Trends in Mortality and the Effects of the Coronavirus Disease 2019 Pandemic on Mortality Among Adult Patients With Staphylococcus aureus Bacteremia, 20 US Counties, 2005–2022

Abstract

Background

Trends in Staphylococcus aureus bacteremia (SAB) mortality in the United States, including the impact of coronavirus disease 2019, have not been well described.

Methods

Active, laboratory- and population-based surveillance data from 20 counties (7 states) were used to examine annual, 30-day mortality rates among US adult patients with methicillin-resistant and methicillin-susceptible Staphylococcus aureus (MRSA and MSSA, respectively) bacteremia during 2005–2022. Medical records and state vital statistics data were linked to ascertain death. Multivariable survival analysis estimated the association between recent severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infection (a positive SARS-CoV-2 test of ≤30 days before incident culture) and mortality among SAB cases, adjusting for other risk factors. Analyses were stratified by MRSA/MSSA and epidemiologic class (hospital-onset [HO], healthcare-associated community-onset [HACO], and community-associated [CA]).

Results

Overall, SAB mortality was 21%. Annual SAB mortality rates were generally stable prepandemic before increasing by 33% and 41% among HO MRSA and MSSA cases, respectively, and by 49% among CA MRSA cases, but not among HACO or MSSA CA cases, from 2019 through 2021. Among HO cases with recent SARS-CoV-2 infection, annual mortality rates were as high as 75%. Adjusted hazard ratios for recent SARS-CoV-2 infection ranged from 1.9 (95% confidence interval: 1.5–2.3) (HACO MRSA) to 3.0 (95% confidence interval: 2.4–3.7) (HO MSSA).

Conclusions

Recent SARS-CoV-2 infection was a significant risk factor for death among patients with SAB during 2020–2022. Strengthening infection prevention efforts in similar respiratory pandemics in the future could help prevent increases in SAB mortality rates.

Graphical Abstract

Graphical Abstract.

This graphical abstract is also available at Tidbit: https://tidbitapp.io/tidbits/trends-in-mortality-and-the-effects-of-the-covid-19-pandemic-on-mortality-among-adult-patients-with-staphylococcus-aureus-bacteremia-20-us-counties-2005-2022?utm_campaign=tidbitlinkshare&utm_source=IO.

Recent severe acute respiratory syndrome coronavirus 2 infection was a significant risk factor for death among US patients with Staphylococcus aureus bacteremia during 2020–2022. Strengthening healthcare-associated infection prevention to reduce secondary S aureus bacteremia among hospitalized patients could prevent deaths during future viral respiratory pandemics.

Staphylococcus aureus is a leading cause of hospital- and community-acquired bacteremia in the United States [1–4]. Staphylococcus aureus bacteremia (SAB) causes significant morbidity and mortality, with an estimated 1-month mortality rate of 19.6% in a global meta-analysis [5]. Although national trends in invasive S aureus infections (including SAB) have been described elsewhere, trends in SAB mortality in the United States, including how mortality changed during the coronavirus disease 2019 (COVID-19) pandemic, are not well understood [6–10].

Prior studies have shown that patients with both severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), the virus that causes COVID-19, infection and SAB (compared to SARS-CoV-2 infection or SAB alone) have higher mortality rates [11–14] and that the prevalence of recent SARS-CoV-2 infection is higher among hospital-onset (HO) SAB cases compared to community-onset SAB cases [11]. However, many prior studies were conducted early in the COVID-19 pandemic, were limited to in-hospital mortality or localized patient populations (eg, a single hospital or city), and/or did not control for potential confounders [11–14]. Furthermore, neither trends in prepandemic SAB mortality rates nor the indirect effects of the pandemic (eg, a strained health system or delays in seeking care) on SAB mortality in the United States have been examined [15].

Using a robust longitudinal surveillance dataset combined with state vital statistics data from multiple states, we aimed to describe (1) annual trends in SAB mortality rates, both before and during the COVID-19 pandemic; (2) the association between recent SARS-CoV-2 infection and mortality among SAB cases, adjusted for patient characteristics; and (3) the direct, indirect, total, and overall effects of COVID-19 on mortality among SAB cases.

