You are on page 1of 11

The n e w e ng l a n d j o u r na l of m e dic i n e

Original Article

A Placebo-Controlled Trial of Antibiotics
for Smaller Skin Abscesses
Robert S. Daum, M.D., C.M., Loren G. Miller, M.D., M.P.H., Lilly Immergluck, M.D.,
Stephanie Fritz, M.D., M.S.C.I., C. Buddy Creech, M.D., M.P.H.,
David Young, M.D., Neha Kumar, M.D., Michele Downing, R.N., M.S.N.,
Stephanie Pettibone, B.S., Rebecca Hoagland, M.S., Samantha J. Eells, M.P.H.,
Mary G. Boyle, R.N., M.S.N., Trisha Chan Parker, M.P.H.,
and Henry F. Chambers, M.D., for the DMID 07-0051 Team*​​


Uncomplicated skin abscesses are common, yet the appropriate management of the From the University of Chicago Hospi-
condition in the era of community-associated methicillin-resistant Staphylococcus aureus tals, Chicago (R.S.D., N.K.); Harbor–
UCLA Medical Center and Los Angeles
(MRSA) is unclear. BioMedical Research Institute at Harbor–
UCLA Medical Center, Los Angeles (L.G.M.,
METHODS S.J.E.), and University of California, San
We conducted a multicenter, prospective, double-blind trial involving outpatient adults Francisco–San Francisco General Hospital,
and children. Patients were stratified according to the presence of a surgically drainable San Francisco (D.Y., M.D., H.F.C.); More-
house School of Medicine and Emory
abscess, abscess size, the number of sites of skin infection, and the presence of non- University–Grady Memorial Hospital and
purulent cellulitis. Participants with a skin abscess 5 cm or smaller in diameter were Children’s Healthcare of Atlanta, Atlanta
enrolled. After abscess incision and drainage, participants were randomly assigned to (L.I., T.C.P.); Washington University School
of Medicine–Barnes-Jewish Hospital and
receive clindamycin, trimethoprim–sulfamethoxazole (TMP-SMX), or placebo for 10 days. St. Louis Children’s Hospital, St. Louis
The primary outcome was clinical cure 7 to 10 days after the end of treatment. (S.F., M.G.B.); Vanderbilt University School
of Medicine and Vanderbilt University
RESULTS Medical Center, Nashville (C.B.C.); EMMES
We enrolled 786 participants: 505 (64.2%) were adults and 281 (35.8%) were children. A Corporation, Rockville, MD (S.P.); and
total of 448 (57.0%) of the participants were male. S. aureus was isolated from 527 par- Cota Enterprises, Meriden, KS (R.H.).
Address reprint requests to Dr. Daum
ticipants (67.0%), and MRSA was isolated from 388 (49.4%). Ten days after therapy in at the Dept. of Medicine, University of
the intention-to-treat population, the cure rate among participants in the clindamycin Maryland School of Medicine, 655 W.
group was similar to that in the TMP-SMX group (221 of 266 participants [83.1%] and Baltimore St., Baltimore, MD 21201, or at
215 of 263 participants [81.7%], respectively; P = 0.73), and the cure rate in each active-
treatment group was higher than that in the placebo group (177 of 257 participants * A list of additional investigators in the
Division of Microbiology and Infectious
[68.9%], P<0.001 for both comparisons). The results in the population of patients who Diseases (DMID) 07-0051 Team is pro-
could be evaluated were similar. This beneficial effect was restricted to participants with vided in the Supplementary Appendix,
S. aureus infection. Among the participants who were initially cured, new infections at available at

1 month of follow-up were less common in the clindamycin group (15 of 221, 6.8%) N Engl J Med 2017;376:2545-55.
than in the TMP-SMX group (29 of 215 [13.5%], P = 0.03) or the placebo group (22 of DOI: 10.1056/NEJMoa1607033
Copyright © 2017 Massachusetts Medical Society.
177 [12.4%], P = 0.06). Adverse events were more frequent with clindamycin (58 of 265
[21.9%]) than with TMP-SMX (29 of 261 [11.1%]) or placebo (32 of 255 [12.5%]); all
adverse events resolved without sequelae. One participant who received TMP-SMX had a
hypersensitivity reaction.
As compared with incision and drainage alone, clindamycin or TMP-SMX in conjunction with
incision and drainage improves short-term outcomes in patients who have a simple abscess.
This benefit must be weighed against the known side-effect profile of these antimicrobials.
(Funded by the National Institutes of Health; number, NCT00730028.)

n engl j med 376;26  June 29, 2017 2545
The New England Journal of Medicine
Downloaded from on June 28, 2017. For personal use only. No other uses without permission.
Copyright © 2017 Massachusetts Medical Society. All rights reserved.

