Journal
of Hospital
Infection
(1991)
18, 219-230
An investigation into Clostridium perfringens enterotoxin-associated diarrhoea S. C. Samuel, Department
of Microbiology,
P. Hancock
and D. A. Leigh
Wycombe General Buckinghamshire
Accepted for publication
Hospital,
23 April
High
Wycombe,
1991
Summary: An investigation was carried out to establish the incidence of diarrhoea associated with the presence of enterotoxigenic Clostridium perfringens. The results indicate a high risk group, namely elderly hospitalized patients, who should be investigated for this organism in a similar way to Clostridium di$j‘kiZe if symptoms occur. The significance of antibiotic association is suggested and cross-infection was shown to be a possibility. Detection of the enterotoxin was accompanied in all cases by the presence of high faecal counts of enterotoxigenic strains of C. perfringens. Keywords:
Enterotoxin;
diarrhoea;
C. perfringens.
Introduction
During late 1985 and early 1986 repeated sporadic outbreaks of diarrhoea occurred in the geriatric wards of Wycombe Health Authority. On routine culture no faecal pathogens were isolated. Samples were also investigated for possible virological causes and these were also negative. All the patients who developed diarrhoea ultimately recovered but the duration of symptoms was prolonged. Most of these cases had previously received antibiotics and this led the laboratory to look for Clostridium dificile and its associated toxin. Selective culture yielded no C. dificile but the presence of large numbers of Clostridium perfringens was noted. Cell lines used for the detection of C. dificile toxin did exhibit a cytopathic effect (C.P.E.) but this was not neutralized by the appropriate Clostridium sordellii antitoxin. This initiated an investigation to see whether the C. perfringens that had been noted before could possibly be the cause of diarrhoea in these patients. The C. perfringens Type A bacterium occurs in man as part of the normal bowel flora in relatively small numbers, up to lo3 colony forming units (cfu) -l.l However, when Yamagishi et a1.2 studied faecal carriage of C. er rzn ens in Japanese geriatric patients their results showed persistent Ff.g high numbers of C. perfringens in faeces up to lo’-lo9 cfu g-‘. Stringer, Correspondence 0195-6701/91;070219+
to: S. C. Samuel 0 1991 The Hospml
12 103 0”:O
219
lnfrctmn
Smety
S. C. Samuel ef al.
220
Watson & Gilbert3 studied faecal carriage of C. perfringens in younger, as well as older adults and found results similar to those produced on earlier occasions.‘*2 It is, therefore, essential to examine stools for the presence of enterotoxin to establish a possible causal agent of diarrhoea.4 This organism and its associated toxin have long been known as a source of food poisoning worldwide.5 There are reports of cases of clostridial diarrhoea where no food source could be traced.6 Clostridium di&cile is already well documented as a cause of antibiotic-associated diarrhoea7 and Borriello et ~1.’ have described C. perfringens as another possible cause of antibiotic-associated diarrhoea. In an epidemiological study carried out during an outbreak of diarrhoea in two hospital wards’ the use of serotyping showed that as well as being antibiotic associated, enterotoxigenic C. perfringens may act as a true infectious agent and that some cross-infection between patients may occur. The aims of this study were to investigate the incidence and treatment of C. perfringens Type A enterotoxin-associated diarrhoea, to establish the incidence of sporadic or antibiotic-associated cases, and to investigate further for evidence of cross-infection between patients. Materials
and methods
Eight hundred and forty-one faecal specimens were examined for the presence of enteric pathogens. Seven hundred and twenty-one diarrhoeal samples were received from symptomatic patients from both general practice and hospitals, and 120 normal samples were received from volunteers within the hospital and the community. As well as looking for the more common pathogens, colony counts of C. perfringens and detection of its associated enterotoxin were carried out. Of the 120 normal samples, 50 were received from within the community and 70 from hospital patients (Table I). A wide age range of volunteers were screened which included roughly equal numbers of males and females. On receipt in the laboratory all samples were cultured for Salmonella, Shigella and Campylobacter species, C. dificile and Escherichia coli
I. Origin of the 120 normal ‘control’ samples
Table Age (years)
Hospital Male
Cl-20 2140 41-60 61-80 81+ Total
3 5 5
Community Female
Male
Female
3 i
5
6
118
139
t
:
32
38
23
27
C. perfringens
diarrhoea
221
0157-H7. Yersinia enterocolitica and Vibrio species were looked for in appropriate clinical cases. Samples were also examined for parasites and the presence of viruses. A reversed passive latex agglutination test (Oxoid Pet-RPLA) was used for the detection of C. perfringens enterotoxin,” positive samples being titred out. Alcohol shock treatment” was used to assess the numbers of C. perfringens present in the faeces. This was carried out using Miles and Misra techniquei on blood agar containing 100 mg l- ’ neomycin. All enterotoxin positive isolates of C. perfringens were confirmed using Vero cell culture technique13 by the Clinical Research Centre, Harrow, Middlesex, and serotyped using slide agglutination by the Food Hygiene Laboratory, Colindale, London.’ Sterilin cotton-tipped swabs were moistened in sterile normal saline and used to take environmental samples from Wards A-D at possible harbouring sites of C. perfringens. These were inoculated immediately into Robertson’s cooked meat broth (Southern Group). Sites included window ledges, floors, sinks, room corners, bed rails, commodes, baths and bed-pan washers. These samples were collected during the later outbreaks. Results Of the 841 faecal specimens examined between January and October 1987, 25 were shown to contain C. perfringens enterotoxin and all had high numbers of clostridia, i.e. > lo3 cfu g-‘, 175 had other enteric pathogens present, 521 samples were from patients with diarrhoea of unknown aetiology and 120 were from normal formed stool samples. Table II shows the distribution amongst these samples of C. perfringens enterotoxin and the numbers of sporulating clostridia present. Seventy-five percent of the normal stool samples had counts of clostridia up to lo3 cfu g-’ compared with only 10% of diarrhoeal samples with other pathogens present, and 30% of diarrhoeal samples of unknown aetiology. Only 25% of normal samples had < 10 cfu g-’ compared with 90% of diarrhoeal samples with other pathogens present and 75% of diarrhoeal samples of unknown aetiology. Table
II.
