Randomized Controlled Trial of Cephalexin Versus Clindamycin for Uncomplicated Pediatric Skin Infections

Abstract

OBJECTIVE:

To compare clindamycin and cephalexin for treatment of uncomplicated skin and soft tissue infections (SSTIs) caused predominantly by community-associated (CA) methicillin-resistant Staphylococcus aureus (MRSA). We hypothesized that clindamycin would be superior to cephalexin (an antibiotic without MRSA activity) for treatment of these infections.

PATIENTS AND METHODS:

Patients aged 6 months to 18 years with uncomplicated SSTIs not requiring hospitalization were enrolled September 2006 through May 2009. Eligible patients were randomly assigned to 7 days of cephalexin or clindamycin; primary and secondary outcomes were clinical improvement at 48 to 72 hours and resolution at 7 days. Cultures were obtained and tested for antimicrobial susceptibilities, pulsed-field gel electrophoresis type, and Panton-Valentine leukocidin status.

RESULTS:

Of 200 enrolled patients, 69% had MRSA cultured from wounds. Most MRSA were USA300 or subtypes, positive for Panton-Valentine leukocidin, and clindamycin susceptible, consistent with CA-MRSA. Spontaneous drainage occurred or a drainage procedure was performed in 97% of subjects. By 48 to 72 hours, 94% of subjects in the cephalexin arm and 97% in the clindamycin arm were improved (P = .50). By 7 days, all subjects were improved, with complete resolution in 97% in the cephalexin arm and 94% in the clindamycin arm (P = .33). Fevers and age less than 1 year, but not initial erythema > 5 cm, were associated with early treatment failures, regardless of antibiotic used.

CONCLUSIONS:

There is no significant difference between cephalexin and clindamycin for treatment of uncomplicated pediatric SSTIs caused predominantly by CA-MRSA. Close follow-up and fastidious wound care of appropriately drained, uncomplicated SSTIs are likely more important than initial antibiotic choice.

WHAT'S KNOWN ON THIS SUBJECT:

Community-associated methicillin-resistant Staphylococcus aureus (MRSA) causes the majority of purulent skin infections in children. The choice of adjuvant antibiotic varies widely in clinical practice, with a recent trend toward the empiric use of antibiotics with activity against MRSA.

WHAT THIS STUDY ADDS:

This randomized controlled trial of cephalexin versus clindamycin for uncomplicated, purulent skin infections showed equally high rates of improvement at 48 to 72 hours and 7 days, which suggests that using antibiotics with in vitro MRSA activity may not be critical for successful treatment.

Since the mid-1990s, an epidemic of community-associated methicillin-resistant Staphylococcus aureus (CA-MRSA) infections in healthy, nonhospitalized children has spread throughout the United States and worldwide.^1–3 The majority of these CA-MRSA infections are purulent skin and soft tissue infections (SSTIs), whereas more invasive infections such as sepsis, necrotizing pneumonia, osteomyelitis, septic arthritis, and pyomyositis occur less frequently.⁴

Although incision and drainage (ID) is considered essential for the management of abscesses,^5,6 the role of adjuvant antibiotics is unclear. Results of a small, blinded, placebo-controlled trial of adults undergoing ID for uncomplicated skin abscesses that was conducted the mid-1980s showed no advantage of adjuvant antibiotics compared with placebo.⁷ Since the emergence of CA-MRSA, findings from observational studies have been inconsistent in regard to the advantage of treating CA-MRSA SSTIs with antibiotics selected on the basis of their in vitro activity.^8,9 During the CA-MRSA era, results of randomized trials in which cephalexin was compared with placebo in adults¹⁰ and trimethoprim-sulfamethoxazole was compared with placebo in children¹¹ have failed to demonstrate a substantial difference in outcome. Since the emergence of CA-MRSA, however, no trial has included a comparison of traditional antistaphylococcal antibiotics with MRSA-active adjuvant antibiotics in the treatment of adults or children.

This trial was designed to enable us to compare the clinical outcomes of children with uncomplicated purulent SSTIs who were randomly assigned to receive treatment with cephalexin (a traditional antistaphylococcal antibiotic without activity against MRSA) or clindamycin (an antibiotic with high clinical activity against CA-MRSA in our area).