.g. perirec- tal. vided written informed consent and assent. evaluated manually by measuring the abscess ever. drainable collection of pus) with a great- antibiotic therapy. and the presence of surgical site or prosthetic device infection. Torrance. 2017 The New England Journal of Medicine Downloaded from nejm. local  June 29.1 Abscesses are the emergency departments. or hand infection). such as bacteremia. Participants with a single systemic antistaphylococcal antibacterial therapy skin abscess 5 cm in diameter or smaller were in the previous 14 days.. Harbor–University of California. impetigo). All rights reserved. Medical Center. evidenced by (>5 cm) or multiple skin abscesses and cellulitis two or more of the following signs or symptoms (either purulent or nonpurulent). cur in rare cases. human or animal bite. Copyright © 2017 Massachusetts Medical Society. dex (the weight in kilograms divided by the cally drainable abscess. Serious complications. and Morehouse School of Medicine– has not been defined. placebo-controlled trial involving procedure was implemented by the treating physi- children and adults with a drainage and TMP-SMX had a slightly higher Participants were eligible for participation if cure rate than those treated with incision and they had a single abscess (defined as a circum- drainage and placebo. Trial Design and Population presence of systemic inflammatory response We conducted a multicenter. or placebo in addition to incision and ization. Emory University. Washington University. the trial did not include the evaluation of a cavity length in three dimensions (width. infection at a body site and drainage. TMP-SMX. cian. Abscess size was similar to those associated with TMP-SMX. or an immunocompromising condition (e. in the United States. ran. had cancer or an inflammatory disorder treated 2546 n engl j med 376. double-blind. when tient treatment of abscesses because of their low age-appropriate. All Clindamycin and trimethoprim–sulfamethox. or placebo after incision fections (e. Participants were ineli- randomly assigned to receive oral clindamycin. a body-mass in- stratified according to the presence of a surgi. with packing of the abscess as needed.0°C for children 6 to 11 months of age). small abscess. genital. Nashville (added causes most skin infections. immunosuppressive therapy domized. ticipants 6 to 11 months of age and ≤4 cm for volving outpatient adults and children with large participants 1 to 8 years of age). or nonpurulent cellulitis. the number square of the height in meters) higher than 40. The protocol was approved by cost and in vitro activity against community-asso. and affiliated clinics at most common of these infections. partici- treatment for skin infections annually pants were recruited from urgent care clinics. aureus (MRSA) strains. purulent drainage. bilt University Medical Center. oral temperature higher than 38.3. in 2011). and ten- The cure rates associated with clindamycin were derness to pain or palpation. placebo group. syndrome criteria.1. San Francisco General Hospital. Vander- methicillin-resistant S. had drainage. Atlanta (added in 2012).4 but the appropri. placebo-controlled clini. The n e w e ng l a n d j o u r na l of m e dic i n e M ore than 4 in 100 people seek From May 2009 through January 2015.1 ter. Participants were betes or chronic renal failure). as appropriate. dia- cal trial involving on June 28. which supports a role for scribed.g. and the ma. For personal use only. including Los Angeles. Here we report the results of a and depth). Louis (add- ate strategy for the treatment of these infections ed in 2012).5°C (or Me thods >38. St.7 Participants for at least 24 hours: erythema. abscess size.8 Participants were randomly assigned to receive Exclusion criteria included superficial skin in- clindamycin. ciated MRSA and methicillin-susceptible strains.g.6 We previously evaluated clinda­ est diameter of 5. oc. lived in a long-term care facility. gible for participation if they required hospital- TMP-SMX. Chicago. length. No other uses without permission.2 Staphylococcus aureus. How. San Francisco.0 cm or less (≤3 cm for par- mycin and TMP-SMX in a randomized trial in. A standardized incision-and-drainage double-blind. requiring specialized management (e.26 nejm. of sites of skin infection. pleteness of the data and the fidelity of the trial One randomized trial showed that outpatients to the protocol..5 The authors vouch for the accuracy and com- but data on their safety and efficacy are limited. . 2017. swelling or indu- underwent incision and drainage. available with the full text of with skin abscesses treated with incision and this article at NEJM. six sites: the University of Chicago Medical Cen- jority of patients are treated as outpatients. the institutional review board at each institution. prospective. ration. participants or their parents or guardians pro- azole (TMP-SMX) are recommended for outpa.

an analysis of variance The statistical analy- of the microbiologic results. Table S3 in the Supplementary power to detect a 10-percentage-point absolute Participants and all study staff were unaware difference in cure rates (e. Trial of Antibiotics for Smaller Skin Abscesses in the previous 12 months. or hos- The suspensions did not differ in appearance or pitalization related to the infection. analyses were performed: one in the intention- ter (EMMES Corporation). and placebo cap. clindamycin. The and drug accountability for those participants trial was designed as a superiority trial with 80% who returned blister packs or suspension bottles. the National 786 participants were required (262 per group). TMP-SMX. Compari- ing in accordance with Clinical and Laboratory sons among the groups were performed with the Standards Institute–approved methods9 by the use of Pearson’s chi-square test. Two primary efficacy independent statistics and data-coordinating cen. and at the 1-month follow-up (day 40). 2017. . Informa- able at NEJM. to-treat population (all participants who under- Clindamycin was given as two 150-mg tablets went randomization) and the other in the popu- three times daily. with an alpha of 0. or any one of the following: recurrence doses were adjusted according to weight (Table at the original site of infection or occurrence of S4 in the Supplementary Appendix). tee with the use of an O’Brien–Fleming monitor- n engl j med 376. were the cure rates at the end-of-treatment and 1-month follow-up visits. Participants who were randomly assigned to re. tion about clinical response and possible side effects of treatment or placebo were obtained Randomization and Study Agents with the use of standardized forms. Fisher’s exact clinical microbiology laboratory at each participat. Vari. and liquid a skin infection at a new body site.05. as appropriate. Pediatric 48 hours. among the three study groups in the population tion of the research pharmacists. end-of-treatment and test-of-cure visits) (Fig. who determined that could be evaluated. taste and were provided in identical medicine The primary null hypothesis was that clindamy- bottles.. the cure rates in adults Microbiologic and Demographic Data and children. No other uses without permission. 1). and susceptibility test. and placebo would have equiva- 10 days. or a logistic ing institution. test. Clindamycin. at the test-of-cure visit (7 to 10 days of inclusion and exclusion criteria is provided in after the prescribed 10-day course of therapy). the correct dosing. tion. All rights reserved. The study medication was Under the assumption of a 20% attrition rate.g. Copyright © 2017 Massachusetts Medical Society. Adherence was assessed by self-report lent rates of cure at the test-of-cure visit. 85% vs. surgical treatment of the skin infection. and adverse-event rates. For personal use only. an inability to continue taking the study pants. The investigators were unaware regression. with the excep. or TMP-SMX. Pills were over-encapsulated to of resolution of signs or symptoms of the infec- prevent identification by study staff and partici. After incision and drainage of the abscess and determination of the size of the abscess. Institute of Allergy and Infectious Diseases of The prespecified exploratory secondary outcomes the National Institutes of Health. 2017 2547 The New England Journal of Medicine Downloaded from nejm. other strains. stratified ac- able-block randomization was performed by an cording to study group. partici.26 nejm. agent because of adverse effects within the first sules were identical in on June 28. TMP-SMX was given as two lation that could be evaluated (participants who tablets (each containing 80 mg of trimethoprim received treatment or placebo and completed the and 400 mg of sulfamethoxazole) twice daily. MRSA. A lack of clinical cure was assessed by the re- ceive TMP-SMX were given a placebo pill for the search nurse at each site and was defined as lack midday dose. independent data and safety monitoring commit- ment failure. unplanned suspensions were available for pediatric dosing. although the results sis plan is available at NEJM. TMP-SMX. or had had major Participants were seen at the end of treatment surgery in the previous 12 months. species with methicillin-susceptible S. 95%) of the study-group assignments. The study agents were administered for  June 29. avail. the cure rates for patients infected Abscess fluid was submitted for culture. the time of the test-of-cure visit. or identification of isolates. aureus. The full list (day 12). Statistical Analysis pants were randomly assigned in a 1:1:1 ratio to The primary trial outcome was clinical cure by receive placebo. purchased by the study sponsor. Interim analyses could be obtained on request by an independent for safety and efficacy were performed by an data and safety monitor in the case of a treat.

