Atypical symptoms in emergency department patients with urosepsis challenge current urinary tract infection management guidelines

The Infectious Diseases Society of America (IDSA)'s guidelines for managing asymptomatic bacteriuria (ASB) advise against antimicrobial treatment for potential urinary tract infection (UTI) except when patients have localizing genitourinary (GU) symptoms, fever, or hemodynamic instability.1 For patients lacking specific UTI signs and symptoms, including older patients with acute mental status changes or falls, they advise close observation instead of antibiotics. The authors of the IDSA guidelines classified the strength of their recommendations as "strong" while simultaneously categorizing the quality of the underlying evidence for these recommendations as "low" or "very low" (per the GRADE framework) due to reliance on studies limited by setting, sample size, and confounding effects.1, 2 The Choosing Wisely campaign—an effort to combat unnecessary tests and treatments—draws on these recommendations, similarly advising against urine testing and treatment, even in the presence of delirium, unless there are localizing urinary symptoms.3 At the same time, the Surviving Sepsis Campaign emphasizes early recognition and antibiotic treatment of sepsis.4 Up to 80% of patients with sepsis receive initial care in the emergency department (ED), and urosepsis represents a common form.5 Prior reports on ED sepsis presentation describe that many patients lack hemodynamic instability on presentation,6, 7 and many septic patients do present with acute confusion.4 Given the Surviving Sepsis Campaign's continued recommendation of prompt administration of antibiotics for the treatment of sepsis, and the lack of any clinical data from EDs cited in the IDSA guidelines, we explored the presenting symptoms of a cohort of ED patients with urosepsis. We hypothesized there may be a subset of patients at risk of delayed antibiotic treatment under the IDSA guidelines. In this letter, we report a secondary analysis of a data set of adult sepsis patients treated in the ED of an urban academic medical center from April 1, 2014, to March 31, 2016. The study protocol was approved by our institutional review board with a waiver of informed consent. The data set included all ED patients meeting sepsis criteria based on the CMS Severe Sepsis and Septic Shock (SEP-1) definition.8 Additional details about this database have been reported.6 The original data set also contained a set of control, nonseptic patients randomly selected from the same time interval who had vital sign abnormalities (qSOFA positive, systolic blood pressure [SBP] < 100 mm Hg, or shock index >1 [SBP < heart rate]), for use as a comparator to the septic cohort.9 All data were sourced retrospectively from the hospital's medical record electronic data warehouse. Presenting symptoms were abstracted by two trained reviewers who independently evaluated portions of the ED clinical notes (triage note as well as nursing and providers' initial documented patient histories) and completed a standardized data entry form. The reviewers were blinded to the clinical assessments, plans, details of ED management, diagnostic findings, and final diagnosis. Septic and nonseptic cases were evaluated in a random order. Fever as a symptom included subjective or measured fever reported by the patient or temperature greater than or equal to 100.4°F measured at triage. GU symptoms were those classically associated with a GU disease process: hematuria, cloudy urine, increased urinary frequency, dysuria, back pain, flank pain, or GU pain. The fatigue/malaise/weakness/lethargy symptom complex also included dizziness, near syncope, and lightheadedness. Next, reviewers were unblinded, reviewed the remainder of the ED and hospital documentation, and abstracted whether an acute infection warranting treatment was documented in the hospital admission note. If yes, reviewers also documented which infection source, e.g., urinary, was diagnosed. For every subject, the coding of each reviewer was compared, and any disagreements were resolved by majority vote in a review session that included a third (physician) reviewer. In this novel secondary analysis, we hypothesized that some patients with urosepsis would present atypically, without meeting the IDSA criteria for treatment. We characterized the presenting symptoms for that subset of septic patients diagnosed with a urinary source, i.e., the urosepsis cohort. We also hypothesized that presenting characteristics of urosepsis patients would differentiate them from the randomly selected cohort of ED patients with abnormal vital signs. Regression analysis was performed to evaluate which symptoms were discriminatory, using SAS Version 9.4 (SAS institute). As investigational predictors, we analyzed all symptoms that had a prevalence of ≥15% in the urosepsis cohort. The original data set also included 576 randomly selected ED patients without sepsis whose primary pathologies spanned toxicologic, cardiac, respiratory, gastrointestinal, vascular, and traumatic diagnoses. Substantial additional details about the urosepsis cohort and the nonseptic cohort are available in the Data S1 on our lab GitHub website.10 The urosepsis cohort was compared with the randomly selected cohort of patients with abnormal vital signs (Table 1). In this retrospective analysis of an ED cohort of urosepsis patients, we found that 28% presented with atypical symptoms