Does geodemographic segmentation explain differences in route of cancer diagnosis above and beyond person-level sociodemographic variables?

BACKGROUND

Emergency diagnosis of cancer is associated with poorer short-term survival and may reflect delayed help-seeking. Optimal targeting of interventions to raise awareness of cancer symptoms is therefore needed.

METHODS

We examined the risk of emergency presentation of lung and colorectal cancer (diagnosed in 2016 in England). By cancer site, we used logistic regression (outcome emergency/non-emergency presentation) adjusting for patient-level variables (age, sex, deprivation and ethnicity) with/without adjustment for geodemographic segmentation (Mosaic) group.

RESULTS

Analysis included 36 194 and 32 984 patients with lung and colorectal cancer. Greater levels of deprivation were strongly associated with greater odds of emergency presentation, even after adjustment for Mosaic group, which nonetheless attenuated associations (odds ratio [OR] most/least deprived group = 1.67 adjusted [model excluding Mosaic], 1.28 adjusted [model including Mosaic], P < 0.001 for both, for colorectal; respective OR values of 1.42 and 1.18 for lung, P < 0.001 for both). Similar findings were observed for increasing age. There was large variation in risk of emergency presentation between Mosaic groups (crude OR for highest/lowest risk group = 2.30, adjusted OR = 1.89, for colorectal; respective values of 1.59 and1.66 for lung).

CONCLUSION

Variation in risk of emergency presentation in cancer patients can be explained by geodemography, additional to deprivation group and age. The findings support proof of concept for public health interventions targeting all the examined attributes, including geodemography.

Population trends in emergency cancer diagnoses: The role of changing patient case-mix

BACKGROUND

Diagnosis of cancer through an emergency presentation is associated with worse clinical and patient experience outcomes. The proportion of patients with cancer who are diagnosed through emergency presentations has consequently been introduced as a routine cancer surveillance measure in England. Welcome reductions in this metric have been reported over more than a decade but whether reductions reflect true changes in how patients are diagnosed rather than the changing case-mix of incident cohorts in unknown.

METHODS

We analysed 'Routes to Diagnosis' data on cancer patients (2006-2015) and used logistic regression modelling to determine the contribution of changes in four case-mix variables (sex, age, deprivation, cancer site) to time-trends in emergency presentations.

RESULTS

Between 2006 and 2015 there was an absolute 4.7 percentage point reduction in emergency presentations (23.8%-19.2%). Changing distributions of the four case-mix variables explained 19.0% of this reduction, leaving 81.0% unexplained. Changes in cancer site case-mix alone explained 16.0% of the total reduction.

CONCLUSION

Changes in case-mix (particularly that of cancer sites) account for about a fifth of the overall reduction in emergency presentations. This would support the use of adjustment/standardisation of reported statistics to support their interpretation and help appreciate the influence of case-mix, particularly regarding cancer sites with changing incidence. However, most of the reduction in emergency presentations remains unaccounted for, and likely reflects genuine changes during the study period in how patients were being diagnosed.

Risk of Suicide After a Cancer Diagnosis in England: A Population-Based Study

Background: Previous research has identified an increased risk of suicide among cancer patients, however this has not been investigated at a population level in England. Those subgroups of patients most at risk need to be identified to ensure appropriate access to psychological support. Aim: To examine the variation in suicide risk among individuals diagnosed with cancer in England. Methods: We identified 4,453,547 individuals (21 million person-years at risk) aged 18 to 99 years at diagnosis of cancer during 1995 to 2015 from the national cancer registry, and followed them up until 31 August 2017. The outcomes of interest were both suicide and open verdicts (ICD-10 X60-X84, Y87.0, Y10-Y34 [excluding Y33.9, Y87.2]). Population-based expected deaths were as published by ONS [2]. We calculated standardized mortality ratios (SMRs) and absolute excess risks (AERs), and explored variation in suicide risk by cancer type, age at death, sex, deprivation, ethnicity, and years since cancer diagnosis. Results: 2352 cancer patients died by suicide. This was 0.08% of all deaths. The overall SMR for suicide was 1.19 (95% CI 1.14-1.24) and AER per 10,000 person-years was 0.18 (0.13-0.22). The risk was highest among individuals diagnosed with mesothelioma, with a 4.34-fold risk corresponding to 4.00 extra deaths per 10,000 person-years. This was followed by pancreatic (3.94-fold), esophageal (2.53-fold), lung (2.52-fold), and stomach (2.14-fold) cancer (all significantly elevated). Suicide risk was highest in the first 6 months following cancer diagnosis (SMR: 2.64 [2.42-2.89]), but a significantly increased risk persisted for 2 years (SMR: 1.21 [1.08-1.35]). Conclusion: Despite low numbers, the elevated risk of suicide in patients with certain cancers is a concern, representing potentially preventable deaths. The increased risk in the first 6 months after diagnosis, which is consistent with previous studies, highlights unmet needs for psychological support delivered alongside cancer diagnosis and treatment. Our findings suggest a need for improved risk stratification across cancer services, followed by targeted psychological support.

