Antibiotic Resistant Infections among COVID-19 Inpatients in U.S. Hospitals

Abstract We described bacterial/fungal co-infections and antibiotic resistant infections among inpatients diagnosed with COVID-19 and compared findings with inpatients diagnosed with influenza-like-illness. Less than 10% of COVID-19 inpatients had bacterial/fungal co-infection. Longer lengths of stay, critical care stay, and mechanical ventilation contribute to increased incidence of hospital-onset infections among COVID-19 inpatients.

during surges of COVID-19 hospitalizations, our objective was to determine the rate of bacterial or 7 fungal co-infections and certain antibiotic resistant infections among inpatients with 7]. 8 Further, we compared these findings with a cohort of inpatients diagnosed with influenza-like-illness 9 (ILI) discharged during 2019. 10 11 Methods: 12 We conducted a retrospective study using adult and pediatric inpatient discharge and 13 microbiology data from U.S. hospitals included in the Premier Healthcare Database Special  Release (PHD-SR, release date 10/12/2021) [8]. The PHD-SR contains discharge records for all 15 inpatients discharged from participating acute care, general, non-Federal US hospitals. Inpatient visit 16 records included diagnostic and procedure codes, demographic information, admission and discharge 17 dates, and facility characteristics [9]. Hospitals were limited to those with available microbiology data.

18
Microbiology data included detailed information such as genus and species of bacterial isolates, day and 19 time stamp of specimens, and associated antimicrobial sensitivity testing results [8].  To identify co-infections among the COVID-19 and ILI cohorts, we first identified patients with 6 at least one microbiology specimen collected between 3 days prior to admission and 3 days post 7 discharge. Specimens with a test result of "bacteria identified", "fungus identified", or with an 8 antimicrobial sensitivity result were considered positive bacterial or fungal cultures. We retrospectively 9 calculated proportions of inpatients with a positive bacterial or fungal culture among inpatients with ILI 10 and COVID-19. Organisms identified among specimens with a sensitivity result taken from ILI and 11 COVID-19 inpatients were described with frequencies.

12
In order to examine antibiotic resistance, we further limited specimens to those that yielded an 13 organism of interest and had susceptibility testing results to determine whether the isolate had the inpatients with more than one positive clinical culture, cultures from a sterile body site were prioritized 24 over those with non-sterile body site specimen sources within 14 days. Clinical cultures were defined as community-onset (CO) when the culture was taken between the 3 days preceding admission and the first 1 3 days after admission. Clinical cultures taken on day 4 or later after admission were considered 2 hospital-onset (HO). 3 We calculated the rates of resistant infections as the number of positive specimens with the 4 resistance phenotype of interest per 10,000 discharges stratified by epidemiology classification 5 (HO/CO). In addition, we compared the rates of infections for each phenotype and epidemiology 6 classification using a multivariable logistic model and adjusting for age group, gender, race/ethnicity for 7 both CO and HO infections. To evaluate whether differences in HO AR infections between ILI and 8 COVID-19 inpatients were driven by higher severity of illness, we further adjusted for length of stay, 9 critical care stay (yes or no), and mechanical ventilation (yes or no) in secondary models. Models 10 account for inter-facility correlation using generalized estimating equations.

11
This activity was reviewed by CDC and was conducted consistent with applicable federal law  Table A2). On average, inpatients 21 with COVID-19 had longer lengths of stay (LOS) than inpatients with ILI (mean: 8.3 vs. 6.1 days). The

22
proportion of inpatients who spent at least one day in a critical care unit among those diagnosed with 23 COVID-19 was 48.3% compared with 46.4% among inpatients with ILI. Among inpatients with -19, 13.0% had at least one day of invasive mechanical ventilation compared with 10.2% of 1 inpatients diagnosed with ILI.

2
The proportion of inpatients with a bacterial or fungal culture obtained was similar for COVID-3 19 and ILI (56.2% and 60.4%, respectively). The percent of discharges with a positive culture 4 categorized as CO was lower among inpatients with COVID-19 compared with inpatients with ILI 5 (7.0% vs. 10.4%). However, the percent of discharges with a positive culture categorized as HO was 6 higher among inpatients with COVID-19 (4.1% vs. 2.4%). The most common organisms identified 7 among inpatients diagnosed with ILI and COVID-19 were similar while HO sterile specimen sources for 8 inpatients with COVID-19 were more common (Supplemental Appendix Table A3 and Table A4). CO 9 infection rates of MRSA, ESBL, CRE, VRE, CRAB, and CRPA tended to be lower, but HO infection 10 rates were higher among COVID-19 inpatients compared with those diagnosed with ILI across all 11 pathogens (Table 1).

12
Multivariable logistic models showed significantly lower odds of CO MRSA, CRPA, and CRAB 13 among inpatients diagnosed with COVID-19 compared with inpatients with ILI (Table 1) Table A5).

Discussion:
19 Based on our findings, the incidence of CO bacterial and fungal co-infection, including with 20 antibiotic resistant pathogens, among patients admitted with COVID-19 was less than 10% of patients,  We used a large administrative dataset representing information from a diverse sample of 8 hospital sizes, teaching status, urban/rural locations, and geographic divisions. However, our study had 9 several limitations: the study did not include hospitals located in the Mountain U.S. census division; 10 administrative data are collected primarily for billing purposes and adapted for research resulting in 11 possible misclassification in clinical and facility information, which is most likely non-differential; and 12 the study did not include molecular diagnostics for identification of potential infections. However, we 13 defined antibiotic resistant pathogens using previously validated methods for identifying these infections  .89, J12.9, J18, J18.1, J18.8, J18.9, J20.9, J40, R05, R50.9.
2 COVID-19 discharges were defined as hospitalizations with a primary or secondary ICD-10-CM code of U07.1 discharged during April 2020-June 2021 or a primary or secondary ICD-10-CM code of B97.29 discharged during March-April 2020 and admitted during February-April 2020.
3 Multivariable logistic models adjusted for gender, age group, race, and ethnicity.