Longer Steroid Treatment Increases Secondary Bloodstream Infection Risk Among Patients With COVID-19 Requiring Intensive Care : Infectious Diseases in Clinical Practice

Secondary Logo

Journal Logo

Original Article

Longer Steroid Treatment Increases Secondary Bloodstream Infection Risk Among Patients With COVID-19 Requiring Intensive Care

Dupper, Amy C. MPH, MA; Malik, Yesha MD; Cusumano, Jaclyn A. PharmD, BCIDP†,‡; Nadkarni, Devika BS; Banga, Jaspreet MD; Berbel Caban, Ana MD; Twyman, Kathryn PhD§; Obla, Ajay PhD; Patel, Dhruv MBBS∗,¶; Mazo, Dana MD, MSc∗,†; Altman, Deena R. MD, MS∗,∥

Author Information
Infectious Diseases in Clinical Practice: July 2022 - Volume 30 - Issue 4 - IPC.0000000000001188
doi: 10.1097/IPC.0000000000001188
  • Open

Abstract

The severe acute respiratory syndrome coronavirus 2 pandemic began as an outbreak in Wuhan, China, in late 2019 and rapidly spread into an ongoing global pandemic.1 The New York City metropolitan area was one of the first epicenters of the coronavirus disease 2019 (COVID-19) pandemic in the United States, with the first case diagnosed late in February 2020.2 The pandemic has caused health care systems around the world to face an unprecedented surge of critically ill patients. A significant proportion of patients hospitalized for COVID-19 require intensive care unit (ICU) level care during their illness.3

Historically, secondary bacterial and fungal infections contribute substantially to overall morbidity and mortality associated with viral pandemics, the most famous being the 1918 influenza pandemic.4 Patients hospitalized with COVID-19 are exposed to numerous factors known to increase the risk of secondary infections, including prolonged hospitalization and extended use of medical devices.5 Risks for secondary infections may be further increased with immunomodulatory therapies that have been incorporated into clinical guidelines for COVID-19, such as corticosteroids and inhibitors of cytokines such as interleukin-6.6–8

There are limited data comparing critically ill COVID-19 patients who develop bloodstream infections (BSIs) to patients who do not, particularly evaluating use of corticosteroids, considering their use is now standard of care therapy for critically ill patients. We aimed to assess risk factors, including steroid treatment days, and outcomes associated with development of bacterial and fungal BSI in patients admitted to the ICU with COVID-19 during the peak of the first wave in New York City. Using data from before the results of the RECOVERY trial6 led to standardization of steroid treatment allows for the comparison of risks of different steroid practices, which can help inform mitigation strategies. We hypothesized that there will be clinical differences among patients critically ill with COVID-19, specifically regarding steroid administration, between those who did and did not develop secondary BSI.

METHODS

Study Population and Inclusion Criteria

The Mount Sinai Hospital (MSH) is an academic, quaternary facility in New York City that serves a diverse patient population from around the metropolitan area. This retrospective case-control study included adult patients (≥18 years) who were admitted to MSH with positive severe acute respiratory syndrome coronavirus 2 by nasopharyngeal polymerase chain reaction between April 1, 2020, and April 30, 2020, and were subsequently admitted to the ICU during their hospital stay. The study protocols were reviewed and approved by the Icahn School of Medicine at Mount Sinai Institutional Review Board. Use and disclosure of protected health information was approved under a waiver of authorization, and a waiver of informed consent was obtained (protocol HS nos. 20-00625 and 20-00703). This study was performed in accordance with the Helsinki Declaration of 1964 and its later amendments. Clinical data were collected from the electronic medical record and entered into the secured online database REDCap.9

We limited our investigation to the month of April 2020, which was the peak of the first surge, to minimize variations in surge level and its impact on the ability to provide care, patient population, circulating variants, and treatment guidelines. Cases were identified as patients with COVID-19 who developed BSI during ICU admission. Controls were patients admitted to the ICU for COVID-19 but who did not develop BSI. Patients transferred to MSH from an outside hospital more than 24 hours after their admission to the outside hospital were excluded. We further excluded patients with a documented BSI before ICU admission, those who had an ICU admission less than 3 days, and those who did not have an infectious diseases consultation. At the height of the COVID-19 pandemic, temporary ICUs were created to accommodate the critically ill. Because of the rapid pace at which these units were created, only patients admitted to units mapped as critical care in the National Healthcare Safety Network before February 2020 were included in the study.