Detection
Samples
of C. perfringens Number
enterotoxin
and enumeration
C. perfringens
C. perfringens toxin
clostridia
(cfu g-‘)
lo’-lo3
lo3
25
25
0
0
25
175
0
160
15
0
521 120
i
347 30
174 90
0 0
222
S. C. Samuel
et al.
Of the 25 cases of C. perfringens diarrhoea only seven had not received antibiotics preceding the diarrhoea. Two of these seven were in close contact with other C. perfringens cases. The range of antibiotics received varied and included the penicillins, cephalosporins, amoxycillin/clavulanate (Augmentin), erythromycin, trimethoprim, nitrofurantoin and gentamicin (Table III). In ward outbreaks it was found that cases tended to occur when patients were in close proximity to each other. Only two patients were under the age of 60 years, the majority being over 70 years old. There were 18 positive female cases as opposed to only seven male cases.
Table Patient
III.
Ward Age
Clinical details of patients with C. perfringens-associated Diagnosis
Sex
(years)
Antibiotics
Comment
Erythromycin
Profuse weeks
diarrhoea
boil
diarrhoea
for 3
1
A
73
M
Infected
2
A
72
M
Stroke
Amoxycillin/ clavulanate, cefadroxil
Close contact 1
3
A
87
F
Chest infection
Pivampicillin
Patient’s general health deteriorated rapidly until completion of course of metronidazole. Close contact of patient 1
4
A
82
F
Breast
cancer
None
Close contact 1
5
A
91
F
Austin
Moore
Amoxycillin/ clavulanate
General health severely deteriorated after profuse diarrhoea for 1 week. Close contact of patient 1
6
A
81
F
Chest infection
Erythromycin, amoxycillin/ clavulanate
Profuse diarrhoea. Close contact of patient 1
7
B
83
M
Cel lulitis
1) Cefuroxime, erythromycin, flucloxacillin 2) Amoxycillin/ clavulanate, flucloxacillin
Two separate incidents of infection both after different antibiotic courses
8
B
84
M
1. Urinary tract infection 2. Chest infection
1) Amoxycillin/ clavulanate, 2) Cefuroxime, cefadroxil, amoxycillin/ clavulanate
Recurrent diarrhoea with continued antibiotic usage. Close contact of patient 7
9
B
82
M
Bronchiectasis
Amoxycillin/ clavulanate, amoxycillin, erythromycin, gentamicin
Diarrhoea for 3 weeks. Close contact of patient 7
of patient
of patient
C. perfringens Table
diarrhoea
223
III-Continued
Patient
Ward Age Sex (years)
Diagnosis
Antibiotics
Comment
10
B
81
F
Chest infection
Cefuroxime, amoxycillin/ clavulanate
Diarrhoea vomiting. of patient
11
c
71
F
Urinary infection
Trimethoprim
Bloody diarrhoea vomiting
12
C
78
F
Ischaemic
None
Close contact 11
of patient
13
C
64
F
Chronic
Nitrofurantoin, penicillin, trimethoprim
Close contact 11
of patient
14
D
77
F
Carcinoma
Penicillin
Recurrent infection. Close contact of patient 15
15
D
77
F
Cholecystectomy, diverticulitis
Trimethoprim
Diarrhoea, vomiting and abdominal pain. Close contact of patient 14
16
E
84
F
Gastroenteritis
Pivampicillin
13 days profuse diarrhoea
17
F
87
F
Chest infection
Amoxycillin/ clavulanate
Intermittent diarrhoea
18
G
79
F
Stroke
Amoxycillin/ clavulanate
Recurrent diarrhoea
19
H
82
F
Gangrenous
Erythromycin
Diarrhoea
20
I
85
M
Urinary infection
Trimethoprim
Diarrhoea
21
J
75
F
Cancer
None
Diarrhoea
22
K
57
F
Pancreatitis
None
Short period diarrhoea
23
GP
64
F
Rheumatoid arthritis
None
Diarrhoea
for 3 weeks
24
GP
94
M
Rehabilitation
None
Diarrhoea
for 6 weeks
25
GP
45
M
Gastroenteritis
None
Intermittent for 1 year
Wards varied.
A, geriatric long-stay; GP, General practice.
tract colitis leg ulcers
toe
tract
B, geriatric
assessment;
C, female
medical;
D, female
and Close contact 7 and
episodes episodes
of
of
of
diarrhoea
surgical;
E-K,
A total of 155 additional samples were examined from the 25 positive patients. It was found that if the enterotoxin continued to be present, diarrhoea occurred in 67 (99%) of 68 cases. If there was no enterotoxin, then diarrhoea only occured in four (2%) of 87 cases. Where high numbers of clostridia were present, diarrhoea occurred in 64 (98%) of 66 cases. If low numbers were present, diarrhoea only occurred in eight (10%) of 81 cases. Multiple specimens were received from most of the patients during both acute and convalescent stages of their infection. Two specimens were
224
S. C. Samuel
et al.
received before treatment with metronidazole, one or more during treatment, and several post treatment including monthly follow-ups. Two main pictures of infection were exhibited. One showed a single episode of infection (Figure 1) and the other showed recurrence of infection after treatment (Figure 2). In the majority of the patients the diarrhoea, enterotoxin and clostridial count concurrently subsided when metronidazole 400 mg t.d.s. for 10 days was administered. Five of the patients showed recurrence of enterotoxin and symptoms, two of these being after further administration of antibiotics other than metronidazole for other clinical conditions. The clostridial count ranged from 104-lo’* cfu g-’ when the enterotoxin was
ProgressIon
of mfectlon
(specimens)
Figure 1. Clostridiumperfringens counts (---) and enterotoxin a single episode of infection (patient 15); MZ = metronidazole.
concentrations
(---)
during
C. perfringens
diarrhoea
225
present in the stools. In the absence of the enterotoxin, the count was never more than 10” cfu g-‘. All the patients had high concentrations of enterotoxin (> 1280 ng ml-‘) in their stools. On eight occasions, all mid-treatment, diarrhoea was noted when there were < 10” cfu g-’ present in faecal samples. It was found that in each ward outbreak a single serotype was isolated from all the affected patients (Table IV). Ward A had serotype 36, wards B and C had serotype 3 1, 25 while ward D had serotype 67. The remainder of the wards and the isolates from general practice patients produced varied
Dlarrhaea
Dlarrhaea
--
1280-
1 640 I I T E \ F ;
IO7
-
IO6
-
105
100 -
I04
320
;: E b E
80
-
10~
$?
