Article Text
Abstract
Objective Pain associated with colonoscopy is a major burden for patients. We investigated modifiable factors associated with patient-reported pain during and after colonoscopy.
Design This cross-sectional analysis included database records from 23 centres participating in a population-based colonoscopy screening programme in Poland. Colonoscopies were performed under three sedation modalities: none, benzodiazepine-opioid sedation or propofol sedation. We used Gastronet (a validated tool) to assess patients’ pain during and after colonoscopy; pain was scored on a four-point scale (no, little, moderate or severe pain), with moderate to severe defined as painful. We used multivariate logistic regression models to estimate ORs for painful colonoscopy and calculated risk-adjusted ratios of painful colonoscopies per endoscopist and compared it to the mean rate.
Results Of 35 216 screening colonoscopies in 2014 and 2015 included in our study, 22 725 (64.5%) patients returned valid Gastronet questionnaires. The proportion of examinations described as causing pain during (after) the procedure was 22.5% (14.2%) for unsedated, 19.9% (13.5%) for benzodiazepine-opioid sedation and 2.5% (7.5%) for propofol sedation. Propofol sedation, higher case volume of endoscopists, newest endoscope generation and adequate bowel preparation were significantly associated with lower odds of painful colonoscopy. Pain scores after colonoscopy showed similar associations. Adjusted pain rates during and after colonoscopy varied 11 and over 23-fold, respectively, between endoscopists.
Conclusion We identified several independent, modifiable factors associated with pain during and after colonoscopy, of which individual endoscopist was the most important. Dedicated training should be considered to decrease variability among endoscopists.
- colonoscopy
- abdominal pain
- colorectal cancer screening
Statistics from Altmetric.com
Significance of this study
What is already known on this subject?
Colonoscopy pain is a major barrier for participation in colorectal cancer screening.
Several factors associated with colonoscopy pain were identified previously (female sex, low body mass index, previous abdominal surgery); however, they are unmodifiable.
Sedation can reduce colonoscopy pain rates; however, effectiveness of different types of sedation has not been fully quantified.
What are the new findings?
High case volume of endoscopist, newer endoscope generation and adequate bowel preparation are modifiable factors associated with lower odds for painful colonoscopy.
Propofol sedation is very effective in prevention of pain during colonoscopy, whereas benzodiazepine-opioid sedation is not significantly better than no sedation.
Rates of painful colonoscopies vary widely between endoscopists, even after adjusting for all modifiable and unmodifiable risk factors.
How might it impact on clinical practice in the foreseeable future?
Modifiable factors identified in the study, as well as previously identified unmodifiable factors should be taken into consideration when sedation is proposed or denied.
If sedated colonoscopy is considered necessary, propofol sedation rather than benzodiazepine-opioid sedation should be the method of choice.
Targeted training to reduce endoscopist’s variation in painful colonoscopy rates is needed.
Introduction
Colorectal cancer (CRC) is the second leading cause of death from cancer in Europe.1 Several modalities of CRC screening exist, including primary colonoscopy, primary sigmoidoscopy and faecal occult blood testing.2 Each screening modality includes colonoscopy as a primary or secondary test following positive sigmoidoscopy or faecal occult blood test.
Patients perceive colonoscopy as a painful and unpleasant procedure. This is one of the major barriers for participation in CRC screening,3 particularly since those being screened are usually healthy and asymptomatic.
Some patient-related factors associated with colonoscopy pain have been identified, but for the most part they cannot be modified (eg, female sex, low body mass index (BMI), and history of abdominal surgery).4–6 Of the procedure-related factors, sedation greatly reduces pain associated with colonoscopy. Several studies have shown better procedure tolerance when patients are sedated.7–9 The two main types of sedation are propofol based or a combination of benzodiazepines and opioids. In most countries, an anaesthesiologist must perform propofol sedation, whereas benzodiazepine-opioid sedation can be performed by endoscopists themselves. The disadvantages of sedated colonoscopy are higher cost and greater risk of sedation-related (eg, pneumonia and cardiovascular events) and procedure-related (eg, bowel perforation and bleeding) complications.10 Moreover, it reduces the capacity of colonoscopy screening facilities, as it requires more time, resources and personnel.11
Since 2014, the Polish Colorectal Cancer Screening Programme (PCSP) has routinely used the Gastronet questionnaire to measure patient-reported outcomes.12 The Gastronet is a Norwegian quality assurance programme that was initiated in 2003.13 14 Patients fill in the questionnaire at home 1 day after colonoscopy and send it back via traditional mail. Questions on pain are answered on a four-point scale: no pain, slight pain, moderate pain and severe pain. This scale has been validated previously.14
Our aim in the present analysis was to identify modifiable factors associated with patient-reported pain during and after colonoscopy performed in a programmatic screening setting.
