The Epidemiology of Invasive Group A Streptococcal Infection and Potential Vaccine Implications: United States,

Transcription

1 MAJOR ARTICLE The Epidemiology of Invasive Group A Streptococcal Infection and Potential Vaccine Implications: United States, Rosalyn E. O Loughlin, 1,2 Angela Roberson, 1 Paul R. Cieslak, 5 Ruth Lynfield, 6 Ken Gershman, 7 Allen Craig, 8 Bernadette A. Albanese, 9 Monica M. Farley, 3,4 Nancy L. Barrett, 10 Nancy L. Spina, 11 Bernard Beall, 1 Lee H. Harrison, 12 Arthur Reingold, 13 and Chris Van Beneden, 1 for the Active Bacterial Core Surveillance Team 1 Respiratory Diseases Branch, National Center for Immunization and Respiratory Diseases, and 2 Epidemic Intelligence Service, Centers for Disease Control and Prevention, 3 Emory University School of Medicine, and 4 Atlanta Veterans Affairs Medical Center, Atlanta, Georgia; 5 Oregon Department of Human Services, Portland; 6 Minnesota Department of Health, Minneapolis; 7 Colorado Department Public Health and Environment, Denver; 8 Tennessee Department of Health, Nashville; 9 New Mexico Department of Health, Santa Fe; 10 Connecticut Department of Public Health, Hartford; 11 New York State Department of Health, Albany; 12 Johns Hopkins Bloomberg School of Public Health, Baltimore; and 13 University of California, Berkeley (See the editorial commentary by Cohen-Poradosu and Kasper on pages 863 5) Background. Invasive group A Streptococcus (GAS) infection causes significant morbidity and mortality in the United States. We report the current epidemiologic characteristics of invasive GAS infections and estimate the potential impact of a multivalent GAS vaccine. Methods. From January 2000 through December 2004, we collected data from Centers for Disease Control and Prevention s Active Bacterial Core surveillance (ABCs), a population-based system operating at 10 US sites (2004 population, 29.7 million). We defined a case of invasive GAS disease as isolation of GAS from a normally sterile site or from a wound specimen obtained from a patient with necrotizing fasciitis or streptococcal toxic shock syndrome in a surveillance area resident. All available isolates were emm typed. We used US census data to calculate rates and to make age- and race-adjusted national projections. Results. We identified 5400 cases of invasive GAS infection (3.5 cases per 100,000 persons), with 735 deaths (case-fatality rate, 13.7%). Case-fatality rates for streptococcal toxic shock syndrome and necrotizing fasciitis were 36% and 24%, respectively. Incidences were highest among elderly persons (9.4 cases per 100,000 persons), infants (5.3 cases per 100,000 persons), and black persons (4.7 cases per 100,000 persons) and were stable over time. We estimate that ,500 cases of invasive GAS infection occur in the United States annually, resulting in deaths. The emm types in a proposed 26-valent vaccine accounted for 79% of all cases and deaths. Independent factors associated with death include increasing age; having streptococcal toxic shock syndrome, meningitis, necrotizing fasciitis, pneumonia, or bacteremia; and having emm types 1, 3, or 12. Conclusions. GAS remains an important cause of severe disease in the United States. The introduction of a vaccine could significantly reduce morbidity and mortality due to these infections. Invasive infections caused by group A Streptococcus (GAS) or Streptococcus pyogenes include sepsis, bacteremic pneumonia, necrotizing fasciitis (NF), and streptococcal toxic shock syndrome (STSS). GAS also causes noninvasive disease, most commonly manifested as Received 9 February 2007; accepted 23 May 2007; electronically published 29 August Reprints or correspondence: Dr. Rosalyn O Loughlin, Centers for Disease Control and Prevention, 1600 Clifton Rd. NE, MS C-23, Atlanta, GA (bwf0@cdc.gov). Clinical Infectious Diseases 2007; 45: by the Infectious Diseases Society of America. All rights reserved /2007/ $15.00 DOI: / pharyngitis, suppurative complications (such as otitis media), and nonsuppurative sequelae (such as acute rheumatic fever and acute glomerulonephritis). GAS infection causes significant morbidity and mortality, with an estimated 500,000 deaths worldwide, most of which are attributable to invasive infection, acute rheumatic fever, and subsequent rheumatic heart disease [1, 2]. In the United States, the annual health care costs associated with GAS are $493 million [3]. Much of this cost is attributable to cases of pharyngitis in children. Clinical management of invasive GAS infection focuses on accurate diagnosis and timely, appropriate use Epidemiology of GAS in the US CID 2007:45 (1 October) 853

