Introduction

Lower gastrointestinal motility disturbances frequently occur in intensive care unit (ICU) patients [1]. Constipation, in some cases progressing to a paralytic ileus, has been shown to occur secondary to medication (opiates [2], vasoactives [3, 4, 5, 6]), neurogenic imbalance [7], hypoperfusion and shock [8], endotoxin [9, 10], nitric oxide overproduction [11, 12], or a combination. Diarrhea, on the other hand, might be multifactorial and related to inflammatory mediators, splanchnic ischemia, enteral feeding and to microbiological imbalance or overgrowth [1]. Microbiological imbalance is induced by stasis and by systemic antibiotics, which reach variable but relevant concentrations in the lumen of the digestive tract [13, 14]. Both result in a reduction of the endogenous anaerobic and Gram-positive flora with overgrowth of Gram-negative bacteria and increased production of endotoxin. Selective decontamination of the digestive tract (SDD) reduces the need for systemic antibiotics, and aids in maintaining the normal flora [15]. SDD reduces the endoluminal endotoxin level, both by minimizing the number of endotoxin-producing bacteria as well as by binding endotoxin [16].

These arguments made us hypothesize that severity of illness, medication and SDD in intensive care patients alter the pattern of defecation, and that constipation is unfavorable [17]. The aim of this study was to describe the defecation pattern of critically ill, mechanically ventilated patients and to investigate the relation between defecation and severity of illness, medication and SDD.

Methods

This study was a separate sub-study on patients who participated in a trial investigating selective decontamination of the digestive tract (SDD) [18]. Fifty consecutive mechanically ventilated patients with a length of stay (LOS) in the ICU of at least 7 days who were enrolled in the SDD trial were included. Written informed consent was obtained from patients or their legal representatives. Demographic data, acute physiology and chronic health evaluation (APACHE II [19]), sepsis-related organ failure assessment (SOFA [20]) scores and LOS were prospectively collected; the other data were retrieved from the patient charts. Per day, the highest dosage of dopamine and noradrenaline administered for at least 1 h was recorded according to the SOFA definition [20]. Administration of at least 10 mg of morphine or an equipotent dose of opiates was regarded as relevant. Defecation, enteral feeding and medication were routinely recorded on the patient charts during the entire ICU stay. To prevent incomplete data collection, the separately recorded nurse's notes were analyzed as well. Enteral feeding with Nutrison Standard (Nutricia, Zoetermeer, The Netherlands) or in trauma patients with Impact (Novartis, Breda, The Netherlands) was routinely begun within 48 h, aiming at 2,000 ml, or at 1,500 ml within 24h, respectively. Gastric retention was treated with cisapride 40 mg t.i.d., which was still available during the study period. In case of persistent gastric retention, a duodenal feeding tube was inserted. Lactulose 30 ml t.i.d. was administered if no stools were passed after 3–4 days. Neither constipation nor diarrhea led to adjustments in the administration of feeding. If no defecation occurred after lactulose administration, phosphate enemas were administered at the discretion of the physician. Defecation was subjectively scored as “none” (< 100 ml), “normal”, or “diarrhea” (large and watery stools > two times per day). Passage of normal stools or diarrhea before day 6 after admission to the ICU was regarded as early, thereafter as late.

Statistical analysis

Data are shown as mean and standard deviation or median and interquartile range, where appropriate. Differences between groups were tested for significance by Fisher exact test. This non-parametric test was chosen because of the relatively small sample size and skewed distribution of most variables. A two-tailed alpha of 0.05 was considered as statistically significant. Mean dopamine and mean noradrenaline dosages showed a skewed distribution and were transformed by logarithmic before entering linear and logistic analyses.

