Methods

Case Identification and Data Collection

We defined confirmed iGAS cases as S. pyogenes isolated by culture from a normally sterile site (e.g., blood, or cerebrospinal, joint, pleural, or pericardial fluids). Cases of iGAS in Idaho must be reported to the Idaho Division of Public Health or local public health agencies, according to Idaho reportable disease rules (https://publicdocuments.dhw.idaho.gov/WebLink/DocView.aspx?id=6798&dbid=0&repo=PUBLIC-DOCUMENTS). Reportable disease surveillance is a passive surveillance system, and with rare exceptions, reports are submitted by laboratories. After iGAS cases are reported, local public health district epidemiologists collect data and determine whether a given case is associated with a congregate setting or an outbreak. For this retrospective study, we used information about iGAS cases reported during 2008–2019, including patient demographics, laboratory reports, and data obtained from public health investigations, recorded in the Idaho National Electronic Disease Surveillance System Base System (NBS) and from review of medical records to collect additional clinical and risk factor information. We viewed medical records in the Idaho Health Data Exchange or obtained them by faxing medical record requests to hospitals. Because reported iGAS cases increased substantially from 2013 to 2014, we reviewed reporting facilities and submission procedures to determine whether potential changes in reporting methods (for example, an increase in electronic over manual reporting) might have contributed to the increase in reported cases.

We obtained data concerning hospitalization and death from NBS or medical records; however, the records for fatal cases did not always indicate whether iGAS was considered a cause of or contributing factor to death. Case information encompassed patient demographics (i.e., age, sex, race and ethnicity), type of residence, clinical syndrome (i.e., pneumonia, bacteremia, meningitis, cellulitis), and underlying conditions and behavioral risk factors (i.e., diabetes, skin wound or injury, cancer, chronic obstructive pulmonary disease or emphysema, obesity, dialysis or renal failure, immunosuppression, postpartum status, recent surgery, alcohol abuse, injection drug use, current smoking). We selected variables for data abstraction based on the Centers for Disease Control and Prevention (CDC) Active Bacterial Core surveillance (ABCs) case report form (https://www.cdc.gov/abcs/downloads/abcs-case-rpt-form.pdf). We collected data on other risk factors (e.g., GAS pharyngitis, recent or concurrent influenza infection, household member with GAS infection), when available, from medical records or NBS.

Except for results from 2 isolates from 2012 that underwent emm typing at the Boise Veterans Administration Medical Center, emm typing results were provided by the Idaho Bureau of Laboratories, which began emm typing GAS isolates voluntarily submitted by diagnostic laboratories in 2014. emm typing at Idaho Bureau of Laboratories was conducted by PCR amplification and Sanger sequencing of the emm gene according to CDC protocols (https://www.cdc.gov/streplab/protocol-emm-type.html). emm types were assigned using a CDC M type–specific sequence database library within the bionumerics software platform (ftp://ftp.cdc.gov/pub/infectious_diseases/biotech/tsemm). Sequences were verified using the CDC Blast-emm database (https://www2.cdc.gov/vaccines/biotech/strepblast.asp).

Data Analysis

We used Stata software version 14.2 (StataCorp, LLC, https://www.stata.com) to conduct data analysis and compiled categorical data as percentages and continuous data as medians with interquartile ranges (IQRs). We calculated annual incidence per 100,000 persons overall and by age group using annual census data for Idaho population estimates (https://www.census.gov/data.html); we age-standardized incidence rates to the average population distribution for Idaho during 2008–2019 and reported with 95% CIs. We calculated incidence estimates by race and ethnicity using data from the Idaho Bureau of Vital Records and Health Statistics (Appendix, https://wwwnc. cdc.gov/EID/article/28/9/21-2129-App1.pdf).

To investigate the increase in iGAS incidence, we conducted a case-case analysis, comparing cases from the higher incidence period (2014–2019) with cases from the lower incidence baseline period (2008–2013). We used the χ² test to compare associations between periods and disease syndromes, deaths, hospitalization status, and predisposing conditions; we used the Mann-Whitney U test to compare median length of hospital stay. We used logistic regression models to compare demographics, underlying conditions, and other risk factors between the 2 periods to identify any factors that might be positively associated with the higher incidence period (Appendix). We included age group, obesity, residence type, and injection drug use as variables in the multivariable logistic regression model. We reported results as odds ratios for the univariable models and adjusted odds ratios for the multivariable model and calculated 95% CIs for all results. Odds ratios >1 indicated higher odds of cases with that risk factor being in the 2014–2019 period, compared with the baseline period. Idaho Division of Public Health and CDC determined this project to be nonresearch (public health practice). We conducted this activity consistent with applicable federal law and CDC policy.