Article Citation:Lori L. Boland, Joshua S. Huelster, David A. Hildebrandt, Ramiro Saavedra Romero, James P. Normington, Roman R. Melamed, Michael R. Mooney, and Maximilian Mulder (2018) Infection, Timing of Antibiotics and Outcomes in Out-of-Hospital Cardiac Arrest Patients Treated With Therapeutic Hypothermia. Journal of the Minneapolis Heart Institute Foundation: Spring/Summer 2018, Vol. 2, No. 1, pp. 21-27.
Lori L. Boland, MPH
Joshua S. Huelster, MD
David A. Hildebrandt, BSN
Ramiro Saavedra Romero, MD
James P. Normington, MS
Roman R. Melamed, MD
Michael R. Mooney, MD
Maximilian Mulder, MD
Care Delivery Research, Abbott Northwestern Hospital, 800 East 28th Street, Minneapolis, MN; Department of Critical Care Medicine, Abbott Northwestern Hospital, 800 East 28th Street, Minneapolis, MN; Heart Institute, Cedars-Sinai Medical Center, 127 S. San Vicente Blvd, Los Angeles, CA; Minneapolis Heart Institute, Abbott Northwestern Hospital, 800 East 28th Street, Minneapolis, MN
Address for correspondence:
Lori L. Boland, MPH
Care Delivery Research
Abbott Northwestern Hospital
800 East 28th Street
Minneapolis, MN 55407
Studies suggest out-of-hospital cardiac arrest (OHCA) patients treated with therapeutic hypothermia (TH) are at increased risk of infection; however, the optimal use of antibiotics in these patients, particularly timing of initiation, remains unclear. A retrospective cohort study was conducted among adult OHCA patients treated with TH at a single hospital between 2006 and 2013. Medication administration timestamps were used to determine whether antibiotics were initiated within 48 hours of admission (ie, during TH) or more than 48 hours after admission. Discharge diagnosis codes were used to identify infections. Due to concerns about confounding by transition to comfort care, analysis was restricted to patients who survived ≥72 hours. Of the patients studied, 187 of 259 (72%) had discharge codes for infection, the most common being pneumonia (57%) and sepsis (27%). Nearly all patients (96%) received antibiotics, with a median time to administration of 16 hours. Patients with initiation of antibiotics more than 48 hours after admission or not at all (n = 75) had a similar prevalence of infection (73% versus 72%; P = .80), slightly longer hospital length of stay (11 days versus 10 days; P = .03), but lower in-hospital mortality (24% versus 39%; P = .02) when compared to patients who received antibiotics within the first 48 hours of hospitalization (n = 184). Among patients who received antibiotics in the first 48 hours, there was no difference in infection or other outcomes when antibiotics were initiated within the first 6.5 hours after admission (sample median) or ≥6.5 hours after admission. Initiation of antibiotics within 48 hours of admission was not associated with reduced incidence of infection or improved survival in this cohort of TH-treated OHCA patients who survived ≥72 hours.
Keywords: out-of-hospital cardiac arrest, therapeutic hypothermia, antibiotic therapy, infection
Therapeutic hypothermia (TH) is a mainstay in the treatment of patients after out-of-hospital cardiac arrest (OHCA) and is endorsed in both the American Heart Association and the European Resuscitation Council guidelines for post-cardiac arrest care published in 2015.1,2 A recent meta-analysis has suggested an increased incidence of sepsis and pneumonia in TH patients,3 although the impact of infection on outcomes in TH populations has not been shown to be significant.4,5 Due to confounding clinical factors such as the inhibition of fever and nonspecific laboratory and radiographic markers of infection, it is often difficult to accurately diagnose infection in this population, and there is a need to better understand the role of antibiotics in the care of OHCA patients treated with TH. Several recent retrospective studies have suggested there may be benefit with either empiric6 or early antibiotic administration.7,8 We conducted a retrospective cohort study of OHCA patients treated with TH to examine patterns and correlates of antibiotic use and infection and their association with length of stay and mortality. Of particular interest was the impact of timing of the first dose of antibiotics, specifically initiation of antibiotics within 48 hours of admission (ie, during TH) versus more than 48 hours after admission (ie, after rewarming).