METHODS

Data

The US Centers for Disease Control and Prevention (CDC) Emerging Infections Program (EIP) tracks invasive S aureus cases through active, laboratory- and population-based surveillance [6]. Data were collected from 20 counties in 7 states (Supplementary Table 1). For this analysis, cases were defined as isolation of S aureus from an initial blood specimen in a surveillance area resident. Site surveillance staff completed a standardized case report form for all cases based on medical record review, as previously described [1]. Case-level data were entered into Microsoft Access or Research Electronic Data Capture (REDCap) and securely transmitted to CDC monthly [16]. A positive culture collected ≥30 days after incident culture was considered a new case. Cases were categorized by methicillin susceptibility (methicillin-resistant Staphylococcus aureus [MRSA] or methicillin-susceptible Staphylococcus aureus [MSSA]), as reported by local clinical microbiology laboratories. Cases were further categorized by epidemiologic class: hospital-onset (HO) if the incident culture was collected ≥3 calendar days after hospitalization; healthcare-associated community-onset (HACO) if the incident culture was collected in an outpatient setting or <3 calendar days after hospitalization from a patient with 1 or more healthcare risk factors: history of hospitalization, surgery, dialysis, or residence in a long-term care facility (LTCF) in the previous year, or the presence of a central venous catheter within 2 days prior to the incident culture date; or community-associated (CA) if none of the previously mentioned criteria were met. Sites linked SAB cases to state vital statistics data to ascertain death from any cause within 90 days of incident culture. Methicillin-susceptible Staphylococcus aureus surveillance counties (n = 11) were a subset of MRSA surveillance counties (n = 20). The analysis was restricted to adult (≥18 years) cases during 2005–2022 (MRSA) and 2016–2022 (MSSA). Cases were excluded if methicillin susceptibility or age were missing, or if epidemiologic classification could not be determined.

The outcome of interest was all-cause mortality within 30 days of incident culture. This follow-up time captures most SAB-attributable deaths while reducing confounding due to other competing causes of mortality [17, 18]. The outcome was determined using death dates from both the medical record and vital statistics data. Cases were excluded if death dates from the 2 data sources were discrepant, could not be resolved, and would result in discrepant classifications of 30-day mortality.

Recent SARS-CoV-2 infection was defined as a positive viral SARS-CoV-2 test (molecular assay or antigen) within 30 days before or on the incident SAB culture date. Test results for SARS-CoV-2 were ascertained by site surveillance staff through medical record review (all sites), state reportable disease surveillance systems (6 sites), and COVID-19 hospitalization surveillance (COVID-NET) records (6 sites) [19]. Staphylococcus aureus bacteremia cases with incident culture date on or after 21 March 2020 (the earliest incident culture date for a case with recent SARS-CoV-2 infection in our data) were excluded if recent SARS-CoV-2 infection could not be determined. All analyses were stratified by epidemiologic class and methicillin susceptibility.

Trends in Staphylococcus aureus Bacteremia Mortality Rates

Strata-specific annual mortality rates were calculated by dividing the number of SAB cases that died from all causes within 30 days of incident culture by the total number of SAB cases with an incident culture each year. We plotted annual mortality rates and calculated annual percentage changes. Mortality rates during 2020–2022 were further stratified by recent SARS-CoV-2 infection and plotted with overall mortality rates. Confidence intervals (CIs) were calculated using the binomial distribution.

Survival Analysis

We used mixed Cox proportional hazards models to examine associations between recent SARS-CoV-2 infection and mortality among SAB cases during 21 March 2020–31 December 2022. Survival time, in days, was the time from incident culture to either death or censoring at 30 days. Cases with death on the incident culture date were assigned a survival time of 0.01 days.

Using directed acyclic graphs (DAGs), the following variables were identified as confounders and included in regression models: incident culture year, demographics (race/ethnicity, age group, and sex), comorbidities (Charlson Comorbidity Index [CCI] [20] and body mass index [BMI]), and other risk factors (current smoker, alcohol abuse, substance use disorder, experiencing homelessness, and LTCF resident) (Supplementary Table 2). Long-term care facility residency was excluded from CA models, as CA cases could not, by definition, be LTCF residents. An interaction between recent SARS-CoV-2 infection and incident culture year was also assessed. Categorical variables were used for age group, BMI, and CCI [21]. To account for clustering, patient identifier and state were included as random intercepts. The proportional hazards assumption was assessed using Schoenfeld residuals and log–log survival plots. Model fit was assessed by stratifying cases into quintiles of predicted risk and comparing predicted survival curves to observed Kaplan–Meier curves for each quintile.

