Invasive Group A streptococcal infections in Florida.

Background: Several previous studies of invasive Group A streptococcal (GAS) disease have been hindered by small sample sizes ([less than or equal to] 100 patients) and limited generalizability.

Methods: We conducted a population-based study of invasive GAS disease. The objectives of the study were to describe the clinical features of individuals who were hospitalized for invasive GAS disease and to identify risk factors for hospital mortality. The cases were 257 patients who were hospitalized throughout Florida during a 4-year period and reported to the Florida Department of Health. Logistic regression was used to calculate adjusted odds ratios (OR) for mortality and 95% confidence intervals (CI).

Results: The overall mortality was 18% (41 of 228). Admission into an intensive care unit was a strong predictor of mortality (OR, 20.41; 95% CI, 6.41-64.96). Treatment with clindamycin reduced mortality in patients who had necrotizing fasciitis (OR, 0.11; 95% CI, 0.01-0.89) but not in patients who did not have necrotizing fasciitis (OR, 1.01; 95% CI, 0.31-3.33).

Conclusion: Clindamycin reduces mortality in patients with invasive GAS disease who have necrotizing fasciitis.

**********

Group A Streptococcus (GAS), or Streptococcus pyogenes, is one of the most important bacterial pathogens of humans. (1) GAS can cause several conditions including pharyngitis, impetigo, scarlet fever, erysipelas, and puerperal fever. (2) GAS can also cause invasive infections that can be life-threatening such as septicemia, streptococcal toxic shock syndrome, and necrotizing fasciitis. (2) In 1999, approximately 10,000 cases of invasive GAS infection occurred in the United States. (3) This incidence is twice that of the annual national incidence of Escherichia coli O157:H7 infections and four times the annual incidence of meningococcal disease. (3) An impressive change in the epidemiology and severity of invasive GAS infections occurred in the 1980s. (4,5) The recent increase in the incidence of invasive GAS infections, after approximately 50 years of relatively benign disease, indicates there has been a major alteration in GAS virulence properties. (6)

Invasive GAS infections became a reportable condition in the State of Florida in mid-1996. (7) Using surveillance data from the Florida Department of Health, we conducted a retrospective population-based study of invasive GAS infections. The objectives of this epidemiologic study were to describe the clinical features of individuals who were hospitalized for invasive GAS disease throughout Florida between August 1996 and August 2000 and to identify risk factors for hospital mortality.

Several previous studies of risk factors for hospital mortality suffered from small sample sizes ([less than or equal to]100 cases) and only included patients seen at one or two hospitals. (8 10) The external validity, that is, generalizability, of these studies may be limited. A study that only reviews invasive GAS cases that are seen at one hospital may have a sample of patients that is not representative of the population of invasive GAS patients. For example, a 10-year study of GAS bacteremia at a university hospital in Madrid, Spain, found that 62% of the cases were IV drug users. (8) Bacteremia in IV drug users can be a benign disease. (8) Furthermore, IV drug users may have a higher prevalence of human immunodeficiency virus infection than non-IV drug users. Infection with human immunodeficiency virus may reduce the risk of streptococcal toxic shock syndrome by reducing the physiologic response to streptococcal superantigens. Streptococcal toxic shock syndrome is the result, in part, of the reactivity of the immune system to the GAS superantigens. (8) Furthermore, the effects of clindamycin and other antibiotics have not been studied extensively in humans using epidemiologic study designs. (11,12)

