Article Text
Abstract
Background Poorer colorectal cancer survival in the UK than in similar countries may be partly due to delays in the care pathway. To address this, cancer waiting time targets were established. We investigated if socio-demographic inequalities exist in meeting cancer waiting times for colorectal cancer.
Methods We identified primary colorectal cancers (International Classification of Diseases, Tenth Revision C18–C20; n=35 142) diagnosed in the period 2001–2010 in the Northern and Yorkshire Cancer Registry area. Using multivariable logistic regression, we calculated likelihood of referral and treatment within target by age group and deprivation quintile.
Results 48% of the patients were referred to hospital within target (≤14 days from general practitioner (GP) referral to first hospital appointment); 52% started treatment within 31 days of diagnosis; and 44% started treatment within 62 days of GP referral. Individuals aged 60–69, 70–79 and 80+ years were significantly more likely to attend a first hospital appointment within 14 days than those aged <60 years (adjusted OR=1.23, 95% CI 1.12 to 1.34; adjusted OR=1.19, 95% CI 1.09 to 1.29; adjusted OR=1.30, 95% CI 1.18 to 1.42, respectively). Older age was significantly associated with lower likelihood of starting treatment within 31 days of diagnosis and 62 days of referral. Deprivation was not related to referral within target but was associated with lower likelihood of starting treatment within 31 days of diagnosis or 62 days of referral (most vs least: adjusted OR=0.82, 95% CI 0.74 to 0.91).
Conclusions Older patients with colorectal cancer were less likely to experience referral delays but more likely to experience treatment delays. More deprived patients were more likely to experience treatment delays. Investigation of patient pathways, treatment decision-making and treatment planning would improve understanding of these inequalities.
- Cancer
- Inequalities
- Registers
Statistics from Altmetric.com
INTRODUCTION
Survival from many cancers is lower for patients in the UK than in similar countries.1 For example, 5-year age-standardised net survival in the UK for patients diagnosed in 2010–2014 was 58.9% for colon cancer and 62.1% for rectal cancer, lower than in Norway, Ireland, Denmark, New Zealand, Canada and Australia, where rates of 65–71% were reported.2 This pattern has been evident for colorectal (and other) cancers diagnosed since the late 1970s. While UK survival rates have increased in recent years, improvements have also been seen elsewhere,3 so the survival gap has not narrowed.4
Evidence suggests that this survival gap is due, at least in part, to later-stage diagnosis alongside inequalities in access to optimal treatment.5 6 Both diagnosis and treatment can be subject to delays. Indeed, delays can exist across the entire cancer care ‘pathway’ from the first symptom noted by the individual through diagnosis to treatment initiation.7 8 These delays can be attributed to both patient behaviour and limitations in the healthcare provider or system.8
The NHS Cancer Plan for England, published in 2000, recognised the possibility of healthcare provider or system delays in cancer and established, for the first time, cancer waiting time targets.9 A target of 14 days between urgent referral (usually from a general practitioner (GP)) with symptoms suggesting cancer and first assessment by a specialist was established in 2000. For those subsequently diagnosed with cancer, treatment should start within 62 days of referral. Since 2005, there has also been a target of 31 days between the date of the decision-to-treat and start of treatment.
Equity and timeliness are key features of high-quality healthcare.10 Despite this, persistent and pronounced socio-demographic inequalities in colorectal cancer treatment receipt have been reported. Population-based studies show that older patients and those of lower socioeconomic status (SES) are less likely to receive cancer-directed, or guideline-recommended, treatment (see, eg, Vulto et al, 11 Carsin et al,12 Lejeune et al,13 Hayes et al 14).
It is less clear whether similar inequalities exist in relation to the cancer waiting time targets or, more generally, in timeliness of referral, diagnosis or start of treatment. In the only UK study, of lung cancer, older patients (aged 80+ years) were 20% less likely to start treatment within 31 days than younger patients (<60 years).15 North American colorectal cancer studies have reported that older patients have longer times between referral and diagnosis or first treatment.16 17 As regards SES, in Denmark, among a mixed group of patients with cancer, those with higher household incomes experienced shorter healthcare system delays18 while modestly sized, non-population-based, studies in Germany, Jordan and Spain have reported socioeconomic patterning in treatment delays for patients with colorectal cancer.19–21 Evidence from the UK is lacking.
We undertook a large, population-based study to investigate age and socioeconomic inequalities in whether the cancer waiting time targets are met for colorectal cancer in England.
