Associations between patient factors and successful colon capsule endoscopy – A prospective cohort study

To establish patient factors associated with a successful colon capsule endoscopy (CCE) test.

A CCE test can be defined as "successful" if it is complete and the overall bowel preparation is adequate.A CCE can be considered as complete if the whole colon and rectum are visualized within the battery life of the capsule.Published completion rates (55%-100%) and rates of adequate bowel preparation (40%-100%) for CCE vary substantially [3,4].Despite the use of different bowel-preparation and booster regimens, the rates of successful tests for CCE do not match those for colonoscopy or CTC [5].The benefits of the less-invasive CCE are lost if patients who do not obtain a successful test require further assessment by colonoscopy, CTC or flexible sigmoidoscopy rather than proceeding to a targeted therapeutic procedure for the pathology identified by CCE.
Multiple patient factors influence gut motility, and these will affect the successful test rates of patients undergoing CCE [6].
Furthermore, a patient's compliance with the bowel-preparation and booster regimen will also influence the chances of obtaining a successful test.Advanced diverticular disease, colonic elongation or tortuosity, and older age are known predictors of incomplete colonoscopy [7,8].The patient factors influencing the rate of successful CCE tests are, to date, unknown.In this study we aimed to establish the factors associated with a successful CCE test.

Data sources and ethics
Patient data collected in the ScotCap evaluation were used in this prospective cohort study [9].Two types of patients were recruited to the ScotCap evaluation: symptomatic and surveillance.
Symptomatic patients were referred (by their general practitioner) with new gastrointestinal symptoms for investigation and had been assessed as requiring a colonoscopy, by a secondary care consultant.
Surveillance patients were those on the waiting list for surveillance colonoscopy because of increased risk of developing colorectal cancer (CRC).The inclusion and exclusion criteria for the ScotCap evaluation are detailed in Table 1.In the current study, we included patients from a single geographical health board (National Health Service [NHS] Highland), the largest subset of those in the ScotCap cohort, who underwent CCE during the ScotCap evaluation up to 1 April 2020.Ethical approval was not obtained for the use of data for patients from NHS Grampian and Western Isles and for those who underwent CCE after 1 April 2020 and hence these patients were excluded from the study.Further details of patient recruitment and data collection in the ScotCap evaluation are described elsewhere [9].
Patients underwent CCE procedures using the PillCam™ COLON 2 (Medtronic, UK).The bowel preparation and booster regimen and the dietary restrictions all used for the ScotCap evaluation are detailed in Appendix S1 in Table S1.A specialist nurse carried out CCE procedures following a protocol (Appendix S2).
The images obtained following each CCE procedure were reviewed by two NHS Scotland consultant gastroenterologists trained in reading CCE images, who produced a report detailing if the CCE procedure was complete, if the bowel preparation was adequate and details of any pathology detected by the capsule anywhere in the visualized GI tract.Colon capsule endoscopy reports were returned to the referring secondary care clinician who decided on further patient management.
We defined a "successful test" as a CCE procedure that was complete and for which the bowel preparation was adequate.
A "complete test" was defined as a CCE procedure for which the whole colon and rectum were visualized within the battery life of the capsule.The bowel preparation was determined to be "adequate" if rated as at least ''fair'' in all colonic segments and the overall quality was deemed to be adequate by the consultant gastroenterologist who analysed the CCE images.The bowel preparation was assessed and rated according to the Boston Bowel Preparation Scale [10].
Significant pathology, identified using CCE, was defined as any finding requiring follow-up according to European guidelines [11].A patient was regarded as requiring a further procedure if they underwent or were scheduled to undergo a colonoscopy, flexible sigmoidoscopy, CTC, clinic review or laparotomy following CCE by the end of the evaluation, irrespective of whether the CCE procedure was successful or not.
Additional patient data were collected to supplement the ScotCap evaluation data.Patients' medical conditions, medication, previous bowel resection surgery and body mass index (BMI) data were collected by a clinical researcher (C.M.) from patients' secondary care electronic health records.Medical conditions and medications were categorized using The International Classification of Diseases 11th Revision (ICD-11) and the British National Formulary, respectively [12,13].Medical conditions, medications and their categories are listed in Appendix S3 in

What does this paper add to the literature?
Optimizing patient selection may improve the outcomes for CCE.We have identified patient factors associated with a successful test, adequate bowel preparation and the need for further procedure following CCE.These factors may help guide clinicians when choosing patients for CCE.