org  June 29. All rights reserved. 2017. 1 Was nonadherent 2 Underwent randomiza- tion and treatment twice 2 Underwent on June 28. study-agent adminis- tration tration tration 3 Had other illness 7 Had other illness 12 Had other illness or injury or injury or injury 3 Withdrew 1 Withdrew 4 Withdrew 3 Were withdrawn 4 Were withdrawn 3 Were withdrawn by investigator by investigator by investigator 1 Missed a dose 6 Were lost to follow up 12 Were lost to follow up 5 Were lost to follow up 3 Had other reasons 6 Had other reasons 1 Had other reasons 238 Were included in population that 232 Were included in population that 220 Were included in population that could be evaluated and underwent could be evaluated and underwent could be evaluated and underwent primary efficacy analysis at TOC primary efficacy analysis at TOC primary efficacy analysis at TOC 28 Were excluded from 31 Were excluded from 37 Were excluded from primary efficacy primary efficacy primary efficacy analysis at TOC analysis at TOC analysis at TOC 9 Missed a visit 16 Missed a visit 18 Missed a visit 2 Received nonstudy 2 Received nonstudy 19 Were terminated from antibiotics antibiotics study before evaluation 17 Were terminated from 13 Were terminated from 10 Were lost to follow-up study before evaluation study before evaluation 2 Underwent randomiza- 13 Were lost to follow-up 7 Were lost to follow-up tion but were not treated 1 Underwent randomiza. The n e w e ng l a n d j o u r na l of m e dic i n e 786 Patients were included in intention-to-treat population 266 Were assigned to receive clindamycin 263 Were assigned to receive TMP-SMX 257 Were assigned to receive placebo 265 Received clindamycin 261 Received TMP-SMX 255 Received placebo 1 Did not receive clindamycin 2 Did not receive TMP-SMX 2 Did not receive placebo 22 Discontinued study 24 Discontinued study 41 Discontinued study regimen early regimen early regimen early 6 Had an adverse 3 Had an adverse 4 Had an adverse reaction to previous reaction to previous reaction to previous study-agent adminis. event 2 Withdrew tion but were not treated 6 Withdrew 1 Was withdrawn by 3 Withdrew investigator 234 Were included in population that 226 Were included in population that 218 Were included in population that could be evaluated and underwent could be evaluated and underwent could be evaluated and underwent secondary efficacy analysis at secondary efficacy analysis at secondary efficacy analysis at 1-month follow-up 1-month follow-up 1-month follow-up 32 Were excluded from 37 Were excluded from 39 Were excluded from secondary efficacy secondary efficacy secondary efficacy analysis at 1-month analysis at 1-month analysis at 1-month follow-up follow-up follow-up 4 Missed a visit 10 Missed a visit 9 Missed a visit 5 Received nonstudy 4 Received nonstudy 3 Received nonstudy antibiotics antibiotics antibiotics 23 Were terminated from 23 Were terminated from 27 Were terminated from study before evaluation study before evaluation study before evaluation 18 Were lost to follow-up 15 Were lost to follow-up 16 Were lost to follow-up 1 Underwent randomiza.26 nejm. 2017 The New England Journal of Medicine Downloaded from nejm. No other uses without permission. Copyright © 2017 Massachusetts Medical Society. For personal use only. study-agent adminis. 1 Was nonadherent 1 had a serious adverse tion and treatment twice 2 Underwent randomiza. . tion but were not treated 2 Withdrew tion but were not treated 1 Had a serious adverse 1 Was withdrawn by 5 Withdrew event investigator 8 Withdrew 2548 n engl j med 376.

