Systematic review: oral bowel preparation for colonoscopy

Authors


Dr J. Belsey, JB Medical Ltd, The Old Brickworks, Little Cornard, Sudbury CO10 0PB, UK.
E-mail: jbelsey@jbmedical.com

Abstract

Summary

Background

There are many published trials of colon cleansing regimens for colonoscopy but no clear consensus regarding relative performance.

Aim

To identify high quality controlled trials comparing two or more bowel preparation regimens and to compare efficacy and tolerability.

Methods

A comprehensive systematic review was carried out to identify candidate studies. Quality appraisal was carried out on all identified studies. Results were meta-analysed where possible and qualitatively compared if not.

Results

Eighty-two studies qualified for analysis. Polyethylene glycol and sodium phosphate were the most frequently investigated preparations. There was no significant efficacy difference between the two, but sodium phosphate was better tolerated. Sodium picosulphate/magnesium citrate, a commonly prescribed preparation, was investigated in four studies, with no clear benefit over other regimens demonstrated. Safety was not recognized as a problem in the randomized controlled trials. Published case series demonstrate that sodium phosphate is associated with the highest risk of clinically significant electrolyte disturbances.

Conclusion

Shortcomings in study design limit the value of many of the studies. Based on these results, no single bowel preparation emerges as consistently superior. New preparations are required that combine better efficacy and tolerability, in addition to rigorous new validated study designs, allowing unequivocal comparisons to be made.

Introduction

Colonoscopy is considered to be the gold standard investigation for assessing the colonic mucosa. To be an effective imaging and therapeutic tool, it is essential that the organ is visualized from anal verge to ileocaecal valve. Only a minority of UK centres achieve at least a 90% caecal intubation rate with large surveys reporting mean completion rates of 59–78%.1, 2 A number of different causes account for the suboptimal performance of colonoscopy. In 35% of failed procedures, patient discomfort was a major factor, whilst uncontrolled looping accounted for 30%. In approximately one in five incomplete colonoscopies, poor bowel preparation prevented completion. In a recent report from France, an analysis of 7205 colonoscopies reported 96% of procedures successfully completed. However, amongst those that failed, 32.7% were attributable to inadequate bowel cleansing.3 This rate is similar to that reported in an earlier survey carried out in 2000, where one-third of failures were ascribed to poor preparation.4

Given the choice of laxative regimens available, the high proportion of inadequately prepared bowels is surprising. Colon cleansing preparations can be broadly classified into three groups. Osmotic laxatives are the most commonly used and include agents such as sodium phosphate (NaP), magnesium citrate and mannitol. These increase colon water content by attracting extracellular fluid efflux through the bowel wall and maintaining oral fluids in the lumen. Polyethylene glycol (PEG), while similar to the osmotic laxatives, works somewhat differently. It is a high molecular weight non-absorbable macrogol polymer which is administered in a dilute electrolyte solution. As a result of the osmotic effect of the polymer, the electrolyte solution is retained in the colon, where it acts as a bowel cleanser. There is little fluid exchange across the colonic mucosal membrane, thereby limiting the potential for systemic electrolyte disturbance. Available preparations include macrogols of differing mean molecular weight (3550 or 4000 Da) and electrolyte composition. There is no theoretical reason to suppose that these will vary in bowel cleansing efficacy. The third group are the stimulant laxatives such as senna, sodium picosulphate and bisacodyl, which work principally by enhancing bowel wall smooth muscle activity. In addition they may also increase bowel water content.

The choice of bowel preparation actually used does not appear to reflect the evidence base in terms of efficacy. In a survey which sampled bowel preparation in 68 UK hospitals,2 the most widely used bowel preparation regimens were sodium picosulphate + magnesium citrate (36.8%), polyethylene glycol + electrolytes (20.7%) and NaP (15.6%). Whilst sodium picosulphate + magnesium citrate was most commonly prescribed, use of this preparation was also most commonly reported as unsatisfactory. Those administered PEG or sodium sulphate showed no difference in the quality of the preparation and yet remained in the minority.

