Skip Navigation

Hospital-Acquired Invasive Group A Streptococcal Infections in Ontario, Canada, 1992–2000

  1. N. Daneman1,
  2. A. McGeer1,2,
  3. D. E. Low1,2,
  4. G. Tyrrell6,
  5. A. E. Simor1,3,
  6. M. McArthur2,
  7. B. Schwartz7,
  8. P. Jessamine5,
  9. R. Croxford4,
  10. K. A. Green5, and
  11. Ontario Group A Streptococcal Study Group
  1. 1University of Toronto, University of Ottawa, Ontario
  2. 2Mount Sinai Hospital, University of Ottawa, Ontario
  3. 3Sunnybrook & Women's College Health Sciences Centre, University of Ottawa, Ontario
  4. 4Institute for Clinical Evaluative Science, Toronto, Ontario
  5. 5Ottawa Hospital and University of Ottawa, Ontario
  6. 6National Centre for Streptococcus, Edmonton, Alberta
  7. 7Centers for Disease Control and Prevention, Atlanta, Georgia
  1. Reprints or correspondence: Dr. A. McGeer, Rm. 1460, Mount Sinai Hospital, 600 University Ave., Toronto, Ontario, Canada, M5G 1X5 (amcgeer{at}mtsinai.on.ca).

Abstract

Background. A significant proportion of invasive group A streptococcal infections are hospital acquired. No large, prospective studies have characterized this subgroup of cases and evaluated the risk of transmission in hospitals.

Methods. We conducted prospective, population-based surveillance of invasive group A streptococcal infections in Ontario, Canada, from 1992 to 2000. Epidemiologic and microbiologic investigations were conducted to identify cross-transmission.

Results. We identified 291 hospital-acquired cases (12.4%) among 2351 cases of invasive group A streptococcal disease. Hospital-acquired invasive group A streptococcal infections are heterogeneous, including surgical site (96 cases), postpartum (86 cases), and nonsurgical, nonobstetrical infections (109 cases). Surgical site infections affected 1 of 100,000 surgical procedures and involved all organ systems. Postpartum infections occurred at a rate of 0.7 cases per 10,000 live births and exhibited an excellent prognosis. Nonsurgical, nonobstetrical infections encompassed a broad range of infectious syndromes (case-fatality rate, 37%). Nine percent of cases were associated with in-hospital transmission. Transmission occurred from 3 of 142 patients with community-acquired cases of necrotizing fasciitis requiring intensive care unit (ICU) admission, compared with 1 of 367 patients with community-acquired cases without necrotizing fasciitis admitted to the ICU and 1 of 1551 patients with other cases (P < .001). Fifteen outbreaks were identified; 9 (60%) involved only 2 cases. Hospital staff were infected in 1 of 15 outbreaks, but colonized staff were identified in 6 (60%) of 10 investigations in which staff were screened.

Conclusions. Presentation of hospital-associated invasive group A streptococcal infections is diverse. Cross-transmission is common; illness occurs in patients but rarely in staff. Isolation of new cases of necrotizing fasciitis and intervention after a single nosocomial case may also prevent transmission.

After nearly a century of decreases, the past 2 decades have witnessed a resurgence of invasive group A streptococcal (GAS) infections. Several hospital and population-based prospective surveys have described the epidemiology and clinical features of this reemerging disease [139]. A significant and consistent proportion of these infections have been associated with health care. However, there has been no systematic study of this important subgroup of GAS infections.

The risk of cross-transmission is an important characteristic of GAS infection. Many outbreaks of infection have been described, with a substantial proportion of them occurring in hospitals [40106]. Because of this, in 2002, the Centers for Disease Control and Prevention (CDC) published guidelines for the surveillance, investigation, and chemoprophylaxis of GAS infections in postsurgical and postpartum settings [40]. Because of a scarcity of information regarding the epidemiology of GAS infection in hospitals, these recommendations were made on the basis of expert consensus.

Prospective, population-based surveillance of all invasive GAS infections was conducted in Ontario from 1 January 1992 through 31 December 2000 [2, 6, 10]. The purpose of this study was to describe the epidemiology of hospital-associated invasive GAS infections in Ontario and to evaluate the risk of cross-transmission in hospitals.

Methods

Population-based surveillance. Prospective, population-based surveillance of all invasive GAS infections was conducted in the province of Ontario, Canada, from 1 January 1992 through 31 December 2000, as described elsewhere [2, 6, 10]. Briefly, all microbiology laboratories serving Ontario hospitals, as well as the largest outpatient laboratory, telephoned the central study office whenever GAS was isolated from a sterile-site specimen. Annual laboratory audits were conducted to ensure complete case ascertainment.

