Performance Assessment of the Fecal Leukocyte Test for Inpatients

Abstract

Traditionally, fecal leukocyte testing detects large bowel inflammation or disruption, conditions that allow leukocytes into the stool. However, test usefulness with inpatients is unclear. Two hundred five inpatients who had undergone one to three tests were identified, and their FLT results were compared to their gastrointestinal disease diagnoses at time of discharge. A specificity of 92% for detecting intact colonic mucosae in inpatients was found.

In 1972, Harris et al. first suggested that fecal leukocytes result from a breach of the colonic mucosa (3). It was theorized that stool from patients with acute intestinal inflammation with overlying eroded, ulcerated, or otherwise disrupted mucosal epithelium allows neutrophils into the colonic lumen, resulting in the abnormal finding of fecal leukocytes. Since this report, fecal leukocyte testing (FLT) has been used to evaluate large bowel mucosa inflammation and disruption. Despite the limited evidence supporting FLT efficacy, the test is considered valuable by many clinicians. Today FLT is often used to screen for infectious diarrhea (2, 4, 9, 12). However, some investigators have found that FLT has limited usefulness (5, 10). A recent study stated that FLT is a poor predictor of Clostridium difficile infection in inpatients and outpatients and demonstrated a negative predictive value (NPV) in ruling out bacterial diarrhea in outpatients (11); for that reason, discontinuing inpatient FLT at our institution was considered. Therefore, we evaluated FLT with an inpatient cohort, defining the test's sensitivity, specificity, and positive predictive value (PPV), and we better characterized the conditions associated with positive FLTs, providing useful information for laboratories and clinicians.

A retrospective review of the microbiology laboratory database at The Methodist Hospital, Houston, Tex., identified 205 adult inpatients who had undergone one to three FLTs from October 2000 through April 2001. Discharge codes were retrieved. Patients were divided into two groups based on their gastrointestinal diagnoses: group 1 consisted of patients who were likely to have a breach in the colonic mucosa (any infectious or inflammatory condition, blood in the stool, or acute vascular insufficiency), and group 2 consisted of patients who were unlikely to have a breach in the colonic mucosa (no lower gastrointestinal discharge diagnosis, abdominal pain, colostomy, constipation, diarrhea, diverticulosis, hemorrhoids, history of colonic malignancy, impaction, incontinence, irritable bowel syndrome, neoplasm, obstruction, paralytic ileus, polyps, or rectal prolapse). Patients with multiple gastrointestinal diagnoses were classified within group 1 if any diagnosis involved a probable colonic mucosal breach.

Samples were examined within 1 h of receipt by the laboratory. We used Spot Test methylene blue stain (Loeffler, Difco, Detroit, Mich.) per the standard protocol, without additional controls. Only neutrophils with unequivocal segmented nuclei were counted. Test results were graded on the basis of the number of neutrophils per high-power field (HPF) as follows: “many” indicated >10 neutrophils, “moderate” indicated 5 to 10 neutrophils, “occasional” indicated 2 to 4 neutrophils, “few” indicated 1 neutrophil, “rare” indicated <1 neutrophil, and “none” indicated 0 neutrophils/HPF. Multiple specimens from the same patient were collected 24 h apart for up to 7 days; the result with the most fecal leukocytes was used for calculations.

Inpatients (n = 205) were divided into group 1 (n = 72) and group 2 (n = 133) (Table 1). Using a cutoff of <1 neutrophil/HPF (Fig. 1A) rather than >1 neutrophil/HPF (Fig. 1B) affected FLT performance. The levels of agreement between the FLT results for multiple specimens from the same patient (n = 31) were 75% (24 patients with two tests) and 57% (7 patients with three tests), FLT agreement being defined as a change in result from the original interpretation of less than two grading categories. Repeat testing demonstrated that only five patients (16%) were correctly reclassified by eventual positive FLT. Of the patients known to have infectious gastroenteritis (n = 25), 32% had a positive FLT result, including those with C. difficile infection (n = 20; 5 were positive [25%]), viral enteritis (n = 3; 2 were positive), unspecified bacterial enteritis (n = 1; positive), and intestinal tuberculosis (n = 1; negative). Interestingly, the patient with both Giardia lamblia and C. difficile had no detectable fecal leukocytes. There was no statistical difference in the FLT results for patients with infectious or noninfectious gastroenteritis. Of patients with blood in the stool, 54% (7 of 13) had a negative FLT result. Unexpected negative FLT results were also found with patients with endoscopy-proven colonic ulceration (n = 1), appendicitis (n = 1), and anal abscess or fissure (n = 1). No patients with gastrointestinal tumors (n = 8) exhibited fecal leukocytes. Positive FLT results were found for three of four patients with acute intestinal vascular insufficiency and only one of three patients with ulcerative colitis; this strong association between acute vascular insufficiency and the presence of fecal leukocytes was noted but not further characterized. Conversely, 6 of 74 patients (8%) had positive FLT results without gastrointestinal pathology.

