JSTOR

Methods

We examine a population‐based approach to infection control, compared with an intervention‐based one, for controlling MRSA. In the former, the goal is to use team‐based infection control approaches¹⁷ to reduce the total rate of nosocomial infection by 50% or more. This is a reasonable goal, since a recent study in 103 ICUs in Michigan showed that this approach led to a 66% reduction in catheter‐related bloodstream infections.¹⁸

Bloodstream infections acquired in hospitals are the focus of this model, because of their high morbidity, mortality, and costs^19,20 and the remarkable ability to prevent them. Moreover, a recent report from the Centers for Disease Control and Prevention showed that 76% of hospital‐onset, invasive MRSA infections involve bacteremia.² Thus, it is an important focus. Nevertheless, the same arguments could be extended to healthcare‐associated pneumonias and surgical site infections.

Results

Mortality

In a hospital with 10,000 hospitalizations annually, an infection rate of 5%–10% equates to 500–1,000 infections. If 10% of the latter involve the bloodstream, then 50 to 100 bloodstream infections would be expected each year (Table). If 25% of patients with bloodstream infections die, that equates to 13–25 deaths from all nosocomial bloodstream infections and 2–4 deaths from those caused by MRSA. If attributable mortality is 50% of total (crude) mortality, then the numbers of attributable deaths are 7–13 from all bloodstream infections and 1–2 from MRSA bloodstream infections. If half of these lives can be saved, that equates to 4–7 lives saved annually with a population‐based program, compared with 1 life saved annually with a MRSA screening program.

Table.  Comparison of the Effects of Infection Control Approaches on Annual Rates of Bloodstream Infection (BSI) and Attributable Death, Assuming 10,000 Hospitalizations and 500–1,000 Patients With Nosocomial Infections (5%–10% Rate)

Open New Window

Extrapolating from 10,000 hospitalizations to the 35 million hospitalizations nationally, one can expect 10,937 to 21,875 lives to be saved annually with a population‐based approach, compared with 1,531–3,063 lives saved with a MRSA‐based approach. (Population‐based approach: 35,000,000 hospitalizations × 10% infection rate × 10% bloodstream infection rate × 25% crude mortality × 50% attributable proportion of crude mortality × 50% reduction in number of infections = 21,875 lives saved. MRSA‐based approach: 35,000,000 hospitalizations × 10% infection rate × 10% bloodstream infection rate × 14% MRSA infection rate × 25% crude mortality × 50% attributable proportion of crude mortality × 50% reduction in number of infections = 3,063 lives saved.) Even if all MRSA bloodstream infections were prevented by an approach targeted at a single drug‐resistant pathogen (ie, the MRSA‐subset approach), the attributable number of lives saved would be 6,124 at a maximum.

Years of Life Lost

In a similar fashion, if 10 years of life are lost for each death attributable to healthcare‐associated bloodstream infection, a population‐based infection control program could save 109,370 to 218,750 years of life annually, compared with 30,620 to 61,240 saved with an approach targeted at a single drug‐resistant pathogen (Figure 1). Even if all MRSA bloodstream infections were prevented—either in a 5% or a 10% total nosocomial infection rate scenario—the total years of life saved would still be less than that achieved with a population‐based approach.

(18 KB)

Figure 1.  Estimate of the years of life saved in the United States with use of an alternative infection control approach, assuming that 50% of bloodstream infections are prevented. The model assumes 10,937–21,875 lives saved in the population‐based approach, 1,531–3,063 lives saved in the methicillin‐resistant Staphylococcus aureus (MRSA)–subset approach, and 10 years of life lost for each attributable death.

Open New Window

Attributable Cost

If one examines cost and assumes a $20,000 or $40,000 attributable cost for a hospital‐acquired bloodstream infection, the costs avoided with a population‐based program range from $1.75 billion to $7.0 billion. In contrast, an approach targeted at a single drug‐resistant pathogen would save from $245 million to $980 million (Figure 2).

(25 KB)

Figure 2.  Estimate of financial savings in the United States expected from use of alternative infection control approaches, assuming that 50% of bloodstream infections are prevented. The model assumes 35 million hospitalizations, a 5%–10% nosocomial infection rate (1.75 million–3.5 million infections), and that 10% of the infections are bloodstream infections (175,000–350,000 bloodstream infections).

Open New Window

Secular Trends in Resistance

In the United States, increasing recognition of MRSA as a hospital pathogen occurred in the 1980s. The proportion of all S. aureus isolates resistant to methicillin rose from approximately 30% in 1990 to more than 65% in 2008 (Figure 3).