2 of antimicrobial therapy. Few effective prevention tools exist. Development of a GAS vaccine is challenging because of the vast number of emm types (1100) [4] and because of concerns of potential immunologic cross-reactivity between epitopes in the organism s M protein and human tissue proteins that could trigger acute rheumatic fever or other adverse sequelae [5, 6]. Despite these difficulties, several vaccines are in preclinical phases of testing, and a 26-valent, M protein based vaccine has successfully completed a phase 2 trial involving adults [7 9]. In the United States, a GAS vaccine is likely to be marketed on the basis of its efficacy against pharyngitis in children; however, it should also have an impact on invasive pediatric disease. Although the duration of immunity is unknown, an anamnestic response was seen in a phase 1 study of a recombinant multivalent GAS vaccine [10]. The Centers for Disease Control and Prevention (CDC) began surveillance for invasive GAS infections in 5 US sites in 1995 as part of the Active Bacterial Core surveillance (ABCs) program of the Emerging Infections Programs Network. We report the epidemiologic characteristics of invasive GAS infection in the United States for 5 years of surveillance ( ), evaluate trends in incidence over a longer time frame ( ), and estimate the potential impact of a multivalent GAS vaccine. METHODS Surveillance. We reviewed ABCs data on case patients from whom GAS-positive specimens were obtained during the period from 1 January 2000 through 31 December ABCs is active, population- and laboratory-based surveillance that tracks invasive GAS infections in 10 US sites, covering a 2004 population of 29.7 million persons ( 9.5% of the US population). ABCs methods have been described elsewhere [1, 11]. The ABCs sites (and years) included in this analysis are as follows: 3-county San Francisco Bay area, California ( ); the 5-county Denver area, Colorado (mid ); the 20- county Atlanta area, Georgia (2004); the 6-county Baltimore metropolitan area, Maryland ( ); the 3-county Portland area, Oregon ( ); the 15-county Rochester and Albany areas of New York ( ); the 11 urban county area of Tennessee ( ); and the entire states of Georgia ( ), Minnesota ( ), New Mexico (2004), and Connecticut ( ). ABCs personnel routinely contacted all participating clinical microbiology laboratories to identify cases. A case was defined as isolation of GAS from a normally sterile site or from a wound specimen obtained from a surveillance area resident with NF or STSS. Case patients who had GAS-positive blood culture results but no identified clinical syndrome were categorized as having bacteremia without a source. The remaining case patients with bacteremia were categorized as having 1 clinical syndrome. To allow comparisons with previously published data from , we categorized clinical syndromes as occurring among 2 age groups: children aged!10 years and persons aged 10 years [1]. Case patients who had undergone surgery in the 7 days before their first positive culture result were considered to be recent surgery case patients. Nosocomial infections were defined as infections in case patients who were hospitalized at least 48 h before their positive culture result. Descriptive epidemiology. Descriptive analysis included all cases from 1 January 2000 through 31 December Unless otherwise specified, incidences are expressed as the number of cases per 100,000 population. We used regional and national intercensus population estimates for each year as the denominator. For national rate estimates, race- and age-specific disease rates were applied from the aggregate surveillance area to the age and racial distribution of the US population for each year. Individuals of unknown race (13%) were distributed by site on the basis of reported race distribution for known cases. Casefatality rates (CFRs) were calculated using outcomes at the time of hospital discharge. To examine incidence trends over time ( ), data were restricted to sites used in previous incidence estimates: the 8-county Atlanta area; the 7-county Minneapolis/St. Paul area; the 3-county Portland area; the 3-county San Francisco area; and all of Connecticut [1]. GAS typing. All available GAS isolates were confirmed and typed at the CDC s Streptococcus laboratory. emm Typing was performed on the basis of restriction analysis of emm gene amplicons or on the basis of sequencing of the variable M serotype specific region of emm gene amplicons. Concordance has been established between the emm type and M serotype. The protocol for emm typing and a complete, downloadable listing of emm type sequences are available online [4, 12]. Estimates of vaccine benefit. To estimate the potential benefits of a multivalent vaccine, we used available emm typing results to calculate the proportion of disease cases and deaths due to GAS emm types in the proposed 26-valent vaccine (types 1, 1.2, 2, 3, 5, 6, 11, 12, 14, 18, 19, 22, 24, 28, 29, 33, 43, 59, 75, 76, 77, 89, 92, 94, 101, and 114) for age groups at highest risk of developing invasive GAS disease: children aged!5 years and adults aged 65 years. We hypothesized that the vaccine efficacy rate would be 84% on the basis of averaged seroconversion rates from immunogenicity (phase I/II) studies among healthy adults [7, 9, 13], and we hypothesized that vaccine coverage rates would be 70% 90% for infants (on the basis of Haemophilus influenzae type b and pneumococcal vaccination rates [14]) and 60% 70% for adults 65 years (on the basis of influenza and pneumococcus vaccination rates [15]). We used the following formula to calculate the number of GAS cases and deaths that could potentially be prevented with a 854 CID 2007:45 (1 October) O Loughlin et al.

