbidities and polypharmacy.Reference Cherubini, Del Signore, Ouslander, Semla and Michel³¹

Communication problems and non-compliance

A significant proportion of older people may find it difficult to understand and adhere to complex treatment regimens. Those with dementia or other forms of cognitive impairment may encounter problems following more simple prescriptions and may face considerable barriers to compliance in the context of polypharmacy. It has been estimated that, in the UK, about 50% of older people may fail to adhere to their treatment regimens.^{32, 33}

Factors such as patient preference, complex instructions and unpleasant side-effects may lead to non-adherence.Reference Thomson and Crome³⁴ Large quantities of medicines and lack of knowledge as to why a medicine has been prescribed also contribute to non-compliance.Reference Kairuz, Bye and Birdsall³⁵ Patients prescribed drugs with narrow therapeutic indices such as insulin, warfarin and non-steroidal anti-inflammatory drugs (NSAIDs) are frequently ignorant about their side-effects.Reference Burns, Austin and Bax³⁶ To further complicate matters, older patients commonly present with falls, immobility, delirium, incontinence or other geriatric syndromes. All of these may be drug-related events, adding complexity to prescribing decisions, with the risk of a drug cascade developing where a drug is used to treat the adverse effect of another, if it is not identified correctly.

Multi-compartment compliance aids are widely used, despite a limited evidence base. Although some patients see aids as maintaining independence, others see them as casting doubt on their independence.Reference Nunney, Raynor, Knapp and Closs³⁷ Aids are often issued without discussion, which can be interpreted as paternalistic. Views of older people must be considered and respected, and clear communication is vital. The patient role in medical decision-making is often absent from research, and it has been argued that the traditional concept of compliance is outmoded in a modern healthcare system with questioning patients with chronic illnesses.Reference Donovan³⁸

Inappropriate prescribing

As explained above ‘inappropriate prescribing’ is a blanket term and includes unregulated polypharmacy, under-prescribing, the prescription of medications that have more potential risk than benefit and poor prescribing practice leading to the potential of ADEs. The World Health Organisation defined an ADE as ‘a noxious, unintended and undesired effect of a drug, excluding therapeutic failures, intentional or accidental poisoning and drug abuse’.³⁹ Inappropriate medication is linked with worsening physical function and can account for excess health care utilization.Reference Chin, Wang and Jin^40, Reference Lindley, Tully, Paramsothy and Tallis⁴¹

Inappropriate prescription has been reported in about 12–40% of older people residing in community and nursing homesReference Beers, Ouslander and Fingold⁴² assessed by consensus-based explicit criteria.Reference Beers, Ouslander, Rollinger, Reuben, Brooks and Beck⁴³ Despite the attention that has been given to quality of prescription, a recent systematic review found that approximately one in five prescriptions to older patients in primary care was inappropriate.Reference Opondo, Eslami, Visscher, de Rooij and Verheij⁴⁴ A study by Goulding in older ambulatory care patients found that the highest risk of inappropriate prescription came from visits to general and family practitioners, compared with visits to specialists other than internal medicine physicians.Reference Goulding⁴⁵ Other potential risk factors for inappropriate prescribing include female gender, lower socio-economic characteristics, older age, number of health visits, depression, poor health and use of multiple drugs.Reference Goulding^45–Reference Zhan, Sangl and Bierman⁴⁷ One study found that older patients with dementia were more commonly exposed to drug duplication and to drugs acting on the central nervous system such as anti-depressants, anti-psychotic and anxiolytic drugs.Reference Giron, Wang, Bernsten, Thorslund, Winblad and Fastbom⁴⁸ Non-demented patients, however, were more commonly affected by drug–disease interactions, mostly with the use of cardiovascular medications. Common, potentially inappropriate drug treatments include psychoactive medications such as long-acting benzodiazepines and tricyclic anti-depressants, medications which increase the risk of falls such as neuroleptic and vasodilator drugs, NSAIDs and opiates and duplicate drug class prescriptions.Reference Gallagher and O'Mahony⁴⁹

Problems and ADEs may also potentially arise from drives to meet generalized clinical practice guidelines. Strict adherence to guidelines developed from evidence in younger populations may have undesirable effects in older patients with multiple co-morbidities. Furthermore, basing standards of care and pay-for-performance on such adherence may lead to inappropriate judgements.Reference Boyd, Darer, Boult, Frier, Boult and Wu⁵⁰