A case of invasive GAS infection was defined as illness associated with the isolation of GAS from a normally sterile body site. Cases were defined as hospital acquired if disease was neither present nor incubating at the time of hospital admission [107]. Surgical site infections were defined on the basis of National Nosocomial Infection Surveillance System criteria and were deemed to be nosocomial if they occurred ⩽30 days after the operative procedure or ⩽1 year after an operative procedure involving implanted material [107]. Peripartum cases were deemed to be nosocomial unless signs or symptoms of disease were evident before hospital admission.

Cross-transmission was defined as ⩾2 cases of disease that were epidemiologically linked and were caused by isolates of the same M and T type that had indistinguishable PFGE patterns. In outbreaks with >1 hospital-acquired case, invasive nosocomial cases were defined as index cases if they were the first invasive case in the outbreak. Necrotizing fasciitis (NF) and streptococcal toxic shock syndrome (STSS) were defined as described elsewhere [2, 6, 10]. Surgical procedures were coded in accordance with the International Classification of Diseases, 9th Revision, Clinical Modification. The total number of live births in the Province of Ontario during 1992–2000 was obtained from Statistics Canada; the number of surgical procedures was obtained by searching hospital discharge records for the occurrence of a Canadian Classification of Diagnostic, Therapeutic, and Surgical Procedures code indicating a surgical procedure in the discharge record (list of codes available on request).

Investigation of disease transmission. When a nosocomial case of GAS was identified, the study group contacted the hospital's infection-control practitioner and recommended that the hospital consider a limited investigation, consisting of identifying close contacts among patients, staff, and households; asking these contacts about symptoms that may be due to GAS infection; and obtaining throat swabs and possibly vaginal and rectal swabs for GAS culture. Study staff were available for consultation during these investigations, and any isolates obtained were sent to the Group A Streptococcal Study laboratory.

Microbiologic and laboratory methods. Clinical isolates were confirmed to be Streptococcus pyogenes using standard techniques. M protein typing, T agglutination typing, and PCR detection of streptococcal pyrogenic exotoxin genes A and C (speA and speC) were performed at the National Centre for Streptococcus (Edmonton, Alberta) [108111]. PFGE was performed as described elsewhere [112].

Statistical analysis. Surveillance data were entered in duplicate and analyzed in SAS for Windows, version 8 (SAS Institute). Univariate analyses assessed differences in proportions by the χ2 test or Fisher's exact test, and differences in continuous variables were assessed by the Student's t test or the Wilcoxon rank sum test. Backward stepwise multivariable logistic regression, including variables with a P value of <.10 in the univariate analysis, was conducted to assess risk factors for mortality. All P values were 2-tailed and were not adjusted for multiple comparisons.

Results

During the study surveillance period, 2351 cases of invasive GAS disease were identified. of these, 291 (12.4%) were hospital acquired. Patients with hospital-acquired invasive GAS infection had a median age of 47.9 years (range, 1 day to 97 years), and 41.9% had a chronic underlying illness (table 1). NF and STSS were present in 5.9% and 8.1% of patients with nosocomial cases, respectively. The case-fatality rate was 17.0%.

Table 1
Table 1

Demographic and clinical characteristics and outcomes of invasive group A streptococcal disease in Ontario, Canada, 1992–2000.

Nosocomial invasive GAS infections were classified as surgical site infections (n = 96), postpartum infections (n = 86), and “other” (n = 109). The groups differed substantially with respect to sex and age distributions, rates of underlying illness, clinical outcomes, and M serotype distributions (tables 1 and 2).

Table 2
Table 2

Distribution of common M serotypes causing invasive group A streptococcal disease in Ontario, Canada, 1992–2000.

Surgical site infections. The 96 cases of postsurgical invasive GAS infection occurred among 9,078,030 surgical admissions in the province of Ontario, for a rate of 1.1 cases per 100,000 surgical admissions . These infections encompassed the entire range of surgical procedures, including those involving the digestive tract (28%); musculoskeletal system (24%); cardiovascular system (9%); nervous system (11%); skin and soft tissue (9%); female and male genital systems (5% and 3%, respectively); lymphatic system (4%); ear, nose, and throat (3%); endocrine system (1%); eyes (1%); respiratory system (1%); and urinary tract (1%).