MATERIALS AND METHODS
Setting and Patients
This retrospective cohort study examined data from patients who were treated with TH after OHCA at a large quaternary hospital in Minneapolis, Minnesota, between February 2006 and December 2013. The hospital is a TH-capable facility that receives resuscitated cardiac arrest patients from across Minnesota and surrounding areas via an integrated, multidisciplinary system of care that affords regional and timely access to TH.9 Briefly, during the study timeframe, the TH protocol consisted of reducing core body temperature to 33°C using surface or intravascular cooling, maintaining target temperature for 24 hours, and then rewarming the patient at a rate of 0.25°C to 0.50°C per hour until core temperature reached 37°C, typically within 8 to 16 hours. All cases of TH are entered into the International Cardiac Arrest Registry (INTCAR), which was used to identify patients for inclusion in the study. Patients were excluded from analysis if the cardiac arrest occurred in-hospital (including the emergency department) or was related to trauma, if the patient was <18 years of age, or if the patient/family indicated at the time of hospitalization that they did not want their medical record data used for research purposes. The study protocol was approved by an appointed external institutional review board with a waiver of informed consent.Data Collection and Definitions
INTCAR data are collected by specially trained clinical research staff using case report forms, and include the following elements used for standardized reporting of cardiac arrest data per Utstein recommendations10: age; gender; prearrest cerebral performance category (CPC) score; arrest location; initial rhythm; whether the arrest was witnessed; provision of bystander cardiopulmonary resuscitation (CPR); presumed etiology of the arrest; and estimated time between arrest and return of spontaneous circulation (ROSC). The registry also provides details of postarrest intensive care unit (ICU) care and outcomes including presence of a do not resuscitate (DNR) order (yes/no); withdrawal of life support (yes/no); CPC category at discharge (1 to 5); ICU and hospital length of stay; and in-hospital mortality.
Whereas the aforementioned data elements were collected consistently across the study timeframe, information about infections available in INTCAR is limited and was not collected consistently. As a result, International Classification of Diseases, 9th Revision, Clinical Modification (ICD-9-CM) discharge diagnosis codes served as the primary data source used to identify infections among study subjects. For each hospitalization, all discharge codes were extracted from the hospital’s electronic health record (EHR) and reviewed by a study physician (JSH) who classified infections as pneumonia, sepsis/bloodstream, urinary tract, or cellulitis. ICD-9-CM codes have been shown to be a reasonably valid tool for retrospective identification and analysis of a variety of infections.11–13
Administration of antibiotics during the post-arrest hospital course was determined using medication data extracted from the EHR, which included type, route, and time first dose was given. Timing of first dose was used to determine whether antibiotics were administered within the first 48 hours after admission or more than 48 hours after admission. Given that in these patients, the cooling protocol is implemented as soon as possible after admission, that target TH temperature is maintained for 24 hours, and that rewarming occurs over 8 to 16 hours, patients receiving a first dose of antibiotics within 48 hours of admission were categorized as having received antibiotics during TH, while patients who received a first dose of antibiotics 48 hours or more after admission were categorized as having received post-TH antibiotics.
