Potentially inappropriate prescribing in an Irish elderly population in primary care


Dr Stephen Byrne, Senior Lecturer in Clinical Pharmacy, School of Pharmacy, University College Cork, Ireland.
Tel: +353 (021) 4901658
Fax: +353 (021) 4901656
E-mail: stephen.byrne@ucc.ie



• Potentially inappropriate prescribing in older people is a well-documented problem and has been associated with adverse drug reactions and hospitalization.

• Beers' criteria, Screening Tool of Older Persons' potentially inappropriate Prescriptions (STOPP) and Screening Tool to Alert doctors to Right Treatment (START) are screening tools that have been formulated to help physicians and pharmacists identify potentially inappropriate prescribing and potential prescribing omissions.

• The prevalence of potentially inappropriate prescribing and prescribing omissions in the elderly population presenting to hospital with acute illness is high according to STOPP and START criteria.


• Potential errors of prescribing and of omission of medicines are prevalent among medically stable older people in primary care.

• Screening tools should be incorporated into the everyday practice of primary care doctors and community pharmacists as a means of preventing potential errors of prescribing commission and prescribing omission in older people.

AIMS Screening tools have been formulated to identify potentially inappropriate prescribing (IP) in older people. Beers' criteria are the most widely used but have disadvantages when used in Europe. New IP screening tools called Screening Tool of Older Person's Prescriptions (STOPP) and Screening Tool to Alert doctors to Right Treatment (START) have been developed to identify potential IP and potential prescribing omissions (PPOs). The aim was to measure the prevalence rates of potential IP and PPOs in primary care using Beers' criteria, STOPP and START.

METHODS Case records of 1329 patients ≥65 years old from three general practices in one region of southern Ireland were studied. The mean age ± SD of the patients was 74.9 ± 6.4 years, 60.9% were female. Patients' current diagnoses and prescription medicines were reviewed and the Beers' criteria, STOPP and START tools applied.

RESULTS The total number of medicines prescribed was 6684; median number of medicines per patient was five (range 1–19). Overall, Beers' criteria identified 286 potentially inappropriate prescriptions in 18.3% (243) of patients, whilst the corresponding IP rate identified by STOPP was 21.4% (284), in respect of 346 potentially inappropriate prescriptions. A total of 333 PPOs were identified in 22.7% (302) of patients using the START tool.

CONCLUSION Potentially inappropriate drug prescribing and errors of drug omission are highly prevalent among older people living in the community. Prevention strategies should involve primary care doctors and community pharmacists.


Screening tools to detect potentially inappropriate medicines (PIMs) and potential prescribing omissions (PPOs) have been described previously. The Assessing Care of Vulnerable Elders tool is a set of indicators to measure the quality of care provided to elderly patients and includes a medication review tool [1]. Other tools including The Improving Prescribing in the Elderly Tool [2], The Medication Appropriate Index [3], Beers' criteria [4], and Screening Tool of Older Person's Prescriptions (STOPP) and Screening Tool to Alert doctors to Right Treatment (START) [5] have also been developed to identify PIMs and PPOs in older people. A recent review by Laroche et al. examining the strengths and weakness of inappropriate prescribing (IP) screening tools endorses the use of STOPP and START and highlights important deficiencies of Beers' criteria in the European context [6].

Beers' criteria are the most widely cited older person IP criteria in the literature and are designed to quantify potential IP in older people in primary care, secondary care and nursing homes. They consist of two explicit lists, i.e. those medicines that should be avoided independent of diagnosis (ID), and those medicines that should be avoided considering diagnosis (CD) with each potential IP incident described as ‘high severity’ and ‘low severity’. Beers' criteria were originally formulated in 1991 [7], updated in 1997 [8] and recently revised in 2003 [4]. Using Beers' criteria in primary care, IP rates of 9.8–38.5% have been reported in various European countries [9–14], while IP rates of 21.3–28.8% have been reported in the USA [15–17]. Beers' criteria have a number of serious deficiencies in relation to European prescribing patterns. Principally, they contain several medicines that are either not prescribed or not available in most European countries and highlight drugs not considered potentially inappropriate in older people in most European countries, e.g. doxazosin [18]. In addition, Beers' criteria do not identify PPOs, they lack organization and are cumbersome to use in day-to-day practice. The need for an organized, up-to-date, generalizable set of criteria that considers PPOs has recently been highlighted [19].

STOPP and START were formulated and validated to address the perceived deficiencies of Beers' criteria [5]. STOPP, which is based on physiological systems, contains a list of 65 explicit rules for avoidance of certain drugs/drug classes. START is also system based and lists 22 common instances of PPOs in patients with particular medical conditions. A recent study of elderly hospitalized patients in Ireland, using the 2003 Beers' criteria to determine the prevalence of potential IP, reported that 34% of patients had at least one PIM on admission [20]. A follow-up study using the STOPP criteria found at least one PIM among the regular prescriptions of 35% of acutely hospitalized elderly patients [21]. A PPO rate of 57.9% was identified using the START criteria in a similar hospitalized elderly population [22]. Elderly patients admitted to hospital are generally sicker and frailer than elderly patients reviewed in primary care, and the prevalence rates for IP and PPOs among elderly patients obtained in these Irish hospital studies do not reflect the Irish population as a whole. The present study therefore aimed to determine: (i) the rate of potential IP in primary care using Beers' criteria (Appendices 1,2) and STOPP (Appendix 3); (ii) the rate of PPOs using the START tool (Appendix 4); (iii) the relationship between age and number of prescription drugs and IP; and (iv) the specific areas of prescribing that contribute most to IP and PPOs.