Polypharmacy

Polypharmacy refers to concurrent use of multiple medications by a single patient. It can be appropriate in an older patient who is suffering from a number of medical conditions. However, it may also imply excessive or unnecessary use of medications and the risk of ADEs increases with higher numbers of drugs taken concurrently. Patients prescribed five or fewer medications are significantly less likely to present to hospital with a PIM-related ADE than those taking six or more (odds ratio 0.59).Reference Gallagher and O'Mahony⁴⁹ The average number of medications used by older people rises significantly with age.Reference Kennerfalk, Ruigomez, Wallander, Wilhelmsen and Johansson⁵¹ In the United States, fewer than 18% of patients between the ages of 20 and 59 years take three or more drugs compared with 64% of those over the age of 60, of whom 36% take five or more medications.Reference Gu, Dillon and Burt⁵ The reasons for polypharmacy in older patients are summarized in Table 1.

Table 1.

Reasons for polypharmacy in older people

Almost half of people older than 65 years have three chronic conditions, and 21% have five or more. As a result, different drugs are often given at different points along patients’ lives, by an expanding number of specialists following guidelines relevant to their areas of interest. Therefore, a single disease approach is likely to lead to polypharmacy.Reference Garfinkel and Mangin⁵³ To compound this problem, clinicians often find it much easier to start than stop treatments.

A complex drug regimen increases the risk of non-concordance, resulting in sub-therapeutic benefit from treatment.Reference Rollason and Vogt⁵⁶ Polypharmacy increases the risk of drug–drug interactions (DDI), for example, the combination of an angiotensin-converting enzyme inhibitor and an aldosterone antagonist may precipitate life-threatening hyperkalaemia. Other types of drug interactions that may occur include drug–herbal products, drug–alcohol and interaction with illicit substances used by older people.Reference Zarba, Storr and Wagner⁵⁸ Polypharmacy significantly increases the cost of health care and utilization of health care resources.Reference Rollason and Vogt⁵⁶ Polypharmacy correlates with an increased risk of hip fracture in older people, with an odds ratio of 8.42 for patients taking 10 or more drugs.Reference Lai, Liao, Liao, Muo, Liu and Sung⁵⁹ The risks of ADRs are also increased by polypharmacy, with associated increased morbidity and mortality.Reference Hajjar, Cafiero and Hanlon⁶⁰

Underprescribing

Underprescribing or under-utilization of medication occurs when a drug, which has a clear beneficial role in the treatment of a disease, is omitted without appropriate reasons. Inappropriate prescribing encompasses acts of omission as well as commission. Most of the published literature deals with the latter, although underprescribing is increasingly recognized as an important problem. Using the validated START (screening tool to alert doctors to the right treatment) list, one study found one or more prescribing omission in 57.9% of older patients admitted to hospital from the community with acute illness.Reference Barry, Gallagher, Ryan and O'Mahony⁶¹ The probability of an omission was found to be significantly related to age, with those between 65 and 84 years having an approximately 55% likelihood of a medication omission, rising to 72.2% in those aged over 85 years. The most common prescribing omissions were statins in atherosclerotic disease (26%), warfarin in chronic atrial fibrillation (9.5%), anti-platelets in arterial disease (7.3%) and calcium and vitamin D supplementation in symptomatic osteoporosis (6%). Cardiovascular disease is the commonest cause of morbidity and mortality in older adults.Reference Wei, Hazzard, Beirman and Blass⁶² However, other studies have also supported the above findings and found significant underprescribing for cardiovascular disease, including undertreatment with beta-blockers after myocardial infarction.Reference Soumerai, McLaughlin, Spiegelman, Hertzmark, Thibault and Goldman⁶³ Similarly, only 33–75% of patients suffering from heart failure were treated with ACE inhibitors.Reference Chin, Wang, Zhang and Lang^64–Reference Mendelson and Aronow⁶⁶ This undertreatment of common medical problems of old age such as ischaemic heart disease and hypertension is likely to increase the incidence of preventable strokes, myocardial infarctions and associated consequences.

Possible reasons for omission of potentially beneficial drugs include lack of knowledge of evidence-based secondary preventative therapies, low therapeutic expectation in frail elderly patients, multimorbidity, desire to avoid polypharmacy, fear of ADEs, economic problems, focus on palliation of symptoms and ageism.Reference Barry, Gallagher, Ryan and O'Mahony^61, Reference Cherubini, Corsonello and Lattanzio⁶⁷ In some cases there may be rational and appropriate reasons not to prescribe, such as limited evidence for drug therapies in patients aged over 80 years, lack of a favourable risk–benefit ratio, limited life expectancy or patient preference.