Methods
PCSP design and subjects
The design of PCSP has been described previously.12 Briefly, it is a programmatic, primary colonoscopy screening; roll-out began in 2012. All individuals aged 55–64 years who live in the geographic target area of a screening centre are considered eligible, but the programme excludes people with a CRC diagnosis or previous screening colonoscopy. An updated list of eligible individuals is acquired yearly from the national population registry so that letters of invitation can be mailed. The central screening database includes patient data, colonoscopy and histopathology reports, responses to the Gastronet questionnaire and information on screening centres and endoscopists.
In our study, we included all primary colonoscopies performed within the PCSP framework in 2014 and 2015. Figure 1 presents the study flowchart and exclusions. Colonoscopies were performed under three different sedation modalities: no sedation, benzodiazepine-opioid sedation (without an anaesthesiologist) and propofol sedation (with anaesthesiologist).
Gastronet questionnaire
All screenees are given the Gastronet questionnaire to be filled in at home on the day after colonoscopy. The questionnaire is sent back via mail (in a prepaid envelope) to the coordinating office, where the forms are scanned and automatically uploaded into the screening programme database. The questionnaire includes four closed-type questions: (1) Were you satisfied with centre’s quality? (yes; no). (2) Was the procedure painful? (no; yes, slightly; yes, moderately; yes, very). (3) Did you feel any discomfort or colicky abdominal pain after the procedure? (no; yes, slightly; yes, moderately; yes, very). If your answer was yes, how long did you feel the aforementioned symptoms? (<1 hour, 1–3 hours, 3–6 hours, >6 hours). (4) Are you satisfied with information on the procedure itself and its results? (yes, no, moderately) The language of all of the questions and answers has been validated previously.15
Data acquisition
We gathered all patient records from the screening database for the years 2014 and 2015.
Data included:
Gastronet results for question on pain during and after colonoscopy; answers were divided into two clinically relevant5 categories: no or slight pain and moderate or severe pain. The latter was referred to as ‘painful’ colonoscopy.
Patient factors: age, sex, previous abdominal surgery, previous colonoscopy and BMI.
Sedation type: no sedation, benzodiazepine-opioid sedation or propofol sedation.
Procedure factors: bowel preparation for procedure (Boston Bowel Preparation Scale (BBPS) total and segmental scores, with adequate preparation defined as total score ≥6 and each segmental score ≥2), procedure completed to the caecum, most advanced lesion categorisation16 and endoscope generation (full list of endoscope models is presented in online appendix 1).
Endoscopist factors: proportion of examinations with caecum intubation (CIR), proportion of examinations with at least one adenoma detected (ADR),17 specialty of endoscopist (gastroenterology, surgery or other (Polish Society of Gastroenterology certificate of competence in colonoscopy was obligatory)), endoscopist’s age and sex and recent screening colonoscopy case volume (number of colonoscopies in screening setting during 2012–2014).
Supplementary file 1
We did not gather data about CO2 usage, because it was not widely available in the PCSP centres and about withdrawal time, which is not routinely recorded in the PCSP.
Factors not found relevant (even though analysed) are not presented in the results.
Statistical analysis
For each type of pain (during or after colonoscopy), we built univariate and multivariate logistic regression model to estimate ORs and 95% CIs for painful colonoscopies. We used backward stepwise regression with a 0.1 level of significance for included variables and interaction terms between sedation type and sex, previous abdominal surgery and BMI. The standard errors in the models were adjusted for 23 clusters defined by endoscopy centres.