Three large general practices agreed to participate in this study: two urban and one rural, in County Cork in the Munster region of Ireland. All patients were ≥65 years old and were prescribed at least one daily medicine. Patients were excluded if they were nursing home residents or terminally ill, as either of these groups would be more likely to have higher rates of IP [23] and therefore not be representative of independently living community-based elderly patients, the group of particular interest in this study.

Alphabetical lists were compiled from each surgery's electronic database of all patients aged ≥65 years. We focused on patients who had received prescription medicines within the previous 3 months and excluded those who had died since electronic registration (>3 months prior to commencing the study), those not prescribed any regular medicines and those who had not attended the practice in the previous 6 months. Each list was therefore analysed to determine the number of actual patients in each practice aged ≥65 years who had been prescribed at least one regular medicine in the previous 6 months. Patients were then recruited prospectively from the active alphabetical list.

Data collection took place over an 18-month period, between January 2007 and July 2008. Patient data, including medical histories, current diagnoses, current medications and biochemical data, were recorded from a combination of electronic and paper-based records. Data collection was conducted by a research pharmacist (C.R.), supported by an academic consultant physician in Geriatric Medicine, three senior academic pharmacists and the general practitioners in the practices involved. The primary researcher referred to this team of senior academic and clinical staff in the event of uncertainty regarding precise diagnosis, interpretation of clinical and laboratory data and application of the screening tools. Each patient was given a unique identifying number in each surgery to ensure confidentiality and to prevent duplication of data collection.

Patients' lists of diagnoses and medical histories were examined and the Charlson Co-morbidity Index (CCI) was calculated and recorded for each patient. The CCI is a weighted index that takes into account the number and seriousness of comorbid diseases in determining patients' health status [24]. It has been used in various studies to predict patients' long-term outcomes [25] and mortality [26]. It lists 17 clinical conditions that are ranked on a score of 1–6 in terms of the seriousness of comorbid disease. In this study, we used it as a tool to quantify the chronic illness status of the patients and determine the association between the degree of comorbid illness and potential IP.

All patients then had 2003 Beers' criteria, STOPP and START applied to their clinical datasheets. The severity of the potential IP incidents identified by Beers' criteria was determined, based on Beers' criteria classification. All recorded disease states and medical conditions were coded to facilitate data analysis. Disease codes were assigned so that each disease was given a unique number from 1 to 262 and was then grouped according to the principal physiological system affected. We decided not to use the International Classification of Diseases 10 as the diagnoses documented were sometimes insufficiently detailed to meet the classification requirements. Each medicine was assigned a seven-digit code in accordance with the Anatomical Therapeutic Chemical (ATC) Classification System (11th edn, 2008) formulated by the World Health Organization Collaborating Centre for Drug Statistics Methodology [27].

The ATC classification system divides medicines into different groups according to the organ or system affected by their prime mode of action and/or their therapeutic and chemical characteristics, and allows for more efficient and manageable analysis. All data were collated using Microsoft Excel® 2003 and subsequently transferred to SPSS Version 15.0 (SPSS Inc., Chicago, IL, USA) for statistical analysis.

To determine the statistical relationship between the number of medicines prescribed, age, gender, CCI and the occurrence of potential IP, one-tailed biviariate correlations (using Spearman's ρ correlation coefficient) for nonparametric data were calculated. The Wilcoxon signed rank test for nonparametric data was performed to compare the rates of identification of IP using Beers' criteria and STOPP. A probability value of <0.05 was considered significant. Approval for the study was granted by the local research ethics committee.



A total of 1329 patients were recruited into the study. The mean age (± SD) of the patients was 74.9 ± 6.4 years, 60.9% were female. The total number of medicines prescribed was 6684, with a range of 1–19 per patient and a median of five per patient (interquartile range 3–7). The average CCI value for the whole population was 0.67, indicating relative wellness rather than chronic illness (Table 1). As expected, the median number of medicines prescribed increased gradually per age category and decreased for those ≥85 years old (Figure 1).

Table 1. 
Patient demographics
DemographicsTotal (n= 1329)
  • *

    CCI, Charlson Comorbidity Index score.

Male520 (39.1%)
Female809 (60.9%)
Age, mean (years ± SD)74.9 (± 6.4)
Age range (years)65–97
Number of drugs prescribed6,684
Median drug prescriptions per patient (IQR)5 (3–7)
Range of drug prescriptions per patient1–19
Mean CCI* (± SD)0.67 (± 0.93)
Figure 1.

The median number of medicines prescribed per age category (IQR)

Beers' criteria

The application of Beers' criteria to all patient profiles identified 286 PIMs prescribed to a total of 243 (18.3%) patients, with slightly more of the male population taking one or more PIMs (17.9%) than the female population (16.1%). Two hundred and ten (15.8%) patients had one PIM prescribed, and 26 (2.5%) had more than one PIM prescribed (Table 2). Only 19 of the 68 Beers' criteria (27.9%) were used to identify these PIMs (Table 3). One hundred and seventy-seven (61.8%) of the potential IPs identified were of ‘high severity’. Those of ‘low severity’ included: ferrous sulphate at doses >325 mg daily, doxazosin, clonidine and the use of calcium channel blockers or tricyclic antidepressants in patients with chronic constipation.