METHODS

This study was approved by the institutional review board of the Johns Hopkins Medical Institutions. Written informed consent with assent appropriate for age was obtained by either the research assistant or the treating attending physician from all subjects and/or parents/guardians. Enrollment occurred between September 2006 and May 2009. Eligible subjects included patients aged 6 months to 18 years (inclusive) who presented to a pediatric outpatient center at Johns Hopkins (pediatric emergency department [ED] or pediatric outpatient department) with an uncomplicated, purulent SSTI, defined as an abscess (with or without surrounding cellulitis), furuncle, or carbuncle for which outpatient management was anticipated. Exclusion criteria included: hospitalization on initial visit or previous 14 days; hypersensitivity to cephalosporin antibiotics or clindamycin; inherited or acquired altered immunity (such as HIV infection, uncontrolled diabetes mellitus, congenital immunodeficiency); skin infections related to surgical wounds or hardware; and current use of antibiotic therapy. Subjects were equally randomized to receive cephalexin (40 mg/kg per day, taken orally in divided doses administered 3 times per day) or clindamycin (20 mg/kg per day, taken orally in divided doses administered 3 times per day).^12–14 Patient randomization and assignment to study groups were performed in permuted blocks of 4 by the investigational drug research pharmacy at Johns Hopkins, who also concealed and maintained all assignments. Study drugs were dispensed in amber glass bottles if liquid formulation was used, and both subjects and caregivers were blinded to the study drug received.

The objective of this study was to compare the outcomes of children with uncomplicated SSTIs treated with either clindamycin or cephalexin. The primary outcome variable was clinical improvement at 48 to 72 hours from the initiation of treatment, defined as improvement in at least 1 of the measured parameters (overall improvement according to subject or parent/guardian, fever, erythema, pain/tenderness, and drainage) without worsening in any of those parameters. The secondary outcome variable was resolution of disease at 7 days, defined as overall improvement according to the subject or parent/guardian in addition to resolution of all variables (fever, erythema, pain/tenderness, and drainage). Our hypothesis was that clindamycin would be superior to cephalexin for both the primary and secondary outcome variables.

The sample size calculations were performed on the basis of the main outcome variable of comparison of the proportion of subjects with clinical improvement at 48 to 72 hours in the 2 study groups. On the basis of the results of previous studies,⁸ we expected clinical improvement at 48 to 72 hours to occur in ∼95% of subjects treated with cephalexin. Using sample size calculations for the comparison of the 2 proportions, we estimated a sample size of 178 (88 per group) needed to detect a 15% difference between improvement in the clindamycin and cephalexin groups with a power of 80% and a 2-sided significance level of 5%. We increased the planned sample size by ∼10% to 100 per group to account for subjects lost to follow-up.

Clinical data for children who met inclusion criteria were collected during the initial encounter, and then during repeat visits or by telephone follow-up at 2 to 3 days, 1 week, and 3 months. For the 3-month follow-up, subjects or parents/guardians were interviewed by telephone, and hospital medical records were reviewed for all subjects to capture data regarding any visits for repeat infections. A standardized data abstraction form was used for all visits, and data were recorded by the research assistant, the attending physician treating the patient, or the study coordinator. Conditions for which historical information was obtained at the initial visit included: previous skin infections; asthma; eczema; antibiotics used in the preceding 6 months; and chronic medical problems. For the acute infection, clinical information obtained included: fever history; location of abscess; measured erythema; measured induration; presence of warmth, tenderness, fluctuance, and spontaneous drainage; number of abscesses; procedure performed; and amount of purulent debris obtained. Procedures were categorized as ID with packing, ID without packing, expressed pus without ID, or no purulent drainage. The longest diameter of both erythema and induration of the infection site were measured with a measuring tape.

Subjects were instructed to return to either the pediatric ED or the outpatient center for reevaluation 48 to 72 hours after the initial encounter; however, if subjects did not return the follow-up information was obtained by telephone contact. Information obtained at the second visit included whether antibiotics were obtained and taken as prescribed and whether there was a need for change in antibiotics or for additional procedures. A 5-point Likert scale (1, much improved; 2, somewhat improved; 3, unchanged; 4, somewhat worsened; 5, much worsened) was used for the following assessments: overall subjective improvement (according to subject or parent/guardian); fever; erythema (actual repeat measurements were performed if patients were seen in the ED or clinic); pain/tenderness; and drainage.