In one study, the estimated probability of underprescribing increased significantly with the number of drugs a patient was prescribed. Amongst patients taking five drugs or more, there was a 42.9% chance of being undertreated compared with 13.5% in those taking four drugs or fewer.Reference Kuijpers, van Marum, Egberts and Jansen⁶⁸ This suggests that efforts to avoid polypharmacy may contribute to underprescribing.

Poor prescribing practice

The Royal College of Physicians of London (RCP) has cited a number of physician-related factors that are causes of sub-optimal prescribing in older people.⁶⁹

These include:

• • Inadequate clinical assessment leading to incorrect diagnosis

• • Failure to record current medications, including over-the-counter medications

• • Failure to monitor response to treatment

• • Failure to document previous adverse drug reactions

• • Excessive prescribing

• • Inappropriate prescribing

• • Failure to review repeat medication

• • Failure to take account of altered pharmacokinetics and pharmacodynamics

To reduce or prevent the causes of inappropriate prescribing a variety of approaches are required, including education, production of guidelines, patient and carer involvement and methods of monitoring and auditing the prescription of medicines.

Tools for appropriate prescribing

Indicators have been developed to evaluate prescribing performance with a view to identifying inappropriate prescribing and improving practice. Many different approaches have been adopted. These include the development of lists of medications, dosages or drug–drug combinations (explicit) known to cause harmful effects in older people. An alternative approach (implicit) uses expert clinical judgement based upon individual patients and the individual prescribing situation. Significant developments have occurred over recent years, which we will discuss in the following sections.

Explicit prescribing indicators

Several groups have published lists of explicit criteria for defining and detecting PIM use in older people. Table 2 includes a number of the most frequently cited. Most have been developed via a consensus approach, commonly using the Delphi process, often based on initial evidence drawn from literature reviews. Explicit prescribing indicators are therefore dependent not only upon available evidence but also the ability to routinely include new evidence.Reference Tully and Cantrill⁷⁹ The ideal features of explicit prescribing criteria are described in Table 3.

Table 2.

Comparison of criteria for appropriateness of prescribing

Table 3.

Features of ideal, explicit prescribing criteria

The most widely known are the Beer's criteria, developed in the United States. Initially, a list of medications to be avoided in nursing home residents was formulated.Reference Beers, Ouslander, Rollinger, Reuben, Brooks and Beck⁴³ This list was later revised to include more drugs and categories for ambulatory people aged over 65 years from any background.Reference Beers⁷⁰ The most recent revision of this list was published in 2012 by the American Geriatric Society.¹⁷ The development process involved systematic review and grading of the evidence on ADEs with an interdisciplinary team of 11 experts in geriatric care and pharmacotherapy applying a modified Delphi method to reach consensus. The final updated criteria included 53 medications or medication classes. These were divided into three categories of PIMs and drug classes: (i) avoid in older patients, (ii) avoid in older patients with certain diseases and syndromes, and (iii) use with caution in older adults. The authors estimate that as many as 40% of older adults receive one or more medications on this list.

Limitations of the Beer's list include that it does not address potential PIMs that are not unique to ageing, including drug–drug interactions, dosing of primarily renally excreted medications and therapeutic duplication. A significant issue and source of debate since the publication of the first Beer's criteria over two decades ago is the lack of a consistent, statistically significant, association between PIMs and the incidence of ADEs.Reference Onder, Landi, Liperoti, Fialova, Gambassi and Bernabei^80–Reference Budnitz, Shehab, Kegler and Richards⁸² Most studies have had significant limitations, including retrospective design. Furthermore, there is little evidence to support a clear benefit from the application of Beer's criteria in terms of clinical outcomes.Reference Spinewine, Swine and Dhillon⁸³

Validated criteria have also been developed in Canada by a 32-member national consensus panel, originally published by McLeod et al.Reference McLeod, Huang, Tamblyn and Gayton⁷⁵ These were applied to a series of hospital in-patients, and the potentially inappropriate prescribing practices detected were used to develop the Improving Prescribing in the Elderly Tool (IPET), consisting of a list of 18 drugs that should generally be avoided in older people, 16 drug–disease interactions and four drug–drug combinations.Reference Naugler, Brymer, Stolee and Arcese⁷⁶ The authors reported excellent inter-rater reliability of the IPET, with a kappa co-efficient of 1.0.

As a result of perceived deficiencies of Beer's criteria, other groups have also sought to construct and validate alternative criteria with more relevant PIM lists. Notably, the Screening Tool of Older Person's Prescriptions (STOPP) was developed using a Delphi consensus technique with an 18-member expert panel from centres in Ireland and the UK.Reference Gallagher, Ryan, Byrne, Kennedy and O'Mahony⁷² Important differences from Beer's criteria are that STOPP includes drugs currently in widespread use in Europe, several PIMs not in Beer's criteria and it addresses drug–drug interactions and duplicate drug class prescribing.