Patients and Methods

Case Definition

Invasive GAS disease became a reportable disease in Florida in mid-1996. The surveillance case definition of invasive GAS issued by the Florida Department of Health to the Florida county health departments is as follows: isolation of GAS from a normally sterile site (eg, blood, cerebrospinal fluid, joint fluid, pleural fluid, or pericardial fluid) and a clinically compatible presentation. The definition of a clinically compatible presentation is as follows: one of several entities, including pneumonia, bacteremia in association with cutaneous infection (eg, cellulitis, erysipelas, or infection of a surgical of nonsurgical wound), deep soft-tissue infection (eg, myositis or necrotizing fasciitis), meningitis, peritonitis, osteomyelitis, septic arthritis, postpartum sepsis (ie, puerperal fever), neonatal sepsis, and nonfocal bacteremia. Between 1996 and 2000, the Florida county health departments also reported cases of necrotizing fasciitis to the Florida Department of Health if GAS was isolated from a nonsterile site. These cases of necrotizing fasciitis were included in this study. Previous investigations of invasive GAS disease, for example, the Active Bacterial Core Surveillance program operated by the Centers for Disease Control and Prevention, have included cases of necrotizing fasciitis even if GAS was not isolated from a normally sterile site. (13)

For each case of invasive GAS that is reported by a county health department to the Florida Department of Health, the epidemiology staff at the reporting county health department must complete a three-page case report form. The case report form contains clinical information including the type of cultures, such as blood cultures, that are positive for GAS, treatment received during the hospital stay, and hospital mortality. The case report form is usually completed after reviewing the patient's hospital medical record. The majority (95%) of invasive GAS cases reported to the Florida Department of Health were hospitalized. The Florida Department of Health's Bureau of Epidemiology and the University of South Florida Medical Institutional Review Board approved this study protocol.

Inclusion and Exclusion Criteria 

A total of 270 case report forms were reviewed for inclusion in this study. These cases were reported to the Florida Department of Health between July 1996 and August 2000. Five percent of these cases (n = 13) were not hospitalized. This study only included hospitalized patients. The 13 patients who were not hospitalized were excluded from the study. The final sample size was 257, of whom 41 patients died and another 29 patients had unknown mortality status. The group of 29 patients whose mortality status was unknown was excluded from the analysis of risk factors for hospital mortality.

The initial descriptive analysis examined several variables, including overall mortality, demographic variables, and the frequency distribution of disease (eg, primary bacteremia, pneumonia, meningitis) caused by GAS. The design of the analytic study was a retrospective cohort. The clinical and epidemiologic literature was reviewed to identify factors that may be associated with hospital mortality and whose associations would be biologically plausible.

The following exposure variables were studied: sex, race, age, presence of necrotizing fasciitis, admission to an intensive care unit, presence of diabetes, treatment with the antibiotic clindamycin during the hospital stay, treatment with [beta]-lactam antibiotics during the hospital stay, and treatment with fluoroquinolone antibiotics during the hospital stay. The outcome in each analysis was hospital mortality. A [beta]-lactam antibiotic was defined as penicillin, any semisynthetic penicillin, or a cephalosporin. Age was originally a continuous variable; however, the risk of hospital mortality did not increase linearly with age and therefore this variable was converted to a categorical variable ([greater than or equal to]55 yr, <55 yr). All of the cases with a known race, except one, were identified as white or black (1 Asian, 50 black, 200 white, 6 unknown race). The case that was not identified as white or black was excluded from the logistic regression analyses.

Statistical Analysis

Data were primarily analyzed by the SAS system statistical package, version 8.00 for Windows (SAS Institute, Inc, Cary, NC). Logistic regression was used to examine the associations between the nine exposure variables listed above and the risk of hospital mortality. (14) All nine variables were entered into one full model. Each odds ratio (OR) was adjusted for any confounding by the remaining variables. Crude and adjusted ORs with 95% confidence intervals (CI) were calculated in the traditional manner.

Finally, a likelihood ratio test was performed to determine whether there was a significant interaction between the clindamycin variable and the necrotizing fasciitis variable. (15,16) The full model contained all of the independent variables plus a clindamycin-by-necrotizing fasciitis interaction term. The reduced model contained all of the independent variables except the interaction term. The significance level used was 0.10.

Multiple logistic regression analyses can only be performed if every patient record has complete data for all of the independent variables and/or the dependent variable. After deleting records that had missing values for any of the independent variables and/or the dependent variable, there were 174 patient records available for logistic regression analyses. An additional 21 patients had complete data on their independent variables but had missing values for their outcome. A standard technique was used to predict the outcomes of these 21 patients. (16) A logistic regression prediction equation was developed using the group of 174 patients. A goodness-of-fit [X.sup.2] test indicated that this prediction model fit the observed data well. The final sample size available for logistic regression was 195 (174 + 21). Results from the smaller group (n = 174) are available upon request from the first author (ZDM).