(25 KB)

Figure 3.  Timeline showing the progression in the estimated prevalence in the United States of antimicrobial resistance for selected pathogen‐antimicrobial pairs.^24‐34

Open New Window

What began to be recognized in the 1990s was a group of new antibiotic‐resistant pathogens (listed here with current estimates of the proportion of isolates that are antibiotic resistant): enterococci resistant to vancomycin (30%), Acinetobacter baumannii strains resistant to imipenem (30%), P. aeruginosa resistant to imipenem (30%), and Candida species resistant to fluconazole (10%)^25‐34 (S. Fridkin, Centers for Disease Control, personal communication, January 2008) (Figure 3).

Discussion

In the late 1970s, one of the authors witnessed the explosive outbreak of MRSA infection in a university hospital. Within a year, the proportion of all nosocomial S. aureus infections that were caused by methicillin‐resistant isolates was 38% for postoperative wound infections, 31% for pulmonary infections, and 24% for bloodstream infections.³⁵ In response to the epidemic, 3 infection control measures were introduced: daily clinical laboratory surveillance, monthly prospective microbiological surveys of patients at high risk, and identification of previously infected or hospitalized patients.³⁶ All MRSA‐positive patients were placed under contact precautions. In the following year, the monthly prevalence decreased steadily from 33 to 6 patients, and, within that group, the number with new cases decreased from 20 patients to 1 patient per month. To our knowledge, this was the first study to show the effectiveness of periodic and selective screening for MRSA carriers and isolation of infected and colonized patients during an outbreak in the hospital.

Subsequently, use of a topical antibiotic with activity against both methicillin‐susceptible S. aureus and MRSA showed remarkable ability to reduce or eliminate carriage.^37‐39 Furthermore, the Dutch program for controlling MRSA built upon these observations and showed that the isolation of all patients who tested positive for MRSA and treatment of medical personnel who were carriers were associated with a very low prevalence of this organism in the country’s hospitals.⁴⁰ In the United States, similar improvements in the control of MRSA were observed after a screening and isolation intervention.⁴¹

Proponents of MRSA screening and isolation have subsequently advocated strongly for legal and administrative mandates for such interventions in all hospitals.⁴² The debate about their value continues in part because of the limited reports of success so far, some reports of the failure of screening,^43,44 the costs of screening and isolation, the unwanted side effects of patient isolation, and the inability to find sufficient isolation rooms in some older hospitals with many 2‐bed patient rooms.^43,45 We add the concern that methicillin‐susceptible S. aureus still constitutes an important infection control challenge in US hospitals. A recent observational study of universal surveillance for MRSA reported a decline in numbers of MRSA infections at each body site but no change in the number of cases of bacteremia due to methicillin‐susceptible S. aureus.⁵

Two recent hospital‐based studies show that 5%–6% of MRSA screening test results are positive.^5,43 Screening and isolation may have adverse consequences, as noted above. Imagine a screening test with a sensitivity and specificity each of 95% or another with a sensitivity and specificity each of 97%. Even in a hospital with a 10% prevalence of MRSA, the positive predictive value of the former test would be 68% and of the latter 78%. Thus, 32% or 22% of patients isolated, respectively, would not even have MRSA colonization or infection, yet they would be subject to the adverse consequences of isolation.

Strong proponents for a nasal screening program to identify MRSA carriers may argue that they also advocate for additional infection control measures. So far, however, no such hospital showing a reduced rate of MRSA infections has reported a 50% reduction in the total infection rate. In contrast, the population‐based approach that focuses on process measures (checklists) was associated with a 66% reduction in the number of all catheter‐related bloodstream infections in 103 critical care units in the state of Michigan.¹⁸ Recently, at the Medical College of Virginia Hospital, data on the rates of device‐related MRSA infections obtained over a 4‐year period showed statistically significant declines achieved by means of evidence‐based interventions without use of active surveillance cultures for MRSA.⁴⁶ In the neuroscience, medical, and surgical ICUs, the rates of device‐related bloodstream infection, of urinary infection, and of ventilator‐associated pneumonia decreased more than 40% in each unit, and the rate of MRSA infection decreased more than 48% in each unit.

We argue that the vocal advocates for an approach focusing on a single drug‐resistant pathogen have failed to appreciate the increasing complexity of the problem of drug‐resistant pathogens in hospitals (Figure 3). These include the almost uniform methicillin resistance among coagulase‐negative staphylococci (thought to be the source of the mec gene’s original transfer to S. aureus to create MRSA⁴⁷), the 30% rate of vancomycin resistance among enterococci,⁴⁸ the 10% rate of resistance to first‐generation triazoles among Candida isolates, and the increasing resistance among gram‐negative rods to imipenem and sometimes to all drugs except colistimethate sodium.⁴⁹ One could imagine an advocacy group proposing the screening of all patients for gastrointestinal carriage of vancomycin‐resistant enterococci or for Candida species, because vancomycin‐resistant enterococci currently ranks higher than MRSA as a cause of bloodstream infections¹⁴; and for both vancomycin‐resistant enterococci and for Candida species, the crude and attributable mortalities are greater than those associated with MRSA bloodstream infections.^50,51 We would make the same arguments against an intervention focused on vancomycin‐resistant enterococci or on Candida species as we have against a MRSA‐only intervention.

