Objective
The use of mechanical bowel preparation prior to laparoscopy is common in gynecology, but its use may affect the rates of perioperative events and complications. Our objective was to compare different mechanical bowel preparations using decision analysis techniques to determine the optimal preparation prior to laparoscopic gynecological surgery.
Study Design
A decision analysis was constructed modeling perioperative outcomes with the following mechanical bowel preparations: magnesium citrate, sodium phosphate, polyethylene glycol, enema, and no bowel preparation. Comparisons were made using published utility values. Secondary analyses included the percentages that had 1 or more preoperative events and 1 or more intra- or postoperative complications.
Results
Overall, the highest utility values were for no bowel preparation (0.98) and magnesium citrate (0.97), whereas the other values were as follows: enema (0.95), sodium phosphate (0.94), and polyethylene glycol (0.91). The difference between no bowel preparation and magnesium citrate was less than the published minimally important differences for utilities, so there is likely no real difference between these strategies. The probability of having at least 1 preoperative event was lowest for no bowel preparation (1%), whereas the probability of having at least 1 intra- or postoperative complication was lowest with magnesium citrate (8%).
Conclusion
The highest utilities were seen with no bowel preparation, but the absolute difference between no bowel preparation and magnesium citrate was less than the minimally important difference. With similar overall utilities, our model raises questions as to whether mechanical bowel preparation is a necessary step prior to laparoscopic gynecological surgery. However, if a surgeon prefers a bowel preparation, magnesium citrate is the preferred option.
The use of mechanical bowel preparation prior to laparoscopy is common in gynecology as well as other surgical specialties, yet studies across specialties have questioned its merit. Historically, gynecologists have prescribed preoperative mechanical bowel preparations in attempts to decrease the risk of infection while also providing easier bowel manipulation and better visualization. However, many of these proposed benefits have never been proven, and bowel preparations may actually increase the risk of a surgical site infection.
Despite such data, a survey of gynecological oncologists found that approximately half of the respondents still prescribe bowel preparations despite 77% acknowledging that there are no data to support such use.
Two recent randomized trials have shown that for both vaginal prolapse repairs and laparoscopic hysterectomies, mechanical bowel preparation with saline enemas conferred no benefit for surgeon visualization and ultimately decreased patient satisfaction.
This is similar to studies of both simple and more complex gynecological laparoscopy, which also found no difference in the surgical field or operative difficulty if either an oral sodium phosphate or saline enema was used preoperatively. Unfortunately, these studies used various mechanical bowel preparations and various routes of surgery, thus making comparisons difficult.
Our objective was to compare multiple different mechanical bowel preparations, including no mechanical bowel preparation in a decision analysis model, to determine the optimal bowel preparation prior to laparoscopic gynecological surgery.
Materials and Methods
After institutional review board approval was obtained, a decision analysis model was created using TreeAge Pro (TreeAge Software, Inc, Williamstown, MA). The decision node included the following possible mechanical bowel preparations: magnesium citrate, sodium phosphate, polyethylene glycol, enema, or no mechanical bowel preparation. Each subsequent subtree was identical for each of the 5 bowel preparation options. The differences in the subtrees were in the probability of each perioperative event or complication occurring with the different bowel preparations.
The first branch point in each subtree was a specific preoperative, intraoperative, or postoperative complication/event dichotomized to present or absent. Subsequent branches were for other possible complications. For example, with 2 complications there were 4 terminal branches in the tree, representing 4 possible outcomes. The result could be both complications, neither, only the first, or only the second. This simplified version of the tree is displayed in the Figure .
Using this methodology, our tree was exponentially expanded, accounting for all 9 perioperative events. The model was based on a 7 day follow-up because most surgical complications related to bowel preparation are noted in this time frame. This shorter time frame improved our ability to distinguish differences between the preparations.
The model included weighted average probabilities for perioperative events and complications from 28 published trials. When laparoscopic gynecological trials were not available, colorectal, urologic, and general surgery literature was used. Eight trials taken from the colonoscopy literature evaluated the preoperative patient outcomes specifically related to the bowel preparations because of the greater robustness of the preoperative outcomes that would occur, regardless of whether a colonoscopy or gynecological surgery follows the bowel preparation.
