OBJECTIVES: The objective of this study was to describe trends in antibiotic management for Staphylococcus aureus infections among hospitalized children from 1999 to 2008.
METHODS: A retrospective study was conducted by using the Pediatric Health Information Systems database to describe antibiotic treatment of inpatients with S aureus infection at 25 children's hospitals in the United States. Patients who were admitted from 1999 to 2008 with International Classification of Diseases, Ninth Revision, Clinical Modification codes for S aureus infection were included. Trends in the use of vancomycin, clindamycin, linezolid, trimethoprim-sulfamethoxazole, cefazolin, and oxacillin/nafcillin were examined for percentage use and days of therapy per 1000 patient-days.
RESULTS: A total of 64813 patients had a discharge diagnosis for S aureus infection. The incidence of methicillin-resistant S aureus (MRSA) infections during this period increased 10-fold, from 2 to 21 cases per 1000 admissions, whereas the methicillin-susceptible S aureus infection rate remained stable. Among patients with S aureus infections, antibiotics that treat MRSA increased from 52% to 79% of cases, whereas those that treat only methicillin-susceptible S aureus declined from 66% to <30% of cases. Clindamycin showed the greatest increase, from 21% in 1999 to 63% in 2008. Similar trends were observed by using days of therapy per 1000 patient-days.
CONCLUSIONS: Antibiotic prescribing patterns for the treatment of S aureus infections have changed significantly during the past decade, reflecting the emergence of community-associated MRSA. Clindamycin is now the most commonly prescribed antibiotic for S aureus infections among hospitalized children. The substantial use of clindamycin emphasizes the importance of continuous monitoring of local S aureus susceptibility patterns.
WHAT'S KNOWN ON THIS SUBJECT:
An increase in the incidence of S aureus infection among hospitalized children has been observed as a result of the increase in community-associated MRSA infections. Data are lacking regarding the impact of changing S aureus epidemiology on antibiotic use.
WHAT THIS STUDY ADDS:
This study shows a significant shift in the use of antibiotics to treat S aureus infections. Clindamycin now represents the most commonly used antibiotic among hospitalized children with S aureus infections.
Staphylococcus aureus is a significant cause of skin and soft tissue infections (SSTIs) as well as a variety of invasive diseases in both children and adults.1,2 Although first successfully treated with β-lactam antibiotics >60 years ago, S aureus developed resistance relatively quickly, leading to the isolation of methicillin-resistant S aureus (MRSA) within 1 year of exposure to this class of antibiotics.3,–,6 Initial MRSA infections were almost exclusively nosocomial, confined to patients with a history of repeated hospitalizations, lengthy or repeated antibiotic treatments, or close contact with health care workers.7,8 More recently, however, an epidemic of MRSA infections that affect individuals without such risk factors has emerged.9 These community-associated MRSA (CA-MRSA) infections have primarily manifested as SSTIs among previously healthy children and adults. This evidence, coupled with molecular and genetic studies demonstrating the emergence of genetically unique strains, suggested novel transmission routes outside of health care settings.10 More recent evidence suggests that this trend has led to an increase in MRSA infections among hospitalized children.11
Before this shift in epidemiology, empiric β-lactam therapy for hospitalized patients with suspected S aureus infection remained commonplace. The rapid emergence of CA-MRSA, however, has led to recommendations for targeting both methicillin-susceptible S aureus (MSSA) and MRSA in patients with suspected S aureus infections.12,–,16 This approach is complicated by the absence of an antibiotic that is ideally suited for this goal. For example, although vancomycin, clindamycin, trimethoprim-sulfamethoxazole (TMP/SMX), and linezolid each provide adequate coverage against most S aureus strains, including MRSA,14 each drug has limitations: despite broad-spectrum coverage, vancomycin has poor anti-staphylococcal activity relative to β-lactams and is available only intravenously; clindamycin is limited by increasing resistance in certain regions of the United States17; TMP/SMX is variably active against Streptococcus species (coverage of which is often desirable for such infections); and linezolid is prohibitively expensive. Accordingly, the extent to which antibiotics with activity against MRSA have supplanted β-lactams in the treatment of S aureus infections, in hospitalized children and the preferred anti-MRSA agent chosen in such circumstances is unknown. The objective of this study was to describe national trends in antibiotic management of S aureus infections from 1999 to 2008 in US children's hospitals.
