Jonathan D Baghdadiv

Background: Surgical site infection (SSI) surveillance can be time consuming and resource intensive. This study investigates the potential of generative artificial intelligence (GenAI) to augment the detection and classification of SSIs. Methods: A case control study of patients with SSI following spine surgery at one US hospital. SSIs were classified into superficial, deep, and organ space. All SSIs were confirmed by infection prevention (IP) experts as they occurred from October, 2023 to September, 2025 and matched 1:1 by year to surgeries deemed non-SSI. A secure GenAI was used to determine if patients had an SSI based on standardized prompts and clinical data. IP nurses used GenAI output to review cases with the ability to ask GenAI questions within the data provided or independently open the medical record. We compared GenAI determinations to initial IP nurses’ determinations. Results: A total of 555 patients had spine surgeries. All 16 SSIs were matched by year to 16 non-SSI. All SSIs were correctly identified by GenAI (sensitivity 100%, 16/16) and only 1 non-SSI was incorrectly identified as SSI (specificity 93.7%, 15/16). Although GenAI accurately identified all SSI cases, it was discordant with original review at classifying the level of infection in 37.5% (6/16) of cases. Upon final IP physician review, GenAI was correct in 66.7% (4/6) of discordant cases (often determining “organ space infections” rather than “deep”). Median time to complete GenAI assisted SSI reviews was 9 minutes (IQR 7–21). Conclusion: GenAI is a promising tool to assist in SSI surveillance following spinal surgery that could improve efficiency.

Background: Respiratory illness is the most frequent reason for unnecessary antibiotic use among hospitalized adults. In randomized trials, procalcitonin and respiratory virus testing without guidance on test interpretation do not influence antibiotic decision-making. Methods: We conducted a pragmatic, randomized, controlled trial of antimicrobial stewardship-guided test interpretation versus usual care among hospitalized adults receiving antibiotics for suspected respiratory infection with either low procalcitonin or positive respiratory virus testing at 2 hospitals. The intervention involved a templated note in the electronic health record interpreting test results in terms of the post-test probability of bacterial pneumonia and antibiotic decision-making. When probability of bacterial pneumonia was low, discontinuation of antibiotic therapy was recommended. The primary outcome was in-hospital antibiotic days of therapy. Results: Between 1 November 2023 and 10 January 2025, 107 adults were enrolled, including 65% with low procalcitonin, 30% with positive respiratory virus testing, and 5% with both. The intervention decreased antibiotic use by an average of 4.1 in-hospital days of therapy (7.5 versus 11.6, P = .006). All respiratory antibiotics were discontinued within 5 days of initiation for 76% of intervention patients versus 49% with usual care (P = .004). Length of stay (5.5 days intervention versus 6.6 days usual care, P = .16) and 30-day readmission (7% intervention versus 19% usual care, P = .079) did not significantly differ between groups. Conclusions: In this proof-of-concept study, antimicrobial stewardship-guided interpretation of laboratory tests for viral infection using a simple template safely decreased unnecessary antibiotic use for hospitalized adults with community-acquired respiratory illness. Clinical Trial Registration. NCT05976581.

A retrospective cohort study compared generative artificial intelligence (GenAI) versus infection control expert for catheter-associated urinary tract infection (CAUTI) detection. Sensitivity was 95.2% (20/21), and specificity was 76.2% (16/21), improving to 90% with expert review of GenAI output. GenAI is a promising tool to assist in CAUTI surveillance.

Among all 2024 emergency department visits for acute respiratory illness or conjunctivitis across a statewide healthcare system (n = 13 494), generative artificial intelligence–based surveillance with adjunctive human review rapidly and cost-effectively identified patients with potential avian influenza exposure. Generative artificial intelligence–based approaches may offer a promising opportunity for expanding public health surveillance in clinical settings.

