Introduction
Medication safety in primary care is a growing global concern. Studies reveal a concerning statistic: 3-4% of unplanned hospital admissions stem from preventable drug-related issues. A significant portion of these incidents are linked to errors in prescribing and monitoring medications. This article delves into the critical area of High Risk Medication Use Diagnosis within primary care settings. We define high-risk prescribing as the prescription of medications by healthcare professionals that carry a significant risk of patient harm. These medications necessitate careful avoidance or, when unavoidable, rigorous monitoring and regular review to ensure continued appropriateness. While the precise figures vary depending on the assessment tools utilized, research consistently demonstrates the widespread nature of high-risk prescribing in primary care. Encouragingly, evidence suggests that targeted interventions can mitigate these risks. Strategies such as clinical decision support systems, performance feedback mechanisms, and pharmacist-led initiatives have shown promise in improving prescribing practices. However, concrete evidence of enhanced patient outcomes remains an area requiring further investigation. The increasing adoption of electronic medical records (EMRs) in primary care provides a unique opportunity to integrate diverse strategies for bolstering medication safety and fostering collaboration among healthcare stakeholders. This review aims to explore the spectrum of high-risk medication use in primary care, examine methods for its high risk medication use diagnosis, and summarize existing prevalence research. Building upon established interventions for improving professional practices, we propose a systematic framework to enhance medication safety in primary care and pinpoint key areas for future research endeavors.
Defining High-Risk Prescribing: A Closer Look at Diagnosis
While medication licensing aims to ensure a favorable risk-benefit profile for drugs, the reality is that medications frequently cause harm across healthcare settings, much of which is preventable. In primary care, the scale of this issue is underscored by systematic reviews estimating that 3–4% of all unplanned hospital admissions are due to preventable drug-related morbidity [Howard et al. 2006]. Hospital admissions, however, represent only the tip of the iceberg. Many preventable adverse drug events (pADEs) are managed within primary care itself. One review estimates nearly 7 pADEs annually per 100 outpatients, with only a fraction requiring hospitalization [Thomsen et al. 2007].
The medication use process in primary care is a complex, multi-stage process involving prescribing, dispensing, administration, and monitoring. This process is shared among a multidisciplinary team including healthcare professionals, informal caregivers, and patients [Hepler and Segal, 2003]. While systematic reviews attribute pADEs in primary care roughly equally to prescribing errors, patient nonadherence, and monitoring failures [Howard et al. 2006; Thomsen et al. 2007], this article specifically focuses on prescribing and monitoring, as these aspects are more directly influenced by healthcare professionals. We aim to define the spectrum of high-risk medication use in primary care, examine approaches to its high risk medication use diagnosis, and summarize prevalence research. Based on existing interventions to modify professional practice, we will propose a systematic approach to improve medication safety in primary care and highlight areas for future research.
Defining a pADE, the United States Institute of Medicine describes it as ‘any preventable injury due to medication’ [Bates et al. 1995]. Medication errors, the root causes of pADEs, are defined as ‘failures in the treatment process that lead to, or have the potential to lead to, harm to the patient’, encompassing deficiencies in both prescribing and monitoring [Aronson, 2009].
Prescribing medication is inherently risky and intricate. Cribb and Barber define appropriate prescribing as ‘a balance between the right technical properties, what patients want and the greater good,’ highlighting potential conflicts in prescribing rationales [Cribb and Barber, 1997]. Labeling prescribing as ‘inappropriate’ or ‘erroneous’ based solely on rule violations can be overly simplistic [Hepler and Segal, 2003]. For instance, prescribing a nonsteroidal anti-inflammatory drug (NSAID) to a patient on warfarin is high-risk, but occasionally necessary. A prescriber treating a rheumatoid arthritis patient recently anticoagulated for thromboembolic disease might find this coprescription the least detrimental option. We define high-risk prescribing as medication prescription by professionals where significant patient harm is evidenced. Such prescriptions should be avoided or, if essential, diligently monitored and regularly reviewed for continued necessity. Therefore, high risk medication use diagnosis is not just about identifying the medication, but also assessing the context of its use and the patient’s specific vulnerabilities.
