Methods

Study design and hospitals. This cohort study was conducted at Duke University Medical Center (DUMC) and at 8 community hospitals in Virginia and North Carolina that are members of the Duke Infection Control Outreach Network (DICON). DUMC is a tertiary care hospital in Durham, North Carolina, with 750 beds, including 138 intensive care unit beds. The annual number of admissions to DUMC is 37,793, and DUMC has 7535 staff members. DUMC serves a community of 210,553 persons and is also a tertiary care referral center with a large catchments area. DICON is a network of community hospitals located primarily in the Southeastern United States. At present, DICON includes 36 hospitals. The 8 participating community hospitals included a total of 2037 beds (mean number of beds, 255; range, 102–537) and a total of 180 intensive care unit beds (mean number of beds, 22.5; range, 8–97). The mean number of annual admissions to these hospitals is 12,234 (median number of hospital admissions, 10,327; range, 5380–30,001), and the mean number of hospital staff members is 1530 (median number of hospital staff members, 1244; range, 633–4282). These community hospitals serve an average population of 90,090 (median population, 38,344; range, 4403–276,093). Only 1 of these hospitals had infectious diseases consultants during the study period, and 3 are teaching hospitals. The structure and function of DICON has been described elsewhere [7, 8].

Surveillance methods for MRSA were similar among study hospitals. Patients with ⩾1 culture positive for MRSA were prospectively identified. Prospectively recorded variables included patient medical record number, location at time of culture, anatomic site of culture, and date of culture. Prospective surveillance for MRSA has been ongoing at DUMC since 1994. Thus, patients included in this study were hospitalized at DUMC from 1994 through 2002. Data from the 8 DICON hospitals were available from 1 January 1999 through 31 March 2003.

Informed consent was waived by the institutional review boards of participating hospitals. Data were de-identified and managed according to the standards required by the institutional review boards of participating hospitals.

Cohort definition. Patients were included if they had either a bloodstream infection (BSI) or a deep or organ space surgical site infection (SSI) due to MRSA [9]. If patients had a history of BSI or SSI due to MRSA prior to the study period, they were excluded. The study included pediatric and adult patients.

Variables. After cases were prospectively identified and entered into study databases, additional variables were retrospectively collected from the patient chart. Data were collected on demographic characteristics, comorbid conditions (including Charlson score [10]), acute severity of illness (McCabe score [11]), dates of hospitalization and time spent in the intensive care unit, presence of indwelling devices, functional status (evaluated according to the Katz criteria [12]), surgical procedures performed, and antimicrobial therapy received in the hospital. BSIs were classified as community acquired, health care associated, or nosocomial [13]. BSIs were defined as primary (e.g., catheter associated) or secondary (arising from an infection at another anatomic site) [9]. SSIs were defined according to standard Centers for Disease Control and Prevention definitions [9]. The day of diagnosis for BSIs and SSIs was defined as the day on which the first blood sample or operative culture specimen was obtained that subsequently had results positive for MRSA.

Data were also collected on mortality during the year following infection, duration of hospitalization, and readmissions to the hospital during the 90-day period following infection. Outcome data were obtained from the patient chart and the Social Security Death Index [14].

Statistical analysis. All analyses were performed using SAS, version 8.2 (SAS). Bivariable analysis was performed using the Student's t test or Wilcoxon rank-sum test for continuous variables. Dichotomous and ordinal variables were analyzed using the Fischer's exact test or χ² test. Multivariable analyses were performed using logistic regression.

Results

A total of 693 patients with invasive infections due to MRSA were identified and included in the cohort: 129 patients had SSI (classified as deep or organ space), and 564 had BSI. The mean age of the cohort was 63.1 years. Approximately one-half of these patients (350 patients; 50.5%) were male, and 382 (55.1%) were white.

SSI

Demographic characteristics and comorbid conditions of the SSI cohort. Patients with SSI had a mean age (±SD) of 62.0±15.3 years, 86 (67.7%) were white, and 60 (46.5%) were male (table 1). Almost one-half of patients underwent surgery in a community hospital, and more than three-quarters were admitted to the hospital from home. Common comorbid conditions included diabetes mellitus, heart disease, and chronic obstructive pulmonary disease. More than 30% of patients required assistance with ⩾1 activity of daily living.

