Colonoscopic polyp detection rate is stable throughout the workday including evening colonoscopy sessions

Introduction

Colonoscopy is an important screening and surveillance tool. Much attention has focussed on optimising its effectiveness at diagnosing pre-malignant lesions. Polyp detection rate (PDR) or adenoma detection rate (ADR) can be used as markers for colonoscopy quality. The implication is that careful and optimal mucosal visualisation enhances polyp detection and is a reliable marker of a colonoscopist performing appropriately and effectively^1,2. Recent data have demonstrated the importance of adenoma detection rate as not only a metric of colonosocopy performance but as an independent predictor of protection against subsequent colorectal cancer³. PDR is defined as the detection of one or more polyps at colonoscopy. Several factors may influence PDR, obviously including patient selection and case mix, but recently it has been suggested that the time the procedure is performed and the order of colonoscopy within a session also influence colonoscopy performance.

Inconsistent results examining the PDR of colonoscopies throughout the workday have been reported. A number of studies have reported that PDRs decrease as the day^4,5 or endoscopy list⁶ progress, and that PDRs in morning lists are higher than afternoon lists⁷. However this effect does not appear to be universal across endoscopy units with other studies reporting no such reduction in polyp detection overall⁸ or when colonoscopies are performed in half day blocks⁹ or 3-hour shifts¹⁰.

This effect, where shown, has been generally attributed to endoscopist fatigue^4,11 although this was not independently assessed in those studies. This effect appears logical and correlates with increasing errors in other repetitive or tiring vocations such as errors by house officers related to time in hospital¹², inadvertent errors at simulated laparoscopy as mental and physical workload increases¹³ and impairment in perceptual and motor ability in nurses toward the end of nightshifts¹⁴. Queue position has been suggested as a novel surrogate measure for operator fatigue⁶. There has also been a suggestion that endoscopists may rush procedures later in the list¹⁵. Another possible hypothesis is that bowel preparation may be a factor, the inference being that bowel preparation is often prescribed appropriately for morning endoscopies but the regimen is not appropriately altered for optimum bowel preparation in the afternoon.

This apparent effect of reduced PDR throughout the work day has been superimposed on top of the difference in performance between endoscopists, which has been reported on multiple occasions; even within a group of fully competent endoscopists there appears to be spectrum of polyp detection rates^3,16,17. This endoscopist effect has not been consistently accounted for in the published data examining the effect of timing on PDR⁷.

The optimisation of colonoscopy will impact upon its safety and cost-effectiveness. Increasing numbers of endoscopies are being performed both in the UK¹⁸ and USA¹⁹, partly as a result of successful screening programmes¹⁸. This has led to higher patient flow though endoscopy units. To accommodate this increase in demand, units have had to expand their provision of colonoscopies by number of endoscopy rooms, length of lists or extra lists.

To accommodate this extra-workload, our unit has run regular evening lists (6–9pm) for over two years to meet this increasing demand and improve patient convenience and we are aware of other endoscopy units developing similar strategies. However it is not known whether evening colonoscopy is as effective as traditional day-time colonoscopy. As evening colonoscopy makes up a significant and increasing amount of our overall workload, it is important to establish that the service in terms of performance metrics is appropriate. It is also unknown whether colonoscopy performance declines in the evening; this specific effect, as opposed to morning or afternoon colonoscopy has not been examined in the previous studies. This study aims to review polyp detection rates in our routine clinical practice, with particular reference to this evening session. We aimed to test the hypothesis that polyp detection would be reduced in the evening sessions. The majority of evening colonoscopists will have already worked a full working day. Although not specifically assessed, they may be assumed to be more fatigued than those working in the morning or afternoon lists. We also aim to review PDRs in relation to other factors in routine clinical practice.

