Introduction
Colonoscopy is the gold standard to detect and remove precancerous colonic lesions,
such as adenomas. Colonoscopy, performed under proper conditions, reduces cancer morbidity
and mortality [1]. Optimal bowel preparation is a key prerequisite to achieve a high adenoma detection
rate. Inappropriately cleaned colons result in suboptimal detection of relevant lesions
and lead to repeated colonoscopies and shorter surveillance intervals [2]
[3].
There are a number of patient-related factors associated with poorly prepared colons,
such as acceptance of the volume of bowel preparation, inability to follow the instructions,
reduced awareness of health behavior, and health illiteracy [4]
[5]. Several strategies have been used to improve bowel preparation through optimizing
patient education. This can be achieved by the use of simple instruction tools [6]. Efforts that involve direct patient contact, such as patient navigators or nurses
with face-to-face counselling are the most effective [7]. This is also paramount for increasing adherence to colonoscopy screening programs
[8].
Face-to-face patient counselling is resource-rich and time-consuming. The high demand
for colonoscopy services as a result of colorectal screening programs has led to a
surge in interest for more efficient strategies with less personnel, while maintaining
quality levels.
There is evidence to suggest that e-health interventions are effective in improving
information transfer to patients [9]. Internet-based education offers a number of advantages: it visualizes information
in a comprehensible format; it is consistent and accessible at any desired moment;
and it provides the option to remind patients in a timely fashion [6].
We have developed a web-based platform consisting of 3 D animations, video, and voiceover
text to inform patients about the colonoscopy procedure and the preparations needed
[10]. This program mimics the patient journey from pre-colonoscopy consultation in the
outpatient clinic to discharge after the procedure. A single center observational
study compared this platform to nurse counselling, with patients who followed this
program having adequate bowel preparation [11]. Subsequent efforts helped to evolve this program into an interactive computer-based
education (CBE) tool. The main improvement consists of the addition of two-way communication
to make home-based use feasible, with substitution of all of the elements of nurse
counselling [12].
The effects of home-based CBE performance in terms of the quality of bowel preparation
or number of repeated procedures because of inadequately prepared colons are unknown.
We hypothesized that CBE as a modality for patient education is equally effective
as nurse counselling for optimal bowel preparation. We report here on our multicenter
randomized controlled trial with a head-to-head comparison of traditional nurse counselling
versus CBE.
Methods
Study design and patients
We performed a multicenter, prospective, endoscopist-blinded, randomized controlled
trial. Patients were recruited from the gastroenterology departments at four hospitals
in the Netherlands, including one academic, two urban, and one rural-based hospital.
Patients were recruited and underwent their colonoscopy between September 2015 and
December 2017.
We included adults referred for complete colonoscopy requiring bowel preparation who
could provide informed consent. The exclusion criteria were: patients unwilling to
participate; no internet access or relative with internet access; Dutch illiteracy;
audiovisual or mental disabilities. An extensive version of the trial protocol was
published previously [10].
Patients were involved in the development of the CBE, using focus groups for content
feedback. The study group promoted the CBE to the wider public, being nominated in
several jury and public award contests. The CBE won the public vote for the yearly
award issued by the Dutch low literate patient society in 2015.
Ethics
The study was performed according to the principles of the 1975 Declaration of Helsinki
and the CONSORT guidelines for reporting of results were followed. This study was
approved by the Institutional Review Board (Medical Research Ethics Committee of the
region Arnhem-Nijmegen, number 2015 – 1765). The trial is registered under the Dutch
Trial Registry, NTR 5475. No commercial support was provided for the trial.
Randomization
Patients referred for colonoscopy were contacted in person or by a telephone call
to judge their willingness to participate. We employed a structured script to explain
the goals of the trial. The reasons for a patient’s unwillingness to participate were
recorded. After the patient had given consent to participate, the patient identification
number and their email address were entered on a secure online tool (Appendix 1s, see online-only Supplementary Material). This tool randomized patients in a 1:1
ratio per trial site and subsequently informed them by automated email of the type
of education they would receive, being either nurse counselling or CBE. All participating
patients provided written informed consent.
