- Email: [email protected]
Risk factors for Clostridium difficile toxin-positive nosocomial diarrhoea
International Journal of Antimicrobial Agents 28 (2006) 231–237
Risk factors for Clostridium difficile toxin-positive nosocomial diarrhoea David Raveh a,∗ , Bella Rabinowitz b , Gabriel S. Breuer c , Bernard Rudensky d , Amos M. Yinnon a,e a
d
Infectious Diseases Unit, Shaare Zedek Medical Center, Jerusalem, affiliated with the Faculty of Health Sciences, Ben-Gurion University, Beer-Shevah, Israel b School of Pharmacy, Faculty of Medicine, The Hebrew University, Jerusalem, Israel c Department of Medicine, Shaare Zedek Medical Center, Jerusalem, affiliated with the Faculty of Health Sciences, Ben-Gurion University, Beer-Shevah, Israel Clinical Microbiology Laboratory, Shaare Zedek Medical Center, Jerusalem, affiliated with the Faculty of Health Sciences, Ben-Gurion University, Beer-Shevah, Israel e School of Medicine, Faculty of Medicine, The Hebrew University, Jerusalem, Israel Received 29 December 2005; accepted 10 April 2006
Abstract Data were retrieved from the records of all patients from whom stool was sent for Clostridium difficile toxin testing during the year 2001. Toxin-positive and -negative patients were compared by bivariate analysis and regression models. Eight hundred samples from 610 patients were sent for C. difficile toxin testing. Charts of 535 patients (88%) were available for analysis. Of those, 17% had a positive toxin test whilst 83% had a negative toxin test. There was no difference in the number of daily bowel movements between the two groups. Toxin-positive patients were older (P < 0.0001), more often came from nursing homes (P < 0.05), had higher leukocyte counts (P < 0.001), higher blood urea nitrogen (P < 0.01), lower serum albumin (P < 0.01) and more often received diuretics (P < 0.01) and clindamycin (P < 0.05). Logistic regression analysis showed that previous antibiotic-associated diarrhoea was the most significant risk factor for toxin-positive diarrhoea (P < 0.001), followed by clindamycin treatment (P < 0.005), diuretics (P < 0.005) and older age (P < 0.05). Another logistic model showed the contribution of macrolides (P < 0.05) to the development of hospital-acquired diarrhoea. © 2006 Published by Elsevier B.V. and the International Society of Chemotherapy. Keywords: Diarrhoea; Clostridium difficile; Antibiotic-associated colitis; Pseudomembranous colitis; Nosocomial infections
1. Introduction Nosocomial diarrhoea, an important cause of morbidity in hospitalised patients, may have an infectious or noninfectious aetiology. The most common infectious cause is toxigenic Clostridium difficile [1,2]. Non-infectious causes include chemotherapy, stool softeners, hyperosmolar enteral feeding formulas and antibiotics. Antibiotics can induce diarrhoea by altering colonic flora [3], increasing peristalsis and acting as colonic irritants. Exposure to antibiotics appears to ∗ Corresponding author. Present address: Shaare Zedek Medical Center, P.O. Box 3235, Jerusalem 91031, Israel. Fax: +972 2 666 6342. E-mail address: [email protected] (D. Raveh).
create a niche that is exploited by C. difficile [1,4,5]. In hospitalised patients, antibiotic-associated diarrhoea (AAD) has been associated with increased mortality and longer length of stay [1,6,7]. Clostridium difficile is estimated to colonise 3% of healthy adults [1,2]. In hospitalised adults, the rate of asymptomatic colonisation ranges from 15% to 25% [2,5,8]. The incidence of nosocomial diarrhoea associated with C. difficile ranges from 1 to 30 cases per 1000 hospital discharges [9]. However, the majority of hospital C. difficile-associated diarrhoea has been shown to be caused by endogenous rather than hospital strains [10], which means that low-level C. difficile carriage is not detected by routine methods and its expression is encouraged by hospital environmental factors such as medications,
0924-8579/$ – see front matter © 2006 Published by Elsevier B.V. and the International Society of Chemotherapy. doi:10.1016/j.ijantimicag.2006.04.013
232
D. Raveh et al. / International Journal of Antimicrobial Agents 28 (2006) 231–237
worsening medical conditions, etc. Higher asymptomatic carriage has been observed among elderly patients admitted to long-stay wards compared with those admitted to acute shortstay wards [2,3,8,9]. Known precipitants of C. difficile-associated diarrhoea (CDAD) include antibiotics, cancer chemotherapy and gastrointestinal illness or manipulation [1,2,6,9,11]. Other potential risk factors are advanced age and serious illness [1,2,7,12]. Host factors are increasingly recognised as critical determinants of disease expression. The magnitude of AAD and CDAD in different patient populations has not been thoroughly described. AAD frequencies of 10–30% have been reported among hospitalised patients treated with antibiotics. The aetiology and risk factors of AAD not associated with C. difficile are controversial [8,13,14]. Broad-spectrum antimicrobial treatment has been associated with C. difficile infection (CDI), mainly with cephalosporins and clindamycin [1,5,8,9]. The role of quinolones [15–18], penicillins combined with -lactamase inhibitors [19] and carbapenems has not been well studied. Interestingly, macrolides have been proposed to be a risk factor for the development of CDI [1,2,5,11,15,20,21] and, recently, linezolid [22]. We aimed to study the characteristics and risk factors of patients who developed hospital-acquired diarrhoea with either positive or negative C. difficile stool toxin.
2. Patients and methods Shaare Zedek Medical Center is Jerusalem’s second largest teaching hospital with 550 beds. It includes all medical and surgical services except for transplantations and neurosurgery. This retrospective study is based on the laboratory list of all inpatients from whom stool was sent to detect C. difficile toxin during the year 2001. Relevant data were retrieved from patients’ records: date of admission; age and gender; whether the admission was from home, nursing home or another institution; hospitalisation within the preceding 2 months; patient mobility; admitting ward; history of any surgery during the present hospitalisation and/or 3 months prior to it; history of any endoscopic procedure during the current hospitalisation; order and duration of antibiotic treatment during the 3 months before onset of the current episode of diarrhoea, either during the present admission or prior to that; history of gastrointestinal or other diseases; any medications given for constipation and for peptic ulcer disease; results of laboratory tests from the day when stool was sent for Clostridium toxin testing; and treatments and complications of hospital-acquired diarrhoea. Detection of C. difficile toxins A and B was performed using an ELISA kit (TechLab, Blacksburg, VA). 2.1. Statistical analysis Data were entered into and processed by Epi Info 6.04d (CDC, Atlanta, GA). Student’s t-test and 2 (or Fisher’s exact) test were performed using this program. SPSS ver-
sion 11.0 and Epi Info 2000 were used for logistic regression analysis. A P-value <0.05 was defined as significant. Analysis of antibiotic use is always difficult because of the high number of combinations and frequent changes in empirical and culture-tailored treatments. We therefore decided to focus on the type and treatment duration of each antibiotic. For practical purposes, antibiotics were grouped by classes.