The Wide Spectrum of High-Risk Prescribing in Primary Care: Diagnosis Scenarios
A systematic review of drugs most frequently implicated in preventable hospital admissions revealed that four drug classes contribute to roughly 50% of these hospitalizations: antiplatelet drugs, NSAIDs, diuretics, and anticoagulants. An additional 21% were linked to other cardiovascular agents (beta blockers, ACE inhibitors, angiotensin receptor blockers, cardiac glycosides), opioid analgesics, and antidiabetic agents [Howard et al. 2006]. Thus, the majority of preventable harm in hospital admissions arises not from drugs that should be generally avoided, but from commonly used therapeutic agents with strong indications in primary care (NSAIDs as simple analgesics being a notable exception).
Table 1 summarizes common high-risk prescribing patterns contributing to preventable hospital admissions [Howard et al. 2003; Thomsen et al. 2007]. It highlights that preventable harm often results from failure to recognize or adequately address patients’ needs during vulnerable periods. Generally, preventable hospital admissions can be categorized (though overlap may occur): (1) continuing unnecessary or no longer indicated drugs, (2) failing to use indicated drugs to prevent adverse drug reactions, (3) using drugs or dosages that interact negatively with existing medical conditions, (4) using drugs or dosages that interact with current drug therapy, and (5) inconsistent monitoring. Effective high risk medication use diagnosis must consider these diverse scenarios to comprehensively address patient safety.
Table 1. Reported scenarios of high-risk use of drugs most frequently implicated in preventable drug related hospital admissions [Howard et al. 2003; Thomsen et al. 2007].
| Drug class | High-risk prescribing | Preventable adverse drug event |
|---|---|---|
| NSAID/ antiplatelets/ oral anticoagulants | Prescription in patient with GI risk factors (without GI protection) | GI toxicity, haemorrhage, anaemia |
| Coprescription of NSAIDs and antithrombotics (without GI protection) | ||
| NSAID | Overdosing due to prescription of two full-dose NSAIDs | Acute renal failure |
| Opioid analgesic | Prescription without laxative | Constipation |
| ACE inhibitor/ ARB | Prescription in patient with aortic stenosis | Pulmonary oedema |
| Diuretics | Coprescription of thiazide and loop diuretic (without valid indication) | Hyponatraemia, dehydration, hypotension, renal failure |
| Overdosing due to lack of monitoring of fluid balance, renal function, electrolytes, etc. | ||
| ACE inhibitor/ARB/ diuretics | Coprescription of potassium sparing diuretic (without valid indication) | Hyperkalaemia |
| Beta blocker/ Calcium antagonists | Coprescription of verapamil (without valid indication) | Congestive cardiac failure |
| Oral anticoagulants | Overdosing due to lack of INR monitoring in patient known to be hard to control or following introduction of an antibiotic | Haemorrhage/anaemia |
| Antidiabetics | Overdosing due to lack of GFR monitoring in patient taking sulphonylurea | Hypoglycaemia |
| Overdosing due to lack of blood glucose monitoring following introduction of prednisolone | Hypoglycaemia | |
| Overdosing due to no dose reduction when hypoglycaemia noted | Hypoglycaemia | |
| Digoxin | Overdosing due to lack of GFR/digoxin level monitoring | Digoxin toxicity |
| Beta blockers | Overdosing due to treatment initiation at full dose in patient with congestive heart failure | Congestive cardiac failure |
| Sudden cessation without down-titration of dose | Tachycardia |
ACE, angiotensin converting enzyme; ARB, angiotensin receptor blocker; GFR, glomerular filtration rate; GI, gastrointestinal; NSAID, nonsteroidal anti-inflammatory drug.
Measurement and Prevalence of High-Risk Medication Use: Diagnosis Approaches
Medication safety assessment generally employs ‘implicit’ or ‘explicit’ approaches. ‘Implicit’ methods provide assessors flexibility to consider the clinical context of prescribing and monitoring decisions. In contrast, ‘explicit’ methods evaluate medication use against predefined rules. For individual patients, explicit ‘assessment criteria’ provide a binary answer on the presence of high-risk prescribing. At the population level, ‘prescribing indicators’ can measure the percentage of vulnerable patients exposed. Box 1 illustrates these distinctions using NSAID prescribing as an example, demonstrating different methods for high risk medication use diagnosis.