METHODS
Setting and data sources
Data were abstracted on primary colorectal cancers (International Classification of Diseases, Tenth Revision C18–C20) diagnosed in the period 2001–2010 from the Northern and Yorkshire Cancer Registry (NYCRIS), which covers a population of 6.8 million resident in the North East and most of the Yorkshire and the Humber regions. Cancer notification is mandatory; hospitals report cases directly to NYCRIS and this is supplemented by manual data collection. The area comprises some densely population urban centres and large rural expanses. Cancer care is almost entirely provided in National Health Service (NHS) hospitals. At the time of the study, the area had three cancer networks, each with several cancer units and a cancer centre providing integrated services. The Bowel Cancer Screening Programme, offering home-based faecal occult blood tests, commenced in 2006.
Data were available on age at diagnosis, deprivation category of area of residence at diagnosis (a proxy for SES measured using the rank of the income domain of the Index of Multiple Deprivation (IMD)),22 sex, diagnosis date, tumour site, stage and cancer-directed treatments (surgery, radiotherapy, chemotherapy) received by 31 December 2011. Where available, information was extracted on dates of referral to hospital by a GP, first hospital appointment and first cancer-directed treatment. Each record was linked to NHS Hospital Episode Statistics (HES) data to provide information on comorbidities. Cancers registered from death certificates only were excluded (n=644) as were those with no HES link (n=8950), leaving 35 142 patients in the data set.
The study was approved by the NRES Committee East of England REC on 13 December 2011 (reference 11/EE/0537) and performed in accordance with the Declaration of Helsinki.
Outcome variables
Three binary outcome variables were created representing whether the interval (1) between referral and first hospital appointment was ≤14 days (henceforth ‘referral interval’); (2) between diagnosis and first treatment was ≤31 days (‘diagnosis–treatment interval’); and (3) between referral and first treatment was ≤62 days (‘referral–treatment interval’). Cases with implausible intervals (eg, treatment date before referral date) were excluded. The referral interval was created for patients for whom dates of GP referral and first hospital appointment (surrogate for date of first assessment by a specialist) were recorded (n=19 798). The diagnosis–treatment interval was created for patients for whom dates of cancer diagnosis (surrogate for the date of the decision-to-treat) and first cancer-directed treatment were recorded (n=29 445). The referral–treatment interval was created for patients for whom dates of GP referral and first cancer-directed treatment were recorded (n=17 622). Where dates were not recorded for patients, we assumed they did not apply (eg, the patient was not referred by a GP).
Explanatory variables
The two main explanatory variables were age (categorised as <60, 60–69, 70–79 and 80+ years) and deprivation category (grouped into fifths, based on the national distribution, with the lowest fifth (‘1’) representing least deprived areas and ‘5’ representing most deprived). IMD2007 was used for patients diagnosed in the period 2001–2006 and IMD2010 for those diagnosed 2007 onwards.
To explore whether timeliness had changed over time, year of diagnosis was grouped: 2001–2005/2006–2010. Cancer stage, assigned using TNM, was categorised as I, II, III, IV, unstaged or staged post-treatment. Cancer site was grouped as colon or rectum. Comorbidities influence treatment decisions and prognosis in colorectal cancer.23 24 NYCRIS provided a weighted comorbidity score based on the Charlson comorbidity index (CCM),25 using the number of relevant in-patient admissions (excluding metastatic cancer) recorded in HES in the 3–18 months before the cancer diagnosis (online supplemental table S1). This score was categorised based on the observed distribution: none/1–2/3+ comorbidities. Patients were categorised by the first cancer-directed treatment they received (cancer-directed surgery/radiotherapy/chemotherapy).
Supplemental material
Statistical analyses
For each interval (referral; diagnosis–treatment; and referral–treatment), the number and percentage of cases referred and treated within target were calculated overall and by age group, deprivation category, and for the other variables. Logistic regression models were used to estimate the odds of referral or treatment within target by age group and deprivation category with and without adjustment for other variables. The three outcomes were modelled separately. We considered variables that were significant in univariate analyses for inclusion in multivariable models. We took care to avoid multicollinearity and also eliminated variables that caused poor model fit. Cancer site and type of first treatment are strongly correlated; we therefore adjusted the referral interval model for site (as this is more pertinent to diagnosis) and the diagnosis–treatment and referral–treatment interval models for first treatment. A variable categorising referral interval (≤14/>14 days) was considered for inclusion in the referral–treatment model. Likelihood ratio tests were used to determine the overall statistical significance of each variable in the final models. Final models had adequate fit.
