- Email: [email protected]
Mortality of patients with antibiotic-associated diarrhoea: the impact of Clostridium difficile
Journal of Hospital Infection (2008) 68, 308e314
Available online at www.sciencedirect.com
www.elsevierhealth.com/journals/jhin
Mortality of patients with antibiotic-associated diarrhoea: the impact of Clostridium difficile J. Bishara a,*, N. Peled b, S. Pitlik a, Z. Samra c a
Infectious Diseases Unit, Rabin Medical Centre, Beilinson Hospital, Petah-Tiqwa and Sackler School of Medicine, Tel-Aviv University, Tel-Aviv, Israel b Institute of Pulmonology, Rabin Medical Centre, Beilinson Hospital, Petah-Tiqwa and Sackler School of Medicine, Tel-Aviv University, Tel-Aviv, Israel c Laboratory of Clinical Microbiology, Rabin Medical Centre, Beilinson Hospital, Petah-Tiqwa and Sackler School of Medicine, Tel-Aviv University, Tel-Aviv, Israel Received 5 December 2006; accepted 30 January 2008 Available online 19 March 2008
KEYWORDS Clostridium difficile; Antibiotic; Diarrhoea; Mortality; Long term
Summary Previous studies have shown conflicting results concerning mortality related to Clostridium difficile infection. The objective of this study was to determine the impact of C. difficile infection on short- and long-term mortality in hospitalised patients with antibiotic-associated diarrhoea. We therefore undertook a prospective caseecontrol study of 217 hospitalised patients who received antibiotics, developed diarrhoea and underwent stool enzyme immunoassay for C. difficile TOX A/B. The KaplaneMeier and the log-rank test were used to determine univariate survival analysis and a Cox regression model for multivariate analysis of 28 day and long-term mortality. Fifty-two (24%) of the 217 patients who met the study criteria were positive for C. difficile TOX A/B. The crude 28 day and long-term mortality rates of the entire cohort were 12.4% and 56%, respectively. On Cox regression analysis, hypoalbuminaemia, impaired functional capacity and elevated serum urea levels were found to be the only independent and statistically significant variables associated with long-term mortality. C. difficile toxin positivity per se was not associated with increased short- or long-term mortality rates. In conclusion, hypoalbuminaemia, renal failure, and impaired function capacity predict mortality due to antibiotic-associated diarrhoea, but C. difficile involvement by itself does not further increase the risk of death in these patients. ª 2008 The Hospital Infection Society. Published by Elsevier Ltd. All rights reserved.
* Corresponding author. Address: Infectious Diseases Unit, Rabin Medical Centre, Beilinson Hospital, Jabotinsky Street, 49100 Petah-Tiwqa, Israel. Tel.: þ972 3 9377511; fax: þ972 3 9377513. E-mail address: [email protected] 0195-6701/$ - see front matter ª 2008 The Hospital Infection Society. Published by Elsevier Ltd. All rights reserved. doi:10.1016/j.jhin.2008.01.033
C. difficile-associated mortality
Introduction Clostridium difficile-associated diarrhoea (CDAD) is the most common type of infectious nosocomial diarrhoea in adults in the developed world.1 C. difficile infection is implicated in 20e30% of all cases of antibiotic-associated diarrhoea, 50e75% of cases of antibiotic-associated colitis and more than 90% of cases of antibiotic-associated pseudomembranous colitis in hospitalised patients.2 The reported mortality rates associated with C. difficile diarrhoea in observational and caseecontrol studies vary from 0.6% to 83%.3e7 The objective of this prospective non-interventional study was to determine the impact of C. difficile infection on short- and long-term mortality in two hospital patient cohorts; one consisting of patients who received antibiotics and developed CDAD, and the other consisting of patients who received antibiotics and developed diarrhoea unrelated to C. difficile.
Methods Patients All patients with diarrhoea, hospitalised in our centre from October 15, 1999 to January 15, 2000 and whose stool samples were collected for C. difficile toxin assay in the Microbiology Laboratory, were identified and followed prospectively. Only patients who had received antibiotics within 40 days prior to the diarrhoeal episode were included. Diarrhoea was defined as the passage of three or more unformed stools for at least two days.8 CDAD was defined as diarrhoea unattributable to any other cause and associated with a positive stool TOX A/B test. Stool Tox A/B immunoassay was repeated in those patients with antibiotic-associated diarrhoea who had negative results from their first stool samples. Patients were classified as non-CDAD on the basis of two negative stool tests. Stool samples were tested either on the day of collection or, if delayed, they were always stored at 4 C, overnight prior to testing next day. The decision to test for C. difficile was made by physicians uninvolved in the study. Each patient was interviewed by one of the investigators using a questionnaire that included demographics, underlying conditions, previous drug and antibiotic therapy, clinical and laboratory findings and outcome. Functional capacity was graded as follows: class 1, patient is independent; class 2, patient requires assistance in daily activities; class 3, patient is bedridden. Duration of follow-up, expressed in days, was defined as the
309 interval from the first day of hospitalisation to the day of death (as recorded in the Central Population Registry) or, if alive, up to 1 December, 2004. Death of an Israeli resident is reported and registered in the Central Population Registry within 48 h.
Toxin assay The stools were analysed with the TOX A/B enzyme immunoassay kit (TechLab, Inc., Blacksburg, VA, USA), according to the manufacturer’s instructions. In a previous study by our group, the TOX A/B test was found to have a sensitivity of 95.3% and specificity of 100% compared to the cytotoxic assay, and a correlation rate of 99.4%.9
Statistical analysis Patients with a positive or negative A/B TOX test were compared for survival rates. Means, standard deviations and P-values (analysis of variance and Chi-squared tests) were calculated. The Chisquared test was used to test contingency table statistical significance. The KaplaneMeier analysis was used for univariate survival analysis, and the log-rank test to determine the statistical significance of differences between survival times; the Cox regression model was used for multivariate analysis of survival. Data were handled and analysed using the Statistical Package for Social Sciences software (SPSS Corporation-11).