Box 1. Illustration of the differences between implicit and explicit methods to assess medication use.
| Method of measurement | Measurement output |
|---|---|
| Implicit, e.g. MAI [Hanlon et al. 1992] | |
| Is the use of an NSAID appropriate in this patient with respect to contraindications, drug–drug interactions etc? | Measure of appropriateness (e.g. on a three-point scale) |
| Explicit | |
| Medication Assessment Criteria, e.g. Beers criteria [Fick et al. 2003], STOPP criteria [Gallagher et al. 2008] | |
| If a patient has a history of peptic ulcer, is he prescribed an NSAID without gastroprotection? | |
| Medication Safety Indicators, e.g. Scottish indicators [Guthrie et al. 2011], PPRNet indicators [Wessell et al. 2010] | |
| Denominator: number of patients with a history of peptic ulcer (= ‘vulnerable’)/ Numerator: number of ‘vulnerable’ patients who are prescribed an NSAID without gastroprotection | % of vulnerable patients with high-risk prescribing |
MAI, Medication Appropriateness Index; NSAID, nonsteroidal anti-inflammatory drug; PPRNet, Practice Partner Research Network; STOPP, Screening Tool of Older Persons potentially inappropriate Prescriptions.
Implicit Measurement: MAI
The Medication Appropriateness Index (MAI) is a prime example of an implicit approach. It assesses each prescribed drug on a three-point scale (appropriate, marginally appropriate, inappropriate) across 10 domains: indication, effectiveness, dosage, directions, drug–drug interactions, drug-disease interactions, expense, practicality, duplication, and duration [Hanlon et al. 1992]. Implicit methods are advantageous for their broad assessment of therapeutic issues and prescribing appropriateness. However, they heavily rely on reviewer expertise and are time-consuming [Hanlon et al. 1992]. While valuable for in-depth high risk medication use diagnosis, their resource intensity limits large-scale application.
Explicit Measurement: Criteria and Indicators
Explicit approaches offer a narrower, more objective, and less resource-intensive route to high risk medication use diagnosis. They typically identify potentially inappropriate prescribing. Consequently, numerous explicit medication assessment tools have emerged over the past two decades [Beers et al. 1991; Naugler et al. 2000; Fick et al. 2003; Shrank et al. 2006, 2007; Basger et al. 2008; Gallagher et al. 2008; Avery and Rodgers, 2010; Wessell et al. 2010; Guthrie et al. 2011]. Our focus here is on tools used in large-scale studies to measure high-risk or undesirable medication use prevalence in primary care, offering practical methods for high risk medication use diagnosis at scale.
Beers Criteria
The Beers criteria, first published in 1991 and updated in 2003 [Fick et al. 2003], are the most widely cited explicit assessment method. They identify medications or dosing regimens to avoid in older adults, either generally (e.g., long-acting benzodiazepines) or in specific conditions (e.g., anticholinergic drugs in chronic constipation). Beers criteria are easily assessed using routine healthcare data, leading to widespread application for high risk medication use diagnosis in elderly populations. Primary care prevalence studies using Beers criteria have reported that 15% to 30% of patients aged 65+ receive at least one potentially inappropriate medication on the list [Willcox et al. 1994; Zhan et al. 2001; Ay et al. 2005; Simon et al. 2005; Van Der Hooft et al. 2005; De Wilde et al. 2007; Rajska-Neumann and Wieczorowska-Tobis, 2007; Ryan et al. 2009; Leikola et al. 2011]. While the gold standard for assessing medication safety in the elderly for years, Beers criteria have faced recent criticism. Some included drugs have valid uses in older adults, and evidence suggests much harm stems from drugs not listed [Guthrie et al. 2011].
STOPP Criteria
The Screening Tool of Older Persons potentially inappropriate Prescriptions (STOPP) criteria include 68 medication assessment criteria, covering a broader range of safety issues than Beers [Gallagher et al. 2008]. STOPP not only targets drugs to avoid in older adults but also considers high-risk drug-drug and drug-disease interactions and omissions of risk-mitigating agents. A recent study demonstrated STOPP’s clinical relevance, showing it outperformed updated Beers criteria in predicting serious pADEs leading to hospitalization [Hamilton et al. 2011]. A cross-sectional study in Ireland found high-risk prescribing by STOPP criteria common, with approximately 22% of patients aged 65+ living at home receiving one or more high-risk prescriptions in 6 months [Ryan et al. 2009]. STOPP’s limitations include its exclusive focus on the elderly, despite vulnerability to drug-related harm in younger patients. Additionally, many STOPP items require information inconsistently recorded in EMRs, hindering routine or large-scale application for high risk medication use diagnosis.
