Methicillin-resistant Staphylococcus aureus (MRSA) is cross-transmitted in hospital settings, and has a high impact not only on patient morbidity and mortality but also on hospitalization costs. Worldwide, it has been endemic in many healthcare facilities since the 1990 s [1]. MRSA remains a major pathogen in nosocomial infections in developing countries [2] and in Latin America, according to SENTRY [3].

The fact that a patient can harbor MRSA at hospital admission has consequences not just for the choice of patient treatment: it also impacts on the effectiveness of infection control in the hospital. MRSA reservoir at hospital can make other measures of infection control not as effective, thereby causing pathogen transmission to continue [4]. Guidelines aimed at controlling the spread of MRSA therefore propose to systematically search for colonized patients, and then to isolate and decolonize them [5]. This policy has not been tested rigorously in methodologically-sound randomized trials, and most of the evidence comes from observational and quasi-experimental studies [6]. There is also concern about the cost of such measures and the lack of available rooms for isolation [7].

Recently-published studies have arrived at different results regarding identification and isolation of MRSA colonized patients. These differences may be attributable to many factors, ranging from differences between the settings and patients to methodological issues, and the multifaceted nature of infection control practices. In one study, universal MRSA surveillance reduced the infection risk during hospitalization and 30 days after discharge [8], while another study evaluated MRSA screening at admission in surgical patients in an endemic setting of MRSA, and found no decrease in surgical site infections and nosocomial acquisition of MRSA [9]. Yet another study in an intensive care unit (ICU) compared two interventions to reduce transmission of MRSA, after identification of colonized patients by the pathogen: cohort-isolation o single room isolation, and found no difference in cross-transmission between the two periods [10].

In Brazil, MRSA is endemic in hospitals [3], but few studies have looked at MRSA colonization. One study that screened patients in an emergency department found 0.7% of patients colonized with MRSA at admission [11]. In ICUs, one study reported 13% of patients colonized at admission [12], and another [13] found 46% patients colonized. In this ICU, 51% of non-colonized patients acquired MRSA colonization [13].

We agree that the identification of baseline rates and the associated risk factors of colonization are necessary to estimate the burden of colonization and the demands for isolation facilities [14], and to help adopt a cost-effective strategy of patient screening which would take into consideration the population prevalence of MRSA and the structure of the hospital [15]. Thus, use of own data, obtained from a study that uses local resources can help to propose a policy that fits the institutional needs and lead to satisfactory outcomes regarding MRSA control.

The present study aims to estimate the prevalence of MRSA colonization and infection in clinical adult and pediatric patients at the time of admission to hospital; the incidence of colonization and infection during hospitalization; and the potential risk factors for both, in a hospital in southern Brazil, in order to obtain information to support infection control planning for MRSA policy in the institution.

From May, 2006 to March, 2007, 1,561 patients met the inclusion criteria, and 580 were selected (Figure 1). The final sample consisted of 473 patients (297 adults, 176 children). For incidence analysis, we followed 276 patients who had at least a second swab during hospitalization, and who had a negative culture at admission. There were 5,414 days of hospitalization. The first swabs were collected within 24 hours of admission for 377 patients, up 48 hours after admission for 95 patients and up 72 hours of admission for one patient.

Sixteen adult patients had been colonized with MRSA at admission (5.4%; 95%CI, 3.1% to 8.6%), and two were infected as shown by blood and skin ulcer secretion. Three (1.7%; 95%CI 0.4% to 4.9%) pediatric patients were colonized and one was infected (sputum) at admission. Weighted prevalence of MRSA at admission for the whole sample was 5.3% (95%CI, 4.5 to 6.2%). None of the patients with clinical cultures positive for MRSA at admission had positive nasal swabs, and none of the colonized patients developed a clinical infection with MRSA while hospitalized, although one of them received empirical vancomycin.

Weighted incidence of MRSA was 9.6% (95%CI 8.1 to 11.1) for the patients followed during hospitalization. Twenty adults acquired colonization during hospitalization (incidence rate, 5.5/1,000 patient-days, 95%CI, 3.4 to 8.5), and two children became colonized (incidence rate, 1.1/1,000 patient-days (95%CI, 0.1 to 4.0).

The incidence of colonization increased with increase length of stay, and was statistically significant (P < 0.001). It reached 2.9% in week two, 4.9% in week three, 2.4% in week four, 8.9% in week five, zero in week six, 9.1% in week seven and zero in week eight. For those patients with negative cultures at admission and have acquired colonization, the median number of days from admission until positive cultures were obtained from children was 8 days, and 15 days for adults. Adult patients with positive screening cultures had spent a mean of 4.05 days (median, 3) in the emergency, while patients with negative cultures, 2.72 days (median, 2), (P = 0.147 for the comparison, using a Mann-Whitney test).

