Abstract
The aim of this prospective observational study was to establish associations between the use of high‐risk medicine groups and the study outcome: occurrence of at least 1 moderate or severe preventable medication‐related problem. Data on medication‐related problems, high‐risk medicines, and other potential risk factors were collected from adults on medical wards in 2 UK hospitals. Logistic regression modelling was used to determine relationships between high‐risk medicines and the study outcome. Among 1503 eligible admissions, 6 high‐risk medicine groups were associated with the study outcome on univariable analysis; multivariable analysis found only systemic antimicrobials and epilepsy medicines to be independently associated with the outcome (adjusted odds ratio 1.44, 95% confidence interval 1.08–1.92 and adjusted odds ratio 1.61, 95% confidence interval 1.16–2.25 respectively). Identification of high‐risk medicine groups has potential to permit targeting of patients at highest risk of avoidable medication‐related harm, but multivariable analysis suggests risk is likely to be multifactorial.
Keywords: clinical pharmacology, medication safety, pharmacoepidemiology, epidemiology, health services research
What is already known about this subject
Medicines are integral to healthcare, but there is growing evidence of a need to improve medication safety.
Previous research has identified medicines that are potentially high‐risk in hospital inpatient settings, but no statistical assessment of high‐risk medicines and clinically relevant medication‐related problems has yet been conducted.
What this study adds
Only 2 of 11 groups of high‐risk medicines (systemic antimicrobials and epilepsy medicines) were significantly associated with moderate or severe preventable medication‐related problems on multivariable analysis.
Identification of high‐risk medicines has potential to permit targeting of patients at risk of avoidable medication‐related harm, however, additional risk factors will also need to be considered.
1. INTRODUCTION
Medicines are integral to the prevention, treatment and management of health,1 but there is growing evidence of a need to improve medication safety.1, 2, 3, 4, 5, 6, 7, 8 Regarding hospitalised patients in England, it has been estimated that the prescribing error rate is almost 9%,9 and that 1 in 7 patients experience harm from their medicines.10
Previous research has identified medicines that are potentially high‐risk when used in a hospital inpatient setting,11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21 but no statistical assessment of high‐risk medicines and clinically relevant medication‐related problems (MRPs) has yet been conducted. Statistical assessment has potential to quantify the impact of groups of high‐risk medicines on the risk of developing clinically relevant MRPs, and to explore potential correlation between high‐risk medicines and other risk factors such as age and renal function. Statistical assessment could therefore inform targeting of patients at highest risk of avoidable medication‐related harm. The aim of the present study was therefore to establish the univariable and multivariable associations between high‐risk medicines and the study outcome, which was occurrence of at least 1 moderate or severe preventable MRP.
2. METHODS
2.1. Study design
This prospective observational study involved adults admitted to the medical wards of 2 hospitals in South East England, described in detail elsewhere.22, 23 A summary is provided in Appendix S1.
MRPs were defined as all circumstances involving a patient's drug treatment that actually, or potentially, interfere with the achievement of an optimal outcome.24, 25, 26 Previous research suggests a significant proportion of hospitalised patients experience MRPs, but many are of limited clinical significance.27 We therefore chose a clinically relevant outcome measure based on severity: moderate or severe MRPs. Similarly, we selected preventable MRPs to focus on patients with MRPs amenable to intervention. Severity was assessed by an expert panel comprising the principal investigator, a hospital pharmacist, a senior nurse and a consultant physician using a validated visual analogue scale.28 At the point of identification, pharmacy staff recorded whether they considered the MRP was preventable, expressed as a dichotomous variable of yes or no. A second check was performed by the principal investigator. Examples of MRPs, classified by severity and preventability, have been published previously.23 MRP data were identified and recorded by pharmacists at the study sites as part of their routine daily clinical assessment of patients. Data on 18 potential risk factors, including use of high‐risk medicines, were collected retrospectively by a researcher. Further details are provided in Appendix S1.