Results

A total of 16 cases of invasive GAS disease were reported to the Florida Department of Health in 1996 (data not shown). This frequency increased steadily to 152 cases in 2000. The crude incidence in 2000 was approximately 1 per 100,000 population.

The clinical and demographic profiles of the hospitalized cohort are shown in Table 1. The sample size was 257 patients; however, several case report forms had missing values for one or more of the variables. All of the patients except one were black or white; the ethnicity of the other patient was Asian/Pacific Islander. A large proportion of the patients (83%) had been treated with one or more [beta]-lactam antibiotics. The median age was 52.0 years. Approximately 54% (140 of 257) of the patients were younger than 55 years of age, and their mortality rate was 12.4% (16 of 129; data not shown). The overall mortality rate was 18%.

Table 2 shows the conditions that were caused by GAS in the hospitalized cohort. The skin was the most common focus of infection: 38% of the patients had cellulitis, 18% had necrotizing fasciitis, and 4% had gangrene. The conditions are not mutually exclusive; that is, several patients presented with multiple conditions. Twenty-two percent of the patients had a primary bacteremia. These patients had blood cultures that were positive for GAS but did not have a focus of infection. More than half of the patients (63%) had a bacteremia secondary to another condition, such as pneumonia or pharyngitis.

Table 3 shows the results of logistic regression analyses performed on 195 patient records. Admission into an intensive care unit was the strongest predictor of mortality. After adjusting for the remaining variables, the OR for mortality associated with the intensive care unit variable was 20.41 (95% CI, 6.41-64.96). Black race seemed to strongly protect against mortality. After adjustment, male sex appeared to be a protective factor (OR, 0.41; 95% CI, 0.17-1.02). This result approached statistical significance (P = 0.055). The risk of mortality increased with age. Compared with patients who were 54 years of age or younger, patients who were 55 years of age or older were approximately four times as likely to die in the hospital (OR, 3.80; 95% CI, 1.46-9.93).

Three antibiotic variables were examined. The adjusted OR associated with treatment with a [beta]-lactam antibiotic was 0.40. This result was not statistically significant at the 0.05 level. Of the 195 patients, 54 (28%) received a fluoroquinolone antibiotic (data not shown). Treatment with a fluoroquinolone antibiotic was not beneficial (OR, 1.02; 95% CI, 0.39-2.69). The majority (81%) of these 54 patients received levofloxacin. The remaining patients received ciprofloxacin or ofloxacin. None of the patients received more than one fluoroquinolone.

There was a statistically significant interaction between clindamycin and necrotizing fasciitis. Table 4 shows ORs for hospital mortality when clindamycin is the exposure variable. Treatment with clindamycin protected against mortality in patients who had necrotizing fasciitis (OR, 0.11; 95% CI, 0.01-0.89) but not in patients who did not have necrotizing fasciitis (OR, 1.01; 95% CI, 0.31-3.33).

Discussion

This study reports the clinical profile of invasive GAS disease in Florida. Furthermore, this investigation identified risk factors for hospital mortality in a cohort of patients hospitalized for invasive GAS disease between 1996 and 2000.

The most common focus of infection in this study was skin and soft tissue. Several previous studies of invasive GAS disease also reported the same result. (10,17-19) The proportion of patients who had primary bacteremia in the current study was similar to the proportion in two other studies. (18,19) The prevalence of necrotizing fasciitis in the current study was higher than what has been previously reported. (17-19) This inconsistency could be a result of several factors. For example, the prevalence of necrotizing fasciitis may truly be high in cases of invasive GAS disease hospitalized in Florida. Another explanation is that these diagnoses of necrotizing fasciitis are not accurate. The chart abstractor from the county health department may have miscoded this variable on the case report form or misinterpreted clinical information in the chart and decided to code the necrotizing fasciitis variable as "yes."