Results A total of 217 patients met the inclusion criteria, of whom 52 (24%) were found to have CDAD. The demographic, clinical and laboratory characteristics of the entire cohort and by subgroups with positive or negative stool assays are shown in Table I. The patients with CDAD were significantly older than those without CDAD and had a worse functional capacity on admission to hospital in addition to more comorbidities, such as diabetes mellitus, congestive heart failure and decubitus ulcers. There were no differences between the two groups in the rates of solid and haematological malignancies (Table I). Although all patients had received antibiotics within 40 days prior to the diarrhoeal episode according to their medical history, antibiotic identity was unknown in 58% of cases. The rates of use of the most common antibiotics in the toxin-positive vs the toxin-negative group
310
J. Bishara et al.
Table I
Characteristics of patients with antibiotic-associated diarrhoea (%)
Variable
All (N ¼ 217)
Toxin positive (N ¼ 52)
Toxin negative (N ¼ 165)
P-value 0.0214 0.412 0.255
Age (years)a Male/female, ratio Nursing home
71.5, 18e94 116/101, 1.14 (17.5)
73.5, 32e93 29/23, 1.26 (14)
71, 18e94 87/78, 1.11 (18.8)
Functional capacityb Class 1 Class 2 Class 3
109 (50) 49 (23) 51 (24)
16 (31) 20 (38) 15 (29)
93 (56) 29 (18) 36 (22)
Underlying morbidities Haematological malignancy Solid tumour Diabetes mellitus Congestive heart failure Chronic renal failure Decubitus ulcer
12 40 48 27 27 17
(5.5) (18) (22) (12) (12) (13)
3 (6) 13 (25) 20 (38) 11 (21) 8 (15) 10 (19)
9 (5) 27 (16) 28 (17) 16 (10) 19 (12) 7 (4)
0.556 0.108 0.004 0.037 0.369 0.001
Risk factors Previous hospitalisation Nasogastric tube Laxatives H2-blocker Proton-pump inhibitor Corticosteroids Cytotoxic agents Previous surgery
63 18 15 67 22 37 21 29
(29) (8) (7) (31) (10) (17) (10) (13)
29 (56) 10 (19) 9 (17) 22 (42) 11 (21) 9 (17) 5 (10) 6 (12)
34 (20) 23 (14) 6 (4) 45 (27) 11 (7) 28 (17) 16 (1) 23 (14)
0.0001 0.235 0.002 0.033 0.005 0.55 0.619 0.449
Clinical features Fever Watery diarrhoea Occult blood WBC (103 cell/mm3) Urea (mg/dL) Creatinine (mg/dL) Albumin (mg/dL) Follow-upa (months)
119 (55) 122 (56) 18 (8) 10.5, 0.4e79 35, 4e389 1, 0.4e7.3 3.1, 1.6e4.2 30.9, 0e66.2
31 (60) 42 (81) 3 (6) 16.6, 3.5e42 37, 9e192 1.1, 0.5e5.9 2.5, 1.6e3.8 24.58, 0e66.2
88 (53) 80 (48) 15 (9) 9.2, 0.4e79.0 34, 4e389 0.9, 0.4e7.3 3.2, 1.8e4.2 55.4, 0e65.6
0.307 0.0001 0.342 0.001 0.723 0.552 0.0631 0.0004
Outcome Length of staya (days) 28 day mortality Death (31 Dec 2004)
8.9, 1e38 27 (12.4) 121 (56)
9.5, 2e31 8 (15) 31 (60)
6, 1e38 19 (11.5) 90 (55)
0.1568 0.261 0.316
0.0105
WBC, white blood cells. Data are no. (%) of patients unless otherwise indicated. a Median (range). b Functional capacity: class 1, independent; class 2, needs help in daily activity; class 3, bedridden.
were as follows: penicillins (25% vs 17%); clindamycin (21% vs 14%); cephalosporins (15% vs 5.5%); quinolones (12% vs 9%); and aminoglycosides (2% vs 7%) (P ¼ 0.013). As expected, hospitalisation during the previous months was significantly more common in the CDAD group, as were previous laxatives, H2-blockers and proton-pump inhibitors (Table I). Clinical features found to be significantly more common in the CDAD group were leucocytosis and watery, non-bloody diarrhoea. In addition, in the CDAD group only, during the 28 day follow-up
period, 7.6% (4/52) patients relapsed, 3.8% (2/ 52) developed toxic megacolon and 1.9% (1/52) patient underwent colectomy.
Twenty-eight day mortality Twenty-seven patients (12.4%) died by day 28, including eight (15%) in the toxin-positive group and 19 (11.5%) in the toxin-negative group (P ¼ 0.261; Table I). The variables associated with
C. difficile-associated mortality
311
increased 28 day mortality on univariate analysis were renal failure, leucocytosis, hypoalbuminaemia and occult blood in the stool. It is of interest that age, solid tumour and CDAD per se had no significant effect on the 28 day mortality rate (Table II). On multivariate analysis, the only variable that was significantly associated with higher 28 day mortality was elevated serum urea level (B ¼ 0.0132; Wald ¼ 32.2521; ExpB ¼ 1.013; 95% confidence interval (CI): 1.006e1.013; P < 0.000). Occult blood in the stool was associated with lower shortterm mortality (B ¼ 1.127; Wald ¼ 4.6496; ExpB ¼ 0.3238; 95% CI: 0.1162e0.9024; P ¼ 0.0311).
Long-term mortality During the entire study period, 121 (56%) patients died and 96 survived. Of the 121 patients (27
Table II
Univariate analysis for 28 day mortality
Variable Age (years)a Male Albumin (mg/dL)a Creatinine (mg/dL)a Urea (mg/dL)a WBC (103 cell/mm3)a Functional capacityb 2 Nursing home Decubitus ulcer Post-surgery Solid tumour Diabetes mellitus Nasogastric tube Proton-pump inhibitor Laxatives Corticosteroids Occult blood Follow-up (months)a Positive C. difficile toxin
patients died at 28 days and 94 thereafter), 31 (60%) and 90 (55%) patients in the toxin-positive and toxin-negative groups, respectively (P ¼ 0.316; Table I). On univariate analysis, the variables significantly associated with increased overall longterm mortality were older age, elevated serum urea and creatinine levels, leucocytosis, hypoalbuminaemia, impaired functional capacity and presence of a nasogastric tube (Table III). On Cox regression analysis, hypoalbuminaemia, impaired functional capacity and elevated serum urea levels were found to be the only independent variables significantly associated with crude long-term mortality (Table IV). As before, positive stool C. difficile toxin assay was not associated with a higher crude long-term mortality rate (Table IV, Figure 1).