Missing values were imputed using multiple imputation by chained equations (MICE) [22]. Eight percent of cases had 1 or more missing value(s). Imputation models included all variables from the regression model (except patient identifier), as well as the cumulative hazard rate and stratification variables (methicillin susceptibility and epidemiologic class). Derived variables were passively imputed. Ten datasets were created, and regression results were combined using Rubin's rules [23].

Effects of the Coronavirus Disease 2019 Pandemic on Staphylococcus aureus Bacteremia Mortality

We extended epidemiologic concepts of infectious disease intervention effects to describe the effects of the COVID-19 pandemic on SAB mortality [24]. We examined (1) the direct effect, or the difference between the outcome (died or survived) in a person with SAB with recent SARS-CoV-2 infection and what their outcome would have been had they not had recent SARS-CoV-2 infection; (2) the indirect effect, or the effect of the COVID-19 pandemic on a person with SAB who did not have recent SARS-CoV-2 infection due to the broader societal and healthcare system disruptions; (3) the total effect, or the direct effect of SARS-CoV-2 infection combined with the indirect effect of the COVID-19 pandemic; and (4) the overall effect, or the difference in the outcome in an average person with SAB during the COVID-19 pandemic and what their outcome would have been in the absence of the COVID-19 pandemic (Supplementary Figure 1). We quantified effect estimates using unadjusted 30-day all-cause mortality rate ratios, calculated by dividing mortality rates between the following groups: (1) SAB cases during the pandemic (21 March 2020–31 December 2021) with recent SARS-CoV-2 infection versus those without recent SARS-CoV-2 infection (direct effect); (2) SAB cases during the pandemic without recent SARS-CoV-2 infection versus SAB cases prepandemic (21 March 2018–31 December 2019) (indirect effect); (3) SAB cases during the pandemic with recent SARS-CoV-2 infection versus SAB cases prepandemic (total effect); and (4) all SAB cases during the pandemic versus those prepandemic (overall effect). The year 2022 was excluded to better capture the effects of peak pandemic activity (Supplementary Figure 1). Confidence intervals were calculated using the Koopman asymptotic score method [25].

Sensitivity Analyses

For all analyses, we restricted the data to counties included in MSSA surveillance only to examine whether results were affected by different surveillance populations between MRSA and MSSA. For the survival analysis, we also (1) excluded cases with any missingness to examine the sensitivity of results to imputed data and (2) varied the definition of recent SARS-CoV-2 infection to include positive SARS-CoV-2 tests within 15, 45, and 60 days before or on the incident culture date.

This activity was reviewed by CDC, deemed not research, and was conducted consistent with applicable federal law and CDC policy (see, eg, 45 C.F.R. part 46.102(l)(2), 21 C.F.R. part 56; 42 U.S.C. §241(d); 5 U.S.C. §552a; 44 U.S.C. §3501 et seq). Additionally, each participating site evaluated the protocol and either deemed it a nonresearch surveillance activity or obtained institutional review board approval with a waiver of informed consent.

RESULTS

Trends in Staphylococcus aureus Bacteremia Mortality

A total of 72 920 SAB cases and 15 114 deaths (mortality rate: 21%) during 2005–2022 were included. Of these, 54 043 cases and 11 692 deaths (mortality rate: 22%) were MRSA and 18 877 cases and 3422 deaths (mortality rate: 18%) were MSSA. Fewer than 1% of cases were excluded from all analyses because patient outcome or recent SARS-CoV-2 infection could not be determined (Supplementary Table 3). If medical record data alone had been used, rather than medical record and vital statistics data, 1321 cases (including 354 deaths) would have been excluded due to missing patient outcome information, and an additional 3923 cases that died would have been misclassified as having survived, resulting in an overall 30-day mortality rate of 16%.