47 Area of wound — cm2‡ 3.1) Other or unknown 6 (2. Findings from the trial are de- not treated.0) Hispanic ethnic background — no.6) Hispanic or Latino 49 (18. All rights reserved.6) 626 (79. and 281 (35.1) 24 (3.5 years. The mean TMP-SMX denotes trimethoprim–sulfamethoxazole.44±86. and 1 received the in- correct study agent.8) 4 (1.4) 159 (20.88±4. Enrollment.6) 152 (57.44 4.47 36. ‡ Area was calculated with the use of the formula for an ellipse: (length × width × π)/4.3) 9 (3.2) 208 (79.9) 55 (21. Copyright © 2017 Massachusetts Medical Society.0%) were male (Table 1). 2017 2549 The New England Journal of Medicine Downloaded from nejm.0) Black or African American 165 (62.4) 505 (64. adults.1) 51 (19.1) 73 (28.63±0.0) 152 (57.64±0.* Clindamycin TMP-SMX Placebo All Groups Characteristic (N = 266) (N = 263) (N = 257) (N = 786) Sex — no.1) 4 (1.5) 2 (0.3) 3 (1.82 * Percentages may not total to 100 because of rounding.68±93. R e sult s ated and was included in the secondary efficacy analysis at the 1-month follow-up included participants who Participants missed the test-of-cure visit (TOC) but completed the We enrolled 786 participants: 505 (64.1) 87 (33.4) Asian 8 (3.6) White 80 (30. Demographic and Clinical Characteristics of the Study Population.0) 172 (65.7) 31 (11. 2017.4) 55 (20.7) 448 (57.4) 59 (23. Trial of Antibiotics for Smaller Skin Abscesses Table 1.4) 0 0 1 (0.2%) were 1-month follow-up visit.2) Unknown 1 (  June 29.8%) were children.8) 8 (1.47 36.2) Body temperature — °C 36.43 3.9) 11 (4.7) Age — no.85±104. TMP-SMX denotes trimethoprim–sulfamethoxazole.10 study agent was not on June 28.5) 2 (0.1) 9 to 17 yr 39 (14.5) Multiracial 5 (1.2) 101 (39. (%) Male 140 (52.64±0. Randomization. Patients could have been exclud- ed from the efficacy analyses for more than one reason. 0. age at enrollment was 25.34 29.2) 1 to 8 yr 56 (21.4) 168 (65.4) 111 (42.4) 2 (0.0) 4 (1.90 3.9) 98 (12. † Race and ethnic background were reported by the participants.4) 240 (30.3) 338 (43.4) 3 (0.2) 8 (3.8) 1 (0. For personal use only.1) 202 (78.76±3.11 27. Figure 1 ( facing page).0) Female 126 (47. 1 recalled having taken a nonstudy drug before enrollment.76±53.5) ≥18 yr 165 (62. No other uses without permission.8) 156 (60. (%) <1 yr 6 (2.89±4.8) Hawaiian or Pacific Islander 2 (0. ing boundary.1) Race or ethnic group — no.6) 13 (1.30 Area of surrounding erythema — cm2‡ 26. (%)† Native American or Alaskan Native 0 2 (0. A total of 448 participants (57.8) 167 (65. In the final analysis. 2 of these 5 underwent randomization but scribed in accordance with CONSORT guidelines.0) 484 (61.76 25.46 36.0) 166 (21. The population that could be evalu.26 nejm.67±0. P values of and Follow-up. .04 or lower were considered to indicate statis- Five participants underwent randomization but were tical significance. Clindamy- n engl j med 376.04±4. (%)† Non-Hispanic and non-Latino 216 (81.8) 17 (2.8) 28 (10.8) 14 (1.

8) results of abscess culture were available for 781 Coagulase-negative staphylococcus isolated 104 (13.0) lower than that in the clindamycin group (rate Acinetobacter species 4 (0. (%) group assignments and reasons for withdrawal Culture obtained 781 (99.3) evaluated (Table 3).7% Non–group A and B beta-hemolytic streptococcus 1 (0. 1). No other uses without permission.0%). Other 42 (5. P = 0. 2017 The New England Journal of Medicine Downloaded from nejm.9) (Table 1). streptococcus species in 54 (6. aureus was iso- Streptococcus species isolated 54 (6.9) points.0%). and Group B streptococcus 4 (0.0)† diameter was present in 44. to age and study group (Table 4).9) lated in 527 participants (67.4%).0.0 cm or smaller in Staphylococcus aureus isolated 527 ( on June 28.9) (13. 95% CI.9% Viridans group streptococcus 36 (4. Bacterial growth not otherwise specified 5 (0. The cure rate in the placebo group was significantly Other species isolated 118 (15. the study- no.2) participants (99.9) −  June 29.26 nejm.1) Clinical Cure at the Test-of-Cure Visit Beta-hemolytic group C streptococcus 2 (0. children had a significantly higher cure rate with clindamycin 2550 n engl j med 376.4) dence interval [CI]. A new lesion at a different body Lactobacillus species 1 (0.4%) (Table 2): S. tion that could be evaluated. Corynebacterium species 31 (3. −8.1) (215 of 263) in the TMP-SMX group. The mean abscess depth was Culture obtained but no growth 32 (4. The difference between the cure rate in Eikenella corrodens 3 (0.4) and TMP-SMX.0) P<0.6) the study regimen (Table S1 in the Supplemen- Positive culture results 718 (91. and the mean abscess area was 3.2%).4) in the intention-to-treat population were 83.3) Logistic-regression analysis was performed to * Participants whose lesions grew multiple organisms are counted once for determine whether cure rates differed according each species identified.1% Beta-hemolytic group G streptococcus 1 (0. In the popula- † The culture from one participant grew a methicillin-resistant S.001).3) tary Appendix).7.1) that in the TMP-SMX group (rate difference.4 to 5. . anginosus 1 (0. The MSSA 140 (17.3) common causes of treatment failure in the pla- Prevotella species 2 (0. 95% CI. The n e w e ng l a n d j o u r na l of m e dic i n e Table 2.0 to −6. S.73). An abscess of 2.5) difference. −20. with significantly different Fusobacterium species 1 (0. P<0.89 cm2 Culture obtained but results not available 31 (3. 2017.4) the TMP-SMX group and that in the clindamycin Enterobacter species 2 (0.6% of participants MRSA 388 (49. −14. MRSA in 388 (49.9 percentage points.1) S. TMP-SMX to 263 participants.3 percentage Enterococcus species 7 (0. −22.4) no significant difference between clindamycin Klebsiella species 3 (0. 95% confi- Actinomyces species 3 (0.001) and Citrobacter freundii 1 (0. A total of 343 participants were fully adherent to Culture not obtained 5 (0.9%).* cin was assigned to 266 participants.1) (Table S8 in the Supplementary Appendix).4.1) cure rates for placebo versus either antibiotic but Haemophilus species 3 (0.3) The rates of clinical cure at the test-of-cure visit Beta-hemolytic group F streptococcus 3 (0. aureus (MSSA) isolate. Diphtheroid bacilli 16 (2. agalactiae 1 (0. The results were similar for the population that could be Escherichia coli 2 (0.5) other organisms in 118 (15.8 to −5.4) from the trial are summarized in Figure 1. All rights reserved. Five participants underwent ran- Result (N = 786) domization but were not treated.6) (177 of 257) in the placebo group (Table 3).4) (Table S2 in the Supplementary Appendix).3) group was not significant (−1. aureus (MRSA) isolate and a methicillin-susceptible S. and 68.1) 1. 81.64 cm.1) site or the use of a rescue medication were more Peptostreptococcus species 2 (0. Copyright © 2017 Massachusetts Medical Society.3) cebo group than in the active-treatment groups Proteus mirabilis 9 (1.2 percentage points.1) (221 of 266) in the clindamycin group. For personal use only. and placebo to 257 partici- Patients with Result pants (Fig. Results of Abscess Culture.6) Treatment failure was rarely due to worsening of the original lesion. coagulase-negative staphylococci in 104 Group A streptococcus 7 (0.