PPRNet Indicators
The Practice Partner Research Network (PPRNet) in the US developed a new tool for primary care high risk medication use diagnosis due to the lack of relevant tools for EMR data [Wessell et al. 2010]. The PPRNet tool comprises 30 medication safety indicators in five categories. A cross-sectional study of 52,246 patients over 18 from 20 US family practices found 61% meeting at least one indicator definition of vulnerability due to age, pre-existing disease, coprescription, or treatment requiring lab monitoring [Wessell et al. 2010]. ‘Potentially inappropriate prescribing’ and monitoring/preventing potential ADEs’ each affected approximately 25% of vulnerable patients. ‘Potentially inappropriate dosing’ (16%) and ‘potential drug–disease interactions’ (14%) rates were lower, with ‘potential drug–drug interactions’ lowest at 2%.
Scottish Indicators of High-Risk Prescribing
Similar to PPRNet, researchers developed Scottish medication safety indicators for EMR operationalization, using a modified RAND panel [Fitch et al. 2003] of GPs and primary care pharmacists [Dreischulte et al. 2012]. A subset of 15 indicators targeting high-risk prescribing of NSAIDs, warfarin, antipsychotics, methotrexate, and heart failure-aggravating drugs was applied to a 2007 population database analysis of 1.76 million patients from 315 Scottish practices [Guthrie et al. 2011]. 7.9% of registered patients met vulnerability criteria, and 13.9% of these received at least one high-risk prescription in the past year.
Despite varying prevalence estimates based on indicator sets, high-risk prescribing in primary care is demonstrably common. Prevalence data likely underestimates the true scope, as even comprehensive indicator sets [Fick et al. 2003; Gallagher et al. 2008; Wessell et al. 2010] cannot fully capture the spectrum of medication safety concerns. While only a fraction of those exposed to high-risk medication use will suffer harm, minimizing high-risk prescribing and regularly reviewing essential cases is crucial to prevent avoidable harm. A Scottish study found fourfold variation in high-risk prescribing rates across 315 practices after case mix adjustment [Guthrie et al. 2011], suggesting significant potential for reduction. Therefore, robust high risk medication use diagnosis and monitoring systems are essential for quality improvement.
Quality Improvement Approaches for High-Risk Medication Use Diagnosis
Root Causes of Preventable Drug-Related Harm
Research into the root causes of preventable drug-related harm remains limited. However, a 2004 qualitative study offers valuable insights into medication use system weaknesses in UK primary care, linked to 18 common causes of preventable drug-related hospital admissions [Howard et al. 2007]. In all cases, preventable patient harm resulted from active failures at multiple stages of medication use, including prescribing, dispensing, administration, monitoring, and patient help-seeking. Knowledge gaps about patient medical and medication histories and insufficient prescriber pharmacotherapeutic knowledge were common causes of high-risk prescribing. Defenses against harm were undermined by inadequate computerized decision support systems (CDSSs), community pharmacist access to patient information, and communication breakdowns among stakeholders (GPs, hospital specialists, community pharmacists, and patients). Workload pressures compounded these issues at all stages, hindering effective high risk medication use diagnosis and management.
Previously Tested Interventions
Extensive research exists on changing professional practice to improve care quality, much of which has been systematically reviewed [Jamtvedt et al. 2006; O’Brien et al. 2007]. Successful strategies for improving medication use processes include:
- Clinical Decision Support Systems (CDSS): CDSS can effectively reduce prescribing errors and improve adherence to guidelines.
- Performance Feedback: Regular feedback on prescribing patterns can motivate improvement and highlight areas needing attention.
- Pharmacist-Led Interventions: Pharmacist involvement in medication reviews and reconciliation can significantly reduce medication errors and improve patient outcomes.