Discussion We evaluated the short- and long-term mortality rates in a cohort of hospitalised patients with
Died (N ¼ 27)
Survived (N ¼ 190)
P
70.8 17.0 15 (58) 2.76 0.83
66.5 17.1 101 (53) 3.08 0.65
0.231 0.402 0.026
Table III
2.02 1.69
1.19 0.90
0.000
94.2 96.5 17.54 11.5
43.9 36.8 12.6 10.7
0.000 0.033
16 (52)
84 (46)
0.139
5 2 1 8 4
33 15 28 32 44
(17) (8) (15) (17) (24)
0.493 0.665 0.105 0.081 0.224
6 (23)
27 (14)
0.184
5 (19)
17 (9)
0.103
1 (4) 5 (20) 5 (19) 0.4 0.3
14 (7.4) 32 (17) 13 (7) 38.0 26.4
0.438 0.441 0.048 0.000
8 (31)
44 (23)
0.261
Age (years)a Male Albumin (mg/dL)a Creatinine (mg/dL)a Urea (mg/dL)a WBC (103 cell/mm3)a Functional capacityb 2 Nursing home Decubitus ulcer Post-surgery Solid tumour Diabetes mellitus Nasogastric tube Proton-pump inhibitor Laxatives Corticosteroids Occult blood Follow-up (months)a Positive C. difficile toxin
(19) (8) (4) (31) (15)
WBC, white blood cells. Data are no. (%) of patients unless otherwise indicated. a Mean SD. b Functional capacity 2: class 2, needs help in daily activity; class 3, bedridden.
Univariate analysis for long-term mortality
Variable
Died (N ¼ 121)
Survived (N ¼ 96)
P
70.4 15.6 67 (55) 2.87 0.63 1.43 1.24
62.9 1.9 49 (51) 3.25 0.68 1.12 0.07
0.001 0.309 0.000 0.034
60.3 59.7 14.5 12.7
36.5 29.5 0.000 11.5 7.7 0.039
71 (61)
29 (32)
0.0001
26 12 13 27 28 24 14
12 (13) 5 (5.2) 16 (17) 13 (14) 20 (22) 9 (10) 8 (8)
0.059 0.148 0.142 0.061 0.439 0.026 0.29
9 (8) 23 (19) 11 (9) 13.2 17.6
6 (6) 14 (15) 7 (7) 59.1 12.2
0.468 0.243 0.413 0.000
31 (26)
21 (22)
0.316
(22) (10) (11) (23) (24) (73) (12)
WBC, white blood cells. Data are no. (%) of patients, unless otherwise indicated. a Mean SD. b Functional capacity 2: class 2, needs help in daily activity; class 3, bedridden.
312 Table IV
J. Bishara et al. Cox regression model for long-term mortality (N ¼ 187 patients, P ¼ 0.0000)
Albumin Functional capacity Urea
B
Exp (B)
Wald
95% CI
P
0.6438 0.3304 1.0085
0.5253 1.3915 1.0085
16.9564 8.3952 1.0122
0.424e0.753 1.080e1.680 1.003e1.010
0.0000 0.0038 0.0000
CI, confidence interval.
antibiotic-associated diarrhoea with and without proven C. difficile aetiology. The most interesting finding of this prospective caseecontrol study was that C. difficile infection was not associated with higher overall 28 day or long-term mortality rates in hospitalised patients with antibiotic-associated diarrhoea. Although both the crude 28 day and long-term mortality rates in our entire study population were high (12.4% and 56%, respectively, Table I), there were no significant differences between patients who were positive or negative for C. difficile toxin. Previous studies have shown conflicting results concerning mortality attributable to C. difficile infection. In 1989, Eriksson and Aronsson reported a three-fold higher mortality rate in patients with CDAD than control patients matched for age, sex, and underlying diseases.10In contrast, MacGowan et al. noted no differences between patients with CDAD and controls well-matched
for infective diagnosis, severity of illness, and coexistent diseases.11 In a prospective study, Kyne et al. reported that CDAD was associated with an increase in estimated total hospital costs and length of hospital stay.3 Although the crude mortality rate was higher for patients with CDAD than for patients without CDAD, the difference was related to other variables. The authors concluded that CDAD was not associated with excess 3 month or 1 year mortality after adjustment for age, comorbidity and disease severity. It is noteworthy that asymptomatic patients were included as a control group in this study.3 Our findings add validity to the observations made by these investigators. Conversely, Changela et al., in a retrospective caseecontrol study, demonstrated that patients with CDAD had a higher mortality rate than patients without diarrhoea, matched for age and time of hospitalisation.12
1.1 1.0
Cumulative survival (%)
0.9 0.8 0.7 0.6 TOXIN 0.5
1.00 1.00-censored
0.4
0.00 0.00-censored
0.3 0
10
20
30 40 50 Follow-up (months)
60
70
Figure 1 KaplaneMeier curves for long-term mortality. Pink line: patients with positive C. difficile toxin stool assay; green line: patients with negative C. difficile toxin stool assay. x-Axis shows duration of follow-up expressed in months since hospitalisation to death; y-axis shows percentages of survivors (log-rank P ¼ 0.1766).