View larger version:

• In this window
• In a new window

• Download as PowerPoint Slide

Table 1

Summary of demographic characteristics, hospital admission data, and perioperative data for 129 patients with methicillin-resistant Staphylococcus aureus (MRSA) surgical site infection (SSI) and 564 patients with MRSA bloodstream infection (BSI).

Perioperative variables. The most common types of surgery were coronary artery bypass grafting, open-reduction internal fixation of fractures, and hip prosthesis insertion (table 1). Almost one-half of the operative procedures were orthopedic (64 procedures). The majority of patients had a preoperative American Society of Anesthesiologists score >3, and the duration of surgery was greater than the National Nosocomial Infection Surveillance 75th percentile cut off point in almost one-half of patients.

Variables related to SSI. Patients were diagnosed with SSI a mean of 39.1 days after their initial surgical procedure (table 2). Eight percent of patients were cared for in the intensive care unit during the 7 days before receiving a diagnosis of SSI.

View larger version:

• In this window
• In a new window

• Download as PowerPoint Slide

Table 2

Infection and therapeutic data for 129 patients with methicillin-resistant Staphylococcus aureus (MRSA) surgical site infection (SSI) and 564 patients with MRSA bloodstream infection (BSI).

Antibiotic therapy variables. Vancomycin was administered to 57 patients (44.2%) on the day that an SSI was first diagnosed (table 2). Other antibiotics prescribed on the day of diagnosis of SSI due to MRSA are described in table 2. By day 7 after diagnosis, only 94 patients (72.9%) were receiving vancomycin.

Outcomes. Only 3 patients with SSI due to MRSA (2%) died before hospital discharge (table 3). Most patients were discharged to home (47 patients; 37.9%).

View larger version:

• In this window
• In a new window

• Download as PowerPoint Slide

Table 3

Clinical and microbiologic outcomes for 129 patients with methicillin-resistant Staphylococcus aureus (MRSA) surgical site infection (SSI) and 564 patients with MRSA bloodstream infection (BSI).

Approximately three-fourths of patients (93 patients; 73.8%) with an SSI due to MRSA were readmitted to the hospital within 90 days after their initial surgical procedure (table 3). The mean number of days that patients spent in the hospital after they developed SSI was 22.3 days. The 1-year mortality rate among patients with SSI due to MRSA was 21.7%.

SSI cohort, comparing patients in community and tertiary care hospitals. Compared with patients cared for in community hospitals (n=55), patients cared for in the tertiary care hospital (n=74) were more likely to have a McCabe score classified as rapidly fatal (0 [0%] of the community hospital patients vs. 3 [7%] of the tertiary care hospital patients; P<.001), and they were more likely to be admitted from another institution, as opposed to admitted from home (6 [10%] of the community hospital patients vs. 23 [31%] of the tertiary care hospital patients). Other variables were similar among patients in community and tertiary care hospitals.

In community hospitals, however, appropriate empirical antimicrobial treatment that included coverage for MRSA was provided to only 10 patients (19%). The rate of administration of appropriate empirical therapy was higher at the tertiary care hospital (47 patients [64%]; OR, 7.83; 95% CI, 3.40–18.0; P<.001), but almost 1 in 3 patients with invasive infection due to MRSA in this setting did not receive appropriate initial empirical therapy. In addition, on day 7 of infection, only 33 patients (59%) in community hospitals were receiving appropriate therapy (excluding deaths). The frequency of administration of appropriate therapy on day 7 at the tertiary care hospital was 84% (62 patients) and was significantly greater than the rate at community hospitals (OR, 3.44; 95% CI, 1.52–7.82; P=.003). Despite these treatment differences, outcomes were similar among patients cared for in community and tertiary care hospitals (table 4).

View larger version:

• In this window
• In a new window

• Download as PowerPoint Slide

Table 4

Treatment and outcomes for patients hospitalized at community hospitals, compared with patients hospitalized at a tertiary care hospital.