Aims
The primary aim was to identify patient factors (Appendix S4 in Table S1) associated with a successful CCE procedure (a complete test with adequate bowel preparation).A secondary aim was to determine any patient factors associated with further procedure requirement following CCE.We also analysed whether any patient factors were associated independently with a complete test (visualization of the whole colon and rectum) or adequate bowel preparation (a bowel preparation rating of at least fair in all colonic segments and determined as adequate overall by the consultant gastroenterologist who analysed the CCE images).

Statistical analysis
Statistical analysis was conducted using R. Symptomatic and surveillance cohorts were analysed separately.Baseline characteristics were expressed as means with standard deviation for continuous variables and absolute numbers and percentages for categorical variables.Univariate and multivariate logistic regressions were performed, along with least absolute shrinkage and selection operator (LASSO) regression for variable selection.The LASSO method was used as it can effectively shrink the coefficients for nonessential variables to zero while retaining the most relevant ones for predicting desired clinical outcomes [14].The model used for this data analysis was calibrated to manage the number of variables being examined.
Cross-validation (10-fold) was used during LASSO regression to find the value of lambda, which was identified when the cross-validation error was at a minimum.The variables selected were considered in a multivariate logistic regression.The results from both the univariate and multivariate regression analysis were expressed as odds ratio and 95% confidence interval.To minimize reduction in sample size, variables such as BMI, haemoglobin (Hb) count, estimated glomerular filtration rate (eGFR) and faecal immunochemical test (FIT) results were not included in the multivariate logistic regression because of the volume of missing values.Instead, univariate logistic regression was performed on these variables.For all these tests, a significance level of 0.05 was used.

RE SULTS
Of the 733 patients in the ScotCap evaluation, 401 NHS Highland patients were included in this study (Figure 1).Data for 263 symptomatic patients and 137 surveillance patients were analysed.The demographics and CCE outcomes for these patients are described in Tables 2 and 3.The mean age of patients was 60 and 62.7 years in symptomatic and surveillance cohorts, respectively.The proportion of patients who were female was 44.5% and 42.3% in symptomatic and surveillance cohorts, respectively.In the symptomatic cohort, the most common referral urgency rating was urgent suspected cancer (39.2%), and the most common referral symptom was change in bowel habit (68.1%).The proportions of patients in the symptomatic cohort with FIT results of <10, 10-399 and >400 μg/g were 35%, 30% and 1.5%, respectively.The remaining 33.5% of symptomatic patients did not have a FIT result available.The most common reason for patients undergoing colonic surveillance was previous polyps (53.6%).The proportions of patients obtaining a complete test, adequate bowel preparation, a successful test and requiring a further procedure in the symptomatic cohort were 72.6%, 80.2%, 66.2% and 62.4%, respectively.In the surveillance cohort, the proportions of patients obtaining a complete test, adequate bowel preparation, a successful test and requiring a further procedure were 71.5%, 65%, 55.5% and 67.9%, respectively.
Patient factors with a statistically significant association with a successful test, adequate bowel preparation and further procedure

DISCUSS ION
In this prospective cohort study, we found that the age of symptomatic patients was associated with a successful CCE test and adequate bowel preparation.Age appears to be a potential predictive factor for patients to obtain a successful test when undergoing CCE.The underlying mechanism for these results may be the ability of this group to complete the bowel-preparation regimen required for CCE.Although these results do not imply causality and the associations found for both patient groups were of low magnitude (OR = 0.97 for both), clinicians could consider patient selection Establishing the optimal FIT range for patients undergoing CCE should be an aim of future research as the use of FIT prior to CCE will be important when determining which patients are best suited for the test.
Recruitment to the surveillance cohort during the ScotCap evaluation was halted because of a higher follow-up procedure rate compared with that for the symptomatic group [9].However, these results suggest that patients undergoing surveillance colonoscopy as a result of previous CRC or having had a previous large-bowel resection were associated with a lower rate of follow-up procedures.
The presence of colorectal pathology is comparatively low in patients who are undergoing surveillance colonoscopy following CRC surgery, and a shorter colon will reduce the amount of mucosa requiring visualization by the capsule [19,20].The use of beta blocker medication was also found to be associated with a reduced rate of follow-up procedure requirements in the surveillance cohort on univariate testing.The mechanism responsible for this result is unclear and should be further investigated in future work.
A comparable study to ours, conducted by Moen et al., has recently been published [21].In their similarly sized study (n = 451), that used in our study, which may explain the differences in results.
In addition, the CCE test completion rate was higher in our study (72.3% vs 51.9%), which may have provided greater power for statistical analysis.These contrasting results may also reflect the challenges of generalizing results to different populations and patient selection criteria for CCE.
Our results should also be compared with those reported in a recently published observational cohort study by Gimeno-García et al [22].Their results demonstrated that no variables were associ-