40
-
IO’
s k 9 i,
20
-
IO’ 0
Progression
Figure patient
2. Clostridium perfringens Counts (-) with recurrence of infection (patient
of infectm
(specmens)
and enterotoxin 7).
concentrations
(-
-) in one
S. C. Samuel Table
IV. Serotypes
Patient 1 2 3 4 5 6 7 : 10 11 12 13 14 15 16 17 18 19 20 21 22 23
Faecal
V. Incidence
isolates from patients
Ward
Serotype
A A A A A A B B B B
36 36 36 36 36 36 31, 25 31,25 31, 25 31,25 31,25 31,25 31, 25 67 67 31, 25 38, 27 38, 27 38, 27 NSA PS68 36 PS73 4 36
K GP GP GP
NSA, Non-specific
Table
of C. perfringens
J
;:
et al.
agglutination;
PS, American antisera.
of enteric pathogens isolatedfrom with infective diarrhoea
pathogens
Campylobacter jejuni Salmonella spp. Enterotoxigenic C. perfringens Cryptosporidium Shigella spp. C. d$cile E. coli 0157-H7 Yersinia enterocolitica Vibrio parahaemolyticus
No.
200 patients
(%)
67 57 25 20 12 9 7 2 1
Total
serotypes. All environmental swabs taken from wards A-D yielded high numbers of C. perfringens. The environmental serotypes included 23, 24, 35, 38, 55, 68 and 71. Infecting serotypes 36, 67 and 31, 25 were not isolated from any of the environmental sites.
C. perfringens
diarrhoea
227
It was possible to establish the incidence of C. perfringens diarrhoea when compared with that of other causes of enteritis (Table V). The number of enterotoxigenic C. perfringens cases was only exceeded by the number of salmonella and campylobacter infections. Discussion
None of the 120 normal formed samples, including those from geriatrics, had counts of C. perfringens over lo3 cfu g- ‘. This disagrees with previous work on the faecal carriage of C. perfringens in geriatric patients. Both Yamagishi et ~1.~ and Stringer et ~1.~ found numbers > lo7 cfu g-’ in the faeces of healthy geriatric patients. This was not the case in any single control sample from geriatrics in this study. There was no significant difference in the numbers of C. perfringens present in the faeces of long-stay geriatric hospital in-patients and geriatrics in the community. The method of spore selection was tested and compared with a batch also done using heat activation at 100°C for 30 min. The counts were exactly the same, showing the alcohol shock treatment to be accurate and reliable. The difference in the results of this study may be explained by the smaller sample size employed in previous studies. Yamagishi et a1.2 found persistently high numbers of C. perfringens in the faeces of only five of 30 geriatric patients and Stringer et al. 3 found the same in five of 21 elderly patients. It is possible that too few patients were examined and that their findings were not truly representative. Of the 721 diarrhoeal samples screened, enteric pathogens excluding viruses and parasites (but not cryptosporidia) were isolated from 200 different patients. Enterotoxigenic C. perfringens was the third most common pathogen to be isolated from all the faecal specimens, exceeded only by the number of campylobacter and salmonella infections. Commonly accepted pathogens with a lower incidence than C. perfringens included cryptosporidia, Shigella species, C. dificile, E. coli 0157-H7, Y. enterocolitica and V. parahaemolyticus. Of 546 cases of diarrhoea with no other accepted pathogens detected, 25 had enterotoxigenic C. perfringens present. Of the 2.5 cases of C. perfringens diarrhoea there was a marked female predominance but this finding was not statistically significant on comparison with control groups. Hospitalization was an important factor with 22 out of 2.5 of the cases being in-patients. Another significant finding was the correlation of old age with the likelihood of diarrhoea associated with C. perfringens infection. These results agree with the work of Borriello et ~1.’ Similar antibiotics had been administered to the majority of the patients in both studies including the penicillins, cephalosporins and trimethoprim. Borriello et ~1.’ were the first investigators to report cases of C. perfringens diarrhoea with a possible antibiotic association. Having completed a similar study, the results produced demonstrate this antibiotic
228
S. C. Samuel
et al.