Study procedure
After enrolment in the trial, patients were either invited for a nurse-counselling
session or the CBE. The intervention group received a unique link via email (known
in cybersecurity terms as a hash) that provided access to the web-based platform [11]. In addition to the original software, patients had to complete an online questionnaire
on medication use and their medical history. This form was returned via the secure
tool to the endoscopy unit. An automated evaluation tool screened for potential risk
factors for undergoing colonoscopy. If no red flags (e. g. use of anticoagulant or
antidiabetic drugs, or history of severe cardiopulmonary condition) were noted, the
patient was directly invited for colonoscopy. In other cases, patients were contacted
by telephone or scheduled for an additional outpatient visit. This was also recorded.
Each participating hospital’s CBE had a tailored video and site-specific instructions.
The application outlook used video scripts and 3 D animations that were the same in
all four sites ([Fig. 1]). The CBE customized to the academic trial site is openly accessible via https://trials.medify.eu/cbe-colonoscopy.
The control group was invited by mail to our outpatient clinic for a routine nurse-counselling
visit. The nurse explained the procedure in full, acquired the relevant information
on sedation pre-assessment, and handed out a short written leaflet on purgative use
[13].
After completing either the CBE or nurse counselling, patients were scheduled for
colonoscopy.
Fig. 1 An overview of screenshots taken from the computer-based education that was used
in this trial before colonoscopy, illustrating all the steps in the patients’ journey
(the individuals depicted are actors).
Study design
We used patient reported study questionnaires at several time points ([Fig. 2]). At the first time point (T1), baseline demographic characteristics, previous experience
with colonoscopy, patient satisfaction, and validated questionnaires for e-health
literacy and patient productivity were recorded [14]
[15]. After receiving the patient education (either nurse counselling or CBE), the level
of trait and state anxiety were measured [16].
Fig. 2 Flowchart showing the assessments performed at the various time points.
On the day of colonoscopy, we collected patient information prior to colonoscopy (T2).
Data on laxative use, the information recall test, and patient state anxiety were
collected. Here, we also noted information on sickness absence leave [15]. In the CBE group, the need for additional contact events was scored.
The quality of the bowel preparation during colonoscopy was assessed by the Boston
Bowel Preparation Scale (BBPS) [17]. Colonoscopy-specific data (indication, type of sedation and analgesic, ASA classification)
were collected. Finally, prior to discharge (T3) patient satisfaction measures were
recorded [18].
Endoscopy
The attending endoscopists were blinded to the type of education patients had received.
All were familiarized or updated with the use of the BBPS before onset of the trial.
The trial sites used either polyethylene glycol or sodium picosulfate-based standard
split dose regimes for bowel preparation [19].
Outcomes
The primary outcome of the trial was quality of bowel preparation, as assessed with
the BBPS. We recorded the need for repeat examinations owing to inadequate BBPS scores.
The secondary outcome measures were the patient-related outcome measures, including
sickness absence leave, anxiety levels after instruction and prior to colonoscopy,
satisfaction, and information recall.
Questionnaires
At baseline (T1), we included a validated questionnaire on health literacy (Dutch
validated Health Literacy Scale) [14]. Anxiety levels were assessed at T1 using the State-Trait Anxiety Inventory (STAI).
This commonly used 20-item self-report instrument provides scores ranging from 20
(absence of anxiety) to 80 (high anxiety) and combines both state (dynamic, at one
particular moment) and trait (static, based on character) anxiety levels [16]. Patients reported sickness absence leave at T2 using the adapted iMTA Productivity
Cost Questionnaire to evaluate the macroeconomic effect (Appendix 2s) [15]. Information recall was tested at T2 using the same 10-item information recall test
used in our prior pilot study [11].
For patient satisfaction, two measures were recorded at T3. First, the patients were
asked about their willingness to return to the endoscopy unit. This is commonly used
in the context of patient satisfaction in endoscopy [20]. Next, the Net Promoter Score (NPS) was used with the question “Would you recommend
this endoscopy unit to your peers?” Patients’ scores range from 1 (not at all likely)
to 10 (extremely likely). The NPS is calculated as “% promoters” (scores 9 – 10) – “%
Detractors” (scores 1 – 6) [18].