3 GAS vaccine: efficacy (%) coverage (%) the percentage of vaccine emm types in the 26-valent vaccine. Predictors of death. We evaluated predictors of death using a conditional logistic regression model stratified by ABCs site. Case patients with multiple clinical syndromes were classified in the category of highest severity on the basis of the CFR for each syndrome. In addition, cutaneous and soft-tissue infections were included as one syndrome. Underlying illnesses were grouped into the following 4 categories: no underlying illness, skin condition (e.g., burns, varicella, and wounds), chronic illness (e.g., diabetes and chronic obstructive pulmonary disease), and immunosuppression (e.g., AIDS and asplenia). A person with multiple underlying illnesses was placed in the category of highest severity on the basis of the following decreasing levels of severity: immunosuppression, chronic illness, and skin condition. The model included only case patients for whom information on all variables was available. One site that did not record information on HIV/AIDS status was excluded, as were case patients with unknown ( n p 148) or nontypable ( n p 1) emm types or with unknown outcomes ( n p 32). Var- iables associated with death ( P!.1) on univariate analysis were entered in the multivariable model and were retained if the P value was!.05. Collinearity and all 2-way interactions were evaluated. All analyses were conducted using SAS software, version 9.1 (SAS Institute). RESULTS Demographic characteristics and disease rates. ABCs identified 5400 cases of invasive GAS infection during the period from 1 January 2000 through 31 December The median age of case patients was 50 years (range, years); 53% were male ( n p 2845). The majority of patients were white (3307 patients [61%]), with black patients (1093 patients [20%]), Asian or American Indian patients (215 patients [4%]), patient of other race (93 patients [2%]), and patients with unknown race (692 patients [13%]) making up the remainder. The distribution of unknown race varied by site, ranging from 1% in Maryland to 38% in New Mexico. The average annual incidence for the 5 years of surveillance was 3.5 invasive GAS cases per 100,000 persons (table 1). Incidence varied by site, ranging from 2.0 to 5.9 cases per 100,000 persons in Georgia and Colorado, respectively, and most sites had slight year-to-year variation. Rates for sites where data were available over a continuous 9-year period showed no significant trend (1996, 3.5 cases per 100,000 persons; 1997, 3.6 cases per 100,000 persons; 1998, 3.5 cases per 100,000 persons; 1999, 3.5 cases per 100,000 persons; 2000, 3.7 cases per 100,000 persons; 2001, 3.7 cases per 100,000 persons; 2002, 3.1 cases per 100,000 persons; 2003, 4.0 cases per 100,000 persons; and 2004, 3.0 cases per 100,000 persons; P p.2). Seasonal variation was observed, with most disease occurring in the winter and early Table 1. Incidence of invasive group A streptococcal disease, by site and year, Active Bacterial Core Surveillance (ABCs), ABCs site Year of surveillance (n p 5238) year average a California Colorado Connecticut Georgia b Maryland Minnesota New Mexico New York Oregon Tennessee All sites Estimated United States c NOTE. Data are no. of cases per 100,000 persons. a The 4-year average is shown for Colorado, and the 1-year incidence is shown for New Mexico. b Includes the entire state of Georgia for and 20 metropolitan counties for c Race- and age-specific rates of disease were applied from the aggregate surveillance area to the age and racial distribution of the US population for each year. spring; this same pattern was noted by syndrome and site, with minor variations. Incidences were highest among patients aged 65 years (9.4 cases per 100,000 persons), followed by children aged!1 year (5.3 cases per 100,000 persons) (figure 1). The incidence was also significantly higher among black patients (4.7 cases per 100,000 persons) than among nonblack patients (3.2 cases per 100,000 persons; OR, 1.5; 95% CI, ). The majority of case patients were hospitalized (4915 patients [91%]), and 734 died (CFR, 13.7%). Compared with hospitalized patients, case patients who were not hospitalized presented more frequently with bacteremia (51% vs. 27%; P!.001) and less frequently with cellulitis (20% vs. 38%; P!.001). The outcome was unknown for 32 case patients (0.6%). The CFR was highest in the oldest age group (22.8%) and was lower among black patients (12.1%) than among nonblack patients (14.1%), although this difference was not significant (OR, 0.9; 95% CI, ). Projecting to the US population, with adjustment for race and age, we estimate that ,500 invasive GAS infections occurred in the United States annually, resulting in deaths. Underlying disease. Underlying disease status was available for 5151 patients (95%); 67% had at least 1 underlying condition, and 35% had at least 2. The most common underlying conditions and definitions are shown in table 2 and include heart disease, diabetes mellitus, and skin conditions. For most conditions, there was significant variation across sites, with the Epidemiology of GAS in the US CID 2007:45 (1 October) 855