We conducted a retrospective observational study to examine the use of commonly prescribed antimicrobial agents for S aureus infections among inpatients at 25 freestanding, tertiary care children's hospitals throughout the United States from 1999 to 2008.
Data Source and Quality
We used data from the Pediatric Health Information System (PHIS), an administrative database that is maintained by the Child Health Corporation of America (CHCA) located in Mission, Kansas. The CHCA is a collaboration of 43 not-for-profit, tertiary care pediatric hospitals across the United States aimed at improving outcomes and reducing costs. Member hospitals represent 17 of the 20 major metropolitan areas across the United States. The PHIS database contains patient-level demographic, diagnostic, procedural, and resource use (eg, pharmaceuticals, laboratory and imaging studies, other patient services) data from CHCA member hospitals. To date, data on >19 million patient encounters are included in the PHIS database.
Data quality and reliability are ensured through a joint effort among the CHCA, Thomson Healthcare (contracted data manager), and contributing hospitals. Data are accepted into the database once classified errors occur in <2% of a hospital's quarterly data. When a hospital's quarterly data are unacceptable according to these limits, all of their quarterly data are rejected; however, these data can be corrected, then resubmitted and reevaluated before inclusion in the database. Hospitals were excluded from our study when they did not contribute complete resource use data for the entire study period. Twenty-five hospitals had complete data and were included in our analysis.
The cohort for this investigation included patients who were younger than 18 years at time of admission to 1 of the 25 study hospitals between January 1, 1999, and December 31, 2008, with an International Classification of Diseases, Ninth Revision, Clinical Modifications (ICD-9-CM) code for S aureus infection (041.11), S aureus septicemia (038.11), pneumonia caused by S aureus (482.41), or MRSA (041.12, 038.12, 482.42, or V09.0) infection at discharge. Thus, patients were classified as having a S aureus infection when ≥1 of the aforementioned codes was in the patient's discharge data, with additional classification of S aureus as MRSA or MSSA on the basis of the presence (or absence) of a discharge code for MRSA infection. Patient encounters that contained multiple S aureus–related codes were counted only once. Patients with S aureus infections were further subdivided into those with SSTIs, osteomyelitis, endovascular infections (endocarditis and septic thrombophlebitis), and pneumonia by using additional ICD-9-CM codes. This approach was previously implemented by using the PHIS database.11 Site-of-infection codes were available for 57% of the study population.
Antibiotic Use Measures
Antibiotics that were selected for investigation included the anti-MRSA antibiotics vancomycin, clindamycin, TMP/SMX, and linezolid and the anti-MSSA β-lactam antibiotics oxacillin, nafcillin, and cefazolin. Two measures were used to examine antibiotic use for each agent: (1) percentage use and (2) days of therapy (DOT) per 1000 patient-days. Percentage use was defined as the proportion of patients who had S aureus infections and received 1 of the specified antibiotics at any time during the hospitalization; this measure assigns equal weight to a single antibiotic dose or multiple days of therapy and is therefore independent of therapy duration. To account for variations in therapy duration, we also measured antibiotic use by calculating DOT, normalized to 1000 patient-days. One day of therapy was defined as receiving ≥1 dose in a given day; patients were counted as having 1 day of therapy regardless of the number of doses administered. By using these measures, we examined antibiotic use for all patients with S aureus and subsequently stratified them by the presence of codes for MRSA and site of infection. All trends were examined annually.
Summary statistics were constructed by using frequencies and proportions for categorical data elements and medians or means for continuous variables. All temporal trends were examined by using the Mantel-Haenszel χ2 test. Because of large sample sizes, significance was determined on the basis of P < .001. SPSS Base 16.0 (SPSS, Inc, Chicago, IL) was used for all statistical analyses.
