Factors influencing late diagnosis in patients with terminal cancer.

Sarah Mills
Prof Blair H Smith, Prof Bruce Guthrie, Dr Deans Buchanan
Author institutions: 
University of Dundee, NHS Tayside


The incidence of cancer is increasing, as is the number of people dying from cancer each year in the UK. Late diagnosis of cancer contributes substantially to poor cancer survival rates. The UK has one of the worst records in Europe for both identification of cancer and cancer survival rates. Improving timeliness of cancer diagnosis is a key priority for the NHS. Even for patients with terminal cancer, earlier diagnosis allows earlier initiation of palliative care, which enables better symptom control, and empowers patients to have greater choice in the manner of their death.


This analysis is a retrospective cohort study of NHS Tayside residents who died from cancer between 01/01/2012-30/06/2015. Patients were identified posthumously using General Register Office ‘Cause of Death’ Data. Routinely collected clinical data from all unscheduled care during their last year of life was collected using Community Health Index (CHI) number searches. The CHI is a unique patient identification number used in all clinical encounters. Clinical data was linked to demographic data, including date of cancer diagnosis and date of death.


There were 3,093 people who met the cohort inclusion criteria. Lung cancer was the commonest cancer in the population (19.7%), followed by haematological malignancies (10%), and bowel cancer (9.9%). Two-thirds (66.5%) of patients in this cohort were diagnosed within last year of life. Sadly, over a quarter of the cohort (28.7%) were given a cancer diagnosis less than three months before their death; 13.5% of whom were within 1 month of death at their time of diagnosis, and 3.5% of whom were diagnosed less than a week before their deaths. Gender and age were not significantly associated with receiving a late cancer diagnosis. Rurality (p=0.004) and deprivation (p=0.22) were significantly correlated with receiving late cancer diagnosis. Cancer type was a factor (p=0.002) in determining when patients diagnosis was made relative to their date of death. Use of unscheduled care (GP Out-of-hours and Accident and Emergency departments) was not significantly associated with receiving a late cancer diagnosis.


Multiple patient and practice level factors influence whether patients dying from cancer receive a late diagnosis. Understanding why some people dying from cancer are diagnosed much later than others is vital to determining what interventions could reduce unnecessary delayed diagnoses and allow for more timely diagnosis of terminal cancers, allowing patients to access necessary care and support at the end of life.


Submitted by: 
Sarah Mills
Funding acknowledgement: 
Funding for data collection and analysis costs has been gratefully received from Tayside Oncology Research Committee (TORC) Research Grant, and from a Palliation and the Caring Hospital (PATCH) Scotland National Research Grant. My salary is funded through a Scottish School of Primary Care (SSPC) GP Academic Fellowship.