None listed
Conditions
Brief summary
Early identification of patients at high risk of medication harm can allow for timely and targeted patient review and proactive monitoring to mitigate harm. An emerging approach involves predictive risk models, which use statistical algorithms to quantify the probability that an individual patient will experience medication harm, and facilitate timely, patient-specific interventions (e.g. therapeutic drug monitoring and deprescribing). A number of risk prediction models for medication harm, in various health settings, have been reported. A locally developed risk model, the Adverse Inpatient Medication Event Model (AIME), was developed and internally validated at the Princess Alexandra Hospital (PAH) using general medical and geriatric patient data from July to December 2017, which has been published in the British Journal of Clinical Pharmacology (Falconer, N, et al. Development and validation of the Adverse Inpatient Medication Event model (AIME). Br J Clin Pharmacol. 2021; 87: 1512– 1524.) where the AIME model evaluated and modified (using frailty as a variable), in a multisite retrospective study conducted using data from Princess Alexandra Hospital (PAH) and Logan Hospital (LGH). Frailty - which can also increase risk of medication harm and poor patient outcomes was also added to this model. In the new study incorporating frailty we modified the AIME model and tested its predictive performance in 3948 patients with median (IQR) age 67 (28) years. The mean (SD) HFRS was 6.2 (+/-5.9). When the HFRS was divided into three groups, 51% of patients were in the low-risk, 40% in intermediate-risk and 9% in high-risk group. A total of 187 (4.7%) patients experienced one or more medication harm events, including bleeding and severe hypoglycaemia. As a measure of predictive accuracy, the area under the curve (AUC) of the AIME-FRAIL was 0.79 (95% CI: 0.76-0.83), an improvement on the AUC of 0.70 of the original AIME model. Given this improved performance, we plan to conduct a pragmatic impact evaluation of the AIME-FRAIL model in the subacute medical setting (Geriatric and Rehabilitation Unit at PAH). We will use the model to estimate patient risk of medication harm (expressed as a risk score) and pharmacists will then use this score to prioritise their patients for timely and comprehensive medication review, and undertake any other necessary actions to avoid medication harm.
Interventions
For the intervention wards each patient will be reviewed weekly by a research pharmacist or a clinical assistant, who will calculate a patient’s risk score using the AIME-FRAIL tool criteria. Briefly the criteria consist of: 1) patient frailty (as per the Clinical Frailty Scale - CFS), 2) length of stay (LOS) in hospital (> 14 days), 3) Insulin use, 4) Anticoagulants, 5) Antiarrhythmics (according to Australian Medicines Handbook classification), 5) Antipsychotic use, 6) Immunosuppressants use,
For the intervention wards each patient will be reviewed weekly by a research pharmacist or a clinical assistant, who will calculate a patient’s risk score using the AIME-FRAIL tool criteria. Briefly the criteria consist of: 1) patient frailty (as per the Clinical Frailty Scale - CFS), 2) length of stay (LOS) in hospital (> 14 days), 3) Insulin use, 4) Anticoagulants, 5) Antiarrhythmics (according to Australian Medicines Handbook classification), 5) Antipsychotic use, 6) Immunosuppressants use, 7) Antibiotic use, and 8) Opioid use. This tool is a risk prediction tool which predicts risk of the patient having an adverse reaction/ or experiencing harm from their medicines. One point will be given to each ‘yes’ response, and a 0 to each ‘no’ response. The patients’ score will be summed to give their total risk score. Patients with a score above 5 will be highlighted as high-risk, for a comprehensive clinical review of their medications by the ward pharmacist (ideally within 24 hours of admission). We anticipate that to complete the risk tool it will take approximately 15 minutes for each participant (the data will be check for against the patients chart and clinical notes and entered as a 1(=yes) or 0 (=no) for each criteria in an Excel spread sheet to generate the risk score). The score will be calculated at admission and updated weekly for the duration of the patient's hospitalisation. The study database will be audited by the principal researcher to ensure adherence to the protocol and data capture requirements. A random sample of 10% of patients will have risk scores validated by a different member of the research team to ensure accurate data entry and risk score calculations. Two wards of similar patient cohorts and casemix will be included, and one will be allocated to intervention and the other to the control. Patient allocation to wards is not randomised but based on hospital resources and clinical needs. From the two selected wards one will be allocated to the intervention (using the tool) and one to control (usual practice). The ward for intervention will be selected based on resourcing and staff availability and skills to optimise ensure patient safety.
Sponsors
AIME-FRAIL Research Group
Study design
Allocation
Non-randomised trial
Intervention model
Parallel
Primary purpose
Prevention
Masking
Open (masking not used)
Eligibility
Sex/Gender
All
Age
18 Years to No maximum
Healthy volunteers
No
Inclusion criteria
All patients admitted to the two selected wards at the Princess Alexandra Hospital will be eligible for inclusion.
Exclusion criteria
The following exclusion criteria will be applied: • Patients admitted prior to the commencement of the study • Patients who are receiving end of life care • Patients admitted and discharged prior to the research pharmacist being able to calculate their risk score • Patients who are transferred to the acute wards for urgent care (e.g., surgery) within the first 14 days of admission to the GARU wards.
Outcome results
None listed