The models were built on a random sample of two-thirds of all analysed colonoscopies (derivation dataset) and then tested for goodness of fit on the remaining one-third of all analysed colonoscopies (validation dataset). In sensitivity analyses, the same models were created for each type of sedation (no sedation, benzodiazepine-opioid sedation and propofol sedation) and pain (during and after colonoscopy), separately.
We treated age as a continuous variable. Reference values for each factor were: female sex, BMI <25 kg/m2, suboptimal bowel preparation (BBPS <2/2/2), no previous abdominal surgery, no previous colonoscopy, endoscopist CIR <95%, recent screening colonoscopy case volume <100, endoscopist specialty in gastroenterology and the first generation of endoscopes (see online appendix 1) (tables 1 and 2 and online appendix 2).
We performed a Hosmer–Lemeshow test to assess the models’ goodness of fit, where p<0.05 was considered to denote significant lack of fit.18 Based on the multivariable model results, we calculated risk-adjusted ratios for painful colonoscopy for each endoscopist. To ensure the robustness of the estimates, we performed calculations only for endoscopists who had performed ≥30 colonoscopies with a given sedation type.19
We used t-test, χ2 test and Fisher’s exact test to compare variables’ distributions between the groups and the McNemar test to determine marginal homogeneity (ie, if there was a difference between patients experiencing pain during and after colonoscopy). All tests were two sided. A p value of <0.05 was considered statistically significant. All analyses were performed with Stata software, V.13.1 (Stata). Figures were prepared with R statistical software V.3.0.1 (R Development Core Team).
Results
Baseline characteristics
Of 43 277 colonoscopies performed in 2014 and 2015, 35 216 procedures were eligible. The final analysis included 22 725 valid Gastronet questionnaires that had been returned by patients (response rate of 64.5%). Figure 1 presents the study flowchart. The male-to-female ratio was 1:1.06, and the mean age was 59.9 (SD 2.7) years. Eighty-seven endoscopists performed colonoscopies in 24 centres.
The majority of colonoscopies (63.4%) were performed without any sedation; 22.3% were performed with benzodiazepine-opioid sedation and 14.3% with propofol sedation. Severe or moderate pain during colonoscopy was reported by 21.9% of unsedated patients, 19.2% of patients under benzodiazepine-opioid sedation and 1.6% of patients under propofol sedation. Severe or moderate pain after colonoscopy was reported by 14.2% of unsedated patients, 13.5% of patients after benzodiazepine-opioid sedation and 7.5% of patients after propofol sedation. Online appendix 3 presents detailed information on pain levels by sedation type. In the pie charts in figure 2, each section represents patients who experienced no pain at all, only one type of pain (either during or after colonoscopy), or both types of pain, by sedation type. There was a significant difference in margin totals between pain during and after colonoscopy (p<0.001 for each sedation type).
CIR adjusted by pain and sedation type
Significant CIR variation was observed depending on sedation modality and reported pain level. Generally, caecum was reached in 98.2% of unsedated, 97.9% of sedated with benzodiazepine-opioid and 99.1% of propofol sedated procedures. These differences were statistically significant (p<0.001); however, in all cases the benchmark value of 95% was reached. Caecum was reached significantly more frequently when the patient reported no pain as compared with painful colonoscopies (99.2% vs 94.1%, respectively, p<0.001). Similar difference was observed for pain after colonoscopy (98.8% vs 95.2%, p<0.001). We have also observed CIR variation for painful colonoscopies, depending on sedation type: 94.1% unsedated, 94.1% sedated with benzodiazepine-opioid and 96.2% propofol-sedated painful procedures were complete (difference is not statistically significant with p=0.930).
Modifiable factors associated with pain during colonoscopy
Online appendix 2 contains the univariate model. In multivariate regression model clustered for endoscopy centre, among the modifiable factors, propofol sedation, adequate bowel preparation, endoscopists’ high case volume of screening colonoscopies and fourth endoscope generation reduced the odds of painful colonoscopy for pain experienced during colonoscopy (table 1). The most pronounced effects were observed for propofol sedation (OR of 0.03 for women and 0.13 for men for painful colonoscopy) and fourth-generation endoscopes (OR 0.52). Of unmodifiable or poorly modifiable factors, the male sex both of patients and endoscopists, patient BMI≥25 were associated with significantly lower odds for painful colonoscopy. By contrast, ORs for painful colonoscopy were increased in older patients, with previous abdominal surgery and previous colonoscopy. ADR was not found to be associated with painful colonoscopy. Hosmer–Lemeshow test revealed (table 1) no significant lack of fit. Additional models (separate for each sedation type) are presented in online appendix 4.
