BACKGROUND AND OBJECTIVES: Skin and soft tissue infections (SSTIs) are an increasingly common cause of pediatric hospital visits among infants. The optimal evaluation strategy for younger infants with SSTI is unknown because there is little information about outcomes including risks of concomitant bacterial infections and treatment failure. This study was designed to determine rates of concomitant invasive bacterial infection and hospital revisits for treatment failure as well as factors associated with treatment failure in infants presenting with SSTI.
METHODS: Retrospective study of patients ≤90 days of age who received care from the 22 emergency departments and hospitals in the Intermountain Healthcare system from July 1, 2004 to December 31, 2011, with a primary discharge diagnosis of SSTI. Concomitant bacterial infections were defined as urinary tract infection (UTI; culture-confirmed) or invasive bacterial infection (IBI; culture-confirmed bacteremia and/or meningitis). Treatment failure was defined as any unplanned change in care at hospital revisit within 14 days of discharge.
RESULTS: The study included 172 infants; 29 (17%) were febrile, and 91 (53%) had ≥1 sterile site culture performed. One case of bacteremia in a febrile infant was identified giving an overall proportion with UTI/IBI of 0.58% (95% confidence interval 0.01%–3.2%). Sixteen infants (9.3%; 95% confidence interval 5.4%–14.7%) returned for treatment failure. Perianal location (P = .03) and private insurance status (P = .01) were associated with more treatment failures compared with other locations or payer types. No patients returned for missed UTI/IBI.
CONCLUSIONS: Concomitant bacterial infections were rare in infants with SSTI, with none identified in afebrile infants. Treatment failure of SSTI leading to hospital revisit was common.
Skin and soft tissue infections (SSTIs), most commonly cellulitis or cutaneous abscess, are an increasingly common cause of pediatric hospital visits, including among infants.1,2 Although recent national guidelines for SSTIs have evaluation and treatment recommendations in children and adults,3 young infants with SSTI represent a unique challenge to the clinician. Infants ≤90 days of age who present with fever without a source are often evaluated for urinary tract infection (UTI) and invasive bacterial infection (IBI) because the rate of UTI/IBI in this age group can reach 8% to 17% even if the infant is well-appearing.4–7Given the risks of UTI/IBI in febrile infants in this age group, many infants who present with SSTI are evaluated for additional concomitant bacterial infections regardless of fever status.8–10 Previous single-center studies in tertiary children’s hospitals in urban settings have explored risks of UTI or IBI in infants with SSTI or Staphylococcus aureus infections; however, revisits for treatment failures or missed UTI/IBI outcomes may not have been captured because of the study setting.8,10,11
Our study, using a multicentered approach over a range of hospital types including rural and urban settings in a network well suited to capture missed concomitant infections or treatment failures, could help to guide optimal clinical management including indication for workup of concomitant infections and follow-up planning. Therefore, our objective was to describe clinical outcomes for infants presenting to the hospital for concern of SSTI (with and without fever) including rates of UTI, IBI, and hospital revisit for treatment failure or missed concomitant bacterial infection as well as factors associated with treatment failure.
Population and Data Source
This was a retrospective study of infants ≤90 days age who received care at an emergency department (ED) or observation/inpatient unit at a hospital in the Intermountain Healthcare system between July 1, 2003 and December 31, 2011. Intermountain Healthcare is a not-for-profit, integrated health care system that provides care for ∼85% of Utah’s children (>95% of infants <90 days) with 22 hospitals across Utah and Idaho and 1 tertiary children’s hospital (Primary Children’s Hospital in Salt Lake City).
Infants ≤90 days of age with a primary or secondary International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9) discharge diagnosis code consistent with SSTI were assessed. Ultimately only patients with primary discharge ICD-9 code for SSTI were included. Similar to previously published studies on SSTIs, our study used ICD-9 codes to identify potential subjects.1,2,12,13 See online Supplemental Appendix 1 for inclusion and exclusion ICD-9 diagnosis, procedure, and Current Procedural Terminology codes. To minimize misclassification, we reviewed the charts of all included patients with primary ICD-9 for SSTI as well as a random sample of 20% of the patients with a secondary ICD-9 for SSTI. Further exclusions included ICD-9 codes for birth hospitalization, chronic medical condition,14 immunocompromise,15 or codes for complicated infection (such as burn, surgical site, or device-related).
All data were obtained from Intermountain Healthcare’s Enterprise Data Warehouse that contains both clinical and administrative data elements. The electronic medical records were reviewed to verify clinical diagnosis and to determine additional clinical information not obtained by administrative data.
