Influence of age on surgical treatment and postoperative outcomes of patients with colorectal cancer in Denmark and Yorkshire, England

Denmark and Yorkshire are demographically similar and both have undergone changes in their management of colorectal cancer to improve outcomes. The differential provision of surgical treatment, especially in the older age groups, may contribute to the magnitude of improved survival rates. This study aimed to identify differences in the management of colorectal cancer surgery and postoperative outcomes according to patient age between Denmark and Yorkshire.


INTRODUC TI ON
Survival rates for patients with colorectal cancer have been shown to vary across Europe [1,2]. In particular, it has been demonstrated that survival rates in Denmark and England were lower in the 1990s and early 2000s than in many other countries with comparable populations and health systems [2,3]. In response, both these countries have instigated interventions to improve colorectal cancer outcomes.
While there are common areas in which interventions have been implemented in both countries, such as the introduction of, and training in, total mesorectal excision for rectal cancer [4,5], there have been some differences in the approaches taken. In Denmark, these interventions have included detailed reviews of readily available observational data to quantify patterns of practice, identifying areas of concern and then focusing action to improve care [6]. There is strong evidence that this has radically changed practice and has coincided with improved outcomes. For example, 30-day postoperative mortality rates have fallen dramatically and survival rates are now more comparable with those of neighbouring Scandinavian countries [7]. Danish 5-year net survival improved from 49% to 66% for colonic cancer and from 48% to 69% for rectal cancer, compared with corresponding increases of 47% to 59% and 48% to 62% in the UK [8].
In 2016, a similar data-driven programme was implemented in Yorkshire, England. It aimed to quantify, in depth, the patterns of care and outcomes [9]. Interventions can then be developed and deployed by Yorkshire's clinical colorectal cancer community to try to eliminate any disparities in quality of care and improve outcomes. Understanding how the surgical management of patients in the region compares with that in populations with similar demographics is a key element of developing these interventions. The UK and Denmark have similar populations, life expectancy, Healthcare Access and Quality Index, smoking rates and alcohol consumption [10,11].
Major surgical resection to remove the tumour and surrounding tissue, i.e. bowel resection including regional lymphadenectomy, is the mainstay of treatment for colorectal cancer. It has recently been suggested that observed survival differences may stem from differences in patient selection for surgical resection, especially in older age groups [12]; if more patients undergo a potentially curative treatment then this could lead to more patients surviving in the longer term. Although survival following surgery may be decreased in the older age groups, longer-term survival may be comparable to that of younger patients [13]. It has also been demonstrated in England that older rectal cancer patients selected for surgery have comparable outcomes to their younger counterparts [14].
Given the recent substantial improvements in survival that have been observed following the clinical interventions in Denmark, comparisons of surgical practice and postoperative outcomes between Denmark and Yorkshire should help to identify areas for improvement. Therefore, this study aimed to identify any differences in the approach to colorectal cancer resection according to patient age between Denmark and Yorkshire, and to investigate the potential impact of these on postoperative mortality and survival rates.

ME THOD
This was a retrospective population-based study, that included firsttime primary colorectal cancers (ICD-10; C18-C20, excluding ma-  [15][16][17]. The DCCG captures the type of surgical procedure performed, if any, for all registered patients. Categorization of these procedures allows all patients who underwent a major surgical resection to be identified (Table S1 in the Supporting Information).
Like Denmark, the Yorkshire region has a total population of 5.7 million. The region accounts for approximately 10% of the colorectal cancer cases in England. The data for Yorkshire were sourced through the UK Colorectal Cancer Intelligence Hub's COloRECTal Repository (CORECT-R) [18]. Specifically, data from the cancer registry (National Cancer Registration and Analysis Service) were linked to hospital admission data (Hospital Episode Statistics) to identify all colorectal patients and those who underwent a major surgical resection, as described in CORECT-R's methodology [18]. The case ascertainment rate in the Yorkshire region was estimated to be 99% when compared with the Hospital Episodes Statistics dataset for the period 2001-2007 [19].
Patients were deemed to have undergone a major surgical resection if the operation date was within 1 month prior to, and up to 1 year after, the date of diagnosis. All analyses were performed separately for colon (C18-C19) and rectal cancer (C20), and further The observed percentage of patients treated with major resection in each stratum was calculated using the total number of cases