C. difficile-associated mortality Pepin et al. compared mortality and total length of hospital stay among inpatients in whom nosocomial CDAD developed and among control subjects without CDAD matched for sex, age, and disease severity.13 One year after diagnosis of CDAD, mortality rates of patients with CDAD were significantly higher than those in the control group. The attributable mortality was measured during a CDAD epidemic in Quebec, Canada, caused by a hypervirulent toxinotype III ribotype 027 strain, which produced levels of toxins A and B 16e23 times higher than those of contemporary toxinotype 0 strains. Therefore, these results cannot be extrapolated to centres outside Quebec, where the strain might be absent or uncommon. However, the same strain has been found in several US states and more recently in the UK and the Netherlands.13 Whether a similar ribotype 027 strain is circulating in our hospital is unknown at present. Although many previous studies have analysed prognostic risk factors and associated mortality in patients with CDAD, our study differs in several ways. First, none of the earlier studies compared patients with CDAD to patients with antibioticassociated diarrhoea not associated with C. difficile infection. Second, the methodology of these studies was different, i.e. either retrospective,6,12e15 lacking a control group,14,16,17 or asymptomatic patients were used as controls.3,12 The strongest significant and independent predictor of short- and long-term mortality in our study was hypoalbuminaemia, in concurrence with previous studies.2,6,15,18 Other significant predictors were renal failure, noted also by others, and impaired functional capacity on admission, regardless of age.7 The lack of a significant effect of age has been reported previously.2,15 In addition, several researchers reported functional impairment to be a predictor of increased mortality in patients with CDAD, but not in patients with antibioticassociated diarrhoea, as we found.16,19 Our study had certain limitations. Data on diarrhoeal disease severity, scores of disease severity, such as the Acute Physiological Assessment and Chronic Health Evaluation (APACHE) II, and antimicrobial therapy for C. difficile were not included, and their impact on outcome was not evaluated. In conclusion, antibiotic-associated diarrhoea is associated with high rates of short- and long-term mortality. Hypoalbuminaemia, renal failure and impaired functional capacity predict mortality. Although C. difficile involvement itself does not impact on mortality in these patients, our results should not underestimate the serious
313 consequences of C. difficile infection on morbidity, cost and length of hospitalisation.
References 1. Dharmarajan TS, Sipalay M, Shyamsundar R, Norkus EP, Pitchumoni CS. Co-morbidity, not age, predicts adverse outcome in Clostridium difficile colitis. World J Gastoenterol 2000;6:198e201. 2. Kelly CP, Pothoulakis C, LaMont JT. Clostridium difficile colitis. N Engl J Med 1994;330:257e262. 3. Kyne L, Hamel MB, Polavaram R, Kelly CP. Health care costs and mortality associated with nosocomial diarrhea due to Clostridium difficile. Clin Infect Dis 2002;34: 346e353. 4. Olson MM, Shanholtzer CJ, Lee Jr JT, Gerding DN. Ten years of prospective Clostridium difficile-associated disease surveillance and treatment at the Minneapolis VA Medical Center, 1982e1991. Infect Control Hosp Epidemiol 1994;15: 371e381. 5. Lesna M, Parham DM. Risk of diarrhoea due to Clostridium difficile during cefotaxime treatment; mortality due to C. difficile colitis in elderly people has been underestimated. Br Med J 1996;312:778. 6. Ramaswamy R, Grover H, Corpuz M, Daniels P, Pitchumoni CS. Prognostic criteria in Clostridium difficile colitis. Am J Gastroenterol 1996;91:460e464. 7. Rubin MS, Bodenstein LE, Kent KC. Severe Clostridium difficile colitis. Dis Colon Rectum 1995;38:350e354. 8. Kyne L, Warny M, Qamar A, Kelly C. Asymptomatic carriage of Clostridium difficile and serum levels of IgG antibody against toxin A. N Engl J Med 2000;342:390e397. 9. Samra Z, Talmor S, Bahar J. High prevalence of toxin A-negative toxin B-positive Clostridium difficile in hospitalised patients with gastrointestinal diseases. Diagn Microbiol Infect Dis 2002;43:189e192. 10. Eriksson S, Aronsson B. Medical implications of nosocomial infection with Clostridium difficile. Scand J Infect Dis 1989;21:733e744. 11. McGowan AP, Feeney R, Brown I, McCulloch SY, Reeves DS, Lovering AM. Health care resources utilization and antimicrobial use in elderly patients with community-acquired lower respiratory tract infection who develop Clostridium difficile-associated diarrhea. J Antimicrob Chemother 1997;39:537e541. 12. Changela U, Cannon JP, Aniziokoro C, Shah PS, Thottapurathu L, Lentino J. Risk factors and mortality associated with Clostridium difficile-associated diarrhea at a VA hospital. Int J Antimicrob Agents 2004;24:562e566. 13. Pe ´pin J, Valiquette L, Cossette B. Mortality attributable to nosocomial Clostridium difficile-associated disease during an epidemic caused by a hypervirulent strain in Quebec. Can Med Assoc J 2005;173:1037e1042. 14. Morris AM, Jobe BA, Stoney M, Sheppard BC, Deveney CW, Deveney KE. Clostridium difficile colitis. An increasingly aggressive iatrogenic disease? Arch Surg 2002;137:1096e 1100. 15. Dharmarajan TS, Patel B, Sipalay M, Norkus EP. Clostridium difficile colitis in older adults in long-term facilities and the community: do their outcomes differ? J Am Med Dir Assoc 2000;2:58e61. 16. Kyne L, Merry C, O’Connell B, Kelly A, Keane C, O’Neill D. Factors associated with prolonged symptoms and severe disease due to Clostridium difficile. Age Ageing 1999;28: 107e113.
314 17. Miller M, Hyland M, Ofner-Agostini M, Gourdeau M, Ishak M. Morbidity, mortality, and healthcare burden of nosocomial Clostridium difficile-associated diarrhea in Canadian hospitals. Infect Control Hosp Epidemiol 2002;23:137e140. 18. Nair S, Yadav D, Corpuz M, Pitchomoni CS. Clostridium difficile colitis: factors influencing treatment failure and
J. Bishara et al. relapse, a prospective evaluation. Am J Gastroenterol 1998;93:1873e1876. 19. Anand A, Bashey B, Mir T, Glatt A. Epidemiology, clinical manifestations and outcome of Clostridium difficile-associated diarrhoea. Am J Gastroenterol 1994;89: 519e523.