Data Collection and Definitions

We defined BSI as a positive blood culture for 1 or more bacterial or fungal organisms, and only the first episode of BSI was included. Provider notes were reviewed by 2 separate infectious disease–trained investigators to determine whether the organism was clinically significant versus a contaminant and to identify the source of BSI. In instances of ambiguity, a third infectious disease–trained specialist was consulted. In assessing vascular access as a presumed source of BSI, a clinical and not surveillance definition was used. Organism identification and susceptibility was determined by standard clinical practices by the Clinical Microbiology Laboratory at MSH. The cumulative number of corticosteroid days was obtained from the Mount Sinai Data Warehouse and manually verified with the electronic medical record. The number of corticosteroid days included oral or intravenous medications equivalent to ≥20 mg of prednisone prescribed specifically as COVID-19 therapy. The number of days from initial insertion of a central line, initiation of mechanical ventilation, and days of corticosteroids from admission were measured until the date of first positive BSI for cases. The number of days for controls was measured until the ICU day of the positive blood culture of the matched case.

Statistical Analysis

This was a retrospective case-control study of patients hospitalized with COVID-19 with the outcome defined as development of BSI during ICU admission. Controls and cases were matched on sex, age at time of infection, and length of ICU stay until BSI. Each case was matched with 2 controls using Mahalanobis distance with nearest neighbor matching as described by Rubin and Thomas.10 Balance was visually assessed using a Q-Q plot.11 Descriptive variables were placed into a univariate logistic regression model, and all variables with P < 0.20 in univariate analysis were then placed in a multivariable logistic regression model. The final model was determined using bidirectional elimination using the Akaike information criterion for model selection. Because the cases and controls were matched on sex, age at time of infection, and length of ICU stay until BSI, these variables were not included in the final multivariable model. The number of days on mechanical ventilation and presence of a central venous catheter were found to be collinear with the number of days spent in the ICU and were therefore not included in the multivariable analysis. Variables with P < 0.05 were considered statistically significant. All analyses were performed in R (version 4.0.4).12

RESULTS

Baseline Demographics and Clinical Characteristics

During the study period, 194 patients were admitted to MSH with a COVID-19 diagnosis and subsequently admitted to the ICU. Of these patients, 63 were excluded based on our exclusion criteria (Fig. 1). Of the remaining 131 patients, 32 were identified as cases and were matched 1:2 to 64 controls resulting in a total of 96 patients included in this study. Baseline clinical characteristics of cases and controls are described in Table 1. Among cases, there were more men than women (n = 19 [59%]) with a median age of 65 years (interquartile range [IQR], 59.50–70.25 years) at the time of infection. The largest racial/ethnic groups represented were Hispanic (n = 14 [44%]) followed by non-Hispanic Black (n = 10 [31%]) and non-Hispanic White (n = 5 [16%]). More than half of cases reported never smoking (n = 20 [63%]), and the most common comorbidity noted at the time of admission was hypertension (n = 25 [78%]), followed by immunosuppression (n = 20 [63%]) and coronary artery disease (n = 7 [22%]). Only 1 person received treatment with an interleukin-6 inhibitor in this study.