In a study of 715 acutely ill elderly patients, STOPP identified PIMs in 35% of patients compared with Beer's criteria identifying them in 25%. Furthermore, STOPP PIMs contributed to 11.5% of admissions compared with only 6% of Beer's criteria PIMs.Reference Gallagher and O'Mahony⁴⁹ Another prospective study of 600 patients aged 65 years or older detected 329 ADEs in 26.3% of patients. Of these, 66.6% where deemed to be causal or contributory to admission, of which 68.9% were avoidable or potentially avoidable as defined by Hallas criteria. Serious avoidable ADE detection increased when STOPP PIMs were prescribed (odds ratio 1.85) compared with Beer's criteria PIMs. Definitely or possibly avoidable ADEs relevant to admission were 2.76 times more frequently detected with STOPP criteria PIMs.Reference Hamilton, Gallagher, Ryan, Byrne and O'Mahony⁷³ This evidence suggests that STOPP criteria are more sensitive to PIMs that result in ADEs, but RCT evidence is needed to prove their value as an interventional tool.

There has been debate about whether indicators are generalizable across countries. Many drugs included in Beer's criteria are infrequently used in Western Europe, and prescribing practices vary significantly between countries. As a result, a number of PIM lists have been developed over recent years. A French consensus panel used a two-round Delphi method to create a list of 36 criteria applicable to people aged 75 years or older, with 29 medications or medication classes to be avoided in all patients and five medications to avoid in specific medical conditions.Reference Laroche, Charmes and Merle⁸⁴ More recently, the PRISCUS list of 83 PIMs was generated by expert consensus by a German group, also utilizing a two-round Delphi process.Reference Holt, Schmiedl and Thurmann⁷⁴ Importantly, therapeutic alternatives are suggested and, for cases where the administration of a PIM is clinically necessary, recommendations for precautions, including monitoring and dose adaptations, are made.

A key limitation in the development of PIM lists and the widespread use of the Delphi process is the lack of methodologically high-quality pharmacological studies and clinical trials in older patients. Due to the paucity of evidence to guide evaluation of PIMs and potential therapeutic alternatives, the use of expert consensus and the Delphi method, despite their subjectivity, can be seen as an acceptable way to develop criteria.

The Assessing Care of Vulnerable Elders (ACOVE) project created a set of indicators designed to measure the quality of care of vulnerable adults in the USA.Reference Wenger and Shekelle⁷⁸ These indicators go further than identifying a list of medications or classes of drugs to be avoided, since they include indicators for appropriate medication.Reference Knight and Avorn⁸⁵ A review of these indicators was undertaken in 2007 using the original methodology, and is set out in Table 4. In ACOVE-3 several indicators were added so that, overall, it incorporates 392 quality indicators, comprehensively covering domains of care including screening and prevention, diagnosis, treatment, follow-up and continuity, which can form the basis of interventions to improve care.Reference Wenger, Roth and Shekelle⁸⁶

Table 4.

Quality indicators for medication use (ACOVE)

Implicit prescribing indicators

Implicit indicators allow clinicians to use information from the patient and published literature to make judgements about the appropriateness of medications. They may allow important factors such as co-morbid diseases and patient preference to be taken into account, but the trade-off is that they are more time consuming and dependent on the knowledge of the clinician. They are based on more generic quality criteria than explicit indicators that are applied using expert clinical judgement.

The Medication Appropriateness Index (MAI) is a measure of appropriate prescribing developed in the USA. It is not drug specific, and focuses on individual patients by using ten criteria.Reference Hanlon, Schmader and Samsa⁷⁷ As an implicit indicator, the MAI requires clinical judgement; however, it also incorporates more explicit instructions that help standardize the process, resulting in good inter-rater reliability and an overall kappa of 0.83. There is evidence to support the reliability of the MAI as an instrument to evaluate medication appropriateness in ambulatory elderly patients and by community pharmacies.Reference Fitzgerald, Hanlon and Shelton^87, Reference Kassam, Martin and Farris⁸⁸ A similar system has been developed by a UK group that produced the Prescribing Appropriateness Indicators.Reference Cantrill, Sibbald and Buetow⁸⁹Table 5 compares the indicators used in both of these tools.

Table 5.