Strengths and limitations
One of the strengths of this study is the data source.The ScotCap evaluation is the largest evaluation of CCE use in a symptomatic patient population.In addition, CCE procedures in the ScotCap evaluation were carried out in line with a protocol using a consistent bowel-preparation and booster regimen and dietary instructions TA B L E 5 Factors associated with further procedure requirement in patients of the surveillance group.throughout to improve the reproducibility of CCE outcomes.
However, further research may be required to establish if these results are applicable to different bowel-preparation regimens and newer CCE devices, if available.
There are some limitations to our study.Only data from one NHS Scotland health board were used in this study, potentially limiting generalizability.Some data (such as BMI or FIT) were not available for all patients, limiting their inclusion in multivariate analysis.We also acknowledge that additional data collection for this study was carried out retrospectively with a risk of reduced data quality and completeness.Finally, patients' medical conditions and medications were grouped into broad categories, which may have reduced the specificity of the result analysis.

CON CLUS ION
In this study, we found that younger age was associated with a successful CCE test result.Symptomatic patients with a lower FIT result or those undergoing surveillance as a result of previous CRC may also be more suitable for CCE if the need for further procedures is

ACK N O WLE D G E M ENTS
We would like to thank all patients and staff involved in the ScotCap evaluation.We would also like to thank The Data Lab for their contribution to the project.

FU N D I N G I N FO R M ATI O N
This study was funded by an NHS Highland Research Development and Innovation grant.

CO N FLI C T O F I NTER E S T S TATEM ENT
We declare no conflicts of interest related to this article.

TA B L E 1
Abbreviation: FIT, faecal immunochemical test.
a concern.Findings are limited to the sample used in this study and therefore further research on different or larger datasets may be required to support these results given the novelty of this work.These results may be considered by clinicians when selecting patients to undergo CCE to achieve a better successful test rate and lower follow-up procedure rate.AUTH O R CO NTR I B UTI O N S C MacLeod: Conceptualization; Methodology; Funding acquisition; Writing -original draft; Writing -review & editing; Data curation.A Foxton: Software; Formal analysis; Validation.P Wilson: Conceptualization; Supervision; Writing & review & editing.S Treweek: Conceptualization; Supervision; Writing & review & editing.AJM Watson: Conceptualization; Funding acquisition; Investigation; Methodology; Supervision; Resources.

Table S1 .
Baseline characteristics and colon capsule endoscopy (CCE) outcomes of patients in the surveillance group.Factors associated with a successful test, adequate bowel preparation and further procedure requirement in patients of the symptomatic group.
[16][17][18]ofile flow diagram.CCE, colon capsule endoscopy; NHS, National Health Service.based on these results to improve the rate of successful testing in clinical practice.No patient factors examined were independently associated with an improved CCE test completion rate.Colon capsule endoscopy test completion rates may be limited by current technology as battery life is a likely limiting factor.However, more promisingly, the introduction of prucalopride as a booster has recently been shown to increase the CCE complete test rate[15].The use of prucalopride should be explored in other patient populations as our findings suggest that refining patient selection for CCE may not improve the complete test rate.Patient selection will be important to reduce the need for further procedures post CCE and to avoid associated costs for both the patient and the health service.Our results suggest that symptomatic patient age and FIT results between 10 and 399 μg/g TA B L E 2 Baseline characteristics and colon capsule endoscopy (CCE) outcomes of patients in the symtomatic group.Note: Values are given as n, mean ± SD or n (%), unless stated otherwise.aMorethanonesymptomcouldberecorded.TA B L E 2 (Continued)are associated with a further procedure following CCE.It is well established that FIT is a valuable test for predicting the risk of underlying colorectal pathology in those with relevant symptoms; however, the risk of harbouring CRC varies significantly, from 1.9% to 22.4%, in patients with a FIT result of 10-399 μg/g[16][17][18].TA B L E 3Note: Values are given as n, mean ± SD or n (%) unless otherwise stated.TA B L E 4Abbreviations: FIT, faecal immunochemical test; LASSO, least absolute shrinkage and selection operator.
Moen et al. found a range of predictors associated with a complete CCE test, including increased patient age, which contradicts our results.The data used in Moen et al. were from patients in an epidemiological study and the booster regimen used was also different from