3. Results From January 2001 until February 2002, 800 stool specimens from 610 inpatients with diarrhoea were sent to the laboratory to test for the presence of C. difficile toxin. Of the 610 patients, 75 records (12%) were unavailable for the study. Consequently, 535 patients were studied from whom stool specimens were sent for testing. Of these 535 patients, 361 (67%) were from medical departments, 51 (10%) from the emergency department, 46 (9%) from paediatric wards and the remainder were from the surgical and gynaecological services. In 104 patients (19%) more than one stool test was sent during the same hospitalisation (mean ± standard deviation (S.D.), 2.5 ± 1.1; range, 2–10). Ninety-one patients (17%) had toxin-positive diarrhoea whilst 444 (83%) had toxin-negative diarrhoea. The ratio between patients tested for Clostridium toxin and the total number of patients in that specific department (over the study period) was calculated and showed that the average ratio in the medical departments was ± 6%. The highest ratio reached 86% of patients in the six-bed intensive care unit. The mean patient age was 68 ± 25 years; 54% of patients were female. Although the laboratory requirement for stool testing for C. difficile toxin is that the number of diarrhoeal stools is six or more per day [14], chart review revealed that the bowel movement frequency ranged from 2 to 20 per day (mean ± S.D., 6 ± 4.2). Patients with a positive toxin test had 6.5 ± 4.1 bowel movements per day, whilst those with a negative test had 5.8 ± 4.1 bowel movements per day (not significant). Table 1 shows the background information and Table 2 shows the relevant clinical and laboratory data from the day the stool sample was sent for toxin testing. Although 400 patients (75%) came from home and 135 patients (25%) came from nursing homes, the charts did not reveal the percentage of patients coming from home who were actually independent. Among the complications of antibiotic-associated colitis (AAC), only 1 patient underwent total colectomy, 13 had electrolyte imbalance and 32 (6%) had a relapse of diarrhoea. Sixty-four patients (12%) died during the relevant admission: no autopsy was performed, but in none of the cases was death attributed to AAC or to failure to perform total colectomy in time, but rather to the overall condition of the patient. Table 3 presents the antibiotic regimens that the patients received during the 3 months prior to the toxin test, including the duration and chronological order of administration. Thirty-five percent were treated with cefuroxime, 28% with
D. Raveh et al. / International Journal of Antimicrobial Agents 28 (2006) 231–237
233
Table 1 Background data of all patients with nosocomial diarrhoea (N = 535)
Table 2 Clinical presentation of all patients with nosocomial diarrhoea (N = 535)
Variable
n
%
Characteristic
n
Mean ± S.D. (range) or %
Age (years) Male gender Pregnancy
528 247 7
68 ± 25 (0–100)a 46 1
Daily functioning Nursing home Home
135 400
25 75
Clinical presentation No. of stools per day Fever Temperature (◦ C) Loss of appetite Abdominal pain
134 296 287 38/41 111/131
6 ± 4 (2–20) 55% 38 ± 1 (35.6–40.1) 93% 85%
Stool test (% from tested patients) Positive occult blood Presence of RBCs Presence of WBCs Colonoscopy done
50/109 5/49 5/48 49
46% 10% 10% 9%
Toxin test result Negative Positive More than one sample sent
444 91 104
83% 17% 2.5 ± 1.1 (2–10)
Laboratory data WBC count (×103 /mm3 ) Neutrophils (%) ESR (mm/h) BUN (mg/dL) Creatinine (mg/dL) Albumin (g/dL)
511 336 126 498 499 356
15.1 ± 11.6 (0.6–100) 73 ± 20 (2–98) 71 ± 37 (5–137) 29 ± 30 (2–230) 1.3 ± 1.1 (0.2–8.4) 2.9 ± 0.5 (1.5–4.3)
Medical background Ischaemic heart disease Hypertension Cerebral vascular accident Malignancy Diabetes mellitus Collagen diseases Renal failure Dialysis Surgery in the last 3 months Additional gastrointestinal diseases
235 231 123 100 139 36 61 10 74 218
44 43 23 19 26 7 11 2 14 41
Treatment NSAIDs ACE inhibitors Beta blockers Peptic ulcer disease drugs Diabetes mellitus drugs Anticoagulants Laxatives Calcium channel blockers Antiarrhythmics Digoxin Diuretics
156 148 141 228 102 114 61 123 54 43 182
29 28 26 43 19 21 11 23 10 8 34
26 60 8 10
25 58 8 10
Immunosuppressive treatment (n = 104) Chemotherapy Steroid treatment Chemotherapy + steroids Other Antibiotics in the last 3 months Days without antibiotics Previous AAC No. of months after previous AAC
411 258 28 26
77 1 ± 4 (0–30)a 5 2.7 ± 3.8 (0.1–18)a
NSAIDs, non-steroidal anti-inflammatory drugs; ACE, angiotensinconverting enzyme; AAC, antibiotic-associated colitis. a Mean ± standard deviation (range).
ciprofloxacin, 25% with metronidazole, 25% with aminoglycosides and 20% with ampicillin. Table 3 demonstrates the complexity of antibiotic treatment, as they are usually prescribed in combination with frequent changes. Table 4 shows only the clinically relevant and statistically significant factors detected by bivariate analysis: it compares the characteristics of patients who had a positive toxin result (n = 91) with those of patients with a negative result (n = 444). For those patients who had multiple tests sent, we referred to the first test only. Factors that were not found to be significant in the bivariate analysis were: gender; ethnic origin; admitting ward; existing heart disease; hypertension; past cerebrovascular accident; malignancy; diabetes mellitus; collagen disease; chronic renal failure; chronic dialysis; surgery in the preceding 3 months; other gastrointestinal diseases;
Clinical response No. of days until temperature normalised No. of days until diarrhoea stopped No. of days until abdominal pain stopped No. of days until appetite returned
116
4.7 ± 4.6 (1–30)
198
6 ± 7 (1–40)
29
7 ± 6 (2–30)
14
11 ± 11 (1–35)
RBC, red blood cell; WBC, white blood cell; ESR, erythrocyte sedimentation rate; BUN, blood urea nitrogen.
pregnancy; antibiotic treatment in the preceding 3 months; immunosuppression; time since the previous AAC; fever; number of bowel movements per day; loss of appetite; occult blood in the stool; number of samples sent for Clostridium toxin testing; imaging; order of utilisation of metronidazole, vancomycin and cholestyramine; days to defervescence; resolution of diarrhoea and abdominal pains; rate of appetite return; mortality (although, as previously mentioned, this was never considered to result from AAC); and electrolyte imbalance. The blood count parameters did not differ between the patients with positive or negative stool toxin tests, except for the total number of white blood cells, which differed significantly (P < 0.001) [23]. Biochemical parameters did not differ between patients with positive or negative stool toxin results except for blood urea nitrogen (BUN) (P < 0.01) and serum albumin (P < 0.01). Of all the classes of drugs administered to patients (other than antibiotics), diuretics were associated with a positive toxin test (P < 0.01), with an almost significant association for anticoagulants (P = 0.056) and peptic ulcer drugs (P = 0.065). There was no difference in laxative consumption between the groups. A documented past episode of AAD (P < 0.0001)
234
D. Raveh et al. / International Journal of Antimicrobial Agents 28 (2006) 231–237
Table 3 Antibiotics given to patients during the 3 months before toxin testing Drug
Cefuroxime Ciprofloxacin Metronidazole Aminoglycosides Ampicillin Ceftazidime Vancomycin Co-amoxiclav Piperacillin/tazobactam Nitrofurantoin Imipenem Cefazolin Cefepime Ceftriaxone Chloramphenicol Fluconazole Mezlocillin Co-trimoxazole Amphotericin B Clindamycin Cloxacillin
Order of prescriptiona
%c
1st
2nd
3rd
4th
75 52 45 40 55 18 11 17 4 16 1 9 8 6 3
29 31 29 35 14 30 13 24 14 5 2 7 8 7 7 2 1 3
9 13 8 8 3 14 8 4 3 2 7 2 3 1 3 1 1 1 3 2 1
3 2 2 2
6 2 2 1
1
5th
6th
7th
8th
1 1 2
4 7
2 1 1 1 1 4 1
6 4 1 3
1 1
1 1 1 1
1 1
3
1
1 1
69 52 49 44 35 36 30 21 18 16 21 18 10 9 7 9 5 4 4 3 1
Duration of treatment (days) Mean ± S.D.
Yb 35 28 25 25 20 19 13 13 11 8 7 7 6 5 4 3 3 2 2 2 1
9.4 8.9 11.1 7.6 7.3 6.6 8.6 8.2 10.4 8.3
± ± ± ± ± ± ± ± ± ±
6.6 5.7 8.1 6.6 4.4 3.5 6.1 7.7 8.4 4.1
15.2 7.5 4.3 4.0
± ± ± ±
16.7 0.7 3.2 2.0
7.5 ± 2.1
Min.