A small number (n = 21) of patient records had complete data on the nine independent variables that were studied but were missing values for the outcome variable (hospital mortality). The missing values were predicted using a logistic regression equation. The original group contained 174 patient records. Results from the two groups, that is, the group with 174 records (data not shown) and the one with 195 records (Table 3), were similar.

We report an overall case-fatality rate of 18%. Davies et al (17) observed a similar case-fatality rate (approximately 16%) in their study of invasive GAS infections in the province of Ontario, Canada. Hoge et al (20) conducted population-based surveillance of invasive GAS infections in Pima County, AZ, between 1985 and 1990. They detected 128 cases and reported a mortality rate of 20%. The current study found that the risk of mortality increased with increasing age. This result is consistent with those from several previous studies. (10,17,18)

Black race seemed to be a protective factor. This result is hard to explain. In a study conducted by Navarro et al, (10) nonwhite race appeared to slightly increase the risk of hospital mortality in patients hospitalized for GAS bacteremia (univariate OR, 1.11; 95% CI, 0.53-2.35). However, a recent study of racial differences in hospital mortality among men hospitalized in the US Veterans Affairs health care system found that black patients admitted for common medical diagnoses such as pneumonia and diabetes had lower mortality rates than white patients. (21) The protective effect of black race persisted even after adjusting for several factors, including diagnosis, age, financial assets, length of stay, and presence of comorbid disease (adjusted relative risk, 0.74; 95% CI, 0.64-0.86). The authors considered several explanations for this result but could not identify a definite cause or causes. In the current study of invasive GAS disease, it is possible that the black patients were more educated than the white patients and/or more likely to have health insurance than the white patients, and/or had a better social support network. All of these factors are potential confounders of the association between race and survival.

The current study found that men were less likely than women to die in the hospital even after adjusting for age, diabetic status, and the other six variables. This association cannot be attributed to a difference in antibiotic treatment. The association between sex and risk of hospital mortality was not confounded by any of the three antibiotics that were studied. Overall, in the group of 195 patients, 35% of the men received clindamycin whereas 32% of the women received this antibiotic. In this same group, treatment with a [beta]-lactam antibiotic was also equally distributed between the sexes (84% of the men and 85% of the women were prescribed one or more [beta]-lactam antibiotic). The use of fluoroquinolones was also similar. Twenty-six percent of the men received a fluoroquinolone antibiotic, whereas 30% of the women received a fluoroquinolone antibiotic. It is possible that even after adjusting for disease severity using the intensive care unit variable, there may have been residual confounding by the severity of illness. Also, there may have been sex differences in health insurance status and other predictors of mortality that were not recorded on the case report form.

The crude OR for hospital mortality associated with the presence of necrotizing fasciitis was 1.96. In contrast, Bernaldo de Quiros et al (8) found that cutaneous necrosis was a very strong risk factor for hospital mortality in patients hospitalized for GAS bacteremia (univariate OR, 47.2; 95% CI, 15.0-424.0). It is possible that our crude OR was attenuated as a result of misclassification. Patients with milder cutaneous infections may have erroneously been given the diagnosis of necrotizing fasciitis.

The current study detected a strong association between admission to an intensive care unit and the risk of a fatal outcome. This association was also reported by Bernaldo de Quiros et al. (8) Unlike previous studies, this study examined the effect of fluoroquinolone antibiotics on hospital mortality. Certain fluoroquinolones, such as levofloxacin and ciprofloxacin, may be used to treat GAS infections (22); however, the results of this study indicated that treatment with a fluoroquinolone did not reduce the risk of hospital mortality.

A study in mice and one in humans found that clindamycin is the antibiotic of choice in the treatment of patients who have invasive GAS disease accompanied by a deep soft-tissue infection such as myositis or necrotizing fasciitis. (12,23) The results of the current study support these previous observations. In this study, clindamycin appeared to strongly protect against hospital mortality in patients who had necrotizing fasciitis but not in patients who did not have necrotizing fasciitis.