For the 1-week follow-up visit, the subjects/family were contacted by telephone to determine adherence to medications, occurrence of adverse events, and overall subjective improvement. The day of resolution was recorded for the following variables: fever; erythema; pain/tenderness; and drainage.

For the 3-month follow-up, subjects/family were contacted by telephone to obtain a history of interval complications from the initial infection or medications, and any interval SSTIs in subjects. For all subjects the electronic medical record from our institution was also reviewed to extract available data from intervening visits and laboratory results related to SSTIs.

Wound specimens were obtained for culture by using the BactiSwab II culturette with modified Stuart's medium (Remel, Lenexa, KS) according to standard ED culturing practices. Specimens were then plated onto BBL MacConkey II sheeps's blood, chocolate, and CNA (colimycin nalidixic) agar; after overnight incubation at 37°C, all cultures were read at 24 and 48 hours. Any potential pathogen was investigated according to standard laboratory protocol, which included identification and susceptibility testing using the BD Phoenix automated microbiology system (BD Diagnostics, Sparks, MD).¹⁵ Isolates were tested for susceptibility to a standard array of antibiotic agents, including cotrimoxazole, clindamycin, erythromycin, tetracycline, moxifloxacin, and vancomycin. The disk-diffusion induction test (D test)¹⁶ was used to detect inducible clindamycin resistance in erythromycin-resistant, clindamycin-susceptible staphylococcal isolates.

All S aureus isolates were additional subjected to pulsed-field gel electrophoresis performed with standard extraction methods¹⁷ and evaluated for strain relatedness as described by Tenover et al.¹⁸ The first 20 isolates and all subsequent isolates that were not identical to the USA300 strain were tested by polymerase chain reaction for the presence of the genes encoding LukS-PV and LukF-PV, 2 proteins that comprise the Panton-Valentine leukocidin toxin.¹⁹

We used χ² or Fisher's exact test to compare proportions of outcomes or characteristics according to study drug assignment. Stata 11 (Stata Corp, College Station, Texas) was used for all analyses, which were performed on an intention-to-treat basis. A P-value < .05 was considered statistically significant.

RESULTS

Enrollment and Demographic Characteristics

Of 220 subjects screened, 200 were enrolled in the study. One hundred subjects were randomly assigned to receive cephalexin and 100 subjects were randomly assigned to receive clindamycin. Three subjects in each arm were lost to follow-up by the first visit, and 1 additional subject in the cephalexin arm and 2 subjects in the clindamycin arm were lost to follow-up at the 7-day visit. In the cephalexin arm 34 of 97 subjects (35%) and in the clindamycin arm 37 of 97 subjects (38%) were followed up at 48 to 72 hours by telephone instead of in the ED or outpatient clinic (P = .67). Twenty-eight subjects in the cephalexin arm and 26 subjects in the clindamycin arm could not be reached for the 3-month follow-up. There were no statistically significant differences between subjects assigned to the cephalexin and clindamycin arms for any of the demographic variables (Table 1).

TABLE 1.

Demographic Characteristics of Study Subjects According to Study Drug Assignment

	Overall	Cephalexin Arm	Clindamycin Arm	P
Total no. of subjects	200	100	100	NA
Prior skin infections	69	36	33	.61
Location of enrollment				.55
ED	188	95	93
Clinic	12	5	7
Black	188	92	96	.23
Age				.82
6–12 mo	18	9	9
1–5 y	75	35	40
6–12 y	45	22	23
13–18 y	62	34	28
Female gender	115	55	60	.47
History of asthma	58	25	33	.21
History of eczema	68	33	35	.77
Antibiotic use ≤6 mo	49	28	21	.25
Underlying CMC	19	10	9	.81
Hospitalization in previous year	14	4	10	.10

NA indicates not applicable; CMC, chronic medical condition.

Clinical Characteristics at Baseline

There were no significant differences in any characteristics of infection at baseline between subjects in the cephalexin and clindamycin treatment arms (Table 2).