Max.
nd
2 1 2 2 2 2 2 2 4 1 25 7 7 3 2
30 21 30 30 21 14 22 30 25 14 45 8 8 6
32 14 13 17 17 15 8 19 5 8 1 5 2 3 3
2 6
9
1 2
45
1
S.D., standard deviation. a Chronological order of prescription of antibiotics: each cell gives the number of patients who received the antibiotics in this order. b No. of cases receiving that antibiotic but where the prescription order was unknown. c Total no. of cases receiving the specific antibiotic (known and unknown order), expressed as percentage of total cases (N = 535). d No. of cases for whom the duration of treatment was known.
and presence of abdominal pain (P = 0.042) were associated with the development of toxin-positive diarrhoea. The order of antibiotic administration, the duration of each administration and the various combinations were too complex to consider in the analysis. We therefore performed two
types of logistic regression analysis: first, the contribution of each drug in a dichotomic mode (whether it was given or not), regardless of duration, order and combination. The results show that none of the antibiotics were associated with toxin-positive diarrhoea, except for clindamycin (P = 0.012),
Table 4 Comparison of Clostridium difficile toxin-positive and -negative diarrhoea patients (mean ± S.D.) Variable Age Residence Home Nursing home Peripheral leukocytes (×103 /mm3 ) Blood urea nitrogen (mg/dL) Serum albumin (g/L) Received diuretics Received clindamycin Days without antibiotics Previous AAC Stool erythrocytes Stool leukocytes >1 sample sent for Clostridium toxin test Abdominal pain Treatment with vancomycin Treatment with metronidazole Combined therapy (metronidazole + vancomycin) Probiotics (additional treatment) Days until diarrhoea stopped Days until abdominal pain stopped
Clostridium toxin-negative (n = 444) 66 ± 26 353 (81%) 82 (19%) 14.4 ± 11.5 29 ± 32 2.9 ± 0.5 139 (32%) 4 (1%) 0.54 ± 2.2 16 (4%) 3 (7%) 3 (7%) 76 (17%) 89 (82%) 27 (6%) 135 (31%) 5 (0.3%) 7 (2%) 5.3 ± 5.1 5.1 ± 3.3
S.D., standard deviation; AAC, antibiotic-associated colitis.
Clostridium toxin-positive (n = 91)
P-value
76 ± 20
<0.0001 <0.05
64 (72%) 25 (28%) 19.1 ± 11.7 32 ± 23 2.7 ± 0.5 42 (50%) 4 (4%) 2.3 ± 5.1 13 (14%) 2 (67%) 2 (67%) 28 (31%) 20 (100%) 28 (31%) 69 (77%) 9 (10%) 7 (8%) 9.7 ± 9.0 11.6 ± 8.6
<0.001 <0.01 <0.01 <0.01 <0.05 <0.05 <0.0001 <0.001 <0.001 <0.005 <0.05 <0.0001 <0.0001 <0.0001 <0.001 <0.0001 <0.05
D. Raveh et al. / International Journal of Antimicrobial Agents 28 (2006) 231–237
despite the rare use of this drug. Ceftriaxone was the next most closely associated (P = 0.086). Second, we assigned each antibiotic a numeric value of days administered, starting from zero (= drug not given). This model included aminoglycosides, ampicillin, amoxicillin/clavulanic acid (coamoxiclav), four generations of cephalosporins, chloramphenicol, clindamycin, imipenem, macrolides, ciprofloxacin [16–18] and piperacillin/tazobactam [19]. This ‘duration’ model showed that only the macrolide group was associated with toxin positivity, with P-values of 0.03–0.05 in various models. A significant association was also found for the presence of erythrocytes and leukocytes in stool samples of toxinpositive patients (P < 0.005) [23]. However, ordering this simple, useful, fast and inexpensive stool test was unfortunately rare, which limits our ability to draw conclusions. There was no association between the results of imaging tests and positive toxin tests. The recovery time for toxin-positive patients was longer than for the toxin-negative group, as measured by the time to resolution of abdominal pain (P < 0.05) and diarrhoea (P < 0.0001). The logistic regression analysis was based on factors found to be significant in the bivariate analysis. We tested various models because several variables had many missing values (e.g. stool erythrocyte and leukocyte tests, etc.). Forward stepwise regression analysis showed that previous AAC was the most significant risk factor for diarrhoea with a positive toxin test (P < 0.001), followed by clindamycin treatment (P < 0.005), diuretics (P < 0.005) and older age (P < 0.05).
4. Discussion In this retrospective study, we sought to determine risk factors associated with the development of toxin-positive hospital-acquired diarrhoea. Clostridium difficile toxin test was determined for 535 hospitalised patients. Despite our laboratory’s demand that samples are sent for toxin testing only from patients with at least six loose stools per day [14], this requirement was not always strictly followed by the referring physicians. Only 134 cases (25%) of the 535 patients included a recorded number of stools passed per day; the maximum number of recorded stools per day reached 20. In several cases there were problems that prevented counting the number of bowel movements, i.e. patients wearing diapers or patients suffering from dementia. Our finding of an association between the age of patients and stool toxin positivity is consistent with earlier data [1,2,5,8]. In our study, the mean ± S.D. age of toxin-negative patients was 66 ± 26 years, whilst toxin-positive patients were almost 10 years older (76 ± 20 years). There is in vitro evidence that colonic contents of younger people inhibit C. difficile more effectively than that of older people [8]. Furthermore, in elderly patients there are more concomitant risk factors, such as longer hospitalisation, more additional
235
diseases and complicated treatments [1,8,24]. Patients who were admitted from other hospitals or nursing homes were at greater risk for development of CDAD than patients who came from home. The percentage of women with hospitalacquired diarrhoea was almost 10% higher than men (54% vs. 46%), but we did not find a gender difference in the toxin positivity rate. Some studies have shown that women are more likely to develop AAD (60%), presumably because they are frequently treated with antibiotics for urinary tract infections [2], whilst others have shown the opposite [1,2,7,20]. Similar to Wistrom et al. [8], we did not find an association between toxin positivity and chronic diseases such as diabetes, renal failure and malignancy, in contrast to other studies [1,2,7,11,24]. We also did not find an association between CDAD and immunosuppressive therapy, and the only drugs found to be associated were diuretics as well as an almost significant association with anticoagulants and antacid drugs. We speculate that reduction in gastric acidity by medications decreases acidic gastric killing and enhances colonisation with C. difficile. We tried to detect surrogate markers for CDAD, its severity and outcome. As in previous studies [1,8,13,14], the white blood cell count was higher in toxin-positive patients compared with toxin-negative ones [23]. Other reported but less frequent factors are dehydration, electrolyte imbalance and hypoalbuminaemia [5,8,13,14]. We did not find additional significantly different laboratory results, except for serum albumin, which was much lower in the toxin-positive group both by bivariate and multivariate analysis. An unanswered question is whether the low albumin level is a risk factor for CDAD, a known marker for very sick patients in general, or a result of the AAD, which also causes protein-losing enteropathy. Comparing renal function tests for the two groups showed higher BUN among patients with a positive toxin test, without a difference in serum creatinine values. This might suggest that dehydration is an important risk factor of CDAD, either due to the diuretic drugs or fluid loss or fluid shift into the third space owing to hypoalbuminaemia. However, we did not find an additional association between toxin positivity and dehydration markers such as hypernatraemia or elevated haematocrit. In order for C. difficile to colonise the gut of a normal individual, the resident normal flora must be altered [3]. Anaerobic gut bacteria are believed to be crucial to ‘colonisation resistance’, but the precise components involved have not been clearly defined. Contradicting evidence derives from the fact that cephalosporins cause more CDAD than piperacillin/tazobactam or carbapenems, despite the fact that the latter agents efficiently kill most of the anaerobic flora of the colon, unlike cephalosporins. There is in vitro evidence that faecal contents of elderly patients are less inhibitory to the growth of C. difficile [8]. The main antibiotics associated with CDAD are secondgeneration and more advanced generation cephalosporins and clindamycin [1,9], and the duration of antimicrobial
236
D. Raveh et al. / International Journal of Antimicrobial Agents 28 (2006) 231–237