Table 2. 
Number of patients with potentially inappropriate prescriptions identified by Beers' criteria and STOPP
 Beers' criteriaSTOPP
Number of potentiallyTotal (%)Total (%)
inappropriate prescriptions(n= 1329)(n= 1329)
  1. STOPP, Screening Tool of Older Persons' potentially inappropriate Prescriptions.

1210 (15.8)232 (17.46)
226 (2.0)42 (3.16)
34 (0.3)10 (0.75)
43 (0.2) 
Total243 (18.28)284 (21.37)
Male (n= 520)93 (17.88)102 (19.6)
Female (n= 809)130 (16.07)182 (22.5)
Total potential inappropriate prescriptions286346
Table 3. 
Potentially inappropriate prescriptions identified using Beers' criteria Independent of Diagnosis (ID) and Considering Diagnosis (CD)
  1. NSAIDs, nonsteroidal anti-inflammatory drugs.

 Oxybutynin (unless XL)7
 Short-acting benzodiazepines: (doses >) 
 Lorazepam 3 mg2
 Temazepam 15 mg8
 Triazolam 0.25 mg3
 Long-acting benzodiazepines: 
 Ferrous sulphate >325 mg day−113
 Long-term long half-life NSAIDs: 
 Short-acting nifedipine2
 Total ID212
 Heart failure: high sodium content medicines1
 Peptic ulcer disease: NSAIDs2
 Long-term benzodiazepine55
 Sympatholytic agents
 Chronic obstructive pulmonary disease 
 Long-acting benzodiazepines4
 β-blocker: propranolol3
 Calcium channel blockers7
 Tricyclic antidepressant2
 Total CD74
Total (ID and CD)286

The prescribing of benzodiazepines accounted for 68 (31.9%) of the PIMs identified by Beers' ID criteria. Doxazosin was prescribed on 86 occasions, accounting for 40.6% of PIMs identified by Beers' ID criteria. None of the patients who were prescribed doxazosin received it as first-line therapy. The adjusted potential IP rate, excluding doxazosin, was 12.6%. A total of 74 PIMs were identified using Beers' CD, the highest proportion being benzodiazepines in patients with concurrent depression. This instance accounted for 74.3% of the PIMs identified by Beers' CD.

STOPP criteria

The STOPP criteria identified a total of 346 PIMs prescribed for 284 (21.4%) patients, with slightly more of the female population taking one or more PIM (22.5%) than the male population (19.6%). Two hundred and thirty-two (17.5%) patients had one PIM and 52 (3.9%) had more than one PIM prescribed (Table 2). Of the 65 criteria in STOPP, 28 (43.1%) were used to identify potential IP. The highest prevalence of potential IP (102) was in relation to the gastrointestinal system [in particular, proton pump inhibitors (PPIs)], followed by drugs whose primary effect is on the central nervous system, musculoskeletal system and cardiovascular system. Prescribing of duplicate drug classes accounted for a total of 29 PIMs, the endocrine system accounted for six and only one inappropriate prescription was identified for the respiratory system.

Three hundred and forty-six (79.1%) instances were attributed to the six groups of medicines: PPIs, benzodiazepines, nonsteroidal anti-inflammatory drugs (NSAIDs), β-blockers, tricyclic antidepressants and calcium channel blockers (Table 4).

Table 4. 
Potential inappropriate medicines identified by STOPP
  1. STOPP, Screening Tool of Older Persons' potentially inappropriate Prescriptions; COPD, chronic obstructive pulmonary disease; HF, heart failure; LA, long acting; LT, long term; NIDDM, non-insulin-dependant diabetes mellitus; NSAID, nonsteroidal anti-inflammatory drug; OA, osteoarthritis; PPI, proton pump inhibitor; PUD, peptic ulcer disease; TCA, tricyclic antidepressant.

Cardiovascular System 
Digoxin >125 µg day−12
Loop diuretic first line for hypertension1
Cardioselective β-blocker and COPD22
β-Blocker and verapamil4
Calcium channel blockers and constipation9
Thiazide diuretic and gout6
Aspirin and warfarin without H2 antagonist/PPI4
Aspirin and history of PUD without H2 antagonist/PPI3
Aspirin >150 mg daily4
Aspirin – not indicated3
Central nervous system 
TCA and glaucoma1
TCAs and constipation6
TCAs and opiate or calcium blockers5
TCA and urinary retention1
LT/LA benzodiazepine and with LA metabolites69
>1 week first-generation antihistamines2
Respiratory system 
Oral steroids instead of inhaled steroids for COPD1
Gastrointestinal system 
PPI for PUD at full therapeutic dosage for >8 weeks102
Musculoskeletal system 
NSAID and history of PUD9
NSAID and hypertension39
Long-term continuous NSAID for OA14
NSAID with chronic renal failure1
Long-term NSAID/colchicine for gout – no contraindication to allopurinol2
Endocrine system 
β-Blocker and frequent hypoglycaemic attacks3
Glibenclamide/chlorpropamide and NIDDM3
Duplicate class29
Total potential inappropriate prescriptions346

The potential IP associated with NSAIDs occurred in patients who had moderate to severe hypertension, their long-term use for osteoarthritis, and in patients with a history of peptic ulcer disease, gout, heart failure and chronic renal failure. The potential IP relating to β-blockers occurred in patients with concurrent chronic obstructive pulmonary disease and for diabetic patients reporting frequent hypoglycaemic attacks. Potential IP relating to tricyclic antidepressants was noted in patients with constipation, glaucoma and urinary retention and in patients receiving concurrent opiates or calcium channel blockers.

