Additional risk factors for infection by multidrug-resistant pathogens in healthcare-associated infection: a large cohort study

Traditionally, infections have been classified as community or hospital-acquired, according to their place of acquisition, and this classification is still used to guide diagnosis and treatment of infections [1, 2]. However, an increasing number of patients reside in nursing homes, the use of aggressive medical therapies (intravenous therapy, wound dressing) at home is more common, an increasing number of invasive therapies (hemodialysis, chemotherapy, radiotherapy) are performed in outpatient clinics, and there is a greater population of older patients, with more chronic diseases and frequent utilization of medical resources. This has led to the creation of a new group among the traditional classification of infections, termed “healthcare-associated infections” (HCAI).

In 2002, Deborah Friedman and colleagues [3] proposed a definition of HCAI including the above subgroups of patients, but despite being widely used in clinical studies [4‐9], there is a lack of consensus regarding risk factors, and more recent studies have included additional risk factors such as an immunocompromised state, hospitalization in the previous year and prior antibiotic therapy [5, 10]. Unfortunately, most of the studies performed using the HCAI classification have been restricted to respiratory infections [4, 5, 8, 10], bloodstream infections [3, 6, 11] or a single pathogenic agent [7, 9, 12], creating the need to widen the study of risk factors for infection by multidrug-resistant (MDR) pathogens to all infected patients hospitalized from the community.

The objective of this study was to identify additional risk factors associated with infection by MDR pathogens in all infected patients admitted from the community into the hospital setting (patients with community-acquired and HCAI, including all foci of infection and associated pathogens), and not included in the adopted definition of HCAI [3].

During the study period a total of 3733 patients were assessed and 1035 (28%) met the inclusion criteria of having infection according to the CDC definitions of infection. The mean ± SD age of the patients included was 65 ± 20 years, and the mean SAPS II was 29 ± 13; 49% of patients were male, and the overall hospital mortality rate was 13% (Table 1). Of these, 718 were admitted to hospital from the community with infection: 493 (48%) with a CAI and 225 (22%) with a HCAI. These populations were significantly different from patients with HAI, with a lower proportion of immunosuppressed patients (19% vs 27%, p = 0.007), a lower proportion of hospitalizations in the previous year (33% vs 55%, p < 0.001) and less likelihood of antibiotic therapy in the last month (19% vs 73%, p < 0.001). Hospital mortality was significantly lower in patients from the community (CAI or HCAI) than in those with HAI (11% vs 19%, p < 0.001) (Table 1). Of the 718 patients with CAI or HCAI, 439 (61%) had microbiologic documentation; of these, 123 were MDR (28%), 104 (85%) with MDR-GN and 41 (33%) with ESKAPE organisms.

From the univariate analysis, age, previous antibiotic therapy, hospitalization in the previous year (from 4 months to 1 year), chronic heart failure, chronic respiratory disease, atherosclerosis and a Karnofsky index <70 were selected for multivariate analysis of independent risk factors associated with infection by MDR pathogens, including infection by MDR-GN (Table 2). For the ESKAPE group, previous antibiotic therapy, hospitalization in the previous year, and a Karnofsky index <70 were the variables selected for the multivariate model (Table 2).

Regarding infection by MDR pathogens, the final model retained age (adjusted OR per category = 1.7, p = 0.001) and hospitalization in the previous year (from 4 months to 1 year) (adjusted OR = 2.0, p = 0.008) (Table 3) as independent risk factors. The final model for MDR-GN infection also retained age (adjusted OR per category = 1.5, p = 0.009) and hospitalization in the previous year (from 4 months to 1 year) (adjusted OR = 1.7, p = 0.048) (Table 3). For ESKAPE pathogen infection, the final model retained previous antibiotic therapy (adjusted OR = 7.2, p < 0.001) and Karnofsky index <70 (adjusted OR = 3.7, p = 0.003) (Table 3).

Patients with infection by a pathogen belonging to one of the three MDR groups had a significantly higher rate of inadequate antibiotic therapy: 16% in the study population, 46% in the group of patients with infection by MDR pathogen, 44% in the group with infection by MDR-GN, and 61% in the group with an ESKAPE group infection (p < 0.001). Despite that, median hospital length of stay was similar in all groups, as were SAPS II and hospital mortality rate (Table 4).

