Journal Logo


The Role of β-Hemolytic Streptococci in Causing Diffuse, Nonculturable Cellulitis

A Prospective Investigation

Jeng, Arthur MD; Beheshti, Manie MD; Li, John MD; Nathan, Ramesh MD

Author Information
doi: 10.1097/MD.0b013e3181e8d635
  • Free



Soft-tissue infections continue to be among the most common diseases seen in the hospital setting. In the presence of a culturable source (for example, abscess, wound, ulcer), the etiologic pathogen can be identified and therapy can be tailored specifically. Commonly, however, patients have soft-tissue infections without a culturable source, such as non-abscess or non-ulcer-associated cellulitis, and in such cases, therapy has been empirically started for the 2 most common skin pathogens: β-hemolytic streptococci (BHS) and Staphylococcus aureus. The BHS mainly consist of Streptococcus pyogenes (group A streptococcus or GAS) but also include Streptococcus agalactiae (group B streptococcus or GBS) and Streptococcus dysgalactiae subsp. equisimilis, which are primarily carbohydrate grouped as C (GCS) and G streptococcus (GGS). Now several decades old, historical studies, which looked at skin infections in dog models of lymphedema, measured streptococcal serologies in patients with cellulitis, cultured incision and drainage sites in acute lymphangitis, and used direct immunofluorescence technique on skin biopsy specimens, have shown BHS to be the predominant cause of such infections, with S aureus making up the rest.5,21,27,41,46,62,65 Fortunately, the gram-positive β-lactam antibiotics, such as oxacillin, nafcillin, and cefazolin, possessed excellent activity against both bacteria, and thus identifying the responsible pathogen was largely unnecessary.

The recent epidemic of community-associated methicillin-resistant S aureus (CA-MRSA), however, which is not susceptible to the current β-lactam antibiotics, raises the question of whether empiric therapy with such agents is still feasible.15 Our emergency department spearheaded a nationwide study showing that 76% of culturable soft-tissue infections were due to S aureus, with methicillin-resistant S aureus (MRSA) responsible for 59% of cases overall.47 But the question remains whether this disturbing trend can be extrapolated to the case of diffuse, nonculturable cellulitis, which may represent a different pathogenic process.

This issue is critical when formulating guidelines for the empiric treatment of soft-tissue infections. If necessary, a change in empiric therapy to cover MRSA for every cellulitis case would have some undesirable consequences, as vancomycin, the antibiotic of choice for severe MRSA infections, is significantly less potent than the β-lactams, is more toxic, and does not have an oral formulation for completion of therapy. Oral regimens also pose challenges, as such agents, although perhaps adequate for MRSA, have limitations for BHS, such as decreased potency (trimethoprim-sulfamethoxazole), high resistance rates (tetracyclines), and low threshold for development of resistance (rifampin); or they are cost prohibitive (linezolid). Furthermore, the routine use of more than 1 antibiotic to cover both BHS and MRSA, if not necessary, would needlessly place the patient at risk for developing adverse drug reactions, promote the emergence of drug-resistant bacteria, and raise health care costs.

A new, current study is needed to determine if BHS remain the main etiologic agents of nonculturable cellulitis. Because wound cultures cannot be performed on this infection type, acute/convalescent serologies are among the only means to answer this question, with anti-streptolysin O (ASO) antibodies rising after infections from GAS, GCS, and GGS, and anti-deoxyribonuclease-B (anti-DNase-B or ADB) antibodies after GAS infections.3 Using these serologic methods, as well as blood cultures, we performed a prospective investigation to determine the prevalence of BHS in causing diffuse, nonculturable cellulitis, and also analyzed the effectiveness of empiric gram-positive β-lactam antibiotics for the treatment of such infections.



We conducted a prospective investigation on the cause of diffuse, nonculturable cellulitis between December 2004 and June 2007 at Olive View-UCLA Medical Center, a county hospital of Los Angeles. The protocol was approved by the institutional review board, and all participants signed an informed consent. All patients aged 18 years or older admitted with the diagnosis of "cellulitis" or "soft-tissue infection" were screened during the study period. Patients were recruited for the study if the cellulitis was diffuse and lacked a culturable source (for example, abscess, furuncle, wound, ulcer). Exclusion criteria were infections involving periorbital, perineal, and groin regions; animal/human bites; and foreign objects; and patients with neutropenia (absolute neutrophil count <500/mm3), necrotizing fasciitis, gangrene, myositis, or osteomyelitis and having had a soft-tissue infection or pharyngitis within the past year, due to confounding issues with interpretation of the serologies.

Serologic Analysis

For all participants, information on demographic characteristics, clinical presentation, potential risk factors for cellulitis, and treatment in the outpatient/inpatient setting was recorded on a standardized form. All patients underwent a venipuncture during their hospital stay for ASO and ADB antibodies, which represented the "acute serologies." ASO is expected to rise within 2 weeks of infection for GAS, GCS, and GGS, and ADB in a similar time period for GAS; both antibody titers should decline in 3-6 months.3 Patients were asked to return to the Infectious Diseases Clinic within 2-12 weeks to assess clinical response and to obtain a second blood draw for ASO and ADB, which represented the "convalescent serologies." ASO serologies were performed at the reference laboratory Focus Diagnostics (Cypress, CA)25 using the Dade Behring N Latex ASL assay, and ADB serologies were performed at the reference laboratory Quest Diagnostics (San Juan Capistrano, CA)52 using the Wampole Streptonase-B assay. A positive result was defined by seroconversion from acute to convalescent phase (0.2 log rise from acute to convalescent, with final titer >200 IU/mL for ASO and/or ≥120 U/mL for ADB) or 2 persistently high positive titers (ASO >200 IU/mL and/or ADB ≥120 U/mL).3,23,25,36,52,69

Studies on normal population sera have shown ASO titers to vary by age, but the mean ranges from 83 to 159 IU/mL (standard error of mean, 7-11),56 and 98.2% of healthy adults will have titers <170 IU/mL;36 this has led to the international standardization of 200 IU/mL as the upper limit of normal in adults.25 Such studies on normal adult sera have shown ADB titers also to vary by age, but the mean ranges from 11 to 63 U/mL (standard error of mean, 15-48),56 and 95% of healthy adults will have titers <120 U/mL;36 this has led reference laboratories to use 85 U/mL as the upper limit of normal.52 For the current study, we selected ≥120 U/mL as a positive value for ADB to increase specificity, at the expense of some sensitivity. Patients were also considered positive for BHS if they were bacteremic with BHS, regardless of serologic results; this seronegativity with bacteremia was only relevant for GBS, which would not be expected to elicit an ASO or ADB response.