Overall, the results demonstrate that existing cleansing protocols do not provide sufficiently predictable bowel clearance to guarantee colonoscopic visualization of the entire colon. Several studies have recently evaluated the impact of colonoscopy preparation quality on the detection of suspected colonic neoplasia.5–7 In a retrospective analysis of 93 004 colonoscopies5 the detection of small polyps (<10 mm) was significantly associated with quality of the preparation. In a prospective, multinational observational study of 5832 patients, the detection of polyps of any size was found to depend on the quality of the bowel preparation,6 a conclusion that was mirrored in a third, smaller study evaluating the association between preparation quality and adenoma detection rate in 260 patients undergoing flexible sigmoidoscopy.7

Adequate bowel cleansing and its accurate description and documentation is also a factor considered in malpractice litigation alleging improper performance of colonoscopy when incident cancer is discovered after negative ‘total’ colonoscopy.8 The quality of bowel preparation also impacts on cost of colonoscopy and imperfect bowel preparation has been estimated to result in an increase in the cost of colonoscopy of 12–22%.9

In 1998, a meta-analysis suggested that NaP offered a more effective and more readily completed preparation than PEG-based regimes and advocated its use as a bowel preparation of choice in patients without comorbidities.10 The validity of these conclusions is limited by the small study sample size and methodological limitations of the component trials. Another review has recently addressed the same comparisons and drew similar conclusions.11 In this recent meta-analysis the search strategy was restricted to three electronic databases and only assessed studies published after 1990 and carries the danger of publication and selection bias.12

A joint consensus document on bowel preparation before colonoscopy, approved by the American Society of C & R Surgeons, the American Society of Gastrointestinal Endoscopy and the Society of American Gastrointestinal and Endoscopic Surgeons13 (GastroIntest Endosc 2006) advises that choice be made according to comorbidities, with PEG being favoured for elderly patients or those with renal insufficiency. It should be borne in mind that this was a consensus document rather than a systematic review. Although expert opinion is an important aspect of clinical guidance, the omission of a number of relevant clinical trials and the failure to quality appraise those studies means that this cannot be considered as evidence-based

In practice, choice of bowel preparation tends to be determined either by departmental policy or individual clinician's preference. In this review we have systematically identified high quality controlled trials which compared two or more bowel preparation regimens. We have acquired this information to establish relative performances and have used the outcomes to help clinicians make an informed and evidence-based choice of preparation in patients undergoing colonoscopy.

Methods

Literature review

Text Word searches were carried out on MEDLINE, EMBASE, CINNAHL and Cochrane CENTRAL databases up to January 2006. The search terms are indicated in Table 1.

Table 1.   Search strategy
Search 1
 Colonoscopy AND {Randomised OR Randomized} AND
 Polyethylene OR Macrogol OR PEG OR Klean-Prep OR Golytely OR Nulytely OR Isocolan OR Coloprep OR Colopeg OR Colyte
Search 2
 Colonoscopy AND {Randomised OR Randomized} AND
 Sodium phosphate OR Sodium phosfate OR Fleet OR Phospho-soda OR Visicol
Search 3
 Colonoscopy AND {Randomised OR Randomized} AND
 Senna OR X-prep OR Citrate OR Magnesium OR Citramag OR Picosulphate OR Picolax OR Mannitol

A further online search was then carried out using the Google Scholar search engine, utilizing key words: Colonoscopy, Preparation, Randomised. The resulting abstracts were scrutinized to ascertain whether they fulfilled the following inclusion criteria: (i) randomized controlled trial; (ii) comparing two or more orally administered bowel preparation regimes; (iii) patients, undergoing colonoscopy; (iv) assessment of quality of bowel preparation using a categorical measure; (v) assessment of patient tolerability included; and (vi) results published in a peer review journal.

Where abstracts indicated that a study was likely to fulfil these criteria, the full text was retrieved. Reference lists of these full texts were further scrutinized to identify studies that had not been identified in the primary searches. The journals most frequently publishing relevant studies were identified from the search results and these formed the basis of a supplemental hand search, in which the past 20 years’ editions of the following journals were scrutinized: (i) American Journal of Gastroenterology; (ii) Diseases of Colon and Rectum; (iii) Gastroenterology; (iv) Gastrointestinal Endoscopy; (v) Journal of Gastroenterology and Hepatology; (vi) Journal of Pediatric Gastroenterology and Nutrition; and (vii) Endoscopy.

In order to minimize the risk of biasing the analysis, all identified studies were quality appraised before inclusion. We used a simple scoring system which has previously been validated to maximize inter- and intra-observer consistency.14 Each study was assessed according to a five-point scale, with one point being awarded for each of the following criteria:

  • • Was the study randomized?
  • • Were details of randomization method given?
  • • Was the study double-blinded?
  • • Were details of blinding method given?
  • • Was information on study withdrawals given?

Any study scoring <3 points was excluded from further consideration.