Study physicians (JSH, RS) reviewed the medical records of all patients to manually collect information needed to compute the sequential organ failure assessment (SOFA) score14 on admission, to assess whether patients were using antibiotics at the time of arrest, and to describe why the care team may have administered antibiotics in cases where no ICD-9-CM codes for infection were found.Exclusions
Among the 418 cardiac arrest patients entered into the TH registry during the study timeframe, 103 were excluded from the initial analysis: 84 were in-hospital arrests, 2 patients were less than 18 years of age, 2 experienced a traumatic arrest, and 15 patients/families had indicated at the time of hospitalization that they did not want their EHR data used for research (Figure 1). A strong concern for the effect that early movement to comfort care would have on both the use of antibiotics and the detection and documentation of infection in the 315 patients that initially remained for analysis was confirmed by the observation that 77% (24/31) of those who had not received any antibiotics had had life support withdrawn, while only 41% (117/284) of those who had received antibiotics during the post-arrest course had had life support withdrawn (P < .001). Also, the prevalence of infection (41% versus 72%; P < .001) and the use of antibiotics (64% versus 96%; P < .001) were markedly lower in patients who survived less than 72 hours when compared with those who survived 72 hours or more. To address this, confounding analyses were restricted to the 259 patients who had survived ≥72 hours. Despite not being part of formal TH or post-arrest protocols, the use of antibiotics was almost universal in this remaining cohort (96%). Only 11 of the 259 patients ultimately included in analysis did not receive antibiotics during the post-arrest hospitalization, and for purposes of analysis, these 11 patients were combined with patients who received a first dose of antibiotics more than 48 hours after admission (Figure 1).
Flow diagram of study cohort. ABX, antibiotics.
Patient and event characteristics were described using medians (range) and proportions. Differences between the <48 hour and 48+ hour antibiotic groups were tested using Wilcoxon rank-sum tests (continuous variables) and Pearson’s chi-square statistics (categorical variables). Prevalence of categorical outcomes and means of continuous outcomes were compared in the <48 hour and 48+ hour antibiotic groups. Patients who had received antibiotics within 48 hours of admission were further dichotomized into “early” and “late” administration of antibiotics using the median time to initiation of antibiotics in the subgroup (6.5 hours; Figure 1). Chi-square tests were used to evaluate proportional differences in categorical outcomes and two-sample Wilcoxon-Mann-Whitney rank sum tests were used to examine differences in continuous outcomes. All analyses were conducted using Stata 12.1 (Stata 12.1; StataCorp LP, College Station, Texas).
Of the 259 patients who were hospitalized more than 72 hours, the mean age was 62 and 72% were male (Table 1). With the exception of hypertension, patients who received antibiotics within 48 hours of admission had a higher prevalence of comorbidities than those receiving a first dose of antibiotics more than 48 hours after admission or not at all, particularly diabetes and chronic obstructive pulmonary disease (COPD). Patients receiving antibiotics within 48 hours of admission were also much more likely to have had a non-shockable initial rhythm at the time of arrest, a statistically significantly higher SOFA score at admission, a higher prevalence of a DNR order, and a greater likelihood of ultimately having life support withdrawn. Among patients who had life support withdrawn, declaration of poor neurologic prognosis was the primary reason for the withdrawal of life-sustaining treatment in 79% of cases (70/89). Declarations of poor neurologic prognosis were made based on a multimodal approach to prognostication including clinical examination, continuous video electroencephalography, and imaging at the minimum.
Characteristics of post-cardiac arrest patients treated with therapeutic hypothermia, overall, and by timing of antibiotics during the post-arrest hospitalization.
Overall in-hospital mortality in the cohort was 33% (Table 2). Discharge codes for infection were identified in 187 patients (72%), with pneumonia (57%), sepsis/bloodstream (27%), and urinary tract infections (8%) being the most common. The distribution of infection types was similar in the <48 hours and 48+ hours antibiotic groups. Initiation of antibiotics more than 48 hours after admission versus within the first 48 hours was associated with a slightly longer hospital length of stay (11 days versus 10 days; Wilcoxon rank sum, P = .03), but with lower in-hospital mortality (24% versus 39%; Pearson’s chi-square, P = .02). Among the 169 patients who survived to discharge, 96% had CPC scores of 1, 2, or 3.
Infection and outcomes in post-cardiac arrest patients treated with TH, overall, and by timing of antibiotics during the post-arrest hospitalization.
A total of 58 patients who received antibiotics during the post-arrest hospitalization had no ICD-9-CM codes for infectious processes among their discharge diagnoses. Upon medical record review, probable clinical indications for antibiotics could be identified among provider notes, imaging, or lab results for most of these patients, the most common being suspicion of aspiration (n = 28), pulmonary infiltrates (n = 11), fever or positive culture (n = 7), and suspicion of sepsis or presence of leukocytosis (n = 6).