This study was a retrospective study, and therefore no cultures were available with which to perform molecular biologic techniques such as M protein serotyping, pulsed field gel electrophoresis, and polymerase chain reaction amplification of streptococcal pyrogenic exotoxin genes. This laboratory information could be used to identify outbreaks and monitor trends in the virulence of GAS. A limitation of this study is the potential misclassification of the clinical presentation that was recorded on the surveillance case report form (Table 3). For example, county public health department staff recorded the diagnosis of streptococcal toxic shock syndrome. We did not have access to the patients' medical records and therefore could not classify patients ourselves using the case definition proposed by the Working Group on Severe Streptococcal Infections. (24) The strengths of this study include its relatively large sample size and generalizability. Furthermore, the clinical profile and outcomes of invasive GAS disease in Florida had not been previously reported.

Conclusions

Invasive GAS disease is a disease of public health importance. Annually, more than 10,000 cases of invasive GAS disease are detected nationwide. (3) The case-fatality rate for patients with necrotizing fasciitis in this study was 24% (10 of 42), and the case-fatality rate for patients who developed streptococcal toxic shock syndrome was 40% (4 of 10). Physicians who are treating patients hospitalized for invasive GAS infections should bear in mind that the risk of mortality increases with age and that treatment with [beta]-lactam antibiotics rather than fluoroquinolone antibiotics may reduce the risk of hospital mortality. This study also found that clindamycin protects against mortality in patients who have necrotizing fasciitis (OR, 0.11; 95% CI, 0.01-0.89).

Give people not only your care, But also your heart.

--Mother Teresa

Table 1. Clinical and demographic characteristics of patients
hospitalized for invasive Group A streptococcal disease in Florida,
 1996-2000 (a)

Variable                                         No. of      Total no.
                                              patients (%)  of patients
                                                            in sample

No. of black patients                            50 (20%)      250

No. of men                                      136 (53%)      257

No. admitted to an intensive care unit           94 (42%)      225

No. treated with clindamycin during              73 (32%)      231
  hospital stay (b)

No. treated with [beta]-lactam antibiotics      191 (83%)      231
  during hospital stay (b)

No. treated with fluoroquinolone antibiotics     66 (29%)      231
  during hospital stay (b)

Diabetes mellitus                                44 (19%)      237

Mortality rate                                   41 (18%)      228

(a) Median age, 52 yr (range, 0-103).

(b) Patients may have received more than one antibiotic during their
 hospital stay.

Table 2. Clinical presentation of 256 patients hospitalized for invasive
 Group A streptococcal disease in Florida, 1996-2000 (a)

Condition                              No. of
                                    patients (%)

Skin and soft tissue infection
  Cellulitis/abscess                 98 (38%)
  Necrotizing fasciitis              45 (18%)
  Gangrene                            9 (4%)
Primary bacteremia (no focus)        57 (22%)
Pneumonia                            43 (17%)
Pharyngitis                          11 (4%)
Septic arthritis                     10 (4%)
Streptococcal toxic shock syndrome   10 (4%)
Osteomyelitis                         6 (2%)
Peritonitis                           6 (2%)
Endometritis/postpartum sepsis        2 (1%)
Meningitis                            1 (0.4%)
Polyarthritis                         1 (0.4%)
Secondary bacteremia                161 (63%)

(a) Patients may have had more than one condition.

Table 3. Crude and adjusted odds ratios for hospital mortality, with 95%
confidence intervals, in 195 patients hospitalized for invasive Group A
streptococcal disease in Florida, 1996-2000 (a)

Variable                         Crude    95%      Adjusted     95%
                                  OR       CI       OR (b)       CI

Age [greater than or equal        2.59  1.21-5.57    3.80    1.46-9.93
to]55 yr (compared with <55 yr)