In this study, age >60 years, hospitalization in the previous year (in the last 4 to12 months), previous antibiotic therapy (last month) and Karnofsky index <70 were found to be significantly and independently associated with infection by MDR pathogens. These risk factors were not considered in the definition of risk factors of HCAI proposed by Friedman et al. Patients with infection by MDR microorganisms had a significantly higher rate of inadequate initial antibiotic therapy, but did not have a longer hospital length of stay or higher hospital mortality rate. This is probably one of the first studies containing a thorough description of independent and significant risk factors associated with infection by pathogens of the ESKAPE group as well as their impact regarding inadequate antibiotic therapy, hospital length of stay and hospital mortality.

In order to define the new group of infections across community and hospital settings, a definition of all risk factors for MDR pathogen infections is required. The definition proposed by Friedman et al. [3] is one of the most used in clinical studies regarding HCAI, but it was developed for bloodstream infections. Further studies followed, using similar definitions, but again only for a single focus of infection or single pathogens. Our study identified additional risk factors for MDR pathogen infection.

Micek et al. [10] performed a retrospective cohort study of hospitalized patients with pneumonia, and identified immunocompromised state and hospitalization in the previous 12 months as risk factors for HCAI. They found a higher proportion of MDR organisms in the HCAI group, with a significant association with inappropriate antibiotic therapy and hospital mortality rate. One limitation discussed by the authors was the fact that they did not evaluate severity-of-illness scores. In the current study severity scores were considered and they were similar in all groups, with no significant differences in hospital mortality rate between groups. Park et al. [5] also performed a retrospective cohort study on HCAI-pneumonia, using the Friedman et al. definition and prior antibiotic therapy in the 30 days before the pneumonia episode, and found a higher proportion of MDR pathogens when compared with community-acquired pneumonia, and a significant association with inappropriate initial antibiotic therapy. Brito and Niederman [18] performed a review of all articles published since the American Thoracic Society guidelines were published [2], and concluded that patients at risk of pneumonia by MDR pathogens were those who had severe illness, were hospitalized in the past 90 days, had antibiotic therapy in the past 6 months, had poor functional status as defined by activities of daily living score, and were immunosuppressed, similar to the risk factors described here.

Wright et al. [19] performed a retrospective cohort study to determine risk factors for MDR pathogens among patients in the Emergency Department with urinary tract infections, and described the following independent risk factors: age ≥65 years, diabetes, prior use of antibiotics, and urinary catheterization. Chen et al. [20] studied patients from the community with bacteremia but with prior hospitalization from periods of 3–90, 91–180 to up to 181–360 days, and concluded that the presence of antimicrobial-resistant bacteria in the past 360 days was an independent risk factor for antimicrobial-resistant bacteremia, reaffirming that hospitalization in the previous year represented a risk factor for MDR pathogen infection. Pop-Vicas et al. [21], in a 6-year study of patients with positive cultures for MDR organisms, noted an increasing prevalence of MDR-GN at admission to a tertiary-care hospital (positive cultures within the first 48 h after hospital admission). Factors independently associated with isolation of the resistant organisms were age ≥65 years, prior antibiotic therapy for ≥14 days and prior residence in a long-term-care facility.

The current study has the major advantage of being prospective, with clear and consensual definition of infection [1]. It included all major foci of infection that drove patients into hospital care and not only pneumonia, bacteremia or urinary infection as in other previous studies. It clearly identified the place of acquisition and the appropriateness of initial antibiotic therapy. Data regarding severity of illness and the impact of infection by MDR organisms in the appropriateness of initial antibiotic therapy, hospital length of stay and mortality rate were also investigated. Data collection was performed by a single trained doctor, and all protocols were completed with no missing data thus minimizing information bias. All patients completed follow-up until hospital discharge. Most importantly, risk factors described in this study have been previously mentioned in studies restricted to single focus of infection or pathogen, by other authors from different parts of the world, suggesting important external applicability of these conclusions.

The study also has certain limitations that should be acknowledged. The research was performed in a single institution and the number of patients from the community was relatively small. There is accumulating evidence that HCAI seems to be a group encompassing CAI and HAI, but specific definition of risk factors are lacking. The diversity of definitions used has led to diverse findings not comparable between studies, and therefore not consequential for specific treatment recommendations.

Implementation of standardized approaches for the treatment of HCAI is desirable, since failure to recognize this group of infected patients appears to be associated with administration of inappropriate antibiotic therapy in hospitalized patients with serious infection, and may be related to poor prognosis.

The results of this study suggest that the inclusion of additional risk factors for MDR pathogen infection in HCAI definition, namely age >60 years, Karnofsky index <70, hospitalization in the previous year, and previous antibiotic therapy, may be clinically beneficial because of early diagnosis and a decrease in the rate of inadequate antibiotic therapy in these patients. Further research involving a large number of patients from different institutions and geographic areas is warranted to confirm these findings.