Outcomes and Definitions
All UTI or IBI were culture-confirmed. UTI was defined as urine culture with ≥50 000 colony forming units/mL of a single organism.16 IBI was defined as bacteremia or meningitis with positive growth of a pathogenic organism on culture. Treatment failure was defined as any unplanned change in care (eg, medication change, drainage procedure, or hospital admission for ongoing treatment of SSTI) at a hospital revisit encounter (defined as ED, observation, inpatient, or same-day surgery) within 14 days of discharge from the original encounter. Missed UTI/IBI was defined as UTI/IBI not identified on initial hospital encounter, but identified on hospital revisit within 14 days of discharge.
Antibiotic use was categorized as “previous antibiotics,” “encounter antibiotics,” and “discharge antibiotics.” Previous and discharge antibiotics were defined as antibiotics reported by caregiver to have been used before encounter or recommended by the provider at the time of discharge, respectively, and were obtained by review of clinical documentation. Encounter antibiotics were defined as antibiotics given during the hospital encounter and were obtained via administrative charge code for the antibiotic. Encounter antibiotics were classified as potentially active against methicillin-resistant S aureus (MRSA), including trimethoprim/sulfamethoxazole, clindamycin, vancomycin, or linezolid or inactive against MRSA. Infection location was categorized as primarily occurring in perianal, extremity, trunk, or head based on medical provider examination in clinical documentation. Fever was defined as parental report or recording of temperature ≥38.0°C (100.4°F) at the time of intake to the hospital encounter. To capture procedures not captured by coding, incision and drainage procedure was defined as description of incision or drainage by clinician in clinical documentation.
Descriptive statistics were used to describe patient characteristics and the rates of primary and secondary outcomes. χ2 or Fisher’s exact test (dichotomous or unordered categorical), Wilcoxon-Mann-Whitney (ordered categorical) or independent groups t test (continuous) were used as appropriate for the variable type to compare the frequency of treatment failure in subpopulations. All confidence intervals (CIs) were 2-sided 95%. P values < .05 were considered statistically significant. Stata 12.0 was used for all analyses.
One hundred eighty-five infants were identified by using primary ICD-9 codes. After medical record review, 13 infants were excluded (3 with no records for date of encounter, 5 with omphalitis, 2 with deep neck abscess, 2 with surgical site infection, 1 with oral thrush) leaving 172 infants included in the study cohort. A further 184 patients had a secondary ICD-9 code for SSTI; a random sample 20% of these charts were reviewed, and none would have met study criteria and as such were excluded.
Of the 172 included infants, 29 (17%) were febrile on presentation. Fever status was unknown in 4 infants. Fifty-three percent of patients (91 of 172) had ≥1 culture performed for UTI/IBI evaluation. Febrile infants were more likely to have a culture performed for UTI/IBI evaluation (26 of 29 or 89.7% of febrile infants vs 54 of 139 or 38.9% of afebrile infants, P < .001).
The overall proportion of infants with either UTI or IBI identified was 1 of 172 (0.58%; 95% CI 0.01%–3.2%). Of patients with blood cultures obtained, bacteremia was identified in 1 of 76 (95% CI 0.03–7.1). The patient with bacteremia was part of the febrile group, making the proportion of bacteremia in this febrile group 1 of 25 (4%; 95% CI 1.0–20.0), vs 0 of 50 (1 patient had unknown fever status) in the nonfebrile group (0; 95% CI 0.0–7.1). Thirteen of 42 patients (31%) with urine culture obtained had fever and may have been evaluated for UTI because of an existing care process model for the evaluation of febrile infants in this age group.17 The total proportion with UTI among infants with urine cultures obtained was 0 of 42 (95% CI 0.0%–8.4%). All patients with cerebrospinal fluid culture obtained had fever; the total proportion with meningitis was 0 of 28 (95% CI 0.0%–12.3%).
One infant had IBI identified. The 66-day-old infant was febrile on presentation and had β-hemolytic Streptococcus (no further speciation documented in microbiology results) bacteremia in addition to groin cellulitis. Chart review indicates this patient was discharged from the hospital from the ED and was treated as an outpatient with an oral third-generation cephalosporin with no hospital revisits.
Of the 172 infants, 16 (9.3%; 95% CI 5.4%–14.7%) returned to the hospital setting for an encounter consistent with treatment failure. Of the 16 patients with treatment failure, 8 had a change in antibiotics, 11 had a new procedure such as incision and drainage, and 3 were admitted (patients may have had >1 type of treatment failure). Characteristics showing significant difference between treatment failure and non–treatment failure groups in univariate analysis were perianal location (11 of 16 or 68.8% and 61 of 156 or 39.6%, respectively, P = .03) and private insurance status (13 of 16 or 81.3% and 63 of 156 or 40.4%, respectively, P = .01). There was no difference in treatment failure based on the performance of incision and drainage or anti-MRSA antibiotics at the initial encounter (P = .36 and P = .56, respectively). See Table 1. No patients returned to the hospital setting for missed UTI/IBI.