Response to β-Lactam Antibiotics

To gain insight into the role of MRSA in these nonculturable infections, all participants who received ≥48 hours of treatment and who did not receive >1 dose of antibiotic active against MRSA (for example, vancomycin, clindamycin, trimethoprim-sulfamethoxazole, tetracycline, daptomycin, linezolid, tigecycline) were analyzed for their response to the β-lactam antibiotics. The admitting physicians were asked to use a gram-positive β-lactam, preferably cefazolin or oxacillin, on study patients. β-Lactam success was defined when the patient was clinically improved on the β-lactam antibiotic alone, and did not need any anti-MRSA antibiotics during their hospitalization (beyond the 1 dose, in those patients who received it). Patients were reassessed at follow-up visits to ascertain sustaining response. We also performed this analysis on the subgroup of patients who were dropped from the serologic analysis because they were lost to follow-up, as their response to inpatient β-lactam therapy could still be evaluated.

Statistical Analysis

Descriptive statistics were used to quantify the proportion of cellulitis due to BHS as defined by seroconversion/positivity or by blood culture positivity and also within the subgroups by risk factors. Descriptive statistics were also used to quantify the percentage response rate to β-lactam antibiotics in the patients who could be evaluated.

Funding Sources

This study was funded by grants from Olive View-UCLA Central Resource Management (CRM) and Olive View SEED Awards (2005, 2006) via the Education and Research Institute (ERI). The funding sources had no role in designing the study; collecting, analyzing, and interpreting the data; or approving the final manuscript.


Patient Characteristics

We enrolled 248 eligible patients; 69 patients were excluded for the following reasons: 62 were lost to follow-up (lacking convalescent serologies), 4 were diagnosed with abscesses after enrollment, and 3 were diagnosed with osteomyelitis (Figure 1). The remaining 179 patients completed the study. The mean age of patients with cellulitis was 47.9 years (range, 18-93 yr, Table 1), with a slight male predominance (67.6%). Slightly over half (58.7%) of the patients had at least 1 of 5 common risk factors for cellulitis, including diabetes mellitus, cirrhosis, lymphedema, history of recurrent cellulitis, and/or obesity. The majority of patients had cellulitis of the lower extremity (75.8%), followed by upper extremity (15.2%), trunk (5.6%), and face (3.9%).

Participant flow diagram. *These patients could not be evaluated for β-lactam response because they empirically received >1 dose of an antibiotic active against MRSA. †Of patients who were dropped from primary analysis because of being lost to follow-up and/or lacking convalescent serologies (62 patients), these could not be evaluated for β-lactam response due to empiric use (>1 dose) of antibiotic active against MRSA (25 patients) or premature discharges, such as leaving against medical advice or transferring to another facility (12 patients).
Patient Characteristics, Risk Factors, and Prevalence of BHS

Serologic Analysis

Overall, 73.2% (131/179) of patients were diagnosed with BHS infections, by virtue of positive serologies and/or blood cultures, with 126 patients seropositive for ASO, ADB, or both. See Figure 2 for each patient's serologic value plotted graphically in the acute and convalescent phase. Regarding how patients serologically responded to BHS, half the patients mounted both ASO and ADB antibodies (63/126, 50%), although a significant proportion responded with either ADB (37/126, 29%) or ASO (26/126, 21%) alone. Because DNase-B should be made only by GAS, while streptolysin-O is produced by GAS as well as by GCS and GGS, 1 interpretation of these results is that the majority (79%) of seropositive BHS cases was caused by GAS (all cases with a positive ADB), while 21% were caused by GAS, GCS, or GGS (ASO-positive/ADB-negative cases). Four patients with diabetes mellitus and 1 with cirrhosis had negative serologies but were diagnosed with BHS because they were bacteremic with GBS. Slightly more men than women were diagnosed with BHS infections (78.5% vs 62%), and in all recorded associated risk factors, BHS was the major cause of cellulitis, with the highest seen in the "recurrent cellulitis" group (88.2%) and the lowest in patients with diabetes (63.3%) (Table 1).

Distribution of patients' ASO and ADB titers. Shown are each patient's acute (drawn during hospitalization) and convalescent (drawn 2-12 wk later) ASO and ADB titers. A. ASO titers: BHS diagnosed if patients seroconverted or had 2 highly positive titers (>200 IU/mL).3,23,25,36,69 Data points at top of graph represent values ≥800, usually in several thousands (maximum patient titer = 6437 IU/mL). B. ADB titers: BHS diagnosed if patients seroconverted or had 2 highly positive titers (≥120 IU/mL).3,23,36,52 Data points at top of graph represent values ≥500, usually above limit of detection (>1360 IU/mL).

Patients With Positive Blood Cultures

In total, 8 patients had positive blood cultures, although not all of the enrolled patients had blood cultures drawn before receiving antibiotics. After careful chart review, we found that 82 of the 103 sets of blood cultures were performed before receipt of antibiotics (from home or emergency department), giving a more accurate blood culture positivity rate of 9.7% (8/82). Of these bacteremic patients, 5 had GBS, 1 had GCS, 1 had GAS, and 1 had MRSA. Four of 5 patients with GBS bacteremia had diabetes mellitus, and the remaining patient with GBS, as well as the MRSA case, had cirrhosis. The patient with GCS bacteremia had psoriasis and recurrent cellulitis and mounted a robust ASO response (228-1012 IU/mL). The patient with GAS had lymphedema and mounted a vigorous ADB response (170-960 U/mL). It is noteworthy that the 1 patient with MRSA bacteremia also had highly positive ADB and ASO serologies, representing possible co-infection. As expected, all patients with GBS had negative serologies, as did all patients who were erroneously enrolled and dropped from the study because they had abscesses that grew S aureus. Of the 49 patients with diabetes enrolled in the study, 29 had blood cultures sent, although only 24 of these were done before receipt of antibiotics; therefore, for those done correctly, blood culture positivity was 16.7% (4/24) in the diabetic population.