Statistical analysis
Data was analyzed on an intention-to-treat (ITT) and per-protocol basis. Presentation
of the data included the mean, median, and standard deviations (SD) for quantitative
data. The point estimates are presented with 95 % confidence intervals (CI). For categorical
data, counts and frequencies were used.
We used the relative risk of an inadequately prepared colon to compare the two groups.
In comparable bowel preparation studies, a 90 % success rate (for an adequately prepared
colon) is commonly used, with a 10 % non-inferiority margin as the maximum clinically
acceptable difference [19]
[21]. The non-inferiority power calculation resulted in 180 patients per group, 360 patients
in total. With a margin of ± 60 % attrition of patients before completing the protocol,
based on earlier research, the target number of patients to approach was set at 1000
to acquire an adequate per-protocol sample.
Comparisons between groups were assessed using bivariate analyses. In addition to
the non-inferiority analyses in the ITT and per-protocol population, superiority analyses
(chi-squares, t test, and ANCOVA) were conducted to investigate effects on the secondary outcome
measures. Possible differences between the groups concerning secondary outcomes were
assessed using two-sided testing. P values < 0.05 were considered statistically significant for the secondary outcomes
and no correction for multiple testing was performed as these analyses were considered
exploratory.
Results
From September 2015 to December 2017, a total of 1035 patients were assessed for eligibility.
Of those, 190 patients declined to participate for the reasons shown in [Fig. 3]. A total of 845 patients underwent randomization in four hospitals. After randomization,
161 patients were excluded from further analysis for the following five reasons: missing
data (n = 35), premature withdrawal from the trial (n = 34), not receiving a scheduled
colonoscopy (n = 5), an incomplete colonoscopy due to pain / stenosis (n = 41), or
absence of BBPS score (n = 46). This resulted in 684 patients who were included in
the ITT analysis population. In these patients, the age and sex were not significantly
different between the groups.
Fig. 3 Flowchart of patient participation through the trial.
A total of 497 patients were entered for the per-protocol analysis. A total of 217
patients received nurse counselling, while 280 patients were assigned to CBE ([Fig. 3]). All patients included in the analysis had 100 % adherence to nurse counselling
and to the complete CBE. Of these 497 patients, 100 % completed the baseline study
forms at T1, with lower response rates on pre-colonoscopy (T2) forms (55.6 %) and
post-colonoscopy (T3) forms (47.3 %). The baseline characteristics of the patients
were not significantly different with respect to age, sex, educational level, or ethnicity.
Prior experience with colonoscopy was comparable between the two groups, with 46.8 %
in the nurse-counselled group and 51.5 % in the CBE group ([Table 1]).
Table 1
Baseline characteristics of the 497 patients included in the per-protocol analysis.
|
Nurse counselling
(n = 217)
|
Computer-based education
(n = 280)
|
P value
Nurse versus computer-based education[1]
|
|
Sex, n (%)
|
0.54
|
|
|
106 (48.8)
|
129 (46.1)
|
|
|
111 (51.2)
|
151 (53.9)
|
|
Age, mean (SD), years
|
56 years (14.5)
|
56 years (14.5)
|
0.80[2]
|
|
Ethnicity, n (%)
|
0.76
|
|
|
194 (89.4)
|
242 (86.4)
|
|
|
22 (10.1)
|
30 (10.7)
|
|
Educational level, n (%)[3]
|
0.98
|
|
|
27 (12.4)
|
41 (14.6)
|
|
|
121 (55.8)
|
154 (55.0)
|
|
|
69 (31.8)
|
85 (30.4)
|
|
Prior experience with colonoscopy, n (%)
|
101 (46.8)
|
140 (51.5)
|
0.30
|
1 Chi-squared test, unless otherwise specified.
2
t test.
3 Highest completed educational level was split into three levels: “low,” no education
through to lower secondary education; “middle,” upper secondary and middle vocational
education; “high,” higher vocational and tertiary education.
Primary endpoint
BBPS scores were collected for all 684 patients. In the ITT and per-protocol population,
mean BBPS scores in both groups exceeded the threshold of 6 that is considered adequate.
In the per-protocol population, the mean BBPS in the nurse-counselling group was 8.00
(95 %CI 7.78 to 8.21), which was comparable to that of the CBE group at 7.81 (95 %CI
7.62 to 8.00; P = 0.21).