The median time between surgery and the day that the sterile-site sample used for the culture yielding S. pyogenes was drawn was 5 days (range, <24 h to 261 days). Twenty-two (23%) of 96 infections were associated with implanted material; these infections had a later onset (median, 30 days; range, 1–261 days; P < .001). There was no association between morbidity or mortality and the time elapsed between surgery and the onset of illness (data not shown).

Postpartum infections. During our surveillance period, there were 86 postpartum invasive GAS infections among the province's 1,269,722 live births (0.68 cases per 10,000 live births). Patients with postpartum infection were less likely than others to have chronic underlying illness, to have infection complicated by NF or STSS, to require surgery or intensive care unit (ICU) admission, and to die (table 1). When women with postpartum invasive GAS infection were compared with healthy women of childbearing age (15–45 years) with community-acquired infection, women with postpartum infection were less likely to have STSS (2.4% vs. 15.0%; P = .001) or NF (0% vs. 12.3%; P = .001) and were less likely to die (case-fatality rate, 1.2% vs. 6.2%; P = .11).

Two infants of infected mothers had noninvasive, laboratory-confirmed GAS infection at 12 days of age and 3 weeks of age (one had pustulosis, and the other had cellulitis). No other maternal infections were associated with infections in infants. There were 5 early-onset neonatal infections (incidence, 3 cases per million live births) during the surveillance period.

Nonsurgical, nonobstetrical nosocomial GAS infection. of the 291 hospital-acquired invasive GAS infections, 109 were neither postsurgical nor postpartum in origin. These infections occurred after a median length of hospital stay of 10.5 days (range, 2 days to >1 year). The most common syndromes were primary bacteremia (33% of subjects), nonnecrotizing soft-tissue infection (32%), lower respiratory tract infection (21%), and NF (6%). Other, less common manifestations were adenitis, septic arthritis, gynecologic infection, osteomyelitis, peritonitis, and pharyngitis. of the 35 cases of non-NF soft-tissue infection, 32 were associated with preexisting skin breakdown (intravenous device insertion sites, 16 cases; other devices, such as gastrostomy tube or tracheostomy tube, 6 cases; chronic ulcers, 5 cases; traumatic lesions, 2 cases; burns, 1 case; and other skin lesions, 2 cases). Patients with these infections were older, were more likely to have a chronic underlying illness, and were more likely to die than were patients with postsurgical infection, postpartum infection, or community-acquired disease (table 1).

Mortality. The overall 30-day case-fatality rate was 17.0%; 33 (67%) of 49 deaths occurred <72 h after the sample for the culture that yielded GAS was obtained. In multivariable analysis, only the presence of STSS, older age, underlying illness, and having a nonobstetrical, nonsurgical infection were significantly associated with mortality (table 3). Early provision of antibiotics was associated with improved survival in univariate analysis, but not in multivariable analysis. Patients with nonsurgical, nonobstetrical infections were somewhat less likely to have received antibiotics within 24 h after the onset of symptoms than were patients with surgical or obstetrical infections (68% of the former but 80% of the latter received antibiotics within 24 h after onset of symptoms; P = .08).

Table 3
Table 3

Factors influencing the risk of mortality due to nosocomial, invasive group A streptococcal infection.

Risk of cross-transmission. Overall, 20 invasive GAS episodes were associated with in-hospital cross-transmission to ⩾1 other patient. The risk of there being at least 1 associated hospital-acquired invasive GAS infection was 5 (0.2%) of 2060 after the admission of a patient with a community-acquired case and 15 (5.2%) of 291 after the occurrence of an index nosocomial case (P < .001). The risk of transmission from patients with community-acquired cases varied according to the type of infection and the admitting ward. Transmission was most common from patients with community-acquired cases of NF who required ICU admission (3 [2.1%] of 142), somewhat less common for patients with nonnecrotizing fasciitis admitted to the ICU (1 [0.27%] of 367; P = .07), and least common from patients with community-acquired cases who did not require ICU admission (1 [0.06%] of 1551; P < .0001, for comparison with patients in the ICU with NF; P = .34, for comparison with patients in the ICU with other diagnoses).

After the occurrence of an index hospital-acquired invasive GAS infection, the probabilities of a second invasive GAS infection in the same hospital were 4.2%, 6.3%, 12%, 24%, and 40% within 1 week, 1 month, 3 months, 6 months, and 12 months, respectively. The proportion of second cases that were truly related to the first case decreased from 67% (8 of 12 cases), if the cases occurred within 1 week of the first, to 0% (0 of 96 cases), if they occurred >1 week later.