Response to β-Lactam Antibiotics

We analyzed the clinical effectiveness of β-lactam antibiotics for the empiric treatment of nonculturable cellulitis. This enabled us to investigate the role of MRSA in such cases, as MRSA should not respond to β-lactam antibiotics. Eighty-three of the enrolled patients could not be evaluated because of empirical (>1 dose) use of agents active against MRSA (see Figure 1). The vast majority (80/96, 83%) of evaluable patients was treated with cefazolin 1 g intravenous (IV) every 8 hr. Other patients were treated with oxacillin 1 g IV every 6 hours (4% of evaluable patients), ampicillin/sulbactam 3 g IV every 6 hours (4%), and penicillin G 2 million units IV every 6 hours (4%), with 1 patient each (1%) treated with ceftriaxone 1 g IV daily, ertapenem 1 g IV daily, piperacillin/tazobactam 3.375 g IV every 6 hours, and amoxicillin/clavulanic acid 875 mg orally twice daily.

Of the evaluable patients, 96% (92/96) had a successful response to β-lactams. In the subgroup of patients diagnosed with BHS infections, 97% (71/73) clinically responded to the β-lactams. One failed case involved a patient who was treated with cefazolin for 1 day and subsequently switched to oxacillin and clindamycin. The second patient coded as a failure had vigorous serologic responses to ASO and ADB but also grew MRSA in the blood. This patient responded clinically to cefazolin and was discharged with cephalexin, but upon identification of MRSA in the blood cultures, he was called back to receive vancomycin for 2 days, was ruled out for endocarditis, and was subsequently discharged with the additional prescription for trimethoprim-sulfamethoxazole.

In patients without evidence of streptococcal infection, 91% (21/23) of patients responded to β-lactam antibiotics, suggesting a low incidence of MRSA, even in those patients who did not have BHS identified as the causative agent. Both of the patients with β-lactam failure were initially treated with cefazolin (for 1 d and 6 d, respectively) and subsequently switched to clindamycin. Additionally, we evaluated 25 of the patients who were lost to follow-up for their inpatient response to β-lactams, of which 1 was a failure. Overall, 95.8% (116/121) of patients who could be evaluated for β-lactam response were successfully treated with these antibiotics (Figure 1).


We conducted this large prospective study and have found that in the current era of CA-MRSA, BHS appear to remain the primary cause (73%) of diffuse, nonculturable cellulitis. Because the investigation studied patients over several years, enrolled a large number of patients, and took place in 1 of the largest, most ethnically diverse cities in the United States, we believe the results are fairly generalizable. The investigation commenced in the months immediately following the notable multicenter emergency department study showing that S aureus was responsible for 76% of culturable soft-tissue infections nationwide, with a 59% overall prevalence of MRSA, and a 51% prevalence of MRSA at our institution.47

The present investigation enrolled patients excluded from the previous study, namely those with cellulitis lacking a culturable source. These results are consistent with results of much older studies,5,21,27,41,46,62,65 and also with results of more recent studies that have attempted to investigate this issue via more indirect means. One group looking at the relationship between tinea pedis and recurrent cellulitis found that 85% of interdigital swabs grew BHS, of which GGS represented 53% and GAS 24%, with S aureus being isolated in a smaller percentage (45%).59 A more recent (2008) case-controlled study60 from Finland demonstrated that in 90 patients admitted for cellulitis, 29% of skin abrasion/lesions, fissured toe webs, or blood cultures grew BHS, most often GGS (22%) followed by GAS (7%), with S aureus recovered as the sole pathogen in 11% of cases.

It is inherently challenging to investigate the cause of diffuse, nonculturable cellulitis given that, by definition, no focus on the skin is culturable in these patients. Cultures of skin abrasions and interdigital spaces are indirect methods, and bacteria isolated from these areas may or may not represent the true pathogen. More direct methods, such as obtaining cultures subcutaneously, are relatively invasive while yielding disappointing results in adults, ranging from 2% for leading edge aspirates to 20% in skin biopsy cultures,22,33 suggesting that the inoculum of the offending bacteria may be low subcutaneously and/or its recovery may be inhibited by prior antibiotic use or local anesthetics. Blood cultures have traditionally also yielded low recovery rates, ranging from 0 to 24%,44 being positive primarily in more severe cases or in those who are immunocompromised.49

Analysis of acute and convalescent serologies is a powerful tool commonly used in epidemiologic investigations of disease outbreaks and is 1 of the few feasible methods by which to investigate etiologies of difficult-to-culture infections. As such, institutions, such as the Centers for Disease Control and Prevention, have utilized serologies to diagnose infections from numerous fastidious organisms, such as viruses (West Nile virus, measles, mumps, dengue)9,11,12,14 and other difficult-to-culture bacteria (Coxiella burnetii/Q fever, Legionella species).10,13

The serologies used in the current investigation have been extensively studied in diagnosing infections involving BHS. In 1932, Todd67 found that neutralizing antibodies to streptolysin-O were moderately elevated in those with recent GAS infections and highly elevated in those with acute rheumatic fever. Since then, numerous researchers have found ASO titers to rise after recent GAS infections, rheumatic fever, and glomerulonephritis.2,43,48,55,63,70 Measurements of ASO titers from patients in convalescence from recent GAS infections (pharyngitis, tonsillitis, scarlet fever) were shown to be significantly higher than those from control subjects.53 Specifically, in streptococcal pharyngitis, ASO titers were found to reach a peak in 2-3 weeks, remaining elevated for 3-6 months, before declining thereafter;3 a 0.2 log rise in titers has been found to be indicative of recent streptococcal infection.69

Because GCS and GGS also produce streptolysin-O, these 2 BHS also cause elevations in ASO titers,32,58 making this test useful for capturing infections caused by 3 of the BHS: GAS, GCS, and GGS. Unfortunately, ASO antibody responses have been shown to be weak in the streptococcal skin infection pyoderma (impetigo), seen in only 27%-43% of such patients in outbreak studies,20,34 presumably because the enzyme is bound by skin lipids/cholesterol thus preventing its interaction with immune cells,35 whereas ADB antibody responses have been shown to be more robust, seen in 56%-67% of such patients. Like ASO, DNase-B antibody titer elevations have been shown to be indicative of recent GAS infections when compared with levels of uninfected, control groups, both in impetigo and pharyngeal infections,6 as well as in its immunologic sequelae, rheumatic fever (82% had elevation)63 and post-streptococcal glomerulonephritis (85%-90% had elevation).4,20,34 Because these 2 serologies have not been extensively studied for diffuse, nonculturable cellulitis, it was unclear which test would be best for use in this infection type. A historical study using such acute/convalescent serologies in necrotizing fasciitis reported that 64% (9/14) of these patients mounted an ASO, and 86% (12/14) mounted an ADB response.40 A similar study performed on cellulitis and erysipelas demonstrated that 74% (26/35) exhibited a rise in ASO and 43% (15/35), in ADB titers.41 Since measurement of both serologies has shown the highest sensitivity for streptococcal infections in general (pharyngitis, scarlet fever, impetigo), ranging from 85% to 90%,23,26 we selected these 2 tests for the current study, to help investigate not only the prevalence of BHS in diffuse cellulitis, but also how useful each serology is for this infection type.

