1.1 Background

The greater part of newly diagnosed colorectal cancer cases are amongst older adults (≥ 70 years), and it is also a significant cause of death in this age group [1-3]. Although the treatment of colorectal cancer has improved, rendering better overall post-operative survival, this is only partly the case amongst older patients. Especially, the immediate post-operative mortality and morbidity are increased amongst older adults [4-8]. We have previously shown that patients ≥ 70 years have an increased 90-day mortality rate post colorectal cancer surgery compared to younger individuals [9].

Although increased chronological age is a negative prognostic factor, there is a vast inter-individual diversity within an age group [1, 10, 11]. Frailty is a clinical syndrome which can be used as a marker of biological age [12, 13]. Frailty has previously been described as an independent risk factor associated with adverse effects such as increased post-operative mortality, functional dependence, readmission to hospital and need of residential care [5, 14-16]. Frail elderly patients are a high-risk group also in colorectal cancer surgery [10, 11, 17, 18] and that frailty is more strongly associated with negative outcome than age [5].

One of the most commonly used screening instruments for frailty is the Clinical Frailty Scale (CFS-9), which can be assessed during a routine patient visit [12, 14, 19-22]. CFS-9 is developed from the accumulation of deficits model of frailty and is based on the care-takers evaluation of the patient's level of fitness, presence of disease symptoms, the degree of dependence from others and of cognitive function [23, 24]. The instrument is a 9-item tool where patients who are assessed as CFS 4–8 are considered frail [25]. The assessment is intended to be done during a clinical interview, but recent studies have shown a reliable correspondence between standard evaluation and retrospective valuation through medical chart review [26-29]. The aim of this study was to investigate whether frailty is an explanatory factor of the poorer post-operative outcome in older adults in Sweden.

1.2 Objectives

The aim of this study was to evaluate if frail patients in this cohort had poorer post-operative outcome following colorectal cancer resection surgery compared to non-frail. The hypothesis was that frailty is a risk factor for post-operative mortality and morbidity following CRC resection surgery in older adults. The primary endpoint was differences in 90-day overall survival, secondary endpoints were post-operative complications, reoperations, intensive care unit (ICU) care, post-operative hospital length of stay (LOS), readmissions within 30 days post-surgery, discharge destination (own housing, nursing facility) and all cause 1-year mortality.

2.1 Study Design, Setting and Participants

This is a retrospective, registry-based, cohort study performed in a Swedish Region of 1.7 million inhabitants during January 2016 to June 2020. The study population included all patients aged ≥ 70 surgically treated for colorectal cancer in Region Västra Götaland (VGR) during the stated period (n = 2378). The study was approved by the Swedish Ethical Review Authority (Dnr 2021-02099 and Dnr 2022-03759-02).

A sample size calculation based on a 5% difference in 90-day mortality between non-frail (CFS-9 1–3) and frail (CFS-9 ≥ 4) patients was conducted. The presence of frailty in the current material was 56% and 90-day mortality in the non-frail group was assumed to be 1.5%. To detect a difference with 80% power and at a significance level of 0.05, a total of 480 patients was needed.

2.2 Variables and Data Sources

A cohort of 500 patients was randomly selected from all patients diagnosed and operated for colorectal cancer during 2016 and 2020 in the studied region and assigned a frailty score according to CFS-9, through retrospective chart review. The medical charts of these patients were analysed regarding previous medical history and comorbidities, current medications, recent visits to the emergency department or hospitalisation episodes, recent investigations, and physical examinations. Background information such as housing situation, social history, access to formal and informal supports, use of gait aid, cognitive status and activities of daily life (ADL) function were also assessed through the medical records. These factors were used to assign patients a CFS-score of 1–9 and the available information was sufficient to issue a CFS-score for all subjects in the cohort.

Baseline data for all subjects were retrospectively obtained from the Swedish Colorectal Cancer Registry (SCRCR), a nationwide registry of patients with colorectal cancer, with high coverage (> 98%) [30]. The following baseline variables known to influence death after colorectal cancer surgery were obtained from the registry: age, sex, tumour location, stage according to TNM, American Society of Anaesthesiologists (ASA) classification and elective versus urgent surgery. It is known that elderly patients are more often treated with urgent surgical resection compared to younger individuals [4, 6], hence, this variable was treated as a mediator rather than a confounder and was not adjusted for. A Directed Acyclic Graph (DAG) showing the relationship between the variables and the exposure (frailty) and outcome (mortality) can be found in the Supporting Information.

2.3 Statistical Methods

Analyses of dichotomous outcome variables were made through logistic regression and Fisher's exact test and presented as odds ratio (OR) with 95% confidence interval (CI). Adjusted analyses were made through multivariable logistic regression and presented as OR with 95% CI. Post-operative hospital LOS was analysed using Fisher's non-parametric permutation test and multiple logistic regression, presented as mean difference with 95% CI. For all models, the independent variable CFS-9 was treated as follows: the subgroup of non-frail patients (CFS 1–3) served as reference group and was compared with each step from 4 to 9. CFS-9 was also analysed as a continuous variable presenting the risk when increasing CFS-9 one step. A dichotomised analysis between non-frail patients (CFS 1–3) and frail patients (CFS ≥ 4) was also performed. All tests were two-tailed and conducted at 0.05 significance level.

3.1 Patient Characteristics

Baseline characteristics of the cohort are described in Table 1. The prevalence of frailty in the entire cohort was 56%. Non-frail patients were significantly younger than frail patients (76.2 vs. 80.0, p < 0.001), the sex distribution was similar but there was a slight overrepresentation of women in the frail group. There were more ASA III–IV individuals in the frail group compared to the non-frail group, and non-frail patients were treated for more advanced disease and more often went through elective surgery.

3.2 Outcome Variables

Multivariable analyses showed a significantly increased 90-day mortality risk for frail patients and there was an evident risk increase for each step of the CFS (Table 2 and Figure 1). This was also confirmed in analyses using CFS as a continuous variable (data not shown). Further, an increased risk of post-operative surgical complications for frail patients was revealed, however, this was not shown in analyses of overall complications (Table 3). It was more common amongst frail patients to be treated in ICU postoperatively compared to non-frail patients, when the CFS was analysed as a continuous variable (1.38 [1.08–1.78], p 0.011). Post-operative LOS compared between non-frail and frail patients was significantly longer (7.63 vs. 11.0 days, p < 0.001). Frail patients were also more likely to be discharged to a nursing home compared to non-frail patients (Table 3). One-year mortality was significantly increased for all stages of frailty. There was no significant increased risk for reoperations or readmissions within 30 days for frail patients.