Overall, 27 (9.3%) of the 291 invasive hospital-acquired GAS infections were linked to other cases of GAS disease. Fifteen hospital outbreaks were identified: most (60%) involved only 2 cases of laboratory-confirmed disease (range, 1–10 cases), and the median duration was 7 days (range, 1–13 days). There was only 1 outbreak of disease in which health care workers exhibited symptomatic GAS infection (6 staff members had pharyngitis, 3 cases of which were culture confirmed) [46]. However, in the 10 outbreak investigations in which health care workers were screened, 6 (60%) yielded at least 1 specimen from an asymptomatic health care worker found to have a GAS isolate of the same M and T type as other outbreak isolates and to be indistinguishable by PFGE.

Cases that were part of outbreaks of infection were distributed among the different nosocomial categories. of the 27 cases (15 index cases), 6 (3 index cases) were surgical site infections, 10 (6 index cases) were postpartum infections, and 11 (6 index cases) were nonsurgical, nonobstetrical infections. Neither index cases nor secondary hospital-acquired cases differed from sporadic, hospital-acquired cases in terms of underlying illness, timing of onset after hospital admission, or M and T type (data not shown).

Discussion

This study provides the first long-term, population-based analysis of hospital-acquired invasive GAS infections. We found that 12.4% of all infections were hospital acquired, which is in keeping with the values reported in prior studies of invasive GAS infections (median rate, 13.5%; range, 0%–66%) [139]. The apparent difference in the proportion of overall cases that were surgical site infections reported in our data (4.1% of all GAS infections) and reported by the US Active Bacterial Core (ABC) surveillance system (1.5% of all GAS infections [1]) is related to definitions. The ABC surveillance system considered infections to be related to surgical procedures only if they occurred within the first 7 days after the operation; according to our data, slightly less than one-half of the GAS surgical site infections met this standard.

Postpartum invasive GAS infections occurred at a rate of 0.7 cases per 10,000 live births, a rate very similar to that reported in 2 recent studies from the United States [113] and the United Kingdom [114]. Our results regarding postpartum infection were strikingly similar in other ways to those from surveillance conducted across several US counties [113]. Despite differences in both time and geography, the proportion of cases that were postpartum infections (2.2% vs. 3.7%), case-fatality rates (3.4% vs. 1.2%) and M serotype distribution (M28 was more common than M1, M4, and M11) reported in our study were very similar to those presented by the US ABC surveillance system. In the US study, neonatal disease was not reported. In contrast to our data, in which neonatal infection is extremely rare and is usually not severe, Barnham and Weightman [114] from the United Kingdom reported that 4 of 6 mothers with invasive postpartum GAS infection had infants who were also affected (3 of whom died), and Udagawa et al. [115] reported from Japan that neonates were infected in 10 of 38 identified episodes of GAS toxic shock syndrome. Whether these differences are due to differences in infecting strains, host susceptibility, or management of perinatal care is not clear.

In Ontario, postpartum invasive GAS infections were less severe than invasive GAS infections in healthy women of child-bearing age. Thus, although the better prognosis of postpartum invasive GAS infection is in part due to the young age and lack of underlying illness among this patient population, it appears also to be associated with other factors, perhaps related to infecting strains or the pathogenesis of postpartum disease. The preponderance of M28 strains does not explain the difference. In our surveillance, overall, M28 strains were not associated with a decreased case-fatality rate [2, 10]; neither of the fatal cases in Barnham and Weightman's [114] surveillance had isolates available for testing. The greater prevalence of M28 strains in cases of postpartum infection, compared with other GAS infections, is, however, consistent with previous demonstrations that M28 infections cluster among children aged <10 years and adults aged 30–40 years [111, 116]. Serotype M28 strains may have a tropism for perineal colonization and infection, because M28 has also been shown to be the predominant M type in studies of perianal or perineal streptococcal infection in children [117, 118]. M28 strains frequently express surface protein R28, which is related to cell-surface molecules in type V group B streptococci and which enhances binding to cervical epithelial cells [119].