A significant correlation was found between the number of medicines prescribed and the occurrence of IP when calculated using Beers' criteria (rs= 0.270, P < 0.01) and STOPP (rs= 0.356, P < 0.01) using Spearman's ρ correlation test. There was also a positive correlation between age and the occurrence of IP using Beers' criteria (rs= 0.068, P < 0.01) and STOPP (rs= 0.071, P < 0.01). Similarly, there was a significant correlation between increasing CCI score and potential IP identified by STOPP (rs= 0.210, P < 0.01).

The number of PIMs identified was significantly lower using Beers' criteria than STOPP (Wilcoxon signed ranks test Z =–2.769; P < 0.01).

START criteria

START identified a total of 333 PPOs in 302 (22.7%) patients (Table 5). The incidence of PPOs was significantly higher in women (27.8%) than in men (14.8%) (P < 0.001). The mean age (± SD) of those identified with a PPO was 74.5 ± 6.2 years, while the mean age (± SD) of those without a PPO was 75.1 ± 6.5 years (difference not significant).

Table 5. 
The number of patients identified with a potential prescribing omission (PPO) by START
Number of PPOsTotal (%)
(n= 1329)
  1. START, Screening Tool to Alert doctors to Right Treatment.

1274 (20.62)
225 (1.88)
33 (0.23)
Total patients302 (22.72)
Male77 (14.81)
Female225 (27.81)
Total PPOs333

Fifteen of the 22 criteria (68.2%) in START identified the PPOs in this study. The cardiovascular system accounted for most of the PPOs. Aspirin was the commonest cardiovascular PPO identified. Omissions of calcium and vitamin D supplements were the commonest PPO for the musculoskeletal system. Statins were often omitted from both a cardiovascular and an endocrine perspective. No PPOs were identified under the central nervous system or the gastrointestinal system criteria (Table 6). The relationship between the number of medicines prescribed and prescribing omissions was not significant (rs= 0.016, P= 0.28).

Table 6. 
Details of potential prescribing omissions (PPOs) identified by the START tool
  1. START, Screening Tool to Alert doctors to Right Treatment; AF, atrial fibrillation; ACE, angiotensin converting enzyme; CHF, chronic heart failure; MI, myocardial infarction; COPD, chronic obstructive pulmonary disease; DMARD, disease-modifying antirheumatic drug; RA, rheumatoid arthritis.

Cardiovascular system185
 Warfarin and AF1
 ACE inhibitor and CHF4
 ACE inhibitor and acute MI3
Respiratory system14
 β2 agonist for COPD14
Musculoskeletal system79
 DMARD and RA1
 Ca2+ and Vitamin D3 supplement59
Endocrine system55
 ACE inhibitor4
Total PPOs333


This study indicates that the rate of potential IP in primary care in south-west Ireland is substantial, i.e. 21.4% using STOPP criteria and 18.3% using Beers' criteria. This is the first study using STOPP criteria in primary care, and therefore there are no other studies with which to compare the current STOPP data. The potential IP rate using Beers' criteria lies within the range referred to earlier, 9.8–38.5%, identified in European studies. The reason for the relatively high rate of IP in Polish primary care (28.2%) is unclear; however, over-the-counter medicines accounted for 5.5% of the potential IP identified [10]. A high proportion of the potential IP identified in a Portuguese primary care study, using the Beers' ID tool only, was attributable to ticlopidine (14.9%) [13], a medicine not licensed for use in Ireland. A large number of general practices (n= 131) were involved in the study conducted in the UK, which reported a potential IP rate of 24.8% [14]. However, the rate of potential IP is less in other European centres (9.8% in Turkey and 12.5% in Finland) [9, 11] compared with centres in the USA (21.3%, 23.5%, 28.8%). The reasons for the variation in IP rates are not clear from the literature. Differences in drug availability, prescribing practices and pharmacist routine review of prescriptions are possibilities.