Prepandemic MRSA mortality rates ranged from 28% to 37% (HO), 19% to 23% (HACO), and 11% to 17% (CA) during 2005–2019 (Figure 1; Supplementary Table 4). During the pandemic, MRSA HO mortality rates increased from 33% in 2019 to 41% in 2020 (a 25% increase), remained elevated in 2021 at 44%, and then declined to prepandemic levels (29%) in 2022. Overall, MRSA HO mortality rates increased by 33% from 2019 to 2021. Methicillin-resistant Staphylococcus aureus HACO mortality rates remained stable, ranging between 19% and 22% during 2019–2022. Finally, MRSA CA mortality rates increased from 11% in 2019 to 18% in 2020 (a 65% increase) and remained elevated in 2021 and 2022. Overall, MRSA CA mortality rates increased by 49% from 2019 to 2021. Among all epidemiologic classes, mortality rates for MRSA cases with recent SARS-CoV-2 infection were 2–3 times greater than mortality rates for cases without recent SARS-CoV-2 infection. The highest mortality rate (75%) occurred in 2020 among MRSA HO cases with recent SARS-CoV-2 infection. In sensitivity analyses restricting MRSA data to MSSA counties, mortality rate trends were generally the same; however, the increase in MRSA CA mortality rates during 2019–2020 was slightly attenuated (Supplementary Figure 2).

Figure 1.

Annual 30-day all-cause mortality rates^a and 95% confidence intervals^b for adult Staphylococcus aureus bacteremia cases by methicillin susceptibility^c and recent SARS-CoV-2 infection^d, both overall (panels A–B) and by epidemiologic class^e (panels C–H), 20 US counties, 2005–2022^f. Abbreviations: methicillin-resistant Staphylococcus aureus, MRSA; methicillin-susceptible Staphylococcus aureus, MSSA; hospital-onset, HO; healthcare-associated community-onset, HACO; community-associated, CA. ^aStrata-specific annual mortality rates were calculated by dividing the number of S aureus bacteremia cases that died from all causes within 30 days of incident culture by the total number of S aureus bacteremia cases with an incident S aureus culture in a given year. ^bConfidence intervals were calculated using the binomial distribution. ^cCases were categorized as either MRSA or MSSA (ie, methicillin susceptibility), as determined by microbiologic testing. ^dRecent SARS-CoV-2 infection was defined as a positive viral SARS-CoV-2 test (molecular assay or antigen) on or within 30 days before the incident S aureus culture date. ^eMRSA and MSSA cases were further categorized into 3 mutually exclusive epidemiologic classes: HO, if the incident S aureus culture was collected on or after the third calendar day of hospitalization; HACO, if the incident S aureus culture was collected from an outpatient setting or before the third calendar day of hospitalization from a patient with 1 or more risk factors: a history of hospitalization, surgery, dialysis, or residence in a long term care facility (LTCF) in the previous year or the presence of a central venous catheter within 2 days prior to the incident S aureus culture; or CA, if none of the previously mentioned criteria were met. ^fCases with incident S aureus culture dates within the date range were included in the analysis. MRSA surveillance began in 2005; MSSA surveillance began in 2016.

Prepandemic MSSA mortality rates ranged from 25% to 28% (HO), 15% to 18% (HACO), and 12% to 16% (CA) during 2016–2019 (Figure 1; Supplementary Table 5). Methicillin-susceptible Staphylococcus aureus HO mortality rates increased during the pandemic (from 28% in 2019 to 30% in 2020 and 39% in 2021) before declining in 2022. Overall, MSSA HO mortality rates increased by 41% from 2019 to 2021. Methicillin-susceptible Staphylococcus aureus HACO mortality rates were generally stable during the pandemic (ranging from 18% to 20% during 2019–2022). Methicillin-susceptible Staphylococcus aureus CA mortality rates were generally stable during the pandemic (ranging from 14% to 17% during 2019–2022). Among all epidemiologic classes, MSSA mortality rates for cases with recent SARS-CoV-2 infection were ∼2–3 times greater than mortality rates for cases without recent SARS-CoV-2 infection. The highest mortality rate (64%) occurred in 2021 among MSSA HO cases with recent SARS-CoV-2 infection.

Survival Analysis

A total of 15 055 cases and 3172 deaths were included (mortality rate: 21%) in the survival analysis examining the association between recent SARS-CoV-2 infection and mortality among SAB cases during 2020–2022; 1383 (9%) cases had recent SARS-CoV-2 infection. Recent SARS-CoV-2 infection was ∼3–4 times more common among HO cases (22% and 25% of MRSA and MSSA cases, respectively) compared to HACO (8% and 6%) and CA (6% and 5%) (Supplementary Tables 6 and 7). Overall mortality rates for cases with and without recent SARS-CoV-2 infection were 45% and 19%, respectively. The interaction between recent SARS-CoV-2 infection and year was not significant and was excluded. The proportional hazards assumption was not grossly violated, and predicted survival probabilities were similar to observed.

