A multicenter analysis of the clinical microbiology and antimicrobial usage in hospitalized patients in the US with or without COVID-19

Coronavirus disease 2019 (COVID-19), which is caused by severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), emerged in Wuhan, China in December of 2019 and became a pandemic. Past respiratory virus epidemics suggest an increased risk for bacterial, fungal, or other viral coinfections or superinfections due to biological factors, underlying conditions, and healthcare exposures [1, 2]. Varying coinfection and superinfection rates have been reported in patients with COVID-19, ranging from 5 to 33% depending on methods, populations, and geographic location (reviewed by Clancy and Nguyen [3] and Rawson et al. [4]). Overall, there is currently limited and often divergent information available on the frequency and microbiology of coinfections and superinfections in patients with COVID-19 [5]. Differences in the methods used for testing criteria, obtaining samples (particularly for nasopharyngeal sources), and storing specimens complicates data comparison and interpretation.

There are also limited data on antimicrobial usage in COVID-19 patients; available studies suggest that the use of broad-spectrum antibiotics is widespread in this population [4]. Such treatment patterns have potential implications for development of antibiotic resistance in both hospital and community settings [6], increased risk for other infections, such as Clostridium difficile, and potential adverse events and toxicity. More complete data on pathogen rates and antimicrobial usage are needed to support appropriate clinical management and antimicrobial stewardship policies during the pandemic [3, 4].

In this study, we used data from a large US hospital database to analyze and compare pathogen rates, antimicrobial utilization patterns, and health outcomes among SARS-CoV-2-positive, SARS-CoV-2-negative, and SARS-CoV-2-untested hospitalized adult patients.

From March 1 to May 31, 2020, 141,621 of 590,960 patients (24.0%) were tested for SARS-CoV-2 in 241 medical facilities and 449,339 (76.1%) were not tested. A total of 17,003 (12.0%) tested patients were positive for SARS-CoV-2 and 124,618 (88.0%) were negative (Table 1). The proportion of male patients was significantly higher in SARS-CoV-2-positive (53.1%) vs –negative (46.5%) or untested (44.5%) patients. Mean (standard deviation [SD]) age was significantly higher in the SARS-CoV-2-positive group (61.7 [18.0] vs 58.5 [20.9] years for SARS-CoV-2-negative and 48.7 [27.3] years for untested patients). Medical facility characteristics and geographic regions by SARS-CoV-2 status are shown in Supplementary Table 1 (see Additional file 1). Of the 241 facilities, 82.6% were urban hospitals; the bed count was < 100 in 34.9%, 100 to 300 in 39.4%, and > 300 in 25.7%. The census division with the greatest number of facilities was West South Central (25.3%) followed by East North Central (17.4%). At the patient level, the highest SARS-CoV-2-positive rates were in the Northeast (4392/22,364 tested patients [19.6%]) and in hospitals that were urban (16,649/131,726 [12.6%]), teaching hospitals (10,405/78,250 [13.3%]), and with > 300 beds (10,475/79,922 [13.1%]).

Most patients tested for SARS-CoV-2 had specimens collected for microbiologic testing for additional pathogens, although the rate was slightly higher for SARS-CoV-2-positive patients (97.8%) compared with SARS-CoV-2-negative patients (91.9%). Among those not tested for SARS-CoV-2, 41.9% of patients had specimens collected for microbiologic testing. The proportion of patients with non-SARS-CoV-2 pathogens detected was highest in patients not tested for SARS-CoV-2 (27.9%) and similar between the SARS-CoV-2-positive (20.9%) and −negative (21.3%) groups (Table 1). The presence of multiple non-SARS-CoV-2 pathogens was slightly more common in SARS-CoV-2-negative patients (37.2% versus 34.3% for SARS-CoV-2-positive and 34.1% for untested patients) (Table 1).

Experts have identified an urgent need for accurate information on potential coinfection and superinfection rates in patients with SARS-CoV-2 to help inform diagnosis, disease management, and antimicrobial stewardship during the COVID-19 pandemic [3‐6]. In our study, there was a high rate of testing for non-SARS-CoV-2 pathogens among those tested for SARS-CoV-2 infection; approximately 20% of patients tested for SARS-CoV-2 had a positive specimen for an additional pathogen, regardless of SARS-CoV-2 status. However, SARS-CoV-2-positive patients had higher rates of hospital-onset pathogens, greater antimicrobial usage, and extended hospital and ICU LOS compared with SARS-CoV-2-negative or −untested patients.