5–89.8–86.26 nejm. For personal use only.3–100. % (95% CI) Total No.5) 59/65 90.2–97.6) 69/76 90. and the population that could be evaluated includes participants who received treatment or placebo and completed the required study visits. 2551 .9 (72.0 (67.5) 116/168 69.9) 29/44 65.3–87.0–99.0–98.1–91.5–83.1–97.7 (75.* Group Clindamycin TMP-SMX Placebo No. % (95% CI) Total No.0 (61.3) 177/220 80.0–71.5–91.8) Adults Intention-to-treat population 131/165 79.8 (56.5) 69/83 83.3–87.7 (88.1 (74.3–78.9–85.0) 75/81 92.4–91.3) 59/72 81. with Cure/ Total No.0) 102/160 63.2) 110/117 94.8 (83.8 (74.6 (86.0 (98.6 (78.3–98.0 (89.5 (74.7) Population that could be evaluated 90/92 97. with Cure/ No.4 (72.3 (61.2) Population that could be evaluated 116/126 92.8) 61/89 68.8–81.6 (67. The intention-to-treat population includes all participants who underwent randomization.1) n engl j med 376.4–98.3) Population that could be evaluated 131/146 89.9) Population that could be evaluated 221/238 92.0–90.7 (80.1 (88.9) 140/172 81.9) 61/74 82. 2017.0–91.0) 39/43 90.0–88.9 (50. % (95% CI) All participants Intention-to-treat population 221/266 83.6% after adjustment for the interim analysis.2) 116/146 79.0–85.1 (86.7–72.9–97.2) 39/49 79.5 (72.0–  June 29.1 (82.8– on June 28. Downloaded from nejm.9–89.1) 149/179 83.7 (89. Table 3. Cure Rate at Test-of-Cure Visit in the Overall Population and Relevant Subgroups.6–86.1) Population that could be evaluated 41/41 100.1) Children Intention-to-treat population 90/101 89.5–95.3–98. with Cure/ No. Intention-to-treat population 57/68 83.8 (83.5 (58.9 (53.3–93.0) 140/151 92.9–74.9) * The actual confidence interval was 95.0) 29/38 76.2 (77. aureus isolated Intention-to-treat population 157/188 83.5 (77.1 (78.5–91.9 (89.2) 73/116 62.4) 215/232 92.9–97.7 (76.4 (73.3) S.8 (94.4 (74.8–100.5 (82.4) 110/130 84.9) 149/160 93.3–91.7) 73/96 76.7) MRSA isolated Intention-to-treat population 116/142 81.9–100.7) 75/91 82.1 (68.1 (79. No other uses without permission. All rights reserved.9 (62.7 (84.5–95.9) 215/263 81.0 (90.3–96.1) Trial of Antibiotics for Smaller Skin Abscesses No S.7) 177/257 68.5) Population that could be evaluated 157/167 94.4) 102/134 76.8–86. 2017 MSSA isolated The New England Journal of Medicine Intention-to-treat population 41/46 89. aureus isolated Copyright © 2017 Massachusetts Medical Society.6–97.4 (75.8) Population that could be evaluated 57/63 90.