Across all strata, recent SARS-CoV-2 infection was strongly and significantly associated with 30-day all-cause mortality in both unadjusted and adjusted analyses (Table 1; Supplementary Tables 8–13 and Figure 3). Unadjusted hazard ratios (HRs) for recent SARS-CoV-2 infection ranged from 2.1 (95% CI: 1.7, 2.6) (HACO MSSA) to 2.8 (95% CI: 2.3, 3.4) (HO MSSA); adjusted HRs ranged from 1.9 (95% CI: 1.5, 2.3) (HACO MRSA) to 3.0 (95% CI: 2.4, 3.7) (HO MSSA). Hazard ratios did not change meaningfully in sensitivity analyses when MRSA data were restricted to MSSA counties, cases with missing data were excluded, or the definition of recent SARS-CoV-2 infection was varied (Supplementary Tables 14 and 15).

Table 1.

Hazard Ratios and 95% Confidence Intervals for Recent SARS-CoV-2 Infection^a From Cox Proportional Hazards Models Examining Associations Between Recent SARS-CoV-2 Infection and 30-Day All-Cause Mortality^b Among Adult Staphylococcus aureus Bacteremia Cases, Stratified by Methicillin Susceptibility and Epidemiologic Class, United States, 2020–2022

Susceptibilityc	Epi Classd	Total No. of Cases	Cases With Recent SARS-CoV-2 Infection			Recent SARS-CoV-2 Infection Hazard Ratio (95% CI)e
			No. of cases	Survived n (%)	Died n (%)	Unadjusted	Adjusted
MRSA	HO	1012	224	79 (35.3)	145 (64.7)	2.8 (2.2, 3.4)	2.7 (2.2, 3.4)
	HACO	4282	338	210 (62.1)	128 (37.9)	2.3 (1.9, 2.8)	1.9 (1.5, 2.3)
	CA	1490	87	58 (66.7)	29 (33.3)	2.3 (1.6, 3.5)	2.2 (1.5, 3.4)
MSSA	HO	1263	313	128 (40.9)	185 (59.1)	2.8 (2.3, 3.4)	3.0 (2.4, 3.7)
	HACO	4268	272	183 (67.3)	89 (32.7)	2.1 (1.7, 2.6)	2.1 (1.6, 2.6)
	CA	2740	149	106 (71.1)	43 (28.9)	2.3 (1.7, 3.2)	2.3 (1.7, 3.2)

Abbreviations: CA, community-associated; CI, confidence interval; HACO, healthcare-associated community-onset; HO, hospital-onset; MRSA, methicillin-resistant Staphylococcus aureus; MSSA, methicillin-susceptible Staphylococcus aureus.

^aRecent severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infection was defined as a positive viral SARS-CoV-2 test (molecular assay or antigen) on or within 30 days before the incident Staphylococcus aureus (S aureus) culture date.

^bThe outcome variable was all-cause mortality within 30 days of the incident S aureus culture date, as determined by patient medical record documentation and state death registry data.

^cCases were categorized as either MRSA or MSSA (ie, methicillin susceptibility), as determined by microbiologic testing.

^dCases were categorized into 3 mutually exclusive epidemiologic classes: HO, if incident S aureus culture was collected on or after the third calendar day of hospitalization; HACO, if incident S aureus culture was collected from an outpatient setting or before the third calendar day of hospitalization from a patient with 1 or more risk factors: a history of hospitalization, surgery, dialysis, or residence in a long-term care facility (LTCF) in the previous year or the presence of a central venous catheter within 2 days prior to incident S aureus culture; or CA, if none of the previously mentioned criteria were met.

^eModels were stratified by methicillin susceptibility (MRSA or MSSA) and epidemiologic class (HO, HACO, or CA) and included random intercepts for state and patient identifier. Unadjusted models included a fixed variable for recent SARS-CoV-2 infection. Adjusted models included fixed variables for recent SARS-CoV-2 infection, year, race/ethnicity, age, female sex, Charlson Comorbidity Index, body mass index, smoking, alcohol abuse, substance abuse, and homelessness. Models for HO and HACO also included LTCF residency.