We found that half the patients mounted an antibody response to both ASO and ADB (50%) if they had cellulitis with GAS, although a significant proportion responded only with ADB (29%), as was traditionally seen in cases of pyoderma, perhaps due to similar inhibitory mechanisms. Approximately 21% of patients mounted an ASO elevation without an ADB response, perhaps due to infections with GCS or GGS, neither of which possess DNase-B, although infection with GAS is still possible. In this study, ASO elevation was seen in 71% and ADB in 79% of patients diagnosed serologically, underscoring the importance of both tests for evaluating BHS in cellulitis. Although measurement of antibodies for GBS and S aureus would have been helpful, no tests with acceptable sensitivity/specificity exist for these 2 bacteria, so it was not possible in this investigation.

Much has been published on the value of blood cultures in cellulitis, with positivity ranging from 0 to 24%.44 In the current study, blood cultures were positive in ∼10% of patients who had them drawn correctly. Caution should be exercised in interpreting published reports on blood culture positivity in patients with cellulitis and other infections, since, after more careful scrutiny of our data, we found that a sizable proportion of the cultures had been sent after antibiotics were either taken at home or given in the emergency department, effectively rendering them useless. Studies have shown that blood cultures are of significantly higher yield in patients with more severe cellulitis or with immunocompromising disorders, such as diabetes mellitus and cirrhosis.49 In the current investigation, the blood culture positivity in patients with diabetes was ∼17%, all of which were from GBS, an interesting finding that would not have been revealed had blood cultures not been drawn. This bacteremia rate in diabetics is similar to that (18.8%) found in multicenter clinical trials on the treatment of diabetic foot ulcer infections.42 The low incidence of GAS bacteremia in enrolled patients is somewhat curious. GAS blood culture positivity was present at our institution during the study period and occurred more frequently (11 additional cases) than GBS, but these patients were not captured in the current study because they were diagnosed clinically with necrotizing fasciitis (5/11) or had acquired immunodeficiency syndrome (AIDS) (1/11), both exclusion criteria, or did not have a soft tissue source (2 cases of necrotizing pneumonia, 1 septic arthritis, 1 severe pharyngitis, and 1 malignant scarlet fever). Hence, as blood cultures remain the only means, aside from serologies, to identify the etiologic pathogen in nonculturable cellulitis, it may be prudent to obtain them in more severe cases or in patients with immunocompromising disorders (for example, diabetes mellitus, cirrhosis), as the rate of bacteremia in these groups is higher. Other conditions where bacteremia rates have been found to be higher, thus increasing the yield and value of blood cultures, include cellulitis associated with lymphedema, fevers/chills, water exposure, and buccal/periorbital involvement.57

Consistent with the finding that most cases of diffuse cellulitis were caused by BHS, analysis of the response to β-lactams revealed that these antibiotics remain very effective for empiric treatment (∼96% response), even though this class does not possess MRSA activity. Most (83%) patients in this study were treated with the gram-positive β-lactam cefazolin. This finding also suggests that most of the patients who were not diagnosed with BHS either had methicillin-susceptible S aureus (MSSA) or still had BHS (for example, GBS) but were not diagnosed as such by our methods. The latter is especially probable in the patients with diabetes who had a lower serologic response to ASO and ADB (63%) but who responded favorably to β-lactams; as this group was notable for bacteremia with GBS, it is likely that some of these serologically negative individuals were infected with GBS, and therefore truly had BHS, but were not classified as having BHS if they were not bacteremic. Because some of the β-lactams used for treatment, such as cefazolin and ampicillin/sulbactam, do have some activity against gram-negative bacteria, the question may be raised whether this bacterial group may be involved, especially in patients who did not have evidence of BHS infections. However, gram-negative bacteria have never been shown to be a significant cause of uncomplicated cellulitis in the non-neutropenic host. Some rare exceptions of this include cellulitis involving water exposure (Vibrio, Aeromonas species), mammalian bites (Pasteurella, Eikenella species), or unimmunized children (Hemophilus influenza), all of which were excluded or were not factors in the current investigation. The success rate of β-lactams for the empiric treatment of uncomplicated cellulitis in this study is consistent with that found in several β-lactam treatment trials performed several years earlier, with efficacy rates ranging from 83% to 94%.7,17,28,30,39

The results of the current study and other recent investigations47 can be used to formulate guidelines for the empiric treatment of uncomplicated soft-tissue infections. One algorithm being used successfully at our institution calls for determining initially whether the patient has a purulent/culturable focus (abscess, furuncle, ulcer) of the soft-tissue infection (Figure 3). Given the prevalence of MRSA in these cases, empiric treatment for MRSA should be given (for example, vancomycin). Culture and drainage should be performed on the pyogenic focus, and subsequent outpatient antibiotic management can be based on the culture/susceptibility results. If the patient has diffuse, nonculturable cellulitis, then a gram-positive β-lactam (oxacillin, cefazolin) can be empirically administered, given that most of these cases are caused by BHS, which have remained exquisitely sensitive to and have been unable to evolve or acquire resistance to this highly bactericidal antibiotic class.

Proposed cellulitis treatment algorithm. Per methods, algorithm would exclude infections involving periorbital, perineal, and groin regions; animal/human bites; foreign objects; and patients with neutropenia, necrotizing fasciitis, gangrene, myositis, or osteomyelitis. *MRSA treatment would primarily be vancomycin, given its comprehensive coverage of both S aureus and BHS, but other options could include daptomycin, telavancin, linezolid, and clindamycin. The former 2, like vancomycin, are available only via the IV route, while the latter 2 are available in oral formulations; linezolid is mainly limited by its cost in oral formulation, and clindamycin by its appreciable resistance rates in BHS and S aureus. Abbreviations: CT = computerized tomography, I&D = incision and drainage.