Our study had 2 main findings. First, using a large cohort of patients, we found a low rate of identified invasive bacterial infections (0.58% overall, 0% in afebrile infants), among infants ≤90 days old presenting for a hospital encounter for SSTI. Although not all patients had cultures for UTI/IBI obtained on initial encounter, there were no patients who returned with “missed” UTI/IBI. Second, SSTI treatment failure was a common cause for hospital revisit (9.3%), particularly in patients with perianal location of infection.
The low rate of concomitant UTI or IBI in infants with SSTI in our study, particularly those who are afebrile, is consistent with previous literature.8,10,11 One study of 104 afebrile infants ≤28 days who presented to the ED for SSTI found that none had IBI.10 Another study of infants ≤30 days identified as having superficial or invasive S aureus infection found that of the 68 patients (total study cohort 126 patients) with cellulitis/abscess, 2 (3%) had bacteremia and none had concomitant UTI or meningitis.11 A third study assessed the rate of UTI/IBI in well-appearing infants (aged <60 days) presenting to the ED and diagnosed with focal bacterial infections- including otitis media, lymphadenitis, and a variety of SSTIs (cellulitis, abscess, impetigo, paronychia). Although the study included a total of 197 patients, in the subset of patients with SSTIs (n = 136), 3 patients (2.2%, all of whom were febrile) had UTI or IBI (1 with Streptococcus pneumoniae bacteremia and 2 with Escherichia coli UTI).8 Because the mechanism of UTI with SSTI is not apparent, it is worth noting that the authors of this article did describe a limitation of their study as being the definition of UTI based on culture (with no colony-forming unit specification) alone without supportive urinalysis findings.8 Although our study used a similar culture-specific definition of UTI, we did maintain a threshold of ≥50 000 colony forming units/mL of a single organism.
Our study adds to these previous single-center tertiary care children’s hospital studies in urban settings by using data from a large health care system that includes community/rural and regional centers as well as a tertiary care children’s hospital, allowing us to capture hospital revisits across a variety of hospital types for missed concomitant invasive bacterial infections and treatment failures.
Although our study found a low rate of identified bacteremia (0.58%), the CI (95% CI 0.01%–3.2%) included the approximate rate (∼2%) for bacteremia in febrile infants without a source.6,18 Furthermore, because not all patients had cultures obtained, it is possible that patients with bacteremia or other IBI were not identified. No patients returned for “missed” bacteremia or other IBI, but this may reflect adequate treatment of (unidentified) bacteremia with oral antibiotics, which are often used to treat SSTI.
Given that UTI and meningitis would be extremely rare (no described cases of the latter in the aforementioned literature or in our study) in infants presenting with SSTI, evaluation for these concomitant infections may not be indicated in a clinically well-appearing infant, potentially sparing the patient from invasive procedures such as lumbar puncture. Although the risk of bacteremia in infants with SSTI and fever is not dissimilar to the risk in a patient with fever without a source, the clinician must ask whether identification of the bacteremia in a patient they planned to treat with antibiotics regardless would affect management before obtaining the blood specimen.
The second main finding of our study was that treatment failure occurred at a rate of 9.3% in our cohort. It is possible that the proportion with treatment failure in our cohort may have been >9.3% because we only captured treatment failures associated with new hospital encounters and may have missed treatment changes made in other outpatient settings such as a primary care physician’s office or an urgent care center. Despite this limitation, we found a similar treatment failure rate as the 6% to 9% rate found in 1 outpatient study of patients aged 0 to 17 with SSTI in Tennessee.12 This comparative effectiveness study identified use of trimethoprim-sulfamethoxazole or β-lactams, particularly if a drainage procedure was performed, as an association with increased risk of treatment failure compared with use of clindamycin.12 To our knowledge, no previous studies have assessed risks or associated factors for treatment failure in neonates/young infants presenting to the hospital with SSTI.
Our study found that treatment failure occurred significantly more often in infants with SSTI involving the perianal region. We suspect that this perianal location is at high risk for treatment failure due to the difficulty of keeping this area clean and the polymicrobial nature of such infections. The reason for the higher rate of treatment failure leading to repeat hospital encounter we observed among privately insured patients is uncertain but may reflect differences in access to care or copays associated with follow-up hospital or ED visits.