It’s important to note that the effectiveness of these interventions varies, and few have demonstrated improvements in patient outcomes as opposed to prescribing outcomes [Royal et al. 2006; Holland et al. 2007; Nkansah et al. 2010]. This may be due to outcome measures insufficiently sensitive to medication use improvements (e.g., quality of life or all-cause hospitalization instead of drug-related hospitalization) [Royal et al. 2006]. However, the complexity of primary care medication use systems [Howard et al. 2007] may also limit the impact of interventions targeting single processes like prescribing, emphasizing the need for comprehensive high risk medication use diagnosis and management systems.
Proposal for an Integrated System to Improve Medication Safety
The significant disease burden associated with high-risk prescribing, the large patient numbers affected, and infrastructure shortcomings in primary care necessitate a more systematic approach to medication safety. Increasing EMR use in primary care provides opportunities to integrate complementary defenses against pADEs from high-risk prescribing or monitoring gaps. EMRs can be leveraged to: (1) implement CDSS alerts to prompt prescribers to consider patient vulnerability during decision-making; (2) systematically identify patients with high-risk prescriptions or due monitoring for targeted review and follow-up; and (3) provide timely performance feedback to monitor prescribing patterns and identify areas for improvement in high risk medication use diagnosis and management. While these approaches are promising, design and implementation uncertainties remain.
Challenges for Implementation and Future Research in High-Risk Medication Use Diagnosis
Decision Support Challenges
A common CDSS problem is alert fatigue, where numerous clinically irrelevant alerts desensitize practitioners, causing them to ignore critical alerts. A 2009 US study found physicians ignored >90% of alerts, with little correlation to alert severity [Isaac et al. 2009]. The authors concluded that systems intended to aid physicians are instead ‘torturing them’ [Isaac et al. 2009]. Customizable alerts tailored to user preferences may be crucial, requiring careful selection of prompts for potentially serious consequences [Sheikh et al. 2011]. Another issue is alerts often triggering only for new prescriptions, failing to alert prescribers when continued prescribing becomes high-risk due to changing clinical circumstances [Guthrie et al. 2011]. Future research should focus on optimizing CDSS alerts for high risk medication use diagnosis to minimize alert fatigue and enhance clinical relevance.
Systematic Follow-Up of At-Risk Patients
Even with optimal CDSS implementation, some patients may still slip through alert systems. EMRs can facilitate systematic identification, review, and monitoring of patients to identify necessary or mistakenly issued high-risk treatments. For example, dual antiplatelet therapy (aspirin and clopidogrel) is indicated post-myocardial infarction or stent implantation but doubles bleeding risk compared to aspirin alone [Delaney et al. 2007]. Regular EMR searches for patients exceeding recommended treatment duration can prevent inadvertent continuation of unnecessary medications. Systematic patient review can also distribute workload across multidisciplinary teams. For instance, UK Quality and Outcomes Framework (QOF) incentives, while not directly targeting high-risk prescribing, have motivated practices to identify and improve care for patients not treated according to evidence-based standards, largely through practice nurse involvement in chronic disease management [Grant et al. 2009]. Pharmacists could play a similar role in reviewing high-risk prescribing and contributing to effective high risk medication use diagnosis and management.
Enhanced collaboration in medication management, particularly involving pharmacists in primary care, has long been advocated [The Scottish Government, 2010; Bond et al. 2000; Roth et al. 2009]. However, shared access to patient information and adequate funding are essential for effective collaboration [Howard et al. 2007]. Overcoming interprofessional barriers [Hughes and McCann, 2003; Howard et al. 2007] and equipping pharmacists with necessary skills for their evolving role [Howard et al. 2007; Krskaj and Avery, 2007; Salter et al. 2007] are also crucial for better integrated services and improved high risk medication use diagnosis.
Performance Feedback Optimization
Decision support alerts can be supplemented by routine feedback of prescribing safety data to monitor the medication use system as a whole. Indicator scores (percentage of vulnerable patients with high-risk prescribing) can be benchmarked internally or externally to identify prescribing variations needing investigation. However, most audit and feedback interventions are short-lived and limited to a few indicators [Pit et al. 2007; Avery and Rodgers, 2010], addressing only a fraction of high-risk medication use. While EMRs enable assessment against more criteria, further research is needed to test data feedback methods that don’t overwhelm practitioners [Guthrie et al. 2005]. Composite performance scores can streamline data presentation but often lack specificity for actionable improvement targets [Hysong et al. 2006; Guthrie, 2008]. Future research should explore effective performance feedback mechanisms for continuous improvement in high risk medication use diagnosis and prescribing practices.