therapy is also an important risk factor [5,8]. Macrolides in general, and in particular clarithromycin, have been proposed to be risk factors for the development of CDI. Aminoglycosides appear to have reduced propensity to induce CDI, probably owing to their lack of effect on the endogenous anaerobic gut bacteria. In addition, there has been little evidence linking quinolones with CDAD. Older quinolones (including ciprofloxacin) have poor antianaerobic activity, whereas newer agents such as moxifloxacin and gatifloxacin have significantly increased activity against anaerobes [17,18]. Data from several studies indicate that antipseudomonal penicillins such as piperacillin and ticarcillin appear to share the low-risk status [5,8,19] for CDAD compared with cephalosporins, but the reason for this has not been clearly found. Other perceived low-risk agents include penicillin, trimethoprim, rifampicin, fusidic acid and nitrofurantoin. It is notable that there are very few published reports associating co-amoxiclav with CDAD, despite its good anti-anaerobic activity, widespread utilisation and tendency to cause abdominal cramps and loose stools. In vivo antimicrobial activity and, therefore, probably the risk of CDAD, can be markedly influenced by factors such as drug penetration into the large bowel lumen, specific and non-specific antibiotic binding, gut pH and redox potential [5]. We are not familiar with any statistical model that could take into consideration all these factors and give each its appropriate weight. We applied two types of logistic regression analysis: the dichotomic ‘yes/no’ for each antibiotic; and the continuous number of administration days from zero (= not given). In the first type, only clindamycin was associated with AAC despite the infrequent use of this drug in our hospital setting. Ceftriaxone use followed clindamycin, but with a P-value of 0.086 [16]. In the second analysis, macrolides were the only significant factor associated with toxin positivity. In our medical centre at the time of the study period, oral roxithromycin was used for respiratory infections and oral clarithromycin as a part of the Helicobacter pylori treatment protocol. We found a strong correlation between toxin positivity and the presence of occult blood, white blood cells and red blood cells in stool, as reported by other investigators [5,8,24,25]. However, only in less than 10% of cases was a direct stool smear performed. This quick, simple, informative and inexpensive test can be helpful in the diagnostic work-up of diarrhoea and should be used much more often. Following the course to recovery, we found that toxin-positive patients had significantly longer time until resolution of diarrhoea and abdominal pain than patients with toxin-negative diarrhoea. No difference was found for time until defervescence or for return of appetite. We also found that patients who had a previous history of AAD were at increased risk for development of CDAD. In this study, we observed a tendency of physicians to repeat and send stool samples for toxin testing, sometimes following the initial test, for symptomatic patients. Previous
studies [5] have shown that the toxin test may remain positive in 25% of patients long after recovery from the clinical disease. In addition, there is a possibility of false-negative results. Both facts emphasise the lack of benefit for sending repeat tests in toxin-positive patients. There could be a rationale for repetitive testing in symptomatic toxin-negative patients. One limitation of our study is its retrospective nature. The optimum study would be prospective, screening patients on admission for toxin and C. difficile, and following patients throughout the admission, assessing for acquisition of C. difficile and production of toxin while recording all medications administered. Regarding antibiotic treatments, a comprehensive mathematical model should be constructed to define risk factors for development of hospital-acquired diarrhoea. Another limitation is the lack of a diarrhoea-free control group, although our study looked at ‘hospital-acquired diarrhoea’ and compared patients with toxin-positive diarrhoea with those with toxin-negative diarrhoea. In conclusion, in this study of 535 patients with hospitalacquired diarrhoea, several epidemiological, clinical and laboratory risk factors for development of C. difficile toxinpositive diarrhoea were identified. Targeting these patients for preventive measures may possibly contribute to reduced incidence and associated morbidity.
References [1] Thielman NM, Wilson KM. Antibiotic-associated colitis. In: Mandell GL, Bennett JE, Dolin R, editors. Principles and practice of infectious diseases. 6th ed. Philadelphia, PA: Churchill Livingstone; 2005. p. 1249–63. [2] Klingler PJ, Metzger PP, Seelig HS, Pettit PD, Knudsen JM, Alvarez SA. Clostridium difficile infection: risk factors, medical and surgical management. Dig Dis 2000;18:147–60. [3] Tonna I, Welsby PD. Pathogenesis and treatment of Clostridium difficile infection. Postgrad Med J 2005;81:367–9. [4] Hogenauer C, Hammer HF, Krejs GJ, Reisinger EC. Mechanisms and management of antibiotic-associated diarrhea. Clin Infect Dis 1998;27:702–10. [5] Wilcox MH. Gastrointestinal disorders and the critically ill. Clostridium difficile infection and pseudomembranous colitis. Best Pract Res Clin Gastroenterol 2003;17:475–93. [6] McFarland LV, Mulligan ME, Kwok RY, Stamm WE. Nosocomial acquisition of Clostridium difficile infection. N Engl J Med 1989;320:204–10. [7] McFarland LV, Stamm WE. Review of Clostridium difficileassociated diseases. Am J Infect Control 1986;14:99–109. [8] Wistrom J, Norrby SR, Myrhe EB, et al. Frequency of antibioticassociated diarrhoea in 2642 antibiotic-treated hospitalized patients: a prospective study. J Antimicrob Chemother 2001;47:43–50. [9] Gross MH. Management of antibiotic-associated pseudomembranous colitis. Clin Pharm 1985;4:304–10. [10] Noren T, Akerlund T, Back E, et al. Molecular epidemiology of hospital-associated and community-acquired Clostridium difficile infection in a Swedish county. J Clin Microbiol 2004;42:3635–43. [11] Anand A, Glatt AE. Clostridium difficile infection associated with antineoplastic chemotherapy: a review. Clin Infect Dis 1993;17: 109–13.
D. Raveh et al. / International Journal of Antimicrobial Agents 28 (2006) 231–237 [12] Chang VT, Nelson K. The role of physical proximity in nosocomial diarrhea. Clin Infect Dis 2000;31:717–22. [13] Bulusu M, Narayan S, Shetler K, Triadafilopoulos G. Leukocytosis as a harbinger and surrogate marker of Clostridium difficile infection in hospitalized patients with diarrhea. Am J Gastroenterol 2000;95:3137–41. [14] Jacobs J, Rudensky B, Dresner J, et al. Comparison of four laboratory tests for diagnosis of Clostridium difficile-associated diarrhea. Eur J Clin Microbiol Infect Dis 1996;15:561–6. [15] Arfons L, Ray AJ, Donskey CJ. Clostridium difficile infection among health care workers receiving antibiotic therapy. Clin Infect Dis 2005;40:1384–5. [16] Muto CA, Pokrywka M, Shutt K, et al. A large outbreak of Clostridium difficile-associated disease with an unexpected proportion of deaths and colectomies at a teaching hospital following increased fluoroquinolone use. Infect Control Hosp Epidemiol 2005;26:273–80. [17] Katikireddi V. UK launches inquiry into Clostridium difficile outbreak. Can Med Assoc J 2005;173:138. [18] Gaynes R, Rimland D, Killum E, et al. Outbreak of Clostridium difficile infection in a long-term care facility: association with gatifloxacin use. Clin Infect Dis 2004;38:640–5.
237
[19] Baines SD, Freeman J, Wilcox MHJ. Effects of piperacillin/ tazobactam on Clostridium difficile growth and toxin production in a human gut model. J Antimicrob Chemother 2005;55:974–82. [20] Schwaber MJ, Simhon A, Block C, Roval V, Ferderber N, Shapiro M. Factors associated with nosocomial diarrhea and Clostridium difficile-associated disease on the adult wards of an urban tertiary care hospital. Eur J Clin Microbiol Infect Dis 2000;19:9–15. [21] Martirosian G, Szczesny A, Cohen SH, Silva Jr J. Analysis of Clostridium difficile-associated diarrhea among patients hospitalized in tertiary care academic hospital. Diagn Microbiol Infect Dis 2005;52:153–5. [22] Zabel LT, Worm S. Linezolid contributed to Clostridium difficile colitis with fatal outcome. Infection 2005;33:155–7. [23] Bouza E, Munoz P, Alonso R. Clinical manifestations, treatment and control of infections caused by Clostridium difficile. Clin Microbiol Infect 2005;11(Suppl. 4):57–64. [24] Wilkins TD, Lyerly DM. Clostridium difficile testing: after 20 years, still challenging. J Clin Microbiol 2003;41:531–4. [25] Zuckerman E, Kanel G, Ha C, Kahn J, Gottesman BS, Korula J. Low albumin gradient ascites complicating severe pseudomembranous colitis. Gastroenterology 1997;112:991–4.