Blacks (compared with  whites)    0.31  0.09-1.08    0.33    0.08-1.45

Male sex                          0.60  0.29-1.27    0.41    0.17-1.02

Diabetes mellitus                 1.02  0.39-2.69    0.62    0.17-2.18

Necrotizing fasciitis             1.96  0.82-4.70    0.80    0.23-2.78

Intensive care unit              12.82  4.70-34.98  20.41    6.41-64.96

[beta]-Lactam antibiotic          0.57  0.23-1.40    0.40    0.12-1.31

Clindamycin                       1.05  0.49-2.28    0.58    0.20-1.74

Fluoroquinolone antibiotic        2.00  0.93-4.30    1.02    0.39-2.69

(a) OR, odds ratio: CI. confidence interval.
(b) Each OR is adjusted for every other variable shown in the table.

Table 4. Interaction between treatment with clindamycin and presence of
necrotizing fasciitis: adjusted odds ratios for hospital mortality
(clindamycin compared with no clindamycin) in 195 patients hospitalized
for invasive Group A streptococcal disease in Florida, 1996-2000 (a)

Necrotizing    Total      Adjusted  95% CI
fasciitis    sample size   OR (b)

Yes              33        0.11     0.01-0.89
No              162        1.01     0.31-3.33

(a) OR, odds ratio: CI, confidence interval.
(b) Adjusted for age, race, sex, diabetes, intensive care unit, [beta]-
lactam, fluoroquinolone.

Acknowledgements

We thank Donald Ward, Dr. William Bigler, and Dr. Richard Hopkins of the Florida Department of Health in Tallahassee for providing access to the data required to conduct this study. We also thank Richard J. Duma, MD, PhD, for his thoughtful comments.

From the Department of Epidemiology and Biostatistics, University of South Florida, Tampa, and the Bureau of Epidemiology, Florida Department of Health, Tampa, FL.

Supported by the Florida Department of Health.

Reprint requests to Zuber D. Mulla, PhD, School of Public Health, University of Texas-Houston, 1100 N. Stanton Street, Suite 110, El Paso, TX 79902. Email: [e-mail]

Accepted October 28, 2002.

Copyright [c] 2003 by The Southern Medical Association 0038-4348/03/9610-0968

References

1. Bisno AL. Streptococcus pyogenes, in Mandell GL, Bennett JE, Dolin R (eds): Mandell, Douglas, and Bennett's Principles and Practice of Infectious Diseases, New York, Churchill Livingstone, 1995, ed 4, pp 1786-1799.

2. American Public Health Association. Streptococcal diseases caused by Group A ([beta]-hemolytic) streptococci, in Chin JE (ed): Control of Communicable Diseases Manual. Washington, DC, American Public Health Association, 2000, ed 17, pp 470-476.

3. Centers for Disease Control and Prevention. Summary of notifiable diseases. United States, 1998. MMWR Morb Mortal Wkly Rep 1999;47(53): ii-92.

4. Basma H. Norrby-Teglund A, Guedez Y, McGeer A, Low DE, El-Ahmedy O. et al. Risk factors in the pathogenesis of invasive Group A streptococcal infections: Role of protective humoral immunity. Infect Immun 1999;67:1871-1877.

5. Weiss K, Laverdiere M, Lovgren M, Delorme J, Poirier L, Beliveau C. Group A Streptococcus carriage among close contacts of patients with invasive infections. Am J Epidemiol 1999;149:863-868.

6. Stevens DL. Streptococcal toxic shock syndrome: Spectrum of disease, pathogenesis, and new concepts in treatment. Emerg Infect Dis 1995;1:69-78.

7. Florida Bureau of Epidemiology. Streptococcal disease, invasive Group A, in Florida Morbidity Statistics, 1996. Tallahassee, Florida Department of Health, Bureau of Epidemiology, p 83.

8. Bernaldo de Quiros JC, Moreno S, Cercenado E, Diaz D, Berenguer J, Miralles P, et al. Group A streptococcal bacteremia: A 10-year prospective study. Medicine (Baltimore) 1997;76:238-248.

9. Francis J, Warren RE. Streptococcus pyogenes bacteraemia in Cambridge: A review of 67 episodes. Q J Med 1988;68:603-613.

