Long‐term health‐related quality of life following colorectal cancer surgery: patient‐reported outcomes in a remote follow‐up population

Remote follow‐up (RFU) after colorectal cancer (CRC) surgery allows delivery of surveillance tests without the need for regular outpatient clinical appointments. However, little is known about health‐related quality of life (HRQoL) in RFU patients. The main aim of this study was to quantify HRQoL in our RFU population to identify particular patient groups that may benefit from a more personalised approach to follow‐up, including access to a survivorship clinic.


INTRODUC TI ON
Colorectal cancer (CRC) is the third most common malignancy in the UK, with over 41 000 new cases being diagnosed each year [1].
With curative surgery as the mainstay of CRC treatment, survivorship is increasing and age-standardised 5-year survival rates are now 60.1% [2]. The randomised Follow-up After Colorectal Surgery (FACS) trial found that carcinoembryonic antigen (CEA) monitoring (initially 3-monthly for 2 years, then 6-monthly for 3 years) and CT of the chest abdomen pelvis (CTCAP; 6-monthly for 2 years, then annually for 3 years) resulted in improved detection of potentially curable recurrence [3]. The National Institute of Health and Care Excellence (NICE) thus advocates regular CTCAP, CEA level monitoring and colonoscopy to detect recurrence for 5 years after completion of treatment [4]. However, no consensus exists as to how follow-up should be delivered [5] and there is significant variation in clinical practice at both national and international levels [6].Clinician-led follow-up requires patients to attend regular clinic appointments over 5 years [7].
This method is resource heavy and increasing survival rates can overwhelm outpatient services [8]. Timing of clinic visits may sometimes adversely affect follow-up schedules; more importantly, administrative errors around significant results or 'lost to follow-up' issues present a significant governance risk. Meta-analysis of randomised controlled trials has found no evidence that face-to-face follow-up is required for effective surveillance [9] and attendance at clinical appointments has been recognised to increase patient anxiety [10].
'Remote' follow-up (RFU) enables timely delivery of surveillance tests and negates the need for regular clinic attendance. This form of follow-up, also referred to as 'personalised stratified follow up', forms part of the National Health Service (NHS) Long Term Plan for Cancer [11]. Robust protocol-driven RFU schemes have been demonstrated to be safe, acceptable to patients and cost-effective [6,12]. Patients undergo tests at the scheduled interval, results administration can be protocolised and 'well survivors' need only return to clinic if their results are abnormal. The potential drawback of RFU is that problems affecting survivors' quality of life may not be addressed. The National Cancer Survivorship Initiative emphasises the importance of quality of life assessment in patients living beyond a cancer diagnosis [13]. Siddika

Aims
Long-term HRQoL after CRC surgery in patients under RFU is of interest due to a lack of literature describing outcomes in this group.
The primary aim of this study was to quantify HRQoL in our RFU population to identify particular patient groups that may benefit from a more personalised approach to follow-up, including access to a survivorship clinic.

ME THOD
In 2011 Nottingham University Hospitals Trust (NUH) adopted a RFU approach for those who had undergone surgery for CRC. Patients are typically reviewed once in a postoperative clinic to address problems related to surgery and subsequent symptoms. If required at this time further adjuvant treatment is arranged and delivered by the oncology team. All patients are simultaneously enrolled into RFU, which begins at time of treatment completion. This service is coordinated and run by a specialist cancer nursing team. Patient demographics and details regarding their diagnosis and treatment are entered prospectively into a RFU database (Microsoft Access™, Seattle, Washington, USA).
When the database started a small number of patients with neuroendocrine tumours and polyps were included, but we planned to exclude these from the analysis of CRC. This database is used to identify when patients require blood tests, CT scans and colonoscopy at appropriate time intervals (see Appendix 1 for the full protocol). The team then orders the required tests, reviews the results and communicates the results to the patient; if they are abnormal, the patient is referred to the clinician-led multidisciplinary team. Figure 1 illustrates the typical journey of a patient and entry into the RFU programme. It is important to note that during RFU patients may request to be seen on an ad hoc basis in a colorectal clinic if they have any troubling symptoms that require further management.
We undertook a cross-sectional study of all patients in RFU using three validated questionnaires to ensure coverage of a wide breadth of HRQoL domains. Prior to distribution, permission to use each questionnaire for the purposes of this study was granted by EuroQol for the EQ-5D-5L [14] and EORTC for QLQ-C30 [15] and QLQ-C29 [16].
The widely used EQ-5D-5L was selected to provide an insight into general HRQoL. This uses a five-point scale (ranging from 'no problems' to 'extreme problems') to measure everyday function across the five domains of mobility, self-care, usual activities, pain and anxiety.
Responses can then be used to generate a single 'index' score which