Data extraction and analysis

Studies included in the analysis were grouped according to the treatment comparisons. Information was extracted describing the endoscopist's blinded assessment of the adequacy of preparation and the patient's assessment of the tolerability of treatment. At the outset, it was uncertain whether this information would be sufficiently homogenous to allow meaningful data pooling. To address this issue, a preliminary pooling plan was prepared based on a sample of studies examined in advance. Assessments of adequacy and/or tolerability to a single categorical variable were made and the results of individual studies aggregated using a DerSimonian Laird random effects model.15 Where such an analysis was not possible or meaningful, simple descriptive statistics, extracted from the original studies, were retained.

Results

The search strategy identified 112 studies, 82 of which were included in the final analysis.16–97 Twenty-eight studies were excluded because of inadequate quality while two used non-categorical methods to assess quality of bowel preparation, making comparison of study results impossible.98, 99 The bowel preparations included in the studies considered satisfactory for analysis are indicated in Table 2.

Table 2.   Included studies broken down by comparison type
Regimen*TrialsReference
  1. PEG, polyethylene glycol.

  2. * Some studies included more than one comparison.

PEG vs. sodium phosphate25[16–40]
PEG formulations and dose regimens33[19–22, 25, 33, 41–67]
Sodium phosphate formulations and dose regimens7[28, 68–73]
Miscellaneous regimens28[47, 54, 70, 74–97]

PEG vs. sodium phosphate – efficacy

There were 24 studies comparing plain PEG with oral NaP. In 22 of these, representing 3748 patients, the endoscopist's assessment was based on a four or five point qualitative scale. ‘Good’ and ‘excellent’ ratings were converted into a positive categorical measure and a pooled odds ratio was calculated to achieve this outcome for PEG vs. NaP (Figure 1). In two studies22, 35 the data were reported in manner that made conversion and pooling impossible. One further study compared NaP to PEG administered with bisacodyl40 and was therefore excluded from the meta-analysis.

Figure 1.

 Random effects pooling of efficacy assessments for polyethylene glycol vs. sodium phosphate (excellent/good vs. all other outcomes). * n/N indicates the number of patients with a good or excellent colonoscopy result, expressed as a proportion of the total number of patients randomised to that study group.

There is substantial heterogeneity between the studies (I2 = 81.7%; P < 0.00001). This indicates that, despite apparently similar trial protocols, there are significant differences either in the way in which bowel preparations were administered, or the way study results were recorded. There was no overall difference in efficacy between the two treatments (OR = 1.00; 95% CI = 0.67–1.50; P = 0.99).

PEG vs. sodium phosphate – tolerability

Statistical pooling of tolerability data was not possible because the data had not been gathered in a sufficiently consistent fashion. However, of the 25 studies that reported tolerability, NaP was reported to be superior in 14, there was no significant difference in 10 and in only one was PEG considered the better tolerated preparation. The commonly cited reasons for poor tolerability of PEG were the flavour and the requirement to consume a large volume (3–4 L PEG compared with 1.5–2 L for NaP). Assessment of adverse events failed to reveal a consistent pattern, although NaP tended to cause more dizziness and anal irritation, whilst PEG was more prone to causing nausea and bloating.

PEG vs. PEG

There are a number of distinct comparisons made in these studies (Table 3).

Table 3.   Comparisons of varying PEG formulations and co-treatments
  1. One study45 evaluated both dose timing and the addition of simethicone.

  2. PEG, polyethylene glycol.

Comparison of different polymer/electrolyte formulations19–21, 45, 46
Comparison of different dose-timing regimes42, 45, 46, 48, 50, 60, 62
Comparison of standard PEG with PEG + prokinetic agent22, 33, 43, 44, 61, 64, 67
Comparison of standard PEG with PEG + simethicone45, 52, 56, 58, 59, 65
Comparison of high-volume PEG with low-volume PEG47, 66
Comparison of high-volume PEG with low-volume PEG + prokinetic agent25, 41, 49, 55, 57, 63

The six direct PEG formulation comparisons evaluated versions of the lavage solution based on either PEG 3350 or PEG 4000, using a variety of different electrolyte combinations. Although one study suggested that a low-sulphate formulation was associated with improved efficacy,20 none of the other studies demonstrated any systematic between-preparations differences in either efficacy or tolerability.

The seven studies of dose timing examined a number of slightly different regimens. The general conclusion was that splitting the dose into two equal segments, separated by 12 h, resulted in improvements in both bowel cleansing and patient acceptability. The closer the final dose is taken to the time of colonoscopy, the more effective the final cleansing.