Compared to MSSA CA cases, MRSA CA cases were more likely to be current smokers (46% vs 31%), have a substance use disorder (36% vs 23%), and be experiencing homelessness (17% vs 11%) (Supplementary Tables 6 and 7). When MRSA data were restricted to MSSA counties, these differences between MRSA and MSSA CA cases persisted (Supplementary Table 16).

Effects of the Coronavirus Disease 2019 Pandemic on Staphylococcus aureus Bacteremia Mortality

For the direct effect (how recent SARS-CoV-2 infection changed SAB mortality), cases during the pandemic with recent SARS-CoV-2 infection had 2.1 (95% CI: 1.8, 2.4)–2.9 (95% CI: 2.0, 4.0) times the risk of dying compared to cases without recent SARS-CoV-2 infection, depending on methicillin susceptibility and epidemiologic class (Table 2). For the indirect effect (how the COVID-19 pandemic influenced mortality for patients without recent SARS-CoV-2 infection), the only significant effect was among MRSA CA cases; during the pandemic, MRSA CA cases without recent SARS-CoV-2 infection had 1.3 (95% CI: 1.1, 1.7) times the risk of dying compared to MRSA CA cases before the pandemic. For the total effect (the direct effect of recent SARS-CoV-2 infection combined with the indirect effect of the COVID-19 pandemic), mortality rate ratios ranged from 2.0 (95% CI: 1.7, 2.4) to 3.8 (95% CI: 2.6, 5.3). Finally, for the overall effect (mortality before vs during the COVID-19 pandemic), the effect was significant for all epidemiologic classes combined (mortality rate ratio: 1.2, [95% CI: 1.1, 1.3] for both MRSA and MSSA). For specific epidemiologic classes, the overall effect ranged from 1.0 (95% CI: 0.9, 1.1) (no effect) to 1.5 (95% CI: 1.2, 1.7). Finally, in a sensitivity analysis restricting MRSA data to MSSA counties, results did not change meaningfully; however, the indirect effect of the COVID-19 pandemic on MRSA CA mortality was slightly attenuated (Supplementary Table 17).

Table 2.

Effects of the COVID-19 Pandemic on Mortality Among Adult Patients With Staphylococcus aureus Bacteremia, United States, 2018–2021

Susceptibilitya	Epi Classb	Mortality Rate (%)c				Effect Estimate (Mortality Rate Ratio [95% CI])d
		Before pandemice	During pandemicf			Directh	Indirecti	Totalj	Overallk
			All cases	Cases with recent SARS-CoV-2 infectiong	Cases without recent SARS-CoV-2 infectiong
All	All	18.4	22.0	52.2	19.2	2.7 (2.5, 2.9)	1.0 (1.0, 1.1)	2.8 (2.6, 3.1)	1.2 (1.1, 1.3)
MRSA	All	20.1	23.7	54.1	20.7	2.6 (2.3, 2.9)	1.0 (0.9, 1.1)	2.7 (2.4, 3.0)	1.2 (1.1, 1.3)
	HO	31.6	43.3	71.9	34.5	2.1 (1.8, 2.4)	1.1 (0.9, 1.3)	2.3 (1.9, 2.7)	1.4 (1.2, 1.6)
	HACO	21.1	21.5	42.1	19.9	2.1 (1.8, 2.5)	0.9 (0.9, 1.0)	2.0 (1.7, 2.4)	1.0 (0.9, 1.1)
	CA	11.4	16.5	43.8	15.1	2.9 (2.0, 4.0)	1.3 (1.1, 1.7)	3.8 (2.6, 5.3)	1.4 (1.2, 1.8)
MSSA	All	16.9	20.6	50.4	17.9	2.8 (2.5, 3.1)	1.1 (1.0, 1.2)	3.0 (2.7, 3.3)	1.2 (1.1, 1.3)
	HO	24.6	36.3	61.5	26.6	2.3 (1.9, 2.7)	1.1 (0.9, 1.3)	2.5 (2.1, 3.0)	1.5 (1.2, 1.7)
	HACO	16.9	19.2	40.5	18.2	2.2 (1.7, 2.8)	1.1 (1.0, 1.2)	2.4 (1.9, 3.0)	1.1 (1.0, 1.3)
	CA	14.2	15.3	31.9	14.6	2.2 (1.5, 3.0)	1.0 (0.9, 1.2)	2.2 (1.5, 3.1)	1.1 (0.9, 1.3)

Abbreviations: CA, community-associated; CI, confidence interval; HACO, healthcare-associated community-onset; HO, hospital-onset; MRSA, methicillin-resistant Staphylococcus aureus; MSSA, methicillin-susceptible Staphylococcus aureus.