Among patients who received antibiotics, 58% received their first dose within 24 hours of admission, and the median time to first dose was 16 hours (interquartile range: 5–50; Table 3). Most commonly ordered were vancomycin, penicillins, fluoroquinolones, and cephalosporins. Interestingly, half of all patients who were administered antibiotics within the first 48 hours of hospitalization were given the first dose fairly rapidly, ie, within 6.5 hours of admission, and these may represent instances of empiric therapy (Figure 2). Among the 184 patients who received their first dose of antibiotics within 48 hours of admission there were no statistically significant differences in infection or other outcomes when patients who received antibiotics very early were compared with those who received the first dose later in the TH phase (Table 4).
Details of antibiotic administration in 248 patients who received antibiotics during post-arrest hospitalization.
Frequency histogram of time to first antibiotics.
Infection and outcomes by timing of first dose of antibiotics in patients who received antibiotics within 48 hours of admission.
Despite recent evidence that TH is associated with an increased risk of infection,3,4prior studies of infection, antibiotic use, and outcomes in OHCA patients treated with TH have failed to produce clear evidence that empiric use of antibiotics improves mortality or functional outcomes.15 Gagnon et al6 found that while prophylactic antibiotic use in TH patients was associated with a 4-fold decrease in pneumonia, it was not associated with functional outcomes. In a comprehensive retrospective study of OHCA patients treated with TH in 371 Japanese hospitals between 2007 and 2013, antibiotic administration within 48 hours of admission was not associated with 30-day mortality.16 An earlier retrospective cohort study reported that overall, TH patients treated with antibiotics had lower mortality, but consistent with our findings the association did not persist when the analysis was restricted to patients who survived at least 72 hours.7 Several studies have reported that infectious complications in OHCA patients do not increase mortality,4,5,17 and thus although antibiotics may reduce the incidence of infection, this may not necessarily translate into improved survival.
Antibiotic use was nearly universal in both the current cohort (96%) and in another contemporary cohort of comparable patients (97%),8 but was reported to be only 50% to 87% in earlier cohorts of OHCA patients.4–6 As heightened concerns about the increased risk of infection with TH cause antibiotic use to become more ubiquitous in practice, observational studies will need to focus on elucidating the impact of the type, timing, dose, and duration of antibiotic therapy on outcomes. Two prior studies that have presented data on antibiotic type reveal a predominance of piperacillin-tazobactam and vancomycin,7,8 however regional and hospital practices are known to vary widely. With regard to timing, we report a median time to first dose of antibiotics of 16 hours, with 58% and 22% of those who received antibiotics within the first 48 hours after admission receiving the first dose within 24 hours and 6.5 hours of admission, respectively. Only two previous studies have quantified time to first dose of antibiotics in OHCA patients treated with TH, reporting a median of 8.7 hours8 and a mean of 2.2 days.7
Our examination of the impact of timing of antibiotic administration on outcomes in patients who survived at least 72 hours yielded two key findings. First, the initiation of antibiotics during the first 48 hours of hospitalization (ie, during TH) versus later or not at all was not associated with hospital-documented infections, but was associated with slightly higher in-hospital mortality and shorter hospital length of stay. Second, among patients who received their first dose of antibiotics within 48 hours of admission, rapid initiation of antibiotics (ie, within 6.5 hours of admission) conferred no statistically significant difference in infection, length of stay, mechanical ventilation hours, or in-hospital mortality. In the only prior studies to quantitatively describe time to first antibiotic dose and its association with outcomes in OHCA patients receiving TH, earlier antibiotic administration was not associated with pneumonia or mortality.8,16Disentangling the association between antibiotic use and outcomes in TH patients is unquestionably complicated in retrospective observational studies since the clinical decision to prescribe antibiotics may be partially driven by the very risk factors that predispose patients to poorer outcomes. In our study, patients who received antibiotics within the first 48 hours had a poorer prognostic profile on admission. They were more likely to have diabetes, COPD, non-shockable initial rhythms, DNR orders, and higher SOFA scores. We cautiously suggest that clinicians may more promptly and readily order prophylactic antibiotics in “sicker” patients and that it is these underlying factors, and not the earlier administration of antibiotics, that likely explain the higher mortality and shorter length of stay we observed in these patients.