Concomitant use of metoclopramide, bisacodyl, cisapride, senna or magnesium citrate with PEG did not offer additional benefit, either in terms of efficacy of bowel cleansing, patient tolerability, or objective adverse events. In five out of six studies where the antifoaming agent simethicone was added, a significant reduction in visible bubbles was reported with consequent improved visibility.52, 56, 58, 59, 65 Two of the simethicone studies addressed patient tolerability. In one, the authors reported benefit with reduced general malaise and sleep disturbance,56 whilst in the other, patients reported disliking the taste of the simethicone combination.45

Two small studies, involving a total of 241 patients,47, 66 reported that low-volume PEG (1.5–2 L) yield similar bowel cleansing efficacy as that achieved with the standard regimen (3–4 L). One of these studies demonstrated improved tolerability.47 Most studies using this strategy, however, have evaluated the use of low-volume PEG in combination with a prokinetic agent (bisacodyl, senna or magnesium citrate). In two of these combination studies, the high-volume regimen had superior efficacy,25, 57 in three studies there was no significant difference,41, 49, 55 whilst in one study, the low-volume regime was found to be superior.63 In all five studies that investigated tolerability, the low-volume regime was preferred by patients.

Sodium phosphate vs. sodium phosphate

The seven NaP studies were designed as dose-finding comparisons, evaluating doses ranging from 45 to 180 mL of solution and 28 to 40 tablets. These studies demonstrated a clear dose–response both in terms of efficacy and tolerability, with the principle adverse events including nausea, vomiting and asymptomatic hyperphosphataemia. Dividing the doses reduced the incidence of nausea without sacrificing efficacy,69, 70 a conclusion supported by one of the high quality studies excluded from the analysis on technical grounds.99 Adding prophylactic antiemetics were shown to be relatively ineffective when used with high-dose NaP.71 Adverse events were reduced with the tablet formulation, although the large number of tablets required reduced patient acceptability and bowel cleaning efficacy was not as effective as the solution.68

Other comparisons

The remaining studies identified by the search criteria compared a wide range of differing bowel preparation regimes, frequently involving combinations of dietary manipulation, oral laxatives and enemas. As such, it would be complex to list the individual results for each of these studies. There are, however, four trials of oral sodium picosulphate with magnesium citrate (Picolax, Picoprep-3)77, 84, 94, 97 and these have been detailed because this formulation is commonly used in the UK.

Two studies compared NaP with sodium picosulphate/magnesium citrate preparations. In one study, no difference was observed in quality of bowel preparation but the picosulphate regimen was better tolerated.93 In the second study, the NaP yielded significantly better bowel preparation and there was no difference in tolerability.97 Two further studies compared sodium picosulphate/magnesium citrate with PEG-based regimes. In one report PEG yielded superior bowel preparation with no difference in tolerability,77 whilst in the second study there was no difference in efficacy but the picosulphate preparation was associated with superior tolerability.84

Safety

No clinically significant complications were reported in any of the randomized controlled trials identified by this review. A number of studies recorded electrolyte changes before and after treatment. Amongst the comparisons of PEG and NaP, eight studies measured a range of different biochemical variables.16, 18, 19, 23, 25, 26, 28, 39 All these studies recorded changes in sodium, potassium and inorganic phosphate. Sodium changes were minimal with PEG, ranging from a fall of 1.0 mmol/L to an increase of 1.4 mmol/L. NaP treatment was always associated with mild hypernatraemia, ranging in magnitude from 1 to 4.1 mmol/L. Potassium decreased in all cases: in PEG-treated patients this decrease ranged from 0.1 to 0.3 mmol/L whilst in NaP patients the effect was greater, ranging from 0.4 to 0.6 mmol/L. PEG had no significant effect on serum phosphate levels, whilst NaP caused consistent hyperphosphataemia, ranging in magnitude from 0.5 to 4.4 mmol/L.

Six of the studies assessed calcium levels,16, 18, 19, 23, 25, 26, 28 which dropped in both treatment groups in all studies. Reductions in the PEG group ranged from 0.03 to 0.1 mmol/L whilst in NaP-treated patients reductions were greater, ranging from 0.08 and 0.2 mmol/L.