What does this paper add to the literature?
Remote follow-up after colorectal cancer surgery allows safe delivery of surveillance tests and obviates the need for regular clinic appointments. However, there is a paucity of information on patient-reported quality of life within this set-up. This study found that women, patients with a right-sided resection and patients with a stoma may require additional clinical reviews. is a summary of respondents' answers to the five domain questions standardised to the UK general population [17]. The index score can range between −0.594 and 1; 1 corresponds to perfect health and less than 0 corresponds to health states which are 'worse than dead' [18].
EORTC produces questionnaires to enable assessment of HRQoL specifically in cancer patients. We selected the general oncological QLQ-C30 and the complementary CRC-specific QLQ-C29 for use in this study. The answers to symptom-specific questions are recorded on a four-point scale ranging from 'not at all' to 'very much'. For QLQ-C30, answers to several questions can be combined to provide an overall score for items such as 'physical function' and 'emotional function'. QLQ-C30 also has two questions about overall health and quality of life with a seven-point scale ranging from 'very poor' to 'excellent'.
For these questions an overall quality of life score can be derived [19].

Data collection
All patients gave permission to be contacted when they initially Questionnaires were produced in a computer-readable format.
Returned questionnaires were scanned and transformed into an electronic database using Teleform Scan Station, Teleform Reader and Teleform Verifier software produced by OpenText ™ [20]. At the time of scanning, all software output was manually checked against the physical questionnaires to ensure accurate transfer of information and corrected accordingly. Ambiguous responses and questions left blank were treated as missing data. The electronic output was re-checked by an external validator (AG) against the physical forms and any discrepancies were amended.
For patients on the database, information on demographics, year of RFU, site of cancer, operation type and recurrence details was collected prospectively. We undertook retrospective review of this information for all questionnaire returners to ensure accuracy. Further data were collected including Dukes stage at operation, operative details, presence of stoma, whether neoadjuvant and/or adjuvant treatment was received and site of cancer recurrence. Retrospective database review and additional data were obtained from electronic hospital records. Patients who were seen by a colorectal surgeon after entry into RFU were identified as having 'breached protocol' and these patients provided a comparative group with those who were purely followed up remotely. Details of any clinic attendance within the year prior to questionnaire completion were also recorded. Operation was categorised into 'right-sided resection', 'leftsided resection' or 'other colorectal resection' (Appendix 2). This involved review of clinic letters, multidisciplinary team outcome letters, discharge summaries, pathology results and follow-up imaging reports. Demographic data for nonresponders were also collected for comparison. Questionnaire responses and clinical data were combined for subsequent analysis.
We categorised age into three groups based on age at the time of

Comparative groups
Results for EQ-5D domains were compared with published norms for the general UK population [21]. Overall HRQoL scores and EQ-5D domains were also analysed between patients who breached protocol and those who did not. Further comparisons were made for patients who were seen in the year prior to questionnaire completion to determine whether recent breaches of protocol had any influence on HRQoL.

Data analysis
All statistical analysis was performed using Stata 12.0 [22]. EQ-5D index scores were calculated using the Crosswalk Index Value F I G U R E 1 Timeline illustrating typical journey of patients through diagnosis, treatment and remote follow-up (RFU)

Entry into RFU database
Year 1

Year 2 Year 3 Year 4+
Calculator [17], which is the method advocated by NICE [23]. For QLQ-C30 the symptom, function and overall global quality of life scores were calculated using the linear transformation method described in the EORTC manual [19].
Descriptive statistics were used to report demographics, op-

RE SULTS
A total of 722 patients were contacted and 463 (64.1%) responses were received ( Figure 2

Missing ques tionnaire dat a
Of the 428 patients included in the data analysis, 35 responders were excluded as they had undergone polypectomy alone. A total of 427 returned all three questionnaires. One patient returned F I G U R E 2 Flowchart to illustrate questionnaire response and subsequent details of included and excluded responders the completed EQ-5D and QLQ-C30 but did not return the QLQ-C29. The majority of questionnaires were filled out completely; for EQ-5D answers were complete in 98.4%, for QLQ-C30 in 98.6% and for QLQ-C29 in 91.6%.