Of the 2.4 million patient discharges examined from 1999 to 2008, a total of 64813 (3%) children had a discharge diagnosis code for S aureus infection. Overall, the majority of patients with S aureus infections were male and white, with a median age of 3.0 years (interquartile range: 0.8–10.9 years). Black children were more commonly infected with MRSA than MSSA (P < .001). The most common type of S aureus infection was SSTI, occurring in 39% of all S aureus cases. SSTIs occurred more frequently among patients with MRSA compared with MSSA (P < .001), whereas pneumonia was more commonly observed in children with MSSA (P < .001; Table 1).
Trends in S aureus Infection
During the 10-year study period, the incidence of S aureus infection more than doubled, increasing from 14.8 per 1000 admissions in 1999 to 35.7 per 1000 admissions in 2008 (Fig 1). The incidence of MRSA infections during this period increased 10-fold, from 2.0 cases per 1000 admissions in 1999 to 20.7 cases per 1000 admissions in 2008 (P < .001 for trend), whereas the MSSA infection rate remained relatively stable. By 2008, MRSA accounted for 58% of all S aureus infections in hospitalized children.
Temporal Trends in Antibiotic Use for All Patients With S aureus
Concurrent with this shift in S aureus epidemiology in hospitalized children, antibiotic use changed significantly.
Measured as percentage use, the overall use of anti-MRSA antibiotics for patients with S aureus infections increased from 52% to 79% of cases. Clindamycin showed the greatest overall increase, with percentage use increasing three-fold (21% in 1999 to 63% in 2008; P < .001 for trend; Fig 2). Vancomycin use remained stable (36% in 1999 vs 37% in 2008; P = .003 for trend), whereas TMP/SMX (9% vs 12%; P < .001 for trend) and linezolid (0% vs 5%; P < .001 for trend) showed small increases. In contrast, β-lactam antibiotic use declined significantly during the study period, from 66% of S aureus cases in 1999 to <30% in 2008 (P < .001 for trend).
DOT per 1000 Patient-days
Overall, similar trends were evident when antibiotic selection was examined in terms of DOT per 1000 patient-days (Fig 2). Clindamycin use increased from 53 DOT per 1000 patient-days in 1999 to 192 DOT per 1000 patient-days in 2008 (P < .001 for trend). Smaller increases occurred for TMP/SMX (37 vs 53 DOT per 1000 patient-days; P < .001 for trend) and linezolid (1 vs 27 DOT per 1000 patient-days; P < .001 for trend). Also consistent with the results observed for percentage use, there was a substantial decline in DOT for β-lactam antibiotics, from 33 to 18 DOT per 1000 patient-days (P < .001 for trend). Vancomycin use, however, increased from 152 to 204 DOT per 1000 patient-days during the study period (P < .001 for trend), an increase not apparent when calculated by percentage use.
Temporal Trends in Antibiotic Use Only for Patients With MRSA
When antibiotic selection was examined for the subset of patients with MRSA infections, generally similar trends emerged across both measures of antibiotic use (Fig 2). When measured by either percentage use or DOT, clindamycin use increased more than three-fold (19% to 67% for percentage use and 48–240 DOT per 1000 patient-days; both P < .001 for trend). Smaller increases occurred for TMP/SMX and linezolid (Fig 2). Although we observed a decline in percentage use of vancomycin from 53% in 1999 to 36% of patients in 2008 (P < .001 for trend), we observed an increase in DOT with vancomycin (204–248 per 1000 patient-days; P < .001 for trend). A significant decline occurred for β-lactam antibiotics (P < .001 for trend) with both measures of antibiotic use.
Antibiotic Use on the Basis of Site of Infection
Among the 64 813 children with a diagnosis of S aureus infection, 36 746 (57%) had an associated code indicating the type of S aureus infection. Although use of anti-MRSA antibiotics—clindamycin in particular—increased overall, we observed differential trends across the various types of S aureus infection. The most substantial increases in clindamycin use were for SSTIs (30% in 1999 vs 89% in 2008; P < .001 for trend) and osteomyelitis (17% in 1999 vs 72% in 2008; P < .001 for trend). Although linezolid was not available during the first 2 years of the study period, its use substantially increased between 2001 and 2008. Linezolid use in 2008 was relatively limited for SSTIs (3%) when compared with osteomyelitis (9%), endovascular infection (21%), and pneumonia (10%). Consistent with aggregated trends, use of β-lactams dropped significantly across all infection sites during the study period, the biggest drop seen in SSTIs (78% in 1999 vs 11% in 2008; P < .001 for trend; Table 2).