Nasal carriage of MRSA has been recognized as a source of subsequent bacteremia,⁵² and a program to identify and decolonize such patients is logical. In studies of MRSA carriage, it has been noted that the organism was found in the throat in only 13%–15% of carriers.^53,54 Thus, a substantial proportion of true carriers are misclassified as noncarriers in a nasal screening program for MRSA. Moreover, as community‐associated MRSA strains become more prominent among healthcare‐associated pathogens, newer and more complex issues will face hospital epidemiologists. The community‐associated MRSA strains are thought to be more virulent than traditional hospital‐associated MRSA strains, and fewer of the colonized patients have nasal carriage than is seen in the traditional strains. Currently, strain USA 300 has spread quickly in the United States in a clonal fashion, and its ability to be transmitted person‐to‐person cannot be questioned. However, in a prospective cohort of 51 patients infected with community‐associated MRSA, only 41% were shown to be nasal carriers.⁵⁵ Recently, in a study of heterosexual transmissions of community‐associated MRSA, only 1 of 5 patients tested was also a nasal carrier.⁵⁶ The public health perspective is that failure to identify standard MRSA or the new community‐associated MRSA in a screening program could lead to a seriously false sense of security with poor outcomes. A population‐based intervention program obviates the issue of false‐negative nasal cultures.

There is some confusion about the specific components of an active surveillance program for MRSA. The Society for Healthcare Epidemiology of America Guideline,⁵⁷ which is cited as the reference for policy advocates in developing statewide culturing mandates, addresses the identification of every colonized patient, with subsequent isolation to prevent transmission. No mention is made of treatment of carriers. However, 2 hospital‐based studies suggest that the carriers have an approximately 25% risk of MRSA infection in the year following their identification as carriers.^58,59 We think that there is value in very selective MRSA screening programs⁶⁰ followed by topical antibiotic prophylaxis for patients at high risk for whom infection would carry devastating consequences; for example, patients undergoing open heart surgery. Such a focus would prevent the most serious infections while limiting exposure to mupirocin and minimizing infrastructure costs and selection for resistance of topical antibiotic agents. The critical patient safety question is this: on top of a platform of an effective infection control program—designed to reduce the total infection rate by at least 50%—what is the incremental benefit of an MRSA screening program?

Our perspective is that, in the 1980s and into the 1990s, the single focus on MRSA was reasonable, because the antibiotics available to treat other key pathogens were not so limited. However, with the increasing rates of resistance among key pathogens between the 1980s and 2008 and the limited number of antibiotics in the development pipeline,⁶¹ a population‐based infection control program becomes more important. Added bits of new information since the 1980s and 1990s are the remarkable increases in the rates of handwashing compliance,^62,63 the effective team‐based infection control programs,^17,46 the public’s demand for public reporting of all infection rates,⁶⁴ and the proliferation of demands for improved targets for infection control by the Centers for Medicare and Medicaid Services, the Joint Commission, and industry.

Our very simple model shows that reducing the overall rate of bloodstream infections by 12.5% is equivalent to reducing the rate of MRSA bloodstream infections by 50%. If bloodstream infection rates could be reduced by 25%, that would be equivalent to eliminating all MRSA bloodstream infections. The 25% figure can be readily achieved,^18,46 and will take substantially fewer resources than an MRSA‐only approach.

We argue, furthermore, that a population‐based infection control intervention program utilizing evidence‐based processes can be applied to favorably influence rates of infection at anatomic sites other than the bloodstream in both critical care units and on general wards. Unfortunately, some institutions are implementing wide‐scale MRSA surveillance programs without increasing the numbers of infection control personnel, diverting attention and resources from the platform of an effective program (eg, surveillance and evidence‐based interventions).

In our model, we have examined only hospital‐acquired bloodstream infections and compared outcomes of alternative interventions, assuming 50% success with either a MRSA screening approach or a total population‐based approach. We also assume that a MRSA screening approach would reduce the incidence of MRSA‐associated pneumonia and surgical site infection by 50%. Likewise, a population‐based approach would decrease by 50% the incidence of all types of hospital‐acquired pneumonia and all types of surgical site infection (approximately 15% of each are caused by MRSA). Since MRSA causes a minority of infections at each of these sites, our arguments would hold in the examination of infections at sites other than the bloodstream, adding value to the epidemiologically sound method we advocate.