High‐risk medicines were grouped into 11 categories (Table 1), selected based on previously published research.11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21 Grouping was used rather than considering medicines individually since different organisations are likely to use different specific medicines. It also reduced the risk of model overfitting associated with using too many variables.29 A category of other high‐risk medicines was used to permit inclusion of medicines considered to be high‐risk, but where use was too infrequent to model individually (i.e. <5% of the study population).30, 31 A summary of the medicines included in each category is given in Appendix S1.
Table 1.
Univariable and multivariable associations between high‐risk medicines and outcome events
| High‐risk medicine | Occurrence of outcome event | Univariable analysis | Multivariable analysisa | |||
|---|---|---|---|---|---|---|
| No (%) | Yes (%) | Odds ratio (95% CI) | P valueb | Odds ratio (95% CI) | P valueb | |
| Systemic antimicrobials (excluding aminoglycosides and glycopeptides) | 512 (57.3) | 525 (69.7) | 1.71 (1.37–2.12) | <.001 | 1.44 (1.08–1.92) | .013 |
| Aminoglycosides and glycopeptides | 49 (5.5) | 56 (9.2) | 1.73 (1.17–2.58) | .006 | 1.39 (0.89–2.18) | .144 |
| Antidepressants | 182 (20.4) | 169 (27.7) | 1.50 (1.18–1.91) | .001 | 1.22 (0.93–1.61) | .146 |
| Anticoagulants | 161 (18.0) | 151 (24.8) | 1.50 (1.17–1.92) | .001 | 1.12 (0.83–1.50) | .468 |
| Antidiabetic medication | 146 (16.4) | 153 (25.1) | 1.71 (1.33–2.20) | <.001 | 1.25 (0.92–1.69) | .147 |
| Epilepsy medicinesc | 104 (11.7) | 123 (20.2) | 1.92 (1.45–2.56) | <.001 | 1.61 (1.16–2.25) | .005 |
| Therapeutic heparin | 119 (13.3) | 103 (16.9) | 1.32 (0.99–1.76) | .057 | 1.31 (0.92–1.86) | .134 |
| Antiarrhythmics | 78 (8.7) | 72 (11.8) | 1.40 (1.00–1.96) | .052 | 0.95 (0.64–1.40) | .778 |
| Opioidsd | 76 (8.5) | 69 (11.3) | 1.37 (0.97–1.93) | .075 | 1.01 (0.67–1.52) | .949 |
| Antipsychotics | 50 (5.6) | 43 (7.1) | 1.27 (0.83–1.93) | .270 | 1.18 (0.72–1.94) | .511 |
| Other high‐risk medicinese (antiretrovirals, medicines for Parkinson's disease, theophylline and aminophylline, immunosuppressants, cytotoxics, lithium) | 62 (6.9) | 55 (9.0) | 1.32 (0.92–1.90) | .134 | 1.13 (0.76–1.68) | .557 |
Adjusted for age, socioeconomic status, previous allergy, body mass index, number of hospital admissions in previous 6 months, primary diagnosis, number of comorbidities, history of dementia, number of regular medicines prescribed on the first day of admission, parenteral medicines administration, renal function, liver disease, serum albumin, serum potassium, white cell count, platelet count.
Test for difference between admissions with and without occurrence of outcome event. Obtained from regression modelling.
Includes use for all potential indications.
Codeine, tramadol, meptazinol & dihydrocodeine excluded to restrict analysis to stronger opioids.
High‐risk medicines/groups used in <5% of the study population.
CI = confidence interval
This study received National Health Service ethical approval (16/WA/0016).