All rights reserved. P = 0.63).04 0.98 Population that could be evaluated 0.99) (Table 3). 2017. indicat- ing that the differences in cure rates between children and adults were not significant in each respective study-group comparison (P>0. 62. 2017 The New England Journal of Medicine Downloaded from nejm. No other uses without permission.26).11 0. clindamycin and P = 0. was significantly lower than that in the clindamy- tive for S.001 0. Clindamycin Placebo vs. ated.37 <0.87 Interaction Intention-to-treat population 0.83 Population that could be evaluated 0.03 for (P<0. Clindamycin Placebo vs.001 <0.06 0.6% of those treated with TMP-SMX and (Table 3). and this treat.001) or the clindamycin group for clindamycin vs. The cure rates among participants with an Among MRSA-infected participants in the abscess that did not grow S.003 Population that could be evaluated 0. aureus iso- The cure rates in the clindamycin and TMP-SMX lates that were resistant to clindamycin — 12 2552 n engl j med 376.03 0. P = 0. the differences between placebo and clinda­ mycin and between placebo and TMP-SMX were significant in both the intention-to-treat population and the population that could be evaluated (P<0. group and 65. . placebo).03 for placebo vs.8% in the place. Similar results were results were similar for the population that observed for the population that could be evalu- could be evaluated.74 * The  P values refer to the results of a logistic-regression model incorporating study group (clindamycin on June 28. The cure rate in the placebo group tion-to-treat population who were culture-posi. cant difference in cure rates between adults and Among the participants who were infected children in the comparison between the TMP. TMP-SMX Study group Intention-to-treat population 0.17 0.001). lation that could be evaluated. In tion-to-treat population. Logistic-Regression Model of Cure Rates among Children and Adults. 81.04 SMX group (P = 0. clindamycin). These rates were significantly difference between the cure rate in the clindamy- higher than the cure rate of 63. None of the interaction terms for the logistic-regression models were significant. ment advantage with clindamycin was signifi.87).2% in the TMP-SMX group than that in the TMP-SMX group (P = 0. there were no in the clindamycin group were cured. The n e w e ng l a n d j o u r na l of m e dic i n e Table 4.6% of participants in the TMP-SMX adults in any study-group comparisons (Table 4).17 <0.05 for the intention-to-treat population and the population that could be evaluated).01) but not significantly lower group and 83. whereas the cure rate in the placebo group was cantly greater than that seen among adults (for significantly lower than that in either the TMP- the age-group differences in cure rates. aureus in the inten- SMX group and the placebo group (P = 0.16). Variable and Population P Value in the Logistic-Regression Model* TMP-SMX vs. For personal use only.1% of the participants the intention-to-treat population. groups did not differ significantly (P = 0. cin group and that in the TMP-SMX group was bo group (P<0.001). The results were similar for the popu- clindamycin vs. group (Table 3). The not significant (P = 0.26 nejm. adults).09 0. TMP-SMX and P = 0. TMP-SMX) and age group (children vs. aureus in culture intention-to-treat population.7% of those were similar for all treatment groups in the inten- treated with clindamycin had been cured by the tion-to-treat population and the population that time of the test-of-cure visit. Copyright © 2017 Massachusetts Medical Society.9% of participants in the placebo The cure rates among participants in the inten.11 0.04 for TMP-SMX vs.001 for both comparisons). After controlling for the effect of  June 29.5% in the clindamycin cin group (P = 0. than with TMP-SMX or placebo.001 Age group Intention-to-treat population 0. aureus were 83. The cure rates for clindamycin in the population that could be evaluated were signif- icantly higher among children than among adults (among children. as com- significant differences between children and pared with 79.9% of those who received placebo (Table 3). with methicillin-susceptible S. The (P = 0. as compared with could be evaluated (P = 0.99 for all comparisons) 84.06 0. 89. there was no signifi. There were 13 participants with S.

73. .. rash. drainage plus placebo.2 to 2. fections among children who were treated with Our trial yielded other new findings. 95% CI. and hepatitis. No other uses without permission. aureus The rate of treatment-associated adverse events isolates (7 of 13 [53.8. 2017. The perirectal abscess.01) were significant. P = 0. three epi- TMP-SMX group (−10. ported in 8 participants (Table S7 in the Supple- centage points. 5. TMP-SMX and clindamycin groups was signifi. clindamycin was as effective as TMP-SMX. although the latter reason was cebo-treated participants in conjunction with infrequent.1 incision and drainage that seems limited to pa- percentage points.1. a new CI. and 12. and both cure rates were significantly higher (by cant (6. Only results were similar in the population that could one episode. 95% sodes of worsening cellulitis or abscess. All rights reserved. without sequelae and were judged by the investi- cebo group and the clindamycin group (−15.8%] vs.6 who found higher cure rates among among the reasons for failure at the 1-month TMP-SMX–treated participants than among pla- follow-up visit. The results com- were not significant. were diarrhea (16. There were no epi- ticipants remained cured. thought by the investigator to be related to the pants who had been found to be cured by the study drug (TMP-SMX).001) and between the placebo group and the tus asthmaticus.9. The most common adverse events At the 1-month follow-up visit in the intention.4 percentage points.8. The cure rate among clin­ (95% CI. Trial of Antibiotics for Smaller Skin Abscesses isolates found to be resistant by single-agent compared with 4. 2017 2553 The New England Journal of Medicine Downloaded from nejm. The difference in the rates of incision and drainage plus clindamycin or inci- interval or recurrent infections between the sion and drainage plus TMP-SMX were similar.2%. −18. 0. ver.7 percentage  June 29. 95% CI. P = 0. mentary Appendix).2%. For personal use only. this trial did not in- toward higher rates of interval or recurrent in.3%]. and 2.6% (209 of 266) of the tively) and nausea (2. and the reaction re- time of the test-of-cure visit. −8. group. these percentage points. the Supplementary Appendix). and 6. Our findings show a clini- 95% CI. 13.8 to 12. rates between the TMP-SMX group and the There were nine serious adverse events re- clindamycin group was not significant (−5. a case of sta- P<0.5%) (Table S5 in the Supplemen- Clinical Cure at the 1-Month Follow-up Visit tary Appendix).06) and the difference cal benefit of antibiotic therapy in addition to between the placebo and TMP-SMX groups (−1.5% (29 of 215) of TMP-SMX recipients. −0. and ate and resolved without sequelae (Table S6 in 62. 145 of 170 [85. damycin recipients with clindamycin-resistant isolates was significantly lower than that among Adverse Events participants with clindamycin-susceptible S. new infections at a solved without sequelae.88) tients with S.2 to 6.9 gator not to be related to the study agent.6 per.26 nejm.4%. 12 to 13 percentage points) than that among P = 0. 4.03).8% (15 of Discussion 221) of clindamycin recipients.1%) or the pla- cebo group (12.7% respec- to-treat population.9 percentage points sion (D-zone) testing. aureus infection.4% respec- clindamycin-treated participants. 95% CI.4% (4 of 90) in the clindamycin testing and 1 found to be resistant by disk-diffu.6 percentage on June 28.0. 78. Copyright © 2017 Massachusetts Medical Society. clude children 12 years of age or younger). P = 0. and tions in this group was 13. than in the TMP-SMX group (11. P = 0. as the cure rates associated with either agent were n engl j med 376. The difference in cure sodes of Clostridium difficile–associated diarrhea. a difference of 8. plement findings from the trial conducted by tion or worsening of the original lesion were Talan et al.3%.05). P = 0. 95% CI. different body site or a recurrent infection at the original body site had occurred in 6. included a motor-vehicle accident. 1. Eight of the events resolved The differences in cure rates between the pla. but the difference between the placebo participants who were treated with incision and and clindamycin groups (5. the rate of interval or recurrent infec. and a case of emesis. A new lesion at a new loca.16).0.7 to −2. a hypersensitivity reaction with fe- be evaluated. First. was higher in the clindamycin group (21.6% (161 of 257) of the placebo-treated par.3% (10 of 75).1 to 18.6 to 12. Most adverse events were mild or moder- 263) of the TMP-SMX–treated participants.1.0% (192 of tively). There was also a nonsignificant trend abscess drainage (however. TMP-SMX. a case of pneumonia. thrombocytopenia.0 to −7. −24.4% (22 of 177) of The cure rates for simple abscesses treated with placebo recipients. −13. was At the 1-month follow-up visit among partici.01).9%) P = 0.