D et ails of surgic al treatment and s toma
The specific operation types included in each category are detailed in Appendix 2. At the time of questionnaire completion 27.1% of patients had a stoma.

D emogr aphic s of patient s who breached protocol
The number of responders who breached protocol by being seen in clinic after entry into RFU was 126 (29.4%); 52 (12.2%) of these were seen within the year prior to questionnaire completion. Regarding gender, there was no significant difference between those who were seen in clinic and those who were not (χ 2 = 1.51, P = 0.22).
However, patients breaching protocol were significantly younger (χ 2 = 7.79, P = 0.05) and were significantly more likely to have undergone a left-sided resection or Abdomino Perineal Excision of Rectum (χ 2 = 7.93, P = 0.005). Further demographic details are outlined in Table 3.

O ver all HRQoL
Two overall measures of quality of life were utilised: the index score from EQ-5D and the global quality of life score from QLQ-C30. The distribution of results for each score was negatively skewed, hence we used nonparametric methods to test statistical significance. For QLQ-C30 global HRQoL the median score was 75.0 (IQR 58.3-83.3). For the EQ-5D index score the median was 0.785 (IQR 0.671-1), which corresponds to a health state with no problems with mobility, self-care or depression, moderate problems in usual activities and slight problems with pain. Figure 3 summarises the percentage of patients reporting 'no problems' versus 'problems' across EQ-5D functional domains. No problems in any domain were reported in 26.6% and 10.7% reported problems in every domain.

EQ -5D domain comparison
EQ-5D domain scores for pain, activity, mobility, self-care and anxiety were compared with published norms from a cohort of unselected members of the general UK population [21] (Table 5). Across all domains patients within RFU reported significantly more pain (P < 0.001) and anxiety (P < 0.001) and higher levels of anxiety (P < 0.001), mobility problems (P < 0.001) and difficulty with self-care

TA B L E 2
Demographic and cancerspecific details by site of tumour (P = 0.001). Domains were compared between patients adhering to RFU protocol and those who breached protocol. Statistically significant differences noted were higher rates of pain (P = 0.05) and more limitation to activity (P = 0.043) in the group that breached protocol.

Sy mptom repor ting
Abdominal symptoms such as pain were reported in 28

Sy mptoms in s toma patient s
Rates of abdominal pain and bloating were not significantly different between those with a stoma and those without (P = 0.72,

DISCUSS ION
This study is the first to examine HRQoL in operatively managed Rates of body dissatisfaction and erectile dysfunction were high. Our results suggest that female patients who are older with right-sided resection may require additional clinical reviews rather than just remote follow-up. Additionally, support should be offered regarding sexual dysfunction to those patients in RFU programmes.
Strengths of this study include the response rate of 64.1%, which is higher than that of similar studies in long-term CRC survivors [24][25][26], and the small number of missed answers. Possible limitations are that questionnaire responders were significantly older than nonresponders and hence the results may not be reflective of the experience of younger patients. No baseline data were collected; we therefore only present a snapshot of HRQoL within a RFU population and in comparison with the subgroup of patients who breached protocol, other studies and population norms.
Comorbidity has been shown to have a negative impact on HRQoL in CRC patients [27]; our study did not examine comorbidity as it was felt that retrospective collection of these data would be unreliable due to inconsistency in local reporting. Similarly, lower socio-economic status has a negative influence on HRQoL [28] and these demographic data were unavailable in our study population.