Although penicillin would be sufficient, and even preferable, for the treatment of BHS, a smaller, yet appreciable, proportion (∼27%) of diffuse cellulitis cases in the current study was not diagnosed as being caused by BHS. If all such cases were caused by S aureus, then penicillin could not be used as a single agent empirically, given the high resistance rate (>95%) of S aureus to penicillin. The success rate of β-lactams in the current study suggests that MRSA plays a smaller role in diffuse cellulitis, and, as such, coverage for S aureus can focus on MSSA. For MSSA, the gram-positive β-lactams are also the most potent antibiotic class, and the fact that they are precluded from use in MRSA infections is, in part, why MRSA is associated with such poorer outcomes when compared with MSSA.

Patients who respond to the β-lactam can be easily discharged on the respective oral formulation (dicloxacillin, cephalexin). Thus, due to the high potency, efficacy, and coverage for both BHS and MSSA, and to the ease of oral transition in this infection type where cultures are lacking for guidance in outpatient antibiotic selection, oxacillin/dicloxacillin and cefazolin/cephalexin represent the ideal, empiric antibiotic choice for diffuse cellulitis. This recommendation has been previously reported in a state-of-the-art "Clinical Practice" article but was acknowledged to be done so without a wealth of or recent data, given the difficult nature of studying this disease.57

If the patient does not respond to the β-lactam, as seen in ∼4% of patients in the current investigation, then the clinician should switch antibiotics to cover MRSA and obtain an imaging study, such as computerized tomography (CT) scan with contrast, of the affected area to look for an occult focus that may need to be drained. This algorithm represents a cost-effective, evidence-based approach that will minimize the wanton use of broader or more-than-necessary antibiotics, and thus help to reduce adverse drug reaction rates, health-care costs, and the emergence of resistant bacteria.

Although vancomycin, the antibiotic of choice for MRSA, does possess comprehensive activity against both BHS and S aureus (including MRSA), its empiric use in diffuse, nonculturable cellulitis is less desirable for several reasons. Glycopeptides are well known to be significantly less potent (less rapidly bactericidal) than the β-lactam antibiotics for the treatment of gram-positive bacterial infections, and have been associated with higher failure and slower response rates;24,29,61,64 this may be in part why MRSA infections, which preclude β-lactam usage, have been associated with higher morbidity, mortality, failure rates, and associated hospital days and costs.18,37,38 Vancomycin is also more toxic, primarily to the kidneys, compared with the β-lactam antibiotics. Vancomycin levels need and should be drawn to assure adequate troughs, which incurs greater costs. And, from a practical standpoint, once the patient has responded to this antibiotic and is deemed ready for discharge, no oral, bio-available vancomycin formulation is available for completion of therapy, making outpatient antibiotic selection problematic.

If the patient with diffuse cellulitis does respond to vancomycin, either as monotherapy or in combination, one is not able to exclude MRSA as a possible cause of the infection, which may needlessly drive the practice of prescribing more than 1 oral antibiotic to cover both MRSA and BHS (for example, sulfa/tetracycline and β-lactam), or more costly options (linezolid) on discharge. Antibiotic polypharmacy, if unnecessary, only serves to increase the risk of adverse drug reactions, medication costs, and the emergence of antibiotic resistant bacteria. Finally, unwarranted routine use of vancomycin also encourages the emergence of worrisome bacteria, such as vancomycin-intermediate-resistant and heteroresistant strains of S aureus ("VISA" and "hVISA"), S aureus with increasing minimum inhibitory concentrations to vancomycin ("vancomycin MIC creep"), and vancomycin-resistant enterococci, all of which pose therapeutic challenges, but the last bacteria of which few therapeutic options truly exist. Therefore, vancomycin should be reserved for infections where MRSA is known to be a significant cause of disease, such as that found in culturable soft-tissue infections; the results of the current study indicate that MRSA is not participating in this disease process significantly.

For patients with diffuse cellulitis who are unable to take β-lactams (for example, patients with allergy), the preferable alternative antibiotic would be clindamycin, as this medication, unlike sulfas or tetracyclines, does have activity against both BHS and S aureus, and, unlike vancomycin, does have an oral formulation for completion of therapy. While it has the advantage of possessing fair activity against MRSA, clindamycin is less favorable than the β-lactams for diffuse cellulitis not only because of its poorer tolerability (nausea, vomiting, diarrhea) and its traditional Clostridium difficile risk association, but also because of worrisome trends in resistance rates for both BHS and S aureus. Although microbiological surveys in the United States and Europe have shown constitutive clindamycin resistance to be relatively low in GAS (1.4%-5.2%),19,31,45 they have shown the other BHS to be relatively resistant (GBS 15.8%-26.4%, GCS 15.8%, GGS 33.3% resistance).19,50,68 Furthermore, inducible clindamycin resistance (iMLSb phenotype), where the bacteria appear susceptible on initial testing but express resistance when exposed to the antibiotic, which is already well appreciated in S aureus, is being increasingly recognized in BHS (including GAS), with studies demonstrating that 20%-55% of erythromycin-resistant/clindamycin-susceptible BHS isolates actually had inducible clindamycin resistance when tested for it.19,31,54 And, while clindamycin is widely considered to be a good agent for S aureus, recent microbiological surveys in the United States have shown outpatient S aureus and, specifically, CA-MRSA strains, to possess appreciable clindamycin resistance (15% and 12%-15%, respectively); this percentage rises to 23%-24% when inducible resistance (iMLSb) is tested.16,66

As a further concern about clindamycin, we note that the predominant CA-MRSA strain causing skin infections is from the clonal pulse-field gel electrophoresis (PFGE) type USA-300, which is mostly (95%-96%) susceptible to clindamycin and which usually lacks inducible resistance. However, recent surveys, including a recent multicenter clinical trial on the treatment of complicated soft-tissue infections, have shown that an appreciable amount (16%) of CA-MRSA strains in the United States are not of this PFGE type,8 which may account for the discernible clindamycin resistance seen in the surveys cited above. PFGE type USA-400, the second most common CA-MRSA strain, is predominantly clindamycin resistant. A Canadian emergency department recently reported a lower rate of USA-300 type in their CA-MRSA strains (61%) compared with those in the United States, with a concomitant decrease in CA-MRSA susceptibility to clindamycin (79%).1 These studies show that clindamycin is moderately to fairly active against BHS and S aureus, but its lowered activity against BHS, the predominant cause of diffuse cellulitis, makes it a less attractive first-line option than the β-lactams for this infection type.