We subsequently calculated the risk of obtaining an adequate BBPS ( > 6). In the ITT
population, among the CBE group, 93.4 % of patients reached an adequate BBPS score,
compared with 95.8 % assigned to the nurse-counselling group. The 95 %CI of the relative
risk difference (−2.4 %) was −5.8 % to 0.9 %, which was within the prespecified non-inferiority
margin of 10 %.
In the per-protocol population, among the CBE group, 93.2 % of patients reached an
adequate BBPS score, compared with 94.0 % assigned to the nurse-counselling group. The
95 %CI of the relative risk difference (−0.8 %) was −5.1 % to 3.5 %, which was also
within the prespecified non-inferiority margin of 10 %. Therefore, our findings show
that CBE is non-inferior to nurse counselling in both the ITT and per-protocol population
(i. e. that the null hypothesis was rejected).
The difference of −2.4 % (95 %CI −5.8 % to 0.9 %) and −0.8 % (95 %CI −5.1 % to 3.5 %)
in formal testing for superiority showed no statistical difference between the groups
in either the ITT or per-protocol population. The numbers of repeat colonoscopies
owing to inadequate bowel preparation were not significantly different between the
groups in either the ITT or per-protocol population, being 4 (0.6 %) for nurse counselling
versus 7 (1.0 %) for CBE in the ITT group, and 3 (0.6 %) for nurse counselling versus
6 (1.2 %) for CBE in the per-protocol group ([Table 2]).
Table 2
Bowel preparation assessed by the Boston Bowel Preparation Scale (BBPS) during colonoscopy
for patients included in the per-protocol analysis.
|
Nurse counselling
|
Computer-based education)
|
P value
Nurse versus computer-based education[1]
|
|
Rate of adequate bowel preparation (BBPS ≥ 6), n (%)
|
204 (94.0)
|
261 (93.2)
|
0.72 for superiority
Non-inferiority: δ −0.8 % (95 %CI −5.1 to 3.5)[2]
|
|
BBPS score, mean (95 %CI)
|
8.00
(7.78 to 8.21)
|
7.81
(7.62 to 8.00)
|
0.21[3]
|
|
Decision to repeat colonoscopy owing to inadequate bowel preparation, n (%)
|
3 (1.4)
|
6 (2.1)
|
0.53
|
CI, confidence interval.
1 Chi-squared test, unless otherwise specified.
2 Within margin.
3
t test.
Subsegmental BBPS scores in the right, transverse, and left colon were equally distributed
amongst the groups (Table 1s). For excellent BBPS scores (8 or higher), no significant differences were observed
between the groups (73.3 % for nurse counselling vs. 69.3 % for CBE; P = 0.33) (Table 2s).
Secondary end points
Sickness absence leave
Sickness absence leave was significantly lower in the CBE group at 28.0 % in the nurse-counselling
group and 4.8 % in the CBE group (P < 0.001) ([Table 3]).
Table 3
Comparison of the secondary outcomes (need for sickness leave, anxiety, satisfaction,
and information recall).
|
Nurse counselling
|
Computer-based education
|
P value
Nurse versus computer-based education[1]
|
|
Need for sickness absence leave, n (%)
[number of respondents, (%)]
|
35 (28.0)
[125 (57.3)]
|
7 (4.8)
[145 (51.8)]
|
0.001[2]
|
|
Anxiety (STAI) score[3], mean (SD)
[number of respondents (%)]
|
|
Trait anxiety
|
53.4 (5.3)
[202 (93.1)]
|
53.2 (4.8)
[212 (75.7)]
|
0.52
|
|
State anxiety after patient education
|
55.3 (5.5)
[203 (93.5)]
|
54.4 (5.8)
[218 (77.9)]
|
0.10
|
|
State anxiety pre-colonoscopy
|
58.2 (5.7)
[118 (54.4)]
|
578 (5.3)
[144 (51.4)]
|
0.65
|
|
Rise in state anxiety after education and prior to colonoscopy
|
3.1 (7.0)
[111 (51.2)]
|
2.8 (7.9)
[124 (44.3)]
|
0.44[4]
|
|
Patient satisfaction
[number of respondents (%)]
|
|
Net promoter score[5]
|
+ 40.9 %
[110 (50.7)]
|
+ 46.3 %
[121 (43.2)]
|
0.45
|
|
Willingness to return (scale 1 – 10), mean (SD)
|
8.13 (1.35)
[110 (50.7)]
|
8.51 (1.70)
[121 (43.2)]
|
0.06
|
|
Information recall test[6], mean (SD)
|
7.18 (1.17)
[125 (57.6)]
|
7.24 (1.06)
[144 (51.4)]
|
0.70
|
STAI, State-Trait Anxiety Inventory; SD, standard deviation.