4 Figure 1. Incidence of invasive group A streptococcal disease at 10 Active Bacterial Core Surveillance sites, by age and race, Numbers are incidence rates by age group for all races combined. Age-specific case-fatality rates (CFRs) for all races combined are indicated by the black line. greatest variation noted for alcohol abuse, injection drug use, and skin conditions. There was no significant association between race and underlying condition (OR, 0.9; 95% CI, ). Case patients aged!10 years were significantly less likely to have an underlying disease (22%) than were those aged 10 years (72%; P!.001). Clinical syndromes. The distribution of clinical syndromes, by age group and CFR (with and without additional syndromes), is presented in table 3. More than one-third (36%) of case patients presented with a cutaneous or soft-tissue infection, 29% presented with primary bacteremia, and 15% presented with pneumonia. Compared with case patients aged 10 years, those aged!10 years were significantly less likely to present with a cutaneous or soft-tissue infection, NF, or endocarditis/pericarditis and were more likely to present with abscesses, osteomyelitis, epiglotitis/otitis media, meningitis, or primary bacteremia. Case patients with STSS, meningitis, NF, or pneumonia had the highest CFRs. Among case patients for whom the information was available (94%), 188 (4%) had recently undergone surgery, 118 (2.6%) were pregnant or had recently given birth, and 275 (6%) had nosocomially acquired infection. The distribution of clinical syndromes was similar among case patients with and without nosocomially acquired infections. GAS was most frequently isolated from blood specimens (4131 patients [77%]). Other sources included joint fluid (8%), surgical specimens (6%), peritoneal fluid (2%), pleural fluid (2%), and wounds (2%). Other sources each accounted for 1% of cases. emm Sequence types and potential vaccine-preventable disease. A bacterial isolate was available for typing for 4350 case patients (81%); only 1 isolate was nontypable. Distribution of emm types varied by ABCs site and by year. The 30 most common emm types accounted for 95% of isolates (figure 2). emm Types 1 (22%), 3 (9%), 28 (9%), 12 (9%), and 89 (6%) were the most common and cumulatively accounted for 55% of isolates. The 26 most common emm types accounted for 93% of isolates, whereas emm types in the proposed 26-valent vaccine (which includes M types historically associated with acute rheumatic fever) accounted for 79% of all isolates; 85% and 88% of NF and STSS isolates, respectively; and 79% of isolates in patients who died. The proportion of disease accounted for by emm types in the proposed vaccine varied little over 10 years of surveillance (82%, 83%, 81%, 81%, 77%, 74%, 79%, 83%, 81%, and 76% for each year during , respectively) and was similar among children aged!5 years (83%) and adults aged 65 years (85%). When the efficacy and coverage estimates described earlier were applied, a 26- valent vaccine could potentially prevent 49% 63% of invasive GAS infections among children and 43% 50% of cases among elderly persons, with similar decreases in the number of deaths (table 4). Predictors of death. Increasing age, residence in a nursing 856 CID 2007:45 (1 October) O Loughlin et al.

5 Table 2. Underlying conditions reported among case patients with invasive group A streptococcal disease by surveillance area, Active Bacterial Core Surveillance site Underlying condition a California (n p 685) Colorado b (n p 516) Connecticut (n p 635) Georgia b (n p 757) Maryland (n p 728) Minnesota (n p 742) New Mexico b (n p 88) New York (n p 362) Oregon (n p 225) Tennessee (n p 413) Total (n p 5151) P c Heart disease d 109 (15.9) e 76 (14.7) e 150 (23.6) e 145 (19.2) 136 (18.7) 139 (18.7) 14 (15.9) 106 (29.3) e 52 (23.1) 125 (30.3) e 1052 (20.4) Diabetes mellitus 127 (18.6) 96 (18.6) 146 (23.0) e 140 (18.5) 151 (20.7) 140 (18.9) 19 (21.6) 78 (21.6) 35 (15.6) 103 (24.9) e 1035 (20.1)!.05 Alcohol abuse 129 (18.8) e 117 (22.7) e 52 (8.2) e 47 (6.2) e 71 (9.8) 64 (8.6) e 15 (17.1) 24 (6.6) e 31 (13.8) 22 (5.3) e 572 (11.0) Malignancy f 56 (8.2) e 48 (9.3) 71 (11.2) 90 (11.9) 64 (8.8) 102 (13.8) e 9 (10.2) 49 (13.5) 30 (13.3) 29 (7.0) e 548 (10.7)!.01 Injection drug use 118 (17.2) e 29 (5.6) 40 (6.3) e 12 (1.6) e 161 (22.1) e 11 (1.5) e 6 (6.8) 7 (1.9) e 35 (15.6) e 8 (1.9) e 427 (8.3) Emphysema and/or COPD 50 (7.3) 51 (9.9) e 62 (9.8) e 53 (7.0) 61 (8.4) 43 (5.8) e 6 (6.9) 23 (6.4) 21 (9.3) 25 (6.1) 395 (7.7)!.07 Immunosuppression g 33 (4.8) e 41 (8.0) 60 (9.5) e 53 (7.0) 56 (7.7) 62 (8.4) 5 (5.7) 32 (8.8) 12 (5.3) 17 (4.1) e 371 (7.2)!.01 Renal failure/dialysis 46 (6.7) 28 (5.4) 38 (6.0) 58 (7.7) 37 (5.1) e 53 (7.1) 8 (9.1) 33 (9.1) 8 (3.6) e 43 (10.4) e 352 (6.8)!.01 Skin condition h 104 (15.2) 113 (22.0) e 64 (10.1) e 138 (18.2) 159 (21.8) e 147 (19.8) e 22 (25.0) e 23 (6.4) e 56 (24.9) e 50 (12.1) e 876 (17.0) No underlying illness 214 (31.2) 117 (22.7) e 249 (39.2) e 286 (37.8) e 206 (28.3) e 268 (36.1) 23 (26.1) 125 (34.5) 65 (28.9) 150 (36.3) 1703 (33.1) 11 Underlying illness i 277 (40.4) e 183 (35.5) 237 (37.3) 230 (30.4) e 329 (45.2) e 227 (37.3) e 34 (38.7) 116 (32.0) 93 (36.5) 131 (31.7) 1857 (36.1) NOTE. Data are no. (%) of underlying conditions, unless