RESULTS
Table 1 shows socio-demographic and clinical characteristics of the 35 142 cases. A total of 19 798 (56%) cases were referred by a GP to hospital and were included in the referral interval analysis; 29 445 (84%) received cancer-directed treatment and were included in diagnosis–treatment interval analysis; and 17 622 (50%) were referred by a GP and had cancer-directed treatment and were included in the referral–treatment interval analysis.
Sixteen per cent of the colorectal cancers were aged ≤60 years at diagnosis, 24% were 60–69 years, 34% were 70–79 years and 25% were 80+ years. Slightly less than one-fifth were resident in the least deprived areas, 57% were male, 61% had colon cancer, 35% had stage I or II disease and 13% had comorbidities. In general, compared to this full data set, the cases included in the referral interval, diagnosis–treatment interval and referral–treatment interval analyses were slightly younger and more often earlier stage, but the distribution across deprivation categories was similar.
Referral interval
Overall, 48% of the cases referred to hospital by a GP had their first hospital appointment within 14 days. The median referral interval was 16 days (IQR 9–37 days) (online supplemental table S2).
Supplemental material
Likelihood of attending a first hospital appointment within 14 days of GP referral is shown in table 2. In adjusted analysis, individuals aged 60–69, 70–79 and 80+ years were significantly more likely to attend a first hospital appointment within 14 days than those aged <60 years (OR=1.23, 95% CI 1.12 to 1.34; OR=1.19, 95% CI 1.09 to 1.29; OR=1.30, 95% CI 1.18 to 1.42, respectively). No significant association between timely first hospital appointment and deprivation was found, although the likelihood of timely referral was (non-significantly) 5% lower in those resident in the most, compared to the least, deprived fifth. Other factors significantly associated with being more likely to have timely referral were as follows: having rectal cancer and having disease that was more advanced than stage I.
Diagnosis–treatment interval
Fifty-two per cent of cases started treatment within 31 days of cancer diagnosis. The median diagnosis–treatment interval was 30 days (IQR 5–49 days) (online supplemental table S2).
Older age was significantly associated with lower likelihood of starting treatment within 31 days of diagnosis (table 3); after adjustment those aged 60–69, 70–79 and 80+ years had reduced odds compared to those aged <60 years (OR=0.74, 95% CI 0.68 to 0.80; OR=0.68, 95% CI 0.63 to 0.73]; OR=0.77 , 95% CI 0.71 to 0.84, respectively). Deprivation was significantly associated with lower likelihood of starting treatment within target. Those resident in deprivation categories 4 and 5 were 9% less likely to receive timely treatment than those resident in deprivation category 1. Other variables significantly associated with being less likely to start treatment within 31 days of diagnosis were being male and having chemotherapy and/or radiotherapy as the first treatment. More advanced disease was significantly associated with starting treatment within 31 days.
Referral–treatment interval
Overall, 44% of the cases referred by a GP started treatment within 62 days of referral. The median referral–treatment interval was 61 days (IQR 49–105 days) (online supplemental table S2).
In multivariable analyses, compared to those aged <60 years, the two oldest age groups had 10% and 12% lower odds, respectively, of starting treatment within 62 days (table 4). Likelihood of starting treatment within target decreased with increasing deprivation and was 0.82 (95% CI 0.75 to 0.91) for those in the most, compared to the least, deprived category. Other factors significantly associated with not starting treatment within 62 days of GP referral were: a referral interval of ≤14 days, receiving radiotherapy and/or chemotherapy as the first treatment and more advanced stage disease.
DISCUSSION
During the period 2001–2010, only 48% of the patients with colorectal cancer referred by a GP attended a first hospital appointment within the national target of 14 days, 52% started treatment within 31 days of diagnosis and 44% waited ≤62 days between GP referral and start of treatment. Older age was significantly positively associated with having a first hospital appointment within target but negatively associated with starting treatment within 31 days of diagnosis or 62 days of GP referral. Greater deprivation was significantly associated with being less likely to start treatment within 31 days of diagnosis and 62 days of GP referral; the effect was more pronounced for the referral–treatment interval and remained after adjustment for having a first hospital appointment within 14 days of GP referral.