Our model included 5 preoperative events and complications including nausea/vomiting/abdominal pain, diarrhea/fecal incontinence, anal discomfort, hypotension, and arrhythmia/seizure. We also modeled 4 intra- or postoperative complications including surgical site infection, abdominal infection, ileus/obstruction, and bowel injury/colostomy. The base case probability was a composite of all available data.
Utility values were assigned to each perioperative event and were obtained from published data. Utilities that represent a measure of quality of life ranged from 0, representing death, to 1, representing perfect health.
The primary objective was to determine which mechanical bowel preparation yielded the highest utility value and thus was the preferred treatment option. Secondary analyses included the percentage of patients who had at least 1 preoperative event or complication and the percentage with at least 1 intra- or postoperative complication.
Multiple 1-way sensitivity analyses were performed to test model robustness and to determine whether a threshold value existed in which the results would then favor a different bowel preparation. These analyses were carried out over the entire range of possible probability values for events and complications as well as utilities, 0 to 1.
Results
Complication rates and utility values were obtained from the published literature ( Table ). Using simple roll-back methodology, the overall utility value for each mechanical bowel preparation was calculated. This method calculates the average utility value a patient will experience, given the weighted average of each perioperative event and the resultant decrease in quality of life utility if each of these events occurs.
| Variable | Outcome | Outcome probability (range in published studies) | Utility value | ||||
|---|---|---|---|---|---|---|---|
| No mechanical bowel preparation | Magnesium citrate | Sodium phosphate | Polyethylene glycol, 238 g | Enema | |||
| Preoperative event | Nausea/vomiting/ abdominal pain | 0.0000 a | 0.1307 (0.0214–0.1688) | 0.3055 (0.0313–0.4771) | 0.3067 (0.0700–0.6226) | 0.3265 b | 0.767 (0.68–0.86) |
| Diarrhea/ fecal incontinence | 0.0052 b | 0.2857 c | 0.2857 b | 0.0749 b , | 0.3673 b | 0.769 (0.61–0.95) | |
| Anal discomfort | 0.0000 a | 0.0214 b | 0.0438 b | 0.0469 d | 0.0499 d | 0.700 b | |
| hypotension | 0.0000 a | 0.0143 b | 0.0438 b | 0.0143 c | 0.0000 a | 0.593 b | |
| Arrhythmia/ seizure | 0.0000 (0–0.0078) | 0.0000 b | 0.0063 b | 0.0145 b | 0.0000 a | 0.835 (0.72–0.95) | |
| Intraoperative/ postoperative complications | Surgical site infection | 0.0832 (0.0145–0.1719) | 0.0236 b | 0.0728 (0–0.12) | 0.2012 (0.0667–0.3488) | 0.0411 b | 0.900 b |
| Abdominal infection | 0.0622 (0.0309–0.32) | 0.0279 e | 0.1059 (0.0139–0.1463) | 0.2634 (0.1812–0.3714) | 0.0685 b | 0.696 (0.642–0.75) | |
| Ileus/obstruction | 0.0196 (0.0118–0.0667) | 0.0285 c | 0.0636 (0.0139–0.0854) | 0.0374 b | 0.0137 b | 0.650 (0.6–0.7) | |
| Bowel Injury/colostomy | 0.0021 (0–0.1467) | 0.0011 c | 0.0000 b | 0.00466 (0.0396–0.0667) | 0.0000 b | 0.550 b | |
a Values for no mechanical bowel preparation and enema were assumed to be 0 for certain preoperative outcomes, given the lack of mechanism to cause this event with no or minimal intervention
b Single value reported in the literature so a range is not reported
c This value was not available in the literature, so the baseline value was assumed to be median of other bowel preparation values. This was accounted for in the sensitivity analysis by varying over the entire range of 0–1
d This value was not available in the literature, so baseline value was calculated using the ratio of nausea/vomiting/abdominal pain to anal discomfort from the magnesium citrate and sodium phosphate preparations. The known value for nausea/vomiting/abdominal pain was then divided by this multiplier of 6.5412. This assumption was accounted for in the sensitivity analysis by varying over the entire range of 0 to 1
e This value was not available in the literature, so the baseline value was calculated using the ratio of surgical site infection to abdominal infection from the other 4 preparations. The known value for surgical site infection was then divided by this multiplier of 0.8472. This assumption was accounted for in the sensitivity analysis by varying over the entire range of 0 to 1.
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