TABLE 2.

Clinical Characteristics of Study Subjects According to Study Drug Assignment

	Overall	Cephalexin Arm	Clindamycin Arm	P
Total number of subjects	200	100	100	NA
Fever present	48	22	26	.48
Infection characteristics:
Tenderness present	198	100	98	.16
Fluctuance present	66	35	31	.55
Spontaneous drainage	86	45	41	.57
Location: below waist	119	60	59	.24
Erythema ≤5 cma	144	68	76	.63
Induration ≤5 cmb	169	83	86	.41
Drainage procedure				.72
No procedure	5	2	3
EP without ID	66	33	33
ID	127	65	62
Needle aspiration	2	0	2
Wound culture isolate				.56
MRSA	137	65	71
MSSA	37	19	19
None/other	26	16	10

NA indicates not applicable; EP, expressed pus.

^aErythema: range, 0–20 cm, median, 4 cm, mean, 4.4 cm.

^bInduration: range, 0–16 cm, median, 3.5 cm, mean, 3.8 cm.

Wound Isolates

Of 200 specimens, 137 (69%) grew MRSA, 37 (19%) grew methicillin-susceptible S aureus (MSSA), 16 (8%) grew other organisms (Proteus spp, coagulase-negative staphylococci, streptococci [groups A, B and C], Actinomyces spp), and 10 (5%) were sterile or unable to be collected. According to results of pulsed-field gel electrophoresis, most MRSA (93%) and many MSSA (35%) isolates were identical to USA 300 or related subtypes, and Panton-Valentine leukocidin was detected at high rates in MRSA (99%) and MSSA (82%). Wound isolates demonstrated high rates of susceptibility to clindamycin (91%), tetracyclines (93%), cotrimoxazole (98%), and vancomycin (100%), and lower rates of susceptibility to erythromycin (19%) and moxifloxacin (25%). Three MRSA wound isolates and 1 MSSA wound isolate harbored inducible clindamycin resistance (D-test positive). The proportions of subjects with wound infections caused by MRSA, MSSA, and other pathogens were similar for the cephalexin and clindamycin treatment arms (Table 2).

Outcome at 48 to 72 Hours

Drug assignments of study subjects were unblinded only after interventions and outcomes were assigned at the end of the study. Three subjects in each study arm were lost to follow-up for this primary outcome variable. Of the remaining subjects, 91 of 97 subjects (94%) in the cephalexin arm and 94 of 97 subjects (97%) in the clindamycin arm showed improvement or resolution in their infection (P = .50). The primary infection had worsened in 6 of 97 subjects (6%) in the cephalexin arm and 3 of 97 subjects (3%) in the clindamycin arm (Table 3). For the subset of subjects whose initial wound cultures grew MRSA and were not lost to follow-up (135), 9% (6 of 64) in the cephalexin arm and 3% (2 of 71) in the clindamycin arm had worsened by the 48-to-72–hour visit (P = .15). For the subset of subjects for whom an organism was isolated from the initial wound culture and susceptibility data were known (183), 2% (2 of 111) of those who received an antibiotic with in vitro activity against the isolate versus 10% (7 of 72) of those who received an inactive antibiotic had worsened by the 48-to-72–hour visit (P = .02). There was no association between gender, size of erythema, size of induration, location of infection, or type of initial drainage procedure on primary outcome; however, there was a significantly lower rate of improvement at 48 to 72 hours among subjects aged <1 year and those with fever (Table 4), results that were unrelated to the study drug given.

TABLE 3.

Characteristics of Study Patients Whose Condition Was Worsened at the 48-to-72–Hour Follow-up Visit