Background: In recent years, new broad-spectrum antibiotics targeting Gram-negative organisms have been introduced, including cefiderocol, ceftazidime-avibactam, ceftolozane-tazobactam, eravacycline, imipenem-relebactam, omadacycline, and meropenem-vaborbactam. This study aimed to describe new antibiotic use across a large national cohort. Methods: We performed a retrospective cohort study of hospital discharges from June 2022 to May 2023 using the Premier Healthcare Database. Antibiotic utilization was ascertained from daily charges. Clinical indication(s) were inferred from International Classification of Diseases, 10th revision, diagnosis codes. Antibiotic therapy was considered definitive if continued >3 days. Piperacillin-tazobactam was used as a comparator. Results: Across 832 hospitals, 3 890 557 admissions (61.9% of all admissions) included an antibiotic prescription. New antibiotics were prescribed in 9768 admissions (0.25% of antibiotic-prescribing admissions) across 537 hospitals. Ceftolozane-tazobactam was prescribed in 4157 admissions (42.6% of 9768), ceftazidime-avibactam in 3660 (37.5%), eravacycline in 1213 (12.4%), cefiderocol in 1060 (10.9%), meropenem-vaborbactam in 456 (4.7%), omadacycline in 104 (1.1%), and imipenem-relebactam in 99 (1.0%). In contrast, piperacillin-tazobactam was prescribed in 731 719 (18.8%) and colistin in 570 (0.01%) admissions. Forty-six percent (n = 4647/9768) of new antibiotics were started in the first 3 days of hospital admission, and 70% (n = 6799/9768) were used as definitive therapy. Sepsis (76%), pneumonia (46%), and urinary tract infection (39%) were the most common clinical indications. On average, patients treated with new antibiotics had 8 more comorbid conditions than patients receiving piperacillin-tazobactam. Conclusions: Ceftazidime-avibactam and ceftolozane-tazobactam remain the most frequently prescribed new antibiotics, with uptake of subsequently approved agents trailing. New antibiotics are most commonly used as treatment for sepsis among patients with multiple comorbidities.

Background: Clinicians often start unnecessarily broad-spectrum empiric gram-negative antibiotics out of the concern that delaying effective therapy could lead to a worse clinical outcome. This study examined the consequences of delayed initiation of broad-spectrum gram-negative antibiotics. Methods: In a retrospective cohort of adult inpatients from 928 US hospitals, we compared clinical outcomes after (1) empiric narrow-spectrum antibiotics escalated to broad-spectrum antibiotics (delayed broad-spectrum therapy [DBT]) and (2) empiric broad-spectrum antibiotics continued as post-empiric therapy (early broad-spectrum therapy [EBT]) using Win Ratios. DBT and EBT patients were matched on hospital, admitting diagnosis, and propensity scores incorporating 28 clinical variables. The outcome of interest was a ranked composite of mortality, readmission, and adverse drug events. Results: Out of 746 880 inpatients, 82 276 (11%) received DBT and 664 604 (89.0%) received EBT. Among the 67 046 with DBT who were matched to 67 046 with EBT, mortality was 8.7% after DBT and 9.5% after EBT (P = .022), readmission was 10.5% after DBT and 11.8% after EBT (P < .0001), and the rate of adverse drug events was 8.4% after DBT and 7.2% after EBT (P < .0001). Among matched patients, clinical outcomes were superior after DBT compared with EBT (win-ratio 1.06; P < .0001). Conclusions: On average, among a large sample of adult inpatients who ultimately received broad-spectrum antibiotic therapy, delaying initiation of a broad-spectrum antibiotic was not associated with worse outcomes. Although broad-spectrum empiric therapy is undoubtedly sometimes warranted, this finding challenges the common belief that is it safer to err towards overly broad-spectrum empiric antibiotic therapy.

