Back pain affects people of all ages and is a common reason to consult a GP. Most back pain can be managed by education, reassurance, analgesic medicines, or other therapies. Occasionally, back pain may be a symptom of undiagnosed cancer. Research shows that 1 in 200 people who consult their GP with a new episode of back pain have undiagnosed cancer. GPs must therefore decide if a patient can be managed conservatively or needs further investigation for cancer, as early detection can lead to improved outcomes. We seek to discover new ways to assist GPs in this decision-making process.
Our study will use anonymised patient records to collect information from the clinical consultation for a new episode of back pain over 22 years (2000 – 2022). We will work out the chances of having a diagnosis of cancer over the following 1 year. This will allow us to:
a) improve our understanding of back pain presenting to general practice,
b) estimate what proportion of people with back pain have a diagnosis of cancer within the following year,
c) discover to what amount additional clinical features from the patient’s history and examination indicate cancer – for example fever, pain at rest, unexpected weight loss, or a past history of cancer.
The results of our research will improve guidance for GPs when performing the initial assessment of people with back pain.
---------------------------
The objective is to inform strategies to screen for cancer in patients (≥ 18 years) who present to primary care with back pain within the CPRD network of general practice registered between January 1st 2000 and December 31st 2022.
Aims
1. What is the association between a new episode of back pain presenting to primary care and cancer diagnosis?
2. In people with a new episode of back pain, what is the diagnostic utility of including clinical features from patient profile, history and physical examination to detect cancer within the following 12 months?
Methods
A cohort analysis of the CPRD 2000 – 2022 will be undertaken to 1) Describe how often people present with a new episode of back pain and of these, have cancer diagnosis within the following 12 months; and 2) Calculate the diagnostic predictive value of additional clinical features for cancer. Cancer diagnoses will be obtained from the NCRAS, HES, and ONS (if related to death). Cox modelling will be used to assess the association between back pain and other clinical features with a subsequent diagnosis of cancer within 12 months. Adjusted measures of association will be calculated by including co-variates from the patient profile including age, sex, ethnicity, deprivation, and comorbidity. A 12-month cancer-risk prediction model in patients with back pain will be developed using clinically relevant covariates and covariates with high predictive ability. Performance will be evaluated using diagnostic accuracy measures including sensitivity, specificity, positive and negative predictive values. Measures of overall performance will include area under the receiver operating characteristic curve, D-statistic, Brier score, R-squared, calibration slope, and calibration plots.
The screening model will inform the creation of clinical decision rules to help GPs decide if a new episode of back pain can be managed conservatively or requires referral for cancer investigation.
Prevalence of back pain (Aim 1.1); diagnosis of cancer within the following 12 months (all other Aims).
---------------------------
Pradeep Virdee - Chief Investigator - University of Oxford
Aron Downie - Corresponding Applicant - Macquarie University
Arianne Verhagen - Collaborator - Not from an Organisation
Brian Nicholson - Collaborator - University of Oxford
Christopher Maher - Collaborator - University Of Sydney
Clare Bankhead - Collaborator - University of Oxford
Cynthia Wright Drakesmith - Collaborator - University of Oxford
Gustavo Machado - Collaborator - University Of Sydney
Hazel Jenkins - Collaborator - Macquarie University
Margaret Smith - Collaborator - University of Oxford
Mark Hancock - Collaborator - Macquarie University
Michael Swain - Collaborator - Macquarie University
Peter Stubbs - Collaborator - University of Technology Sydney
Richard Hobbs - Collaborator - University of Oxford
Simon French - Collaborator - Macquarie University
Subhashisa Swain - Collaborator - University of Oxford
Pradeep Virdee - Collaborator - University of Oxford
HES Accident and Emergency;HES Admitted Patient Care;HES Outpatient;NCRAS Cancer Registration Data;No additional NCRAS data required;ONS Death Registration Data;Patient Level Index of Multiple Deprivation