Defining, Measuring and Preventing the Diagnosis of Cancer as an Emergency: A Critical Review of Current Evidence

Background: Many patients with cancer are diagnosed through an emergency presentation, which is associated with inferior clinical and patient-reported outcomes compared with those of patients who are diagnosed electively or through screening. Reducing the proportion of patients with cancer who are diagnosed as emergencies is, therefore, desirable; however, the optimal means of achieving this aim are uncertain owing to the involvement of different tumor, patient and health-care factors, often in combination. Methods: We searched the literature to identify all population-based studies that examined emergency presentation as a diagnosis or independent variable. Results: Most relevant evidence relates to patients with colorectal or lung cancer in a few economically developed countries, and defines emergency presentations contextually (that is, whether patients presented to emergency health-care services and/or received emergency treatment shortly before their diagnosis) as opposed to clinically (whether patients presented with life-threatening manifestations of their cancer). Consistent inequalities in the risk of emergency presentations by patient characteristics and cancer type have been described, but limited evidence is available on whether, and how, such presentations can be prevented. Evidence on patients' symptoms and health-care use before presentation as an emergency is sparse. Conclusion: In this review, we describe the extent, causes and implications of a diagnosis of cancer following an emergency presentation, and provide recommendations for public health and health-care interventions, and research efforts aimed at addressing this underresearched aspect of cancer diagnosis.

Cancer Before, During and After Pregnancy

Background: The occurrence of cancer during pregnancy is uncommon with an incidence rate of ∼1 in 1000 pregnancies. The rate of pregnancy-associated cancer is increasing and this is partly caused by a trend in delaying child bearing to an older age. Aim: With little data in the UK concerning the number of women diagnosed with cancer during pregnancy, the purpose of this study was to compare incidence of cancer in pregnant women to the general female population. Methods: Cancer registry data for England were linked to hospital activity data to establish pregnancy-associated cancers. For this study, women aged 15 to 44 years diagnosed with a malignant cancer between 2012 and 2014 and a pregnancy or delivery code 1 year before or up to 1 year after diagnosis were defined as pregnant women. Age-standardized and age-specific incidence rates of cancer in pregnant women and the general female population in England were compared by 5-year age-group, geographic region of residence, income deprivation quintile and stage of cancer diagnosis. Results: A total of 3272 pregnancy-associated cancers were identified in 2,503,174 pregnancies. The age-standardized incidence rate (ASIR) of cancer in pregnant women was 48% higher than the equivalent ASIR of cancer in the female population aged 15-44 nationally (173 vs 117 per 100,000). This trend of higher incidence of cancer among pregnant women persisted for most regions, ages and stages, and was particularly high in the most deprived quintile. The most common cancers diagnosed around the time of pregnancy were breast (n = 784), melanoma of skin (n = 504), cervical (n = 498), hematologic (n = 286), ovarian (n = 240) and colorectal (n = 188). Comparing the ASIR of cancer in pregnant women with the female population, by site, rates were over 30% higher for breast cancer (55 vs 41 per 100,000 respectively) and around double those for melanoma (26 vs 13 per 100,000). Conclusion: The higher rates of pregnancy-associated cancers compared with the general female population may be due to frequent obstetric examinations which increases the chances of cancer detection. Further work using a more robust maternity dataset would be required to ascertain timing of cancer diagnosis in relation to delivery.