Patient Study No.	Age, y	Gender	Study Drug	Wound Culture	Chronic Condition	Fever at Initial Visit	Location	Erythema, cm	Induration, cm	Spontaneous Drainage	Procedure	Improvement at 48–72 h					Subsequent Procedure	Change in Antibiotics?	Hospitalized?	Secondary Outcome: Status at 7 d
												Overalla	Fever	Erythema	Pain/tenderness	Drainage
89	0.7	Male	Clindamycin	MRSA	No	101.7	Thigh	8	5	Yes	ID-NP	5	3	4	4	4	Repeat ID	Yes	Yes	Resolved
128	0.9	Male	Cephalexin	MRSA	No	102.7	Buttocks	4	7	Yes	ID-NP	1	3	4	2	4	Wound repacked	Yes	Yes	Resolved
170	0.9	Female	Clindamycin	MRSAb	No	101.3	Thigh	3	4	Yes	EP no ID	5	2	2	4	1	None	Yes	No	Improved
43	0.9	Female	Cephalexin	MRSA	No	No	Buttocks	6	5	No	ID-P	2	0	4	2	2	None	Yes	No	Resolved
25	1	Male	Cephalexin	MRSA	No	101.3	Buttocks	2	3.5	No	ID-NP	5	5	4	4	4	Repeat ID; sedation	Yes	No	Resolved
197	1	Female	Cephalexin	MRSA	No	101.5	Groin	6	5	No	ID-NP	4	3	5	3	3	Repeat ID attempted	Yes	No	Resolved
71	8	Female	Clindamycin	MSSA	No	No	Thigh	7	5	No	ID-P	1	0	4	1	1	Repeat ID	Yes	No	Resolved
45	9	Female	Cephalexin	MRSA	No	No	Leg	2	2	No	ID-NP	4	0	2	4	2	None	No	No	Resolved
77	11	Male	Cephalexin	MRSA	No	No	Abdomen	5	3	Yes	ID-P	2	0	4	1	2	None	No	No	Resolved
Summary, median or %	1	44% Male	33% Clinda	89% MRSA	100% No chronic condition	56% Fever	100% Waist and below	5 cm	5 cm	44% Spontaneous drainage	100% Drainage procedure						56% Repeat drainage procedure	78% Antibiotics changed	22% Hospitalized	100% Improved or resolved

ID-NP indicates ID without packing; ID-P, ID with packing; EP, expressed pus.

^aOverall improvement, fever, erythema, pain/tenderness, and drainage were assessed with a 5-point Likert scale: 0, never had symptoms; 1, much improved; 2, somewhat improved; 3, unchanged; 4, somewhat worse; 5, much worse.

^bOf the 9 patients whose condition had worsened at the 48-to-72–hour follow-up, only patient 170 had a wound-culture organism that was resistant to clindamycin (it was also D-test positive).

TABLE 4.

Primary and Secondary Outcomes According to Erythema, Induration, Fever, Type of Procedure, Location of Infection, Gender, and Age

	Primary Outcome: Improved at 48–72 h (n = 194), % (n/N)a	P	Secondary Outcome: Resolved at 7 d (n = 191), % (n/N)a	P
Erythema		.27		.27
<5 cm	96% (134/139)		96% (132/137)
>5 cm	93% (51/55)		93% (50/54)
Induration		.71		.58
<5 cm	95% (156/164)		96% (154/161)
>5 cm	97% (29/30)		93% (28/30)
Fever present		.03b		.03b
Yes	89% (42/47)		89% (42/47)
No	97% (143/147)		97% (140/144)
Procedure		.43		1.00
ID	94% (117/125)		95% (117/123)
EP, no ID	98% (63/64)		95% (60/63)
None	100% (5/5)		100% (5/5)
Location of infection		.09		.31
Above waist	99% (78/79)		97% (77/79)
Below waist	93% (107/115)		94% (105/112)
Gender		1.00		1.00
Male	96% (79/83)		95% (77/81)
Female	95% (106/111)		95% (105/110)
Age		.004b		.17
6–12 mo	76% (13/17)		88% (14/16)
>1 year	97% (172/177)		96% (168/175)

EP, expressed pus.

^aThe total N for both the primary and secondary outcome variable is <200 owing to subjects lost to follow-up.

^bStatistically significant.

Hospitalizations

Four subjects were hospitalized, all within the first week after enrollment; 2 of these hospitalizations were unrelated to the SSTI (1 was for acute gastroenteritis and 1 for injuries from a motor vehicle crash). The other 2 subjects, 1 in each study arm, were hospitalized because of worsening of their initial infection (P = .73, Table 3).