Risk factors for Clostridium difficile toxin-positive nosocomial diarrhoea David Raveh a,∗ , Bella Rabinowitz b , Gabriel S. Breuer c , Bernard Rudensky d , Amos M. Yinnon a,e a
d
Infectious Diseases Unit, Shaare Zedek Medical Center, Jerusalem, affiliated with the Faculty of Health Sciences, Ben-Gurion University, Beer-Shevah, Israel b School of Pharmacy, Faculty of Medicine, The Hebrew University, Jerusalem, Israel c Department of Medicine, Shaare Zedek Medical Center, Jerusalem, affiliated with the Faculty of Health Sciences, Ben-Gurion University, Beer-Shevah, Israel Clinical Microbiology Laboratory, Shaare Zedek Medical Center, Jerusalem, affiliated with the Faculty of Health Sciences, Ben-Gurion University, Beer-Shevah, Israel e School of Medicine, Faculty of Medicine, The Hebrew University, Jerusalem, Israel Received 29 December 2005; accepted 10 April 2006
Abstract Data were retrieved from the records of all patients from whom stool was sent for Clostridium difficile toxin testing during the year 2001. Toxin-positive and -negative patients were compared by bivariate analysis and regression models. Eight hundred samples from 610 patients were sent for C. difficile toxin testing. Charts of 535 patients (88%) were available for analysis. Of those, 17% had a positive toxin test whilst 83% had a negative toxin test. There was no difference in the number of daily bowel movements between the two groups. Toxin-positive patients were older (P < 0.0001), more often came from nursing homes (P < 0.05), had higher leukocyte counts (P < 0.001), higher blood urea nitrogen (P < 0.01), lower serum albumin (P < 0.01) and more often received diuretics (P < 0.01) and clindamycin (P < 0.05). Logistic regression analysis showed that previous antibiotic-associated diarrhoea was the most significant risk factor for toxin-positive diarrhoea (P < 0.001), followed by clindamycin treatment (P < 0.005), diuretics (P < 0.005) and older age (P < 0.05). Another logistic model showed the contribution of macrolides (P < 0.05) to the development of hospital-acquired diarrhoea. © 2006 Published by Elsevier B.V. and the International Society of Chemotherapy. Keywords: Diarrhoea; Clostridium difficile; Antibiotic-associated colitis; Pseudomembranous colitis; Nosocomial infections
1. Introduction Nosocomial diarrhoea, an important cause of morbidity in hospitalised patients, may have an infectious or noninfectious aetiology. The most common infectious cause is toxigenic Clostridium difficile [1,2]. Non-infectious causes include chemotherapy, stool softeners, hyperosmolar enteral feeding formulas and antibiotics. Antibiotics can induce diarrhoea by altering colonic flora [3], increasing peristalsis and acting as colonic irritants. Exposure to antibiotics appears to ∗ Corresponding author. Present address: Shaare Zedek Medical Center, P.O. Box 3235, Jerusalem 91031, Israel. Fax: +972 2 666 6342. E-mail address: [email protected] (D. Raveh).
create a niche that is exploited by C. difficile [1,4,5]. In hospitalised patients, antibiotic-associated diarrhoea (AAD) has been associated with increased mortality and longer length of stay [1,6,7]. Clostridium difficile is estimated to colonise 3% of healthy adults [1,2]. In hospitalised adults, the rate of asymptomatic colonisation ranges from 15% to 25% [2,5,8]. The incidence of nosocomial diarrhoea associated with C. difficile ranges from 1 to 30 cases per 1000 hospital discharges [9]. However, the majority of hospital C. difficile-associated diarrhoea has been shown to be caused by endogenous rather than hospital strains [10], which means that low-level C. difficile carriage is not detected by routine methods and its expression is encouraged by hospital environmental factors such as medications,
0924-8579/$ – see front matter © 2006 Published by Elsevier B.V. and the International Society of Chemotherapy. doi:10.1016/j.ijantimicag.2006.04.013
232
D. Raveh et al. / International Journal of Antimicrobial Agents 28 (2006) 231–237
worsening medical conditions, etc. Higher asymptomatic carriage has been observed among elderly patients admitted to long-stay wards compared with those admitted to acute shortstay wards [2,3,8,9]. Known precipitants of C. difficile-associated diarrhoea (CDAD) include antibiotics, cancer chemotherapy and gastrointestinal illness or manipulation [1,2,6,9,11]. Other potential risk factors are advanced age and serious illness [1,2,7,12]. Host factors are increasingly recognised as critical determinants of disease expression. The magnitude of AAD and CDAD in different patient populations has not been thoroughly described. AAD frequencies of 10–30% have been reported among hospitalised patients treated with antibiotics. The aetiology and risk factors of AAD not associated with C. difficile are controversial [8,13,14]. Broad-spectrum antimicrobial treatment has been associated with C. difficile infection (CDI), mainly with cephalosporins and clindamycin [1,5,8,9]. The role of quinolones [15–18], penicillins combined with -lactamase inhibitors [19] and carbapenems has not been well studied. Interestingly, macrolides have been proposed to be a risk factor for the development of CDI [1,2,5,11,15,20,21] and, recently, linezolid [22]. We aimed to study the characteristics and risk factors of patients who developed hospital-acquired diarrhoea with either positive or negative C. difficile stool toxin.
2. Patients and methods Shaare Zedek Medical Center is Jerusalem’s second largest teaching hospital with 550 beds. It includes all medical and surgical services except for transplantations and neurosurgery. This retrospective study is based on the laboratory list of all inpatients from whom stool was sent to detect C. difficile toxin during the year 2001. Relevant data were retrieved from patients’ records: date of admission; age and gender; whether the admission was from home, nursing home or another institution; hospitalisation within the preceding 2 months; patient mobility; admitting ward; history of any surgery during the present hospitalisation and/or 3 months prior to it; history of any endoscopic procedure during the current hospitalisation; order and duration of antibiotic treatment during the 3 months before onset of the current episode of diarrhoea, either during the present admission or prior to that; history of gastrointestinal or other diseases; any medications given for constipation and for peptic ulcer disease; results of laboratory tests from the day when stool was sent for Clostridium toxin testing; and treatments and complications of hospital-acquired diarrhoea. Detection of C. difficile toxins A and B was performed using an ELISA kit (TechLab, Blacksburg, VA). 2.1. Statistical analysis Data were entered into and processed by Epi Info 6.04d (CDC, Atlanta, GA). Student’s t-test and 2 (or Fisher’s exact) test were performed using this program. SPSS ver-
sion 11.0 and Epi Info 2000 were used for logistic regression analysis. A P-value <0.05 was defined as significant. Analysis of antibiotic use is always difficult because of the high number of combinations and frequent changes in empirical and culture-tailored treatments. We therefore decided to focus on the type and treatment duration of each antibiotic. For practical purposes, antibiotics were grouped by classes.