association in 18 of the cases although other factors such as host condition and environment may play a r6le. Most of the patients had prolonged periods of diarrhoea with evidence of blood in some. The general health of two of the patients deteriorated rapidly and it was only after the complete course of metronidazole that some improvement was seen. The severity of infection was varied, and depended on the physical condition of each patient. All the cases of C. perfringens diarrhoea had high counts, in comparison with the ‘control’ samples. Based on this observation, it could be suggested that a simple screening test could be employed to primarily detect high numbers of C. perfringens in diarrhoeal samples. In any such cases identified, detection of the enterotoxin in faeces would then be carried out to determine if the organism was toxigenic. Comparison of cases of diarrhoea either with other pathogens present or with cases of unexplained aetiology showed that very low numbers of C. perfringens were isolated. This suggests that unless C. perfringens is the causative agent of diarrhoea, high numbers are not present. In all cases the symptoms had resolved and the organism (except one case) and enterotoxin were undetectable on completion of treatment. In some cases symptoms had resolved after only two or three days. This contrasts with a case report on one patient with C. perfringens diarrhoea,14 where it was found that treatment did not eradicate the enterotoxigenic serotype of C. perfringens from the patient’s faeces although the diarrhoea and enterotoxin disappeared. It was also noted that diarrhoea had not been reported previously in patients carrying < 10’ cfu g-i in their faeces. In this study, diarrhoea was found on several occasions when organisms were present at < lo4 cfu g-‘. There were five cases of recurrence of infection after the initial illness had completely resolved following treatment with metronidazole. This agrees with the results of previous studies. Two of these cases were after further administration of antibiotics other than metronidazole. There was no explanation for the other three cases apart from re-infection. Several different serotypes were isolated from those faecal samples containing enterotoxigenic C. perfringens. However, in each individual ward outbreak, single serotypes were isolated from all the patients. This is contrary to previous studiesss’4 in which a variety of serotypes were isolated from the patients involved. Four separate incidents of ward outbreaks were reported in wards A-D. In all four of these outbreaks patients had been clustered together in the same bays or side rooms. Further investigations were carried out to eliminate food poisoning. No single item of food was consumed by all the patients involved and patients on other wards were not affected. The symptoms were not typical of food poisoning caused by C. perfringens and all the patients in each ward started the infection on different days over a period of a few weeks. This agreed with previous studies.8 In two of the ward outbreaks, occurring on wards A and C, one patient in each
C. perfringens
diarrhoea
229
ward had not received antibiotics prior to the onset of diarrhoea. Some factor other than antibiotic association must have been involved. Patient-topatient cross-infection may have occurred or high numbers of the infecting serotype were accumulating in the environment and causing re-infection of early cases as well as producing new cases. The results showed that a variety of environmental serotypes were isolated. This was to be expected in a geriatric ward where there were usually several incontinent patients and high standards of hygiene were not always possible. However, as the infecting serotypes 36, 67 and 31, 25 were not detected in the environment, it can be suggested that due to the close proximity of patients in the ward, patient-to-patient cross-infection occurred. This differs from results of a previous studyI which showed that high numbers of the infecting serotype were present in the environment. The results of this study clearly indicate that cases of diarrhoea caused by enterotoxigenic C. perfringens should be recognized. Antibiotic association in hospitalized elderly patients is probably the highest risk factor. However, there is also an indication that patient-to-patient cross-infection may occur. Preventive measures should be implemented to try to halt the spread of infection and efforts should be made to improve the standards of hygiene. It would seem perfectly reasonable and practical to isolate the affected patients in side rooms or in a separate bay of the ward. Although serotyping has shown that single serotypes were present in outbreaks, further investigations such as plasmid analysis could be undertaken to confirm that the strains all originated from the same source.” Further studies are also required on patient-to-patient spread of infection including investigations of hand carriage of the infecting organism on both patients and staff during an outbreak. An investigation of the pathogenicity of enterotoxigenic C. perfringens could be carried out possibly persuading human volunteers to ingest a known infecting serotype to see if the infection can be reproduced. This has already been accomplished in investigations of C. perfringens food poisoning ” but has yet to be established with C. perfringens-associated diarrhoea from other sources. We thank Dr K. Hoyle, Microbiology Dept., Wycombe General Hospital; Dr S. P. Borriello and Mr A. Welch, Clinical Research Centre, Harrow, for confirmation of positive samples; and Mr P. Berry, Food Hygiene Laboratory, Colindale, for serotyping of isolates.
References 1. Sutton RGA. Enumeration of Clostridizcm welchii in the faeces of varying sections of the human population. J Hyg 1966; 64: 367-374. 2. Yamagishi T, Serikawa T, Morita R, Nakamura S, Nishida S. Persistent high numbers of Clostridium perfringens in the intestines of Japanese aged adults. Jpn J Microbial 1976; 20: 397403. 3. Stringer MF, Watson GN, Gilbert RJ. Faecal carriage of Clostridium perfringens.J Hyg 1985; 95: 277-288.
230
S. C. Samuel
et al.
4. Jackson SG, Yip-Chuck DA, Clark JB, Brodsky MH. Diagnostic importance of Clostridiumperjringens enterotoxin analysis in recurring enteritis among elderly, chronic care psychiatric patients. J Clin Microbial 1986; 23: 748-751. 5. Stringer MF, Turnbull PCB, Gilbert RJ. Application of serological typing to the investigation of outbreaks of Clostridium perjringens food poisoning, 1970-1978. r Hyg 1980; 84: 443-456. SP et al. Diarrhoea due to enterotoxigenic Clostridium 6. Williams R, Piper M, Borriello perjringens: clinical features and management of a cluster of cases. Age and Ageing 1985; 14: 296-302. SP, Larson HE. Antibiotic and pseudomembranous colitis, J Antimicrob 7. Borriello Chemother 1981; 7 (Suppl. A): 53-62. SP, Larson HE, Welch AR, Barclay F, Stringer MF, Bartholomew BA. 8. Borriello Enterotoxigenic Clostridium perjringens: a possible cause of antibiotic associated diarrhoea. Lancet 1984; 1: 305-307. 9. Borriello SP, Barclay F, Welch AR, Sullens K, Stringer MF Watson GN. Epidemiology of Clostridium perfringens enterotoxin-associated with diarrhoea. J Med Microbial 1985; 20: 363-372. 10. Berry PR, Stringer MF & Uemura T. Comparison of latex agglutination and ELISA for the detection of Clostridium perfringens Type A enterotoxin in faeces. Appl Microbial 1986; 2: 101-102. 11. Koransky JR, Allen SD, Dowel1 VR. Use of ethanol for selective isolation of sporeforming micro-organisms. Appl Environ Microbial 1978; 35: 762-765. 12. Miles AA, Misra SS. The estimation of the bactericidal power of the blood. J Hyg (Camb) 1938; 38: 732. 13. Welch AR, Borriello SP, Barclay FE. Simplified procedure for tissue culture in routine detection of cytotoxins J Clin Path01 1985; 38: 835-837. 14. Borriello SP, Williams RKT. Treatment of Clostridium perfringens enterotoxin-associJ Inject 1985; 10: 65-67. ated diarrhoea with metronidazole. 15. Mahoney DE, Stringer MF, Borriello SP Mader JA. Plasmid analysis as a means of strain differentiation in Clostridium perjringens. J Clin Microbial 1987; 25: 1333-1335. R, Uemura T. Experimental diarrhoea in human volunteers following oral 16. Skjelkvale, administration of Clostridium perjringens enterotoxin. J Appl Bacterial 1977; 43: 281-286.