^aCases were categorized as either MRSA or MSSA (ie, methicillin susceptibility), as determined by microbiologic testing.

^bCases were categorized into 3 mutually exclusive epidemiologic (epi) classes: HO, if incident Staphylococcus aureus (S aureus) culture was collected on or after the third calendar day of hospitalization; HACO, if incident S aureus culture was collected from an outpatient setting or before the third calendar day of hospitalization from a patient with 1 or more risk factors: a history of hospitalization, surgery, dialysis, or residence in a long-term care facility in the previous year or the presence of a central venous catheter within 2 days prior to incident S aureus culture; or CA, if none of the previously mentioned criteria were met.

^cStrata-specific mortality rates were calculated by dividing the number of S aureus bacteremia cases that died from all causes within 30 days of incident culture by the total number of S aureus bacteremia cases with an incident S aureus culture for a given timeframe.

^dEffect estimates indicate the effect of the pandemic on all-cause 30-day mortality among patients and were quantified by mortality rate ratios. Confidence intervals were calculated using the Koopman asymptotic score method.

^eBefore pandemic indicates the time period before the coronavirus disease 2019 (COVID-19) pandemic and includes cases with incident S aureus culture dates 21 March 2018–31 December 2019. All of these cases are presumed to not have had recent severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infection.

^fDuring pandemic indicates the time period during the COVID-19 pandemic and includes cases with incident S aureus culture dates 21 March 2020–31 December 2021.

^gRecent SARS-CoV-2 infection was defined as a positive viral SARS-CoV-2 test (molecular assay or antigen) on or within 30 days before the incident S aureus culture date.

^hThe direct effect was calculated by dividing the mortality rate of cases during the COVID-19 pandemic with recent SARS-CoV-2 infection by the case mortality rate of cases during the COVID-19 pandemic without recent SARS-CoV-2 infection.

ⁱThe indirect effect was calculated by dividing the mortality rate of cases during the COVID-19 pandemic without recent SARS-CoV-2 infection by the mortality rate of cases before the COVID-19 pandemic.

^jThe total effect was calculated by dividing the mortality rate of cases during the COVID-19 pandemic with recent SARS-CoV-2 infection by the mortality rate of cases before the COVID-19 pandemic.

^kThe overall effect was calculated by dividing the mortality rate of cases during the COVID-19 pandemic by the mortality rate of cases before the COVID-19 pandemic.

DISCUSSION

After several years of little change, there was a modest increase in overall SAB mortality during the early years of the COVID-19 pandemic. Increases in overall SAB mortality were driven primarily by increases in HO mortality (MRSA and MSSA) and CA mortality (MRSA only). Mortality rates for HACO (both MRSA and MSSA) and CA MSSA did not change meaningfully.

Mortality rates for SAB cases with recent SARS-CoV-2 infection were 2–3 times greater than for SAB cases without recent SARS-CoV-2 infection during 2020–2021. This difference was attenuated in 2022 but persisted. Because recent SARS-CoV-2 infection was much more common among HO cases than HACO or CA cases, it drove increases in HO mortality rates. Among all epidemiological classes, even after adjusting for demographics, underlying conditions, and other risk factors in survival analyses, the risk of death among cases with recent SARS-CoV-2 infection was 2–3 times greater compared to cases without recent SARS-CoV-2 infection. This strong association between SARS-CoV-2 infection and death among SAB cases is consistent with other studies [13, 14].

Our overall 30-day mortality rate of 21% among SAB cases was comparable to a global meta-analysis that reported a pooled mortality rate of 19.6% (95% CI: 18.4%–20.9%) during 1991–2021 [5]. Our MRSA-specific mortality rate (22%) was slightly lower than that reported in the global meta-analysis (26.5% [95% CI: 24.5%–28.5%]), whereas our MSSA-specific mortality rate (18%) was comparable (18.5% [95% CI: 16.9%–20.3%]). Unlike the global meta-analysis, in which mortality rates declined over time, pre-2020 mortality rates in our study appeared in figures to be relatively stable. This could be attributed to differences in case surveillance, ascertainment of death, infection control practices, case management, and years included in other countries and studies.