that did not meet IDSA symptom criteria for antibiotic treatment. Nearly all these patients had either fatigue/weakness or confusion. Advanced age and comorbidities were the other significant factors that helped distinguish urosepsis patients from the random control group. Nearly all had either leukocytosis or hyperlactatemia. Only 22% of ED urosepsis patients had frank hypotension at triage. Although these findings arose from a small cohort, they suggest that early detection and treatment of high-mortality UTI may require sending urinalyses and initiating antibiotics even in patients without fever or localizing symptoms. This is especially true in the presence of fatigue or confusion, the most common presenting symptoms in urosepsis patients with atypical symptoms. The IDSA guidelines state that any bacteriuria should be considered asymptomatic "irrespective of the presence of pyuria" and advise against antimicrobial treatment unless there is fever or localizing GU symptoms, even if older or confused. Furthermore, the guidelines express uncertainty regarding whether older patients with acute confusion should receive antibiotics even if febrile, stating that that "[i]t is unknown whether antimicrobial therapy for ASB in patients with delirium is beneficial when fever or other systemic signs of infection are present and no other localizing source of infection is apparent [emphasis added]." The IDSA guidelines advise that patients with ASB, including those with acute mental status changes, should be observed, although details are not provided for how to safely observe such patients. The IDSA acknowledges that their practice recommendations are based on "very low-quality evidence." There were no ED studies cited in these recommendations.1 Our analysis suggests that the 2019 IDSA guidelines, though well intentioned in the effort to avoid antibiotic overuse, may be too narrow in the ED setting where strict adherence could result in a sizable minority of urosepsis patients failing to receive timely antibiotics. Indeed, we found that fatigue/weakness had a comparable association with urosepsis as localizing GU symptoms did and is therefore important to consider in ED patients (fatigue/weakness OR 3.18 vs. GU symptoms/abdominal pain OR 3.61). Clinicians should balance the priority of antibiotic stewardship with a sufficiently high index of suspicion for sepsis even in patients who present without classic symptoms, especially in patients with mild vital sign abnormalities and/or lab abnormalities consistent with systemic infection. The ideal UTI management guidelines would include clear and safe recommendations for identifying and treating atypical presentations of UTI before patients develop hemodynamic instability and high-mortality sepsis. There are key limitations to this report. First, findings came from a moderate-sized cohort from a single center. Second, sepsis diagnosis was based in part on billing codes in accordance with SEP-1, which may reflect subjective diagnostic judgment of clinicians and billing staff; therefore, some misclassification between the sepsis and the control cohorts is possible. Third, symptoms were based on what was charted in the triage note, provider note, and initial ED nurse note, which are subject to errors in charting or chart review (mitigated in part by the aforementioned process of multistep, multireviewer blinded chart review). Treatment guidelines would benefit from more data from the ED setting, where the majority of urosepsis patients initially present. Future studies should evaluate the significance of atypical symptoms in the diagnosis of UTI to optimize early detection and prevention of urosepsis. Doing so, with data from ED patients, could help define best practices for when to send urinalysis in the ED, when to give antibiotics for abnormal urinalysis in the ED, and how to best observe patients with atypical UTI symptoms and abnormal urinalyses such that they do not necessarily progress to hemodynamic instability before meeting criteria for antibiotics. Study concept and design (Brett Biebelberg, Michael R. Filbin, Thomas Heldt, Andrew T. Reisner), acquisition of the data (Brett Biebelberg, Iain E. Kehoe, Abigail O'Connell, Andrew T. Reisner, Michael R. Filbin, Thomas Heldt), analysis and interpretation of the data (Brett Biebelberg, Iain E. Kehoe, Hui Zheng, Michael R. Filbin, Thomas Heldt, Andrew T. Reisner), drafting of the manuscript (Brett Biebelberg, Iain E. Kehoe, Andrew T. Reisner), critical revision of the manuscript for important intellectual content (Hui Zheng, Abigail O'Connell, Michael R. Filbin, Thomas Heldt), statistical expertise (Brett Biebelberg, Hui Zheng, Andrew T. Reisner), and acquisition of funding (Michael R. Filbin, Thomas Heldt, Andrew T. Reisner). This investigation was supported in part by the MIT–MGH Strategic Grand Challenge Partnership (to Andrew T. Reisner, Michael R. Filbin, and Thomas Heldt) and by grants from the CRICO Risk Management Foundation (to Andrew T. Reisner, Michael R. Filbin) and Nihon Kohden Corporation (to Andrew T. Reisner, Michael R. Filbin, and Thomas Heldt). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. Investigators Andrew T. Reisner, Michael R. Filbin, and Thomas Heldt hold a patent related to sepsis patient management (#WO2016133928A1), which has been licensed to the Nihon Kohden Corporation. The other authors declare no conflicts of interest. Data S1. Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.