What does this paper add to the literature?
Age may influence the decision to surgically treat patients with colorectal cancer and therefore have an impact on overall survival rates. In this population-based study, we compared the use of major surgical resection and the impact this may have had on survival in different age groups over a 12-year period in two demographically similar regions of Europe.
as the denominator, irrespective of treatment intent, which was not available across both datasets. Odds ratios (OR) and 95% confidence intervals (CI) were then calculated comparing Denmark with Yorkshire for each stratum and also combined for the whole study period using Mantel-Haenszel weights.
To investigate the factors associated with use of major resection, we modelled Danish and Yorkshire populations separately using logistic regression with the following covariates: age group, sex, stage of disease and study period. Stage of disease was missing for 11% and 19% of Danish and Yorkshire patients respectively. Therefore, we used ordered logistic imputation to impute missing values and es- To investigate whether differences in the use of major resection may have an effect on overall outcome of colorectal patients in the two populations, we calculated 1-year survival estimates. We used relative survival using the strs [20] function in Stata to estimate survival and to control for any differences in background mortality between Denmark and Yorkshire. The background mortality of the general populations was estimated using life tables by sex, single year of age and calendar year. These were estimated over the three periods of diagnosis for all patients, and both resected and nonresected patients.

Colorectal populations
A total of 51,021 Danish and 39,456 Yorkshire patients with colorectal cancer were included. The age distribution of patients was broadly similar in the two populations at the beginning of the study; however, a higher proportion of patients in Yorkshire were aged ≥80 years during the most recent period for both colonic (30.9% vs 23.0%) and rectal (20.6% vs 16.5%) cancer (Table 1). An increase in the occurrence of Stage I colon cancers over time was observed in both populations, but comparisons between stage groups was difficult due to a differential rate of missing stage over the study period.

Colon cancer resections and outcomes
Overall, the proportion of colon cancer patients treated with major resection was higher in Denmark than in Yorkshire (77.3% vs 63.5%).
This was consistent across all the periods of study and the difference in the use of major resections increased with age (70.5% vs 50.5% for the ≥80 years age group). After adjustment for covariates, the odds of resection for those aged ≥80 years compared with 60-69 years in Yorkshire were OR (95% CI) = 0.26 (0.23, 0.28), lower than the corresponding odds in Denmark [OR (95% CI) = 0.54 (0.50, 0.59)] ( Table 2). The odds of resection were also lower in Stage III patients compared with Stage II patients in Yorkshire [OR (95% CI) = 0.54 (0.49, 0.60)] but not in Denmark [OR (95% CI) = 0.97 (0.82, 1.16)]. Additionally, a significant decrease in the use of resection was observed over the study period for both populations ( Table 2).
Within patients aged ≥80 years, the use of resection decreased from 55.4% to 45.1% in Yorkshire. The corresponding decrease in Denmark was much smaller, from 70.9% to 68.5% ( Figure 1A). To investigate whether the differences in the use of resection across age groups had an effect on short-term outcomes, we com-

Rectal cancer resections and outcomes
Overall, the proportion treated with major resection was higher in Denmark than in Yorkshire (71.2% vs 61.9%); this difference increased with age ( Figure 1A,  Figure 1A).
The difference in use of major resection was significant in all age groups, with an increase in odds of major resection in Denmark

Impact on 1-year relative survival over time
For the entire cohort of patients with colonic cancer, relative survival in Denmark significantly improved over the study period for all patient age groups, including those aged ≥80 years ( Figure 4A).
Survival for patients with colonic cancer in Yorkshire showed a nonsignificant increase in all age groups. In nonresected patients, significant increases in relative survival were observed for all age groups