1 Independent sample t test unless otherwise specified.
2 Chi-squared test.
3 Range from 20 (no anxiety) to 80 (high anxiety).
4 ANCOVA.
5 %promoters minus %detractors.
6 10 basic-item test score before endoscopy (score range 1 – 10).
Anxiety
Anxiety scores were completed by 235 patients in total, 111 in the nurse counselling
and 124 in the CBE group at baseline and sequentially before colonoscopy using the
STAI. The baseline trait and state anxiety scores were equally distributed in the
two groups: 53.4 (SD 5.3) in the nurse counselling versus 53.2 (SD 4.8) in the CBE
group. This was also the case in state anxiety prior to colonoscopy. Comparing both
groups, we noted an expected rise in anxiety scores between the moment of education
and just before colonoscopy. The small difference in the rise in scores between the
groups, indicating a possible benefit of either modality, was not significant: 3.1
(SD 7.0) versus 2.8 (SD 7.9); P = 0.44 ([Table 3]).
Satisfaction
Patient satisfaction scores with the education before colonoscopy (defined as willingness
to return) were high but not statistically different between the groups. The nurse-counselling
group scored a mean of 8.13 (SD 1.35) out of 10, whereas the CBE group scored 8.55
(SD 1.30); P = 0.059. Second, the NET-promoter scores recorded were + 40.9 % versus + 46.3 %,
respectively, which is also not significant amongst groups (P = 0.45) ([Table 3]).
Information recall
Information recall was tested using a 10-item questionnaire. There was no significant
difference between the groups, at 7.18 (SD 1.17) in the nurse-counselling group, versus
7.24 (SD 1.06) in the CBE group ([Table 3]).
Endoscopy
When asked whether they required additional information prior to the colonoscopy,
among the CBE group, 78.5 % of patients responded negatively and were directly scheduled
for colonoscopy after CBE. In 21.5 % of cases, there was an extra contact event: 18.5 %
by telephone call and 3.0 % at the outpatient clinic.
A total of 70 endoscopists were involved in the trial.
Discussion
This multicenter randomized controlled trial evaluated CBE as an educational tool
for patient counselling prior to colonoscopy. We found in our ITT analysis, as well
as on per-protocol analysis, that CBE is non inferior to nurse counselling in terms
of bowel preparation. At the same time, CBE reduced the number of patient visits to
the outpatient clinic by 79 % compared with conventional nurse counselling. An added
value of CBE is the lower proportion of patients who report taking sickness absence
leave prior to endoscopy. CBE, with two-way communication in place, functions therefore
as a time- and resource-effective nexus between patients and the endoscopy unit.
We also investigated the psychological parameters, such as stress or anxiety, that
may accompany (preparation for) a colonoscopy and found that there were no differences
in the trait (or “character”) anxiety scores between the groups. Similarly, the state
(or “moment”) anxiety scores were comparable between the groups both after receiving
education and just prior to colonoscopy. Finally, CBE showed high scores for patient
satisfaction (Appendix 3s) and information recall at levels similar to those seen after nurse counselling.
There have been several comparable studies that have used various means of electronic
communication. One study enriched patient communication by sending a series of 15
text messages to patients and found that this led to better colonoscopy preparation
being achieved [22]. In addition, digital send instructions increase appointment adherence with less
same-day cancellations [23]. Trials using smartphone apps have shown improved bowel preparation [24]
[25]
[26]. An important difference with these studies using text messages via SMS or a smartphone
app is that our approach aimed to achieve patient engagement through the use of visual
3 D animation as a teaching tool to provide better insight and actual visualization
of the procedure. Also, web-based solutions like ours have the benefit over smartphone
apps that they are ubiquitously available on all devices (e. g. desktop computer,
tablet, or smartphone) without the need for users to download anything first. More
importantly, the fact that our CBE platform may substitute nurse counselling, a common
practice in several healthcare services, is a novel element and relevant to policy
makers [27].