Modifiable factors associated with pain after colonoscopy
Online appendix 2 contains the results of a univariate model. In a multivariate regression model, corrected for clustering endoscopy centres, among the modifiable factors, propofol sedation, adequate bowel preparation, endoscopist CIR ≥95% and newer endoscope generation reduced the odds of painful colonoscopy for pain experienced after colonoscopy (table 2). The most pronounced effects were observed for propofol sedation (OR 0.21–0.57, depending on the patient’s sex and previous abdominal surgery status) and fourth-generation endoscopes (OR 0.56). Of unmodifiable or poorly modifiable factors, the male sex and BMI ≥30 were associated with significantly lower odds for pain after colonoscopy. By contrast, ORs for pain after colonoscopy were increased in patients who had colonoscopy prior to screening. ADR was not found to be associated with pain after colonoscopy. Hosmer–Lemeshow test revealed (table 2) no significant lack of fit. Additional models (separate for each sedation type) are presented in online appendix 4.
Variation among endoscopists
We analysed each endoscopist’s performance in terms of painful colonoscopies, and we adjusted for factors that were significant in multivariate models. Figure 3 shows funnel plots for each endoscopist’s performance (adjusted risk ratios) compared with the mean rate of painful colonoscopies for pain during and after colonoscopy. The funnel plots clearly distinguished overperformers and underperformers in terms of painful colonoscopy. Moreover, overperformers and underperformers were inconsistent for each type of pain (data not shown). The adjusted risk ratios of painful colonoscopy were 3.6%–41.4% (mean 18.5%) for pain during colonoscopy and 0%–23.6% (mean 13.1%) for pain after colonoscopy.
Discussion
Our analysis identified several, modifiable factors associated with pain during and after colonoscopy: propofol sedation, endoscope fourth generation, adequate bowel preparation (based on BBPS≥6) and endoscopist’s high case volume. The most effective way of reducing pain during colonoscopy was propofol sedation (OR of 0.03 and 0.13 for women and men, respectively), as identified by this study. However, propofol sedation is not always available and has its drawbacks, including increased cost, organisational complexity and a small but significant increase in complication rates.10 The most commonly used alternative, a benzodiazepine-opioid sedation was not as effective as propofol sedation, with the proportion of painful colonoscopies similar to that of unsedated colonoscopies (figure 2 and online appendix 3). Moreover, both types of sedation were much less effective at reducing pain after colonoscopy. These findings, in addition to the previously reported increased OR for complications after sedated colonoscopy,10 imply that sedation, particularly non-propofol based, is not a universal means of reducing odds for pain associated with colonoscopy. In the Polish Colonoscopy Screening Programme (PCSP), each centre needs to provide sedation free of charge in at least 20% of colonoscopies. On registration, selected patients (after abdominal surgery, women with low BMI, with fear of colonoscopy) are proposed to undergo colonoscopy with sedation; however, this selection is up to individual centre discretion.