3. Results From January 2001 until February 2002, 800 stool specimens from 610 inpatients with diarrhoea were sent to the laboratory to test for the presence of C. difficile toxin. Of the 610 patients, 75 records (12%) were unavailable for the study. Consequently, 535 patients were studied from whom stool specimens were sent for testing. Of these 535 patients, 361 (67%) were from medical departments, 51 (10%) from the emergency department, 46 (9%) from paediatric wards and the remainder were from the surgical and gynaecological services. In 104 patients (19%) more than one stool test was sent during the same hospitalisation (mean ± standard deviation (S.D.), 2.5 ± 1.1; range, 2–10). Ninety-one patients (17%) had toxin-positive diarrhoea whilst 444 (83%) had toxin-negative diarrhoea. The ratio between patients tested for Clostridium toxin and the total number of patients in that specific department (over the study period) was calculated and showed that the average ratio in the medical departments was ± 6%. The highest ratio reached 86% of patients in the six-bed intensive care unit. The mean patient age was 68 ± 25 years; 54% of patients were female. Although the laboratory requirement for stool testing for C. difficile toxin is that the number of diarrhoeal stools is six or more per day [14], chart review revealed that the bowel movement frequency ranged from 2 to 20 per day (mean ± S.D., 6 ± 4.2). Patients with a positive toxin test had 6.5 ± 4.1 bowel movements per day, whilst those with a negative test had 5.8 ± 4.1 bowel movements per day (not significant). Table 1 shows the background information and Table 2 shows the relevant clinical and laboratory data from the day the stool sample was sent for toxin testing. Although 400 patients (75%) came from home and 135 patients (25%) came from nursing homes, the charts did not reveal the percentage of patients coming from home who were actually independent. Among the complications of antibiotic-associated colitis (AAC), only 1 patient underwent total colectomy, 13 had electrolyte imbalance and 32 (6%) had a relapse of diarrhoea. Sixty-four patients (12%) died during the relevant admission: no autopsy was performed, but in none of the cases was death attributed to AAC or to failure to perform total colectomy in time, but rather to the overall condition of the patient. Table 3 presents the antibiotic regimens that the patients received during the 3 months prior to the toxin test, including the duration and chronological order of administration. Thirty-five percent were treated with cefuroxime, 28% with
D. Raveh et al. / International Journal of Antimicrobial Agents 28 (2006) 231–237
233
Table 1 Background data of all patients with nosocomial diarrhoea (N = 535)
Table 2 Clinical presentation of all patients with nosocomial diarrhoea (N = 535)
Variable
n
%
Characteristic
n
Mean ± S.D. (range) or %
Age (years) Male gender Pregnancy
528 247 7
68 ± 25 (0–100)a 46 1
Daily functioning Nursing home Home
135 400
25 75
Clinical presentation No. of stools per day Fever Temperature (◦ C) Loss of appetite Abdominal pain
134 296 287 38/41 111/131
6 ± 4 (2–20) 55% 38 ± 1 (35.6–40.1) 93% 85%
Stool test (% from tested patients) Positive occult blood Presence of RBCs Presence of WBCs Colonoscopy done
50/109 5/49 5/48 49
46% 10% 10% 9%
Toxin test result Negative Positive More than one sample sent
444 91 104
83% 17% 2.5 ± 1.1 (2–10)
Laboratory data WBC count (×103 /mm3 ) Neutrophils (%) ESR (mm/h) BUN (mg/dL) Creatinine (mg/dL) Albumin (g/dL)
511 336 126 498 499 356
15.1 ± 11.6 (0.6–100) 73 ± 20 (2–98) 71 ± 37 (5–137) 29 ± 30 (2–230) 1.3 ± 1.1 (0.2–8.4) 2.9 ± 0.5 (1.5–4.3)
Medical background Ischaemic heart disease Hypertension Cerebral vascular accident Malignancy Diabetes mellitus Collagen diseases Renal failure Dialysis Surgery in the last 3 months Additional gastrointestinal diseases
235 231 123 100 139 36 61 10 74 218
44 43 23 19 26 7 11 2 14 41
Treatment NSAIDs ACE inhibitors Beta blockers Peptic ulcer disease drugs Diabetes mellitus drugs Anticoagulants Laxatives Calcium channel blockers Antiarrhythmics Digoxin Diuretics
156 148 141 228 102 114 61 123 54 43 182
29 28 26 43 19 21 11 23 10 8 34
26 60 8 10
25 58 8 10
Immunosuppressive treatment (n = 104) Chemotherapy Steroid treatment Chemotherapy + steroids Other Antibiotics in the last 3 months Days without antibiotics Previous AAC No. of months after previous AAC
411 258 28 26
77 1 ± 4 (0–30)a 5 2.7 ± 3.8 (0.1–18)a
NSAIDs, non-steroidal anti-inflammatory drugs; ACE, angiotensinconverting enzyme; AAC, antibiotic-associated colitis. a Mean ± standard deviation (range).
ciprofloxacin, 25% with metronidazole, 25% with aminoglycosides and 20% with ampicillin. Table 3 demonstrates the complexity of antibiotic treatment, as they are usually prescribed in combination with frequent changes. Table 4 shows only the clinically relevant and statistically significant factors detected by bivariate analysis: it compares the characteristics of patients who had a positive toxin result (n = 91) with those of patients with a negative result (n = 444). For those patients who had multiple tests sent, we referred to the first test only. Factors that were not found to be significant in the bivariate analysis were: gender; ethnic origin; admitting ward; existing heart disease; hypertension; past cerebrovascular accident; malignancy; diabetes mellitus; collagen disease; chronic renal failure; chronic dialysis; surgery in the preceding 3 months; other gastrointestinal diseases;
Clinical response No. of days until temperature normalised No. of days until diarrhoea stopped No. of days until abdominal pain stopped No. of days until appetite returned
116
4.7 ± 4.6 (1–30)
198
6 ± 7 (1–40)
29
7 ± 6 (2–30)
14
11 ± 11 (1–35)
RBC, red blood cell; WBC, white blood cell; ESR, erythrocyte sedimentation rate; BUN, blood urea nitrogen.
pregnancy; antibiotic treatment in the preceding 3 months; immunosuppression; time since the previous AAC; fever; number of bowel movements per day; loss of appetite; occult blood in the stool; number of samples sent for Clostridium toxin testing; imaging; order of utilisation of metronidazole, vancomycin and cholestyramine; days to defervescence; resolution of diarrhoea and abdominal pains; rate of appetite return; mortality (although, as previously mentioned, this was never considered to result from AAC); and electrolyte imbalance. The blood count parameters did not differ between the patients with positive or negative stool toxin tests, except for the total number of white blood cells, which differed significantly (P < 0.001) [23]. Biochemical parameters did not differ between patients with positive or negative stool toxin results except for blood urea nitrogen (BUN) (P < 0.01) and serum albumin (P < 0.01). Of all the classes of drugs administered to patients (other than antibiotics), diuretics were associated with a positive toxin test (P < 0.01), with an almost significant association for anticoagulants (P = 0.056) and peptic ulcer drugs (P = 0.065). There was no difference in laxative consumption between the groups. A documented past episode of AAD (P < 0.0001)
234
D. Raveh et al. / International Journal of Antimicrobial Agents 28 (2006) 231–237
Table 3 Antibiotics given to patients during the 3 months before toxin testing Drug
Cefuroxime Ciprofloxacin Metronidazole Aminoglycosides Ampicillin Ceftazidime Vancomycin Co-amoxiclav Piperacillin/tazobactam Nitrofurantoin Imipenem Cefazolin Cefepime Ceftriaxone Chloramphenicol Fluconazole Mezlocillin Co-trimoxazole Amphotericin B Clindamycin Cloxacillin
Order of prescriptiona
%c
1st
2nd
3rd
4th
75 52 45 40 55 18 11 17 4 16 1 9 8 6 3
29 31 29 35 14 30 13 24 14 5 2 7 8 7 7 2 1 3
9 13 8 8 3 14 8 4 3 2 7 2 3 1 3 1 1 1 3 2 1
3 2 2 2
6 2 2 1
1
5th
6th
7th
8th
1 1 2
4 7
2 1 1 1 1 4 1
6 4 1 3
1 1
1 1 1 1
1 1
3
1
1 1
69 52 49 44 35 36 30 21 18 16 21 18 10 9 7 9 5 4 4 3 1
Duration of treatment (days) Mean ± S.D.
Yb 35 28 25 25 20 19 13 13 11 8 7 7 6 5 4 3 3 2 2 2 1
9.4 8.9 11.1 7.6 7.3 6.6 8.6 8.2 10.4 8.3
± ± ± ± ± ± ± ± ± ±
6.6 5.7 8.1 6.6 4.4 3.5 6.1 7.7 8.4 4.1
15.2 7.5 4.3 4.0
± ± ± ±
16.7 0.7 3.2 2.0
7.5 ± 2.1
Min.