Available online at www.sciencedirect.com
www.elsevierhealth.com/journals/jhin
Mortality of patients with antibiotic-associated diarrhoea: the impact of Clostridium difficile J. Bishara a,*, N. Peled b, S. Pitlik a, Z. Samra c a
Infectious Diseases Unit, Rabin Medical Centre, Beilinson Hospital, Petah-Tiqwa and Sackler School of Medicine, Tel-Aviv University, Tel-Aviv, Israel b Institute of Pulmonology, Rabin Medical Centre, Beilinson Hospital, Petah-Tiqwa and Sackler School of Medicine, Tel-Aviv University, Tel-Aviv, Israel c Laboratory of Clinical Microbiology, Rabin Medical Centre, Beilinson Hospital, Petah-Tiqwa and Sackler School of Medicine, Tel-Aviv University, Tel-Aviv, Israel Received 5 December 2006; accepted 30 January 2008 Available online 19 March 2008
KEYWORDS Clostridium difficile; Antibiotic; Diarrhoea; Mortality; Long term
Summary Previous studies have shown conflicting results concerning mortality related to Clostridium difficile infection. The objective of this study was to determine the impact of C. difficile infection on short- and long-term mortality in hospitalised patients with antibiotic-associated diarrhoea. We therefore undertook a prospective caseecontrol study of 217 hospitalised patients who received antibiotics, developed diarrhoea and underwent stool enzyme immunoassay for C. difficile TOX A/B. The KaplaneMeier and the log-rank test were used to determine univariate survival analysis and a Cox regression model for multivariate analysis of 28 day and long-term mortality. Fifty-two (24%) of the 217 patients who met the study criteria were positive for C. difficile TOX A/B. The crude 28 day and long-term mortality rates of the entire cohort were 12.4% and 56%, respectively. On Cox regression analysis, hypoalbuminaemia, impaired functional capacity and elevated serum urea levels were found to be the only independent and statistically significant variables associated with long-term mortality. C. difficile toxin positivity per se was not associated with increased short- or long-term mortality rates. In conclusion, hypoalbuminaemia, renal failure, and impaired function capacity predict mortality due to antibiotic-associated diarrhoea, but C. difficile involvement by itself does not further increase the risk of death in these patients. ª 2008 The Hospital Infection Society. Published by Elsevier Ltd. All rights reserved.
* Corresponding author. Address: Infectious Diseases Unit, Rabin Medical Centre, Beilinson Hospital, Jabotinsky Street, 49100 Petah-Tiwqa, Israel. Tel.: þ972 3 9377511; fax: þ972 3 9377513. E-mail address: [email protected] 0195-6701/$ - see front matter ª 2008 The Hospital Infection Society. Published by Elsevier Ltd. All rights reserved. doi:10.1016/j.jhin.2008.01.033
C. difficile-associated mortality
Introduction Clostridium difficile-associated diarrhoea (CDAD) is the most common type of infectious nosocomial diarrhoea in adults in the developed world.1 C. difficile infection is implicated in 20e30% of all cases of antibiotic-associated diarrhoea, 50e75% of cases of antibiotic-associated colitis and more than 90% of cases of antibiotic-associated pseudomembranous colitis in hospitalised patients.2 The reported mortality rates associated with C. difficile diarrhoea in observational and caseecontrol studies vary from 0.6% to 83%.3e7 The objective of this prospective non-interventional study was to determine the impact of C. difficile infection on short- and long-term mortality in two hospital patient cohorts; one consisting of patients who received antibiotics and developed CDAD, and the other consisting of patients who received antibiotics and developed diarrhoea unrelated to C. difficile.
Methods Patients All patients with diarrhoea, hospitalised in our centre from October 15, 1999 to January 15, 2000 and whose stool samples were collected for C. difficile toxin assay in the Microbiology Laboratory, were identified and followed prospectively. Only patients who had received antibiotics within 40 days prior to the diarrhoeal episode were included. Diarrhoea was defined as the passage of three or more unformed stools for at least two days.8 CDAD was defined as diarrhoea unattributable to any other cause and associated with a positive stool TOX A/B test. Stool Tox A/B immunoassay was repeated in those patients with antibiotic-associated diarrhoea who had negative results from their first stool samples. Patients were classified as non-CDAD on the basis of two negative stool tests. Stool samples were tested either on the day of collection or, if delayed, they were always stored at 4 C, overnight prior to testing next day. The decision to test for C. difficile was made by physicians uninvolved in the study. Each patient was interviewed by one of the investigators using a questionnaire that included demographics, underlying conditions, previous drug and antibiotic therapy, clinical and laboratory findings and outcome. Functional capacity was graded as follows: class 1, patient is independent; class 2, patient requires assistance in daily activities; class 3, patient is bedridden. Duration of follow-up, expressed in days, was defined as the
309 interval from the first day of hospitalisation to the day of death (as recorded in the Central Population Registry) or, if alive, up to 1 December, 2004. Death of an Israeli resident is reported and registered in the Central Population Registry within 48 h.
Toxin assay The stools were analysed with the TOX A/B enzyme immunoassay kit (TechLab, Inc., Blacksburg, VA, USA), according to the manufacturer’s instructions. In a previous study by our group, the TOX A/B test was found to have a sensitivity of 95.3% and specificity of 100% compared to the cytotoxic assay, and a correlation rate of 99.4%.9
Statistical analysis Patients with a positive or negative A/B TOX test were compared for survival rates. Means, standard deviations and P-values (analysis of variance and Chi-squared tests) were calculated. The Chisquared test was used to test contingency table statistical significance. The KaplaneMeier analysis was used for univariate survival analysis, and the log-rank test to determine the statistical significance of differences between survival times; the Cox regression model was used for multivariate analysis of survival. Data were handled and analysed using the Statistical Package for Social Sciences software (SPSS Corporation-11).