2.2. Data analysis
We performed univariable analyses to provide data on the unadjusted association between the preselected groups of high‐risk medicines and the outcome event. Multivariable analysis was used to identify the high‐risk medicines that were independently associated with the study outcome after adjusting for other potential risk factors: use of other high‐risk medicines, age, socioeconomic status, previous allergy, body mass index, number of hospital admissions in previous 6 months, primary diagnosis, number of comorbidities, history of dementia, number of regular medicines prescribed on the first day of admission, parenteral medicines administration, renal function, liver disease, serum albumin, serum potassium, white cell count and platelet count. The extent of missing data has been reported previously.23 In summary, of 1503 included admissions, 387 (25.7%) had 1 or more missing data points, accounting for 1.6% of total risk factor data. Exploratory analysis found that data were likely to be missing at random, missing data were therefore handled using multiple imputation. Information on the potential impact of excluded patients is given in Appendix S2.
Logistic regression modelling was performed using a generalised estimating equation approach, this was to account for possible correlation between patients admitted more than once during the study period.32 Odds ratios (OR) were obtained, a P value <.05 was regarded as statistically significant and 95% confidence intervals (CI) are reported.
Results are reported according to the Strengthening the Reporting of Observational Studies in Epidemiology reporting guidelines for observational studies.33 All analyses were conducted using Stata version 14.2.
3. RESULTS
An overview of the 1503 included patient admissions has been described in detail elsewhere.23 A summary is provided in Appendix S2. In total, 610 admissions experienced the study outcome.
3.1. Associations between the high‐risk medicines and study outcome
The results of univariable and multivariable analyses between the preselected groups of high‐risk medicines and the study outcome are shown in Table 1. The univariable analyses showed there was strong evidence for statistically significant associations between the study outcome and the use of the following high‐risk medicines: systemic antimicrobials (P < .001), aminoglycosides/glycopeptides (P = .006), antidepressants (P = .001), anticoagulants (P = .001), antidiabetic medication (P < .001) and epilepsy medicines (P < .001).
Multivariable analysis showed that only 2 medicine groups (systemic antimicrobials and epilepsy medicines) were still associated with increased risk of the study outcome (systemic antimicrobials adjusted OR 1.44, 95% CI 1.08–1.92, P = .013 and epilepsy medicines adjusted OR 1.61, 95% CI 1.16–2.25, P = .005).
4. DISCUSSION
4.1. Key findings
In the univariable analyses 6 of 11 preselected high‐risk medicine groups were significantly associated with the study outcome, experiencing at least 1 moderate or severe preventable MRP: systemic antimicrobials, aminoglycosides/glycopeptides, antidepressants, anticoagulants, antidiabetic medication and epilepsy medicines. However, on multivariable analysis, only 2 high‐risk medicines groups were independently associated with the study outcome: systemic antimicrobials and epilepsy medicines.
4.2. Comparison with previous work
Previous research has identified medicines that are potentially high‐risk when used in a hospital inpatient setting,11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21 but direct comparison between sources is difficult due to differences in study design, outcome measures, and the way medicines were grouped. To our knowledge, the present study is the first to identify high‐risk medicines that are associated with moderate or severe preventable MRPs.
4.3. Strengths and limitations
Strengths of this research include adherence to prognostic modelling recommendations34, 35, 36; this has the potential to enhance the quality of data collection and reduce bias. Other strengths include the relatively large sample size and use of 2 study sites to increase generalisability. The choice of study sites is also a strength; their size, range of services and demographics make them broadly representative of other acute district general hospitals in England.
The observational nature of the study is a potential limitation, as data collection was not carried out under strict trial conditions. This led to a small amount of missing data, and the possibility that MRP prevalence was underestimated. To minimise under‐reporting of MRPs we provided regular training for all staff involved. Another possible limitation is risk of bias due to exclusion of admissions whose prescribing records were not reviewed by a clinical pharmacist during admission. However, as discussed in Appendix S2, this impact is likely to be minimal. Finally, while it was necessary to categorise high‐risk medicines, we appreciate that medicines within the same category may not have equivalent risks, for example insulin and oral diabetes medication.