Results

Five patients were excluded from analysis because of recent gastrointestinal surgery (< 14 days before admission to the ICU), and one patient with a primary indication for parenteral feeding due to a pre-existing short bowel syndrome. The 44 remaining patients showed comparable baseline characteristics (Table 1). These patients spent 767 days in the ICU, 371 in the SDD group and 396 in the controls. The median time until the first passage of stools was 6 days, mean 6.2 days (SD ± 2.48) (Fig. 1); in the SDD group mean 6.3 days, in the control group 6.0 days. The four patients who died, all from the control group, did not differ from survivors in the time to pass stools (mean 6.4 days). Lactulose to promote defecation was administered in 39 patients (SDD 18, control 21), after a mean of 3.8 days. Enemas were administered in 21 patients after mean 4.7 days after admission. After enemas and with continued administration of lactulose, stools were passed after mean 2 days, i.e., 6.8 days after admission. Table 2 summarizes the pattern of defecation. Diarrhea occurred at least once in 27 patients (SDD 12, control 15), of whom 19 received lactulose prior to diarrhea. There were no clinical signs or symptoms of C. difficile infection, but specific cultures or endotoxin assays were not routinely performed.

Fig. 1
figure 1

Kaplan-Meier curve for day of first defecation

Full size image
Table 1 Patient characteristics. No significant differences exist between control and SDD patients. (APACHE acute physiology and chronic health evaluation, IQR interquartile range, LOS length of stay, PM predicted mortality, SDD selective decontamination of the digestive tract)
Full size table
Table 2 Type of defecation per day (95% CI diff 95% confidence interval for the difference in %, SDD selective decontamination of the digestive tract)
Full size table

Twenty patients passed stools before day 6 (early defecation group); half of them were treated with SDD. The mean APACHE II scores did not differ significantly between the early defecation group and the late (19.2 vs. 18.0), but, until the day of defecation, the mean dosage of vasoactive medication per day was lower in the early defecation group (Table 3, Fig. 2a and b). Comparing the first 6 days, the late group received more dopamine, noradrenaline, and morphine and had higher SOFA scores (Table 3); the SOFA scores were still higher if the circulatory component, reflecting dopamine and noradrenaline administration, was left out. Mean dopamine, mean noradrenaline and mean SOFA score were forwardly entered in a stepwise logistic regression analysis with late vs. early defecation as the dependent variable. Mean dopamine dosage was the only factor significantly associated with late defecation in a univariate analysis, but this association was lost in the multiple regression analysis. Up to day 4, equal numbers of patients in the early and late groups received morphine, but thereafter more patients in the late defecation group, the difference being significant on days 5, 6 and 7 (Fig. 2c). The decline in SOFA scores seemed more rapid in the first days in the early defecation group (Fig. 2d). Patients in the early defecation group received more cisapride and lactulose, but there were no differences between the groups regarding the administration of enemas or the amount of enteral feeding (Table 4). No complications were seen due to the administration of cisapride or lactulose. The early defecation group had a shorter duration of artificial ventilation and a shorter LOS (Table 5).

Fig. 2
figure 2

Mean daily maximum dopamine (a) and noradrenaline (b) in μg/kg/min, numbers of patients receiving morphine (shaded area) (c) and SOFA scores for early (< 6 days) vs. late defecation groups (d). Bars denote 95% confidence interval. Days after ICU admission on the x-axis (SOFA sepsis-related organ failure assessment score)

Full size image
Table 3 Mean (± SD) daily doses of dopamine and noradrenaline, % of days on ventilator and % of days receiving at least 10 mg of morphine, SOFA scores and SOFA scores without the circulatory component during the first 5 days for early (< 6 days) and late (≥ 6 days) defecation. 95% CI-diff 95% confidence interval for the difference, SOFA sepsis-related organ failure assessment score, SOFA-circ SOFA score without the circulatory component
Full size table
Table 4 Administration of cisapride, lactulose, enemas and enteral feeding during the first 5 days for early (< 6 days) and late (≥ 6 days) defecation. Percentage (± SD) of days on which the specific drug is given; enteral feeding is expressed as mean total amount of enteral feeding administered per 24 h minus gastric residuals (± SD). 95% CI diff 95% confidence interval for the difference
Full size table
Table 5 Comparison of length of stay and duration of mechanical ventilation for early (< 6 days) and late (≥ 6 days) defecation
Full size table

Discussion

While the pattern of defecation in healthy people has a great variation from passing three stools per day up to three stools per week or less [21], the pattern of defecation in the critically ill is largely unknown. Only three studies address constipation in the critically ill, reporting an incidence of 16–83% [1, 22, 23]. This study is the first to describe the pattern of defecation in critically ill ventilated patients with a prolonged stay in the ICU in relation with severity of organ failure and medication.

