A Mild Outbreak of Gastroenteritis in Long-Term Care Facility Residents Due to Clostridium perfringens,Australia 2009

Abstract

Introduction:

Clostridium perfringens food poisoning is a commonly cited cause of gastroenteritis outbreaks among elderly long-term care facility (LTCF) residents, yet little is known about the natural history of disease in this vulnerable group. In July 2009, an investigation into diarrheal illness among LTCF residents was commenced.

Methods:

An environmental health investigation and retrospective cohort study were undertaken to confirm the outbreak, to identify a source and mode of transmission, and to implement public health measures to prevent further cases. Menu listings and food safety program details were obtained and food-handling practices were observed. Clinical notes of all residents were reviewed. A possible case was defined as any resident developing one or more acute loose stool episodes between the evenings of 23 July and 27 July.

Results:

Fifty-two residents (41%) had been ill with diarrhea, and eight residents had fecal samples positive for C. perfringens enterotoxin. LTCF staff failed to perform routine temperature checks on hot foods before the outbreak. A sweet-and-sour pork lunch served on 23 July was implicated in causing residents' illness, but no residual food remained for microbiological testing. Independent associations with illness were demonstrated among residents living in two wings of the facility that received a standard level of service, whereas an inverse association with illness was shown among residents living in an “extra service” wing. Male residents were also more likely to become ill. Illness was mild with case patients reporting a median of two loose stools (range 1–12).

Conclusions:

C. perfringens is an important cause of both foodborne and nonfoodborne gastroenteritis outbreaks in LTCF, but may be missed due to the often mild nature of illness. This investigation highlights the potential burden of C. perfringens disease among vulnerable LTCF populations. To prevent C. perfringens outbreaks, facilities must adhere to food safety plans and ensure high standards of infection control practice.

Introduction

C lostridium perfringens is an important cause of outbreaks of food poisoning worldwide, with the majority of cases caused by enterotoxin-producing strains of C. perfringens type A (Meer et al., 1997). It is also considered to be a cause of antibiotic-associated (Asha et al., 2006) and sporadic diarrheal illness (Brett et al., 1992). In food poisonings, the most frequently reported food vehicles are protein-based foods of animal origins, such as cooked meats, poultry, stews, meat pies, and gravies. Spores in these foods that have survived normal cooking will then proliferate during long periods of slow cooling or storage at ambient temperatures (Lorber, 2004). People consuming foods with high levels of C. perfringens become ill due to the production of C. perfringens enterotoxin (CPE) in the human gut (Meer et al., 1997). CPE is considered a major virulence attribute and is the principal toxin associated with C. perfringens food poisonings (McClane, 1992).

Illness caused by C. perfringens type A is generally a mild, self-limiting condition characterized by diarrhea and abdominal pain. The incubation period is brief, usually between 7 and 15 hours, with symptoms resolving within 6–24 hours for most cases (Lorber, 2004). Within the setting of long-term care facilities (LTCFs) and other elderly cohorts, C. perfringens food poisoning is usually mild, but can result in an extended spectrum of illness and occasionally death (Jackson et al., 1986; Tallis et al., 1999). Consideration must also be given to the possibility of nonfoodborne causes in such settings (Williams et al., 1985; Jackson et al., 1986). While C. perfringens food poisoning is a commonly suspected cause of outbreaks in Australian LTCFs (Kirk et al., 2010b), such outbreaks are often poorly characterized and prove difficult to investigate thoroughly.

In 2009, a privately operated LTCF advised the Australian Capital Territory Health Protection Service of an outbreak of gastroenteritis affecting residents, which had the appearance of a toxin-mediated etiology. An investigation was commenced to identify the cause of illness in residents and implement public health measures to prevent further cases.

Methods

Case definitions

A possible case was defined as a resident of the LTCF who had one or more acute loose stool episodes between the evenings of 23 July and 27 July 2009. A laboratory-confirmed case was any resident with loose stools in the identified timeframe who provided a fecal specimen that was found to have CPE detected.