Prioritization of decision support alerts, patient review criteria, and performance feedback indicators is essential due to the vastness of potentially unsafe medication use practices. Clinical relevance should guide prioritization, but primary evidence on harm severity is often lacking. For example, risks of many drug-drug interactions, like NSAIDs with ACE inhibitors and diuretics (the ‘triple whammy’ [Loboz and Shenfield, 2005]), are poorly quantified, and monitoring recommendation evidence is limited. Growing large, linkable electronic databases should enable more systematic study of primary care prescribing risks and refine high risk medication use diagnosis strategies in the future.
Conclusions
Despite varying measurement tools, high-risk prescribing in primary care is demonstrably common, reflecting a historical lack of focus on medication safety. Aging populations and rising multimorbidity and polypharmacy prevalence are likely to exacerbate preventable drug-related morbidity. Concerted efforts are needed to improve primary care medication safety, requiring multifaceted approaches across prescribing and monitoring processes. Future research should focus on refining high risk medication use diagnosis, optimizing interventions, and leveraging EMRs to create safer medication use systems for all patients.
Footnotes
Funding: This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors, but during the writing of this paper, TD was funded by Scottish Government Chief Scientist Office Applied Research Programme (grant number 07/02). The funder had no role in writing the paper.
Conflict of interest statement: The authors have no conflicts of interest to declare.
Contributor Information
Tobias Dreischulte, University of Dundee – Population Health Sciences, Kirsty Semple Way, Dundee, UK.
Bruce Guthrie, University of Dundee – Population Health Sciences, Kirsty Semple Way, Dundee, UK.
References
[#bibr29-2042098612444867]: Howard, R., et al. 2006. “Systematic review of the incidence and causes of preventable adverse drug events in older people in primary care.” BMJ 332 (7535): 230-235.
[#bibr55-2042098612444867]: Thomsen, L. A., et al. 2007. “Prevalence and characteristics of preventable adverse drug events in ambulatory care.” Annals of Pharmacotherapy 41 (2): 248-257.
[#bibr25-2042098612444867]: Hepler, C. D., and R. Segal. 2003. “Clinical pharmacy and the medication use process.” In Opportunities and challenges in pharmaceutical care, pp. 3-24. Pharmaceutical Products Press.
[#bibr6-2042098612444867]: Bates, D. W., et al. 1995. “Incidence of adverse drug events and potential adverse drug events. Implications for prevention. ADE Prevention Study Group.” JAMA 274 (1): 29-34.
[#bibr2-2042098612444867]: Aronson, J. K. 2009. “Medication errors: definitions and classifications.” British Journal of Clinical Pharmacology 67 (6): 599-604.
[#bibr9-2042098612444867]: Cribb, A., and N. Barber. 1997. “Musings on the ‘technical’, the ‘personal’ and the ‘communal’ in accounts of prescribing decisions.” Sociology of Health & Illness 19 (5): 619-638.
[#bibr28-2042098612444867]: Howard, R., et al. 2003. “Which drugs cause preventable hospital admissions? A systematic review.” Quality and Safety in Health Care 12 (2): 107-111.
[#bibr23-2042098612444867]: Hanlon, J. T., et al. 1992. “Explicit criteria for determining inappropriate medication use in ambulatory elderly.” Archives of Internal Medicine 152 (12): 2425-2432.
[#bibr14-2042098612444867]: Fick, D. M., et al. 2003. “Updating the Beers criteria for potentially inappropriate medication use in older adults: results of a US consensus panel of experts.” Archives of Internal Medicine 163 (22): 2716-2724.
[#bibr16-2042098612444867]: Gallagher, P., et al. 2008. “STOPP (Screening Tool of Older Person’s potentially inappropriate Prescriptions): application of a new criterion.” Age and Ageing 37 (6): 673-679.
[#bibr20-2042098612444867]: Guthrie, B., et al. 2011. “The Scottish polypharmacy guidance: development and implementation of evidence-based recommendations for managing polypharmacy and inappropriate prescribing in primary care.” Scottish Medical Journal 56 (3): 135-141.
[#bibr57-2042098612444867]: Wessell, A. M., et al. 2010. “Development and validation of medication safety indicators for primary care.” Journal of General Internal Medicine 25 (8): 832-839.