Methods

Endoscopy session data was collected retrospectively from the Hospital’s booking system for all NHS outpatient colonoscopies booked for weekdays in the Norfolk and Norwich University Hospital in 2011. Endoscopy sessions were classified as: ‘Morning’ 08.00 to 12.00, ‘Afternoon’ 13.00 to 17.00 and evening 17.30 to 21.00 with colonoscopies usually afforded 30–45 minutes per slot. The times of colonoscopies were recorded; each colonoscopy was grouped by session (morning, afternoon, evening) and ranked by order within their session. Lists sometimes included other endoscopic procedures which were each assigned a value of 1 in the ranking. For example a colonoscopy performed after two oesophago-gastro-duodenoscopies would be ranked number 3 in its session although it was the 1^st colonoscopy. Private patients were excluded.

The Gastroenterology Endoscopy Unit at the Norfolk and Norwich University Hospital provides a full diagnostic and therapeutic endoscopy service to the local population and tertiary referral services in advanced colonoscopic polypectomy, small bowel enteroscopy, endoscopic ultrasound and upper GI endoscopic mucosal resection. The unit is multidisciplinary with colonoscopy performed by medical gastroenterologists, colorectal surgeons and independent nurse endoscopists, with patients booked to most procedures from a common waiting list, although the specific bowel cancer screening lists are only performed by 6 of the endoscopists. The majority of colonoscopy procedures are performed on colonoscopy-only lists, a smaller number are on mixed lists. It is usual practice for our endoscopists to only perform one colonoscopy session (morning, afternoon or evening) per day; in the relatively rare occurrence of an endoscopist providing two sessions within the same day, our usual policy is to have one as a predominantly gastroscopy session.

The unit is the regional endoscopic training centre and an accredited bowel cancer screening centre. The study was approved by and registered with the Audit and Clinical Governance Department at Norfolk and Norwich University Hospital, (project code 34810) the project was regarded as service evaluation and approval from a Research Ethics Committee was not required.

Computerised records were retrieved individually for each colonoscopy from the endoscopy reporting system, Scribes^® (iSoft Health, Banbury UK). Patient demographics, indication, endoscopist, trainee-involvement and sedation information were recorded retrospectively along with information specific to the procedure itself, namely bowel prep quality, diagnosis and polyps detected.

Colonoscopy patients received 2 sachets of Picolax^® (Sodium picosulfate with magnesium citrate, Ferring Pharmaceuticals Ltd, West Drayton UK) bowel preparation, 1 sachet at 6pm the day prior to procedure and 1 sachet at 6am on the day of procedure, for morning and afternoon endoscopies and at 6am and 2pm on the day of the procedure for evening colonoscopies. Quality of bowel prep was subjectively recorded as poor, satisfactory or good by the endoscopist at the time of the procedure. Throughout the study period Olympus-240 and -260 series video-endoscopes were used. Endoscopists recorded their findings themselves immediately after each procedure with polyps qualitatively described and site recorded as accurately as possible.

The indication for colonoscopy was recorded and later simplified into 6 categories for numerical analysis, such that each category contained had at least 10 colonoscopies in each of morning, afternoon and evening sessions and at least 50 procedures overall. Categories with fewer than this were grouped under ‘other’. Thirty different endoscopists performed colonoscopies throughout the year. Those performing over 30 colonoscopies were treated as independent variables and those with fewer than 30 to their name were grouped under ‘other’ for analysis.

Sample size calculations were not formally performed as the intention was to review one year’s colonoscopy practice; the final number of procedures included exceeds those calculated to be required by other similar studies⁸.

PDR was defined as the detection of one or more polyps at colonoscopy. Unadjusted PDRs were calculated and demographic information, indication and endoscopists assessed for each colonoscopy session. Due to the considerable variability in these factors between these sessions multivariate regression analysis was performed to adjust for differences. Stata^® 12 (StataCorp LP, Texas USA) was used to perform multivariate regression analysis with the variables outlined in the results section below Results are expresses as adjusted odds ratio (OR) for polyp detection with 95% confidence intervals. Differences between groups were considered statistically significant if the “p” value was < 0.05.