Study design

A retrospective cohort study was used to examine risks for residents developing gastroenteritis after an implicated meal. A systematic questionnaire was used to collect data, and care plans, progress notes, and medication sign-off sheets were reviewed.

Study population

The facility provides residential care to ∼130 elderly persons, including those requiring high and low levels of nursing care. Residents are also divided into two service categories, either standard service or extra service, reflecting the level of fees paid to the facility. In particular, extra service residents had a different menu to that offered to residents receiving a standard level of service. Residents were excluded from the study if they were not present in the facility at the time of the outbreak or if their usual bowel habits were characterized by passing of loose stools. This was determined from review of resident's bowel charts for the period May to July 2009.

Data analysis

Data were entered into a Microsoft Excel^® database for descriptive analysis before transfer to STATA^® version 9 for univariate and multivariate analyses, including calculation of relative risks (RR), adjusted relative risks, associated 95% confidence intervals, and p-values using a two-tailed Fisher's exact test. Associations with illness were examined for the following:

the level of service received by the resident;

comorbid factors;

medications used; and

meal size and residents attitude toward food.

Multivariate analysis was performed using backward stepwise logistic regression. Variables considered in the model were those that were biologically plausible or were shown to have potential significance during univariate analysis (p ≤ 0.10). At each step, a likelihood ratio test was performed to assess the explanatory power of the model, with the variable expressing the largest p-value being removed. Significance levels were set at 0.2 for variable removal and 0.1 for variable addition in this model.

Environmental health investigation

The initial inspection conducted at the time the outbreak was notified focused on food-handling and storage procedures. Copies of menu listings, temperature logs, the food safety plan and recent Hazard and Critical Control Point auditing results were requested. Subsequent inspections involved direct observation of lunch preparation, kitchen temperature checks, examination of the hot-hold bain marie, a review of LTCF temperature check documentation, and reinforcement of compliance with food safety protocols.

Laboratory investigations

An Oxoid reverse passive latex agglutination test kit was used to detect the presence of CPE in patient feces. The test was performed in V-well microtiter plates with nonimmune rabbit-globulins as a control. In addition, spore counts were performed on feces samples positive for CPE using 1 in 10 serial dilutions of feces up to 10⁻⁶. The 10⁻³–10⁻⁶ dilutions were then cultured anaerobically at 37°C for 24–48 hours on 5% horse blood agar (Blood Agar Base No.2, Oxoid Ltd). Isolates were identified using the Remel RapID™ ANA II System and long chain fatty acid analysis by gas chromatography (Microbial Identification System; MIDI, Inc.). Counts >10⁶ colony forming units (cfu) per gram of feces were considered indicative of food poisoning by C. perfringens; however, elevated spore levels may be seen in healthy elderly people (Bates and Bodnaruk, 2003).

Results

Descriptive epidemiology

Fifty-two cases of gastroenteritis (attack rate 41%) were identified among residents, including eight standard service residents positive for CPE (Fig. 1). No deaths or hospitalizations occurred among affected residents and no cases were identified among staff. Before the investigation the facility had reported only 26 cases among residents.

FIG. 1.

Epidemic curve detailing an outbreak of Clostridium perfringens gastroenteritis in an Australian long-term care facility.

The median age of cases was 84 years (range 63–97 years old), which was similar to the age of unaffected residents.

Using recognized incubation times for C. perfringens intoxication (Lorber, 2004), and the findings of epidemiological analysis (see below), a single meal—the 23 July lunch—was identified as being most likely responsible for resident's illness. Precise incubation times for individuals could not be calculated due to the paucity of symptom onset detail in clinical notes. The median number of loose stools reported for cases was 2 (range 1–12), whereas for laboratory-confirmed cases it was 3.5 (range 1–8) (Fig. 2). Few other signs and symptoms typical of gastroenteritis were recorded by staff. In addition to diarrhea, one laboratory-confirmed case of C. perfringens gastroenteritis reported accompanying abdominal pain, whereas another possible case also experienced a single episode of vomiting.