We identified the genotype of 36 of the 41 patients with positive swabs. For the positive patients at admission, the SCCmecA genes present were type I in five patients and type III in eight. For the patients that acquired colonization, there were SCCmecA type I in seven patients, SCCmecA type III in twelve and SCCmecA type IV in two.

Our study found a prevalence of MRSA colonization at admission that is higher than that previous reported in patients admitted to emergency service in Brazil [11], and may reflect the chronic nature of patients in a tertiary care hospital. Sixty-four percent of our sample had had at least one hospitalization in the previous year, and this was associated with colonization at admission. The results of SCCmecA, which mainly disclosed those types associated with hospital strains, also confirmed this profile. Moreover, the two strains with SCCmecA type IV acquired after admission are in agreement with a study that reported hospital dissemination of CA-MRSA in hospitals in Detroit [23].

However, even thought 70% of all pediatric patients had been admitted to a hospital in the last year, 19% of them with a neoplastic disease, the prevalence of colonization proved to be low in children. In Brazil, another study also reported low prevalence of MRSA in hospitalized children younger than five years (1%), and all strains harbored SCCmecA type III [24]. The number of positive children was too small for performing a multivariate analysis.

The two risk factors found for colonization at admission in our adult patients have been reported elsewhere in the literature [25]. However, there are also other risk factors associated with MRSA colonization that have been reported, but we found no association with HIV infection or with treatment for hemodyalisis. This may be due to the small number of those patients in the population under study, and the consequent lack of power to identify such associations, which is a limitation of our study. Another limitation may be related to the lack of association between previous drug use (i.e. antibiotic, corticosteroids or other immunosupressors) and MRSA harboring. This may reflect an information bias because most patients in the study were not able to report the name or type of the drugs that they had been using in the recent past.

The high incidence of colonization between hospitalized patients is a matter of concern. Two other studies in Brazil have also shown a high incidence rate of colonization [12, 13], although they were in ICU settings, where patients are more exposed to invasive procedures (intravenous catheters, urinary catheters and mechanical ventilation) with higher rates of colonization. We could not identify any particular risk factor for this acquisition, but we followed a heterogeneous sample of patients, and the CCS may not be the most appropriate tool for use in studies with resistant pathogens [26]. The proportion of patients in study which became colonized each week increased and this is in accordance with the fact that patients who remain longer in hospital are exposed to more interventions (eg, by invasive procedures, contacts with healthcare workers, antimicrobials use) which may raise the risk of acquisition of a pathogen.

The number of colonized patients may have been underestimated since screening used only one site, and this is a limitation of the study. However, we wished to estimate how much work was involved in screening patients in a pragmatic study, in which the primary objective was to characterize the incidence of colonization by MRSA in an environment with high endemic levels. The lack of resources for a comprehensive strategy of universal MRSA screening demands a targeted approach that could result in a reduction of MRSA transmission, using the hospital capacity. Also, we did not adjusted for current use of prophylactic antibiotics used by oncologic patients (mainly, trimethropim-sulfamethozaxole), which may have affected colonization. Indeed, some patients received vancomycin because of suspected MRSA infection; this also could have biased the results of the nasal cultures, but the use of antimicrobial was not associated with colonization.

Almost 40% of the original sample that tested negative at admission was discharged at the weekend, without a second swab. This fact could have overestimated the incidence of colonization, because patients that had a shorter stay were less ill than the other patients, and have had a lower risk of becoming colonized.

Finally, if we consider the risk factors for adults at the time of admission, our findings suggest that the hospital should be able to screen 78% of those patients being admitted (all adult patients with admission in previous year and/or aged 60 years or more). This appears to be a high percentage if we consider isolating the patients while waiting the results of the cultures, as it would put them in isolation for 24% of their days in hospital. But if we look at the incidence rate found (5.5/1,000 patient-days), this measure could help lessen colonization pressure in the hospital. Another important point is that these newly-colonized patients would probably be readmitted, so maintaining the chain of MRSA transmission.

We also consider that this high prevalence of MRSA colonization at admission is a matter for concern, since the hospital emergency department is overcrowded and in this population the median time spent in the emergency department was two days before admission to an acute-care ward. Another Brazilian study has found that patients colonized or infected with multidrug-resistant bacterial organisms had a higher length of stay than controls in the ER, and a higher associated mortality [27]. The presence of colonized patients in a chronically overcrowded and understaffed ER may represent another substantial obstacle to implementing hospital infection control measures [28], especially where resources are limited, and thus requiring specifically customized strategies.

Although MRSA infection is a very frequent pathogen in Brazilian hospitals, [2, 3], there are few data regarding patient colonization [11–13]. Even though the universal screening is not a resolved issue, knowledge of the magnitude of colonization is invaluable for planning to control the pathogen. In a heterogeneous sample of clinical patients admitted to a Brazilian hospital, a high prevalence of MRSA colonization was found, and the results of the study provide the basis for better-targeted MRSA control policies aimed at specific groups of patients. The study also gives an estimate of the resources needed for a particular hospital to implement such interventions, especially where funds are limited.