AP&T correspondence columns are restricted to letters discussing papers that have been published in the journal. A letter must have a maximum of 300 words, may contain one table or figure, and should have no more than 10 references. It should be submitted electronically to the Editors via http://mc.manuscriptcentral.com/apt.
Letters to the Editors: Bowel preparation: which meta-analysis is right? Like the cleansing methods, they are all still imperfect
Version of Record online: 3 SEP 2010
© 2010 Blackwell Publishing Ltd
Alimentary Pharmacology & Therapeutics
Volume 32, Issue 7, pages 934–936, October 2010
How to Cite
Heresbach, D. and Boustière, C. (2010), Letters to the Editors: Bowel preparation: which meta-analysis is right? Like the cleansing methods, they are all still imperfect. Alimentary Pharmacology & Therapeutics, 32: 934–936. doi: 10.1111/j.1365-2036.2010.04421.x
- Issue online: 3 SEP 2010
- Version of Record online: 3 SEP 2010
Sirs, We read with interest the recent article of Juluri et al.,1 which was a meta-analysis of randomized controlled trials (RCTs) comparing 4 L polyethylene glycol (4-L PEG) with 2 × 45 mL doses of sodium phosphate (NaP) as bowel preparation for colonoscopy. They concluded, on the basis of 18 head-to-head RCTs of NaP vs. 4-L PEG, that NaP was more likely to result in excellent or good quality preparation. They noted that two previous meta-analyses2, 3 comparing the two preparations, by ourselves2 and Tan et al.,3 included RCTs with off-label or atypical doses.
However, in contrast to the meta-analyses of Juluri and Tan and their colleagues, we concluded that neither PEG nor NaP emerged as consistently superior in terms of efficacious colon cleansing.
Therefore, we have reviewed our own analysis and also the data extraction and the RCTs included in this recent meta-analysis. In comparison with the meta-analysis of Juluri et al.,1 we included effectively 24 RCTs including those with off-label or atypical dosage, whereas the 18 RCTs that they included used only FDA-approved dosages.
Analysis in depth shows that they have included six trials that were not in our meta-analysis, but only four of them were included in their bowel cleansing quality analysis, the others being included in the acceptability and tolerance (completion) analysis. Among these four RCTs, two concluded that NaP and 4-L PEG preparations were equivalent for cleansing quality,4, 5 and two6, 7 demonstrated that NaP was associated with a higher frequency of excellent or good preparation quality. We did not include the study of Arezzo,6 which in fact was a three-arm trial in which a senna compound was also tested and which compared a posteriori head-to-head PEG with NaP.
We agree with Juluri et al. in that we misinterpreted the data from the study of Marshall et al.5 in concluding a superiority of PEG over NaP preparation, whereas the analysis of Juluri et al.1 considers that the two cleansing methods were equivalent. On the other hand, their data extraction from the trial by Henderson et al.4 shows NaP as being equivalent to PEG, but in Table 3 of the original publication, this was only true for the ‘remainder colon overall’, whereas PEG preparation was significantly better for the right colon. Moreover, from the RCT of Thomson et al.,7 Juluri et al.1 conclude that NaP gave a better preparation quality than 4-L PEG, but in the original publication the colonoscopist assessment found no difference in quality in general (Figure 3 of the article), which was not significantly different overall (P = 0.8), for amount of residual fluid (P = 0.06) and for residual stool (P = 0.9).
So, what is the final position? First, we need to be careful when data are extracted for meta-analysis, especially when the data have not been expressed in exactly the same way. Secondly, we cannot conclude at present that a particular bowel preparation gives a better quality result, despite recent meta-analyses as these are both imperfect. Thirdly, besides the quality of bowel preparation, we also need to consider acceptability issues. Acceptability is very difficult to measure and preparation quality is affected if only half of the product is taken. Finally, given the older patient population, we need to pay attention to adverse effects, especially regarding impairment of renal function and we need more information about this comorbidity and about medication that might have a toxic synergy with NaP.
To conclude, do we need to carry out a fourth meta-analysis to determine which meta-analysis and conclusion are correct? While acknowledging the rigour of the study of Juluri et al. in including only the trials that followed the approved dosages, we all know that in actual everyday practice, different kinds of preparation protocols are used (such as split doses, low residue diets 2–3 days before the procedure, or exclusively liquid diets 1 day before colonoscopy), and that currently other PEG formulations are available with reduced quantities (2 L) that give similar results to 4-L PEG or the two-dose NaP regimen, and finally that NaP preparation is also available as 32 tablets.
Colonoscopy screenees demand a ‘light’ preparation, but this cannot be allowed to impair the quality of the examination, increasing the polyp miss rates. However, it is pointless to provide a full-dose bowel preparation that will not all be used because the screenee will not accept it. Ultimately, the best comparison today would not be another meta-analysis comparing small volume PEG (2L) associated to ascorbic acid versus NaP tablets but a comparison of small volume PEG to picosulfate–citrate magnesium (PCSM) formulation. This comparison was done in one RCT.8
Declaration of personal interests: D Heresbach was a medical expert for Bayer Santé familial in 2009. C Boustière belonged to the French scientific board for the annual AstraZeneca France symposium in 2008 and 2009. D Heresbach and C Boustière belong to the French Norgine Board for the year 2010. Declaration of funding interests: None.