During our 3-month study (March to May), 12% of hospitalized adult patients tested for SARS-CoV-2 in US acute care facilities were positive for this virus. As might be expected, testing for additional pathogens was far more common in patients tested for SARS-CoV-2 (more than 90%) compared with patients that were not tested for SARS-CoV-2 (42%), as many untested patients were likely admitted for noninfectious causes with no reason to expect or test for a clinically relevant pathogen. Nevertheless, the rate of pathogen detection was slightly higher among untested patients (28% versus 21%). Urine was the most frequent specimen source for pathogens in all groups, followed by respiratory samples for SARS-CoV-2-positive patients. The relatively low yield of non-SARS-CoV-2 pathogens across all subgroups may be relevant to hospitals coping with the redeployment of diagnostics to focus on the COVID-19 pandemic [4, 11].

In presenting these data, it is important to note that we do not consider the SARS-CoV-2-negative or −untested subgroups to be “controls” for the SARS-CoV-2-positive patients in this study, as the subgroups differed in many important characteristics including the proportion of patients with hospital-onset pathogens. Rather, we believe the inclusion of these other subgroups provides context for the findings in the SARS-CoV-2 tested subgroup.

Other studies have come to varying conclusions concerning rates of coinfection in patients with SARS-CoV-2 [11‐15]. The differences among studies may relate to study location and to testing procedures. The strength of our study is use of a broader approach on two levels: 1) we assessed all specimens collected during the admission, rather than focusing solely on respiratory pathogens; and 2) we included regionally diverse hospitals throughout the US. Given the extensive geographic variations in SARS-CoV-2 infection rates during the course of the pandemic [16, 17] and region-specific patient characteristics [18], national data may provide a more complete view of SARS-CoV-2 infections and attendant coinfections.

Rates of specific pathogens were generally comparable among the three groups, although rates of some bacteria, including P. aeruginosa, and rates of Candida spp. were significantly higher in SARS-CoV-2-positive patients compared with −negative or −untested patients. Both pathogens are frequently found in the respiratory tract of ventilated patients or critically ill patients, and both are associated with previous antibiotic treatment [19]. In contrast, rates of Aspergillus were highest in SARS-CoV-2-negative patients. Variability in methods used to detect Aspergillus impact reported rates of infection [20]. Studies using various methods have reported pulmonary aspergillosis rates of 20 to 30% in the most severely ill, mechanically ventilated COVID-19 patients [20], but the overall rate is much lower. In line with the numbers reported here, a recent study identified COVID-19-associated pulmonary aspergillosis in 0.3% of all hospitalized COVID-19 patients [21].

In SARS-CoV-2-positive patients, 58% of non-SARS-CoV-2 pathogens were identified in specimens collected within 3 days of admission. The delayed onset of the remaining 42% of positive pathogen detections indicates the increased risk of healthcare exposure for the acquisition of opportunistic pathogens and/or healthcare-associated infections. In contrast, positive pathogen detection in specimens collected within 3 days of admission occurred at higher rates in SARS-CoV-2-negative (78%) and −untested (81%) patients, and rates of subsequent hospital-onset positive pathogen detections were correspondingly reduced. The different temporal distribution of positive specimens among these three groups suggests that characteristics associated with hospitalized COVID-19 patients [3], including comorbidities, the severity of illness associated with SARS-CoV-2, extended hospital lengths of stay, and the frequent need for ventilation [15, 22], perhaps in conjunction with immunomodulatory treatments, may increase susceptibility to infection with other pathogens [3]. In addition, prolonged SARS-CoV-2 infection could itself potentially confer an immunosuppressed state that increases vulnerability to opportunistic infections [23]. Rates of healthcare-associated infections may have also been impacted by the heavy burden on hospitals during COVID-19 surges, leading US medical professional societies to urge the Department of Health and Human Services to suspend reimbursement penalties for healthcare-associated infections during the pandemic [24]. Given the fact that urine and respiratory sources predominated as sources of pathogens in SARS-CoV-2-positive patients, it is feasible that infection prevention bundles and the use of personal protective equipment may have been adversely affected as a result of the pandemic [25, 26].

Histopathologic studies support an important role for S. pneumoniae and other Gram-positive pathogens, including Streptococcus pyogenes and S. aureus, in contributing to lung damage and causing bacterial pneumonia in patients with influenza [27]. Gram-negative pathogens have been identified as the major cause of bacterial pneumonia in critically ill patients with COVID-19 [28]. In addition to lung damage, pathogens can influence a number of other pathologic processes. For instance, during the influenza A H1N1 pandemic in 2009, coinfection with S. aureus was reported to be associated with an increased risk of severe coagulopathy in children [29]. It should be noted, however, that the correlation between copathogens and other severe clinical outcomes is not necessarily causative; the presence of copathogens might instead act as a marker for more severe chronic conditions or reflect differences in individual immune responses.

