Our study spanning 8 months prior to the COVID-19 pandemic through October 2021 and encompassing almost 300,000 hospital admissions with positive bacterial cultures found that SARS-CoV-2-positive patients had significantly higher rates of antibiotic IET compared with the overall patient population during the pre-pandemic period and with SARS-CoV-2-negative and not tested patients during the pandemic period. We further observed elevated IET rates in patients with AMR- or MDR-positive cultures; the IET rate for MDR pathogens in this study (45% for evaluated pathogens across all culture sources) was similar to rates recently reported for patients with bloodstream infections caused by carbapenem-resistant Enterobacterales (44.7%) or VRE (39.6%) [7], providing further confirmation of these data. Our finding that IET was associated with additional DOT and longer hospital and ICU stays is consistent with other studies[1–3, 5, 6, 22] and documents the substantial burden not only for patients receiving IET, but also for hospital facilities, particularly during surge capacity periods.
It is clear from our data that AMR is closely connected with IET. We found a 2.59-fold increased risk of IET with AMR-positive cultures and an MDR-positive culture was associated with an additional 1.86-fold increase in IET compared with AMR-positive cultures. Over one-third (34.9%) of patients with AMR-positive cultures and 45.0% of patients with MDR-positive cultures received IET upon hospital admission, a finding that highlights difficulties in choosing empiric therapy for patients with potentially resistant pathogens. Of the many factors we assessed in multivariate analyses, including age, comorbidities, isolate source, and ventilator/ICU status, only specific, highly-resistant pathogens (Enterococcus and S. maltophilia/Acinetobacter complex) were associated with a higher risk of IET than AMR. Other factors with significant contributions to high IET rates included respiratory source, hospital-onset infections, and certain underlying conditions (heart failure/myocardial injury, liver dysfunction, and renal failure/insufficiency), older age, and prior admissions. IET risk factors identified in our study may serve as predictors of patients in need of aggressive initial antibiotic treatment and as candidates for future studies aimed at optimizing initial therapy choices in high-risk patients.
The SARS-CoV-2 pandemic has further exacerbated challenges with IET. Although overall IET rates remained similar in the pre-pandemic and pandemic periods, during the SARS-CoV-2 period we observed significant increases in IET over pre-pandemic rates in patients with AMR-positive cultures and those who were SARS-CoV-2 positive; hospital-onset infections were a key contributing factor. Patients with SARS-CoV-2 and bacterial infections are at higher risk for mortality than SARS-CoV-2-negative patients [1, 23] and have more antimicrobial usage and longer hospital and ICU LOS [8]. IET, which is associated with increased LOS and mortality [1, 7], likely contributes to these impaired outcomes in SARS-CoV-2-positive patients.
The strong association between AMR and IET was retained during both the pre-pandemic and pandemic periods. A recent report estimated that bacterial AMR was associated with an estimated 4.95 million deaths worldwide in 2019 [24]. There is evidence that the COVID-19 pandemic may have contributed to increases in bacterial AMR in hospitalized patients [17, 25]. During peak capacity periods associated with the COVID-19 pandemic, hospital systems were forced to decrease diagnostic and antimicrobial susceptibility testing and reallocate staff from antimicrobial stewardship activities to COVID-related priorities while at the same time increasing antibiotic consumption [13]. The increased AMR rates we have observed in hospitalized SARS-CoV-2-positive patients may be a reflection of these factors [17]. Future longitudinal studies will be needed to explore ongoing changes in IET and AMR in US hospitals. A study of 38 Michigan hospitals found that early increases in antibiotic consumption diminished over time as more experience was gained with managing SARS-CoV-2 infections [26], perhaps augmented by the less severe disease observed later in the pandemic [27]. Lag times of approximately 3 to 6 months have been reported between antibiotic changes and resistance levels in a pre-pandemic study [28], so reductions in AMR may not be immediately apparent.
Irrespective of pandemic-related factors, however, IET remains an important challenge in the treatment of hospitalized patients. Although one option to address this problem is expanded use of broad-spectrum antibiotics, these antibiotics are associated with increased AMR and therefore have the potential to actually compound IET challenges. In addition, broad-spectrum drugs can have negative clinical consequences, including increased risk of Clostridioides difficile infection [29, 30] and higher rates of severe sepsis following hospital discharge [31]. Accordingly, the solution to IET does not appear to be indiscriminate broad-spectrum antibiotic use, but rather a tailored therapy approach based on risk factor assessment, diagnostic testing, and antimicrobial stewardship efforts, including vaccination programs to reduce infectious diseases [32–35]. Although the study reported here focused on IET in culture-positive patients, it is important to note that a substantial proportion of culture-negative or not tested hospitalized patients receive prolonged antibiotic therapy (> 3 days) [17]. This patient population would benefit from antimicrobial stewardship programs as well.
Study limitations include the use of facility reports rather than a central laboratory for SARS-CoV-2 and antimicrobial susceptibility tests. Different laboratories may have used different testing systems and breakpoints for determination of resistance, thereby potentially affecting the AMR and MDR rates reported here. Analyses were based on positive SARS-CoV-2 tests and not on symptomatic infection, so asymptomatic patients admitted for other causes may have been included. Similarly, patients with positive bacterial cultures did not necessarily have a confirmed bacterial infection. However, our testing algorithm was designed to remove admissions with contaminating or colonizing bacteria [20]. Selection bias (e.g., a greater likelihood of collecting bacterial culture data on more severely ill patients) may have influenced reported resistance rates. Certain geographic areas and smaller hospitals may have been underrepresented in our database.