FIG. 2.

Stool frequency histogram for C. perfringens gastroenteritis cases in an Australian long-term care facility.

Analytical epidemiology

Univariate analysis

The level of service offered to residents (which reflects meal options) showed that those residing in standard service wings carried an increased risk of developing gastroenteritis (RR 1.8, 95% confidence intervals [CI] 1.0–3.1) (Table 1). Conversely, residents residing in wings receiving extra service meal options had a reduced risk (RR 0.6, 95% CI 0.3–1.0, p = 0.04). Living in one of two wings was associated with illness (Wing C RR 1.6, 95% CI 1.1–2.5; Wing B RR 1.6, 95% CI 1.0–2.4), whereas residents of other wings were less likely to become ill. Importantly, Wings B and C were high care and dementia-specific areas with residents receiving standard service meal options.

Table 1.

Univariate Analysis of Risk Factors for Clostridium perfringens Gastroenteritis Among Elderly Long-Term Care Facility Residents

Risk factor	No. of patients/no. of exposed	Attack rate exposed (%)	No. of patients/no. of not exposed	Attack rate unexposed (%)	RR (95% confidence interval)	p-Value
Male	28/51	55	24/75	32	1.72 (1.14–2.59)	0.01
Resident—Wing A^a	10/40	33	42/96	44	0.76 (0.44–1.33)	0.31
Resident—Wing B^a	15/26	58	37/100	37	1.56 (1.03–2.37)	0.06
Resident—Wing C^a	16/27	59	36/99	36	1.63 (1.08–2.45)	0.03
Resident—Wing D^a	1/6	17	51/120	43	0.39 (0.07–2.47)	0.24
Resident—Wing G^b	4/12	33	46/114	40	0.83 (0.06–2.38)	0.21
Resident—Wing H^b	1/17	6	51/109	47	0.13 (0.02–0.85)	<0.01
Resident—Wing J^b	5/8	63	47/118	40	1.57 (0.88–2.80)	0.21
Std service wing	42/89	47	10/37	27	1.75 (0.98–3.10)	0.04
Extra service wing	10/37	27	42/89	47	0.57 (0.32–1.02)	0.04
Small serving	4/23	17	48/103	47	0.37 (0.15–0.93)	0.01
Medium serving	21/49	43	31/77	40	1.06 (0.70–1.62)	0.77
Large serving	8/18	44	44/108	41	1.09 (0.62–1.92)	0.77
Meals—independent	2/11	18	50/114	44	0.38 (0.11–1.37)	0.07
Meals—some assistance	38/87	44	14/38	37	1.17 (0.73–1.89)	0.51
Meals—full assistance	12/24	50	40/101	40	1.21 (0.75–1.95)	0.45
Meals—any assistance	50/111	45	2/15	13	3.38 (0.91–12.48)	0.02
Diet—normal	31/84	37	21/42	50	0.74 (0.49–1.11)	0.16
Diet—soft	12/24	50	40/102	39	1.27 (0.80–2.03)	0.33
Diet—puree	10/19	53	42/107	39	1.34 (0.82–2.18)	0.28
Diet—soft or puree	22/43	51	30/83	36	1.42 (0.94–2.13)	0.10
Dementia	41/86	48	11/39	28	1.67 (0.97–2.89)	0.05
Underlying gastrointestinal disease	8/32	25	44/93	47	0.51 (0.27–0.97)	0.02
Regular GI system medication	30/80	38	22/46	48	0.78 (0.52–1.19)	0.26
Laxatives <24 hours before symptoms	21/58	36	30/66	45	0.78 (0.51–1.21)	0.26

Standard service residential wing.

Extra service residential wing.

RR, relative risk.

Residents' preferences for meal sizes were assessed, with small meal servings showing an apparently protective effect (RR 0.4 95% CI 0.2–0.9). Assisted feeding was also associated with illness (RR 3.4, 95% CI 0.9–12.5). We did not observe associations between illness and eating pureed or softened meals.