F1
FIGURE 1:
Flowchart outlining the selection of cases and controls with predefined inclusion and exclusion criteria, resulting in 32 cases and 64 matched controls for 1:2 matching analysis.
TABLE 1 - Baseline Clinical Characteristics and Univariate and Multivariable Analyses of Risk Factors for Developing BSI During ICU Admission
Characteristics Cases (n = 32) Controls (n = 64) Univariable Analysis Multivariable Analysis
OR (95% CI) P OR (95% CI) P
Sex, no (%)*
 Male 19 (59) 38 (59) Reference
 Female 13 (41) 26 (41) 1.00 (0.42–2.37) 1.00
Age at time of infection, median (IQR)* 65.0 (59.5–70.3) 65.5 (58.0–71.3) 1.00 (0.97–1.05) 0.81
Days in ICU, median (IQR)*,† 11.0 (7.8–14.0) 11.0 (7.0–16.0) 1.01 (0.95–1.07) 0.80
Race/Ethnicity, no (%)
 Non-Hispanic White 5 (16) 15 (23) Reference
 Non-Hispanic Black 10 (31) 8 (13) 3.75 (0.99–15.92) 0.06
 Hispanic/Latino 14 (44) 31 (48) 1.35 (0.43–4.82) 0.62
 Asian 2 (6) 1 (2) 6.00 (0.48–146.87) 0.18
 Not reported 1 (3) 9 (9) 0.33 (0.02–2.53) 0.35
Smoking status, no. (%)
 Never smoker 20 (63) 45 (70) Reference
 Current smoker 4 (12) 1 (2) 9.00 (1.34–182.15) 0.06
 Former smoker 7 (22) 10 (16) 1.58 (0.51–4.71) 0.42
 Unknown 1 (3) 8 (12) 0.28 (0.01–1.68) 0.25
BMI, no. (%)
 18.5–24.9 kg/m2 9 (28) 13 (20) Reference Reference
 25.0–29.9 kg/m2 13 (41) 20 (31) 0.94 (0.31–2.86) 0.91 0.61 (0.17–2.08) 0.43
 ≥30.0 kg/m2 10 (31) 31 (48) 0.47 (0.15–1.42) 0.18 0.29 (0.08–0.97) 0.05
Admission source, no. (%)
 Home 27 (84) 59 (92) Reference
 Other facility 5 (16) 5 (8) 2.19 (0.56–8.48) 0.25
Comorbidities at admission, no. (%)
 Hypertension 25 (78) 34 (53) 3.15 (1.24–8.85) 0.02 3.78 (1.36–11.92) 0.01
 Immunosuppressed§ 20 (63) 35 (55) 1.38 (0.58–3.35) 0.47
 Coronary artery disease 7 (22) 8 (13) 1.96 (0.62–6.06) 0.24
 Kidney disease 5 (16) 9 (14) 1.13 (0.32–3.61) 0.84
 Solid organ transplant 5 (16) 7 (11) 1.51 (0.41–5.16) 0.51
 Asthma/COPD 3 (9) 8 (13) 0.72 (0.15–2.72) 0.65
 Congestive heart failure 3 (9) 4 (6) 1.55 (0.29–7.49) 0.58
 None 3 (9) 4 (6) 1.55 (0.29–7.49) 0.58
 Cancer 2 (6) 2 (3) 2.07 (0.24–17.91) 0.48
 Rheumatologic disease 1 (3) 3 (5) 0.66 (0.03–5.36) 0.72
Corticosteroid days, median (IQR)# 7.50 (5.50–14.00) 6.00 (1.00–8.25) 1.11 (1.03–1.21) 0.009 1.13 (1.04–1.25) 0.006
Days with CVC, median (IQR) 9.00 (5.75–12.25) 7.00 (3.50–10.50) 1.02 (0.97–1.08) 0.46
Ventilator days, median (IQR) 9.50 (6.00–13.25) 7.50 (4.00–13.00) 1.05 (0.98–1.14) 0.17
Bold indicates significance at P ≤ 0.05.
BMI indicates body mass index; COPD, chronic obstructive pulmonary disease; CVC, central venous catheter.
*Variables were used for matching and were not included in the multivariable analysis.
“Days in ICU”: measured from admission to ICU to first positive culture for cases, and controls were measured based on the number of days for their matched case.
“Other Facility”: includes admission to hospital from a nursing home, rehabilitation facility, LTACH, other hospitals including MSH satellite locations, and homeless shelters.
§“Immunosuppressed”: comorbidity at the time of hospital admission of diabetes, chronic corticosteroids/immunosuppressants use, cirrhosis, and/or HIV.
#“Corticosteroid Days”: number of corticosteroid days or oral or intravenous medications including prednisone, prednisolone, methylprednisolone, dexamethasone, or hydrocortisone equivalent to ≥20 mg of prednisone prescribed specifically as COVID-19 therapy.