Comparative groups
Younger patients, those with who underwent left-sided resection and those with recurrent cancer were more likely to breach protocol and be seen in clinic. No overall differences were found in the subgroup of patients who breached protocol by being seen in clinic following entry into RFU. This suggests that the extra support required by these patients was provided appropriately through an ad hoc clinic visit.   Our findings reiterate previous UK-based studies which have found that stoma presence [25,29] and cancer recurrence [25] have a negative impact on HRQoL in CRC patients. There is variation in the reported influence of gender on HRQoL depending on the population studied. In general population terms it is well recognised that women report lower HRQoL scores than men [30]. Finnish and Iranian studies focusing on CRC patients found no difference between male and female responses to EQ-5D and QLQ-C30 [31,32]. We found significantly lower scores in women, which has been previously observed in the UK and Japanese cohorts [25,33]. Within our RFU patients high rates of abdominal symptoms and sexual dysfunction were found, and both of these sequelae have been widely reported in CRC survivors [26,[34][35][36][37].
Persistence of abdominal symptoms over time was reported in CRC patients at 1 and 3 years postdiagnosis, and our findings reflect this [29].
Downing et al. [25] reported that 34.5% of CRC patients be- HRQoL correlates with severity of low anterior resection syndrome (LARS) [34] and this impact has also been shown to persist over time [35].There is, however, a deficit of literature comparing outcomes between patients with right-and left-sided resection. One small casecontrol study which reported no difference in EQ-5D scores was stratified by resection side [38]. Recently, Buchli et al. reported on HRQoL and LARS stratified by resection side [39]. That study found that major LARS symptoms were more frequently experienced by patients with right-sided resection and that major symptoms were an independent predictor of lower HRQoL scores. Our data corroborate this: within our study population lower HRQoL scores were associated with right-sided resection. Our findings highlight that the long-term HRQoL outcomes of patients with right-sided resection should be of clinical concern. The outcomes in this patient group have perhaps been overshadowed by the current focus on LARS.

CON CLUS ION
Our findings provide us with confidence that patients enrolled in our RFU programme experience high HRQoL which remains stable.
We have identified factors which make a negative contribution to HRQoL; this information will be a useful tool in future service planning and patient counselling. Patients who breached protocol did not differ on overall HRQoL score but were more likely to experience pain and limitation of activity. Patients with right-sided resection reported significantly worse HRQoL, and we therefore highlight this patient group as a focus for further investigation. Overall, these findings suggest that even within a RFU setting, targeted clinics dedicated to addressing these specific problems and patient groups could mitigate deterioration in HRQoL after CRC surgery. A targeted clinic for these patients is being planned for those in the third year of follow-up as this was the postoperative time point with the lowest overall HRQoL scores. Given the ongoing global challenges with the COVID-19 pandemic this is likely to be delivered virtually.

E THI C S APPROVAL AND CON S ENT TO PARTI CIPATE
Following assessment with the UK Health Research Authority (HRA) decision tool, it was ruled that no formal ethics approval was required for this particular study. Patients returned the quality of life questionnaire packs if they first consented to participate.

CONS ENT TO PUB LIS H
Not applicable. No individual-level data are included in this paper.

ACK N OWLED G EM ENTS
We authors thank all our patients in RFU and the Nottingham Colorectal Service for their continued support and participation.
We would like to personally thank Charlene Atherton, Wayne Croves, Kate Roggan, Charlotte Ryton, Denyse Whitehead and Louise Williams who make up the specialist nursing team delivering RFU, Lisa Janiec the manager of the Macmillan Cancer Pathways Programme and Oliver Ng who set up the RFU database.

E TH I C S A PPROVA L A N D CO N S E NT TO PA RTI CI PATE
Following assessment with the UK Health Research Authority (HRA) decision tool, it was ruled that no formal ethics approval was required for this particular study. Patients returned the quality of life questionnaire packs if they firstly consented to participate.

CO N S ENT TO PU B LI S H
Not applicable. No individual-level data are included in this paper.

CO N FLI C T O F I NTE R E S T
None of the authors have any conflicts of interest to declare.

AUTH O R CO NTR I B UTI O N S
FLM, AA, AB, JW and DJH participated in the study concept and design, data collection, data analysis, reviewed the paper and approved the final paper for submission. DW and AG participated in the study design, data collection, reviewed the paper and approved the final paper for submission.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data in this publication are confidential. Any data requests should be made to the corresponding author.