Results

Data from 2576 colonoscopies were included: 1163 (45.1%) were performed in the morning, 1123 (43.6%) in the afternoon and 290 (11.3%) in the evening.

Mean age was lower in the evening sessions (58.2) compared to morning (64.7) and afternoon (62.3). Table 1 demonstrates the breakdown of demographic and endoscopic information between the three sessions. Trainee-involvement was higher in the evening (17.9%) and afternoon (14.6%) compared to the morning sessions (5.5%). Caecal intubation rates (uncorrected for bowel preparation and impassable strictures) were above 94% in each of the three sessions without any significant difference between sessions. Full data on all procedures included are included in the associated data files.

Unadjusted PDR in the morning, afternoon and evening session were 46.4%, 35.9% and 37.2% respectively.

Factors associated with polyp detection were assessed by multivariate logistic regression (Table 2). Male gender (odds ratio (OR) = 1.76, 95% confidence intervals (CI) 1.48 - 2.11, p<0.001), increasing age (OR = 1.04, 95%CI = 1.03 - 1.06, p<0.001) and successful caecal intubation (OR = 2.48, 95%CI = 1.53 - 4.01, p<0.001) were all significantly independently associated with higher polyp detection. The indications ‘faecal occult blood (FOB) screening’ (those subjects in the UK national bowel cancer screening programme) (p<0.001) and ‘polyp surveillance’ (p<0.001) were strongly positively associated and ‘anaemia’ (p=0.01) negatively associated with PDR. Subjectively ‘Good’ bowel preparation was significantly associated with a higher PDR as compared to ‘Poor’ bowel prep. As Table 1 demonstrates, a considerable proportion of the morning endoscopies were performed for FOB screening. Indeed 54.90% of all FOB screening colonoscopies at our unit were performed in the morning sessions. This correlates with the increased mean age in the morning session and high unadjusted PDR in the morning.

Following standardisation of covariates, there was no significant difference in PDR between sessions. With the morning as the reference value, the odds ratio for polyp detection in the afternoon and evening were 0.93 (95%CI = 0.72 - 1.18) and 1.15 (95%CI = 0.82 - 1.61) respectively. PDR was not shown to be affected by rank order of colonoscopy within a list (p=0.904), sedation type or trainee-involvement.

Discussion

Our data correlate with much larger cohorts that have purely analysed colonoscopic diagnoses without any reference to time of procedure in demonstrating a higher PDR in males than females and with increasing age²⁰. Overall, the PDR and caecal intubation rates are comparable to, and in fact slightly better, than those reported in the recent UK-wide audit of colonoscopy practice (which reported an uncorrected caecal intubation rate of 92.3% and polyp detection rate of 32.1%)²¹. In our study PDR was significantly higher with complete caecal intubation demonstrating the validity of this as a performance metric of colonoscopy quality. When more of the colon is visualised, more polyps will be detected and it is also likely that the most skilled and careful colonoscopists are both more likely to complete a colonoscopy to the caecum and visualise the mucosal completely.

Our data do not support our original hypothesis that evening PDR might be lower than other sessions in the day. Our data suggest there is no significant difference in detection of polyps in either the evening compared to the morning, or based on an endoscopy’s rank order within a session. This is contrary to some previous reports in the literature^4–6. The reasons for this are unclear. Our unit aims to minimise pressures on endoscopists by allowing 30 minutes per colonoscopy (45 minutes for bowel cancer screening colonoscopy) and by strictly avoiding more than 5 colonoscopy procedures per endoscopy list. Also, as demonstrated by our large number of endoscopists (30 in this study), we believe our staff are less likely to fatigue. Endoscopy remains part of a varied working life for both physicians and surgeons within our hospital, rather than operators performing solely endoscopy. Equally apart from the bowel cancer screening specific lists, the majority of colonoscopy lists include a variety of patients with different indications. During the course of this study evening colonoscopy was performed by some consultant staff either working the evening as a planned session within their job plan or by other consultant staff or nurse endoscopists working and remunerated for extra sessions above their full job plan. Arguably the latter group may be more fatigued, but again our data show no deterioration of colonoscopy performance when this group was analysed separately.