IMPORTANCE: Sepsis, a leading cause of death in the hospital, is a heterogeneous syndrome without a defined or specific set of symptoms. OBJECTIVES: We conducted a survey of clinicians in practice to understand which clinical findings they tend to associate with sepsis. DESIGN, SETTING, AND PARTICIPANTS: A survey was distributed to physicians and advanced practice providers across a multihospital health system during April 2022 and May 2022 querying likelihood of suspecting sepsis and initiating sepsis care in response to various normal and abnormal clinical findings. ANALYSIS: Strength of association between clinical findings and suspicion of sepsis were based on median and interquartile range of complete responses. Comparisons between individual questions were performed using Wilcoxon rank-sum testing. RESULTS: Among 179 clinicians who opened the survey, 68 (38%) completed all questions, including 53 (78%) attending physicians representing six different hospitals. Twenty-nine respondents (43%) worked primarily in the ICU, and 16 (24%) worked in the emergency department. The clinical findings most strongly associated with suspicion of sepsis were hypotension, tachypnea, coagulopathy, leukocytosis, respiratory distress, and fever. The abnormal clinical findings least likely to prompt suspicion for sepsis were elevated bilirubin, elevated troponin, and abdominal examination suggesting ileus. On average, respondents were more likely to suspect sepsis with high temperature than with low temperature (p = 0.008) and with high WBC count than with low WBC count (p = 0.003). CONCLUSIONS: Clinicians in practice tend to associate the diagnosis of sepsis with signs of severe illness, such as hypotension or respiratory distress, and systemic inflammation, such as fever and leukocytosis. Except for coagulopathy, nonspecific laboratory indicators of organ dysfunction have less influence on decision-making.

Background: COVID-19 changed the epidemiology of community-acquired respiratory viruses. We explored patterns of respiratory viral testing to understand which tests are most clinically useful in the postpandemic era.Methods: We conducted a retrospective observational study of discharge data from PINC-AI (formerly Premier), a large administrative database. Use of multiplex nucleic acid amplification respiratory panels in acute care, including small (2–5 targets), medium (6–11), and large panels (>11), were compared between the early pandemic (03/2020–10/2020), late pandemic (11/2020–4/2021), and prepandemic respiratory season (11/2019 - 02/2020) using ANOVA.Results: A median of 160.5 facilities contributed testing data per quarter (IQR 155.5–169.5). Prepandemic, facilities averaged 103 respiratory panels monthly (sd 138), including 79 large (sd 126), 7 medium (sd 31), and 16 small panels (sd 73). Relative to prepandemic, utilization decreased during the early pandemic (62 panels monthly/facility; sd 112) but returned to the prepandemic baseline by the late pandemic (107 panels monthly/facility; sd 211). Relative to prepandemic, late pandemic testing involved more small panel use (58 monthly/facility, sd 156) and less large panel use (47 monthly/facility, sd 116). Comparisons among periods demonstrated significant differences in overall testing (P < 0.0001), large panel use (P < 0.0001), and small panel use (P < 0.0001).Conclusions: Postpandemic, clinical use of respiratory panel testing shifted from predominantly large panels to predominantly small panels. Factors driving this change may include resource availability, costs, and the clinical utility of targeting important pathogenic viruses instead of testing “for everything.”

Background: Clostridioides difficile infection (CDI) may be misdiagnosed if testing is performed in the absence of signs or symptoms of disease. This study sought to support appropriate testing by estimating the impact of signs, symptoms, and healthcare exposures on pre-test likelihood of CDI. Methods: A panel of fifteen experts in infectious diseases participated in a modified UCLA/RAND Delphi study to estimate likelihood of CDI. Consensus, defined as agreement by >70% of panelists, was assessed via a REDCap survey. Items without consensus were discussed in a virtual meeting followed by a second survey. Results: All fifteen panelists completed both surveys (100% response rate). In the initial survey, consensus was present on 6 of 15 (40%) items related to risk of CDI. After panel discussion and clarification of questions, consensus (>70% agreement) was reached on all remaining items in the second survey. Antibiotics were identified as the primary risk factor for CDI and grouped into three categories: high-risk (likelihood ratio [LR] 7, 93% agreement among panelists in first survey), low-risk (LR 3, 87% agreement in first survey), and minimal-risk (LR 1, 71% agreement in first survey). Other major factors included new or unexplained severe diarrhea (e.g., ≥ 10 liquid bowel movements per day; LR 5, 100% agreement in second survey) and severe immunosuppression (LR 5, 87% agreement in second survey). Conclusion: Infectious disease experts concurred on the importance of signs, symptoms, and healthcare exposures for diagnosing CDI. The resulting risk estimates can be used by clinicians to optimize CDI testing and treatment.