The development of a prediction model to enable clinical (hospital) pharmacists to prioritise older patients at risk of drug related problems

Brady (née Jenkins), Amanda Jane (2025). The development of a prediction model to enable clinical (hospital) pharmacists to prioritise older patients at risk of drug related problems. University of Birmingham. Other

Preview

Brady2025DPharm_Redacted.pdf
Text - Redacted Version
Available under License All rights reserved.
Download (5MB) | Preview

Abstract

This thesis describes the research undertaken to develop a prioritisation tool for Clinical Pharmacists at Sligo University Hospital (SUH). The aim was to develop a tool to prioritise older (medical) patients most at risk of drug related problems (DRPs), and most likely to need Clinical Pharmacist Interventions, for early Clinical Pharmacist review.

A systematic review (SR) was conducted to identify (existing) screening tools and potential risk factors associated with DRPs (or equivalent) which would increase the need for early Clinical Pharmacist review and intervention(/s). Eight medical (/science/healthcare) literature databases were searched and a number of relevant search terms used to identify relevant published primary literature. The search of the eight databases identified 1030 titles and an additional 11 papers were identified from other sources (e.g. search of references for an included study) making a total of 1041 records. Titles, abstracts and, where deemed appropriate, full text articles were reviewed and nineteen publications met the criteria for inclusion in the SR.

From the included studies, potential risk factors for DRPs were identified; individual studies included between 3 and 38 variables. The selection of potential risk factors for the current primary research was based on the overall (collective) findings of the studies included in the SR, whilst considering the practicalities of collecting data on these variables at SUH. These potential risk factors related to patients’ age, gender, number of comorbidities, number of medications taken, antimicrobial therapy (current/recent), falls, high risk medications, renal impairment, hepatic impairment and current/recent infection.

Primary research was conducted at SUH to determine if any of the potential risk factors, identified from the SR, were predictive of Clinical Pharmacist Interventions in older medical in-patients at SUH. Older patients (aged 65/+ years old) who were medically admitted during the four month study period (1st September – 31st December 2020) were prospectively identified by the Clinical Pharmacists providing Clinical Pharmacy Services to the included wards (if they met the inclusion criteria). Data onpatient demographics, the potential risk factors and the outcome (number of Clinical Pharmacist Interventions received at first Clinical Pharmacist review) were collected retrospectively by the Principal Researcher (AB), using the relevant medical notes (retrieved through the Medical Records department) and computerised laboratory results. Data collection and the sample size included were negatively impacted by a number of challenges including the COVID-19 pandemic, delays in launching a new drug kardex and changes in tracking of medical records, and this resulted in delays in data collection and the sample size being less than anticipated with data collection completed for 121 admissions.

The study outcome was the number of Clinical Pharmacist Interventions received at the time of first Clinical Pharmacist review; for the 121 patient admissions included, 246 interventions were made, equivalent to approximately 2 interventions per patient. The most common type of interventions related to medication reconciliation issues (accounting for 68.7% of interventions made) followed by those belonging to the advice/recommendation category (25.6%). The statistical program Stata was used to describe and analyse the results to determine if any of the variables appeared to be predictive for Clinical Pharmacist Interventions. Backward Poisson regression was performed to develop a risk score/model. Two risk factors were found to be predictive of (the number of) Clinical Pharmacists interventions at first review in the sample population, namely number of pre-admission medications (PAMs) taken (p=0.000) and gender, with the female gender found to be associated with a reduction in Clinical Pharmacist Interventions (p=0.013) whilst increasing the number of PAMs taken was associated with an increase in Clinical Pharmacist Interventions.

However, due to the comparatively small sample size (n=121) and without more specialist statistical support it was not possible to conduct internal validation of the developed score (/model) nor determine the goodness of fit of the score. Therefore, further research would need to be conducted to (internally) validate the score, determine its goodness of fit and, if necessary, refine the score before any recommendations can be made regarding its use locally at SUH. Furthermore, external validation would need to be conducted before considering the broader application of the score outside SUH, its generalisability and its potential role in enabling Clinical Pharmacists to prioritise those older medically admitted patients most likely to benefit from early Clinical Pharmacist review.

Type of Work:

Thesis (Other)

Award Type:

Other

Supervisor(s):