Discussion

The high toll that MRSA takes on patients cared for in tertiary care hospitals is generally recognized as a major problem and challenge by health care providers [4, 15, 16]. However, few if any data exist pertaining to MRSA in community hospitals. Thus, the impact of invasive infections due MRSA and their treatment has received insufficient attention. Our data show that MRSA is now as severe a problem in community hospitals as it is in our tertiary care university hospital.

In this study, empirical treatment choices for patients with invasive infections due to MRSA and mortality rates due to these infections were poor in all study settings. Antimicrobial therapy and outcomes were particularly poor among patients cared for at community hospitals. Approximately 40% of patients with SSI and/or BSI in community hospitals (123 patients; 41%) did not receive an antimicrobial agent active against MRSA during the first 7 days after their infection was diagnosed. Patients with BSI due to MRSA who were cared for in community hospitals were >2-fold less likely to receive an agent active against MRSA at 7 days after diagnosis, compared with patients in the tertiary care hospital (OR, 2.07; 95% CI, 1.45–2.97). Patients cared for at community hospitals were less frequently discharged to home and had higher rates of 1-year mortality, compared with patients at the tertiary care hospital. However, compared with patients who were cared for in the tertiary care hospital, patients treated in community hospitals had poorer functional status, were more likely to be incontinent of urine, and were more likely to have dementia. The presence of these patient characteristics might explain why patients treated in community hospitals had higher mortality rates.

The explanation for the high frequency of administration of inappropriate antimicrobial therapy for BSI and SSI due to MRSA is unclear. The high rate of administration of inappropriate treatment for BSIs due to MRSA (which occurred in more than one-third of study patients) was alarming. In a secondary analysis, we were unable to identify significant predictors of receipt of inappropriate therapy. We believe that these results represent a significant lack of understanding, particularly in community hospitals, regarding 3 important issues: the emergence of MRSA as a community and SSI pathogen, the importance of administering early effective therapy, and the need for effective, culture-based, and prolonged (i.e., administered for at least 2 weeks) systemic therapy for serious MRSA infections. Health care providers must become more familiar with patients who present with invasive “community-onset” infection due to MRSA and must provide early, optimal treatment modalities.

This study was limited by only including a single tertiary care hospital. Referral bias may have existed in the tertiary care hospital cohort. Similarly, severely ill or moribund patients may have been “allowed to die” and purposefully not given appropriate therapy following MRSA infection. We were unable to ascertain data regarding these practices in our cohort. The proportions of patients with a McCabe score of 1 (indicating the presence of a rapidly fatal condition), however, were similar at the community and tertiary hospitals.

This study had several strengths. It included a larger multicenter cohort that included many patients who were treated in community hospitals. Several measures of illness severity were analyzed. All patients were prospectively identified as having BSI or SSI. Finally, data were collected on both short-term and long-term outcomes.

Although empirical therapy was not provided for MRSA in many study patients, a particularly concerning finding was that >30% of patients did not receive effective antimicrobial therapy for MRSA by day 7 of hospitalization. Additional well-designed studies are needed to examine the relationship between the timing of effective antimicrobial treatment and outcome of invasive MRSA infections. In addition, further study is necessary to better elucidate why so many patients with invasive infection due to MRSA receive inadequate antimicrobial therapy. Interventions aimed toward improving the efficacy of antimicrobial therapy should be designed, trialed, and implemented.

The evolving epidemic of community-onset and community-acquired MRSA will likely greatly increase the number of patients with MRSA infection in community hospitals and simultaneously increase the clinical challenge presented by these infections [17]. Thus, ongoing epidemiologic study of trends in the incidence of serious MRSA infection in community hospitals is important. If MRSA becomes a common cause (if not the most common cause) of invasive SSI and community-based wound infections, further study of antimicrobial therapy will be needed. Community hospitals can no longer rely on microbiological surveillance data from large tertiary care hospitals, because these reports might be misleading. Hospitals must rely on local surveillance and therapeutic data and feed these data back to local physicians. Local knowledge of MRSA trends, associated outcomes, and therapeutic practices should greatly impact management decisions.

Finally, the focus of national quality-improvement projects on the prevention and control of infections due to MRSA [18] should also include measures as to whether appropriate empirical therapy is administered to patients with invasive infections due to MRSA.