We began active, statewide, population- and laboratory-based surveillance for invasive GAS disease in April 1995 through Active Bacterial Core surveillance (ABCs), part of the Emerging Infections Program of the Centers for Disease Control and Prevention (CDC) (20). The population of Minnesota was 4.9 million in 2000. Invasive GAS disease is defined as GAS isolated from a normally sterile site such as blood or cerebrospinal fluid or from a wound when accompanied by STSS or necrotizing fasciitis (20). To ensure complete case capture, laboratories either submit computerized lists of all GAS-positive cultures from normally sterile sites at least monthly or are contacted twice monthly by Minnesota Department of Health (MDH) staff, and audits are completed routinely. Hospital infection control practitioners then complete standard report forms for cases of invasive disease, and all GAS isolates are sent to the MDH Public Health Laboratory. All GAS isolates undergo pulsed-field gel electrophoresis (PFGE) with Sma1 by methods described elsewhere, with the exception that an Enterococcus isolate was used as the standard (21). PFGE patterns are evaluated both visually and with BioNumerics software (Applied Maths, Kortrijk, Belgium) by using the dice coefficient. For patterns to be considered indistinguishable, they must visually appear identical and the DNA patterns must differ by <1.5% with respect to molecular weight. Isolates are also sent to CDC for emm typing (22).

Beginning in 1995, information about a case-patient’s residence was collected on the case report form, including street address, city, state, and ZIP code. In 1998, a question was added about whether the case-patient had been a resident of a nursing home, long-term care facility (LTCF), or other chronic care facility for at least 30 days before the date the culture was collected. Persons living in group homes, prisons, rehabilitation hospitals, or who were going to facilities for daily outpatient therapy were not included. Beginning in 2002, the name of the facility was collected. Addresses for case-patients >55 years of age were also checked by a reverse address directory to see whether they corresponded with that of an LTCF. Information about size, location, and classification of LTCFs was collected from a directory of Minnesota licensed, certified, and registered healthcare facilities.

For the purposes of this study, only case-patients who could be confirmed as residents of nursing homes were included because accurate denominator data were only available for this group. In Minnesota, a nursing home is defined as a facility that provides nursing care to >5 persons who are not in need of acute care facilities but require nursing supervision on an inpatient basis. Denominators for calculating incidence were derived from the 2000 US Census data as reported by the Minnesota State Demographic Center, which describe the population living in group quarters by age and type of quarters (23).

We defined a nursing home cluster as >2 cases in residents of a nursing home in which isolates were nearly identical as determined by PFGE (PFGE patterns within a 3-band difference [24]) during a 12-month period. PFGE patterns were used for cluster identification because they were more discriminating than emm types (e.g., several PFGE patterns were typically found to correspond to 1 emm type, while PFGE patterns were not found to have multiple emm types), and PFGE was readily available in our laboratory. We chose 12 months because we observed that invasive cases with indistinguishable patterns sometimes occurred many months apart within a facility. Beginning in 1995, case reports were reviewed regularly by address, facility name, and PFGE pattern to look for clusters.

EpiInfo version 6.0 (www.cdc.gov/epiinfo/Epi6/EI6dnjp.htm) was used for statistical analysis. χ² test was used to determine statistical significance of differences in proportions for discrete variables, and a t test was used to determine whether the difference in means was significant for continuous variables.

Since 1997, whenever a cluster was identified through ABCs, the nursing home was contacted by MDH and encouraged to conduct retrospective and enhanced prospective surveillance for invasive and noninvasive GAS infections. This surveillance included reviewing culture logs to identify noninvasive GAS infections and residents with chronic or recurrent infections and reviewing reported staff illnesses to identify a possible source of GAS. Clinical examples of possible GAS infections were provided so the facility could consider the possibility of earlier undiagnosed GAS disease and recognize new suspect cases. For prospective surveillance, the nursing home was encouraged to obtain cultures for any suspected infection and to ask laboratories to save GAS isolates for PFGE and emm typing. A follow-up letter and packet of information about GAS and infection control measures were also sent to the nursing home. We offered assistance of further investigation but most often had no further contact with the facility. In 2004, after noting that we seldom observed another case at a facility after we contacted them regarding a cluster, we began to contact a facility any time we received a report of a single case.

In Minnesota, MDH staff cannot conduct an investigation in a nursing home without an invitation from the facility. Only 2 facilities with clusters requested our assistance. These investigations have been described in detail elsewhere (25,26), but in both instances cultures were collected from residents and staff (all residents and staff at facility G; all but 1 resident who refused and 3 staff members in the affected unit at facility B). Those with positive GAS cultures were treated with antimicrobial agents (clindamycin at facility G and penicillin and rifampin at facility B). Formal infection control educational sessions were provided for staff on the same day that cultures were collected.