Addressing this issue, authors of a recent pharmaco-economic decision-analysis investigating the most cost-effective outpatient antibiotic for diffuse, nonculturable cellulitis concluded that cephalexin (or dicloxacillin) was optimal, over clindamycin and trimethoprim/sulfamethoxazole, when using various assumptions, including that BHS caused 63% and S aureus 37% (of which 27% was MRSA) of such cases.51 In their model, clindamycin becomes more cost-effective than β-lactams only when S aureus causes 41%-80% of cases, and trimethoprim/sulfamethoxazole is best when S aureus causes >80% of cases. The results of the present study add clinical evidence and scientific support to these theory-driven conclusions.

The current study design has several limitations. Although measuring acute and convalescent serologies has long been used as a means of diagnosing infections where obtaining cultures is problematic or impossible, false negatives may occur in individuals who do not mount an antibody response to the offending pathogen. And because reliable serologies for GBS are not available, and thus are not used in the current study, this type of BHS as a cause of cellulitis is likely underreported, especially in the diabetic population. Thus, the ∼73% prevalence rate of BHS in causing diffuse cellulitis may be an underestimation. This study also does not address which bacteria caused the cellulitis in those who were not diagnosed with BHS. Data from past studies and the high response rate to cefazolin suggest that these patients had MSSA or other streptococci (for example, GBS). The question of co-infection with other pathogens aside from BHS (especially S aureus) is also not addressed by this investigation, and thus co-infection remains a possibility; the 1 patient with vigorous serologic responses to ASO and ADB while bacteremic with MRSA may represent such a case. However, diffuse cellulitis, as with most other infections, such as pneumonia, pyelonephritis, meningitis, and even abscess-associated soft-tissue infections, is generally considered to be primarily a monomicrobial disease process. Again, the high success rate of cefazolin is reassuring that diffuse cellulitis is mainly caused by BHS as the sole pathogen or, if co-infecting bacteria are indeed present, they are responsive to these antibiotics (for example, MSSA or other streptococci such as GBS). Barring the development of highly sensitive/specific serologies for S aureus and GBS, the only means by which to investigate the possibility of co-pathogens would be through molecular analysis of the subcutaneous tissue (for example, via polymerase chain reaction). This methodology may be an avenue of future research, although its sensitivity in identifying the offending pathogen in subcutaneous tissue is currently unknown; its over-sensitivity in picking up bacterial skin flora not involved in the infection is also of concern.

In conclusion, in the current prospective investigation we found that even in this era of CA-MRSA, BHS remain the primary cause of diffuse, nonculturable cellulitis. Treatment with β-lactam antibiotics appears to be highly efficacious, and a cost-effective algorithm can be useful for the empiric management of uncomplicated soft-tissue infections based on the presence or absence of a culturable source.


The authors thank the Olive View-UCLA CRM and ERI for the grants that made this study possible.