of Hospital
Infection
(1991)
18, 219-230
An investigation into Clostridium perfringens enterotoxin-associated diarrhoea S. C. Samuel, Department
of Microbiology,
P. Hancock
and D. A. Leigh
Wycombe General Buckinghamshire
Accepted for publication
Hospital,
23 April
High
Wycombe,
1991
Summary: An investigation was carried out to establish the incidence of diarrhoea associated with the presence of enterotoxigenic Clostridium perfringens. The results indicate a high risk group, namely elderly hospitalized patients, who should be investigated for this organism in a similar way to Clostridium di$j‘kiZe if symptoms occur. The significance of antibiotic association is suggested and cross-infection was shown to be a possibility. Detection of the enterotoxin was accompanied in all cases by the presence of high faecal counts of enterotoxigenic strains of C. perfringens. Keywords:
Enterotoxin;
diarrhoea;
C. perfringens.
Introduction
During late 1985 and early 1986 repeated sporadic outbreaks of diarrhoea occurred in the geriatric wards of Wycombe Health Authority. On routine culture no faecal pathogens were isolated. Samples were also investigated for possible virological causes and these were also negative. All the patients who developed diarrhoea ultimately recovered but the duration of symptoms was prolonged. Most of these cases had previously received antibiotics and this led the laboratory to look for Clostridium dificile and its associated toxin. Selective culture yielded no C. dificile but the presence of large numbers of Clostridium perfringens was noted. Cell lines used for the detection of C. dificile toxin did exhibit a cytopathic effect (C.P.E.) but this was not neutralized by the appropriate Clostridium sordellii antitoxin. This initiated an investigation to see whether the C. perfringens that had been noted before could possibly be the cause of diarrhoea in these patients. The C. perfringens Type A bacterium occurs in man as part of the normal bowel flora in relatively small numbers, up to lo3 colony forming units (cfu) -l.l However, when Yamagishi et a1.2 studied faecal carriage of C. er rzn ens in Japanese geriatric patients their results showed persistent Ff.g high numbers of C. perfringens in faeces up to lo’-lo9 cfu g-‘. Stringer, Correspondence 0195-6701/91;070219+
to: S. C. Samuel 0 1991 The Hospml
12 103 0”:O
219
lnfrctmn
Smety
S. C. Samuel ef al.
220
Watson & Gilbert3 studied faecal carriage of C. perfringens in younger, as well as older adults and found results similar to those produced on earlier occasions.‘*2 It is, therefore, essential to examine stools for the presence of enterotoxin to establish a possible causal agent of diarrhoea.4 This organism and its associated toxin have long been known as a source of food poisoning worldwide.5 There are reports of cases of clostridial diarrhoea where no food source could be traced.6 Clostridium di&cile is already well documented as a cause of antibiotic-associated diarrhoea7 and Borriello et ~1.’ have described C. perfringens as another possible cause of antibiotic-associated diarrhoea. In an epidemiological study carried out during an outbreak of diarrhoea in two hospital wards’ the use of serotyping showed that as well as being antibiotic associated, enterotoxigenic C. perfringens may act as a true infectious agent and that some cross-infection between patients may occur. The aims of this study were to investigate the incidence and treatment of C. perfringens Type A enterotoxin-associated diarrhoea, to establish the incidence of sporadic or antibiotic-associated cases, and to investigate further for evidence of cross-infection between patients. Materials
and methods
Eight hundred and forty-one faecal specimens were examined for the presence of enteric pathogens. Seven hundred and twenty-one diarrhoeal samples were received from symptomatic patients from both general practice and hospitals, and 120 normal samples were received from volunteers within the hospital and the community. As well as looking for the more common pathogens, colony counts of C. perfringens and detection of its associated enterotoxin were carried out. Of the 120 normal samples, 50 were received from within the community and 70 from hospital patients (Table I). A wide age range of volunteers were screened which included roughly equal numbers of males and females. On receipt in the laboratory all samples were cultured for Salmonella, Shigella and Campylobacter species, C. dificile and Escherichia coli
I. Origin of the 120 normal ‘control’ samples
Table Age (years)
Hospital Male
Cl-20 2140 41-60 61-80 81+ Total
3 5 5
Community Female
Male
Female
3 i
5
6
118
139
t
:
32
38
23
27
C. perfringens
diarrhoea
221
0157-H7. Yersinia enterocolitica and Vibrio species were looked for in appropriate clinical cases. Samples were also examined for parasites and the presence of viruses. A reversed passive latex agglutination test (Oxoid Pet-RPLA) was used for the detection of C. perfringens enterotoxin,” positive samples being titred out. Alcohol shock treatment” was used to assess the numbers of C. perfringens present in the faeces. This was carried out using Miles and Misra techniquei on blood agar containing 100 mg l- ’ neomycin. All enterotoxin positive isolates of C. perfringens were confirmed using Vero cell culture technique13 by the Clinical Research Centre, Harrow, Middlesex, and serotyped using slide agglutination by the Food Hygiene Laboratory, Colindale, London.’ Sterilin cotton-tipped swabs were moistened in sterile normal saline and used to take environmental samples from Wards A-D at possible harbouring sites of C. perfringens. These were inoculated immediately into Robertson’s cooked meat broth (Southern Group). Sites included window ledges, floors, sinks, room corners, bed rails, commodes, baths and bed-pan washers. These samples were collected during the later outbreaks. Results Of the 841 faecal specimens examined between January and October 1987, 25 were shown to contain C. perfringens enterotoxin and all had high numbers of clostridia, i.e. > lo3 cfu g-‘, 175 had other enteric pathogens present, 521 samples were from patients with diarrhoea of unknown aetiology and 120 were from normal formed stool samples. Table II shows the distribution amongst these samples of C. perfringens enterotoxin and the numbers of sporulating clostridia present. Seventy-five percent of the normal stool samples had counts of clostridia up to lo3 cfu g-’ compared with only 10% of diarrhoeal samples with other pathogens present, and 30% of diarrhoeal samples of unknown aetiology. Only 25% of normal samples had < 10 cfu g-’ compared with 90% of diarrhoeal samples with other pathogens present and 75% of diarrhoeal samples of unknown aetiology. Table
II.