The median number of days between admission to the hospital and the first positive BSI was 14 days (IQR, 9.75–22.25 days). No difference in the mean number of days between admission to the hospital and admission to the ICU was observed between cases and controls (4.13 vs 4.39 days, P = 0.81). The most prevalent organism identified was Staphylococcus aureus (n = 6 [19%]) followed by Staphylococcus epidermidis (n = 5 [16%]), Enterococcus faecalis/faecium (n = 5 [16%]), and Pseudomonas aeruginosa (n = 4 [13%]; Table 2). The most common presumed source of BSI was vascular access (n = 13 [41%]) followed by an unknown source (n = 11 [34%]) and pneumonia (n = 4 [13%]). Twenty-five percent of the enterobacteriales isolates were classified as extended-spectrum β-lactamase. None of the gram-negative isolates were carbapenem resistant, and more than half (60%) of the enterococcus isolates were resistant to vancomycin.

TABLE 2 - BSI Characteristics and Outcomes of Cases
Characteristic Cases (n = 32)
Days from hospital admission and positive BSI, median (IQR) 14.00 (9.75–22.25)
Organism, no. (%)
Staphylococcus aureus 6 (19)
Staphylococcus epidermidis 5 (16)
Enterococcus species* 5 (16)
Pseudomonas aeruginosa †,‡ 4 (13)
Streptococcus 2 (6)
Klebsiella pneumoniae 2 (6)
Escherichia coli 2 (6)
Candida glabrata 1 (3)
Presumed source of BSI, no. (%)
 Vascular access 13 (41)
 Unknown 11 (34)
 Pneumonia 4 (13)
 Urinary source 3 (9)
 Intra-abdominal infection 2 (6)
WBC count 48 h before BSI, median (IQR), ×103/μL 12.50 (9.80–17.40)
Procalcitonin 48 h before BSI, median (IQR), ng/mL§ 1.40 (0.45–2.67)
Cumulative days on antibiotics, median (IQR) 16.00 (10.00–27.00)
Positive BSI within 48 h of death, no. (%)
 Yes 5 (24)
 No 15 (71)
*Includes E. faecalis and E. faecium; 3 (60%) of these were vancomycin resistant.
None of the Enterobacterales species or P. aeruginosa were carbapenem resistant.
Two were extended-spectrum β-lactamase producers.
§Missing = 10 cases.
WBC indicates white blood cell count.

Matched Case-Control Analysis

Cases had higher odds of a preexisting diagnosis of hypertension at the time of admission compared with controls (odds ratio [OR], 3.15; 95% confidence interval [CI], 1.24–8.85; P = 0.02). All patients in the study received corticosteroids before BSI for cases or for controls, before day of BSI of matched case. For every 1 day of steroid use, there was an 11% increase in the odds of BSI (OR, 1.11; 95% CI, 1.03–1.21; P = 0.009; Table 1). A multivariable logistic regression model revealed that cases had higher odds of having a diagnosis of hypertension at hospital admission (OR, 3.78; 95% CI, 1.36–11.92; P = 0.01) and that for every 1 day of steroid treatment, the odds of developing BSI increased by 13% (OR, 1.13; 95% CI, 1.04–1.25; P = 0.006) compared with controls. Alternatively, cases had lower odds of having body mass index greater than or equal to 30.00 (OR, 0.29; 95% CI, 0.08–0.97; P = 0.05) compared with controls. The majority (86%) of patients included in the study received antibiotics within the first 3 days of admission. Of those who received early antibiotic therapy, no difference between cases and controls was observed (78.81% vs 90.62%, P = 0.17). As was standard practice at the time of the study, most (75%) patients in the study received hydroxychloroquine, and there was no difference in the receipt of hydroxychloroquine among cases and controls (68.75% vs 78.13%, P = 0.56).