Given the evidence that most postpartum GAS infections arise from GAS carriage in the genital tract [113, 114, 120], it is reasonable to ask whether the identification and treatment of vaginal GAS carriage late in the course of pregnancy would prevent these maternal infections. Vaginal GAS carriage in pregnant women occurs at a rate of ∼3 cases per 10,000 pregnant women, a rate ∼4-fold greater than reported rates of invasive disease [121]. Although screening for GAS carriage on its own would not be cost-effective, it is reasonable to speculate that the addition of GAS screening to existing recommended, universal GBS screening protocols for rectovaginal swabs late in the course of pregnancy would be logistically feasible and relatively inexpensive [122]. Surveillance studies are needed to identify whether women who are colonized with GAS (as determined with postpartum specimens submitted for screening for group B streptococci) are at higher risk of acquiring postpartum infections.

Nearly 40% of hospital-acquired invasive GAS infections were neither postsurgical nor postpartum in origin. These infections span a broad range of clinical syndromes, can occur at any time during hospitalization, and have a strikingly high case-fatality rate. Our study is the first to suggest improved survival with earlier antibiotic provision after the onset of symptoms in serious nosocomial infections, although earlier provision of appropriate antibiotics has previously been shown to be important in patients with community-acquired pneumonia and sepsis requiring ICU admission [123125]. This observation reinforces the need for physicians to be alert to the development of nosocomial infections in their patients and to the need for early treatment of the subgroup of elderly and compromised patients at highest risk for mortality.

Current guidelines recommend contact isolation for patients with soft-tissue infection only if wound drainage cannot be contained [126, 127]. In our surveillance, secondary invasive disease associated with community-acquired GAS infection occurred rarely. However, 2.2% of cases of NF requiring ICU admission were associated with secondary invasive disease. Similarly, necrotizing soft-tissue infection occurred in 3 of 4 published reports of transmission from patients with community-acquired disease [42, 64, 66, 101]. Diagnosis of NF can often be made clinically; we believe that, in the absence of further data, contact precautions should be used for patients with NF who require ICU admission until the presence of GAS is ruled out. The majority of transmissions occur with exposure during the first 48 h of the hospital stay.

The recently published CDC consensus guidelines for the prevention of postpartum and postsurgical cases of GAS disease recommend enhanced surveillance and storage of the GAS isolate when 1 case is identified, as well as enhanced surveillance and typing if 2 GAS isolates are identified in a 6-month period [40]. Our data suggest that, if 2 cases occur >1 month apart, they can reasonably be treated as 2 individual cases until typing results are available. Two cases that occur within 1 week of each other, however, are likely to be related. Because of the short interval between cases in outbreaks, epidemiologic investigation (including screening of health care workers) should be initiated before typing findings are available. Our data also suggest that investigations should not be limited to obstetrical and postsurgical cases, because all types of hospital-acquired infections have the same risk of being associated with an outbreak.

The CDC guidelines focus on preventing outbreaks of infection but do not consider whether the first transmitted case of GAS infection can be prevented [40]. In contrast to outbreaks that result in publications about cases, the majority of episodes of hospital transmission detected by surveillance involve only a single secondary case, with no more than a few days between cases. Prevention of the majority of hospital-transmitted GAS infections thus requires very rapid investigation and intervention once a single hospital-acquired case is identified. Our data do not allow us to assess whether the usually small number of cases in outbreaks in our study was the result of early intervention or the extent to which rapid investigation might prevent a second case. However, hospital-acquired GAS disease, while uncommon, is associated with substantial morbidity and mortality, and 10% of cases are associated with transmission. We believe that the currently available data support a recommendation for the initiation of a limited investigation after a single case to ensure that second cases are promptly identified and to ensure that no health care workers in close contact with the patient have been ill or are colonized with S. pyogenes.

In summary, our study provides the first prospective, population-based analysis of hospital-acquired invasive GAS infection. We have identified nonsurgical, nonobstetrical infections as an important category of disease, and we have highlighted areas for future investigation and intervention, including the potential benefit of identifying GAS, in addition to group B streptococci, when screening pregnant women for carriage at 36–37 weeks, as well as the apparent importance of early therapy of nosocomial GAS infections. Our results also offer insight into the prevention of cross-transmission. The occurrence of 2 nosocomial cases within 1 week in 1 hospital should be treated as an outbreak until typing results are available. Contact isolation of community-acquired cases of NF and initiation of outbreak investigations after a single nosocomial case may be of benefit in preventing cross-transmission.

Acknowledgments

We are grateful to the many patients, physicians, microbiology technologists, infection-control practitioners, and public health unit staff who have collaborated in the surveillance and outbreak investigations across Ontario. We thank MDS Laboratories for assistance with specimen transport.

Potential conflicts of interest. All authors: no conflicts.

  • Received February 17, 2004.
  • Accepted March 7, 2004.

References

| Table of Contents