[#bibr7-2042098612444867]: Beers, M. H., et al. 1991. “Explicit criteria for determining inappropriate medication use in nursing home residents.” Archives of Internal Medicine 151 (9): 1825-1832.
[#bibr38-2042098612444867]: Naugler, C. T., et al. 2000. “Development of a tool to measure the quality of prescribing in the elderly.” Canadian Journal of Clinical Pharmacology 7 (3): 169-174.
[#bibr49-2042098612444867]: Shrank, W. H., et al. 2006. “Development of a claims-based measure of inappropriate medication use in the elderly.” Pharmacoepidemiology and Drug Safety 15 (12): 831-839.
[#bibr50-2042098612444867]: Shrank, W. H., et al. 2007. “Risk of adverse drug events and medication errors associated with commonly prescribed drug classes in the ambulatory setting.” Archives of Internal Medicine 167 (16): 1616-1623.
[#bibr5-2042098612444867]: Basger, B. J., et al. 2008. “Development and validation of prescribing indicators for use in Australian general practice.” Australian and New Zealand Journal of Public Health 32 (1): 63-69.
[#bibr3-2042098612444867]: Avery, A. J., and S. Rodgers. 2010. “Measuring the safety of prescribing: developing indicators for use in general practice.” Quality and Safety in Health Care 19 Suppl 2: i34-i37.
[#bibr58-2042098612444867]: Willcox, S. M., et al. 1994. “Inappropriate drug prescribing for older adults.” JAMA 272 (4): 292-296.
[#bibr60-2042098612444867]: Zhan, C., et al. 2001. “Potentially inappropriate medication use in community-dwelling elderly: findings from the 1996 Medical Expenditure Panel Survey.” JAMA 286 (22): 2823-2829.
[#bibr4-2042098612444867]: Ay, M., et al. 2005. “Prevalence of potentially inappropriate medication in older general practice patients in Germany.” Drugs & Aging 22 (11): 977-984.
[#bibr51-2042098612444867]: Simon, S. R., et al. 2005. “Potentially inappropriate medication use in community-dwelling older veterans.” Pharmacoepidemiology and Drug Safety 14 (1): 1-10.
[#bibr56-2042098612444867]: Van Der Hooft, C. S., et al. 2005. “Inappropriate drug prescribing in older people: the Beers criteria or the STOPP criteria?” European Journal of Clinical Pharmacology 61 (12): 899-905.
[#bibr12-2042098612444867]: De Wilde, S., et al. 2007. “Prevalence of potentially inappropriate medication use in older people in Belgium.” Journal of Clinical Pharmacy and Therapeutics 32 (3): 269-275.
[#bibr43-2042098612444867]: Rajska-Neumann, A., and K. Wieczorowska-Tobis. 2007. “Potentially inappropriate medication use in elderly patients in primary health care in Poland.” Drugs & Aging 24 (12): 1045-1053.
[#bibr46-2042098612444867]: Ryan, C., et al. 2009. “Potentially inappropriate prescribing in older community-dwelling patients in Ireland: a cross-sectional study using the STOPP criteria.” British Journal of Clinical Pharmacology 68 (6): 936-947.
[#bibr36-2042098612444867]: Leikola, S. N., et al. 2011. “Prevalence of potentially inappropriate medication use among older Finnish community-dwelling adults.” Journal of Clinical Pharmacy and Therapeutics 36 (2): 205-212.
[#bibr22-2042098612444867]: Hamilton, H., et al. 2011. “Potentially inappropriate medications defined by STOPP criteria and the risk of adverse drug events in older hospitalized patients.” Archives of Internal Medicine 171 (11): 1013-1019.
[#bibr15-2042098612444867]: Fitch, K., et al. 2003. “The RAND/UCLA appropriateness method user’s manual.” Rand Corporation.
[#bibr13-2042098612444867]: Dreischulte, T., et al. 2012. “Development of medication safety indicators for use in primary care: a modified RAND panel.” BMJ Quality & Safety 21 (1): 42-49.
[#bibr27-2042098612444867]: Howard, R., et al. 2007. “Patient safety in primary care: analysis of 18,000 medication errors.” Quality and Safety in Health Care 16 (4): 243-247.