1. Al-Rawahi GN, Reynolds S, Porter SD, Forrester L, Kishi L, Chong T, Bowie WR, Doyle PW. Community-associated CMRSA-10 (USA-300) is the predominant strain among methicillin-resistant Staphylococcus aureus strains causing skin and soft tissue infections in patients presenting to the emergency department of a Canadian tertiary care hospital. J Emerg Med. 2008 Mar 4. (Epub ahead of print).
2. Anderson HC, Kunkel HG, McCarthy M. Quantitative antistreptokinase studies in patients infected with group A hemolytic streptococci: a comparison with serum antistreptolysin and gamma globulin levels with special reference to the occurrence of rheumatic fever. J Clin Invest. 1948;27:425.
3. Ayoub EM. Immune response to group A streptococcal infections. Pediatr Infect Dis J. 1991;10:S15-S19.
4. Ayoub EM, Wannamaker LW. Evaluation of the streptococcal deoxyribonuclease B and diphosphopyridine nucleotidase antibody tests in acute rheumatic fever and acute glomerulonephritis. Pediatrics. 1962;29:527-538.
5. Bernard P, Bedane C, Mounier M, Denis F, Catanzano G, Bonnetblanc JM. Streptococcal cause of erysipelas and cellulitis in adults. A microbiologic study using direct immunofluorescence technique. Arch Dermatol. 1989;125:779-782.
6. Bisno AL, Nelson KE, Waytz P, Brunt J. Factors influencing serum antibody responses in streptococcal pyoderma. J Lab Clin Med. 1973;81:410-420.
7. Bucko AD, Hunt BJ, Kidd SL, Hom R. Randomized, double-blind, multicenter comparison of oral cefditoren 200 or 400 mg BID with either cefuroxime 250 mg BID or cefadroxil 500 mg BID for the treatment of uncomplicated skin and skin-structure infections. Clin Ther. 2002;24:1134-1147.
8. Campbell SJ, Deshmukh HS, Nelson CL, Bae IG, Stryjewski ME, Federspiel JJ, Tonthat GT, Rude TH, Barriere SL, Corey R, Fowler VG Jr. Genotypic characteristics of Staphylococcus aureus isolates from a multinational trial of complicated skin and skin structure infections. J Clin Microbiol. 2008;46:678-684.
9. Centers for Disease Control and Prevention (CDC). Dengue hemorrhagic fever-U.S.-Mexico border, 2005. MMWR Morb Mortal Wkly Rep. 2007;56:785-789.
10. Centers for Disease Control and Prevention (CDC). Follow-up on respiratory illness-Philadelphia, 1977. MMWR Morb Mortal Wkly Rep. 1997;46:50-56.
11. Centers for Disease Control and Prevention (CDC). Mumps epidemic-Iowa, 2006. MMWR Morb Mortal Wkly Rep. 2006;55:366-368.
12. Centers for Disease Control and Prevention (CDC). Outbreak of West Nile-like viral encephalitis-New York, 1999. MMWR Morb Mortal Wkly Rep. 1999;48:845-849.
13. Centers for Disease Control and Prevention (CDC). Q fever outbreak-Germany, 1996. MMWR Morb Mortal Wkly Rep. 1997;46:29-32.
14. Centers for Disease Control and Prevention (CDC). Update: measles-United States, January-July 2008. MMWR Morb Mortal Wkly Rep. 2008;57:893-896.
15. Chambers HF. The changing epidemiology of Staphylococcus aureus? Emerg Infect Dis. 2001;7:178-182.
16. Como-Sabetti K, Harriman KH, Buck JM, Glennen A, Boxrud DJ, Lynfield R. Community-associated methicillin resistant Staphylococcus aureus: trends in case and isolate characteristics from six years of prospective surveillance. Public Health Rep. 2009;124:427-435.
17. Corwin P, Toop L, McGeoch G, Than M, Wynn-Thomas S, Wells JE, Dawson R, Abernethy P, Pithie A, Chambers S, Fletcher L, Richards D. Randomised controlled trial of intravenous antibiotic treatment for cellulitis at home compared with hospital. BMJ. 2005;330:129-135.
18. Cosgrove SE, Sakoulas G, Perencevich EN, Schwaber MJ, Karchmer AW, Carmeli Y. Comparison of mortality associated withi methicillin-resistant and methicillin-susceptible Staphylococcus aureus bacermia: a meta analysis. Clin Infect Dis. 2003;36:53-59.
19. Diaz ML, Sanchez Torres MJ, Martin JA. Prevalence and mechanisms of erythromycin and clindamycin resistance in clinical isolates of β-haemolytic streptococci of Lancefield groups A, B, C and G in Seville, Spain. Clin Microbiol Infect. 2008;14:85-87.
20. Dillon HC Jr, Reeves MS. Streptococcal immune responses in nephritis after skin infection. Am J Med. 1974;56:333-346.
21. Drinker CK, Field ME, Ward HK, Lyons C. Increased susceptibility to local infection following blockage of lymph drainage. Am J Physiol. 1935;112:74-81.
22. Duvanel T, Auckenthaler R, Rhoner P, Harms M, Saurat JH. Quantitative cultures of biopsy specimens from cutaneous cellulitis. Arch Intern Med. 1989;149:293-296.
23. El-Khateeb MS. Comparison of antibody titres to streptococcal extracellular antigens. J Trop Pediatr. 1989;35:159-162.
24. Engemann JJ, Carmeli Y, Cosgrove SE, Fowler VG, Bronstein MZ, Trivette SL, Briggs JP, Sexton DJ, Kaye KS. Adverse clinical and economic outcomes attributable to methicillin resistance among patients with Staphylococcus aureus surgical site infection. Clin Infect Dis. 2003;36:592-598.
25. Focus Diagnostics, Incorporated. Directory of Laboratory Services. Updated January 2010. Last accessed January 4, 2010.
26. Gerber MA, Wright LL, Randolph MF. Streptozyme test for antibodies to group A streptococcal antigens. Pediatr Infect Dis J. 1987;6:36-40.
27. Ginsberg MB. Cellulitis: analysis of 101 cases and review of the literature. South Med J. 1981;74:530-533.
28. Giordano PA, Elston D, Akinlade Bk, Weber K, Notario GF, Busman TA, Cifaldi M, Nilius AM. Cefdinir versus cephalexin for mild to moderate uncomplicated skin and skin structure infections in adolescents and adults. Curr Med Res Opin. 2006;22:2419-2428.
29. Gonzalez C, Rubio M, Romero-Vivas J, Gonzalez M, Picazo JJ. Bacteremic pneumonia due to Staphylococcus aureus: a comparison of disease caused by methicillin-resistant and methicillin-susceptible organisms. Clin Infect Dis. 1999;29:1171-1177.
30. Grayson ML, Mcdonald M, Gibson K, Athan E, Munckhof WJ, Paull P, Chambers F. Once-daily intravenous cefazolin plus oral probenicid is equivalent to once-daily ceftriaxone plus oral placebo for the treatment of moderate to severe cellulitis in adults. Clin Infect Dis. 2002;34:1440-1448.
31. Hasenbein ME, Warner JE, Lambert KG, Cole SE, Onderdonk AB, McAdam AJ. Detection of multiple macrolide- and lincosamide-resistant strains of Streptococcus pyogenes from patients in the Boston area. J Clin Microbiol. 2004;424:1559-1563.
32. Hill HR, Caldwell GG, Wilson E, Hager D, Zimmerman RA. Epidemic of pharyngitis due to streptococci of Lancefield group G. Lancet. 1969;2:371-374.
33. Hook EW, Hooton TM, Horton CA, Coyle MB, Ramsey PG, Turck M. Microbiologic evaluation of cutaneous cellulitis in adults. Arch Intern Med. 1986;146:295-297.
34. Kaplan EL, Anthony BF, Chapman SS, Ayoub EM, Wannamaker LW. The influence of the site of infection on the immune response to group A streptococci. J Clin Invest. 1970;81:1405-1414.
35. Kaplan EL, Wannamaker LW. Suppression of antistreptolysin-O response by cholesterol and by lipid extracts of rabbit skin. J Exp Med. 1976;144:754-767.
36. Klein GC, Baker CN, Jones WL. "Upper limits of normal" antistreptolysin O and antideoxyribonuclease B titers. Appl Microbiol. 1971;21:999-1001.
37. Kollef MH. Limitations of vancomycin in the management of resistant Staphylococcal infections. Clin Infect Dis. 2007;45(Suppl 3):S191-S195.