To the best of our knowledge, this is the first study to show an association between newest endoscope generation and lower painful colonoscopy rate. We observed the reduction in OR for painful colonoscopy with the fourth-generation endoscopes (online appendix 1). The main differences between these models and previous ones are variable stiffness (also present in some third-generation models) and responsive insertion technology. We were unable to compare fourth-generation models from different manufacturers because the included centres did not use them. Previous studies have revealed that variable stiffness has advantages, including lower patient-reported pain levels, reduced loop formation and higher CIR.20 21 Some previous studies found an association between responsive insertion technology and reduced odds of painful colonoscopy; however, the data were not consistent.22–24 Responsive insertion technology was associated with parameters reflecting easier colonoscope use and reduced insertion time. Moreover, a reduced bending propensity with endoscope usage25 likely contributes to higher odds for painful colonoscopy with older generation endoscopes relative to newer generation endoscopes. Although we do not have data regarding overall endoscope usage (outside of screening), we assume that the older the endoscope, the more it had been used. It may be assumed that more skilled endoscopists have access to newer technologies; however, due to the fact that (1) the multivariate models were adjusted for potential clustering at centre or endoscopist level and (2) the variation between endoscopists exists after adjustment for relevant factors, we have presumably taken into account such confounding. We did not find that paediatric endoscopes were superior to traditional adult colonoscopes in terms of painful colonoscopy (data not shown), which is consistent with prior studies.21 26 27
We also showed the impact of adequate bowel preparation on pain scores during and after colonoscopy. An adequate bowel preparation could reduce the OR for painful colonoscopy by decreasing the procedure time (unfortunately, the PCSP did not measure insertion and withdrawal times) and making endoscope insertion easier, thereby contributing to better procedure tolerance.28 29 Moreover, adequate bowel preparation may facilitate complete gas removal during colonoscope withdrawal, thereby contributing to pain reduction after colonoscopy. This strengthens the rationale for adequate bowel preparation as an important colonoscopy quality measure.
Another important endoscopist factor was case volume, represented by the number of screening colonoscopies performed in the previous 3 years. We found that a high endoscopist case volume was associated with a reduced risk of painful colonoscopy, as previously reported.30 We did not, however, record case volume outside of screening setting.
Among unmodifiable and poorly modifiable factors, we confirmed an association between painful colonoscopy and female sex, lower BMI and previous abdominal surgery.4–6 The increased OR for painful colonoscopy that we found in patients with a history of previous colonoscopy was unexpected. Since we excluded patients who had undergone previous screening colonoscopy in the PCSP, we believe that those patients included in our study who had undergone a previous colonoscopy were likely symptomatic at that time. It is possible that they continued to have symptoms of irritable bowel syndrome, which is known to impair colonoscopy tolerance.31 32 Unfortunately, we did not collect data on abdominal symptoms, therefore this hypothesis needs further verification.
We also analysed factors associated with pain after colonoscopy, which is a different phenomenon, attributable mostly to insufficient gas removal during endoscope withdrawal.33 34 Gas removal may be impaired due to polypectomy, as colon distention is more pronounced and procedure time is prolonged in these cases. We did not, however, find significant differences in reported pain level between colonoscopies with or without polypectomy (data not shown). Previous studies showed that, compared with air, CO2 insufflation may significantly reduce post-procedure pain.15 35 Unfortunately, CO2 is not routinely used in the PCSP. We found an association between pain after colonoscopy and factors similar to those associated with pain during endoscopy. However, a considerable subset of patients experienced moderate-to-severe pain after, but not during colonoscopy (figure 2). This was particularly true for sedated procedures.
We found that the likelihood of painful colonoscopy varied widely among endoscopists (figure 3), resulting in more than 11-fold variation of painful colonoscopy rate (more than 23-fold variation for pain after colonoscopy). The importance of this finding is strengthened by the fact that we calculated rates of painful colonoscopies after adjusting for factors the multivariate regression model found to be statistically significant, including sedation (figure 3). Moreover, endoscopists who performed outside of the 95% CI were inconsistent for pain during and after colonoscopy meaning that endoscopists represented by red dots in plot for pain during colonoscopy were different from endoscopists represented by red dots in plot for pain after colonoscopy (data not shown).
Previous studies have shown differences among endoscopists.4 13 14 36 However, we are unaware of a previous study that comprised a population-based screening setting with three sedation modalities, a large number of endoscopists and adjustments for many patient, endoscopist and centre factors. Moreover, the multivariate models were clustered for centre (tables 1 and 2) and endoscopists factor (data not shown; similar factors with similar ORs were identified in these models), excluding one of potential selection biases. These adjustments allowed us to attribute the differences that we observed for painful colonoscopy, with high probability, to the technical and non-technical skills possessed by each endoscopist.9 These skills comprise both the interpersonal skills required to establish a meaningful patient–endoscopist relationship and individual manual skills.