R. (UL1RR033176. NIAID) for assistance and local prevalence of resistance should be used by advice. Copyright © 2017 Massachusetts Medical Society. Chambers. No other uses without on June 28. Abbott.S. Participants infected with a clinda­ ment failure at the test-of-cure visit. outcomes among patients with uncomplicated The cumulative data from our investigation cutaneous abscesses.6 call into question the by S. not be assessed because there were no resistant In conclusion.. other potential conflict of interest relevant to this article was reported. . We recurrences or new infections after completion followed participants for 1 month. aureus isolate who were treated was not evaluated at the 1-month follow-up visit). particularly in children. particularly those caused and that of Talan et al. Such information and the biology and Infectious Diseases. our results show that short-term isolates in vitro. Fourth. these data underscore the poten. a longer fol- of therapy. and Hyung Koo. ings suggest that there is a trade-off between Disclosure forms provided by the authors are available with more adverse effects and a lower likelihood of the full text of this article at NEJM. Sciences. Achaogen.6 Third. doxycycline (which is con- that trial. R.. Allergan.. infection recurrence when one uses clindamycin We thank Christine children and adults both ben.N. B.17 The potential for participants to have effective in preventing recurrent or new infec. Miller. and lower-quality non. Pfizer. (all at the Division of Micro- rather than TMP-SMX. inferiority trials11-14 — that cure rates do not The content is solely the responsibility of the authors and improve with the addition of systemic antibiotic does not necessarily represent the official view of the National treatment after incision and drainage. receiving grant support from SMX was associated with a hypersensitivity reac. but there are others that may be just as tered for 10 days instead of the 7 days used in effective. and consulting fees from Theravance. In this trial. if any.N. underpowered. Dr. and clindamycin was associated with more Kline. clindamycin may er cure rates that we found. although TMP-SMX was adminis. to abscess incision and drainage. Creech.15 These Institutes of Health.. All rights reserved. For personal use only. Maureen Mehigan.. grant support and consulting fees from GlaxoSmith- tion. Our find. Two antibiotics that participants in either antibiotic group. The n e w e ng l a n d j o u r na l of m e dic i n e higher than that associated with placebo. treatment failure but be cured of infection sub- tions. against MRSA strains. such as at the 1-month follow-up visit. traindicated for children younger than 8 years of efited from active therapy.6 Finally. Second. UL1TR000124). No known side effects must be considered. Dr. Merck. The contribution of re. aureus. now at the National Center for Advancing Antibiotic-related side effects. therapy improves cure rates for skin abscesses Chambers) and the National Center for Research Resources and decreases the recurrence rate. are improved by antibiotic treatment perception — largely based on expert opinion or with either clindamycin or TMP-SMX in addition smaller.D. the marginal bene- be more effective than TMP-SMX in preventing fit. if a patient had a treat- clindamycin. Table treating physicians and policy makers when S8 in the Supplementary Appendix shows that choosing an antibiotic for adjunctive therapy of new infections developed more frequently in par.N. and Dr.26 nejm. the patient mycin-resistant S. 2554 n engl j med 376. Pfizer. and Cepheid and consulting fees from Tetraphase. particularly if Translational Sciences. cutaneous abscesses. perhaps a low-up period may have captured more recur- higher dose of TMP-SMX would have been more rences. should be taken into ac. and Cempra.S. cussed provide a potential direction for future sistance to TMP-SMX to treatment failure could analyses. difficile–associated diarrhea or severe fees for serving on an advisory board from AstraZeneca. B. adverse events than TMP-SMX. would probably be small at best. are commonly used and recommended for the TMP-SMX was effective at half the dose used by treatment of uncomplicated skin infections were Talan et al. given the high- SMX among children. sequently. studied. with clindamycin had cure rates similar to those The exploratory secondary analyses we have dis- who were given placebo. Theravance. allergic reactions were observed. tial clinical relevance of in vitro resistance to was not assessed (i.e. although clindamycin age and was not studied in this trial) is active may have performed slightly better than TMP. For example. two larger trials show that adjunctive antibiotic Supported by a contract from the National Institute of Allergy and Infectious Diseases (NIAID) (HHSN272200700031C to Dr. ticipants who received placebo than among Our trial has limitations. these and other Cubist (now Merck). receiving grant support from Gilead abscess with an antibiotic. M. Dr. Although no previously holding stock in Merck and receiving grant support and fees for serving on an advisory board from Genentech and cases of C. TMP. Daum reports receiving fees for serving on an advisory frequent or serious. 2017.16  June 29. board from Pfizer and Dynavax and grant support from Thera- count when deciding whether to treat a drained vance and Merck.. 2017 The New England Journal of Medicine Downloaded from nejm.