otherwise indicated. COPD, chronic obstructive pulmonary disease. a Some patients had 11 underlying condition. The underlying condition was unknown for 249 case patients (4.6%). Those conditions that totaled!5% of cases are not presented here and include cerebrovascular accident (4.1%), HIV infection/aids (3.2%), cirrhosis/liver failure (2.7%), and asplenia/sickle cell disease (0.8%). c e b Colorado data are from mid-2000 through 2004; for Georgia, data are for the entire state for and for 20 metropolitan counties for 2004; New Mexico data are for 2004 only. Variation in the proportion of each underlying condition across all sites; determined by x 2 test. d Includes heart failure or atherosclerotic cardiovascular disease. P!.05 for all; the P value indicates the variation across sites when each site was compared with the remaining sites combined. Includes leukemia, myeloma, Hodgkin disease, or other malignancies. g Resulting from immunoglobulin deficiency, treatment of systemic lupus erythematosus, nephrotic syndrome, organ transplantation, or receipt of immunosuppressive therapy. h Includes burns, varicella, blunt or penetrating trauma, and surgical wounds. Data are no. (%) of case patients who have 11 underlying condition; denominator is the total no. of patients with an underlying condition, by site. f i

6 Table 3. Clinical syndromes and case-fatality rates (CFRs) among case patients with invasive group A streptococcal (GAS) disease, by age group, Syndrome a All (n p 5400) Case patient group Patients aged!10 years (n p 572) b Patients aged 10 years (n p 4826) b Cutaneous and/or soft-tissue infection d No. (%) of case patients 1947 (36.4) 124 (21.7) 1823 (37.7) CFR, % (CFR when only the specified syndrome was present, %) 8.0 (6.8) 0 (0) 8.5 (7.2) Bacteremia without a source No. (%) of case patients 1586 (29.4) 204 (35.7) 1382 (28.6) CFR, % (CFR when only the specified syndrome was present, %) 16.7 (17.0) 3.4 (3.6) 18.5 (19.0) Pneumonia.26 No. (%) of case patients 792 (14.7) 93 (16.3) 699 (14.5) CFR, % (CFR when only the specified syndrome was present, %) 22.9 (23.0) 11.8 (10.7) 24.3 (24.7) Septic arthritis.7 No. (%) of case patients 416 (7.7) 47 (8.9) 369 (8.5) CFR, % (CFR when only the specified syndrome was present, %) 2.7 (2.2) 0 (0) 3.0 (2.5) Necrotizing fasciitis e No. (%) of case patients 389 (7.2) 5 (0.9) 384 (8.0) CFR, % (CFR when only the specified syndrome was present, %) 23.7 (25.4) 20.0 (33.3) 23.7 (25.3) Streptococcal toxic shock syndrome e.20 No. (%) of case patients 309 (5.7) 26 (4.6) 283 (5.9) CFR, % (CFR when only the specified syndrome was present, %) 35.9 (43.2) 7.7 (7.1) 38.5 (49.4) Abscess No. (%) of case patients 329 (6.1) 57 (9.9) 272 (5.6) CFR, % (CFR when only the specified syndrome was present, %) 4.0 (3.6) 0 (0) 4.8 (4.6) Abdominal and/or peritoneal infection.59 No. (%) of case patients 152 (2.8) 18 (3.2) 134 (2.8) CFR, % (CFR when only the specified syndrome was present, %) 12.6 (12.2) 11.1 (7.7) 12.7 (12.9) Osteomyelitis No. (%) of case patients 186 (3.4) 34 (5.9) 152 (3.2) CFR, % (CFR when only the specified syndrome was present, %) 1.1 (0) 0 (0) 1.3 (0) Pregnancy-related infection f NA No. (%) of case patients 57 (1.1) NA 57 (1.2) CFR, % (CFR when only the specified syndrome was present, %) 7.1 (4.3) NA 7.1 (4.3) Endocarditis and/or pericarditis!.01 No. (%) of case patients 75 (1.4) 1 (0.2) 74 (1.5) CFR, % (CFR when only the specified syndrome was present, %) 4.0 (6.1) 0 (0) 4.1 (6.3) Meningitis and/or CNS infection No. (%) of case patients 52 (1.0) 17 (3.0) 35 (0.7) CFR, % (CFR when only the specified syndrome was present, %) 26.9 (17.1) 23.5 (18.2) 28.6 (16.7) Epiglottitis and/or otitis media No. (%) of case patients 72 (1.3) 22 (3.9) 50 (1.0) CFR, % (CFR when only the specified syndrome was present, %) 12.5 (5.9) 4.6 (0) 16.0 (8.1) Other g.003 No. (%) of case patients 29 (0.5) 8 (1.4) 21 (0.4) CFR, % (CFR when only the specified syndrome was present, %) 3.5 (0) 0 (0) 4.8 (0) NOTE. NA, not applicable. a Case patients may have had 11 syndrome, with the exception of bacteremia without a source. Clinical syndromes, such as cutaneous and/ or soft-tissue infection and pneumonia, had to be accompanied by recovery of an isolate from a normally sterile site or specimen, such as blood, to meet case definition. b The ages for 2 case patients were missing. c Comparison by age category of the proportion of invasive GAS infections attributable to a particular syndrome; determined by x 2 test. d Includes cellulitis ( n p 1838), infected ulcers, erisypelas, wound infections, phlebitis including thrombophlebitis, lymphangitis, lymphadenitis, bursitis without a GAS-positive joint fluid culture, and gangrene; excludes necrotizing fasciitis. e Sixty-two case patients (1.2%) presented with both streptococcal toxic shock syndrome and necrotizing fasciitis (combined CFR, 35.5%). f Includes endometritis ( n p 38), septic abortion ( n p 6), chorioamnionitis ( n p 6), and puerperal sepsis ( n p 11). g Other includes hemolytic uremic syndrome ( n p 2) and unknown syndromes ( n p 27). P c

