Review of the identified 354 articles that included typed MRSA isolates with genotyping data identified 236 studies that included unique years of bacterial isolation and specified the US states where they were isolated. Within these 236 studies, 19,748 MRSA isolates were reported, of which 8,092 were classified as USA300 and 2,595 as USA100. Among isolates with any reported anatomic site of isolation, skin and soft tissue infections accounted for 62.6% of USA300 and 19.1% of USA100 isolates in studies that reported specific years and geographic locations and 58.8% of USA300 and 7.0% of USA100 in all studies, inclusive of those not reporting state locations or study years. Of all the USA300 and USA100 isolates in spatiotemporal analyses that reported a site of isolation, 38.5% of USA300 and 80.9% of USA100 were known invasive infections, whereas the full 33,543-isolate dataset designated 41.7% of USA300 and 93.0% of USA100 isolates as invasive infections.

Of isolates with reported spa types under the Ridom typing system, t008 and t002 were the dominant spa types during 2000–2004, 2005–2009, and 2010–2013 (Figure 1, panel A). Although t008 isolates made up the largest percentage during all 3 time groups, the share of isolates with this type decreased from 82% during the first period to 59% during the third, even as the total number of isolates with reported spa types increased. The share of isolates categorized as t002 rose from 10% during the first period to ≈20% during the 2 subsequent periods. ST8 and ST5 were the 2 most common MLST types reported during all 3 periods (Figure 1, panel B), as the number of isolates with reported MLST results declined during the study.

Figure 2

Figure 2. Proportions of methicillin-resistant Staphylococcus aureus isolates, United States 2000–2013. A) USA300 strain type. B) USA100 strain type. Darker shading indicates higher proportions of types reported in studies conducted during those years.

The proportion of MRSA isolates defined as USA300 increased during the study period and increased in some states earlier than others (Figure 2). Simultaneously, the proportion of isolates defined as USA100 decreased in many states.

Figure 3

Figure 3. Proportions of methicillin-resistant Staphylococcus aureus isolates in each state that were defined as USA300, USA100, or other strain types, United States 2000–2013.

Although USA100 isolates were reported during all years of the study period, pie charts indicate a gradual increase in the proportion of all isolates that were USA300 over time and space (Figure 3; Technical Appendix). In addition, the proportion of studies reporting USA100 simultaneously declined, and a higher percentage of studies reported other types of MRSA.

Figure 4

Figure 4. Weighted mean geographic center for proportions of methicillin-resistant Staphylococcus aureus (MRSA) USA300 strain type, United States, 2000–2013. This map shows the likely trend in the spread of USA300 as a proportion...

The weighted mean geographic center of USA300 shifted from the start of the study period (2000–2001 studies) to the end of the study period (2012–2013 studies) (Figure 4). The general movement was from west to east; the early weight of USA300 isolates was heaviest in the Midwest, gradually moving toward the mid-Atlantic states during the study period. The mean center was pulled westward during 2002–2005 by studies performed in Alaska and Hawaii during those years.

Figure 5

Figure 5. Proportion of methicillin-resistant Staphylococcus aureus USA300 and USA100 strain types and total sample size in 4 Census regions, United States 2000–2013. A) West. B) Midwest. C) Northeast. D) South. Linear regression...

When studies were assigned to 1 of 4 Census regions and the USA300 and USA100 proportions were charted over time (Figure 5), regional differences in the proportions of these 2 types differed by region. The number of isolates reported in each region also differed during the study. For the West, the general pattern over time was a series of peaks and troughs of USA300 reporting and a gradual decline in USA100. In the South and Northeast the general pattern over time was increasing USA300 proportions and decreasing USA100 proportions. The decrease in USA100 appeared particularly sharp in the Northeast. By contrast, in the Midwest, both USA300 and USA100 gradually increased, although the share of USA300 percentages was consistently higher. Cochran-Armitage tests for trend indicated statistically significant (p<0.0001) trends in USA300 and USA100 across all 4 Census areas.