Male residents were more likely to develop illness (RR 1.7, 95% CI 1.1–2.6), as were residents with dementia (RR 1.7, 95% CI 1.0–2.9). Around 70% of residents in the facility suffered from dementia. People with an underlying gastrointestinal condition, such as peptic ulcer disease, diverticulosis, and gastroesophageal reflux disease, were less likely to become ill (RR 0.5, 95% CI 0.3–1.0).

Multivariate analysis

Eleven variables were included in the initial model. After backward stepwise regression, independence with meals, underlying dementia, preference for small meal size, any assistance with meals, and regular use of cardiac medication were excluded. Using this approach cases were more likely to have been male residents and reside in Wings B or C (Table 2).

Table 2.

Multivariate Analysis of Risk Factors for Clostridium perfringens Gastroenteritis Among Elderly Long-Term Care Facility Residents

Risk factor	Adjusted RR (95% confidence interval)	p-Value
Male	2.22 (0.98–4.99)	0.05
Resident—wing H^a	0.07 (0.01–0.70)	0.02
Extra service wing^a	2.29 (0.70–7.52)	0.17
Resident—wing C^b	3.41 (1.16–10.07)	0.03
Resident—wing B^b	3.26 (1.09–9.70)	0.03
Underlying gastrointestinal disease	0.49 (0.18–1.32)	0.16

Extra service residential wing.

Standard service residential wing.

Laboratory investigations

Of the 14 samples submitted for testing, CPE was detected in eight specimens collected from standard service residents. Only two of the CPE-positive specimens contained sufficient fecal material to undertake spore count testing. These specimens contained 6.7 × 10⁶ cfu/g feces and 2.3 × 10⁶ cfu/g feces.

Environmental investigation

Kitchen inspections, including direct observation of lunchtime meal preparation and service, revealed food-handling procedures to be of a high standard. Temperatures for hot meals, including all roasts, chickens, and wet dishes, are routinely checked with a probe thermometer to ensure a minimum core temperature of 75°C, whereas chops and other grilled meats are cooked until juices run clear. Hot foods not immediately assembled are kept in a bain marie at a minimum 60°C. Following assembly, all meals are covered with a lid, before transport (at minimum 60°C) to residential wings using an airline cart system. There was around a 3-hour duration between the start of on-site meal preparation and service to residents. Measurements taken for cook and hot-hold temperatures were within appropriate ranges. However, temperatures for hot foods, including gravy, mashed vegetables, puree, and soup, had not been recorded appropriately in the weeks preceding the outbreak. Hand-washing facilities were satisfactory and staff demonstrated a good understanding of food safety and hygiene.

The 23 July lunch was identified as being the meal most likely to be responsible for residents' illness. Separate meal options were available based on the level of service offered to individual residents. Standard service residents received sweet-and-sour pork, whereas extra service residents received grilled lamb cutlets. Potatoes and mixed vegetables were offered to both groups. All pureed meals were made from vegetables only. No foods from these meals were available for microbiological testing.

Discussion

This investigation in a vulnerable elderly population highlights that epidemic C. perfringens gastroenteritis can be very mild. Residents receiving a standard service meal option had an increased risk of developing gastroenteritis, and the investigation found breaches in temperature recording for implicated meals, despite the facility having a risk-based food safety plan. Although a foodborne cause was suspected, this was unable to be definitively proven, as no leftover foods were available for testing and human isolates did not undergo further characterization by genotyping to differentiate possible foodborne and nonfoodborne isolates.

Despite LTCFs being one of the most common settings for gastroenteritis outbreaks (Kirk et al., 2010a), public health authorities may easily miss both foodborne and nonfoodborne C. perfringens outbreaks. This could be due to their relatively short duration and mild symptoms or because of reluctance to send specimens to specialist enteric reference laboratories for testing. Further, mild outbreaks of toxin-mediated diarrhea may be incorrectly categorized as suspected viral etiology, as norovirus infections commonly cause gastroenteritis outbreaks in LTCFs (Said et al., 2008). However, the presentation of C. perfringens gastroenteritis is markedly different from norovirus, which manifests with obvious person-to-person transmission and an increased proportion of cases with vomiting and illness among staff (Kirk et al., 2010c).