We have shown that, despite active measures to promote defecation in almost all patients, in the majority of days no stools were passed, and only in 21% of the days a normal formed stool was passed. Diarrhea was seen frequently, but it occurred in the majority of the patients after administration of lactulose and might therefore be considered iatrogenic. Diarrhea without prior lactulose administration occurred in 22% of the patients, and this frequency is in concordance with the literature [22].

The patients who passed stools early, i.e., before day 6 after admission, had less organ failure as reflected in lower SOFA scores and needed less dopamine and noradrenaline. The question is whether late defecation is a symptom of organ failure or the effect of treatment with dopamine and/or noradrenaline. To gain more insight in this issue, we omitted the circulatory component from the SOFA scores and found that the scores were still higher, i.e., the patients were more ill. This suggests that severity of illness is at least partly responsible for a delayed defecation. Administration of dopamine and noradrenaline as well as severity of illness are implicated in impaired motility of the colon [3, 4, 5, 6, 24]. From this study it cannot be distinguished whether the delayed defecation is a consequence of the administration of vasoactive medication itself, or of the severity of illness for which the medication was necessary. The patients in the early defecation group more often received cisapride and lactulose. Ambivalent effects of cisapride on colonic motility have been reported, but no data exist on its effect on defecation in critically ill patients; lactulose has not been investigated in ICU patients. It may be expected, however, that either one of these drugs or both have a prokinetic effect on the colon. Starting early after ICU admission with these drugs may have resulted in early defecation. Morphine probably did not influence defecation in the early days, because in the first 4 days there was no significant difference in the number of patients receiving morphine between the early and the late defecation groups. Thereafter, however, morphine was administered in more patients in the late defecation group. Whether continuing morphine administration delayed defecation is unclear; prolonged administration of morphine might reflect a more critically ill condition, which in itself or via the higher doses of dopamine and noradrenaline caused constipation.

It might be speculated that passing stools early and on a regular basis is beneficial, especially when SDD is used. Stasis of feces in the digestive tract promotes bacterial overgrowth and translocation of bacteria and/or endotoxin, thereby “fueling the motor of organ failure” [13, 14, 25]. Stasis also hampers the efficacy of SDD: as long as the enterally administered non-resorbable antibiotics have not reached the anus, elimination of Gram-negative rods, S. aureus and yeasts will not be accomplished, with subsequent risk for secondary infections. In our patients we could demonstrate a slightly more rapid decline in the SOFA scores when patients had early defecation.

Irrespective of the cause of constipation, the most striking finding is the fact that later defecation was associated with a longer duration of mechanical ventilation and ICU stay. Mostafa also found a correlation between constipation and these outcome variables [23].

Conclusion

In this cohort study of critically ill ventilated patients, the mean time until the first defecation was 6 days, and on more than half of the days in the ICU no defecation occurred. Diarrhea seemed al least partially iatrogenic due to administration of lactulose. SDD resulted in a higher proportion of days with normal defecation and less diarrhea. Higher SOFA scores and higher mean daily doses of dopamine and noradrenaline were associated with a prolonged time to first defecation. Morphine administration in the first 4 days did not influence the time until the first defecation, but prolonged administration may have contributed to delayed defecation. Delayed defecation was associated with longer duration of mechanical ventilation and length of stay in the ICU, but from our data no cause and consequence can be deducted. Administration of cisapride and/or lactulose correlated with earlier defecation, but it cannot be concluded from our data that administration of these drugs is beneficial and results in a shorter length of ICU stay.