38. Kopp BJ, Nix DE, Armstrong EP. Clinical and economic analysis of methicillin-susceptible and -resistant Staphylococcus aureus infections. Ann Pharmacother. 2004;38:1377-1382.
39. Leman P, Mukherjee D. Flucloxicillin alone or combined with benzylpenicillin to treat lower limb cellulitis: a randomized controlled trial. Emerg Med J. 2005;22:342-346.
40. Leppard BJ, Seal DV. The value of bacteriology and serology in the diagnosis of necrotizing fasciitis. Br J Dermatol. 1983;109:36-44.
41. Leppard BJ, Seal DV, Colman G, Hallas G. The value of bacteriology and serology in the diagnosis of cellulitis and erysipelas. Br J Dermatol. 1985;112:559-567.
42. Lipsky BA, Armstrong DG, Citron DM, Tice AD, Morgenstern DE, Abramson MA. Ertapenem versus piperacillin/tazobactam for diabetic foot infections (SIDESTEP): prospective, randomised, controlled, double-blinded, multicentre trial. Lancet. 2005;366:1695-1703.
43. Longcope WT. Studies of the variations in the anti-streptolysin titre of the blood serum from patients with hemorrhagic nephritis. II. Observations of patients suffering from streptococcal infections, rheumatic fever, and acute and chronic hemorrhagic nephritis. J Clin Invest. 1936;15:277.
44. Lutomski DM, Trott AT, Runyon JM, Miyagawa CI, Staneck JL, Rivera JO. Microbiology of adult cellulitis. J Fam Pract. 1988;26:45-58.
45. Michos AG, Bakoula CG, Braoudaki M, Koutouzi FI, Roma ES, Pangalis A, Nikolopoulou G, Kirikou E, Syriopoulou VP. Macrolide resistance in Streptococcus pyogenes: prevalence, resistance, determinants, and emm types. Diagn Microbiol Infect Dis. 2009;64:295-299.
46. Morales-Otero P, Pomales-Lebron A. The development of anti-streptolysins and anti-fibrinolysins following acute attacks of recurrent tropical lymphangitis. Trans R Soc Trop Med Hyg. 1936;20:191-197.
47. Moran GJ, Krishnadasan A, Gorwitz RJ, Fosheim GE, McDougal LK, Carey RB, Talan DA, for the EMERGEncy ID Net Study Group. Methicillin-resistant S. aureus infections among patients in the emergency department. N Engl J Med. 2006;355:666-674.
48. Mote JR, Jones TD. Studies of hemolytic streptococcal antibodies in control groups, rheumatic fever, and rheumatoid arthritis. I. The incidence of antistreptolysin "O," antifibrinolysin, and hemolytic streptococcal precipitating antibodies in the sera of urban control groups; II. The frequency of antistreptolysin "O," antifibrinolysin, and precipitating-antibody responses in scarlet fever, hemolytic streptococcal infections, and rheumatic fever; and III. The magnitude of antistreptolysin "O," antifibrinolysin, and precipitating-antibody responses; the persistence of antibodies, and variations in anti-streptolysin "O," curves in scarlet fever, hemolytic streptococcal infections, and rheumatic fever. J Immunol. 1941;41:35,61,87.
49. Peralta G, Padron E, Roiz MP, De Benito I, Garrido JC, Talledo F, Rodriguez-Lera MJ, Ansorena L, Sanchez MB. Risk factors for bacteremia in patients with limb cellulitis. Eur J Clin Microbiol Infect Dis. 2006;2:619-626.
50. Phares CR, Lynfield R, Farley MM, Mohle-Boetani J, Harrison LH, Petit S, Craig AS, Schaffner W, Zansky SM, Gershman K, Stefonek KR, Albanese BA, Zell ER, Schuchat A, Schrag SJ; Active Bacterial Core surveillance/Emerging Infections Program Network. Epidemiology of invasive group B streptococcal disease in the United States, 1999-2005. JAMA. 2008;299:2056-2065.
51. Phillips S, MacDougall C, Holdford DA. Analysis of empiric antimicrobial strategies for cellulitis in the era of methicillin-resistant Staphylococcus aureus. Ann Pharmacother. 2007;41:13-20.
52. Quest Diagnostics Incorporated. Quest Diagnostics Directory of Services 2010. Updated January 2010. Accessed January 4, 2010.
53. Quinn RW, Liao SJ. A comparative study of antihyaluronidase, antistreptolysin "O," antistreptokinase, and streptococcal agglutination titers in patients with rheumatic fever, acute hemolytic streptococcal infections, rheumatoid arthritis, and non-rheumatoid forms of arthritis. J Clin Invest. 1950;29:1156-1166.
54. Raney PM, Tenover FC, Carey RB, McGowan JE Jr, Patel JB. Investigation of inducible clindamycin and telithromycin resistance in isolates of β-hemolytic streptococci. Diagn Microbiol Infect Dis. 2006;55:213-218.
55. Rantz LA, Boisvert PJ, Spink WW. Antistreptococcal antibodies. Bull U. S. Army M. Dept. 1946;5:40.
56. Renneberg J, Soderstrom M, Prellner K, Forsgren A, Christensen P. Age-related variations in anti-streptococcal antibody levels. Eur J Clin Microbiol Infect Dis. 1989;8:792-795.
57. Schwartz MN. Cellulitis. N Engl J Med. 2004;350:904-912.
58. Schwartz RH, Shulman ST. Group C and group G streptococci: in-office isolation from children and adolescents with pharyngitis. Clin Pediatr. 1986;25:496-502.
59. Semel JD, Goldin H. Association of athlete's foot with cellulitis of the lower extremeties: diagnositic value of bacterial cultures of ipsilateral interdigital space samples. Clin Infect Dis. 1996;23:1162-1164.
60. Siljander T, Karppelin M, Vahakuopus S, Syrjanen J, Toropainen M, Kere J, Vuento R, Jussila T, Vuopio-Varkila J. Acute bacterial, nonnecrotizing cellulitis in Finland: microbiological findings. Clin Infect Dis. 2008;4:855-861.
61. Small PM, Chambers HF. Vancomycin for Staphylococcus aureus endocarditis in intravenous drug users. Antimicrob Agents Chemother. 1990;34:1227-1231.
62. Steel WA. Acute lymphangitis. Am J Surg. 1937;36:37-43.
63. Stollerman GH, Lewis AJ, Schultz I, Taranta A. Relationship of immune response to group A streptococci to the course of acute, chronic and recurrent rheumatic fever. Am J Med. 1956;20:163-169.
64. Stryjewski ME, Szczech LA, Benjamin DK Jr, Inrig JK, Kanafani ZA, Engemann JJ, Chu VH, Joyce MJ, Reller LB, Corey GR, Fowler VG Jr. Use of vancomycin or first-generation cephalosporins for the treatment of hemodialysis-dependent patients with methicillin-susceptible Staphylococcus aureus bacteremia. Clin Infect Dis. 2007;44:190-196.
65. Suarez J. A preliminary report on the clinical and bacteriological findings in 60 cases of lymphangitis associated with elephantoid fever in Puerto Rico. Am J Trop Med. 1930;10:189-190.
66. Tillotson GS, Draghi DC, Sahm DF, Tomfohrde KM, Del Fabro T, Critchley IA. Susceptibility of Staphylococcus aureus isolated from skin and wound infections in the United States 2005-2007: laboratory-based surveillance study. J Antimicrob Chemother. 2008;62:109-115.
67. Todd EW. Antihaemolysin titers in haemolytic streptococcal infections and their significance in rheumatic fever. J Exp Pathol. 1932;13:248.
68. Ulett KB, Benjamin WH Jr, Zhuo F, Xiao M, Kong F, Gilbert GL, Schembri MA, Ulett GC. Diversity of group B streptococcus serotypes causing urinary tract infection in adults. J Clin Microbiol. 2009;47:2055-2060.
69. Wannamaker LW, Ayoub EM. Antibody titers in acute rhematic fever. Circulation. 1960;21:598-614.
70. Wilson MG, Wheeler GW, Leask MM. The relation of upper respiratory infections to rheumatic fever in children. II. Antihemolysin titres in respiratory infections and their significance in rheumatic fever in children. J Clin Invest. 1935;14:333.
© 2010 Lippincott Williams & Wilkins, Inc.