7 Figure 2. Cumulative frequency of the 30 most common emm types, Active Bacterial Core Surveillance data, Black bars indicate the emm types included in the proposed 26-valent vaccine, and white bars indicate the emm types that are not included. Other 26-valent vaccine emm types constituted 1.2% of the total, and other non 26-valent vaccine emm types constituted 3.9% of the total. home, and presence of a specific disease syndrome (i.e., STSS, meningitis, NF, pneumonia, or bacteremia), emm type (1, 3, or 12), or underlying condition (i.e., skin condition, chronic illness, or immunosuppression) were all independent predictors of an increased risk of death (table 5). Male sex was an independent predictor of a reduced risk of death. No statistically significant 2-way interactions were found. DISCUSSION Invasive GAS infection is a significant cause of morbidity and mortality in the United States, with ,500 episodes and deaths each year. The rates of disease are highest among children aged!1 year and persons aged 65 years. Because 91% of case patients are hospitalized, invasive GAS infection also places a significant burden on the health care system. No other data source in the United States provides such an extensive, current description of the frequency of severe disease manifestations of this pathogen. Our findings help address recurrent concerns among health care providers, epidemiologists, and sometimes the general public regarding perceived increases in deadly manifestations of severe GAS infections in the United States. Although rates vary among smaller communities, our study found that the incidence and mortality rate of invasive GAS infections in the United States have remained relatively stable from 1996 to The overall incidence of 3.5 cases per 100,000 persons is consistent with rates found in Canada (3.8 cases per 100,000 persons) [16] and several European countries, including The Netherlands (3.1 cases per 100,000 persons), Norway (3.3 cases per 100,000 persons), and the United Kingdom ( cases per 100,000 persons) [17]. This relative stability in disease incidence contrasts with the epidemiology of invasive GAS disease in the 1980s and early 1990s, when reports from the United States suggested an increasing severity of infection, accompanied by a potential concurrent increase in incidence [18, 19], although this increase was difficult to confirm because of the limited population-based longitudinal surveillance. Invasive GAS infection rates varied both among geographically disparate ABCs sites and within sites over time. Such inter- and intrasite variations have been found previously in the United States [1], Canada [16], and Europe [17] and can be attributed partially to differences found in circulating emm types and in the population s susceptibility to a particular emm type over time. Variations may also be due to site-to-site differences in the prevalence of risk factors for invasive GAS infection, including chronic diseases (e.g., HIV infection, diabetes, cardiac disease, cancer, and substance abuse), race, and number of persons living in the home [20, 21]. In our study, rates of underlying medical conditions in particular, those related to behaviors such as alcohol and drug abuse varied widely between sites. Epidemiology of GAS in the US CID 2007:45 (1 October) 859

8 Table 4. Calculation of potential invasive group A streptococcal (GAS) disease and fatalities prevented by vaccination of infants and elderly persons with a proposed 26-valent vaccine. Age group Children aged!5 years (n p 295) b Assumed vaccine efficacy, % (A) Assumed vaccine coverage, % (B) Case patients with GAS disease due to 26-valent emm type, % a (C) GAS-related deaths associated with a 26-valent emm type, % a (D) Potential cases of disease prevented, %(A B C) Potential deaths prevented, %(A B D) Low coverage assumption High coverage assumption Adults aged 65 years (n p 1368) b Low coverage assumption High coverage assumption NOTE. Proposed 26-valent vaccine from [7]. a Among patients with a typed isolate. b The n value represents the number of patients with an isolate available. The rate of invasive GAS infection among all children aged!5 years was 3.4 cases per 100,000 persons, and the rate among all adults aged 65 years was 9.4 cases per 100,000 persons. Although stable over time, rates of invasive GAS infection measured by ABCs are likely minimum estimates, because ABCs does not capture invasive infections without sterile site isolates unless they are manifested by STSS and NF. A previous study compared the number of patients with STSS and/or NF reported via standard ABCs methods with the number identified through comprehensive chart review, with subsequent reclassification of patients with STSS or NF according to required clinical and laboratory criteria. Patients with invasive infection presented with STSS and NF at least twice as frequently as recognized by standard ABCs methods [22]. We kept our methods similar to those of the analysis to allow direct comparisons [1]. Significant differences include additional variables in the predictors of death model (i.e., nursing home, underlying illness, and sex); a new syndrome consisting of epiglottitis and otitis media, which were previously classified under the category of other syndrome; an expanded definition of the underlying condition termed skin to include blunt or penetrating trauma