4.4. Interpretation
We have quantified the impact of high‐risk medicines on the risk of developing clinically relevant MRPs; this has potential to inform targeting of patients at highest risk of avoidable medication‐related harm. The univariable analyses suggest strong evidence for associations between the study outcome and 6 of 11 high‐risk medicines groups. Despite strong univariable associations, only 2 of these medicine groups remained statistically significant in multivariable analysis: systemic antimicrobials (excluding aminoglycosides and glycopeptides) and epilepsy medicines. This suggests the association between the study outcome and high‐risk medicines can be explained, either partly or fully, by additional risk factors in the multivariable model. While risk of medication‐related harm associated with medicines such as anticoagulants and antidiabetic medicines is well recognised,37, 38 the present study suggests such risk is likely to be multifactorial and subject to residual confounding, with use of high‐risk medicines alone being unlikely to accurately predict the occurrence of moderate or severe preventable MRPs. The present study therefore suggests that, while identification of high‐risk medicines associated with clinically relevant MRPs has potential to inform targeting of patients for medicines optimisation activities, additional risk factors also need to be considered.23 Further studies in different settings are needed to confirm these findings.
5. CONCLUSION
We found that 6 of 11 groups of high‐risk medicines were associated with moderate or severe preventable MRPs on univariable analysis. While identification of these high‐risk groups has potential to permit targeting of patients at highest risk of avoidable medication‐related harm only 2 medicine groups (systemic antimicrobials and epilepsy medicines) were significantly associated with the study outcome on multivariable analysis. This suggests that additional risk factors may also need to be considered when predicting overall risk.
COMPETING INTERESTS
There are no competing interests to declare.
The views expressed are those of the author(s) and not necessarily those of the National Health Service, the NIHR, Public Health England or the Department of Health and Social Care. The funder had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
CONTRIBUTORS
C.G. is the principal investigator, and is responsible for the initial concept, study design and analysis plan. B.D.F. and L.W. refined the design and analysis plan. C.G. applied for NIHR fellowship funding, with the support and guidance of B.D.F. and L.W. C.G. drafted the manuscript, which was then critically reviewed by B.D.F. and L.W. All authors approved the final version.
Supporting information
Appendix S1 Supplementary information on methods
Appendix S2 Supplementary information on results
ACKNOWLEDGEMENTS
We would like to thank the patient and public members of the project steering group (Helen Clothier, Marie‐France Capon, Derek Smith, Jack Wright and Tom Drabble), also Mary Evans and Lindsay Smith, who provided clinical expertise to this group. We also thank the Pharmacy staff at the study sites, who collected outcome data. Thanks must also go to the expert panel members: Sue Lee, Sivanangai Puthrasingam and Ann Williams, and to Jack Glendenning and Colin Merrill, who supported risk factor data collection. We are also grateful to the executive editor and 2 anonymous reviewers for their very helpful feedback on an earlier version of this paper.
This work was supported by a Clinical Doctoral Research Fellowship award from Health Education England and the National Institute for Health Research (NIHR), CDRF‐2014‐05‐033. This article represents independent research supported by the NIHR Imperial Patient Safety Translational Research Centre and the NIHR Health Protection Research Unit in Healthcare Associated Infections and Antimicrobial Resistance at Imperial College in partnership with Public Health England.
Geeson C, Wei L, Franklin BD. High‐risk medicines associated with clinically relevant medication‐related problems in UK hospitals: A prospective observational study. Br J Clin Pharmacol. 2020;86:165–169. 10.1111/bcp.14119
PI statement: This study was an observational study that did not involve interventions with human subjects or require substances to be administered to human subjects/patients (other than required for standard care). As such, the principal investigator for this paper (Cathy Geeson) did not require direct clinical responsibility for patients.
DATA AVAILABILITY STATEMENT
The data that support the findings of this study are available from the corresponding author upon reasonable request.
REFERENCES
- 1. National Institute for Health and Care Excellence . Medicines optimisation: the safe and effective use of medicines to enable the best possible outcomes, NICE guidelines [NG5]. 2015. [PubMed]
- 2. Francis R. Report of the Mid Staffordshire NHS Foundation Trust Public Enquiry Mid Staffordshire NHS Foundation Trust Public Inquiry 2013; 2013.
