The ageing population in the UK means NHS resources are being stretched by the increasing demand for medical treatments. Now, with an ever-growing waiting list nationally, an effective prioritisation and efficient pathway is needed to ensure safe and quick treatments. Specifically, this project will look at orthopaedics where tens of thousands of planned procedures happen annually. 30% of GP consultations are Musculoskeletal (MSK) complaints, with the majority being knee and hip problems. Patients suffering with MSK problems that require surgery have major reduction to their quality of life. The pathways to treatment vary based on the patient’s health information, such as comorbidities and age, as well as the local access to specialist care. We need to understand any bottlenecks and if we can cluster groups of patients who show a similar journey to give a more targeted and efficient pathway. With Coronavirus halting NHS non-critical procedures during the early stages of the pandemic, such as orthopaedics hip and knee replacements, this offers an ideal time to review the pre-COVID patient flow. Operational improvements can be found to aid reductions in both waiting time and resource management as services are returned to normal.
Healthcare aims:
1. To map out patient journeys throughout the healthcare system. Using traditional statistical, artificial Intelligence (AI) and machine learning (ML) methods to extract knowledge on patient journeys and evaluate hospital trust performance.
2. To see how hospital trusts, Clinical Commissioning Groups (CCGs) and healthcare regions performance compares with their differing orthopaedic referral protocols.
A retrospective cohort study utilising CPRD and HES (A&E, APC, DID, OP) linkage will allow us to understand the patient’s pathway throughout their orthopaedic journey. This study will focus on adults who present with symptoms, diagnosis, and treatments between March 2013 and 2018. Using validated codes describing Total Hip or Knee Replacements (THR, TKR) surgeries as well as symptoms, diagnosis, treatments will give us the patient journey from symptom/diagnosis presentation to surgery and outcome. A standard pathway will include a cycle of GP and consultant visits, with the possibility of scans and medical tests, until the patient is accepted for surgery. Pathways at a patient level are subject to a high amount of variation due to the personalisation of care. The Orthopaedic Get It Right First Time Report highlights the “undesirable variation in practice around the country”. Therefore, a prominent outcome to evaluate throughout this project will be the analysis of pathway variations between Clinical Commissioning Groups (CCGs) and regions. With the inclusion of key factors leading to hip or knee replacement surgery, such as age and IMD, combined with the mapped patient journeys, this project will explore ML methods of patient clustering. Methods will include traditional machine learning clustering algorithms such as K-means, optimising Maximum Likelihood Estimation as well as deep learning approaches including autoencoders, deep matrix factorization, and graph embedding (graph2vec). Clustering aims to achieve groupings of patients, which will provide easier treatment planning and pathway prediction allowing the use of simulation methods to optimise patient journeys. Key metrics can be estimated, such as diagnosis to treatment time. These metrics will be calculated at regional levels, comparing the variance between regions, and identifying factors that cause the imbalances. This project will provide a deeper understanding of patient journeys in orthopaedics and how they vary between regions of England.
Primary outcome: Diagnosis and Referral To Treatment (DTT/RTT) time.
Secondary outcomes: Conversion rate from GP referrals to surgery; Waiting times during the clinical pathway; Number of patients completing hip or knee replacement surgeries per unit of time; Patient short-term outcomes; General financial costings of healthcare usage.
Definitions of the key primary and secondary outcomes are included under the section on “Exposures, Outcomes and Covariates”.
Alex Bottle - Chief Investigator - Imperial College London
William Plumb - Corresponding Applicant - Imperial College London
Alex Bottle - Collaborator - Imperial College London
Alex Liddle - Collaborator - Imperial College London
GIULIANO CASALE - Collaborator - Imperial College London
Mark Cunningham - Collaborator - Imperial College London
HES Accident and Emergency;HES Admitted Patient Care;HES Outpatient;Patient Level Index of Multiple Deprivation;CCG Pseudonyms;Rural-Urban Classification