Epidemiologic studies of both children and adults have documented an increase in S aureus infections that largely were attributable to the emergence of CA-MRSA.11,18,–,20 This study is the first to evaluate the change in antibiotic use in hospitalized children with S aureus infections during the decade in which CA-MRSA emerged. Consistent with clinical recognition of this epidemic, we identified a dramatic increase in the use of MRSA-active agents—particularly clindamycin—in conjunction with a significant decrease in β-lactam use.
Although use of all MRSA-active antibiotics significantly increased during the study period, the use of clindamycin increased most dramatically, representing use in almost two-thirds of hospitalized patients with S aureus infection. Considering DOT per 1000 patient-days, clindamycin use for children with S aureus infection increased four-fold, becoming equivalent by this measure to that of vancomycin. In addition, increases in use of clindamycin across various sites of S aureus infection (osteomyelitis: six-fold increase; pneumonia: 1.5-fold increase; and endocarditis: two-fold increase) suggest that clinicians are becoming more comfortable prescribing clindamycin for serious invasive S aureus infections. Although clindamycin is not Food and Drug Administration–approved for these indications, pediatric studies have demonstrated effectiveness in treating osteomyelitis and pneumonia with this agent.21,22 Despite this expanding pattern of use, it is noteworthy that the American Heart Association discourages the use of clindamycin in the treatment of S aureus endocarditis.23 In our study, 87% of patients with documented MRSA endocarditis were treated with vancomycin in 2008.
The observed decrease in β-lactam use in the context of unchanged MSSA incidence rates suggests that another antibiotic—likely clindamycin—may be replacing β-lactams as the preferred empiric therapy for presumed S aureus infections. Although these data do not allow us specifically to distinguish between empiric and culture-directed therapies, it is probable that clindamycin is often the empiric therapy for both SSTIs and more invasive infections. Given this dramatic increase in clindamycin use, practitioners must become aware of its limitations. These include both constitutive and inducible resistance, each of which varies significantly by geography (with rates as high as 80% in some regional reports).17 Furthermore, this surge of antibiotic pressure has the potential to facilitate the development of resistance among circulating clindamycin-sensitive CA-MRSA strains; therefore, it is essential that clinicians be aware of and continue to monitor their institution's antibiograms. Unfortunately, a recent study that evaluated the use of institution-based antibiograms among house officers revealed that 60% never used this information, with the main reason being not knowing where to find it.24 It is imperative that infectious diseases physicians and microbiologists make this information readily available, because it will affect the future treatment of patients with S aureus infection. Ultimately, prospective studies that compare both empiric and definitive antibiotic therapy for MRSA infections in children are required to define the appropriate approach.
We evaluated temporal trends in antibiotic use by measuring both percentage use and DOT per 1000 patient-days. This combination of metrics allowed us to compute a more accurate description of antibiotic use within this setting. Percentage use evaluates how often the antibiotics are being prescribed irrespective of whether the choice is empiric or culture-directed. Although this provides a broad overview of antibiotic choices, percentage use ignores length of antibiotic therapy and, consequently, gives equal weight to empiric (often brief) and culture-directed (generally longer and more definitive) therapy. To account for variations in length of therapy, we also calculated DOT. This measure describes how many days a patient was on a given antibiotic, standardizing to 1000 patient-days. This approach not only addresses variable lengths of therapy but also provides a more accurate picture of use when compared with another common measure—the defined daily dose—especially in pediatric populations.25
The difference between these 2 measures is exemplified by the calculated vancomycin use in our study; although percentage use was unchanged over time, DOTs per 1000 patient-days increased substantially. Given that vancomycin is the empiric antibiotic of choice for patients with suspected severe S aureus infections, this difference may be reconciled by considering the epidemiology of MRSA: in the early study period, we speculate that patients were empirically treated with vancomycin but it was discontinued after the patient's culture and susceptibility results revealed MSSA; however, given that MRSA is more prevalent today, use of MRSA-active drugs such as vancomycin should increase as patients require continuation of this initial empiric therapy, leading to increases in vancomycin DOT.