Variation in 'Fast-Track' Referrals for Suspected Cancer by Patient Characteristic and Cancer Diagnosis: Evidence From 670,000 Patients With Cancers of 35 Different Sites

Background: In England, 'fast-track' (also known as 'two-week wait') general practitioner referrals for suspected cancer in symptomatic patients are used to shorten diagnostic intervals and are supported by clinical guidelines. However, the use of the fast-track pathway may vary for different patient groups. Methods:  We examined data from 669,220 patients with 35 cancers diagnosed 2006-2010 following either fast-track or nonfast track primary-to-secondary care referrals using a bespoke English dataset, the 'Routes to Diagnosis' data. We estimated the proportion of fast-track referrals by sociodemographic characteristics and cancer diagnosis and used logistic regression to estimate respective crude and adjusted odds ratios. We additionally explored whether sociodemographic associations varied by cancer. Results:  There were large variations in the odds of fast-track referral by cancer ( P < 0.001). Patients with testicular and breast cancer were most likely to have been diagnosed after a fast-track referral (adjusted odds ratios 2.73 and 2.35 respectively, using rectal cancer as reference); while patients with brain cancer and leukemias least likely (adjusted odds ratios 0.05 and 0.09 respectively for brain cancer and acute myeloid leukemia). There were sex, age and deprivation differences in the odds of fast-track referral ( P < 0.013), which varied in their size and direction for patients with different cancers ( P < 0.001). For example, fast-track referrals were least likely in younger women with endometrial cancer and in older men with testicular cancer. Conclusion: Fast-track referrals are less likely for cancers characterized by nonspecific presenting symptoms and patients belonging to low incidence demographic strata. Interventions beyond clinical guidelines for “alarm” symptoms are needed to improve diagnostic timeliness.

What Proportion of Cancers in the UK and Its Constituent Countries Could Be Prevented? An Updated Analysis

Background: Understanding population-level exposure to cancer risk factors is vital when devising risk-reduction policies. By reducing exposure to cancer risk factors, many cancers could be prevented. But what impact on cancer incidence do these risk factors have? And what proportion of cancers could be prevented if these risk factors are avoided? Aim: The aim of this analysis was to update the estimates of the number and proportion of theoretically preventable cancers in the UK to reflect the changing behavior as assessed in representative national surveys, and new epidemiologic evidence. Separate estimates were also calculated for England, Wales, Scotland, and Northern Ireland because prevalence of risk factor exposure varies between them. Methods: Population attributable fractions (PAFs) were calculated for combinations of risk factor and cancer type with sufficient/convincing evidence of a causal association. Relative risks (RRs) were drawn from meta-analyses of cohort studies where possible. Prevalence of exposure to risk factors was obtained from nationally representative population surveys. Cancer incidence data for 2015 were sourced from national data releases and, where needed, personal communications. Results: Around four in ten (38%) cancer cases in 2015 in the UK were attributable to known risk factors. The proportion was around two percentage points higher in UK males (39%) than UK females (37%). Comparing UK countries, the attributable proportion for persons was highest in Scotland (41%) and lowest in England (37%). Tobacco smoking contributed by far the largest proportion of attributable cancer cases, followed by overweight and obesity, accounting for 15% and 6%, respectively, of all cases in the UK in 2015. Conclusion: Around four in ten (38%) cancer cases in the UK could be prevented. Tobacco and obesity remain the top contributors of attributable cancer cases. Tobacco smoking has the highest PAF because it greatly increases cancer risk and has a large number of cancer types associated with it. Obesity has the second-highest PAF because it affects a high proportion of the UK population and is also linked with many cancer types. Public health policy may seek to reduce the level of harm associated with exposure or reduce exposure levels - both approaches may be effective in preventing cancer. The variation in PAFs between UK countries is affected by sociodemographic differences which drive differences in exposure to theoretically avoidable 'lifestyle' factors. PAFs at UK country level have not been available previously and they should be used by policymakers in the devolved nations to develop more targeted public health measures. This analysis demonstrates the importance of nationally representative exposure prevalence data and cancer registration in informing evidence-based public health policy.

Variation in 'fast-track' referrals for suspected cancer by patient characteristic and cancer diagnosis: evidence from 670 000 patients with cancers of 35 different sites

BACKGROUND

In England, 'fast-track' (also known as 'two-week wait') general practitioner referrals for suspected cancer in symptomatic patients are used to shorten diagnostic intervals and are supported by clinical guidelines. However, the use of the fast-track pathway may vary for different patient groups.