From April 1995 through 2006, 1,858 cases of invasive GAS disease were reported among Minnesota residents; 642 (35%) were in persons >65 years of age. One hundred seventy-five case-patients were identified as LTCF residents on their case report forms. Twenty-three of these case-patients resided in non–nursing home settings such as assisted living or group homes, and we were unable to determine the type of setting for 18 of those designated as LTCF residents. One hundred thirty-four (7%) of our case-patients were known to be nursing home residents. The number and percentage of all cases associated with nursing homes fluctuated over time; from 6 to 21 cases were identified among nursing home residents annually, representing 3%–12% of all case-patients each year (Figure 1). Seasonal variation of invasive GAS infections was noted among both the general population and nursing home residents (Figure 2), with peak incidence in the winter and spring and little disease noted in late summer and early fall.

The age of nursing home case-patients ranged from 36 to 100 years of age (median 84 years); 58% were women, 87% had positive blood cultures, 36% had bacteremia without another focus of infection, 32% had cellulitis, and 12% had pneumonia. The case-fatality ratio of all case-patients with invasive GAS was 12%. Among case-patients >65 years of age, the case-fatality rate of nursing home resident case-patients (n = 121) was 35% compared with 18% of case-patients who were not nursing home residents (n = 521).

In 2000, 37,542 Minnesota residents >65 years of age lived in nursing homes, and 556,724 lived in their own homes or other group quarters. During 2000, the incidence of invasive GAS infections among Minnesota nursing home residents >65 years was 18.6 cases/100,000 population compared with 6.8 cases/100,000 among those >65 years who did not reside in nursing homes. Estimated annual incidence for nursing home residents >65 years varied from 13.3 cases/100,000 in 1997 to 50.6 cases/100,000 in 2003, while the estimated incidence for non–nursing home residents >65 years was 8.6 and 9.3 cases/100,000 during the same years.

Emm type was available for 117 (87%) of the nursing home case-patient isolates. Of 21 different emm types identified, 4 (emm 1 [21%], emm 89 [15%], emm 28 [13%], and emm 03 [11%]) accounted for 60% of the isolates. Among 1,416 (82%) non–nursing home case-patients, 5 emm types accounted for 62% of the isolates (emm 1 [24%], emm 28 [13%], emm 03 [11%], emm 12 [10%], and emm 89 [4%]). Although total numbers of cases varied considerably from one year to the next, the proportion of disease caused by the most common emm types fluctuated little.

Of the 444 licensed nursing homes in Minnesota, 91 (20%) were known to have at least 1 case of invasive GAS disease during the study period. Sixty-seven (74%) of these facilities had a single case; 13 facilities had 2 cases; and 11 facilities had >3 cases. Of 24 facilities that had >2 cases, 13 (54%) met the definition of a cluster as previously defined (Table). We found that PFGE patterns for isolates from the same facilities were either indistinguishable from each other or distinctly different (>3 bands different).

Four nursing homes that had clusters also had additional cases that did not fit the definition for inclusion in the cluster, either because a case isolate had a distinctly different PFGE pattern, a case did not occur within 1 year of the other cases, or both. One of these facilities had 4 such cases. In addition, 11 other nursing homes each had 2–3 cases that did not fit the definition of a cluster.

All but 2 clusters were caused by the most common emm types. Emm 1 was the most common cause of invasive disease (causing 24% of all cases) and also the cause of 5 (38%) clusters.

Eighteen of 21 pairs (86%) of consecutive cases occurring within 12 months of each other in the same facility had matching PFGE patterns, while 13 (93%) of 14 pairs of consecutive cases occurring within 3 months had matching patterns. The occurrence of a third case in a nursing home was not dependent on the first 2 case isolates having the same PFGE patterns; 6 (46%) of 13 facilities in which the first 2 case isolates had different PFGE patterns and 5 (45%) of 11 facilities in which the first 2 case isolates had matching PFGE patterns subsequently had more cases.

No significant difference was found for age, sex, or type of infection between cluster and sporadic cases. Forty-one percent of case-patients with cluster-associated cases died, compared with 32% of patients with sporadic cases, but this difference was not significant (p = 0.33).

In 32 (86%) of 37 encounters with nursing staff, the person contacted was not aware of a diagnosis of invasive GAS disease among their residents before our call. In addition, even when our contacts (usually either directors of nursing or nurses designated to oversee infection control for the facility) did know of the diagnosis, they generally had little knowledge about GAS disease. Before 2004, we noted that 9 of 12 nursing homes did not identify additional cases of invasive disease after our call. In 2004, we began notifying nursing homes after we identified single cases in their facilities. Since then, we are aware of only 1 facility with a cluster, and that facility had 2 cases 4 days apart.

We collected throat and skin lesion cultures from staff and residents for a unit with a cluster of invasive GAS disease at facility B and from staff and residents of the entire nursing home at facility G. At facility B, 2 (2.7%) of 75 throat cultures from staff and 2 (5.9%) of 34 throat cultures from residents were positive for GAS; 5 (6.2%) of 81 throat cultures from staff and 2 (4.5%) of 44 throat cultures of residents were positive for GAS at facility G. All of those with positive throat cultures were asymptomatic at the time of culture. All except 1 isolate from a staff person at facility G who did not provide direct patient care had PFGE patterns indistinguishable from those associated with the invasive cases at the facility. Those with positive cultures were each treated with a course of antimicrobial drugs, and no additional cases were detected at either facility.