and surgical wounds data not previously collected; and a change in definition of nosocomial infection (reduced from 3 to 2 days after hospital admission to the date of culture positivity). Identification of factors associated with mortality can guide disease-prevention efforts. Vaccines developed to prevent invasive GAS infections should include emm types 1, 3, and 12. The increased risk of death among the elderly, persons with certain underlying diseases, and those residing in a nursing home highlights the importance of following infection control guidelines for long-term care facilities and suggests a potential target population for immunization should a GAS vaccine be shown to be effective in the elderly and licensed for use for this age group. We are unable to explain the apparent protective effect of being male and acknowledge that there may be confounders for which we were unable to control. Current strategies for preventing morbidity and mortality resulting from invasive GAS infections are limited to promoting routine infection control, healthy behaviors and practices, and preventing secondary cases among postpartum and postsurgical patients [23]. We estimate that a proposed 26-valent GAS vaccine could prevent 40% 50% of cases and 50% 60% of deaths due to invasive GAS infections among children and elderly persons, respectively, in the United States. These may be overestimates, because we assumed vaccination of both infants and the elderly, used immunogenicity data from healthy adults, which may not be an adequate proxy for vaccine efficacy, and used coverage estimates based on other vaccines. However, our estimates also do not include a possible benefit resulting from herd immunity. Stability of the distribution of emm types over time is essential for successful prevention by vaccination. Although invasive infections comprise only a small proportion of the total burden of GAS disease, they include a significant proportion of GAS-associated morbidity and mortality in the United States. In industrialized countries, the development and introduction of a GAS vaccine is likely to be driven by the burden of noninvasive disease specifically, pharyngitis in children. The proposed 26-valent GAS vaccine includes immunogens for 86% of GAS isolates obtained from acute pharyngitis surveillance in the United States [24]. Nonsuppurative sequelae, such as acute rheumatic fever, may also be prevented by a GAS vaccine. The limited information available from developing and middle-income countries on the distribution of emm and clonal types, for both invasive and noninvasive GAS disease, show increased diversity and different emm types, compared with distribution in the United States [25 28]. 860 CID 2007:45 (1 October) O Loughlin et al.

9 Table 5. Results of multivariable analysis of factors independently associated with death due to invasive group A streptococcal disease for cases (n p 3974) reported from 1 January 2000 through 31 December 2004, stratified by Active Bacterial Core Surveillance site. Variable OR (95% CI) Age, years!18 Reference ( ) ( ) ( ) Race Nonblack Reference Black 0.9 ( ) Sex Female Reference Male 0.7 ( ) Nursing home residence No Reference Yes 1.9 ( ) Syndrome a Syndrome other than those listed below Reference Streptococcal toxic shock syndrome 12.7 ( ) Necrotizing fasciitis 5.3 ( ) Pneumonia 2.9 ( ) Bacteremia without a source 2.8 ( ) Meningitis and/or CNS infection 8.0 ( ) Cutaneous and/or soft-tissue infection 0.8 ( ) emm Type Not emm1, emm3, or emm12 Reference emm1 1.9 ( ) emm3 2.2 ( ) emm ( ) Health status No underlying illness Reference Skin condition a 1.4 ( ) Chronic illness 1.6 ( ) Immunosuppression 2.2 ( ) NOTE. Data from New York were excluded, because deaths were associated with HIV infection/aids were not reported. a Includes burns, varicella, and wounds. Rapid advances in vaccine technology, particularly the possibility of a protein-based vaccine that may eliminate the need for multivalent conjugate preparations, are promising [24]. Meanwhile, continued surveillance is essential to document disease burden, monitor changes in emm distribution, guide public health prevention activities, identify target populations for vaccination, and provide data for vaccine cost-benefit analyses. The World Health Organization has plans to develop international standards for GAS surveillance, to generate disease burden estimates in multiple countries. This will be important to guide design of vaccines that could have global utility. Acknowledgments We thank the following Emerging Infections Programs staff members: Pam Daily (California); Steve Burnite, Allison Daniels, Nicole Haubert, Joyce Knutsen, and Karen Xavier (Colorado); James Hadler, Susan Petit, and M. Zachariah Fraser (Connecticut); Kathryn E. Arnold and Wendy Baughman (Georgia); Laurie Thomson Sanza (Maryland); Jean Rainbow, Brenda Jewell, and Richard Danila (Minnesota); Joseph Bareta, Kathy Angeles, Lisa Butler, Joanne Keefe, and Karen Johnson (New Mexico); Glenda Smith and Nellie Dumas (New York); Karen Stefonek, Margaret Dragoon, and Anna Zeigler (Oregon); and Brenda Barnes and William Schaffner (Tennessee). We also thank Tami Skoff, Carolyn Wright, and Elizabeth Zell (of the CDC s