10. Navarro VJ, Axelrod PI, Pinover W, Hockfield HS, Kostman JR. A comparison of Streptococcus pyogenes (Group A streptococcal) bacteremia at an urban and a suburban hospital: The importance of intravenous drug use. Arch Intern Med 1993;153:2679-2684.

11. Kaul R, McGeer A, Low DE, Green K, Schwartz B. Population-based surveillance for Group A streptococcal necrotizing fasciitis: Clinical features, prognostic indicators, and microbiologic analysis of seventy-seven cases. Am J Med 1997;103:18-24.

12. Zimbelman J, Palmer A, Todd J. Improved outcome of clindamycin compared with [beta]-lactam antibiotic treatment for invasive Streptococcus pyogenes infection. Pediatr Infect Dis J 1999;18:1096-1100.

13. Centers for Disease Control and Prevention, Division of Bacterial and Mycotic Disease. Active Bacterial Core Surveillance (ABCs) Report, Emerging Infections Program Network, Group A Streptococcus, 1999. Available at: http://www.cdc.gov/ncidod/dbmd/abcs/survreports/gas99.pdf. Accessed April 1, 2003.

14. SAS Institute, Inc. The LOGISTIC procedure, in SAS/STAT User's Guide, version 6. Cary, NC, SAS Institute, 1989. vol 2, ed 4, pp 1071-1126.

15. Greenland S. Modeling and variable selection in epidemiologic analysis, Am J Public Health 1989;79:340-349.

16. Kleinbaum DG, Kupper LL, Muller KE. Applied Regression Analysis and Other Multivariable Methods. Boston, PWS-Kent, 1988, ed 2.

17. Davies HD, McGeer A, Schwartz B, Green K, Cann D, Simor AE, et al. Invasive Group A streptococcal infections in Ontario, Canada. N Engl J Med 1996;335:547-554.

18. Eriksson BK, Andersson J, Holm SE, Norgren M. Epidemiological and clinical aspects of invasive Group A streptococcal infections and the streptococcal toxic shock syndrome. Clin Infect Dis 1998;27:1428-1436.

19. Zurawski CA, Bardsley M, Beall B, Elliott JA, Facklam R, Schwartz B, et al. Invasive Group A streptococcal disease in metropolitan Atlanta: A population-based assessment. Clin Infect Dis 1998;27:150-157.

20. Hoge CW, Schwartz B, Talkington DF, Breiman RF, MacNeill EM, Englender SJ. The changing epidemiology of invasive Group A streptococcal infections and the emergence of streptococcal toxic shock-like syndrome: A retrospective population-based study. JAMA 1993;269:384-389.

21. Jha AK, Shlipak MG, Hosmer W, Frances CD, Browner WS. Racial differences in mortality among men hospitalized in the Veterans Affairs health care system. JAMA 2001;285:297-303.

22. Medical Economics Staff. Physician's Desk Reference 2001. Montvale, NJ, Medical Economics Co., 2001, ed 55.

23. Stevens DL, Gibbons AE, Bergstrom R, Winn V. The Eagle effect revisited: Efficacy of clindamycin, erythromycin, and penicillin in the treatment of streptococcal myositis. J Infect Dis 1988;158:23-28.

24. The Working Group on Severe Streptococcal Infections. Defining the Group A streptococcal toxic shock syndrome: Rationale and consensus definition. JAMA 1993;269:390-391 (comment).

RELATED ARTICLE: Key Points

* The annual incidence of invasive Group A streptococcal disease in the United States is higher than that of other serious bacterial illnesses such as meningococcal disease and Escherichia coli O157:H7 infections.

* We conducted a large 4-year retrospective study of invasive Group A streptococcal infections in Florida.

* We found that treatment with clindamycin protected against hospital mortality among patients with necrotizing fasciitis but did not protect against mortality among the larger group of patients who did not develop necrotizing fasciitis.

Zuber D. Mulla, PHD, MSPH, Paul E. Leaverton, PHD, and Steven T. Wiersma, MD, MPH