A strength of this study is the ability to evaluate the antibiotic treatment of a large number of hospitalized children from across the United States with S aureus infections by using the PHIS database. The PHIS database provides national-level pediatric data, combining patient records from 43 large, tertiary care children's hospitals. PHIS provides up to 21 diagnosis codes per hospitalization, providing more diagnosis data per patient than other administrative data sets. In this study, only 25 of the CHCA hospitals were included because of lack of complete data from 18 of the hospitals. Although the hospitals excluded were similar in that they are freestanding children's hospitals, the exclusion of these hospitals led to a lack of significant representation from the Pacific Northwest and the Northeast. This is potentially important because the rates of MRSA vary on the basis of geographic region; however, a recent study that used PHIS hospitals with a more uniform geographic representation noted a similar increase in MRSA infection.11 Although it is possible that these areas are using antibiotics differently, anecdotal reports suggested that their prescribing practices are likely similar.
Despite the advantages of administrative data, they are limited by the possibility of miscoded information. The accuracy of ICD-9-CM codes for diagnosing either S aureus or MRSA infection has not been validated for this large population, and, although generally specific, ICD-9-CM code–based searches may not have ideal sensitivity, as evidenced by our identification of site codes for less than two-thirds of S aureus infections; therefore, this analysis might underrepresent the actual disease burden. Also, because we were unable to assess microbiologic data, we are limited in distinguishing empiric therapy from culture-directed regimens. Furthermore, we cannot verify that patients received medications for which bills were generated—the source of antibiotic use data. Finally, because the analysis was limited to freestanding children's hospitals, generalizing these findings to other inpatient pediatric settings (eg, general hospitals with pediatric wards) and outpatient settings may not be warranted.
Antibiotic prescribing for hospitalized children with S aureus infections has changed dramatically since the emergence of CA-MRSA, reflecting an increase in MRSA-active therapy coincident with a decrease in β-lactam use. Clindamycin has become the primary antibiotic for the treatment of S aureus infections, including both SSTIs and invasive diseases such as osteomyelitis and pneumonia. It is essential that clinicians be vigilant in monitoring antibiotic susceptibilities as well as patient outcomes to ensure appropriate empiric and culture-directed therapy for S aureus infections.
This study was supported by the Agency for Healthcare Research and Quality: Centers for Education and Research on Therapeutics grant U18-HS10399.
We thank Jeff Barnes and Matt Hall from the CHCA for assistance with the use of the PHIS data.
- Accepted February 22, 2010.
- Address correspondence to Jason G. Newland, MD, Children's Mercy Hospital and Clinics, Section of Infectious Diseases, 2401 Gillham Rd, Kansas City, MO 64108. E-mail:
FINANCIAL DISCLOSURE: The authors have indicated they have no financial relationships relevant to this article to disclose.
- SSTI =
- skin and soft tissue infection •
- MRSA =
- methicillin-resistant Staphylococcus aureus •
- CA-MRSA =
- community-associated MRSA •
- MSSA =
- methicillin-susceptible S aureus •
- TMP/SMX =
- trimethoprim/sulfamethoxazole •
- PHIS =
- Pediatric Health Information System •
- CHCA =
- Child Health Corporation of America •
- ICD-9-CM =
- International Classification of Diseases, Ninth Revision, Clinical Modification •
- DOT =
- days of therapy
- Deresinski S
- Moreno F,
- Crisp C,
- Jorgensen JH,
- Patterson JE
- Dietrich DW,
- Auld DB,
- Mermel LA
- Gerber JS,
- Coffin SE,
- Smathers SA,
- Zaoutis TE
- Miller LG,
- Perdreau-Remington F,
- Bayer AS,
- et al
- Schreckenberger PC,
- Ilendo E,
- Ristow KL
- Baddour LM,
- Wilson WR,
- Bayer AS,
- et al
- Polk RE,
- Fox C,
- Mahoney A,
- et al
- Copyright © 2010 by the American Academy of Pediatrics