Outcome at 7 Days

For the secondary outcome variable of resolution of infection at 7 days, 4 subjects in the cephalexin arm and 5 subjects in the clindamycin arm were lost to follow-up. Of the remaining subjects, 93 of 96 subjects (97%) in the cephalexin arm and 89 of 95 subjects (94%) in the clindamycin arm had clinical resolution by 7 days (P = .33). Only 1 subject developed a new SSTI while on therapy; this subject was on study-assigned clindamycin and had an infection caused by MRSA susceptible to clindamycin. Compliance with taking medications as directed was reported by subjects or parents/guardians for 84 of 96 subjects (88%) in the cephalexin arm and 81 of 95 (85%) in the clindamycin arm (P = .66).

For the subset of subjects whose initial wound cultures grew only MRSA and who were not lost to follow-up (133), 100% (63 of 63) in the cephalexin arm and 94% (66 of 70) in the clindamycin arm had clinical resolution of disease by 7 days (P = .12). For the subset of subjects for whom an organism was isolated from the initial wound culture and susceptibility data were known (180), 95% (104 of 109) of those who received an active antibiotic versus 99% (70 of 71) of those who received an inactive antibiotic had clinical resolution of disease by 7 days (P = .41). Only baseline fever was associated with a significantly lower rate of resolution at 7 days (Table 4).

Three-Month Follow-up

According to data obtained from telephone contact (73%) and chart review (100%) at the 3-month follow-up, 36 subjects (18%) had a recurrent SSTI. The risk of new SSTI did not differ according to isolation of MRSA versus MSSA from initial wound culture (21% MRSA versus 16% MSSA; P = .51) or by cephalexin or clindamycin assignment (20% vs 16%; P = .46).

Adverse Events

There were no serious adverse events related to study treatment. A 13-month-old child developed mild diarrhea with stool positive for Clostridium difficile antigen and toxin 1 week after completing study-assigned treatment with clindamycin. The diarrhea resolved without treatment and repeat stool specimens were negative for C. difficile.

DISCUSSION

Although there is a limited body of evidence to indicate that antibiotics may not be helpful in the management of purulent SSTIs, especially those that have been treated with ID, many clinicians have expressed discomfort with withholding antibiotics for these patients. Results of a recent survey by the New England Journal of Medicine²⁰ revealed significant variability in clinical practice in the treatment of patients with uncomplicated SSTIs. When surveyed on their management of an uncomplicated, purulent SSTI in a college athlete, nearly 31% responded that they would perform ID alone, 28% responded that they would advocate use of ID plus an oral antibiotic with activity against MSSA, and 41% responded that they would advocate use of ID plus an oral antibiotic with activity against MRSA.

Recently, Duong et al reported results of a randomized trial of cotrimoxazole versus placebo for children with uncomplicated SSTIs, which demonstrated no difference in treatment failures between the 2 groups but higher rates of recurrent SSTIs within the first 10 days (but not 3 months) in the placebo group.¹¹ At the time of our study design, we did not feel that there was sufficient evidence to justify a true placebo arm, and therefore for this study we chose to compare an antibiotic (clindamycin) with high in vitro activity against CA-MRSA, the predominant organism causing SSTIs in our area, with a traditional antistaphylococcal antibiotic (cephalexin) that has been associated with good clinical outcomes for CA-MRSA SSTIs in observational studies^3,8,21 despite in vitro resistance. Since mid-2004, it has been standard practice at our institution to use clindamycin for patients who present with uncomplicated SSTIs. This change in practice, which was instituted on the basis of reports of safety and efficacy in children with MRSA infections²² and the perceived importance of antibiotics in the management of SSTIs, has resulted in the significantly increased use of clindamycin despite its numerous disadvantages over traditional antistaphylococcal antibiotics, including increased cost, poor palatability of oral formulations, greater concern for adverse effects (such as C difficile infection), and development of antibiotic resistance.

In this double-blinded, randomized trial of children with uncomplicated purulent SSTIs caused predominantly by CA-MRSA we found no difference in clinical outcomes at 2 to 3 days or at 7 days between those children who received cephalexin and those who received clindamycin. This finding was consistent for the overall study and for only those infections caused by MRSA. Nearly all children had active or passive drainage of purulent collections before treatment with antibiotics. Adverse events, including hospitalization for treatment failure, were uncommon and equal in the 2 study arms. A small percentage of subjects with worsening infection at 48 to 72 hours (<5% of all subjects) were managed with repeat drainage, with or without a change in antibiotics, and all had good outcomes at 7 days. Overall, 95% of SSTIs had resolved within 1 week of entry.