Strengths and limitations
To our knowledge, this is the first UK population-based study of inequalities in timeliness of colorectal cancer diagnosis and treatment. Although the data source (NYCRIS) has a high level of completeness,14 it covers only the north of England, potentially limiting generalisibility. Nationally, in the period 2006–2008, 65% of colorectal cancers were diagnosed through a route originating from a GP referral; the remainder were emergency presentations (26%), screen-detected (2%), identified as having cancer only at the time of death (1%) or had an unknown route (6%).26 In our analysis, 56% (19 798/35 142) of the cases had GP referral and first cancer-directed treatment dates, suggesting that some cases which originated at the GP may be missing from our analyses, perhaps because referral dates were not recorded either in medical records or by the Registry. Since our focus was comparisons within the group of patients referred by GPs, this is unlikely to impact internal validity. Eighty-four per cent (29 445/35 142) of the study population were included in the diagnosis–treatment interval analysis, broadly consistent with the proportion of colorectal cancers reported as having cancer-directed treatment in other European population-based series.12 Although we accounted for comorbidity, the CCM as applied to HES data captures information only on conditions requiring an in-patient stay, meaning we probably underestimated comorbidity. SES was assessed using an area-based measure so may be subject to some misclassification.
GPs can refer patients with suspected cancer symptoms as urgent or non-urgent (standard), and the 14-day target technically applies only to urgent referrals. A weakness of the data is that we were unable to distinguish between urgent and non-urgent referrals. In addition, we did not have access to data after 2010. It is important to determine whether socio-demographic variations remain, especially given that the national operational standards for timely referral and treatment are frequently missed.27
The 31-day target refers to the time from decision-to-treat. NYCRIS does not record this date, so we used date of diagnosis as a proxy. These two dates may differ for some patients, meaning it is likely we have slightly overestimated the length of the diagnosis–treatment interval. Whether this difference will be differential by socio-demographic factors is unclear. In addition, no information was available on any cancer-directed treatments other than surgery, chemotherapy or radiotherapy.
Interpretation of findings
The ‘waiting time paradox’, where people who are more obviously unwell are referred more quickly but have worse outcomes, is well recognised in cancer.28 Older paients with cancer more often have multiple comorbidities.29 Thus, one explanation for the relationship between older age and more rapid referral is that patients who have multiple health problems (and seem sicker) are more likely to be urgently referred. However, we adjusted for comorbidities, although imperfectly, so it seems unlikely that this is the full explanation. Older patients may be more likely to present with ‘alarm symptoms’ (eg, rectal bleeding). New-onset alarm symptoms are more predictive of cancer in older people.30 Because of this, and because colorectal cancer incidence increases with age, GPs’ index of suspicion of cancer may be higher in older than younger patients, hence stimulating urgent referral of older patients. ‘Direct to test’ services—where younger, fit, patients go straight from the GP to colonoscopy, while older patients (especially the oldest old) wait for an initial specialist appointment, then may have a less invasive test (such as computed tomographic colonography or barium enema) before cancer diagnosis—are also likely to result in a shorter interval from referral to cancer diagnosis in younger patients.
There was a modest (although non-significant) association between greater deprivation and lower likelihood of being seen at hospital within 14 days of GP referral, consistent with UK lung cancer findings.15 In general, lower SES is associated with suboptimal healthcare use, perhaps due to more limited health literacy translating into difficulty navigating healthcare systems.31 Research on barriers to medical help-seeking around cancer shows people of lower SES status are less confident talking to the GP about symptoms,32 raising the possibility that patients of higher SES are better able to articulate symptoms and problems in a way that stimulates urgent referral.
The diagnosis–treatment interval is the window within which treatment decisions are made and patients readied for treatment. Assessing patient suitability for treatment may take longer in older patients particularly if geriatric assessments or referral to other specialities (eg, cardiology) are needed; this information is not generally recorded by cancer registries. Limitations in physical functioning and/or nutritional status, such as sarcopenia, malnourishment and iron-deficiency anaemia, are more common in older patients with colorectal cancer.33 34 Patients may need interventions to address these before undergoing surgery. This could delay start of treatment and explain the observed negative association between increasing age and likelihood of starting treatment within 31 days.
The observed association between greater deprivation and lower likelihood of starting treatment within 31 days of diagnosis, or 62 days of referral, is consistent with associations between delayed treatment and lower education and general (vs private) health insurance in Spain21 and Germany,19 respectively. Patients with colorectal cancer with lower SES more often have comorbid conditions,29 which may lengthen treatment planning. We adjusted for comorbidities, but residual confounding is possible.
The increasing emphasis on patient-centred care in the NHS has led to more shared and informed decision-making. Past work shows that lack of comprehension by patients (as a result of poor health literacy) of risks and benefits of elective surgery can lead to decision conflict.35 We might, therefore, speculate that, as a result of limited health literacy, patients with colorectal cancer from more deprived areas may struggle with treatment decision-making, and this could lead to delays; however, we are not aware of any empirical data supporting this.