Formal microbiological identification of the causes of outbreaks in LTCF is very important, as the mitigation strategies for foodborne and nonfoodborne C. perfringens outbreaks will differ. A limitation of this investigation was that isolates were not genotyped, which may have assisted in determining if the outbreak was foodborne. Chromosomal CPE-carrying C. perfringens strains have been associated with food poisoning, whereas the plasmid-borne CPE has been thought to be typical of C. perfringens strains associated with antibiotic or sporadic diarrhea (Sparks et al., 2001). However, more recently evidence has shown that plasmid-borne CPE-carrying C. perfringens could be a more common cause of food poisoning than previously thought (Lahti et al., 2008).

A number of case definitions have been proposed for surveillance of infectious gastroenteritis and the investigation of outbreaks in LTCF. For outbreaks, investigators commonly use a case definition of three or more loose stools in a 24-hour period as inclusion criteria, or two or more loose stools for infection control surveillance (McGeer et al., 1991; Majowicz et al., 2008). In this outbreak, few reported cases met these criteria, with only 16 cases having three or more episodes of loose bowel movements. While 70% of reported cases had fewer than three loose bowel episodes, it is possible that elderly persons are more likely to experience prolonged loose bowel activity after enteric infection when compared with the general population.

The investigation revealed a considerable discrepancy between case numbers reported by the LTCF and those identified during the epidemiological investigation. In outbreaks, the attack rate for C. perfringens gastroenteritis in exposed persons typically averages 50%–60% (Lorber, 2004). Among standard service clients in this outbreak, the attack rate was 47%, approximating typical attack rates and supporting the means of case ascertainment as being thorough.

While C. perfringens gastroenteritis is predominantly a foodborne pathogen, more prolonged antibiotic-associated illness may occur (Williams et al., 1985). One hospital outbreak involving 10 laboratory-confirmed cases of antibiotic-associated illness (median age 85 years) revealed a median diarrhea duration of 5 days, with a range of 2–6 stools per day (Williams et al., 1985). In this investigation a low risk of gastroenteritis associated with recent antibiotic usage among residents was observed (RR 0.80, 95% CI 0.31–2.05).

Incident cases of gastroenteritis occurred among residents up to 4 days after the implicated meal, which may have been due to continued use of left-over food. However, this was unlikely given the LTCFs use of a rotating menu system for all meals. C. perfringens transmission via person-to-person contact or contaminated surfaces and fomites (Brett et al., 1992) are other possibilities, especially given the high proportion of dementia cases in the facility (Table 1). A further explanation is that initial symptom onsets for several cases may have been earlier than the dates identified through the retrospective review of clinical notes and stool charts. This is supported by a strong body of evidence showing that conditions documented using observation and interview in LTCFs are frequently not recorded in resident clinical records (Schnelle et al., 2004).

Conclusions

C. perfringens is an important cause of gastroenteritis outbreaks in LTCFs. Although unable to be definitively shown, this investigation raises the possibility this outbreak was of foodborne origin, reinforcing the need for strict adherence to food safety programs in institutional care settings. Although C. perfringens illness may be mild, it is important for public health agencies to investigate these outbreaks to reveal food-handling faults or other causes, such as poor infection control or cleaning practices, that if left uncorrected could result in potentially serious outcomes for frail elderly residents.