Our study showed that the variation among endoscopists in terms of painful colonoscopy occurred regardless of the sedation type and patient population, as figure 3 presents rates of painful colonoscopies adjusted for all significant factors found in multivariate models. This suggests that directed training should be considered to reduce the rate of painful colonoscopies. Technical and non-technical skills can be either improved or developed with such training, which should focus on reducing the odds for pain during colonoscopy and improving the retrieval of gas during colonoscope withdrawal (particularly when CO2 is not used for insufflation). In our screening setting, the proportion of unsedated painful colonoscopies (21.9%) was lower than those previously reported (ie, 33.8%, 32% and 26%).4 5 13 14 We attribute this to the improvements in specific training required by endoscopists in Poland to enable them to perform unsedated colonoscopy successfully. This training is necessary due to the low availability of anaesthesia services. On the other hand, endoscopist performance in the PCSP clearly shows (figure 3) that a <20% rate of painful colonoscopies is achievable and should be considered as a minimum standard, especially for screening of healthy individuals. Depending on the country or setting, directed training to perform painless colonoscopy37 or more extensive use of sedation should be considered.
The two main strengths of the study are the high number of patients recruited from a population-based setting and the high-quality database, which allowed us to perform meaningful analyses of different factors related to patients, procedures, endoscopists and centres. Moreover, our screening programme is conducted in public academic, public non-academic and private centres.12 To our knowledge, this was the first study to directly compare the effects of sedation with propofol and benzodiazepine-opioid on patient pain during and after colonoscopy. Moreover, we calculated endoscopist performance in terms of adjusted mean rate of painful colonoscopies and found that endoscopists’ individual technical and non-technical skills contributed considerably to pain reported by patients.
Study limitations include a substantial dropout rate due to insufficient data (18.6%, mostly attributable to incomplete database records; see figure 1). Our response rate was 64.5%, which is satisfactory; however, it is far from usual values reported for Scandinavian countries.13–15 36 Yet, in recent publication by Hoff et al, the response rate was similar to currently observed.38 Response rate of 64.5% is consistent with previously reported rates in Poland.2 This is an important limitation, as we were unable to assess what was the reason for not returning the questionnaire. We assume that the proportion of patients withholding critique (for example omitting to report high pain level or dissatisfaction with centre quality) is similar to non-respondents among persons being satisfied and thinking that their feedback will have little impact on the results. We did not record withdrawal time, type of insufflation (however CO2 insufflation is not routinely used in PCSP) and did not have access to data about the endoscopists’ case volume outside the screening programme. However, we believe that their case volume in the screening setting can be regarded as proportional to their total experience. Finally, we did not record doses of propofol or other agents used for sedation. However, all agents were dosed according to manufacturer guidelines.
Female sex, previous abdominal surgery, endoscopist CIR >95% and high screening colonoscopy case volume (higher proportion for assisted sedation) were disproportionately represented in the sedation modality groups (online appendix 5). Due to using multivariable models, however, the final results were not influenced by this disproportion.
In conclusion, we determined that propofol sedation, adequate bowel preparation, newer generation of endoscopes and high case volume of endoscopists were modifiable factors associated with lower rates of painful colonoscopy. The endoscopist per se was the most powerful predictor of painful colonoscopy, independent of sedation type and patient population. Future efforts should aim to reduce the variation in endoscopist performance; this is most likely to be accomplished by means of dedicated training.
Acknowledgments
We would like to thank Malgorzata Pisera for additional help in contacting participants and coordinating Gastronet surveys.
References
Footnotes
Contributors MB, PW, GH, MR and JR, MFK: study design, data acquisition,analysis and interpretation; MB: writing the manuscript; PW, GH, MR, JR, MFK: manuscriptedition; GH: original Gastronet creation; MFK: polish Gastronet version adaptation.
Funding This work was funded by grant Pol-Nor/204233/30/2013 from the Polish-Norwegian Research Programme through the National Centre for Research and Development of Poland.
Competing interests MFK is on the advisory board of Alfa Wasserman and has spoken and taught for Olympus Poland. JR is on the advisory boards of Alfa Wasserman, Ipsen Pharma, Polpharma and Takeda and hasa travel grant from Abbvie. The other authors have no competing interests.
Patient consent Detail has been removed from this case description/these case descriptions to ensure anonymity. The editors and reviewers have seen the detailed information available and are satisfied that the information backs up the case the authors are making.
Provenance and peer review Not commissioned; externally peer reviewed.