Detection
Samples
of C. perfringens Number
enterotoxin
and enumeration
C. perfringens
C. perfringens toxin
clostridia
(cfu g-‘)
lo’-lo3
lo3
25
25
0
0
25
175
0
160
15
0
521 120
i
347 30
174 90
0 0
222
S. C. Samuel
et al.
Of the 25 cases of C. perfringens diarrhoea only seven had not received antibiotics preceding the diarrhoea. Two of these seven were in close contact with other C. perfringens cases. The range of antibiotics received varied and included the penicillins, cephalosporins, amoxycillin/clavulanate (Augmentin), erythromycin, trimethoprim, nitrofurantoin and gentamicin (Table III). In ward outbreaks it was found that cases tended to occur when patients were in close proximity to each other. Only two patients were under the age of 60 years, the majority being over 70 years old. There were 18 positive female cases as opposed to only seven male cases.
Table Patient
III.
Ward Age
Clinical details of patients with C. perfringens-associated Diagnosis
Sex
(years)
Antibiotics
Comment
Erythromycin
Profuse weeks
diarrhoea
boil
diarrhoea
for 3
1
A
73
M
Infected
2
A
72
M
Stroke
Amoxycillin/ clavulanate, cefadroxil
Close contact 1
3
A
87
F
Chest infection
Pivampicillin
Patient’s general health deteriorated rapidly until completion of course of metronidazole. Close contact of patient 1
4
A
82
F
Breast
cancer
None
Close contact 1
5
A
91
F
Austin
Moore
Amoxycillin/ clavulanate
General health severely deteriorated after profuse diarrhoea for 1 week. Close contact of patient 1
6
A
81
F
Chest infection
Erythromycin, amoxycillin/ clavulanate
Profuse diarrhoea. Close contact of patient 1
7
B
83
M
Cel lulitis
1) Cefuroxime, erythromycin, flucloxacillin 2) Amoxycillin/ clavulanate, flucloxacillin
Two separate incidents of infection both after different antibiotic courses
8
B
84
M
1. Urinary tract infection 2. Chest infection
1) Amoxycillin/ clavulanate, 2) Cefuroxime, cefadroxil, amoxycillin/ clavulanate
Recurrent diarrhoea with continued antibiotic usage. Close contact of patient 7
9
B
82
M
Bronchiectasis
Amoxycillin/ clavulanate, amoxycillin, erythromycin, gentamicin
Diarrhoea for 3 weeks. Close contact of patient 7
of patient
of patient
C. perfringens Table
diarrhoea
223
III-Continued
Patient
Ward Age Sex (years)
Diagnosis
Antibiotics
Comment
10
B
81
F
Chest infection
Cefuroxime, amoxycillin/ clavulanate
Diarrhoea vomiting. of patient
11
c
71
F
Urinary infection
Trimethoprim
Bloody diarrhoea vomiting
12
C
78
F
Ischaemic
None
Close contact 11
of patient
13
C
64
F
Chronic
Nitrofurantoin, penicillin, trimethoprim
Close contact 11
of patient
14
D
77
F
Carcinoma
Penicillin
Recurrent infection. Close contact of patient 15
15
D
77
F
Cholecystectomy, diverticulitis
Trimethoprim
Diarrhoea, vomiting and abdominal pain. Close contact of patient 14
16
E
84
F
Gastroenteritis
Pivampicillin
13 days profuse diarrhoea
17
F
87
F
Chest infection
Amoxycillin/ clavulanate
Intermittent diarrhoea
18
G
79
F
Stroke
Amoxycillin/ clavulanate
Recurrent diarrhoea
19
H
82
F
Gangrenous
Erythromycin
Diarrhoea
20
I
85
M
Urinary infection
Trimethoprim
Diarrhoea
21
J
75
F
Cancer
None
Diarrhoea
22
K
57
F
Pancreatitis
None
Short period diarrhoea
23
GP
64
F
Rheumatoid arthritis
None
Diarrhoea
for 3 weeks
24
GP
94
M
Rehabilitation
None
Diarrhoea
for 6 weeks
25
GP
45
M
Gastroenteritis
None
Intermittent for 1 year
Wards varied.
A, geriatric long-stay; GP, General practice.
tract colitis leg ulcers
toe
tract
B, geriatric
assessment;
C, female
medical;
D, female
and Close contact 7 and
episodes episodes
of
of
of
diarrhoea
surgical;
E-K,
A total of 155 additional samples were examined from the 25 positive patients. It was found that if the enterotoxin continued to be present, diarrhoea occurred in 67 (99%) of 68 cases. If there was no enterotoxin, then diarrhoea only occured in four (2%) of 87 cases. Where high numbers of clostridia were present, diarrhoea occurred in 64 (98%) of 66 cases. If low numbers were present, diarrhoea only occurred in eight (10%) of 81 cases. Multiple specimens were received from most of the patients during both acute and convalescent stages of their infection. Two specimens were
224
S. C. Samuel
et al.
received before treatment with metronidazole, one or more during treatment, and several post treatment including monthly follow-ups. Two main pictures of infection were exhibited. One showed a single episode of infection (Figure 1) and the other showed recurrence of infection after treatment (Figure 2). In the majority of the patients the diarrhoea, enterotoxin and clostridial count concurrently subsided when metronidazole 400 mg t.d.s. for 10 days was administered. Five of the patients showed recurrence of enterotoxin and symptoms, two of these being after further administration of antibiotics other than metronidazole for other clinical conditions. The clostridial count ranged from 104-lo’* cfu g-’ when the enterotoxin was
ProgressIon
of mfectlon
(specimens)
Figure 1. Clostridiumperfringens counts (---) and enterotoxin a single episode of infection (patient 15); MZ = metronidazole.
concentrations
(---)
during
C. perfringens
diarrhoea
225
present in the stools. In the absence of the enterotoxin, the count was never more than 10” cfu g-‘. All the patients had high concentrations of enterotoxin (> 1280 ng ml-‘) in their stools. On eight occasions, all mid-treatment, diarrhoea was noted when there were < 10” cfu g-’ present in faecal samples. It was found that in each ward outbreak a single serotype was isolated from all the affected patients (Table IV). Ward A had serotype 36, wards B and C had serotype 3 1, 25 while ward D had serotype 67. The remainder of the wards and the isolates from general practice patients produced varied
Dlarrhaea
Dlarrhaea
--
1280-
1 640 I I T E \ F ;
IO7
-
IO6
-
105
100 -
I04
320
;: E b E
80
-
10~
$?