Footnotes

Acknowledgments

The authors wish to acknowledge the following groups for their assistance with the investigation: staff from Environmental Health, Australian Capital Territory Health Protection Service for inspections and additional follow-up; staff at the NSW Enteric Reference Laboratory, Institute for Clinical Pathology and Medical Research (ICPMR), Westmead Hospital for C. perfringens testing; LTCF staff for their cooperation with this investigation; and Mr. Martyn Kirk, Senior Epidemiologist, OzFoodNet, Australian Government Department of Health and Ageing for providing comments on the manuscript. OzFoodNet is funded by the Australian Government Department of Health and Ageing.

Disclosure Statement

No competing financial interests exist.

References

Asha

, Tompkins

et al. Comparative analysis of prevalence, risk factors, and molecular epidemiology of antibiotic-associated diarrhea due to Clostridium difficile, Clostridium perfringens, and Staphylococcus aureus. J Clin Microbiol, 2006; 44:2785–2791.

Bates

, Bodnaruk

. Clostridium perfringens. Foodborne Microorganisms of Public Health Significance. Hocking

. Sydney: Australian Institute of Food Science and Technology Incorporated, 2003; 479–504.

Brett

, Rodhouse

et al. Detection of Clostridium perfringens and its enterotoxin in cases of sporadic diarrhoea. J Clin Pathol, 1992; 45:609–611.

Jackson

, Yip-Chuck

et al. Diagnostic importance of Clostridium perfringens enterotoxin analysis in recurring enteritis among elderly, chronic care psychiatric patients. J Clin Microbiol, 1986; 23:748–751.

Kirk

, Fullerton

et al. Surveillance for outbreaks of gastroenteritis in long-term care facilities, Australia, 2002–2008. Clin Infect Dis, 2010a. 51:907–914.

Kirk

, Moffatt

et al. The burden of infectious gastroenteritis in elderly residents, staff of long-term care facilities, Australia. Infect Control Hosp Epidemiol, 2010b. 31:860–863.

Kirk

, Veitch

et al. Gastroenteritis, food-borne disease in elderly people living in long-term care. Clin Infect Dis, 2010c. 50:397–404.

Lahti

, Heikinheimo

et al. Clostridium perfringens type A strains carrying a plasmid-borne enterotoxin gene (genotype IS1151-cpe or IS1470-like-cpe) as a common cause of food poisoning. J Clin Microbiol, 2008; 46:371–373.

Lorber

. Gas gangrene and other Clostridium-associated diseases. Mandell, Douglas and Bennett's Principles and Practice of Infectious Diseases. Mandell

, Bennett

, Dolin

. Philadelphia: Elsevier, 2004; 2549–2561.

10.

Majowicz

, Hall

et al. A common, symptom-based case definition for gastroenteritis. Epidemiol Infect, 2008; 136:886–894.

11.

McClane

. Clostridium perfringens enterotoxin: structure, action and detection. J Food Saf, 1992; 12:237–252.

12.

McGeer

, Campbell

et al. Definitions of infection for surveillance in long-term care facilities. Am J Infect Control, 1991; 19:1–7.

13.

Meer

, Songer

et al. Human disease associated with Clostridium perfringens enterotoxin. Rev Environ Contam Toxicol, 1997; 150:75–94.

14.

Said

, Perl

et al. Healthcare epidemiology: gastrointestinal flu: norovirus in health care and long-term care facilities. Clin Infect Dis, 2008; 47:1202–1208.

15.

Schnelle

, Bates-Jensen

et al. Accuracy of nursing home medical record information about care-process delivery: implications for staff management and improvement. J Am Geriatr Soc, 2004; 52:1378–1383.

16.

Sparks

, Carman

et al. Genotyping of enterotoxigenic Clostridium perfringens fecal isolates associated with antibiotic-associated diarrhea and food poisoning in North America. J Clin Microbiol, 2001; 39:883–888.

17.

Tallis

, Ng

et al. A nursing home outbreak of Clostridium perfringens associated with pureed food. Aust N Z J Public Health, 1999; 23:421–423.

18.

Williams

, Piper

et al. Diarrhoea due to enterotoxigenic Clostridium perfringens: clinical features and management of a cluster of ten cases. Age Ageing, 1985; 14:296–302.