We demonstrated that USA300 did not emerge simultaneously throughout the United States. It emerged earlier in the western part of the country and only later on the eastern seaboard. The published literature, used as a proxy for MRSA surveillance, suggests that USA300 appeared during 2000 in several states across the country, including California, Texas, and midwestern states. In subsequent years, USA300 constituted a large share of total reported MRSA isolates.

USA100, the predominant health care–associated MRSA strain type in the United States, in contrast, already constituted a larger proportion of reported MRSA isolates in the earlier years of the study in eastern US states. Over time, USA300 dominated among reported MRSA strain types in the Midwest and East. This finding is most clearly demonstrated by the focus of the geographic center of USA300 in Missouri/Illinois that gradually shifted toward the East during the latter years of the study.

Regionally, USA300 was present in higher proportions during the early years of the study in western states than other Census areas. Although Cochran-Armitage tests indicated statistically significant trends in all regions for both USA300 and USA100, trends toward an increase in the proportion of MRSA isolates that were USA300 were strongest in the Midwest and South, and large declines in USA100 proportions were observed in the Northeast, South, and West. These findings correspond to generally perceived but never formally tested hypotheses on the origins and spread of USA300 MRSA in the United States. The decline in the relative proportion of USA100 isolates reported during the study period most likely resulted from the increased attention to infection control in hospitals and a corresponding decrease in nosocomial health care–associated MRSA (31).

Our study is subject to several limitations. Most important, the data were not derived from a single sample of MRSA isolates collected prospectively or with equal representation of all states or regions. The data were derived instead from the extant literature in which authors chose to perform genotyping. Unlike an ideal prospective surveillance study that would include data selected to represent a sample of the entire population, we had available only published studies, which might have biased our results if in a given period. For example, if studies were more or less likely in 1 region of the country to focus on community-associated or health care–associated infections, if few studies were performed in a given region, or if smaller studies predominated in a given region compared with the rest of the country, our results could be biased. We attempted to correct for this lack of complete data by relying on all available data in each studied 2-year period to determine the geographic center of USA300 among MRSA isolates. The 2012–2013 center in particular was calculated from a small number of studies in the eastern portion of the country.

Second, although the articles included were identified through a rigorous literature review, additional published studies reporting a large number of isolates did not use any genotyping or did not use modalities of genotyping that met our inclusion criteria. Had these studies genotyped their reported isolates or genotyped them using methods that we chose to include, they might have altered the results of the study. Third, we eliminated from consideration in our analysis of geographic spread studies in which no US state of collection was named for studied MRSA isolates. This exclusion might have introduced error into our results, but we have no reason to believe that this error would systematically bias our findings because a specific region is unlikely to be overrepresented or underrepresented given this exclusion criterion. Fourth, although we made every attempt to exclude studies that reported previously published isolates, some isolates might been included more than once in our analyses. We identified specific strain names in the MRSA TypeCat whenever they were included in a published report to avoid repeated entries, but many authors did not identify specific isolate designations in their publications. Finally, many articles with genotyping information were not included because they provided inadequate typing data to identify USA300 or USA100 isolates by the criteria that we used to define these 2 strain types. Some articles, for example, identified isolates that bear the genetic determinants of PVL and SCCmec type IV, strongly suggesting a USA300 isolate, but we did not include these isolates in our analysis. We believe that the criteria that we chose for USA300 isolates, although arbitrary, were appropriately conservative to avoid misclassification of other community-associated MRSA strain types as USA300.

Our study examined the geographic distribution of USA300, a strain type of MRSA that emerged in the late 1990s to cause the fourth wave of resistance in S. aureus and came to predominate as a cause of MRSA infections in the United States during the course of approximately a decade. The reasons for the geographic pattern of emergence of USA300 from west to east are not yet known. However, this study represents an attempt to document the movement of a successful epidemic strain type of MRSA geographically over a prolonged period from its earliest emergence to predominance among MRSA strain types in a country. Our study might provide a model for understanding the emergence of a future, novel, fit strain type of antimicrobial drug–resistant S. aureus that could make up the fifth wave of resistance in S. aureus and is particularly relevant given increased focus and funding from the US government on developing a national strategy to combat antimicrobial drug–resistant bacteria (32). Such a strategy should include a national surveillance program that can detect the regional emergence of virulent new strains to inform local empiric therapy.