Compliance

Compliance with bowel preparations, as evidenced by patients’ ability to consume the complete prescribed treatment, was recorded in 18 of 25 studies comparing PEG with NaP.16–20, 22–28, 31–33, 37–40 In all but one case38 more patients completed treatment with NaP than with PEG, although not all differences were statistically significant. Median completion rate for NaP was 97% (range 67–100%) vs. 89.5% for PEG (range 53–98%). Failure to complete, however, did not equate to total non-compliance. In one study,16 although 32.5% of PEG patients failed to complete the treatment, the mean volume of PEG consumed was 91% of the intended dose, suggesting the shortfall was relatively small. Similar results were found in a second study,18 with a 27% non-completion rate being observed alongside 93% mean volume consumption.

No study explored the reasons behind these differences in a systematic fashion, although there is an common and reasonable assumption that it principally reflects high volumes normally associated with PEG. This is supported by data from three studies comparing low-volume with high-volume PEG that reported treatment completion rates. There were clear compliance benefits associated with lower volumes: 66% vs. 93%,41 80% vs. 94%57 and 92% vs. 100%.66

The relationship between treatment compliance and the efficacy of bowel preparation was not described in any of the studies.

Discussion

Although our review has demonstrated that large numbers of published studies have investigated the optimum regimen for pre-colonoscopy bowel cleansing, the relatively small size of many trials, inconsistent outcomes assessment and design weakness make it difficult to draw clear conclusions. A large proportion of these studies have characterized the two main treatment options currently in use, that is: PEG and NaP. Although no overall difference was demonstrated, it is apparent from the forest plot (Figure 1) that there is considerable heterogeneity in the results of these comparative studies. Around half the studies involved fewer than 100 patients and exhibit markedly different results, both qualitatively and quantitatively. Some showed a clear advantage for PEG whilst others favoured NaP. No clear conclusion can be drawn from these studies, especially those that have small samples. The results of the larger, higher quality studies exhibit a greater degree of consistency, lending a higher degree of confidence in their conclusions. Based on the results of the best quality studies, it seems reasonable to conclude that these two forms of bowel preparation deliver a similar quality of bowel cleansing. No other regimen has demonstrated a superior efficacy profile. In particular, sodium picosulphate/magnesium citrate has been subjected to few randomized controlled trials and those that exist do not demonstrate any consistent benefit over other regimens.

Interpretation of the studies is limited by the inconsistent and poorly defined measures of efficacy outcome. In most studies, a four or five point scale has been used, based on the endoscopist's subjective assessment of the overall quality of the bowel preparation. It is impossible to assess the extent to which these assessments are compatible between studies. Indeed, the degree of heterogeneity seen in the PEG vs. NaP comparisons, even when the analysis is restricted to the higher quality studies, suggests that there is considerable between-observer variation. Refining this aspect of study design is important: as the link between preparation quality and adenoma detection rate has been clearly defined,5–7 some form of quantitative assessment is essential if evidence-based guidance is to be formulated.

Recently, a calibrated, externally validated outcome assessment tool (the Ottawa bowel preparation scale) has been developed to objectively quantify the quality of bowel preparation100 but this measure has yet to be used in a published randomized controlled trial. Until such tools achieve widespread use, achieving validated reproducibility between studies is virtually impossible and consequently, caution must be exercised in comparing studies.

A further consideration relates to the distinction between the quality of preparation achieved in different colon segments. There is good evidence that flat-type colorectal cancer develops preferentially in the ascending colon, particularly amongst older patients.101–103 Diagnosis and treatment at an early stage, however, depends on having a clear view of the mucosa of the ascending colon, in order that dye enhancement techniques can be effectively used.104 Although it is generally recognized that quality of preparation tends to be poorer in the right colon than the left, of the 82 studies in this review, only 22 attempted a segmental analysis of the quality of bowel preparation. At its most rigorous, this type of analysis equates the overall quality of preparation to the least well-prepared segment,20 thereby ensuring that a good result reflects that good visibility has been achieved throughout the bowel. In other circumstances, however, the use of a mean outcome figure may mean that inadequate proximal visibility is concealed by a better distal result.

In the absence of clear differences in efficacy, tolerability is likely to be an important discriminator when choosing treatments. In comparisons of NaP and PEG, there is evidence that PEG is less well tolerated, chiefly because of the volume of liquid that the patient is required to drink. There is some evidence that the volume can be reduced without impairing bowel cleansing efficacy, especially if treatment is taken over a relatively short time scale prior to performing colonoscopy. NaP is a lower volume preparation, although it still needs to be taken with 1.5–2 L of fluids. However, despite better patient acceptability, compared with PEG, this preparation is associated with increased incidence of nausea, vomiting and anal irritation.

