Defining and Reporting Diarrhea in Tube-Fed Patients—What a Mess! D. Z. BLISS,P. A. GUENTER,R. G. SETTLE American Journal of Clinical Nutrition 55:753-759,1992

The purpose of this study is to document the effect that various definitions of diarrhea have on its reported incidence in tube-fed patients. Following a group of 29 patients who were tube-fed for at least 5 days (median duration of feeding, 13 days), the investigators prospectively monitored the occurrence of diarrhea according to each of eight definitions: ≥1, ≥2, ≥3, ≥4 liquid stools/ day; ≥3 stools/day; ≥4 stools/day; ≥4 stools/day for 2 days; and ≥5 stools/day. None of the definitions incorporated measurements of either stool weights or volumes. Because treatment with antibiotics and the presence of hypoalbuminemia have been associated with diarrhea in some reports, the investigators took those into account. Sixty-six percent of the patients (19 patients) were treated with antibiotics and only 7% (2 patients) had serum albumin levels <26 g/L; the mean serum albumin level of the remaining 27 patients was 34 g/L. Only 10% (3 patients) were in critical care units. Because duration of monitoring would be expected to correlate positively with incidence, and is often not reported in studies of tube-feeding-associated diarrhea, the authors reported both the incidence of diarrhea and the percent of days with diarrhea. It is also noteworthy that 31% (9 patients) of the patients received kaolin-pectin for treatment of diarrhea, which probably affected the percent of days with diarrhea.

The more lenient definitions of diarrhea resulted in higher incidence rates and higher percentages of days with diarrhea. The most lenient definition (≥1 liquid stool/day) was associated with an incidence rate of 72%, and 26% of days with diarrhea, whereas the most rigorous definitions (≥4 liquid stools/day and ≥5 stools/day) resulted in 21% incidence and 1.7% of days with diarrhea. The number of days patients were monitored correlated positively with the incidence of diarrhea, but very weakly with the percent of days with diarrhea. Thus, the use of percent of days with diarrhea as the method of reporting reduced the effect of varying durations of monitoring. Antibiotic treatment correlated positively with percent of days with diarrhea, but not with incidence. Serum albumin levels did not correlate with diarrhea, but this may have been caused by the largely normal albumin levels in this population.