More than half of the cases died during their admission (n = 21 [66%]), and of these, 5 cases (24%) had BSI within 48 hours of death (Tables 2, 3). Of the 11 cases that were discharged, 10 (91%) were discharged to a long-term acute care hospital (LTACH) or other facility. Although there were no significant differences in mortality, cases had higher odds of being discharged to facilities such as LTACHs compared with controls (OR, 11.33; 95% CI, 1.85–220.68; P = 0.03; Table 3).

TABLE 3 - Univariate Logistic Regression of Outcomes in Cases and Controls
Outcomes Cases (n = 32) Controls (n = 64) Univariable Analysis
OR (95% CI) P
Location patient was discharged, no. (%)*
 Home 1 (9) 17 (53) Reference
 LTACH or other facility 10 (91) 15 (47) 11.33 (1.85–220.68) 0.03
Patient died during admission, no. (%)
 Yes 21 (66) 32 (50) 1.91 (0.80–4.71) 0.15
 No 11 (34) 32 (50) Reference
Bold indicates significance at P ≤ 0.05.
*Limited to patients discharged alive.

DISCUSSION

This study examined the clinical characteristics and outcomes of patients admitted for COVID-19 and who later developed BSI during admission to the ICU. We specifically focused on ICU patients in April 2020 to evaluate the risk factors for developing secondary BSI in a standardized group of high-risk patients at a time period before the use of corticosteroids was standard clinical practice. Our study revealed that each day of steroids increased the odds of BSI by 13% after adjusting for covariates. This finding expands on the findings of others by demonstrating that patients who developed nosocomial BSI were more likely to have been prescribed corticosteroids before BSI,13 and steroid use raises the odds of developing any secondary infection in ICU patients.14

There is an established association between preexisting hypertension and admission to the ICU among COVID-19 patients,15 and more than half of the patients included in our study had a diagnosis of hypertension at the time of hospital admission. Our study also demonstrated a positive association between hypertension and development of BSI while in the ICU, which has been described in only one other study.16 This finding warrants further examination in future studies.

The onset of BSI, on average, occurred 2 weeks into the hospital course and is consistent with findings that a majority of secondary infections develop late and are often complications of the hospital course.13 Several studies have demonstrated an independent association between hospital-onset BSI and mortality, as well as nosocomial ICU infections and mortality.14,17 Although we did not find mortality to be significant in our cohort, the patients with BSI were more likely to be discharged to a long-term care facility when compared with those who did not develop infection. This highlights the impact on long-term outcomes and recovery that secondary infections have on critically ill COVID-19 patients.

This study used strict inclusion criteria, rigorous case matching, and a narrow study window to minimize bias when comparing COVID-19 patients who did and did not develop BSI. These attempts to minimize potential confounders are especially important considering that, during the first surge, very little was known about treating this novel disease, and guidelines and clinical practices were constantly evolving. This study looked specifically at BSI in ICU patients to both standardize the group of patients evaluated and to focus on a group with high morbidity and mortality from both COVID-19 and BSI. We additionally chose to focus strictly on BSI, because positive culture results from other body sites (e.g., respiratory, urine, wound) are more difficult to distinguish between colonization and true infection. Although other studies have found an association with development of BSI in patients who received both methylprednisolone and tocilizumab,18 the topic was out of scope for our analysis given the few patients in our cohort who received tocilizumab.

There are potential limitations to this study. Using a single urban academic institution in April 2020 could impact generalizability to smaller community hospitals, nonsurge situations, and different treatment practices. The deliberate exclusions imposed on our study population resulted in a small sample size. Smoking status was self-reported in the medical charts and may not represent patients' actual smoking status. Retrospective observational chart review may contain risks of errors in chart abstraction and limits on the ability to control confounders. Because of the nonstandardized use of corticosteroids, the diversity of formulations, and the frequency of dose adjustments during this time, we simplified data collection by categorizing the daily dose of corticosteroid treatments into high (≥20 mg of prednisone or equivalent) and low (<20 mg of prednisone or equivalent). Further dose-response studies are warranted.

CONCLUSIONS

Critically ill patients with COVID-19 receive immunosuppressive treatment per clinical guidelines and are at high risk for both developing and dying from secondary infections. Our study indicates that further investigation of the impact of these immunosuppressive therapies and therapy durations on the development of secondary BSIs is warranted.