Like assessment of efficacy, tolerability outcomes rely on the assessment measure used. The studies used a number of different tolerability assessments, none of which have been validated. Variability is demonstrated by the different outcome measures employed which included: direct assessment of treatment acceptability, clinician assessment of treatment acceptability, willingness of patient to use a treatment again, proportion of prescribed treatment actually taken, proportion of patients reporting adverse events and severity of adverse events

Tolerability is a patient-reported outcome (PRO). Although more difficult to standardize and quantify than an efficacy outcome, PROs must be both explicit and validated if the results are to have scientific credibility.105 This requires prespecification of key outcomes, together with proof of their importance to patients. None of these conditions have been adhered to in the studies chosen for this review and it is therefore impossible to assess which, if any, of these surrogate measures of tolerability has reproducible value. In these circumstances, pooling of results and drawing valid conclusions is not possible.

The most important clinical outcome associated with these bowel preparations relates to electrolyte disturbance. Although no clinically significant adverse events were reported in these studies, the potential for NaP to cause serious electrolyte disturbances is well documented. Hypernatraemia, hypokalaemia, hyperphosphataemia and hypocalcaemia have all been reported, with potential to cause dehydration, metabolic acidosis, renal failure, tetany and even death.106–117 Although other bowel preparations have the potential to disturb electrolyte balance118, 119 NaP appears to be particularly prone to causing imbalance. Whilst increasing age and impaired renal function may alert the clinician to the potential for problems,120 significant electrolyte abnormalities may occur in apparently healthy individuals.115

A recent study121 of 36 patients aged over 65 undergoing bowel preparation with NaP reported that 58% developed significant hypocalcaemia and 56%, hypokalaemia. The frequency and magnitude of this effect correlated with increasing age and creatinine clearance. In another study122 of patients without clinical contraindications to NaP, 39% developed hyperphosphataemia and 5% developed hypocalcaemia.

As a result of these adverse event reports, drug surveillance authorities have issued warnings to prescribers worldwide to be cautious when prescribing NaP solution in higher risk patients,109, 123–126 with particular emphasis being placed on monitoring electrolyte levels both before and after administration.

Conclusions

This review set out to assess the performance of existing bowel preparation regimens and to what extent poor preparation might reflect a difference in efficacy and/or tolerability. Despite the large number of published clinical trials, small patient numbers and methodological limitations severely limit the value of most of these studies as a guide to clinical practice.

The combination of sodium picosulphate and magnesium citrate is the most commonly used preparation in the UK. However, this preparation has been subject to less clinical investigation than either PEG or NaP. There is currently no evidence base to justify its choice as the cleansing preparation of choice. PEG and NaP appear to be equally effective, but both agents have their limitations. Electrolyte abnormalities are common in patients treated with NaP, and may lead to serious clinical consequences especially, but not exclusively, in the elderly and those who are pregnant or who show impaired renal function. Electrolyte disturbances are much less common with PEG, but the high volume of fluid intake required with current preparations makes it unpopular with patients and may lead to nausea and bloating.

The optimum combination of efficacy, tolerability and safety has yet to be defined for bowel preparation for colonoscopy. There is clearly a need for new preparations and an imperative to develop new and validated methods of assessing efficacy and tolerability using rigorous study designs that will allow unequivocal conclusions to be drawn from adequately powered controlled trials. Sensitive and systematic outcome measures need to be developed that clarify the extent of bowel preparation, with particular reference to segmental visualization. The use of explicit measures that allow the distinction between tolerability, patient acceptability and adverse events are essential if Patient Reported Outcomes are to play a reproducible part in the decision-making process.105 Achievement of this goal will necessarily involve an element of patient education and perception of bowel preparation quality for colonoscopy because a unique study has demonstrated that patients are unreliable judges of the quality of their own preparation, tending to overestimate the cleanliness of their colon.127 Other potential variables that warrant consideration include the timing of colonoscopy (afternoon investigations having been shown to have higher failure rates compared with morning colonoscopies) and the impact of walking on bowel clearance, with the consequent advantage for those undergoing preparation as out-patients.128–130

Acknowledgement

Authors' declaration of personal interests: J. Belsey has been paid by Norgine Ltd as a data analyst and medical writer; O. Epstein has been paid by Norgine Ltd to attend two advisory board meetings.

Declaration of funding interests: this systematic review was funded by Norgine Ltd and was carried out and written independently by the three cited authors.

Ancillary