Results A total of 217 patients met the inclusion criteria, of whom 52 (24%) were found to have CDAD. The demographic, clinical and laboratory characteristics of the entire cohort and by subgroups with positive or negative stool assays are shown in Table I. The patients with CDAD were significantly older than those without CDAD and had a worse functional capacity on admission to hospital in addition to more comorbidities, such as diabetes mellitus, congestive heart failure and decubitus ulcers. There were no differences between the two groups in the rates of solid and haematological malignancies (Table I). Although all patients had received antibiotics within 40 days prior to the diarrhoeal episode according to their medical history, antibiotic identity was unknown in 58% of cases. The rates of use of the most common antibiotics in the toxin-positive vs the toxin-negative group
310
J. Bishara et al.
Table I
Characteristics of patients with antibiotic-associated diarrhoea (%)
Variable
All (N ¼ 217)
Toxin positive (N ¼ 52)
Toxin negative (N ¼ 165)
P-value 0.0214 0.412 0.255
Age (years)a Male/female, ratio Nursing home
71.5, 18e94 116/101, 1.14 (17.5)
73.5, 32e93 29/23, 1.26 (14)
71, 18e94 87/78, 1.11 (18.8)
Functional capacityb Class 1 Class 2 Class 3
109 (50) 49 (23) 51 (24)
16 (31) 20 (38) 15 (29)
93 (56) 29 (18) 36 (22)
Underlying morbidities Haematological malignancy Solid tumour Diabetes mellitus Congestive heart failure Chronic renal failure Decubitus ulcer
12 40 48 27 27 17
(5.5) (18) (22) (12) (12) (13)
3 (6) 13 (25) 20 (38) 11 (21) 8 (15) 10 (19)
9 (5) 27 (16) 28 (17) 16 (10) 19 (12) 7 (4)
0.556 0.108 0.004 0.037 0.369 0.001
Risk factors Previous hospitalisation Nasogastric tube Laxatives H2-blocker Proton-pump inhibitor Corticosteroids Cytotoxic agents Previous surgery
63 18 15 67 22 37 21 29
(29) (8) (7) (31) (10) (17) (10) (13)
29 (56) 10 (19) 9 (17) 22 (42) 11 (21) 9 (17) 5 (10) 6 (12)
34 (20) 23 (14) 6 (4) 45 (27) 11 (7) 28 (17) 16 (1) 23 (14)
0.0001 0.235 0.002 0.033 0.005 0.55 0.619 0.449
Clinical features Fever Watery diarrhoea Occult blood WBC (103 cell/mm3) Urea (mg/dL) Creatinine (mg/dL) Albumin (mg/dL) Follow-upa (months)
119 (55) 122 (56) 18 (8) 10.5, 0.4e79 35, 4e389 1, 0.4e7.3 3.1, 1.6e4.2 30.9, 0e66.2
31 (60) 42 (81) 3 (6) 16.6, 3.5e42 37, 9e192 1.1, 0.5e5.9 2.5, 1.6e3.8 24.58, 0e66.2
88 (53) 80 (48) 15 (9) 9.2, 0.4e79.0 34, 4e389 0.9, 0.4e7.3 3.2, 1.8e4.2 55.4, 0e65.6
0.307 0.0001 0.342 0.001 0.723 0.552 0.0631 0.0004
Outcome Length of staya (days) 28 day mortality Death (31 Dec 2004)
8.9, 1e38 27 (12.4) 121 (56)
9.5, 2e31 8 (15) 31 (60)
6, 1e38 19 (11.5) 90 (55)
0.1568 0.261 0.316
0.0105
WBC, white blood cells. Data are no. (%) of patients unless otherwise indicated. a Median (range). b Functional capacity: class 1, independent; class 2, needs help in daily activity; class 3, bedridden.
were as follows: penicillins (25% vs 17%); clindamycin (21% vs 14%); cephalosporins (15% vs 5.5%); quinolones (12% vs 9%); and aminoglycosides (2% vs 7%) (P ¼ 0.013). As expected, hospitalisation during the previous months was significantly more common in the CDAD group, as were previous laxatives, H2-blockers and proton-pump inhibitors (Table I). Clinical features found to be significantly more common in the CDAD group were leucocytosis and watery, non-bloody diarrhoea. In addition, in the CDAD group only, during the 28 day follow-up
period, 7.6% (4/52) patients relapsed, 3.8% (2/ 52) developed toxic megacolon and 1.9% (1/52) patient underwent colectomy.
Twenty-eight day mortality Twenty-seven patients (12.4%) died by day 28, including eight (15%) in the toxin-positive group and 19 (11.5%) in the toxin-negative group (P ¼ 0.261; Table I). The variables associated with
C. difficile-associated mortality
311
increased 28 day mortality on univariate analysis were renal failure, leucocytosis, hypoalbuminaemia and occult blood in the stool. It is of interest that age, solid tumour and CDAD per se had no significant effect on the 28 day mortality rate (Table II). On multivariate analysis, the only variable that was significantly associated with higher 28 day mortality was elevated serum urea level (B ¼ 0.0132; Wald ¼ 32.2521; ExpB ¼ 1.013; 95% confidence interval (CI): 1.006e1.013; P < 0.000). Occult blood in the stool was associated with lower shortterm mortality (B ¼ 1.127; Wald ¼ 4.6496; ExpB ¼ 0.3238; 95% CI: 0.1162e0.9024; P ¼ 0.0311).
Long-term mortality During the entire study period, 121 (56%) patients died and 96 survived. Of the 121 patients (27
Table II
Univariate analysis for 28 day mortality
Variable Age (years)a Male Albumin (mg/dL)a Creatinine (mg/dL)a Urea (mg/dL)a WBC (103 cell/mm3)a Functional capacityb 2 Nursing home Decubitus ulcer Post-surgery Solid tumour Diabetes mellitus Nasogastric tube Proton-pump inhibitor Laxatives Corticosteroids Occult blood Follow-up (months)a Positive C. difficile toxin
patients died at 28 days and 94 thereafter), 31 (60%) and 90 (55%) patients in the toxin-positive and toxin-negative groups, respectively (P ¼ 0.316; Table I). On univariate analysis, the variables significantly associated with increased overall longterm mortality were older age, elevated serum urea and creatinine levels, leucocytosis, hypoalbuminaemia, impaired functional capacity and presence of a nasogastric tube (Table III). On Cox regression analysis, hypoalbuminaemia, impaired functional capacity and elevated serum urea levels were found to be the only independent variables significantly associated with crude long-term mortality (Table IV). As before, positive stool C. difficile toxin assay was not associated with a higher crude long-term mortality rate (Table IV, Figure 1).