40
-
IO’
s k 9 i,
20
-
IO’ 0
Progression
Figure patient
2. Clostridium perfringens Counts (-) with recurrence of infection (patient
of infectm
(specmens)
and enterotoxin 7).
concentrations
(-
-) in one
S. C. Samuel Table
IV. Serotypes
Patient 1 2 3 4 5 6 7 : 10 11 12 13 14 15 16 17 18 19 20 21 22 23
Faecal
V. Incidence
isolates from patients
Ward
Serotype
A A A A A A B B B B
36 36 36 36 36 36 31, 25 31,25 31, 25 31,25 31,25 31,25 31, 25 67 67 31, 25 38, 27 38, 27 38, 27 NSA PS68 36 PS73 4 36
K GP GP GP
NSA, Non-specific
Table
of C. perfringens
J
;:
et al.
agglutination;
PS, American antisera.
of enteric pathogens isolatedfrom with infective diarrhoea
pathogens
Campylobacter jejuni Salmonella spp. Enterotoxigenic C. perfringens Cryptosporidium Shigella spp. C. d$cile E. coli 0157-H7 Yersinia enterocolitica Vibrio parahaemolyticus
No.
200 patients
(%)
67 57 25 20 12 9 7 2 1
Total
serotypes. All environmental swabs taken from wards A-D yielded high numbers of C. perfringens. The environmental serotypes included 23, 24, 35, 38, 55, 68 and 71. Infecting serotypes 36, 67 and 31, 25 were not isolated from any of the environmental sites.
C. perfringens
diarrhoea
227
It was possible to establish the incidence of C. perfringens diarrhoea when compared with that of other causes of enteritis (Table V). The number of enterotoxigenic C. perfringens cases was only exceeded by the number of salmonella and campylobacter infections. Discussion
None of the 120 normal formed samples, including those from geriatrics, had counts of C. perfringens over lo3 cfu g- ‘. This disagrees with previous work on the faecal carriage of C. perfringens in geriatric patients. Both Yamagishi et ~1.~ and Stringer et ~1.~ found numbers > lo7 cfu g-’ in the faeces of healthy geriatric patients. This was not the case in any single control sample from geriatrics in this study. There was no significant difference in the numbers of C. perfringens present in the faeces of long-stay geriatric hospital in-patients and geriatrics in the community. The method of spore selection was tested and compared with a batch also done using heat activation at 100°C for 30 min. The counts were exactly the same, showing the alcohol shock treatment to be accurate and reliable. The difference in the results of this study may be explained by the smaller sample size employed in previous studies. Yamagishi et a1.2 found persistently high numbers of C. perfringens in the faeces of only five of 30 geriatric patients and Stringer et al. 3 found the same in five of 21 elderly patients. It is possible that too few patients were examined and that their findings were not truly representative. Of the 721 diarrhoeal samples screened, enteric pathogens excluding viruses and parasites (but not cryptosporidia) were isolated from 200 different patients. Enterotoxigenic C. perfringens was the third most common pathogen to be isolated from all the faecal specimens, exceeded only by the number of campylobacter and salmonella infections. Commonly accepted pathogens with a lower incidence than C. perfringens included cryptosporidia, Shigella species, C. dificile, E. coli 0157-H7, Y. enterocolitica and V. parahaemolyticus. Of 546 cases of diarrhoea with no other accepted pathogens detected, 25 had enterotoxigenic C. perfringens present. Of the 2.5 cases of C. perfringens diarrhoea there was a marked female predominance but this finding was not statistically significant on comparison with control groups. Hospitalization was an important factor with 22 out of 2.5 of the cases being in-patients. Another significant finding was the correlation of old age with the likelihood of diarrhoea associated with C. perfringens infection. These results agree with the work of Borriello et ~1.’ Similar antibiotics had been administered to the majority of the patients in both studies including the penicillins, cephalosporins and trimethoprim. Borriello et ~1.’ were the first investigators to report cases of C. perfringens diarrhoea with a possible antibiotic association. Having completed a similar study, the results produced demonstrate this antibiotic
228
S. C. Samuel
et al.