TABLE 1.

Results of FLT^a

Breach of colonic mucosa	Lower gastrointestinal discharge diagnosis	Total no. of patients	No. of patients with indicated grade						No. of procedures
			None	Rare	Few	Occasional	Moderate	Many
Yes	Acute vascular insufficiency of intestine	4	1				1	2	2
	Blood in stool	8	5			1		2	2
	C. difficile	19	12	2	3			2	0
	Diverticulitis of colon	3	3						0
	Infectious diarrhea, NOS	5	2			1	1	1	1
	Noninfectious gastroenteritisb	26	16		3	3	3	1	3
	Otherc	4	4						2
	Ulcerative colitis	3	1			1		1	0
No	Abdominal pain	2	2						1
	Benign neoplasm in the large bowel	3	3						2
	Diarrhead	30	21	3	3		1	2	5
	Gastrointestinal malignancy	3	3						1
	Intestinal impaction	4	3	1					1
	Intestinal obstruction	5	4					1	2
	No diagnosise	74	65	3	1	3		2	1
	Otherf	7	3	1	3				2
	Paralytic ileus	5	4			1			0
	Total	205	152	10	13	10	6	14	25

^aPatients are divided into two groups based on gastrointestinal disease discharge diagnoses, i.e., diagnoses for patients likely to have a breach in colonic mucosa and diagnoses for patients not likely to have a breach in the colonic mucosa. The total numbers of patients, number of patients with the listed FLT findings, and numbers of invasive procedures, including endoscopy, biopsy, and/or surgery, are listed.

^bNoninfectious gastroenteritis was defined as inflammation of the gastrointestinal tract without an infectious agent identified.

^cIncluded patients with intestinal tuberculosis with obstruction, anal and rectal abscesses with fissures, appendicitis, and intestinal obstruction with regional enteritis.

^dDiarrhea was likely osmotic, and there was no evidence of mucosal inflammation.

^ePatients who did not have gastrointestinal disease discharge diagnosis and likely represent a group with transient gastrointestinal symptoms.

^fIncluded patients with irritable bowel syndrome (n = 2), a history of colonic malignancy (n = 2), hemorrhoids (n = 1), diverticulosis (n = 1), and constipation (n = 1).

FIG. 1.

Analysis of cutoff values for FLT. (A) Classification of patients with and without a breach in the colonic mucosa in relation to FLT results using a cutoff of <1 neutrophil/HPF as an indication of positivity. The sensitivity was 39%, the specificity was 81%, and the PPV was 53%. (B) Classification of patients with and without a breach in the colonic mucosa in relation to FLT results using a cutoff of >1 neutrophil/HPF as an indication of positivity. The sensitivity was 28%, the specificity was 92%, the PPV was 67%, and the NPV was 70%.

In conclusion, FLT detects colonic mucosal breach and inflammation, but recent results with FLT in predicting positive stool cultures from patients with infectious enteritis have been less promising (1, 11, 13). Other inflammatory gastrointestinal conditions cause erosion or ulceration, resulting in positive FLTs, and provide an explanation for poor FLT performance in differentiating infectious and noninfectious gastroenteritis. Other explanations for FLT variability include the fact that fecal leukocytes degenerate if they are not processed rapidly (6); therefore, since inflammatory processes in the right colon or appendix may permit degeneration of the cells during transit within the colon itself, a false-negative FLT may result. Sample variability, including the distribution of leukocytes within stool specimens, may also influence FLT results. For example, in cases of acute inflammatory disease of the anal region, there may not be sufficient time for leukocytes to become enmeshed in the feces, adding to the potential for false-negative results. However, even with these limitations, the use of a cutoff value of >1 neutrophil/HPF with FLT resulted in 92% specificity for detection of mucosal integrity. This cutoff was also used by Savola et al. (11), who concluded that FLT was not useful for identification of C. difficile infections in hospitalized patients; our results support this finding.

Studies investigating FLT in conjunction with occult blood screening (guaiac testing) for detection of invasive bacterial enteritis have not found improved sensitivity (6, 8, 12). We also found no correlation between FLT, occult blood screening, and bacterial enteritis. Similarly, guaiac testing is an effective screening test for colon cancer, but no fecal leukocytes were found in patients with enteric neoplasm (n = 8). Therefore, it appears that these two tests detect somewhat different physiological processes. Until this difference is clarified, it may be misleading to compare FLT or tests for fecal leukocyte markers to guaiac testing when assessing the performance of FLT, which is a currently accepted practice (7). Our results demonstrate that, even in conjunction with guaiac testing, FLT cannot distinguish between infectious gastroenteritis and noninfectious gastroenteritis in inpatients and should not be used for this purpose. In summary, a test with even an optimized cutoff value of >1 neutrophil/HPF detected breached colonic mucosae only 20% more often than a coin toss (PPV, 67%; NPV, 70%), and repeat testing was not helpful in most cases (82%).