We would therefore recommend evening colonoscopy sessions as a potential solution to safely and effectively meet increasing endoscopy demand. We feel it is important that any unit instigating evening colonoscopy carefully audits performance-metrics to ensure that colonoscopy performance is satisfactory in these later sessions. It may be that factors related to individual endoscopy units are important in explaining the differences in polyp detection reported in relation to endoscopy timings. The setting was an endoscopy training centre with an internal high priority on quality of endoscopy and well-established practices of governance, peer-review and teaching and it is possible these are drivers that help maintain PDR.

Anecdotal evidence within our patient group suggests these sessions are popular with patients also. The evening sessions were popular with a considerably younger demographic (mean age 58.2) which may be assumed to represent more convenience for working individuals.

Our data suggest evening lists allowed a higher degree of trainee-involvement. This may be due to ward or clinic commitments of trainees in the morning, indeed some registrars chose to stay later for evening colonoscopy lists, when they have no other distractions, in order to gain further experience and enhance their portfolio of procedures.

It is interesting to note that type of sedation made no difference to PDR, however, patient comfort or length of procedure was not measured (although ongoing audits separate to this study show low and appropriate pain scores and high patient satisfaction). A large Italian prospective study showed caecal intubation rates of only 81.6% of unsedated patients²², although data from the UK bowel cancer screening programme suggests that, as in the current study, colonoscopy without intravenous analgesia and/or sedation is not associated with impaired caecal intubation or adenoma detection²³. In our study it is also unknown whether a procedure was attempted without sedation and sedation later given (although in practice the later addition of sedation seems unusual).

One of the flaws in this dataset is the reliance on accurate documentation by endoscopists on each endoscopy report. Parts of their report, such as bowel preparation, are easily open to subjectivity. Equally, we used PDR as our performance metric: at present the information technology available is not suitable for readily determining the ADR, but we believe that the PDR is a useful surrogate marker of the care taken during colonoscopy and has the advantage that this can be readily compared to other large datasets such as the UK colonoscopy audit²¹. That the PDR was higher when bowel preparation was recorded as ‘good’ compared to ‘poor’ (OR=1.46, 95%CI = 1.03-2.07) may indeed reflect this subjectivity when endoscopists blame the quality of images for the inability to visualise any polyps.

Although it was not a primary outcome measure, we noted that the proportion of colonoscopies recorded as having ‘good’ bowel preparation was much higher in the afternoon (82.4%) compared to the morning (59.0%) sessions. We had not assessed patients’ adherence to laxative dosing protocols but if adherence were equivalent this may suggest the protocols we use require adjustment. Previous recommendations have proposed split-dose bowel preparation as the standard of care²⁴, which we adhered to.

A validated bowel preparation scale used prospectively may improve the reliability of these data and it is recommended that this be included in future evaluations.

Other limitations of this study are inherent within its design. It was retrospective, without any randomisation to different sessions. The external generalizability of the data may also be questioned although we believe we have a balanced workforce of medical, surgical and specialist-nurse endoscopists that is fairly representative of a modern UK NHS unit. We did not record data on withdrawal time on the non-bowel cancer screening colonoscopies. This would have been useful in comparing these data to other datasets which provide this information.

In conclusion, this study shows that the detection of polyps at colonoscopy is independent of the time of day that the procedure is performed. Specifically, evening colonoscopy is no less effective at detecting polyps than morning colonoscopy. We attribute this consistency to appropriate colonoscopy scheduling and an interested, engaged workforce with varied working patterns, both of which could be directly tested in prospective studies. Evening colonoscopy is a suitable initiative to increase procedure numbers but we recommend continued governance of performance metrics to ensure that timing of procedures does not adversely influence outcomes.

Data availability

figshare: Dataset showing all procedures included in analysis of polyp detection study, http://dx.doi.org/10.6084/m9.figshare.1016982²⁵