DISCUSS ION
This retrospective population-based study has shown a differing approach to surgical management of patients with colorectal cancer between Denmark and Yorkshire, especially in older age groups. A higher proportion of all patients underwent major surgical resection in Denmark and, of these patients, long-term postoperative mortality was as low as that in Yorkshire. High use of major resection in those aged ≥80 years in Denmark has been maintained while still increasing overall rates of survival, whereas decreasing use of resection in Yorkshire patients with colon cancer of the same age has coincided with a period of unchanging survival.
Although differences in the rates of use of resection for colonic cancer were found across all age groups, the most pronounced difference was found in those aged ≥80 years, but also with noticeable differences in those aged 70-79 years. There is a concern that increased use of resection in older age groups will lead to a higher postoperative mortality as such patients are more likely to have existing comorbidity and frailty [13]. Some evidence for this was observed here, as there were increased odds of death within 30 days for Danish patients aged ≥80 years with colonic cancer. However, 1-year postoperative mortality in this age group in Denmark was equivalent to that observed in Yorkshire patients, and was actually lower for the latest study period in patients aged 70-79 years. The potential for a trade-off between increased short-term risk and longer-term benefit has been suggested previously when considering treatment of the older population [13]. It is also worth noting that 30-day postoperative mortality decreased sharply over time in both populations (16.2%-7.7% in Denmark, 15.1%-5.2% in Yorkshire), which suggests that patient care in the ≥80 years age group has improved considerably and needs to be considered when selecting patients for surgery.
Given that relative survival for resected patients aged ≥80 years with colonic cancer in Yorkshire improved over the study period, but not for all patients aged ≥80 years, it is possible that this could be due to a tendency to select fewer elderly patients for major resection. Whereas use of major resection in Yorkshire decreased over the study period, the Danish maintained relatively high use whilst still increasing 1-year relative survival. As more of the Danish patients aged ≥80 years received surgical resection, this could explain the substantial difference in relative survival between the two cohorts of patients in this age group.
Use of major resection was lower for rectal cancer in both countries. This is to be expected, since a number of patients will have alternative treatments including radiotherapy [21] and local surgical resection [22]. However, use of major resection was again higher in the Danish population. Unlike in colon cancer, the higher rate of resection did not coincide with a higher 30-day mortality for Danish patients aged ≥80 years compared with Yorkshire, and the mortality treatment was implemented in Denmark during the 2000s [4,23], but the resulting number of centres is similar to that found in Yorkshire [9] with a population of equivalent size. The increased number of Danish cases shown in the last period of this study is almost certainly due to the later introduction of screening [24]. The impact of screening and centralization will have resulted in a higher number of cases per centre and, possibly, a higher per surgeon workload.
Increased hospital and surgeon workload have shown associations with better outcomes in colorectal cancer [25,26]. Preoperative and postoperative initiatives may also differ; for example, enhanced recovery after surgery (ERAS) has been adopted widely in Denmark [7] but not region-wide within Yorkshire. Since ERAS has been shown to lower the risk of postoperative complications and reduce recovery time [27], having such a policy in place may increase the willingness to operate. Additionally, Denmark is known to have a high uptake of laparoscopic surgery [6], which has been associated with reduced length of hospital stay and 30-day mortality compared with open surgery [28] and could be a contributing factor when considering patient recovery.
There are limitations in this study, and the implications of comparing the two datasets need to be considered. Potential disparities in case ascertianment may have a marginal effect on estimates of ORs when comparing proportions of resection between the two populations. However, we calculated that over 60% of colon cancer cases and over 40% of rectal cancer cases in Denmark would need to be missing in the ≥80 years age group in 2013-2016 for differences in ORs to be nonsignificant.
There are likely to be additional differences between the two datasets that may have an impact when investigating surgical management. The DCCG records the surgical procedure used at the time of operation, whereas the CORECT-R methodology uses an algorithm to retrospectively search hospital admission records and identify those patients who underwent a major resection. In addition, it is important to take into consideration that international comparison of survival estimates may be affected by differences in cancer registration practices [29]. This includes the completeness of the registration source [19,30,31] or errors in registration such as in the date of diagnosis [32]. However, this is unlikely to affect the survival estimates to the extent that it explains the observed differences [33].  This study shows greater use of major resection in the older age groups for the Danish colorectal cancer population when compared with Yorkshire, corresponding to an increased short-term risk in colon cancer patients aged over 80 years but no increased risk in the longer term or in rectal cancer. If confirmed through further study, we should be able to identify more patients from Yorkshire who, with appropriate improvement in selection and improved periand postoperative care, are suitable for potentially curative surgery so improving long-term outcomes. It is important to appropriately communicate the risks of surgery, but it is also possible that many older patients in Yorkshire would benefit from consideration of a major resection.

CO N FLI C T O F I NTE R E S T
The authors declare there are no conflicts of interest.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data used for this study are available from the National Cancer Registration and Analysis Service via application to the Public Helath England Office for Data Release and CORECT-R, and application to the DCCG, subject to relevant approvals.