association in 18 of the cases although other factors such as host condition and environment may play a r6le. Most of the patients had prolonged periods of diarrhoea with evidence of blood in some. The general health of two of the patients deteriorated rapidly and it was only after the complete course of metronidazole that some improvement was seen. The severity of infection was varied, and depended on the physical condition of each patient. All the cases of C. perfringens diarrhoea had high counts, in comparison with the ‘control’ samples. Based on this observation, it could be suggested that a simple screening test could be employed to primarily detect high numbers of C. perfringens in diarrhoeal samples. In any such cases identified, detection of the enterotoxin in faeces would then be carried out to determine if the organism was toxigenic. Comparison of cases of diarrhoea either with other pathogens present or with cases of unexplained aetiology showed that very low numbers of C. perfringens were isolated. This suggests that unless C. perfringens is the causative agent of diarrhoea, high numbers are not present. In all cases the symptoms had resolved and the organism (except one case) and enterotoxin were undetectable on completion of treatment. In some cases symptoms had resolved after only two or three days. This contrasts with a case report on one patient with C. perfringens diarrhoea,14 where it was found that treatment did not eradicate the enterotoxigenic serotype of C. perfringens from the patient’s faeces although the diarrhoea and enterotoxin disappeared. It was also noted that diarrhoea had not been reported previously in patients carrying < 10’ cfu g-i in their faeces. In this study, diarrhoea was found on several occasions when organisms were present at < lo4 cfu g-‘. There were five cases of recurrence of infection after the initial illness had completely resolved following treatment with metronidazole. This agrees with the results of previous studies. Two of these cases were after further administration of antibiotics other than metronidazole. There was no explanation for the other three cases apart from re-infection. Several different serotypes were isolated from those faecal samples containing enterotoxigenic C. perfringens. However, in each individual ward outbreak, single serotypes were isolated from all the patients. This is contrary to previous studiesss’4 in which a variety of serotypes were isolated from the patients involved. Four separate incidents of ward outbreaks were reported in wards A-D. In all four of these outbreaks patients had been clustered together in the same bays or side rooms. Further investigations were carried out to eliminate food poisoning. No single item of food was consumed by all the patients involved and patients on other wards were not affected. The symptoms were not typical of food poisoning caused by C. perfringens and all the patients in each ward started the infection on different days over a period of a few weeks. This agreed with previous studies.8 In two of the ward outbreaks, occurring on wards A and C, one patient in each
C. perfringens
diarrhoea
229
ward had not received antibiotics prior to the onset of diarrhoea. Some factor other than antibiotic association must have been involved. Patient-topatient cross-infection may have occurred or high numbers of the infecting serotype were accumulating in the environment and causing re-infection of early cases as well as producing new cases. The results showed that a variety of environmental serotypes were isolated. This was to be expected in a geriatric ward where there were usually several incontinent patients and high standards of hygiene were not always possible. However, as the infecting serotypes 36, 67 and 31, 25 were not detected in the environment, it can be suggested that due to the close proximity of patients in the ward, patient-to-patient cross-infection occurred. This differs from results of a previous studyI which showed that high numbers of the infecting serotype were present in the environment. The results of this study clearly indicate that cases of diarrhoea caused by enterotoxigenic C. perfringens should be recognized. Antibiotic association in hospitalized elderly patients is probably the highest risk factor. However, there is also an indication that patient-to-patient cross-infection may occur. Preventive measures should be implemented to try to halt the spread of infection and efforts should be made to improve the standards of hygiene. It would seem perfectly reasonable and practical to isolate the affected patients in side rooms or in a separate bay of the ward. Although serotyping has shown that single serotypes were present in outbreaks, further investigations such as plasmid analysis could be undertaken to confirm that the strains all originated from the same source.” Further studies are also required on patient-to-patient spread of infection including investigations of hand carriage of the infecting organism on both patients and staff during an outbreak. An investigation of the pathogenicity of enterotoxigenic C. perfringens could be carried out possibly persuading human volunteers to ingest a known infecting serotype to see if the infection can be reproduced. This has already been accomplished in investigations of C. perfringens food poisoning ” but has yet to be established with C. perfringens-associated diarrhoea from other sources. We thank Dr K. Hoyle, Microbiology Dept., Wycombe General Hospital; Dr S. P. Borriello and Mr A. Welch, Clinical Research Centre, Harrow, for confirmation of positive samples; and Mr P. Berry, Food Hygiene Laboratory, Colindale, for serotyping of isolates.
References 1. Sutton RGA. Enumeration of Clostridizcm welchii in the faeces of varying sections of the human population. J Hyg 1966; 64: 367-374. 2. Yamagishi T, Serikawa T, Morita R, Nakamura S, Nishida S. Persistent high numbers of Clostridium perfringens in the intestines of Japanese aged adults. Jpn J Microbial 1976; 20: 397403. 3. Stringer MF, Watson GN, Gilbert RJ. Faecal carriage of Clostridium perfringens.J Hyg 1985; 95: 277-288.
230
S. C. Samuel
et al.
4. Jackson SG, Yip-Chuck DA, Clark JB, Brodsky MH. Diagnostic importance of Clostridiumperjringens enterotoxin analysis in recurring enteritis among elderly, chronic care psychiatric patients. J Clin Microbial 1986; 23: 748-751. 5. Stringer MF, Turnbull PCB, Gilbert RJ. Application of serological typing to the investigation of outbreaks of Clostridium perjringens food poisoning, 1970-1978. r Hyg 1980; 84: 443-456. SP et al. Diarrhoea due to enterotoxigenic Clostridium 6. Williams R, Piper M, Borriello perjringens: clinical features and management of a cluster of cases. Age and Ageing 1985; 14: 296-302. SP, Larson HE. Antibiotic and pseudomembranous colitis, J Antimicrob 7. Borriello Chemother 1981; 7 (Suppl. A): 53-62. SP, Larson HE, Welch AR, Barclay F, Stringer MF, Bartholomew BA. 8. Borriello Enterotoxigenic Clostridium perjringens: a possible cause of antibiotic associated diarrhoea. Lancet 1984; 1: 305-307. 9. Borriello SP, Barclay F, Welch AR, Sullens K, Stringer MF Watson GN. Epidemiology of Clostridium perfringens enterotoxin-associated with diarrhoea. J Med Microbial 1985; 20: 363-372. 10. Berry PR, Stringer MF & Uemura T. Comparison of latex agglutination and ELISA for the detection of Clostridium perfringens Type A enterotoxin in faeces. Appl Microbial 1986; 2: 101-102. 11. Koransky JR, Allen SD, Dowel1 VR. Use of ethanol for selective isolation of sporeforming micro-organisms. Appl Environ Microbial 1978; 35: 762-765. 12. Miles AA, Misra SS. The estimation of the bactericidal power of the blood. J Hyg (Camb) 1938; 38: 732. 13. Welch AR, Borriello SP, Barclay FE. Simplified procedure for tissue culture in routine detection of cytotoxins J Clin Path01 1985; 38: 835-837. 14. Borriello SP, Williams RKT. Treatment of Clostridium perfringens enterotoxin-associJ Inject 1985; 10: 65-67. ated diarrhoea with metronidazole. 15. Mahoney DE, Stringer MF, Borriello SP Mader JA. Plasmid analysis as a means of strain differentiation in Clostridium perjringens. J Clin Microbial 1987; 25: 1333-1335. R, Uemura T. Experimental diarrhoea in human volunteers following oral 16. Skjelkvale, administration of Clostridium perjringens enterotoxin. J Appl Bacterial 1977; 43: 281-286.