Max.
nd
2 1 2 2 2 2 2 2 4 1 25 7 7 3 2
30 21 30 30 21 14 22 30 25 14 45 8 8 6
32 14 13 17 17 15 8 19 5 8 1 5 2 3 3
2 6
9
1 2
45
1
S.D., standard deviation. a Chronological order of prescription of antibiotics: each cell gives the number of patients who received the antibiotics in this order. b No. of cases receiving that antibiotic but where the prescription order was unknown. c Total no. of cases receiving the specific antibiotic (known and unknown order), expressed as percentage of total cases (N = 535). d No. of cases for whom the duration of treatment was known.
and presence of abdominal pain (P = 0.042) were associated with the development of toxin-positive diarrhoea. The order of antibiotic administration, the duration of each administration and the various combinations were too complex to consider in the analysis. We therefore performed two
types of logistic regression analysis: first, the contribution of each drug in a dichotomic mode (whether it was given or not), regardless of duration, order and combination. The results show that none of the antibiotics were associated with toxin-positive diarrhoea, except for clindamycin (P = 0.012),
Table 4 Comparison of Clostridium difficile toxin-positive and -negative diarrhoea patients (mean ± S.D.) Variable Age Residence Home Nursing home Peripheral leukocytes (×103 /mm3 ) Blood urea nitrogen (mg/dL) Serum albumin (g/L) Received diuretics Received clindamycin Days without antibiotics Previous AAC Stool erythrocytes Stool leukocytes >1 sample sent for Clostridium toxin test Abdominal pain Treatment with vancomycin Treatment with metronidazole Combined therapy (metronidazole + vancomycin) Probiotics (additional treatment) Days until diarrhoea stopped Days until abdominal pain stopped
Clostridium toxin-negative (n = 444) 66 ± 26 353 (81%) 82 (19%) 14.4 ± 11.5 29 ± 32 2.9 ± 0.5 139 (32%) 4 (1%) 0.54 ± 2.2 16 (4%) 3 (7%) 3 (7%) 76 (17%) 89 (82%) 27 (6%) 135 (31%) 5 (0.3%) 7 (2%) 5.3 ± 5.1 5.1 ± 3.3
S.D., standard deviation; AAC, antibiotic-associated colitis.
Clostridium toxin-positive (n = 91)
P-value
76 ± 20
<0.0001 <0.05
64 (72%) 25 (28%) 19.1 ± 11.7 32 ± 23 2.7 ± 0.5 42 (50%) 4 (4%) 2.3 ± 5.1 13 (14%) 2 (67%) 2 (67%) 28 (31%) 20 (100%) 28 (31%) 69 (77%) 9 (10%) 7 (8%) 9.7 ± 9.0 11.6 ± 8.6
<0.001 <0.01 <0.01 <0.01 <0.05 <0.05 <0.0001 <0.001 <0.001 <0.005 <0.05 <0.0001 <0.0001 <0.0001 <0.001 <0.0001 <0.05
D. Raveh et al. / International Journal of Antimicrobial Agents 28 (2006) 231–237
despite the rare use of this drug. Ceftriaxone was the next most closely associated (P = 0.086). Second, we assigned each antibiotic a numeric value of days administered, starting from zero (= drug not given). This model included aminoglycosides, ampicillin, amoxicillin/clavulanic acid (coamoxiclav), four generations of cephalosporins, chloramphenicol, clindamycin, imipenem, macrolides, ciprofloxacin [16–18] and piperacillin/tazobactam [19]. This ‘duration’ model showed that only the macrolide group was associated with toxin positivity, with P-values of 0.03–0.05 in various models. A significant association was also found for the presence of erythrocytes and leukocytes in stool samples of toxinpositive patients (P < 0.005) [23]. However, ordering this simple, useful, fast and inexpensive stool test was unfortunately rare, which limits our ability to draw conclusions. There was no association between the results of imaging tests and positive toxin tests. The recovery time for toxin-positive patients was longer than for the toxin-negative group, as measured by the time to resolution of abdominal pain (P < 0.05) and diarrhoea (P < 0.0001). The logistic regression analysis was based on factors found to be significant in the bivariate analysis. We tested various models because several variables had many missing values (e.g. stool erythrocyte and leukocyte tests, etc.). Forward stepwise regression analysis showed that previous AAC was the most significant risk factor for diarrhoea with a positive toxin test (P < 0.001), followed by clindamycin treatment (P < 0.005), diuretics (P < 0.005) and older age (P < 0.05).
4. Discussion In this retrospective study, we sought to determine risk factors associated with the development of toxin-positive hospital-acquired diarrhoea. Clostridium difficile toxin test was determined for 535 hospitalised patients. Despite our laboratory’s demand that samples are sent for toxin testing only from patients with at least six loose stools per day [14], this requirement was not always strictly followed by the referring physicians. Only 134 cases (25%) of the 535 patients included a recorded number of stools passed per day; the maximum number of recorded stools per day reached 20. In several cases there were problems that prevented counting the number of bowel movements, i.e. patients wearing diapers or patients suffering from dementia. Our finding of an association between the age of patients and stool toxin positivity is consistent with earlier data [1,2,5,8]. In our study, the mean ± S.D. age of toxin-negative patients was 66 ± 26 years, whilst toxin-positive patients were almost 10 years older (76 ± 20 years). There is in vitro evidence that colonic contents of younger people inhibit C. difficile more effectively than that of older people [8]. Furthermore, in elderly patients there are more concomitant risk factors, such as longer hospitalisation, more additional
235
diseases and complicated treatments [1,8,24]. Patients who were admitted from other hospitals or nursing homes were at greater risk for development of CDAD than patients who came from home. The percentage of women with hospitalacquired diarrhoea was almost 10% higher than men (54% vs. 46%), but we did not find a gender difference in the toxin positivity rate. Some studies have shown that women are more likely to develop AAD (60%), presumably because they are frequently treated with antibiotics for urinary tract infections [2], whilst others have shown the opposite [1,2,7,20]. Similar to Wistrom et al. [8], we did not find an association between toxin positivity and chronic diseases such as diabetes, renal failure and malignancy, in contrast to other studies [1,2,7,11,24]. We also did not find an association between CDAD and immunosuppressive therapy, and the only drugs found to be associated were diuretics as well as an almost significant association with anticoagulants and antacid drugs. We speculate that reduction in gastric acidity by medications decreases acidic gastric killing and enhances colonisation with C. difficile. We tried to detect surrogate markers for CDAD, its severity and outcome. As in previous studies [1,8,13,14], the white blood cell count was higher in toxin-positive patients compared with toxin-negative ones [23]. Other reported but less frequent factors are dehydration, electrolyte imbalance and hypoalbuminaemia [5,8,13,14]. We did not find additional significantly different laboratory results, except for serum albumin, which was much lower in the toxin-positive group both by bivariate and multivariate analysis. An unanswered question is whether the low albumin level is a risk factor for CDAD, a known marker for very sick patients in general, or a result of the AAD, which also causes protein-losing enteropathy. Comparing renal function tests for the two groups showed higher BUN among patients with a positive toxin test, without a difference in serum creatinine values. This might suggest that dehydration is an important risk factor of CDAD, either due to the diuretic drugs or fluid loss or fluid shift into the third space owing to hypoalbuminaemia. However, we did not find an additional association between toxin positivity and dehydration markers such as hypernatraemia or elevated haematocrit. In order for C. difficile to colonise the gut of a normal individual, the resident normal flora must be altered [3]. Anaerobic gut bacteria are believed to be crucial to ‘colonisation resistance’, but the precise components involved have not been clearly defined. Contradicting evidence derives from the fact that cephalosporins cause more CDAD than piperacillin/tazobactam or carbapenems, despite the fact that the latter agents efficiently kill most of the anaerobic flora of the colon, unlike cephalosporins. There is in vitro evidence that faecal contents of elderly patients are less inhibitory to the growth of C. difficile [8]. The main antibiotics associated with CDAD are secondgeneration and more advanced generation cephalosporins and clindamycin [1,9], and the duration of antimicrobial
236
D. Raveh et al. / International Journal of Antimicrobial Agents 28 (2006) 231–237