Discussion We evaluated the short- and long-term mortality rates in a cohort of hospitalised patients with
Died (N ¼ 27)
Survived (N ¼ 190)
P
70.8 17.0 15 (58) 2.76 0.83
66.5 17.1 101 (53) 3.08 0.65
0.231 0.402 0.026
Table III
2.02 1.69
1.19 0.90
0.000
94.2 96.5 17.54 11.5
43.9 36.8 12.6 10.7
0.000 0.033
16 (52)
84 (46)
0.139
5 2 1 8 4
33 15 28 32 44
(17) (8) (15) (17) (24)
0.493 0.665 0.105 0.081 0.224
6 (23)
27 (14)
0.184
5 (19)
17 (9)
0.103
1 (4) 5 (20) 5 (19) 0.4 0.3
14 (7.4) 32 (17) 13 (7) 38.0 26.4
0.438 0.441 0.048 0.000
8 (31)
44 (23)
0.261
Age (years)a Male Albumin (mg/dL)a Creatinine (mg/dL)a Urea (mg/dL)a WBC (103 cell/mm3)a Functional capacityb 2 Nursing home Decubitus ulcer Post-surgery Solid tumour Diabetes mellitus Nasogastric tube Proton-pump inhibitor Laxatives Corticosteroids Occult blood Follow-up (months)a Positive C. difficile toxin
(19) (8) (4) (31) (15)
WBC, white blood cells. Data are no. (%) of patients unless otherwise indicated. a Mean SD. b Functional capacity 2: class 2, needs help in daily activity; class 3, bedridden.
Univariate analysis for long-term mortality
Variable
Died (N ¼ 121)
Survived (N ¼ 96)
P
70.4 15.6 67 (55) 2.87 0.63 1.43 1.24
62.9 1.9 49 (51) 3.25 0.68 1.12 0.07
0.001 0.309 0.000 0.034
60.3 59.7 14.5 12.7
36.5 29.5 0.000 11.5 7.7 0.039
71 (61)
29 (32)
0.0001
26 12 13 27 28 24 14
12 (13) 5 (5.2) 16 (17) 13 (14) 20 (22) 9 (10) 8 (8)
0.059 0.148 0.142 0.061 0.439 0.026 0.29
9 (8) 23 (19) 11 (9) 13.2 17.6
6 (6) 14 (15) 7 (7) 59.1 12.2
0.468 0.243 0.413 0.000
31 (26)
21 (22)
0.316
(22) (10) (11) (23) (24) (73) (12)
WBC, white blood cells. Data are no. (%) of patients, unless otherwise indicated. a Mean SD. b Functional capacity 2: class 2, needs help in daily activity; class 3, bedridden.
312 Table IV
J. Bishara et al. Cox regression model for long-term mortality (N ¼ 187 patients, P ¼ 0.0000)
Albumin Functional capacity Urea
B
Exp (B)
Wald
95% CI
P
0.6438 0.3304 1.0085
0.5253 1.3915 1.0085
16.9564 8.3952 1.0122
0.424e0.753 1.080e1.680 1.003e1.010
0.0000 0.0038 0.0000
CI, confidence interval.
antibiotic-associated diarrhoea with and without proven C. difficile aetiology. The most interesting finding of this prospective caseecontrol study was that C. difficile infection was not associated with higher overall 28 day or long-term mortality rates in hospitalised patients with antibiotic-associated diarrhoea. Although both the crude 28 day and long-term mortality rates in our entire study population were high (12.4% and 56%, respectively, Table I), there were no significant differences between patients who were positive or negative for C. difficile toxin. Previous studies have shown conflicting results concerning mortality attributable to C. difficile infection. In 1989, Eriksson and Aronsson reported a three-fold higher mortality rate in patients with CDAD than control patients matched for age, sex, and underlying diseases.10In contrast, MacGowan et al. noted no differences between patients with CDAD and controls well-matched
for infective diagnosis, severity of illness, and coexistent diseases.11 In a prospective study, Kyne et al. reported that CDAD was associated with an increase in estimated total hospital costs and length of hospital stay.3 Although the crude mortality rate was higher for patients with CDAD than for patients without CDAD, the difference was related to other variables. The authors concluded that CDAD was not associated with excess 3 month or 1 year mortality after adjustment for age, comorbidity and disease severity. It is noteworthy that asymptomatic patients were included as a control group in this study.3 Our findings add validity to the observations made by these investigators. Conversely, Changela et al., in a retrospective caseecontrol study, demonstrated that patients with CDAD had a higher mortality rate than patients without diarrhoea, matched for age and time of hospitalisation.12
1.1 1.0
Cumulative survival (%)
0.9 0.8 0.7 0.6 TOXIN 0.5
1.00 1.00-censored
0.4
0.00 0.00-censored
0.3 0
10
20
30 40 50 Follow-up (months)
60
70
Figure 1 KaplaneMeier curves for long-term mortality. Pink line: patients with positive C. difficile toxin stool assay; green line: patients with negative C. difficile toxin stool assay. x-Axis shows duration of follow-up expressed in months since hospitalisation to death; y-axis shows percentages of survivors (log-rank P ¼ 0.1766).
