Routes to Diagnosis - NCIN Data Briefing
The overarching goal of the National Awareness and Early Diagnosis Initiative (NAEDI) is to promote early diagnosis of cancer and thereby improve survival rates and reduce cancer mortality. To help achieve this we need to better understand the different routes taken by patients to their cancer diagnoses, to examine what effect this has on overall outcomes.
For all patients diagnosed with cancer in 2007 we used existing routinely available data sources to work backwards through their cancer journey to examine the sequence of events that took them to that diagnosis. These routes to diagnosis included through inpatients, outpatients, screening and emergency presentation.
We then examined how the routes to diagnosis vary for different cancer types and by age, sex and deprivation, to highlight differences in relative one-year survival rates.
Cancer registration data from the National Cancer Data Repository is the core data source for the project. The results cover all English patients diagnosed in 2007 with malignant cancer; excluding non-melanoma skin cancer, in situ breast and cervical cancers, and patients with multiple tumours. Datasets were obtained for inpatient and outpatient activity from Hospital Episodes Statistics (HES), Cancer Waiting Times and from cancer registries for screening information.
The analysis takes as a starting point the date of cancer diagnosis. By working backwards from this point, retracing the patient journey through the data, a set of rules has been defined to identify the sequence of events that make up the different routes to diagnosis. From a patient perspective a series of appointments and investigations or procedures were the events that led up to the diagnosis of cancer, regardless of whether the diagnosis was as a result of a suspected cancer referral or an incidental finding.
There are clearly limitations using NHS data which was not specifically generated for this purpose, and a set of assumptions have been used within the algorithm which derives the route. It is important that when examining the results of this work, these factors are taken into account.
The eight routes to diagnosis were:
||Flagged by cancer registry as detected via breast or cervical screening programme
|Two Week Wait
||Urgent GP referrals with a suspicion of cancer
||Routine and urgent referrals where the patient was not referred under the Two Week Wait referral route
||An elective route starting with an outpatient appointment that is either a consultant to consultant referral, other referral, self-referral, dental referral or unknown referral
||Where no earlier information can be found prior to admission from a waiting list, booked or planned
||An emergency route via A&E, emergency GP referral, emergency consultant outpatient referral, emergency transfer, emergency admission or attendance
||Diagnosis by death certificate only
||No data available from inpatient or outpatient HES or from cancer waiting times or screening
Routes to diagnosis
Although there are potential limitations in the data and methodology to assign routes, the analysis shows the proportion of patients diagnosed through each route and the corresponding survival rates.
The table below highlights the wide variation across different cancer types in routes to diagnosis. Across all cancers, 25% of patients are being diagnosed through the Two Week Wait, whilst 23% are presenting as emergencies. The percentage of patients in the unknown route varies by cancer type. Some of these could be private patients and there could be data quality issues. This warrants further investigation.
Routes to diagnosis by cancer type for all malignant diagnoses, excluding C44 (non-melanoma skin cancer) and multiples, in England, 2007
The table has been colour coded using a gradation in intensity to highlight data distribution and variation in the percentages, a darker colour indicates a higher value.
Routes to diagnosis and outcomes
Having understood the proportion of patients through each route, it is also possible to calculate the relative one-year survival for patients first diagnosed through these routes. Relative survival looks at the ratio of the observed survival rate in a group of cancer patients to the expected survival rate in a similar group of people from the general population, where they have been matched by age and sex.
An example of the survival differences, for breast cancer, is shown in the graph below, highlighting the poorer survival for those patients diagnosed through the emergency presentation route, and for women aged 85 and over.
One-year relative survival (%) by diagnosis route and age group, malignant breast cancer, England, 2007
Conclusions and summary
These results show that nationally 23% of newly diagnosed cancer patients came through as emergency presentations. The proportion of emergency presentations varied widely between cancer types (e.g. melanoma 3%; brain and central nervous system 58%) and by age. Patients aged under 25 and patients over 75 were the most likely to present as emergencies. A socio-economic gradient was also observed, with more affluent patients being less likely to present as emergencies.
Importantly, for all cancer types apart from acute leukaemia, one-year relative survival rates were lower for patients presenting as emergencies than for those presenting via other routes, including the Two Week Wait urgent referral route and routine outpatient appointments.
Measurement of emergency hospital presentations of new patients with cancer, which correlates closely with poor one-year survival rates, provides a new indicator for the extent of early/late diagnoses in a population.
Because of the caveats highlighted earlier, it is recommended that an audit of patient-level data is carried out to quality assure the project methods and approach. It is also suggested that the results are updated for later years when these data become available. Recommendations for further analysis can be found in the Technical Supplement.
A detailed Technical Supplement is available to accompany this national work, which was undertaken following piloting of the methodology in the South West. The project approach has relied on using multiple datasets which have their own strengths and weaknesses. The screening data were supplied by cancer registries and data coverage is not consistent across England and are likely to be under-recorded.
The matching of HES data to National Cancer Repository data is incomplete for some London Primary Care Trusts (PCTs). This has impacted on assigning the correct routes to diagnosis for nine PCTs. Investigations into why this happened are ongoing, and rather than publish results based on this data the particular PCTs where the problem existed have been excluded from the analysis.
The algorithm which derives each route uses new methods of data analysis, and can inevitably be improved with further, more localised data.
Detailed examination of data quality issues can be found in the Technical Supplement.
This project has been led and managed on behalf of NCIN by Lucy Elliss-Brookes, Associate Director Commissioning and Cancer Intelligence, Avon Somerset and Wiltshire Cancer Services.
Professor Sir Mike Richards, National Cancer Director, has acted as clinical adviser to the project. Analytical and technical support, input and leadership have been provided by members of the South West Public Health Observatory, in particular Dr Julia Verne, Tariq Malik, Alexander Ives, Matt Greenslade, Luke Hounsome, Andy Pring, Davidson Ho and Roy Maxwell.
Find out more
A technical supplement, which describes the methods, algorithms and data quality issues in more detail accompanies this data briefing.
Other useful resources within the NCIN partnership: