This large retrospective analysis of microbiologically-confirmed culture positive pneumonia patients presenting to the hospital reveals that MRSA constitutes an important pathogen in this setting. Similarly, HCAP accounts for a large proportion of all pneumonia patients admitted to the hospital and is nearly as common as CAP. Even though patients with HCAP face an increased risk for MRSA, the prevalence of MRSA in HCAP was less than 20%. A risk scoring tool based on factors independently associated with the recovery of MRSA segregates patients with moderate accuracy. A low total score describes a cohort of individuals unlikely to have MRSA and in whom anti-MRSA therapy can likely be withheld.
Our data add to the growing evidence elucidating the frequency of HCAP relative to CAP. In the initial study examining the epidemiology of HCAP, HCAP accounted for approximately one-third of all pneumonias presenting to the emergency department (ED) . Other reports have documented that HCAP represents between 30% and 60% of pneumonias initially evaluated in the ED [5–7], and this trend is also observed in Asia and Europe [9, 16]. Our study confirms that MRSA is a common pathogen in patients presenting to the hospital with their pulmonary infections. Classically thought only to be significant in nosocomial infections, it is now evident that physicians treating those who come to the hospital with pneumonia must consider MRSA when selecting antibiotic regimens.
Originally, the notion of HCAP was developed to help with this decision making process . The specific components of the HCAP definition were derived from expert opinion and were initially adopted from a single centre study in bacteraemia [3, 17]. HCAP was meant to serve as a tool to help clinicians stratify individuals as to the chance that a resistant pathogen, like MRSA, was a practical issue. Our finding that there are many patients with HCAP who are not infected with MRSA suggests that HCAP alone cannot serve as a precise tool for risk stratification. Our findings further confirm the observations of earlier researchers. For example, Schrieber and colleagues found that HCAP performed poorly as a screening test for MRSA in persons presenting to the hospital with severe pneumonia necessitating mechanical ventilation . Shorr et al. similarly found that HCAP misclassified many patients both as to their risk for resistant pathogens in general, and for MRSA, specifically . Thus, it appears that broad use of the HCAP criteria for classification can lead to overuse of broad spectrum antibiotics which can result in unnecessary cost and antimicrobial resistance.
The proposed risk score is novel and improves on the status quo, as it provides a means for reliably classifying quickly and easily a large cohort of patients at low risk for MRSA. Since anti-MRSA therapies can often be added or withheld without necessarily adjusting the selected Gram-negative coverage, having an effective way to limit the prescription of anti-MRSA treatments becomes crucial. The situation with respect to anti-MRSA treatment options is even more acute given their costs and potential side effects.
Prior reports have proposed alternate risk scoring schemes for determining the probability of resistant organisms in those coming to the hospital with pneumonia [5–7]. However, most of these studies have pooled resistant Gram negative organisms (e.g., P. aeruginosa) with MRSA [5–7]. Our effort is therefore unique in that we have examined MRSA specifically. Our results are also novel in that we identify several factors associated with MRSA that have not previously been linked to pneumonia with this pathogen. For example, extremes of age have not been previously described as related to MRSA in pneumonia. Likewise the description of a nexus between certain co morbid diseases, such as cerebrovascular disease (CVD) and dementia, with MRSA has not been noted before in evaluations of pneumonia in the ED. The mechanism of these associations with MRSA is unclear. These specific factors might actually represent surrogates for other factors that we could neither address nor measure in the dataset. Additionally, it appears that not all factors contribute equally to the risk for MRSA as a cause of pneumonia on hospital presentation. Intense exposure to the healthcare system (e.g., recent inpatient stay) seems disproportionately important as does severity of illness. Moreover, different risk scores utilized thus far in pneumonia (e.g., CURB-65) focus solely on severity of illness while our risk score specifically deals with issues of aetiology.
Our study has several strengths. It is based on a large sample from multiple hospitals across the US. This provides us the power to examine select conditions that other analyses could not evaluate because of their smaller size. It also suggests the generalizability of our findings. We have also internally validated the proposed risk score. No other effort has applied such a split sample approach in pneumonia presenting to the hospital. In fact, many risk stratification schemes for other disease states never undergo any effort at either internal or external validation.
Conversely, the present investigation has a number of significant limitations. First, its retrospective nature exposes it to several forms of bias. This is particularly concerning since we relied on ICD-9-CM coding to identify patients with pneumonia or a primary diagnosis of sepsis with a secondary diagnosis of pneumonia. There certainly are patients we included who likely had some alternate diagnosis. Second, culture data are not always obtained in every case of suspected pneumonia which introduces a selection bias. A priori, physicians may have preferentially opted to obtain cultures in patients when they were already concerned about resistant pathogens. Moreover, cultures can be negative even in the presence of active bacterial infection and positive cultures of the respiratory tract can represent colonization and not true infection. Again, this raises the concern that we could have misclassified patients. On the other hand, since we only studied culture positive patients, our findings likely underestimate the potential for antibiotic overuse and abuse. Third, we did not evaluate patients for co-infection with viral pathogens which could have influenced our findings, especially as a risk factor for MRSA infection. Fourth, despite our attempt to evaluate all components of the HCAP definition, we likely miscategorised some patients because of inadequate information regarding recent IV antibiotic exposure since administrative datasets may not consistently capture this. Some of the factors identified, such as dementia, could represent surrogates for other characteristics about which we lacked data. Finally, our score had only moderate discriminatory power and requires external validation. Although it identified a group of persons as low risk for MRSA, our data illustrate that there is a clear need for better tools for this purpose and that we urgently require rapid diagnostic tests for organism identification.