- 3. Berwick D. A promise to learn – a commitment to act. Improving the safety of patients in England. In: England TNAGotSoPi, editor. 2013.
- 4. The Royal Pharmaceutical Society . Keeping patients safe when they transfer between care providers – getting the medicines right. 2012.
- 5. The Royal Pharmaceutical Society . Medicines Optimisation: Helping patients to make the most of medicines, good practice guidance for healthcare professionals in England. 2013.
- 6. National Institute for Health and Care Excellence . CG138 Patient experience in adult NHS services. 2012. [PubMed]
- 7. World Health Organization . WHO Global Patient Safety Challenge: Medication Without Harm. 2017.
- 8. Elliott RA, Camacho E, Campbell F, Jankovic D, St James MM, Kaltenthaler E, et al. Prevalence and economic burden of medication errors in the nhs in england. Policy Research Unit in Economic Evaluation of Health and Care Interventions. Universities of Sheffield and York; 2018.
- 9. Dornan T, Ashcroft D, Heathfield H, et al. An in‐depth investigation into causes of prescribing errors by foundation trainees in relation to their medical education: EQUIP study. London: General Medical Council; 2009:1‐215. [Google Scholar]
- 10. Davies EC, Green CF, Taylor S, Williamson PR, Mottram DR, Pirmohamed M. Adverse drug reactions in hospital in‐patients: a prospective analysis of 3695 patient‐episodes. PLoS One. 2009;4(2):e4439. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11. Suggett E, Marriott J. Risk factors associated with the requirement for pharmaceutical intervention in the hospital setting: a systematic review of the literature. Drugs ‐ Real World Outcomes. 2016;3(3):241‐263. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12. Saedder EA, Brock B, Nielsen LP, Bonnerup DK, Lisby M. Identifying high‐risk medication: a systematic literature review. Eur J Clin Pharmacol. 2014;70(6):637‐645. [DOI] [PubMed] [Google Scholar]
- 13. Boeker EB, Ram K, Klopotowska JE, et al. An individual patient data meta‐analysis on factors associated with adverse drug events in surgical and non‐surgical inpatients. Br J Clin Pharmacol. 2015;79(4):548‐557. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14. Thomas SK, McDowell SE, Hodson J, et al. Developing consensus on hospital prescribing indicators of potential harms amenable to decision support. Br J Clin Pharmacol. 2013;76(5):797‐809. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15. Roten I, Marty S, Beney J. Electronic screening of medical records to detect inpatients at risk of drug‐related problems. Pharm World Sci. 2009;32(1):103. [DOI] [PubMed] [Google Scholar]
- 16. Cottrell R, Caldwell M, Jardine G. Developing and implementing a pharmacy risk screening tool. Hosp Pharm Eur. 2013(71):58‐60. [Google Scholar]
- 17. Falconer N, Nand S, Liow D, Jackson A, Seddon M. Development of an electronic patient prioritization tool for clinical pharmacist interventions. Am J Health Syst Pharm. 2014;71(4):311‐320. [DOI] [PubMed] [Google Scholar]
- 18. Saedder EA, Lisby M, Nielsen LP, et al. Detection of patients at high risk of medication errors: development and validation of an algorithm. Basic Clin Pharmacol Toxicol. 2016;118(2):143‐149. [DOI] [PubMed] [Google Scholar]
- 19. Hickson RP, Steinke DT, Skitterall C, Williams SD. Evaluation of a pharmaceutical assessment screening tool to measure patient acuity and prioritise pharmaceutical care in a UK hospital. Eur J Hosp Pharm. 2017;24:74–9. 10.1136/ejhpharm-2015-000829 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20. Kaufmann CP, Stämpfli D, Hersberger KE, Lampert ML. Determination of risk factors for drug‐related problems: a multidisciplinary triangulation process. BMJ Open. 2015;5(3):e006376. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21. Institute for Safe Medication Practices . High‐Alert Medications in Acute Care Settings https://www.ismp.org/recommendations/high-alert-medications-acute-list