C. difficile-associated mortality Pepin et al. compared mortality and total length of hospital stay among inpatients in whom nosocomial CDAD developed and among control subjects without CDAD matched for sex, age, and disease severity.13 One year after diagnosis of CDAD, mortality rates of patients with CDAD were significantly higher than those in the control group. The attributable mortality was measured during a CDAD epidemic in Quebec, Canada, caused by a hypervirulent toxinotype III ribotype 027 strain, which produced levels of toxins A and B 16e23 times higher than those of contemporary toxinotype 0 strains. Therefore, these results cannot be extrapolated to centres outside Quebec, where the strain might be absent or uncommon. However, the same strain has been found in several US states and more recently in the UK and the Netherlands.13 Whether a similar ribotype 027 strain is circulating in our hospital is unknown at present. Although many previous studies have analysed prognostic risk factors and associated mortality in patients with CDAD, our study differs in several ways. First, none of the earlier studies compared patients with CDAD to patients with antibioticassociated diarrhoea not associated with C. difficile infection. Second, the methodology of these studies was different, i.e. either retrospective,6,12e15 lacking a control group,14,16,17 or asymptomatic patients were used as controls.3,12 The strongest significant and independent predictor of short- and long-term mortality in our study was hypoalbuminaemia, in concurrence with previous studies.2,6,15,18 Other significant predictors were renal failure, noted also by others, and impaired functional capacity on admission, regardless of age.7 The lack of a significant effect of age has been reported previously.2,15 In addition, several researchers reported functional impairment to be a predictor of increased mortality in patients with CDAD, but not in patients with antibioticassociated diarrhoea, as we found.16,19 Our study had certain limitations. Data on diarrhoeal disease severity, scores of disease severity, such as the Acute Physiological Assessment and Chronic Health Evaluation (APACHE) II, and antimicrobial therapy for C. difficile were not included, and their impact on outcome was not evaluated. In conclusion, antibiotic-associated diarrhoea is associated with high rates of short- and long-term mortality. Hypoalbuminaemia, renal failure and impaired functional capacity predict mortality. Although C. difficile involvement itself does not impact on mortality in these patients, our results should not underestimate the serious
313 consequences of C. difficile infection on morbidity, cost and length of hospitalisation.
References 1. Dharmarajan TS, Sipalay M, Shyamsundar R, Norkus EP, Pitchumoni CS. Co-morbidity, not age, predicts adverse outcome in Clostridium difficile colitis. World J Gastoenterol 2000;6:198e201. 2. Kelly CP, Pothoulakis C, LaMont JT. Clostridium difficile colitis. N Engl J Med 1994;330:257e262. 3. Kyne L, Hamel MB, Polavaram R, Kelly CP. Health care costs and mortality associated with nosocomial diarrhea due to Clostridium difficile. Clin Infect Dis 2002;34: 346e353. 4. Olson MM, Shanholtzer CJ, Lee Jr JT, Gerding DN. Ten years of prospective Clostridium difficile-associated disease surveillance and treatment at the Minneapolis VA Medical Center, 1982e1991. Infect Control Hosp Epidemiol 1994;15: 371e381. 5. Lesna M, Parham DM. Risk of diarrhoea due to Clostridium difficile during cefotaxime treatment; mortality due to C. difficile colitis in elderly people has been underestimated. Br Med J 1996;312:778. 6. Ramaswamy R, Grover H, Corpuz M, Daniels P, Pitchumoni CS. Prognostic criteria in Clostridium difficile colitis. Am J Gastroenterol 1996;91:460e464. 7. Rubin MS, Bodenstein LE, Kent KC. Severe Clostridium difficile colitis. Dis Colon Rectum 1995;38:350e354. 8. Kyne L, Warny M, Qamar A, Kelly C. Asymptomatic carriage of Clostridium difficile and serum levels of IgG antibody against toxin A. N Engl J Med 2000;342:390e397. 9. Samra Z, Talmor S, Bahar J. High prevalence of toxin A-negative toxin B-positive Clostridium difficile in hospitalised patients with gastrointestinal diseases. Diagn Microbiol Infect Dis 2002;43:189e192. 10. Eriksson S, Aronsson B. Medical implications of nosocomial infection with Clostridium difficile. Scand J Infect Dis 1989;21:733e744. 11. McGowan AP, Feeney R, Brown I, McCulloch SY, Reeves DS, Lovering AM. Health care resources utilization and antimicrobial use in elderly patients with community-acquired lower respiratory tract infection who develop Clostridium difficile-associated diarrhea. J Antimicrob Chemother 1997;39:537e541. 12. Changela U, Cannon JP, Aniziokoro C, Shah PS, Thottapurathu L, Lentino J. Risk factors and mortality associated with Clostridium difficile-associated diarrhea at a VA hospital. Int J Antimicrob Agents 2004;24:562e566. 13. Pe ´pin J, Valiquette L, Cossette B. Mortality attributable to nosocomial Clostridium difficile-associated disease during an epidemic caused by a hypervirulent strain in Quebec. Can Med Assoc J 2005;173:1037e1042. 14. Morris AM, Jobe BA, Stoney M, Sheppard BC, Deveney CW, Deveney KE. Clostridium difficile colitis. An increasingly aggressive iatrogenic disease? Arch Surg 2002;137:1096e 1100. 15. Dharmarajan TS, Patel B, Sipalay M, Norkus EP. Clostridium difficile colitis in older adults in long-term facilities and the community: do their outcomes differ? J Am Med Dir Assoc 2000;2:58e61. 16. Kyne L, Merry C, O’Connell B, Kelly A, Keane C, O’Neill D. Factors associated with prolonged symptoms and severe disease due to Clostridium difficile. Age Ageing 1999;28: 107e113.
314 17. Miller M, Hyland M, Ofner-Agostini M, Gourdeau M, Ishak M. Morbidity, mortality, and healthcare burden of nosocomial Clostridium difficile-associated diarrhea in Canadian hospitals. Infect Control Hosp Epidemiol 2002;23:137e140. 18. Nair S, Yadav D, Corpuz M, Pitchomoni CS. Clostridium difficile colitis: factors influencing treatment failure and
J. Bishara et al. relapse, a prospective evaluation. Am J Gastroenterol 1998;93:1873e1876. 19. Anand A, Bashey B, Mir T, Glatt A. Epidemiology, clinical manifestations and outcome of Clostridium difficile-associated diarrhoea. Am J Gastroenterol 1994;89: 519e523.