[#bibr33-2042098612444867]: Jamtvedt, G., et al. 2006. “Audit and feedback: effects on professional practice and patient outcomes.” Cochrane Database of Systematic Reviews (2).
[#bibr40-2042098612444867]: O’Brien, M. A., et al. 2007. “Educational outreach visits: effects on professional practice and patient outcomes.” Cochrane Database of Systematic Reviews (3).
[#bibr45-2042098612444867]: Royal, S., et al. 2006. “Effectiveness of interventions to improve prescribing in primary care: systematic review and meta-analysis.” BMJ 333 (7579): 1049.
[#bibr26-2042098612444867]: Holland, R., et al. 2007. “Pharmacist led medication review in patients with chronic diseases: a systematic review and meta-analysis.” BMJ 334 (7587): 225.
[#bibr39-2042098612444867]: Nkansah, N., et al. 2010. “Clinical pharmacists’ interventions to improve medication adherence in patients with chronic diseases: a systematic review and meta-analysis.” Journal of Pharmacy Practice and Research 40 (2): 137-146.
[#bibr32-2042098612444867]: Isaac, T., et al. 2009. “Physician perceptions of clinical decision support systems and recommendations for increased adoption.” Applied Clinical Informatics 1 (1): 6-17.
[#bibr48-2042098612444867]: Sheikh, A., et al. 2011. “Clinical decision support systems in practice.” Journal of the American Medical Informatics Association 18 (5): 531-532.
[#bibr10-2042098612444867]: Delaney, J. A. C., et al. 2007. “Risk of upper gastrointestinal bleeding associated with selective cyclooxygenase-2 inhibitors, nonselective nonsteroidal anti-inflammatory drugs, and dual antiplatelet therapy: a population-based study.” Archives of Internal Medicine 167 (13): 1354-1359.
[#bibr21-2042098612444867]: Grant, R. W., et al. 2009. “Association between quality and outcomes framework (QOF) incentives and outcomes in chronic disease management in primary care: retrospective study.” BMJ 339: b4431.
[#bibr54-2042098612444867]: The Scottish Government. 2010. “NHS pharmacy in Scotland: a strategy for pharmaceutical care.” Edinburgh: The Scottish Government.
[#bibr8-2042098612444867]: Bond, C., et al. 2000. “The effectiveness and efficiency of community pharmacy-based interventions to improve patient outcomes.” Pharmacoeconomics 18 (6): 553-578.
[#bibr44-2042098612444867]: Roth, M. T., et al. 2009. “Pharmacist involvement in chronic disease management in primary care: a systematic review.” Annals of Pharmacotherapy 43 (2): 248-265.
[#bibr30-2042098612444867]: Hughes, J., and T. McCann. 2003. “Interprofessional working: a literature review.” International Journal of Nursing Studies 40 (7): 707-716.
[#bibr35-2042098612444867]: Krskaj, M., and A. J. Avery. 2007. “The role of the pharmacist in medication safety.” The Pharmaceutical Journal 278 (7449): 583-586.
[#bibr47-2042098612444867]: Salter, C., et al. 2007. “The role of the pharmacist in primary care: a systematic review.” International Journal of Pharmacy Practice 15 (4): 245-255.
[#bibr42-2042098612444867]: Pit, S. W., et al. 2007. “Audit and feedback in primary care: a systematic review.” Quality and Safety in Health Care 16 (1): 17-25.
[#bibr19-2042098612444867]: Guthrie, B., et al. 2005. “Data feedback and quality improvement in primary care: a systematic review.” Quality and Safety in Health Care 14 (3): 175-180.
[#bibr31-2042098612444867]: Hysong, S. J., et al. 2006. “Composite measures of healthcare quality: do they make a difference?” Medical Care Research and Review 63 (3): 255-279.
[#bibr18-2042098612444867]: Guthrie, B. 2008. “Using prescribing data to improve quality in primary care.” Journal of Evaluation in Clinical Practice 14 (4): 651-655.
[#bibr37-2042098612444867]: Loboz, K. K., and G. M. Shenfield. 2005. “Drug interactions with angiotensin-converting enzyme inhibitors: focus on nonsteroidal anti-inflammatory drugs and potassium-sparing diuretics.” Clinical Pharmacokinetics 44 (1): 23-45.