ABCs program) and Varja Sakota, Zhongya Li, Delois Jackson, and Alma Ruth Franklin (of the CDC s Streptococcus laboratory). In addition, we thank Kate O Brien (Johns Hopkins Bloomberg School of Public Health). Financial support. Support for the ABCs is provided by the CDC s Emerging Infections Programs. Potential conflicts of interest. All authors: no conflicts. References 1. O Brien KL, Beall B, Barrett NL, et al. Epidemiology of invasive group A streptococcus disease in the United States, Clin Infect Dis 2002; 35: Department of Child and Adolescent Health and Development, Department of Immunization Vaccines and Biologicals. The current evidence for the burden of group A streptococcal diseases. Geneva: World Health Organization, Institute of Medicine. In: Stratton KR, ed. Vaccines for the 21st century. Washington, DC: National Academy Press, 2001: Streptococcus pyogenes emm sequence database. Available at: wwwcdcgov/ncidod/biotech/strep/strepblasthtm. Accessed 22 May Bisno AL, Rubin FA, Cleary PP, Dale JB. Prospects for a group A streptococcal vaccine: rationale, feasibility, and obstacles--report of a National Institute of Allergy and Infectious Diseases workshop. Clin Infect Dis 2005; 41: Department of Child and Adolescent Health and Development, Department of Immunization Vaccines and Biologicals Group. Group A streptococcal vaccine development: current status and issues of relevance to less developed countries. WHO/IVB/ Geneva: World Health Organization, McNeil SA, Halperin SA, Langley JM, et al. Safety and immunogenicity of 26-valent group A Streptococcus vaccine in healthy adult volunteers. Clin Infect Dis 2005; 41: Hu MC, Walls MA, Stroop SD, Reddish MA, Beall B, Dale JB. Immunogenicity of a 26-valent group A streptococcal vaccine. Infect Immun 2002; 70: McMillan D. StreptAvax (ID Biomedical). Curr Opin Investig Drugs 2006; 7: Kotloff KL, Dale JB. Progress in group A streptococcal vaccine development. Pediatr Infect Dis J 2004; 23: Active Bacterial Core surveillance methodology. Available at: Accessed 22 May CDC s Protocol for emm typing. Available at: Accessed 22 May Kotloff KL, Corretti M, Palmer K, et al. Safety and immunogenicity of a recombinant multivalent group a streptococcal vaccine in healthy adults: phase 1 trial. JAMA 2004; 292: Estimated vaccination coverage with individual vaccines and selected vaccination series among children months of age by state USNIS, Available at: antigen_state.xls. Accessed 22 May Influenza and pneumococcal vaccination coverage among persons aged Epidemiology of GAS in the US CID 2007:45 (1 October) 861

10 65 years and persons aged years with diabetes or asthma United States, MMWR Morb Mortal Wkly Rep 2004; 53: Tyrrell GJ, Lovgren M, Kress B, Grimsrud K. Invasive group A streptococcal disease in Alberta, Canada (2000 to 2002). J Clin Microbiol 2005; 43: Lamagni T, Efstratiou A, Vuopio-Varkila J, Jasir A, Schalen C. The epidemiology of severe Streptococcus pyogenes associated disease in Europe. Euro Surveill 2005; 10: O Brien KL, Levine OS, Schwartz B. The changing epidemiology of group A streptococcus infections. Semin Pediatr Infect Dis 1997;8: Givner LB, Abramson JS, Wasilauskas B. Apparent increase in the incidence of invasive group A beta-hemolytic streptococcal disease in children. J Pediatr 1991; 118: Factor SH, Levine OS, Schwartz B, et al. Invasive group A streptococcal disease: risk factors for adults. Emerg Infect Dis 2003; 9: Factor SH, Levine OS, Harrison LH, et al. Risk factors for pediatric invasive group A streptococcal disease. Emerg Infect Dis 2005; 11: Cowgill KD, Van Beneden C, Wright C, Beall B, Schuchat A; the Active Bacterial Core Surveillance Team. Burden of streptococcal toxic shock syndrome and necrotizing fasciitis United States, 2002 [abstract 238]. In: Program and abstracts of the 43rd Annual Meeting of the Infectious Diseases Society of American (San Diego, California). Alexandria, VA: Infectious Diseases Society of America, 2003: Prevention of Invasive Group A Streptococcal Infections Workshop Participants. Prevention of invasive group A streptococcal disease among household contacts of case patients and among postpartum and postsurgical patients: recommendations from the Centers for Disease Control and Prevention. Clin Infect Dis 2002; 35:950 9 (erratum: Clin Infect Dis 2003; 36:243). 24. Shulman ST, Tanz RR, Kabat W, et al. Group A streptococcal pharyngitis serotype surveillance in North America, Clin Infect Dis 2004; 39: Teixeira LM, Barros RR, Castro AC, et al. Genetic and phenotypic features of Streptococcus pyogenes strains isolated in Brazil that harbor new emm sequences. J Clin Microbiol 2001; 39: Lopardo HA, Vidal P, Sparo M, et al. Six-month multicenter study on invasive infections due to Streptococcus pyogenes and Streptococcus dysgalactiae subsp. equisimilis in Argentina. J Clin Microbiol 2005; 43: Dey N, McMillan DJ, Yarwood PJ, et al. High diversity of group A streptococcal emm types in an Indian community: the need to tailor multivalent vaccines. Clin Infect Dis 2005; 40: Abdissa A, Asrat D, Kronvall G, et al. High diversity of group A streptococcal emm types among healthy schoolchildren in Ethiopia. Clin Infect Dis 2006; 42: CID 2007:45 (1 October) O Loughlin et al.