One potentially important finding was a slight but statistically significant better outcome at 48 to 72 hours for subjects treated with an antibiotic with in vitro activity. An observational study by Ruhe et al⁹ also showed a small increase in the incidence of early treatment failures for antibiotics without in vitro activity compared with antibiotics with in vitro activity. Such results must be interpreted with caution, however, because subjects in our study were randomly assigned to clindamycin versus cephalexin and not to active versus inactive antibiotic, and these results may have been influenced by confounding differences between groups. In practice, there is no way to reliably predict a priori the infecting organism and its susceptibilities. While active development of much-needed technology for rapid organism identification and susceptibility testing is underway, the clinical scenario facing most practitioners today would in fact be that of choosing an empiric antibiotic on the basis of the local prevalence of MRSA and local susceptibility patterns. The results of this study should reassure clinicians that despite this small difference in early treatment failures according to in vitro antibiotic activity, adverse events were quite uncommon and outcomes at 7 days were uniformly good.

Contrary to a previously reported study by Lee et al,⁸ in our study we did not find a significant difference in adverse outcomes associated with the initial size of lesion (≤5 cm versus >5 cm). This lack of association of outcome with lesion size was observed even for subjects who had significant surrounding areas of cellulitis (erythema greater than induration). There did seem to be a statistically significant difference in early treatment failures for younger subjects (children aged less than 1 year) compared with older subjects, as well as those presenting with fever. These data suggest that younger age and presence of fever, rather than initial size of lesion, may be more important for the prediction of adverse outcomes and may be more important for predicting the subsequent need for hospitalization.

There were several limitations to this study. Because of the relatively small sample size, we likely were unable to detect very small differences in outcomes, but the extremely high rate of favorable outcomes in both study arms (94% and 97%) was a reassuring indication that there was no clinically important difference. A second limitation was that not all infections were attributable to MRSA and not all MRSA were susceptible in vitro to clindamycin, which prevented a true random comparison of antibiotics with in vitro activity to those without. Third, the rates of recurrence by 3 months may have been underestimated because they were determined with data obtained from reports by patients or parents/guardians and abstraction of medical records only from our institution. Fourth, there was no formal method of assessing the success of blinding. Difficulty with the administration of clindamycin because of its unpalatability was considered as a potential contributor to poorer compliance (and thus effectiveness) of clindamycin, but there was not a significant difference in reported rates of compliance with medications. Another limitation was that this study lacked a true placebo arm. Some experts have hypothesized that antibiotics without in vitro activity against infecting organisms may exert some clinical benefit through effects such as immunomodulation, activity against coexisting organisms, or overcoming resistance in vivo by other mechanisms,¹⁰ but evidence of these effects in the treatment of CA-MRSA is limited. The high rate of improvement in the placebo arm of the trial by Duong et al¹¹was similar to the rate in this study, and highlights the uncertain benefit of adjunctive antibiotics and the appropriateness of placebo arms in future studies.

In this randomized controlled trial of antibiotic treatment for children with uncomplicated SSTIs, the majority of our subjects had spontaneous drainage or a drainage procedure, the cornerstone of management of purulent infections. Contrary to our initial hypothesis, we did not find any significant difference in outcomes at 48 to 72 hours or 7 days that was attributable to the antibiotic used for initial treatment (cephalexin versus clindamycin), although there was a small but statistically significant difference in early treatment failures that was associated with non–randomly assigned active versus inactive antibiotics. Until additional studies confirm that adjuvant antibiotics offer no benefit in the management of children with uncomplicated, purulent SSTIs, cephalexin remains a viable empiric antibiotic choice (even in areas with a high prevalence of CA-MRSA) in the context of management that already includes careful drainage of purulent collections, attention to wound care, and appropriate follow-up, especially in children of younger age and with fever. Additional studies are needed to define predictors and effective interventions for the prevention of recurrent infections.