- 22. Geeson C, Wei L, Franklin BD. Medicines optimisation assessment tool (MOAT): a prognostic model to target hospital pharmacists' input to improve patient outcomes. Protocol for an observational study. BMJ Open. 2017;7(6):e017509. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23. Geeson C, Wei L, Franklin BD. Development and performance evaluation of the medicines optimisation assessment tool (MOAT): a prognostic model to target hospital pharmacists' input to prevent medication‐related problems. BMJ Qual Saf. 2019;28(8):645‐656. 10.1136/bmjqs-2018-008335 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24. van den Bemt PMLA, Egberts TCG, de Jong‐van den Berg LTW, Brouwers JRBJ. Drug‐related problems in hospitalised patients. Drug Saf. 2000;22(4):321‐333. [DOI] [PubMed] [Google Scholar]
- 25. Krähenbühl‐Melcher A, Schlienger R, Lampert M, Haschke M, Drewe J, Krähenbühl S. Drug‐related problems in hospitals. Drug Saf. 2007;30(5):379‐407. [DOI] [PubMed] [Google Scholar]
- 26. Pharmaceutical Care Network Europe . The PCNE Classification V 7.0 2016 http://www.pcne.org/upload/files/152_PCNE_classification_V7-0.pdf
- 27. Blix HS, Viktil KK, Reikvam Å, et al. The majority of hospitalised patients have drug‐related problems: results from a prospective study in general hospitals. Eur J Clin Pharmacol. 2004;60(9):651‐658. [DOI] [PubMed] [Google Scholar]
- 28. Dean BS, Barber ND. A validated, reliable method of scoring the severity of medication errors. Am J Health Syst Pharm. 1999;56(1):57‐62. [DOI] [PubMed] [Google Scholar]
- 29. Moons KG, de Groot JA, Bouwmeester W, et al. Critical appraisal and data extraction for systematic reviews of prediction modelling studies: the CHARMS checklist. PLoS Med. 2014;11(10):e1001744. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30. Katz MH. Multivariable analysis: a primer for readers of medical research. Ann Intern Med. 2003;138(8):644‐650. [DOI] [PubMed] [Google Scholar]
- 31. Vittinghoff E, McCulloch CE. Relaxing the rule of ten events per variable in logistic and cox regression. Am J Epidemiol. 2007;165(6):710‐718. [DOI] [PubMed] [Google Scholar]
- 32. Kirkwood BR, Sterne JA. Essential medical statistics. Oxford, UK: Blackwell Publishing; 2010. [Google Scholar]
- 33. Vandenbroucke JP, Von Elm E, Altman DG, et al. Strengthening the reporting of observational studies in epidemiology (STROBE): explanation and elaboration. PLoS Med. 2007;4(10):e297. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34. Hemingway H, Croft P, Perel P, et al. Prognosis research strategy (PROGRESS) 1: a framework for researching clinical outcomes. BMJ. 2013;346(feb05 1):e5595. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35. Steyerberg EW, Moons KG, van der Windt DA, et al. Prognosis research strategy (PROGRESS) 3: prognostic model research. PLoS Med. 2013;10(2):e1001381. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36. Hayden JA, van der Windt DA, Cartwright JL, Côté P, Bombardier C. Assessing bias in studies of prognostic factors. Ann Intern Med. 2013;158(4):280‐286. [DOI] [PubMed] [Google Scholar]
- 37. National Institute for Health and Care Excellence . Anticoagulants, including non‐vitamin K antagonist oral anticoagulants (NOACs). 2016.
- 38. Lamont T, Cousins D, Hillson R, Bischler A, Terblanche M. Safer administration of insulin: summary of a safety report from the National Patient Safety Agency. BMJ. 2010;341(oct13 3):c5269. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Appendix S1 Supplementary information on methods
Appendix S2 Supplementary information on results
Data Availability Statement
The data that support the findings of this study are available from the corresponding author upon reasonable request.