ACKNOWLEDGMENTS

We thank Dr Waleed Javaid for his assistance with initial study planning and critical comments on the manuscript. We would like to thank the Mount Sinai Data Warehouse for collating data on patients with COVID-19.

REFERENCES

1. Dong E, Du H, Gardner L. An interactive Web-based dashboard to track COVID-19 in real time. Lancet Infect Dis. 2020;20(5):533–534.
2. Gonzalez-Reiche AS, Hernandez MM, Sullivan MJ, et al. Introductions and early spread of SARS-CoV-2 in the New York City area. Science. 2020;369(6501):297–301.
3. Petrilli CM, Jones SA, Yang J, et al. Factors associated with hospital admission and critical illness among 5279 people with coronavirus disease 2019 in New York City: prospective cohort study. BMJ. 2020;369:m1966.
4. Brundage JF, Shanks GD. Deaths from bacterial pneumonia during 1918-19 influenza pandemic. Emerg Infect Dis. 2008;14(8):1193–1199.
5. Buetti N, Ruckly S, de Montmollin E, et al. COVID-19 increased the risk of ICU-acquired bloodstream infections: a case-cohort study from the multicentric OUTCOMEREA network. Intensive Care Med. 2021;47(2):180–187.
6. Horby P, Lim WS, et alGroup RC. Dexamethasone in hospitalized patients with COVID-19. N Engl J Med. 2021;384(8):693–704.
7. Coutinho AE, Chapman KE. The anti-inflammatory and immunosuppressive effects of glucocorticoids, recent developments and mechanistic insights. Mol Cell Endocrinol. 2011;335(1):2–13.
8. Gordon AC, Mouncey PR, Al-Beidh F, et alREMAP-CAP Investigators. Interleukin-6 receptor antagonists in critically ill patients with COVID-19. N Engl J Med. 2021;384(16):1491–1502.
9. Harris PA, Taylor R, Thielke R, et al. Research electronic data capture (REDCap)—a metadata-driven methodology and workflow process for providing translational research informatics support. J Biomed Inform. 2009;42(2):377–381.
10. Rubin DB, Thomas N. Combining propensity score matching with additional adjustments for prognostic covariates. J Am Stat Assoc. 2000;95(450):573–585.
11. Austin PC. Balance diagnostics for comparing the distribution of baseline covariates between treatment groups in propensity-score matched samples. Stat Med. 2009;28(25):3083–3107.
12. Dessau RB, Pipper CB. “R”—project for statistical computing. Ugeskr Laeger. 2008;170(5):328–330.
13. Kubin CJ, McConville TH, Dietz D, et al. Characterization of bacterial and fungal infections in hospitalized patients with coronavirus disease 2019 and factors associated with health care–associated infections. Open Forum Infect Dis. 2021;8(6):ofab201.
14. Bardi T, Pintado V, Gomez-Rojo M, et al. Nosocomial infections associated to COVID-19 in the intensive care unit: clinical characteristics and outcome. Eur J Clin Microbiol Infect Dis. 2021;40(3):495–502.
15. Roncon L, Zuin M, Zuliani G, et al. Patients with arterial hypertension and COVID-19 are at higher risk of ICU admission. Br J Anaesth. 2020;125(2):e254–e255.
16. Kokkoris S, Papachatzakis I, Gavrielatou E, et al. ICU-acquired bloodstream infections in critically ill patients with COVID-19. J Hosp Infect. 2021;107:95–97.
17. Bhatt PJ, Shiau S, Brunetti L, et al. Risk factors and outcomes of hospitalized patients with severe coronavirus disease 2019 (COVID-19) and secondary bloodstream infections: a multicenter case-control study. Clin Infect Dis. 2021;72(12):e995–e1003.
18. Giacobbe DR, Battaglini D, Ball L, et al. Bloodstream infections in critically ill patients with COVID-19. Eur J Clin Invest. 2020;50(10):e13319.
Keywords:

COVID-19; coinfection; bacteremia; steroids; critical care

Copyright © 2022 The Author(s). Published by Wolters Kluwer Health, Inc.