In our study, SARS-CoV-2-positive patients had longer hospital and ICU LOS compared with SARS-CoV-2-negative or −untested patients. Identification of a pathogen was associated with increased LOS in all groups, but was particularly notable in SARS-CoV-2-positive patients, who had a mean hospital LOS of almost 14 days in the SARS-CoV-2-positive/pathogen specimen-positive group compared with 7 days in the SARS-CoV-2-positive/pathogen specimen-negative group. The presence of multiple pathogens further increased LOS. These data are derived from patients with additional pathogens identified during both the admission onset and hospital onset periods.

Antimicrobial usage was significantly higher in SARS-CoV-2-positive patients (68% vs 45% for SARS-CoV-2-negative and 25% for −untested patients), despite similar or lower rates of pathogen-positive specimens. The higher rate of antimicrobial use may be due to initiation of empiric therapy for suspected pneumonia or sepsis, consistent with guidelines for the latter [30]. Other studies have also observed higher utilization of antibiotics in patients with SARS-CoV-2 with no identified coinfection [4, 31, 32]. Although most drug classes were used at comparable levels in the different groups, the use of macrolides was much higher in SARS-CoV-2-positive patients. This may be due to the widely reported increase in radiographic opacities in COVID-19 patients concurrent with signs and symptoms of pneumonia at presentation [33]. Macrolide use was also likely influenced by early reports concerning the potential efficacy of hydroxychloroquine plus azithromycin in managing COVID-19 [34], which were not supported by subsequent studies [35, 36].

The treatment patterns observed in our study likely reflect difficulties in making treatment decisions in critically ill patients, prolonged turnaround time for SARS-CoV-2 testing, particularly in the early days of the pandemic [37], and concerns over potential bacterial superinfection in SARS-CoV-2, but may also indicate overuse of antimicrobials. Collateral damage from antimicrobial overuse includes increased selection of antimicrobial resistance and opportunistic pathogens such as C. difficile, adverse effects of drugs, and unnecessary treatment costs [3, 4, 9, 38]. Continued monitoring of the utilization and appropriateness of antimicrobial use among SARS-CoV-2 patients is required to identify antimicrobial stewardship opportunities.

As with any database study, the study reported here has important limitations. SARS-CoV-2 and pathogen status were based on reports from institutional facilities; there was no uniform method of testing or central laboratory. The sensitivity and specificity of SARS-CoV-2 PCR assays are known to vary based on several factors, including time from exposure [39, 40]. No case definition for COVID-19 disease was applied, consistent with current medical care practices, and we did not assess disease severity, so it is possible that some of the SARS-CoV-2 patients identified here were asymptomatic and admitted for other causes. Although our established algorithm [10] is designed to remove admissions with colonizing microbes from the analyses, some of the pathogens resulting in positive specimens may not have been associated with clinically significant infections. Certain geographic areas may have been underrepresented by our database, although the northeast, which had high SARS-CoV-2 rates in the earliest phase, is well represented. Influenza is not common during the months included in this study, so our findings concerning co-infection with respiratory viruses may not be representative of other times of the year. Our study reflects an early period in the COVID-19 pandemic; it will be important to follow changes in these patterns over time to assess changes in virus epidemiology and in healthcare resource utilization.

The findings reported in our study raise several important points of clinical relevance. First, potential copathogens (concurrent infections or superinfections) occur in approximately 21% of SARS-CoV-2 patients, so the identification of another pathogen does not rule out the presence of SARS-CoV-2. Second, although overall potential copathogen rates were similar regardless of SARS-CoV-2 status, clinicians should be aware of the increased risk of secondary infections in patients with SARS-CoV-2 and the burden associated with copathogens. The presence of an additional pathogen was associated with a marked increase in hospital and ICU LOS for all groups, but the difference was greatest in SARS-CoV-2-positive patients. Third, about 80% of SARS-CoV-2-positive patients do not have a detectable copathogen, so antimicrobial therapy should be evaluated daily and de-escalated when possible. Our data underscore the need for sensitive and specific diagnostics to help inform appropriate antimicrobial stewardship. The deployment of early respiratory diagnostic tests relative to the first administration of an antimicrobial is therefore a prime consideration, especially for septic patients who will likely receive an antimicrobial within the first hour of acute care.

During the 3-month period encompassed by this study, SARS-CoV-2-positive patients had specimens collected more than twice as often as patients not tested for SARS-CoV-2. Additionally, SARS-CoV-2-positive patients were 1.5-fold more likely to be prescribed antibiotics than SARS-CoV-2-negative patients, and 2.7-fold more likely than untested patients. Our findings document epidemiological variations in specimen collection, pathogen prevalence, and antimicrobial utilization associated with SARS-CoV-2 status that may be useful in helping clinicians assess and manage patients with suspected or confirmed COVID-19, and highlight the considerable operational burden that COVID-19 imposes on healthcare systems nationwide.