In some cases of SSTI, there are appropriate treatment “changes” that can be made, such as redrainage of a maturing abscess or narrowing of antibiotic selection based on culture results. In our study, we elected to define treatment “failure” as new hospital encounters, on the basis that these natural or anticipated changes might ideally be done in the outpatient setting and that a revisit to the hospital might reflect a “failure” of the system. Previous studies have used similar definitions for treatment “failure” (ie, a new outpatient SSTI claim plus an antibiotic prescription or drainage procedure or new inpatient hospitalization with discharge diagnosis code of SSTI within 14 days of the initial SSTI12). In some cases, the treatment failures found in our study may have been “appropriate” (such as if sedation was needed for a drainage procedure), but in other cases, it may reflect failure of the system to organize appropriate outpatient follow-up. Given the high rate of treatment failure found in our study, whether from abscess evolution or a need to modify the antibiotic regimen, we suggest that infants with SSTI have outpatient follow-up within 48 to 72 hours of discharge from hospital encounter.
There were several limitations to this study. The occurrence of the primary and secondary outcomes was rare, which precluded more precise estimates and controlling for confounders. We relied on ICD-9 codes to identify cases, which is subject to misclassification, although misclassification was reduced by medical record review of all included patients as well as a random sample of 20% of patients with secondary discharge diagnosis code for SSTI excluded from our study. Incomplete and nonstandardized clinical documentation during the study period precluded collection of detail on antibiotics used before or after the hospital encounter. Additional encounters could have taken place in the outpatient setting or in settings outside of the Intermountain system, resulting in underestimation of treatment failures.
One additional limitation to our study was the low total number of included patients over the study period. Although this was a multihospital study, low total population in the hospital catchment areas may have been reflected in low total study patients. Data from the 2010 US census showed that Utah had an estimated total population of 2.76 million people (vs Houston, Texas, with 2.1 million people where a single-center study with similar patient numbers discussed earlier was located).11,19 Furthermore, the demographic of Utah (79.7% white, non-Hispanic/Latino vs 62.6% in general United States19) and MRSA rates may differ from other geographic regions. Geographic variation in the incidence of MRSA infections have also been well described,20–22 unlike other geographic areas, Utah did not experience a large increase in community-associated MRSA during the study period.23,24 Unpublished 2014 internal Intermountain Healthcare laboratory data showed that MRSA wound culture rates in pediatric patients across the IH system were 28%, and 33% at the tertiary children’s hospital (K.K., unpublished observations). Unfortunately, inconsistent capture of wound cultures during the study period precluded analysis of this element in our study.
UTI/IBI is rare in infants presenting with SSTI. Clinicians may consider limiting evaluation for UTI/IBI in afebrile infants presenting with SSTI if other criteria for more extensive evaluation are not met and if identification is unlikely to change management. A clear short-term follow-up care plan is important for infants with SSTI, especially those with perianal infections, who, as seen in our study, may experience a higher rate of treatment failure.
The principal investigator (G.H.) thanks Dr Christopher G. Maloney as well as the Academic Pediatrics Association Research Scholars Program scholars and mentors for their feedback on this work.
FINANCIAL DISCLOSURE: Dr Byington has intellectual property and receives royalties from BioFire Diagnostics (Salt Lake City, UT). The other authors have indicated they have no financial relationships relevant to this article to disclose.
FUNDING: Dr Byington is supported by the National Center for Advancing Translational Sciences, National Institutes of Health (grant UL1TR001067) and by the H.A. and Edna Benning Society. The work of Gregory Stoddard and his contribution to this study was supported by the University of Utah Study Design and Biostatistics Center, with funding in part from the National Center for Research Resources and the National Center for Advancing Translational Sciences, National Institutes of Health (grant 8UL1TR000105; formerly UL1RR025764).
POTENTIAL CONFLICT OF INTEREST: Dr Byington has intellectual property and receives royalties from BioFire Diagnostics (Salt Lake City, UT). She reports no conflict of interest for this study. The other authors have indicated they have no potential conflicts of interest to disclose.
- Frei CR,
- Makos BR,
- Daniels KR,
- Oramasionwu CU
- Stevens DL,
- Bisno AL,
- Chambers HF,
- et al
- Byington CL,
- Rittichier KK,
- Bassett KE,
- et al
- Greenhow TL,
- Hung Y-Y,
- Herz AM
- Nguyen R,
- Bhat R,
- Teshome G
- Fortunov RM,
- Hulten KG,
- Hammerman WA,
- Mason EO Jr.,
- Kaplan SL
- Williams DJ,
- Cooper WO,
- Kaltenbach LA,
- et al
- Feudtner C,
- Hays RM,
- Haynes G,
- Geyer JR,
- Neff JM,
- Koepsell TD
- Nakamura MM,
- Lee GM
- Byington CL,
- Reynolds CC,
- Korgenski K,
- et al
- Biondi E,
- Evans R,
- Mischler M,
- et al
- ↵United States Census Data. Available at: www.census.gov/2010census/popmap/ipmtext.php?fl=48:4822. Accessed March 10, 2015.
- Copyright © 2015 by the American Academy of Pediatrics