Randomized. Cosgrove SE. Trimethoprim–sulfamethoxazole community-associated methicillin-resistant es failure and recurrence in methicillin- versus placebo for uncomplicated skin ab. stitute. ment of uncomplicated skin abscesses in a tions in ambulatory and inpatient settings. Trial of Antibiotics for Smaller Skin Abscesses References 1.​31:​828-37. BMC Infect Dis 2015. by community-acquired methicillin-resis. 8.​372:​1093-103. et al. Bayer A. Practice guidelines for the diagnosis 4. J Pedi- 7. McGinnis HD. fections among patients in the emergency 10. et al. Antimicrob Agents Chemo­ X. et al. et al. fections. population at risk for community-acquired 2005-2010. 14. Krishnadasan cated skin abscesses in patients at risk for thoprim-sulfamethoxazole therapy reduc- A. Daum RS. 2017 2555 The New England Journal of Medicine Downloaded from nejm. children. Clin Microbiol Rev 2010. N Engl J Med 2015. Altman on June 28. Abscess incision methicillin-resistant Staphylococcus aure- 2.​169:​128-34. Methods for dilution antimicro. Baren- skin and skin-structure infections: deriva. N Engl J Med 2016. 10th munity-acquired skin abscesses in the 3. thoprim-sulfamethoxazole for uncompli. Duong M. Rios AM. Daum RS. Bayer A. Clin Infect Randomized controlled trial of trime.​357(19):​ us infection. Clin Infect Dis 2011. n engl j med 376. Ann Emerg Med 2010. Liu C.​52(3):​e18- children: executive summary. e55. et  June 29. Dis 2014. ther 2007. et al. Ann Intern Med 2010. tious Diseases Society of America for the tious Diseases Society of America for the tant Staphylococcus aureus. Aten MF. Holmes L. et al. lococcus aureus infections in adults and lococcus aureus infections in adults and 12. kamp S. 2017.​152:​ infections: 2014 update by the Infectious coccus aureus: epidemiology and clinical 726-32. 13. Stevens DL. lines for reporting parallel group ran. domized trials. Schulz KF. Clindamycin versus trimethoprim–sulfa. et al. Eisenberg DF.​23:​123-7. Copyright © 2017 Massachusetts Medical Society. Miller LG. et al. 11. Dis J 2004. Clin Infect consequences of an emerging epidemic.e1. Miller LG. Trime- 6. methoxazole for uncomplicated skin in. Bisno AL. N Engl J Med 2006. Rajendran PM. Mower WR. Emerg Med 2010.​56:​283-7. Cosgrove SE. Peter J. atr 2016. Miller LG. et al. 55:​401-7. Talan DA. Gorwitz ed. Qiao H. Chambers HF. CLSI. Community. Ma C. Sun and drainage. Infect bial susceptibility tests for bacteria which antibiotics in the management of com- Control Hosp Epidemiol 2010. department. Moher D. PA:​ pediatric patient. abscesses after surgical drainage. Markwell S. CLSI document M07-A10. Lipsky BA. grow aerobically:​approved standard. Randomized. Liu C. controlled trial of cephalexin for treat- Incidence of skin and soft tissue infec. Kollef MH. Lee MC. Creech CB. placebo. Nicks BA. Copyright © 2017 Massachusetts Medical Society. controlled trial of tion and validation of a risk score. Moran GJ. 15. and management of skin and soft tissue associated methicillin-resistant Staphylo. Krishnadasan A. Young D. Ann resistant Staphylococcus aureus skin scess. Fitch MT. Pediatr Infect treatment of methicillin-resistant Staphy- treatment of methicillin-resistant Staphy.​355:​666-74. Pariyadath M. et al. CONSORT 2010 statement: updated guide. Johannes RS. 5.​59(2):​e10-e52.​23:​616-87. N Engl J Med 2007. Manthey DE. All rights reserved. Staphylococcus aureus infection. Schmitz GR. e20. Dis 2011. Pitotti R. Maurer T. Tabak YP. January 2015. double-blind. For personal use only. Wayne. Liu H.​52:​285-92. teremia among patients hospitalized for 9. Clinical and Laboratory Standards In. David MZ. Methicillin-resistant S. Management and outcome of children 16. aureus in. No other uses without permission. Predicting bac. .​374:​823-32. Diseases Society of America.​15:​362.26 nejm. Bruner D. 17.​51:​4044-8.​ RJ. with skin and soft tissue abscesses caused Clinical practice guidelines by the Infec- Clinical practice guidelines by the Infec.