The finding of approximately one-fifth of patients in the present study receiving one or more PIM is an important message for Irish primary care prescribers. We stress that both STOPP and Beers' criteria are designed to identify prescription medicines that are potentially inappropriate for older patients. Nevertheless, prescribing of inappropriate drugs is associated with a significant increase in adverse drug events (ADEs) and iatrogenic morbidity [28, 29]. Although STOPP and Beers' criteria contain comparable numbers of potential IP avoidance rules, in this study 43% of the STOPP potential IP rules were breached compared with 28% of Beers' criteria rules. This, and the recent finding that STOPP showed significantly superior detection of ADEs causing hospitalization vs. Beers' criteria [21], suggests that STOPP may be a more relevant potential IP detection tool in primary care in Ireland than Beers' criteria, as many of Beers' criteria are redundant in the Irish setting [30]. Similarly, in a recent Dutch study it was noted that only 24 of the 78 drugs in Beers' list were available in the Netherlands [12]. Conversely, a number of the medicines in the most recent iteration of Beers' independent of diagnosis criteria are not contraindicated in older people according to the latest edition of the British National Formulary [31], e.g. oxybutinin, amitriptyline, nitrofurantoin, naproxen. Also, doxazosin is commonly prescribed in Ireland for hypertension and is not considered inappropriate when prescribed in combination with other antihypertensives or in cases where other antihypertensive drug classes are contraindicated. Similarly, dipyridamole appears in Beers' criteria. According to recent UK guidelines for secondary stroke prevention, dipyridamole is recommended for secondary prevention of stroke in combination with low-dose aspirin [32] (the combination was shown to be superior to aspirin monotherapy in the ESPS-2 trial) [33]. Accordingly, dipyridamole is widely prescribed for stroke prevention throughout Europe.

Nearly 80% of the potential IP detected by STOPP in this study involved five categories of medicines, i.e. PPIs, long-acting benzodiazepines, NSAIDs, nonselective β-blockers and tricyclic antidepressants. It may be argued that long-term, high-dose PPI treatment in older people is relatively harmless in terms of ADEs and this may be true in practice. However, continuation of high-dose PPI treatment without clear indication is expensive and almost always unnecessary. Furthermore, the surge in PPI prescribing in recent years is a cause of major budgetary concern. In Ireland, annual expenditure on PPIs increased from approximately €8 million in 1995 to €64 million in 2002, accounting for >10% of the total expenditure on drugs funded by the Irish government in 2002 [34]. In the current climate of major fiscal pressure on health resources, the overuse of PPIs becomes more relevant in terms of overall drug expenditure by governments globally. Inappropriate prescription of long-acting benzodiazepines in older patients has been highlighted repeatedly in the literature over the last 25 years, in particular given the link with falls and fracture risk and the difficulties with successful withdrawal [35–37]. Despite this, long-acting benzodiazepines continue to be initiated and repeatedly prescribed for older patients in primary and secondary care in Ireland and other countries [10, 12–14]. These realities suggest that long-acting benzodiazepines should not be initiated in older patients, given their high propensity for psychological and physical dependency.

A serious weakness of IP detection tools to date has been the exclusion of prescribing omissions of clinically indicated drugs. The START list of potential errors of prescribing omission has been formulated specifically for use in tandem with STOPP [5] to give a more complete assessment of potential IP in older people. In the present study, 22.7% of patients had one or more clinically indicated medicines omitted from their regular prescriptions without valid reasons. Seventy-five percent of the START rules applied to patients with PPOs, indicating the high degree of relevance of the START rules. The majority of these omissions involved low-dose aspirin, calcium and vitamin D supplements, statins, angiotensin converting enzyme inhibitors and metformin. The evidence for clear-cut benefit from these drugs in secondary prevention of major morbidity and mortality is well established. These drugs are also generally well tolerated in older people. The reasons for omission of indicated medicines in the present study are unclear. Advanced old age did not influence prescribing omission in this study, i.e. there was no significant difference between patients aged 65–74 years compared with patients aged ≥85 years. The reasons for avoidance of indicated medicines in older patients are unclear from the existing literature. Lack of conviction on the part of some physicians as regards efficacy may be a significant reason for nonprescription of indicated statins in some over-80s patients [38]. A desire to avoid major polypharmacy and complexity of treatment regime as well as predictable poor compliance in certain patients may be relevant also. In our study, there was no significant relationship between the number of medicines (indicating degree of polypharmacy) and occurrence of PPOs, i.e. the fact of taking a large number of prescription medicines did not dissuade prescribers from adding more medicines, when clearly indicated. In contrast, Kuijpers et al. found that polypharmacy was related to under-prescribing. The difference is probably due to the relatively healthy population in the present study [39].

The practical applicability of STOPP and START in daily general practice and community pharmacy is not yet established. The STOPP and START tools may be applicable in the completion of Medication Use Reviews (MURs), which are being increasingly requested within the National Health Service at the present time [40, 41]. Although pharmacist-led MURs have not yet been shown to improve patient clinical outcomes [42], they have demonstrated a potential economic saving [43] and a reduction in prescribing [44], with the majority of recommendations being accepted and implemented by general practitioners [45, 46]. Pharmacists could incorporate the STOPP and START tools into their everyday practice, although the lack of complete patient clinical data may limit the full application of the criteria. STOPP and START could also be adapted to existing prescribing systems in primary care as a means of preventing IP in older people at the point of drug initiation. STOPP and START were easy to use and were time efficient. On average, STOPP criteria were fully deployed within 3 min, START criteria within 1 min. However, there is a short but significant learning curve with STOPP and START criteria until one is fully familiar with them.