We found that mortality rates during the pandemic increased substantially for HO SAB cases, primarily due to a high prevalence of recent SARS-CoV-2 infection among these cases. A previous study using the same data source found that HO SAB cases with recent SARS-CoV-2 infection were significantly more likely to be in the intensive care unit or have a central venous catheter in place before SAB onset [11]. This suggests that HO SAB cases with recent SARS-CoV-2 infection were likely patients who had been hospitalized for severe COVID-19 and were already critically ill when they acquired SAB in the hospital, potentially from exposure to critical care environments and healthcare interventions [8, 10, 13]. Mortality rates for these patients were extremely high; approximately three-fourths of HO MRSA cases and almost two-thirds of HO MSSA cases with recent SARS-CoV-2 infection died early in the pandemic. This is similar to a previous study, which reported a mortality rate of 62% in patients coinfected with SARS-CoV-2 and S aureus [13]. Although SAB alone is associated with substantial mortality, exceeding 35% in some years among HO cases in our study, and in-hospital mortality rates for patients with COVID-19 were as high as 16% in 2020 in a previous study [26], the extremely high mortality rates in our study suggest that SARS-CoV-2 infection and SAB have a more than additive effect on mortality.

Mortality rates during the pandemic also increased for MRSA CA cases, which could have been driven by indirect effects of the pandemic rather than by recent COVID-19. In contrast, mortality rates did not increase among MSSA CA cases. This difference in trends may be due to a greater indirect effect of the COVID-19 pandemic on MRSA CA cases, potentially due to differences in underlying risk factors. For example, MRSA CA cases were more likely than MSSA CA cases to have a substance use disorder and be experiencing homelessness, which may have contributed to delays in seeking care and poorer health outcomes.

We note several limitations with the analysis. First, when calculating mortality rates, we were unable to distinguish between deaths due to SAB from deaths due to other causes. We used a 30-day follow-up time to minimize the inclusion of deaths that were not attributed to SAB. Second, we were unable to distinguish between COVID-19 and asymptomatic SARS-CoV-2 infection, or between current and recent SARS-CoV-2 infection. Moreover, there is no established definition of recent SARS-CoV-2 infection. However, in a sensitivity analysis varying the definition of recent SARS-CoV-2 infection, there was no meaningful change in the association between SARS-CoV-2 infection and mortality. Therefore, it is likely that both current and recent SARS-CoV-2 infection up to 60 days prior to the incident culture date negatively affect SAB survival. Third, changes in SARS-CoV-2 testing practices over time (eg, screening testing on hospital admission) likely affected SARS-CoV-2 infection detection rates. To address this, we included infection year in the regression model and excluded 2022 cases when examining the effects of the COVID-19 pandemic on SAB mortality. Fourth, for positive SARS-CoV-2 tests not documented in the medical record, ascertainment of SARS-CoV-2 testing data may have differed between sites depending on state reporting systems. Fifth, there may be residual confounding from patient acuity. However, by controlling for comorbidities, age, and other variables, we expected this residual confounding to be minimal. Sixth, 2 states (3 counties) had noncontinuous reporting for MRSA data and did not begin reporting MSSA data until 2017. However, reporting was continuous for all counties starting in 2017. Finally, much of the data were collected through medical record review and were subject to limitations of those data sources. However, a strength of our study was the use of vital statistics data, which captured more than 4000 deaths not recorded in the medical record.

In summary, recent SARS-CoV-2 infection was strongly associated with death among SAB cases, even after controlling for demographics and other risk factors. During the pandemic, mortality rates increased substantially among HO SAB cases, a group in which recent SARS-CoV-2 infection was more prevalent and which had considerable exposure to intensive healthcare interventions. Mortality rates also increased among CA SAB cases, but only among MRSA cases and not MSSA cases, potentially due to greater indirect effects of the pandemic on mortality among these cases. Mortality rates did not increase among HACO SAB cases. These findings emphasize the importance of infection prevention in patients hospitalized with SARS-CoV-2 infection and could lend support for greater infection prevention measures among patients early in a similar respiratory virus pandemic to reduce secondary SAB and ultimately prevent deaths.

Supplementary Data

Supplementary materials are available at Open Forum Infectious Diseases online. Consisting of data provided by the authors to benefit the reader, the posted materials are not copyedited and are the sole responsibility of the authors, so questions or comments should be addressed to the corresponding author.