In recent decades we have seen major improvements in patient safety, but most of these efforts focused on making treatment safer, underplaying the critical importance of accurate and timely diagnosis. The 2015 (US) Institute of Medicine report 'Improving diagnosis in health care' has highlighted diagnostic delays across different diseases (including cancer) as a major problem for contemporary healthcare systems. Motivated by these realisations we will generate evidence to help improve the accuracy and timeliness of diagnosis in patients who present with symptoms in primary care. We will interrogate the rich Clinical Practice Research Datalink (CPRD) data about events that occur before diagnosis (consultations, prescriptions, and investigations) in patients with symptoms, and which of these patients are then diagnosed with cancer or other serious illnesses. Our findings will update and strengthen current evidence about how presenting symptoms can be interpreted. They will provide the basis for updating clinical practice guidelines to support doctors, when they have to decide whether to refer or investigate a patient for suspected cancer or other serious illness, or to actively monitor them.
The overall research objective is to produce population-based evidence about the likelihood of cancer and serious non-neoplastic diseases* among primary care consultees, taking into account their symptoms and, where applicable, their history of repeat consultations, prescriptions, and investigations. We will estimate the predictive values of:
- Different symptomatic presentations for cancer (overall and by major tumour site) and serious non-neoplastic diseases diagnosed within a year after presentation.
- Symptoms combined with information on pre-diagnostic events (such as use of investigations or prescriptions).
This evidence is needed to support clinical decisions about either specialist referral/investigation, or active monitoring (also known as 'safety netting') for patients 'at low but not no-risk'.
We will perform a cohort study, including patients aged 30 years or older with one or more pre-specified symptoms of interest recorded in CPRD between 2007 and 2016. Using primary and secondary care data linked to cancer registration data we will estimate the predictive values (and 95% confidence intervals) of each selected symptom, both for cancer and non-neoplastic disease. Positive predictive values will correspond to the proportion of patients with a given symptom that are diagnosed with a specific outcome (one of the cancers or non-neoplastic diseases of interest) within 1 year since first symptomatic presentation. Similarly, predictive values of different symptom-prescription-investigation combinations will be estimated. We will also examine the role of covariates (socio-demographic factors, comorbidities) possibly influencing predictive values.
*Hereafter we define serious non-neoplastic disease as disease that requires treatment and/or is progressive. An example is inflammatory bowel disease.
The outcomes will be: diagnosis of cancer; diagnosis of serious non-neoplastic disease.
Georgios Lyratzopoulos - Chief Investigator - University College London ( UCL )
Annie Herbert - Corresponding Applicant - University of Bristol
Arturo Gonzalez-Izquierdo - Collaborator - University College London ( UCL )
Bethany Wickramsinghe - Collaborator - University College London ( UCL )
Cristina Renzi - Collaborator - University College London ( UCL )
Edmund Njeru Njagi - Collaborator - University College London ( UCL )
Emma Whitfield - Collaborator - University College London ( UCL )
Gary Abel - Collaborator - University of Exeter
Irene Petersen - Collaborator - University College London ( UCL )
Jessica Kurland - Collaborator - University College London ( UCL )
Kathy Pritchard-Jones - Collaborator - University College London ( UCL )
Matthew Barclay - Collaborator - University College London ( UCL )
Meenakshi (Meena) Rafiq - Collaborator - UCL Hospital
Monica Koo - Collaborator - University College London ( UCL )
Muhammad Qummer ul Arfeen - Collaborator - University College London ( UCL )
Nadine Zakkak - Collaborator - University College London ( UCL )
Rebecca White - Collaborator - University College London ( UCL )
Ruth Swann - Collaborator - University College London ( UCL )
Samantha (Hiu Yan) Ip - Collaborator - University College London ( UCL )
Sara Benitez Majano - Collaborator - London School of Hygiene & Tropical Medicine ( LSHTM )
Sarah Price - Collaborator - University of Exeter
Tra Pham - Collaborator - University College London ( UCL )
William Hamilton - Collaborator - University of Exeter
yangfan Li - Collaborator - University College London ( UCL )
Aradhna Kaushal - Collaborator - University College London ( UCL )
Helen Fowler - Collaborator - University College London ( UCL )
JIAHUI WANG - Collaborator - University College London ( UCL )
Marta Berglund - Collaborator - University College London ( UCL )
Freya Pollington - Collaborator - University College London ( UCL )
HES Admitted Patient Care;HES Diagnostic Imaging Dataset;HES Outpatient;NCRAS Cancer Registration Data;Patient Level Index of Multiple Deprivation