Discussion

In this multicenter study, illicit drug users were more likely to have a bacteremic infection caused by USA300 MRSA than any other type of S. aureus, regardless of when the patient's infection occurred. The data were consistent with a decrease in the association over the study period, from a RR of 4.63 in the early years of the study to a RR of 2.45 in the late years of the study.

The finding that the association between illicit drug use and USA300 MRSA bacteremia declined over the study period is clinically important. This illustrates that, although USA300 MRSA infections emerged in drug users, this epidemic is now spreading to other populations. The illicit drug-using population has been recognized as a reservoir for USA300 MRSA.^[14] The spread to other populations observed in this study could be the result of transmission and dissemination of the strains among at-risk patients in the community, as suggested by the fact that 43% of cases of bacteremia were not classified as nosocomial infections in this population. This means that patients who have bacteremia could have acquired the bacteria during a previous healthcare exposure or in the community. In contrast, illicit drug users who are colonized or infected with USA300 MRSA could serve as a reservoir for transmission to other patient populations while hospitalized for their infections.^{[12-14,26,27]} Further investigation is warranted.

The increased risk of acquiring an infection caused by USA300 MRSA among illicit drug users found in this study is consistent with the findings of other studies. Gilbert et al. reported that the incidence of USA300 MRSA was higher among high-risk case-patients (defined as including a history of illicit drug use) than low-risk case-patients.^[28]

In a study of the New York State Prison System, inmates with drug charges were more likely to have an infection. The USA300 MRSA strain was the predominant clone that caused infections in inmates.^[29] Although our study used different sampling methods and slightly different case definitions than previous studies, the conclusions are the same. Illicit drug users are at increased risk for an infection caused by USA300 MRSA compared with persons who do not use illicit drugs.

The association between illicit drug use and USA300 MRSA bacteremia could be explained by the fact that S. aureus infections are a common complication of drug use. Drug users are colonized with S. aureus more often than persons who do not use drugs.^[13,14] Colonization is a main risk factor for infection; the patient is usually infected with his own colonizing strain. In addition, invasive infections can occur from transfer of the colonizing strain directly into the patient's bloodstream.^[13,14] If USA300 MRSA is the colonizing strain, the frequent inhalation or injection of drugs could transfer this strain into the bloodstream to cause a life-threatening invasive infection.

Our study has some limitations. First, data regarding illicit drug use at the various VAMCs is based on self-report. Information regarding drug use is of a sensitive nature and, therefore, subject to recall bias. Also, to locate in the patient's electronic medical record whether the drugs were injected depended on whether the healthcare professional specifically asked and made note of this, a situation which is prone to information bias. Second, the ICD-9 codes used to define illicit drug use are imperfect measurements because they are used for insurance billing, rather than clinical purposes. Also, if any illicit drug-using patients were not hospitalized in the year before enrollment, or if drug-using patients did not report drug use during the hospitalization of interest, drug use would have been misclassified. In addition, using ICD-9 codes to define drug use up to 1 year before admission may not accurately measure current patterns of drug use in patients; the result would be a differential misclassification of exposure and could overestimate or underestimate the true association. Third, the study population was comprised only of veterans. The use of such a distinct population could reduce the generalizability of these findings. However, the relationship of illicit drug use with USA300 MRSA among veterans would likely not differ from the relationship among nonveterans, so the results should still be generalizable. Finally, USA300 MRSA isolates could have been misclassified with the use of our laboratory algorithm; however, the validation of isolates by PFGE, which showed a high sensitivity and specificity, makes this unlikely.

This study has several strengths. First, the use of more than 1 study site helped improve the generalizability of these findings. For example, we evaluated whether any of the associations observed may have been due to differences between the 4 VAMCs (this analysis was in response to a finding that patients from the Buffalo VAMC were less likely to be illicit drug users and less likely to have an infection due to USA300 MRSA). After excluding all patients from the Buffalo VAMC from the analysis, we found no difference from the results of the entire cohort (data not shown); therefore, we chose to present the combined data. Additionally, the electronic medical record system used throughout the Veterans Health Administration is known to be a valuable asset because of the completeness of data and for the amount of time it has stored information. The use of this system for chart review helped decrease selection and information bias. Finally, the findings of this study are strengthened because we could provide molecular typing data for each of the isolates.

In conclusion, the data from this study showed that illicit drug users are more likely to acquire a bacteremic infection caused by USA300 MRSA than by all other S. aureus strains. The decrease observed in the association of illicit drug use and USA300 MRSA bacteremia over the study period suggests that the USA300 MRSA epidemic is now spreading from illicit drug users to other patient populations. Focusing infection control efforts on high-risk groups such as illicit drug users might slow the progression of the USA300 MRSA epidemic in areas of the country where the association between illicit drug use and USA300 is still high.