Accurate diagnosis of infection can be challenging in TH due to the suppression of fever and the difficulty of interpreting non-specific laboratory values and radiologic findings. Most previous studies of incident infection in TH patients have used clinical markers (eg, cultures, radiographic findings, purulent secretions) to define infection,6–8but several studies have demonstrated the validity of using discharge ICD-9 codes for retrospective identification of infections in hospitalized patients.11–13 In the current study, 57% of patients had a diagnosis of pneumonia per hospital discharge codes, which is in line with the 40%-61% reported in other cohorts of patients treated with TH after OHCA.4–6,8,17 Our documented prevalence of sepsis was 27%, which is noticeably higher than previous estimates from earlier cohorts of 4% to 13%,5,6,17 but may reflect recent initiatives around early recognition and treatment of sepsis and improved documentation and coding practices for the condition.
Our analysis has several limitations. The nonrandomized nature of these historical observations should be recognized. Virtually all patients in our study received antibiotics during the post-arrest hospitalization and this precluded our ability to examine whether antibiotic use as a dichotomous exposure is associated with outcomes. Residual confounding by transition to comfort care may have compromised our estimates of the association between antibiotic use and outcomes. Many patients who initially survive OHCA to hospital admission are ultimately transitioned to comfort care because of a declaration of poor neurologic prognosis, and all clinical diagnostics and treatments are withheld. However, there is no single reliable indicator variable or timestamp available in the medical record that signals the transition of a patient to comfort care, and although we attempted to address this confounding by restricting analysis to patients who survived ≥72 hours, residual confounding may have persisted. Others have similarly excluded cases of early death from analysis in an effort to address this issue.6,8,16 We were also unable to investigate the specific implications of prophylactic antibiotics due to our inability to definitively characterize use as prophylactic. Although accurate timestamps for the administration of antibiotics were available in the medical record, the precise time of detection and diagnosis of infection is not available in a discrete field for comparison. Clinical information was not reviewed to confirm discharge diagnosis codes for infections, but using discharge diagnosis codes to retrospectively identify pneumonia has been shown to have acceptable accuracy.11–13 Finally, while our study represents a single center experience, it is part of a well-established integrated system that supports patients from a large regional referral network.
Very early initiation of antibiotics (<6.5 hours after admission) or initiation within 48 hours of admission did not confer any benefit in terms of infection or mortality in this report of OHCA patients treated with TH who survived ≥72 hours. Infection is a significant diagnostic challenge in this population, and large randomized studies are needed to elucidate the efficacy and optimal timing of antibiotics in these patients.