There were some limitations to the present study. The comparatively reduced sample size in a restricted geographical area of Ireland limits the generalizability of the findings. The potential IP rate from STOPP criteria may be a conservative estimate, since over-the-counter medicines were not included in the analysis. General practitioners were not given the opportunity to state their reasons for prescribing the identified potential errors of commission or omission due to time constraints. Incomplete documentation of serious falls in patients' case notes may have led to a lower rate of reporting of potential IP in relation to iatrogenically increased falls risk. The clinical and financial benefits from routine application of STOPP and START rules to older people's prescriptions are as yet unknown. Therefore, STOPP and START currently have limited applicability in routine practice. Sufficiently large prospective randomized controlled trials are needed to determine if rigorous application of STOPP and START rules has tangible benefits in terms of reduction of ADEs (e.g. falls), cost, hospitalization and mortality. These questions are currently being studied within our research group.

In conclusion, this study of older people's regular prescriptions in primary care has identified potential errors of prescribing commission and omission in significant proportions of patients. We contend that the majority of these instances are avoidable. Prevention of IP in late life is important for avoidance of predictable ADEs and polypharmacy, as well limiting costs of medication. Presentation of the STOPP/START criteria as a prescribing assistant in a user-friendly electronic automated format to primary care physicians will be necessary for the effective use of the STOPP/START criteria in day-to-day clinical practice. Our research group is currently developing electronic versions of the STOPP/START criteria for this purpose.

Competing interests

None declared.

We wish to thank the primary care physicians who gave us access to their patients' medical records for the purpose of this study. Without their cooperation, this work could not have been completed.


Appendix 1

Beers' criteria considering diagnosis

Considering diagnosis
  1. CNS, central nervous system; COPD, chronic obstructive pulmonary disease; SIADH, syndrome of inappropriate ADH (antidiuretic hormone); GI, gastrointestinal; MAOI, monoamine oxidase inhibitor; NSAIDs, nonsteroidal anti-inflammatory drugs; SSRI, selective serotonin reuptake inhibitor.

Heart failureDisopyramide and high sodium content drugs
HypertensionPhenylpropanolamine hydrochloride, pseudoephedrine, diet pills and amphetamines
Gastric of duodenal ulcersNSAIDs and aspirin
Seizures or epilepsyClozapine, chlorpromazine, thioridazine and thiothixene
Blood clotting disorders or receiving anticoagulant therapyAspirin, NSAIDs, dipyridamole, ticlodipine and clopidogrel
Bladder outflow obstructionAnticholinergics and antihistamines, GI antispasmodic drugs, muscle relaxants, oxybutynin, flavoxate, anticholinergics, antidepressants, decongestants and tolteridine
Stress incontinenceα-Blockers, anticholinergics, tricyclic antidepressants (imipramine hydrochloride, doxepin hydrochloride and amitriptyline hydrochloride) and long-acting benzodiazepines
ArrhythmiasTricyclic antidepressants (imipramine hydrochloride, doxepin hydrochloride and amitriptyline hydrochloride)
InsomniaDecongestants, theophylline, methylphenidate, MAOIs and amphetamines
Parkinson diseaseMetoclopramide, conventional antipsychotics and tacrine
Cognitive impairmentBarbiturates, anticholinergics, antispasmodics, muscle relaxants and CNS stimulants: dextroamphetamine, methylphenidate, methamphetamine and pemolin
DepressionLong-term benzodiazepine use. Sympatholytic agents: methyldopa, reserpine and guanethidine
Anorexia and malnutritionCNS stimulants: dextroamphetamine, methylphenidate, methamphetamine, pemolin and fluoxetine
Syncope or fallsShort to intermediate acting benzodiazepine and tricyclic antidepressants (imipramine hydrochloride, doxepin hydrochloride and amitriptyline hydrochloride)
SIADH/hyponatraemiaSSRIs: fluoxetine, citalopram, fluvoxamine, paroxetine and sertraline
Seizure disorderBupropion
COPDLong-acting benzodiazepines: chlordiazepoxide, chlordiazepoxide-amitriptyline, clidinium-chlordiazepoxide, diazepam, quazepam, halazepam and chlorazepate. β-Blockers: propranolol
ConstipationCalcium channel blockers, anticholinergics, tricyclic antidepressants (imipramine hydrochloride, doxepin hydrochloride and amitriptyline hydrochloride)

Appendix 2

Beers' criteria independent of diagnosis

  1. COX, cyclooxygenase; NSAIDs, nonsteroidal anti-inflammatory drugs.

Propoxyphene and combination productsDiphenhydramine
IndomethacinErgot mesyloids and cyclandelate
PentazocineFerrous sulphate >325 mg
TrimethobenzamideAll barbiturates (except Phenobarbital) except when used to control seizures
Muscle relaxants and antispasmodics: methocarbamol, carisprodol, oxybutynin, chloroxazone, metaxalone and cyclobenzaprine (except extended release oxybutynin)Meperidine
Amitriptyline, chlordiazepoxide-amitriptyline and perphenazine-amitriptylineKetorolac
DoxepinAmphetamines and anorexic agents
MepobramateLong-term use of full dosage, longer half-life, non-COX-selective NSAIDs: naproxen, oxaprozin and piroxicam
Doses of short-acting benzodiazepines: doses greater than lorazepam 3 mg; oxazepam 60 mg; alprazolam 2 mg; temazepam 15 mg; triazolam 0.25 mgDaily fluoxetine
Long-acting benzodiazepinesLong-term use of stimulant laxatives: bisacodyl, cascara sagrada and neoloid except in the presence of opiate analgesic use
Chlordiazepoxide, chlordiazepoxide-amitriptyline, clidinium-chlordiazepoxide, diazepam, quazepam, halazepam and chlorazepateAmiodarone
Digoxin (should not exceed 0.125 mg daily, except when treating atrial arrhythmias)Guanethidine
Short-acting dipyridamole. Do not consider the long-acting dipyridamole except in patients with artificial heart valvesGuanadrel
Methyldopa and methyldopa-hydrochlorothiazideCyclandelate
Reserpine at doses >0.25 mgIsoxsurpine
Gastrointestinal antispasmodic drugs: dicyclomide, hyoscyamine, propantheline, belladonna alkoloids and clinidium-chlordiazepoxideDoxazocin
Anticholinergics and anthistamines: chlorpheniramine, diphenhydramine, hydroxyzine, cyproheptadine, promethazine, tripelennamine and dexchlorpheniramineMethyltestosterone
Short-acting nifedipineClonidine
Mineral oilCimetidine
Ethacrynic acidDessicated thyroid
Amphetamines (excluding methylphenidate hydrochloride and anorexics)Oestrogens only (oral)