Implications
Although evidence is inconsistent on whether more rapid referral influences stage at diagnosis and more rapid start of treatment influences survival (at least when relatively short time windows are considered),36 37 timeliness and equity are key elements of high-quality healthcare.10 Moreover, the time waiting for a hospital appointment to investigate suspected cancer or for treatment to start in those with a confirmed diagnosis is associated with considerable patient anxiety and distress.38 39 Therefore, given that the NHS has set cancer waiting time targets, the system should seek to provide all patients with a definitive diagnosis and to start treatment in a timely manner.
The question then arises of how to achieve this. Some encouraging findings have been reported for patient navigation programmes, which are intended to promote access to timely cancer diagnosis and treatment.40 This raises the possibility that allocating some patients (perhaps those who are older, more deprived, or have inadequate health literacy) a designated navigator, from within the hospital care team, could help shorten the time to treatment and reduce or eliminate some of the disparities observed here.
However, there may be disadvantages of rapid diagnosis or treatment commencement; Canadian studies show that patients with colorectal cancer who waited longer for diagnosis or treatment more often received higher-quality healthcare.41 42 While we were unable to measure clinical aspects of quality of care, these findings indicate that relationships between timeliness, utilisation and care quality are complex.
The findings here—both in terms of the proportions of patients first seen at hospital or treated within target, and the socio-demographic patterning—raise many questions. Research is needed to quantify the following: how many, and which, patients need additional time to make treatment decisions; how often more time is needed for treatment planning (eg, for complex patients) or prehabilitation; and how often failures to meet targets are due to lack of capacity, lack of equipment or delays reporting test results. Recent qualitative research in Northern England among patients with upper gastrointestinal cancer diagnosed through the 2-week wait pathway found that they often have complex diagnostic journeys visiting multiple hospital and primary care sites and may wait considerable time for results of tests and investigations.43 Further research is needed to shed light on patient pathways, treatment decision-making and treatment planning overall and for different patient subgroups. This would help better understand inequalities like those reported here, why the system fails to meet the national operating standards27 and what changes are needed if these standards are to be met.
The 2000 NHS Cancer Plan stated ‘Where patients wait longer, this should be because of the needs of the diagnostic process or their personal choice, not because of built-in delays in the system of care.’ 9 Addressing the socio-demographic inequalities observed here in relation to timely referral and start of treatment for colorectal cancer may require changes in care pathways.
What is already known on this subject
Pronounced socio-demographic inequalities in receipt of colorectal cancer treatment have been demonstrated in many healthcare settings, including the UK.
It is unclear whether similar inequalities exist in timeliness of referral or start of treatment.
What this study adds
In this large population-based study, older patients with colorectal cancer were significantly less likely to experience referral delays but more likely to experience treatment delays.
The most deprived patients were significantly more likely to experience treatment delays.
Timeliness and equity are key features of high-quality healthcare. Implementation of strategies to reduce these inequalities could reduce psychological distress and improve patient experiences of cancer care.
Acknowledgments
We thank staff at the former NYCRIS for the preparation and provision of the data set used in the study.
REFERENCES
Footnotes
Contributors LH undertook the data analyses and co-wrote the paper; LS: advised on the data analysis and co-wrote the paper; JA, IM, LF, MH and MW contributed to the study design, supervision and interpretation of data, and commented critically on manuscript drafts.
Funding This study was supported by the National Institute for Health Research (NIHR) School for Public Health Research (grant reference numbers: SPHR-SWP-AGP-PR3 and PD-SPH-2015), of which Fuse is a member. Fuse is a UK Clinical Research Collaboration (UKCRC) Public Health Research Centre of Excellence. Fuse received funding from British Heart Foundation, Cancer Research UK, National Institute of Health Research, Economic and Social Research Council, Medical Research Council, Health and Social Care Research and Development Office, Northern Ireland, National Institute for Social Care and Health Research (Welsh Assembly Government) and the Wellcome Trust, under the auspices of the UKCRC.
Disclaimer The views expressed are those of the author(s) and not necessarily those of the NIHR or the Department of Health and Social Care. At the time of the data request, the SPHR was a partnership between the Universities of Sheffield, Bristol, Cambridge; UCL; The London School for Hygiene and Tropical Medicine; The Peninsula College of Medicine and Dentistry; the LiLaC collaboration between the Universities of Liverpool and Lancaster and Fuse.
Competing interests None declared.
Patient consent for publication Not required.
Provenance and peer review Not commissioned; externally peer reviewed.
Data availability statement The data used in the study are not owned by the authors. They were released by the data controller to the authors solely for the purpose of the research study and the authors do not have permission to share them with others. Public Health England is the current data controller and data for the same population may be requested through PHE’s Office of Data Release.
Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.