|1.||Callaway CW. Donnino MW. Fink EL.et al. Part 8: post-cardiac arrest care: 2015 American Heart Association guidelines update for cardiopulmonary resuscitation and emergency cardiovascular care. Circulation. 2015;132:S465–S482. [Crossref] [Google Scholar]|
|2.||Nolan JP. Soar J. Cariou A.et al. European Resuscitation Council and European Society of intensive care medicine guidelines for post-resuscitation care 2015: section 5 of the European Resuscitation Council guidelines for resuscitation 2015. Resuscitation. 2015;95:202–222. [Crossref] [Google Scholar]|
|3.||Geurts M. Macleod MR. Kollmar R. Kremer PH. van der Worp HB.Therapeutic hypothermia and the risk of infection: a systematic review and meta-analysis. Crit Care Med. 2014;42:231–242. [Crossref] [Google Scholar]|
|4.||Mongardon N. Perbet S. Lemiale V.et al. Infectious complications in out-of-hospital cardiac arrest patients in the therapeutic hypothermia era. Crit Care Med. 2011;39:1359–1364. [Crossref] [Google Scholar]|
|5.||Nielsen N. Sunde K. Hovdenes J.et al. Adverse events and their relation to mortality in out-of-hospital cardiac arrest patients treated with therapeutic hypothermia. Crit Care Med. 2011;39:57–64. [Crossref] [Google Scholar]|
|6.||Gagnon DJ. Nielsen N. Fraser GL.et al. Prophylactic antibiotics are associated with a lower incidence of pneumonia in cardiac arrest survivors treated with targeted temperature management. Resuscitation. 2015;92:154–159. [Crossref] [Google Scholar]|
|7.||Davies KJ. Walters JH. Kerslake IM. Greenwood R. Thomas MJ.Early antibiotics improve survival following out-of hospital cardiac arrest. Resuscitation. 2013;84:616–169. [Crossref] [Google Scholar]|
|8.||Hellenkamp K. Onimischewski S. Kruppa J.et al. Early pneumonia and timing of antibiotic therapy in patients after nontraumatic out-of-hospital cardiac arrest. Crit Care. 2016;20:31. [Crossref] [Google Scholar]|
|9.||Mooney MR. Unger BT. Boland LL.et al. Therapeutic hypothermia after out-of-hospital cardiac arrest: evaluation of a regional system to increase access to cooling. Circulation. 2011;124:206–214. [Crossref] [Google Scholar]|
|10.||Jacobs I. Nadkarni V. Bahr J.et al. Cardiac arrest and cardiopulmonary resuscitation outcome reports: update and simplification of the Utstein templates for resuscitation registries. A statement for healthcare professionals from a task force of the international liaison committee on resuscitation (American Heart Association, European Resuscitation Council, Australian Resuscitation Council, New Zealand Resuscitation Council, Heart and Stroke Foundation of Canada, InterAmerican Heart Foundation, Resuscitation Council of Southern Africa). Resuscitation. 2004;63:233–249. [Google Scholar]|
|11.||Drahos J. Vanwormer JJ. Greenlee RT. Landgren O. Koshiol J.Accuracy of ICD-9-CM codes in identifying infections of pneumonia and herpes simplex virus in administrative data. Ann Epidemiol. 2013;23:291–293. [Crossref] [Google Scholar]|
|12.||Olsen MA. Ball KE. Nickel KB. Wallace AE. Fraser VJ.Validation of ICD-9-CM diagnosis codes for surgical site infection and noninfectious wound complications after mastectomy. Infect Control Hosp Epidemiol. 2017;38:334–339. [Crossref] [Google Scholar]|
|13.||Tieder JS. Hall M. Auger KA.et al. Accuracy of administrative billing codes to detect urinary tract infection hospitalizations. Pediatrics. 2011;128:323–330. [Crossref] [Google Scholar]|
|14.||Vincent JL. Moreno R. Takala J.et al. The SOFA (sepsis-related organ failure assessment) score to describe organ dysfunction/failure. On behalf of the Working Group on Sepsis-Related Problems of the European Society of Intensive Care Medicine. Intensive Care Med. 1996;22:707–710. [Crossref] [Google Scholar]|
|15.||Newell C. Thomas M.Prophylactic antibiotic therapy for all out-of-hospital cardiac arrest survivors?Resuscitation. 2015;92:A4–A5. [Crossref] [Google Scholar]|
|16.||Tagami T. Matsui H. Kuno M.et al. Early antibiotics administration during targeted temperature management after out-of-hospital cardiac arrest: a nationwide database study. BMC Anesthesiol. 2016;16:89. [Crossref] [Google Scholar]|
|17.||Tsai MS. Chiang WC.et al. Infections in the survivors of out-of-hospital cardiac arrest in the first 7 days. Intensive Care Med. 2005;31:621–626. [Crossref] [Google Scholar]|
|Print ISSN: ||2475-0190|