METHODS

We examined data from 669 220 patients with 35 cancers diagnosed in 2006-2010 following either fast-track or 'routine' primary-to-secondary care referrals using 'Routes to Diagnosis' data. We estimated the proportion of fast-track referrals by sociodemographic characteristic and cancer site and used logistic regression to estimate respective crude and adjusted odds ratios. We additionally explored whether sociodemographic associations varied by cancer.

RESULTS

There were large variations in the odds of fast-track referral by cancer (P<0.001). Patients with testicular and breast cancer were most likely to have been diagnosed after a fast-track referral (adjusted odds ratios 2.73 and 2.35, respectively, using rectal cancer as reference); whereas patients with brain cancer and leukaemias least likely (adjusted odds ratios 0.05 and 0.09, respectively, for brain cancer and acute myeloid leukaemia). There were sex, age and deprivation differences in the odds of fast-track referral (P<0.013) that varied in their size and direction for patients with different cancers (P<0.001). For example, fast-track referrals were least likely in younger women with endometrial cancer and in older men with testicular cancer.

CONCLUSIONS

Fast-track referrals are less likely for cancers characterised by nonspecific presenting symptoms and patients belonging to low cancer incidence demographic groups. Interventions beyond clinical guidelines for 'alarm' symptoms are needed to improve diagnostic timeliness.

Cancer-specific variation in emergency presentation by sex, age and deprivation across 27 common and rarer cancers

BACKGROUND

Although overall sociodemographic and cancer site variation in the risk of cancer diagnosis through emergency presentation has been previously described, relatively little is known about how this risk may vary differentially by sex, age and deprivation group between patients with a given cancer.

METHODS

Data from the Routes to Diagnosis project on 749,645 patients (2006-2010) with any of 27 cancers that can occur in either sex were analysed. Crude proportions and crude and adjusted odds ratios were calculated for emergency presentation, and interactions between sex, age and deprivation with cancer were examined.

RESULTS

The overall proportion of patients diagnosed through emergency presentation varied greatly by cancer. Compared with men, women were at greater risk for emergency presentation for bladder, brain, rectal, liver, stomach, colon and lung cancer (e.g., bladder cancer-specific odds ratio for women vs men, 1.50; 95% CI 1.39-1.60), whereas the opposite was true for oral/oropharyngeal cancer, lymphomas and melanoma (e.g., oropharyngeal cancer-specific odds ratio for women vs men, 0.49; 95% CI 0.32-0.73). Similarly, younger patients were at higher risk for emergency presentation for acute leukaemia, colon, stomach and oesophageal cancer (e.g., colon cancer-specific odds ratio in 35-44- vs 65-74-year-olds, 2.01; 95% CI 1.76-2.30) and older patients for laryngeal, melanoma, thyroid, oral and Hodgkin's lymphoma (e.g., melanoma specific odds ratio in 35-44- vs 65-74-year-olds, 0.20; 95% CI 0.12-0.33). Inequalities in the risk of emergency presentation by deprivation group were greatest for oral/oropharyngeal, anal, laryngeal and small intestine cancers.

CONCLUSIONS

Among patients with the same cancer, the risk for emergency presentation varies notably by sex, age and deprivation group. The findings suggest that, beyond tumour biology, diagnosis through an emergency route may be associated both with psychosocial processes, which can delay seeking of medical help, and with difficulties in suspecting the diagnosis of cancer after presentation.

Routes to diagnosis for cancer - determining the patient journey using multiple routine data sets

BACKGROUND

Cancer survival in England is lower than the European average, which has been at least partly attributed to later stage at diagnosis in English patients. There are substantial regional and demographic variations in cancer survival across England. The majority of patients are diagnosed following symptomatic or incidental presentation. This study defines a methodology by which the route the patient follows to the point of diagnosis can be categorised to examine demographic, organisational, service and personal reasons for delayed diagnosis.

METHODS

Administrative Hospital Episode Statistics data are linked with Cancer Waiting Times data, data from the cancer screening programmes and cancer registration data. Using these data sets, every case of cancer registered in England, which was diagnosed in 2006-2008, is categorised into one of eight 'Routes to Diagnosis'.

RESULTS

Different cancer types show substantial differences between the proportion of cases that present by each route, in reasonable agreement with previous clinical studies. Patients presenting via Emergency routes have substantially lower 1-year relative survival.

CONCLUSION

Linked cancer registration and administrative data can be used to robustly categorise the route to a cancer diagnosis for all patients. These categories can be used to enhance understanding of and explore possible reasons for delayed diagnosis.