Appendix 3

STOPP: Screening Tool of Older People's potentially inappropriate Prescriptions.

The following drug prescriptions are potentially inappropriate in persons aged ≥65 years.

  • A Cardiovascular system
    • 1Digoxin at a long-term dose >125 µg day−1 with impaired renal function*.
    • 2Loop diuretic for dependent ankle oedema only, i.e. no clinical signs of heart failure.
    • 3Loop diuretic as first-line monotherapy for hypertension.
    • 4Thiazide diuretic with a history of gout.
    • 5Noncardioselective β-blocker with chronic obstructive pulmonary disease (COPD).
    • 6β-Blocker in combination with verapamil.
    • 7Use of diltiazem or verapamil with New York Heart Association Class III or IV heart failure.
    • 8Calcium channel blockers with chronic constipation.
    • 9Use of aspirin and warfarin in combination without histamine H2 receptor antagonist (except cimetidine because of interaction with warfarin) or proton pump inhibitor (PPI).
    • 10Dipyridamole as monotherapy for cardiovascular secondary prevention.
    • 11Aspirin with a past history of peptic ulcer disease without histamine H2 receptor antagonist or PPI.
    • 12Aspirin at dose >150 mg day−1.
    • 13Aspirin with no history of coronary, cerebral or peripheral vascular symptoms or occlusive event.
    • 14Aspirin to treat dizziness not clearly attributable to cerebrovascular disease.
    • 15Warfarin for first, uncomplicated deep venous thrombosis for >6 months' duration.
    • 16Warfarin for first uncomplicated pulmonary embolus for >12 months' duration.
    • 17Aspirin, clopidogrel, dipyridamole or warfarin with concurrent bleeding disorder.

*Serum creatinine >150 µmol l−1, or estimated glomerular filtration rate (GFR) <50 ml min−1.

  • B Central nervous system and psychotropic drugs
    • 1Tricyclic antidepressants (TCAs) with dementia.
    • 2TCAs with glaucoma.
    • 3TCAs with cardiac conductive abnormalities.
    • 4TCAs with constipation.
    • 5TCAs with an opiate or calcium channel blocker.
    • 6TCAs with prostatism or prior history of urinary retention.
    • 7Long-term (i.e. >1 month), long-acting benzodiazepines, e.g. chlordiazepoxide, flurazepam, nitrazepam, chlorazepate and benzodiazepines with long-acting metabolites, e.g. diazepam.
    • 8Long-term (i.e. >1 month) neuroleptics as long-term hypnotics.
    • 9Long-term neuroleptics (>1 month) in those with parkinsonism.
    • 10Phenothiazines in patients with epilepsy.
    • 11Anticholinergics to treat extrapyramidal side-effects of neuroleptic medications.
    • 12Selective serotonin re-uptake inhibitors (SSRIs) with a history of clinically significant hyponatraemia.
    • 13Prolonged use (>1 week) of first-generation antihistamines, i.e. diphenydramine, chlorpheniramine, cyclizine, promethazine.
  • C Gastrointestinal system
    • 1Diphenoxylate, loperamide or codeine phosphate for treatment of diarrhoea of unknown cause.
    • 2Diphenoxylate, loperamide or codeine phosphate for treatment of severe infective gastroenteritis, i.e. bloody diarrhoea, high fever or severe systemic toxicity.
    • 3Prochlorperazine (Stemetil) or metoclopramide with parkinsonism.
    • 4PPI for peptic ulcer disease at full therapeutic dosage for >8 weeks.
    • 5Anticholinergic antispasmodic drugs with chronic constipation.
  • D Respiratory system
    • 1Theophylline as monotherapy for COPD.
    • 2Systemic corticosteroids instead of inhaled corticosteroids for maintenance therapy in moderate–severe COPD.
    • 3Nebulized ipratropium with glaucoma.
  • E Musculoskeletal system
    • 1Nonsteroidal anti-inflammatory drug (NSAID) with history of peptic ulcer disease or gastrointestinal bleeding, unless with concurrent histamine H2 receptor antagonist, PPI or misoprostol.
    • 2NSAID with moderate–severe hypertension.
    • 3NSAID with heart failure.
    • 4Long-term use of NSAID (>3 months) for symptom relief of mild osteoarthritis.
    • 5Warfarin and NSAID together.
    • 6NSAID with chronic renal failure*.
    • 7Long-term corticosteroids (>3 months) as monotherapy for rheumatoid arthritis or osteoarthritis.
    • 8Long-term NSAID or colchicine for chronic treatment of gout where there is no contraindication to allopurinol.