Criteria used in the Medication Appropriateness Index and the Prescribing Appropriateness Indicators

The Good Palliative-Geriatric Practice Algorithm for drug discontinuation, developed in Israel, has been shown to be effective in reducing polypharmacy as well as in reducing morbidity, including admission to acute care facilities, and mortality in elderly nursing home patients.Reference Garfinkel, Zur-Gil and Ben-Israel⁹⁰ The group demonstrated a statistically significant difference in 1-year mortality, with a rate of 21% in the study group compared with 45% in the control group.

A feasibility study of the application of the Good Palliative-Geriatric Practice algorithm to a cohort of community-dwelling older patients found that the protocol recommended discontinuation of 58% of prescribed drugs.Reference Garfinkel and Mangin⁹¹ Successful discontinuation was achieved in 81%, including non-consent and failures. Of discontinued drugs, only 2% were restarted because of the recurrence of the original indication, and no significant adverse events or deaths were attributable to discontinuation. On a modified Likert scale, 88% of patients reported a global improvement in health and there was some evidence of improvements of cognitive impairment.

It has been suggested that the use of both explicit and implicit criteria together may be more advantageous than either one alone.Reference Lindley, Tully, Paramsothy and Tallis^41, Reference Buetow, Sibbald, Cantrill and Halliwell⁹² In 1998, the National Health Service in the UK initiated the National Sentinel Clinical Audit, which consisted of projects of national relevance within the health service. One of these was to measure the prescribing performance of health care providers nationally by using prescribing indicators. The indicators were developed to identify and improve suboptimal prescribing in older people and fall into three groups: purely descriptive, indicators of potentially harmful or unnecessary prescribing, and evidence-based indicators.Reference Oborne, Batty, Maskrey, Swift and Jackson⁹³ The audit analysed combinations of prescribed drugs, collection of clinical data for corroboration of prescriptions for commonly overprescribed drugs and shortfall of prescriptions for commonly underprescribed drugs. The initial investigation of the use of these indicators demonstrated the ability to improve certain clinical services in relation to prescribing and in a variety of settings.Reference Grant, Batty and Aggarwal⁹⁴ A disadvantage is that the indicators are dependent on the quality of clinical data and are limited by the standard of documentation in case notes.

Appropriate prescribing towards the end of life

Older patients reaching the end of life are significant consumers of health care and are at increased risk of polypharmacy and ADEs.Reference O'Mahony and O'Connor^95, Reference Maddison, Fisher and Johnston⁹⁶ Unlike the terminal phase of illness, where pharmacotherapy is relatively well defined, there is a much weaker evidence base for the treatment of patients who have not yet reached the terminal phase but have a limited life expectancy of less than 1 year. Frail, older people are usually excluded from randomized controlled trials; therefore trying to apply such evidence to end-of-life treatment may be of limited benefit to the patient and potentially of detriment.

It is of key importance that clinicians recognize patients entering the end-of-life stage and, at this point, carefully review and define treatment goals. A considerable proportion of patients with a known terminal illness continue to take preventative medications for chronic disease until death, despite questionable benefit.Reference Maddison, Fisher and Johnston⁹⁶ As a general rule, drugs for primary prevention, for example statins, are not appropriate in the treatment of end-of-life patients. The same is true for secondary prevention if the time to therapeutic benefit exceeds life expectancy. The concept of ‘oligopharmacy’, the deliberate avoidance of polypharmacy, has been proposed as a key issue in appropriate prescribing in end-of-life care.Reference O'Mahony and O'Connor⁹⁵ Algorithms have been created to help guide deprescribing and a systematic approach to the discontinuation of medications has been shown to be feasible and to reduce morbidity and mortality.Reference Garfinkel, Zur-Gil and Ben-Israel^90, Reference Garfinkel and Mangin^91, Reference Hardy and Hilmer⁹⁷

When people enter the terminal stage and death approaches, the total number of medications often increases. A study of 260 patients in the palliative phase of their illness found that, as death approached, the mean total number of medications increased by 2.5.Reference Currow, Stevenson, Abernethy, Plummer and Shelby-James⁹⁸ This increase in prescribing is likely to be largely appropriate due to the increased use of symptoms-specific medications. The prompt availability of medications to manage symptoms is a key aspect of high-quality end-of-life care and it is generally thought that anticipatory prescribing is good practice in the terminal phase of illness. Challenges are encountered by primary and community health professionals related to anticipatory prescribing, including resourcing concerns, professional expertise and experience, as well as professional relationships with patients and other professionals.Reference Faull, Windridge, Ockleford and Hudson⁹⁹ It is important that such issues are addressed to avoid underprescribing and inadequate symptom control.

Other approaches to enhancing appropriate prescribing

Conflicts of interest

None.