therapy is also an important risk factor [5,8]. Macrolides in general, and in particular clarithromycin, have been proposed to be risk factors for the development of CDI. Aminoglycosides appear to have reduced propensity to induce CDI, probably owing to their lack of effect on the endogenous anaerobic gut bacteria. In addition, there has been little evidence linking quinolones with CDAD. Older quinolones (including ciprofloxacin) have poor antianaerobic activity, whereas newer agents such as moxifloxacin and gatifloxacin have significantly increased activity against anaerobes [17,18]. Data from several studies indicate that antipseudomonal penicillins such as piperacillin and ticarcillin appear to share the low-risk status [5,8,19] for CDAD compared with cephalosporins, but the reason for this has not been clearly found. Other perceived low-risk agents include penicillin, trimethoprim, rifampicin, fusidic acid and nitrofurantoin. It is notable that there are very few published reports associating co-amoxiclav with CDAD, despite its good anti-anaerobic activity, widespread utilisation and tendency to cause abdominal cramps and loose stools. In vivo antimicrobial activity and, therefore, probably the risk of CDAD, can be markedly influenced by factors such as drug penetration into the large bowel lumen, specific and non-specific antibiotic binding, gut pH and redox potential [5]. We are not familiar with any statistical model that could take into consideration all these factors and give each its appropriate weight. We applied two types of logistic regression analysis: the dichotomic ‘yes/no’ for each antibiotic; and the continuous number of administration days from zero (= not given). In the first type, only clindamycin was associated with AAC despite the infrequent use of this drug in our hospital setting. Ceftriaxone use followed clindamycin, but with a P-value of 0.086 [16]. In the second analysis, macrolides were the only significant factor associated with toxin positivity. In our medical centre at the time of the study period, oral roxithromycin was used for respiratory infections and oral clarithromycin as a part of the Helicobacter pylori treatment protocol. We found a strong correlation between toxin positivity and the presence of occult blood, white blood cells and red blood cells in stool, as reported by other investigators [5,8,24,25]. However, only in less than 10% of cases was a direct stool smear performed. This quick, simple, informative and inexpensive test can be helpful in the diagnostic work-up of diarrhoea and should be used much more often. Following the course to recovery, we found that toxin-positive patients had significantly longer time until resolution of diarrhoea and abdominal pain than patients with toxin-negative diarrhoea. No difference was found for time until defervescence or for return of appetite. We also found that patients who had a previous history of AAD were at increased risk for development of CDAD. In this study, we observed a tendency of physicians to repeat and send stool samples for toxin testing, sometimes following the initial test, for symptomatic patients. Previous
studies [5] have shown that the toxin test may remain positive in 25% of patients long after recovery from the clinical disease. In addition, there is a possibility of false-negative results. Both facts emphasise the lack of benefit for sending repeat tests in toxin-positive patients. There could be a rationale for repetitive testing in symptomatic toxin-negative patients. One limitation of our study is its retrospective nature. The optimum study would be prospective, screening patients on admission for toxin and C. difficile, and following patients throughout the admission, assessing for acquisition of C. difficile and production of toxin while recording all medications administered. Regarding antibiotic treatments, a comprehensive mathematical model should be constructed to define risk factors for development of hospital-acquired diarrhoea. Another limitation is the lack of a diarrhoea-free control group, although our study looked at ‘hospital-acquired diarrhoea’ and compared patients with toxin-positive diarrhoea with those with toxin-negative diarrhoea. In conclusion, in this study of 535 patients with hospitalacquired diarrhoea, several epidemiological, clinical and laboratory risk factors for development of C. difficile toxinpositive diarrhoea were identified. Targeting these patients for preventive measures may possibly contribute to reduced incidence and associated morbidity.
References [1] Thielman NM, Wilson KM. Antibiotic-associated colitis. In: Mandell GL, Bennett JE, Dolin R, editors. Principles and practice of infectious diseases. 6th ed. Philadelphia, PA: Churchill Livingstone; 2005. p. 1249–63. [2] Klingler PJ, Metzger PP, Seelig HS, Pettit PD, Knudsen JM, Alvarez SA. Clostridium difficile infection: risk factors, medical and surgical management. Dig Dis 2000;18:147–60. [3] Tonna I, Welsby PD. Pathogenesis and treatment of Clostridium difficile infection. Postgrad Med J 2005;81:367–9. [4] Hogenauer C, Hammer HF, Krejs GJ, Reisinger EC. Mechanisms and management of antibiotic-associated diarrhea. Clin Infect Dis 1998;27:702–10. [5] Wilcox MH. Gastrointestinal disorders and the critically ill. Clostridium difficile infection and pseudomembranous colitis. Best Pract Res Clin Gastroenterol 2003;17:475–93. [6] McFarland LV, Mulligan ME, Kwok RY, Stamm WE. Nosocomial acquisition of Clostridium difficile infection. N Engl J Med 1989;320:204–10. [7] McFarland LV, Stamm WE. Review of Clostridium difficileassociated diseases. Am J Infect Control 1986;14:99–109. [8] Wistrom J, Norrby SR, Myrhe EB, et al. Frequency of antibioticassociated diarrhoea in 2642 antibiotic-treated hospitalized patients: a prospective study. J Antimicrob Chemother 2001;47:43–50. [9] Gross MH. Management of antibiotic-associated pseudomembranous colitis. Clin Pharm 1985;4:304–10. [10] Noren T, Akerlund T, Back E, et al. Molecular epidemiology of hospital-associated and community-acquired Clostridium difficile infection in a Swedish county. J Clin Microbiol 2004;42:3635–43. [11] Anand A, Glatt AE. Clostridium difficile infection associated with antineoplastic chemotherapy: a review. Clin Infect Dis 1993;17: 109–13.
D. Raveh et al. / International Journal of Antimicrobial Agents 28 (2006) 231–237 [12] Chang VT, Nelson K. The role of physical proximity in nosocomial diarrhea. Clin Infect Dis 2000;31:717–22. [13] Bulusu M, Narayan S, Shetler K, Triadafilopoulos G. Leukocytosis as a harbinger and surrogate marker of Clostridium difficile infection in hospitalized patients with diarrhea. Am J Gastroenterol 2000;95:3137–41. [14] Jacobs J, Rudensky B, Dresner J, et al. Comparison of four laboratory tests for diagnosis of Clostridium difficile-associated diarrhea. Eur J Clin Microbiol Infect Dis 1996;15:561–6. [15] Arfons L, Ray AJ, Donskey CJ. Clostridium difficile infection among health care workers receiving antibiotic therapy. Clin Infect Dis 2005;40:1384–5. [16] Muto CA, Pokrywka M, Shutt K, et al. A large outbreak of Clostridium difficile-associated disease with an unexpected proportion of deaths and colectomies at a teaching hospital following increased fluoroquinolone use. Infect Control Hosp Epidemiol 2005;26:273–80. [17] Katikireddi V. UK launches inquiry into Clostridium difficile outbreak. Can Med Assoc J 2005;173:138. [18] Gaynes R, Rimland D, Killum E, et al. Outbreak of Clostridium difficile infection in a long-term care facility: association with gatifloxacin use. Clin Infect Dis 2004;38:640–5.
237
[19] Baines SD, Freeman J, Wilcox MHJ. Effects of piperacillin/ tazobactam on Clostridium difficile growth and toxin production in a human gut model. J Antimicrob Chemother 2005;55:974–82. [20] Schwaber MJ, Simhon A, Block C, Roval V, Ferderber N, Shapiro M. Factors associated with nosocomial diarrhea and Clostridium difficile-associated disease on the adult wards of an urban tertiary care hospital. Eur J Clin Microbiol Infect Dis 2000;19:9–15. [21] Martirosian G, Szczesny A, Cohen SH, Silva Jr J. Analysis of Clostridium difficile-associated diarrhea among patients hospitalized in tertiary care academic hospital. Diagn Microbiol Infect Dis 2005;52:153–5. [22] Zabel LT, Worm S. Linezolid contributed to Clostridium difficile colitis with fatal outcome. Infection 2005;33:155–7. [23] Bouza E, Munoz P, Alonso R. Clinical manifestations, treatment and control of infections caused by Clostridium difficile. Clin Microbiol Infect 2005;11(Suppl. 4):57–64. [24] Wilkins TD, Lyerly DM. Clostridium difficile testing: after 20 years, still challenging. J Clin Microbiol 2003;41:531–4. [25] Zuckerman E, Kanel G, Ha C, Kahn J, Gottesman BS, Korula J. Low albumin gradient ascites complicating severe pseudomembranous colitis. Gastroenterology 1997;112:991–4.