*Serum creatinine >150 µmol l−1, or estimated GFR 20–50 ml min−1.

  • F Urogenital system
    • 1Bladder antimuscarinic drugs with dementia.
    • 2Antimuscarinic drugs with chronic glaucoma.
    • 3Antimuscarinic drugs with chronic constipation.
    • 4Antimuscarinic drugs with chronic prostatism.
    • 5α-Blockers in men with frequent incontinence, i.e. one or more episodes of incontinence daily.
    • 6α-Blockers with long-term urinary catheter in situ, i.e. >2 months.
  • G Endocrine system
    • 1Glibenclamide or chlorpropamide with Type 2 diabetes mellitus.
    • 2β-Blockers in those with diabetes mellitus and frequent hypoglycaemic episodes, i.e. ≥1 episode per month.
    • 3Oestrogens with a history of breast cancer or venous thromboembolism.
    • 4Oestrogens without progestogen in patients with intact uterus.
  • H Drugs that adversely affect fallers
    • 1Benzodiazepines.
    • 2Neuroleptic drugs.
    • 3First-generation antihistamines.
    • 4Vasodilator drugs with persistent postural hypotension, i.e. recurrent >20 mmHg drop in systolic blood pressure.
    • 5Long-term opiates in those with recurrent falls.
  • I Analgesic drugs
    • 1Use of long-term powerful opiates, e.g. morphine or fentanyl as first-line therapy for mild–moderate pain.
    • 2Regular opiates for >2 weeks in those with chronic constipation without concurrent use of laxatives.
    • 3Long-term opiates in those with dementia unless indicted for palliative care or management of moderate–severe chronic pain syndrome.
  • J Duplicate drug classes
  • Any duplicate drug class prescription, e.g. two concurrent opiates, NSAIDs, SSRIs, loop diuretics, ACE inhibitors.

Appendix 4

START: Screening Tool to Alert doctors to Right, i.e. appropriate, indicated but often omitted Treatments.

These medications should be considered for people ≥65 years of age with the following conditions, where no contraindication to prescription exists.

  • A Cardiovascular system
    • 1Warfarin in the presence of chronic atrial fibrillation (AF).
    • 2Aspirin in the presence of chronic AF, where warfarin is contraindicated, but not aspirin.
    • 3Aspirin or clopidogrel with a documented history of atherosclerotic coronary, cerebral or peripheral vascular disease in patients with sinus rhythm.
    • 4Antihypertensive therapy where systolic blood pressure consistently >160 mmHg.
    • 5Statin therapy with a documented history of coronary, cerebral or peripheral vascular disease, where the patient's functional status remains independent for activities of daily living and life expectancy is >5 years.
    • 6Angiotensin converting enzyme (ACE) inhibitor with chronic heart failure.
    • 7ACE inhibitor following acute myocardial infarction.
    • 8β-Blocker with chronic stable angina.
  • B Respiratory system
    • 1Regular inhaled β2 agonist or anticholinergic agent for mild to moderate asthma or chronic obstructive pulmonary disease (COPD).
    • 2Regular inhaled corticosteroid for moderate–severe asthma or COPD, where predicted forced expiratory volume in 1 s <50%.
    • 3Home continuous oxygen with documented chronic type 1 respiratory failure (pO2 <8.0 kPa, pCO2 <6.5 kPa) or type 2 respiratory failure (pO2 < 8.0 kPa, pCO2 > 6.5 kPa).
  • C Central nervous system
    • 1L-DOPA in idiopathic Parkinson's disease with definite functional impairment and resultant disability.
    • 2Antidepressant drug in the presence of moderate–severe depressive symptoms lasting at least 3 months.
  • D Gastrointestinal system
    • 1Proton pump inhibitor with severe gastro-oesophageal acid reflux disease or peptic stricture requiring dilation.
    • 2Fibre supplement for chronic, symptomatic diverticular disease with constipation.
  • E Musculoskeletal system
    • 1Disease-modifying antirheumatic drug with active moderate–severe rheumatoid disease lasting >12 weeks.
    • 2Bisphosphonates in patients taking maintenance corticosteroid therapy.
    • 3Calcium and Vitamin D supplement in patients with known osteoporosis (previous fragility fracture, acquired dorsal kyphosis).
  • F Endocrine system
    • 1Metformin with Type 2 diabetes ± metabolic syndrome (in the absence of renal impairment*).
    • 2ACE inhibitor or angiotensin receptor blocker in diabetes with nephropathy, i.e. overt urinalysis proteinuria or microalbuminuria (>30 mg per 24 h) ± serum biochemical renal impairment*.
    • 3Antiplatelet therapy in diabetes mellitus with co-existing major cardiovascular risk factors (hypertension, hypercholesterolaemia, smoking history).
    • 4Statin therapy in diabetes mellitus if coexisting major cardiovascular risk factors present.

*Serum creatinine >150 µmol l−1, or estimated GFR <50 ml min−1.