Previously, it has been hypothesized that there is a “ceiling effect” of 90 % adequate
bowel preparation score for educational interventions that aim to influence the scores
in any endoscopy unit [28]. As a result, these interventions will be beneficial in underperforming units with
scores well below the 85 % benchmark that has been advised by the US Multi-Society
Task Force on Colorectal Cancer Screening [29]. In several recent (non-Western) studies that demonstrated improved bowel cleanliness
by smartphone intervention, baseline scores in the control group were often below
this point (73.6 % – 77.2 %) [25]
[30]. By contrast, in our four trial units (already performing well above 90 % adequate
bowel preparation in controls), the ceiling effect might have prevented the detection
of any meaningful superiority difference. We therefore adopted the non-inferiority
design, which was novel to this type of research.
Initiatives in other fields using the same functionality of CBE have shown that it
can reduce the number of outpatient visits. For example, use of CBE improves patient
self-management in inflammatory bowel disease, diabetes, asthma, and chronic obstructive
pulmonary disorder [31]
[32].
We realize that CBE is not suitable for every patient. Patients with low (e-health)
literacy are less likely to benefit. In our trial, 3.0 % of patients paid an extra
visit to the hospital despite CBE. CBE should therefore be positioned as an adjunct
to nurse counselling in vulnerable patient groups, as they might need alternative
access to relevant healthcare information.
The implication of our finding is that CBE may save valuable time for nurses and free
up resources. With the growing future need for colonoscopies, due to the national
colorectal cancer screening program and subsequent surveillance colonoscopies, and
the current problems in recruiting nursing staff in Dutch hospitals, this is very
relevant [33]
[34].
Our randomized clinical trial comes with strengths and limitations. In terms of strengths,
our trial was conducted with a large real-life sample of patients. The non-inferiority
hypothesis and power allowed robust statements on the efficacy of CBE. We tested this
CBE in a real-world setting, with patients having a variety of indications (Table 3s), both with and without previous experience of colonoscopy. Also, we used three different
types of endoscopy unit, using a variety of different practices (Tables 4s and 5s), so the results are generalizable to daily practice. In the catchment area of our
endoscopy centers, CBE can be used in up to 94 % of patients undergoing colonoscopy
[11].
On the other hand, our trial comes with limitations. There were a significant number
of drop outs after randomization owing to inclusion failures; however, this did not
result in an unequal distribution regarding baseline characteristics between the arms
in either the ITT or per-protocol populations, limiting the risk of selection bias.
Because of the use of patient reported questionnaires, we do not have 100 % data collection
at all time points, although the trial protocol called for that. While this did not
affect our main outcome, it might have affected the assessment of secondary outcomes,
such as anxiety and satisfaction. Satisfaction was measured several hours after administration
of sedatives. Sedatives may cause a euphoric effect after administration and result
in higher overall scores. However, the types of sedative used were distributed equally
over the groups (data not shown), precluding bias.
We did not collect complete medical histories of our patients, including previous
abdominal surgery, or risk factors for poor bowel preparation, such as diabetes mellitus,
constipation, or the use of motility influencing drugs. We surmise that the effect
of these risk factors on the bowel preparation efficacy in our trial is limited, in
view of the small difference in BBPS scores. We did not collect data on adenoma detection
rate (ADR) as this was outside the remit of this clinical trial. From the literature,
the robust correlation between adequate BBPS and ADR suggests that BBPS is a good
technical proxy parameter [35].
In conclusion, in this trial, we have established non-inferiority for CBE compared
with nurse counselling prior to colonoscopy in terms of bowel preparation. This finding
paves the way for further upscaling of CBE in endoscopy units to prepare their patients
more effectively before colonoscopy. Preparing patients with CBE reduces the need
for outpatient clinic capacity, leading to less absenteeism at work, high satisfaction
scores, and good recollection of information.
