{"id":28811,"date":"2025-11-22T04:19:00","date_gmt":"2025-11-21T20:19:00","guid":{"rendered":"https:\/\/csccm.org.cn\/?p=28811"},"modified":"2025-11-22T06:58:24","modified_gmt":"2025-11-21T22:58:24","slug":"icu-management-practice-%e9%9d%9e%e6%82%a3%e8%80%85%e7%89%b9%e5%be%81%e7%94%a8%e4%ba%8e%e5%8c%bb%e9%99%a2%e8%8e%b7%e5%be%97%e6%80%a7%e8%8f%8c%e8%a1%80%e7%97%87%e7%9a%84%e6%9c%ba%e5%99%a8%e5%ad%a6","status":"publish","type":"post","link":"https:\/\/csccm.org.cn\/?p=28811","title":{"rendered":"[ICU Management &#038; Practice]: \u975e\u60a3\u8005\u7279\u5f81\u7528\u4e8e\u533b\u9662\u83b7\u5f97\u6027\u83cc\u8840\u75c7\u7684\u673a\u5668\u5b66\u4e60"},"content":{"rendered":"\n<h1 class=\"wp-block-heading\">Using Nonpatient Features in Machine Learning for Hospital-Onset Bacteraemia&nbsp;<\/h1>\n\n\n\n<ul>\n<li>In&nbsp;<a href=\"https:\/\/healthmanagement.org\/c\/icu\">ICU<\/a><\/li>\n\n\n\n<li>Wed, 9 Jul 2025<\/li>\n<\/ul>\n\n\n\n<figure class=\"wp-block-image size-large\"><img decoding=\"async\" src=\"https:\/\/res.cloudinary.com\/healthmanagement-org\/image\/upload\/c_thumb,f_auto,fl_lossy,q_90\/v1752068513\/cw\/00130522_cw_image_wi_c201e840f56240b2fe75e90467077c13.webp\" alt=\"\"\/><\/figure>\n\n\n\n<p>Hospital-onset bacteraemia and fungaemia (HOB) are frequent, often preventable hospital complications linked to high mortality, morbidity, and costs. While many studies have focused on patient-related, non-modifiable factors (e.g., demographics, comorbidities), these offer limited prevention opportunities.&nbsp;<\/p>\n\n\n\n<p>Since up to two-thirds of HOB cases may be preventable, exploring nonpatient factors is crucial. The hospital functions as a dynamic network where patient risk is shaped by room locations, prior room occupants, interactions via healthcare workers, and environmental exposures, all of which can influence infection risk. Recent studies have shown that factors like prior room occupants with specific microbes, antibiotic exposure, and nursing variables can increase the risk of hospital-acquired infections.&nbsp;<\/p>\n\n\n\n<p>A new study used multimodal EHR data and machine learning models, incorporating both patient-related and nonpatient-related variables, to identify factors associated with HOB in adult patients, aiming to find actionable prevention targets within the hospital environment. The study analysed adult patients admitted to Barnes-Jewish Hospital in St. Louis in 2021, with analyses conducted from October 2023 to August 2024 and in April 2025. Patient data from electronic health records were used to create nonpatient features, such as interactions with healthcare workers and direct or indirect patient contact through consecutive room occupancy.<\/p>\n\n\n\n<p>HOB was defined as a positive blood culture after hospital day 3, with patients hospitalised longer than 3 days considered at risk. Three gradient boosting models were developed: two predictive (using patient features alone and combined with nonpatient features) and one causal model to test associations between nonpatient features and HOB.&nbsp;<\/p>\n\n\n\n<p>Among 52,442 patients, 34,855 (66.5%) hospitalised over 72 hours were analysed, with 556 (1.6%) developing HOB. The median age was 60 years, 50.5% were female, and obesity was the most common comorbidity (25%). Including nonpatient features, such as prior room occupants treated with antipseudomonal beta-lactams and the average number of healthcare workers interacting daily in the week before HOB, significantly improved model performance. These features were also linked to a higher risk of HOB in the causal model.<\/p>\n\n\n\n<p>The study demonstrates that nonpatient risk factors, such as antibiotic use by prior room occupants and interactions with healthcare workers (HCWs), improved the prediction of hospital-onset HOB beyond individual patient characteristics. Antipseudomonal beta-lactam use by previous room occupants and the number of HCWs interacting with patients were strongly associated with increased HOB risk in both predictive and causal models.<\/p>\n\n\n\n<p>While modifiable factors like indwelling devices and central venous catheter (CVC) care have traditionally been targets for reducing bloodstream infections, this study emphasises that HOB risk reflects the broader complexity of hospital networks involving patient-to-patient contact, HCW interactions, and environmental exposures, not just individual patient factors or device use.&nbsp;<\/p>\n\n\n\n<p>Recent literature supports the notion that hospital-acquired infections (HAIs) arise from transmission within the hospital setting, including exposure to infected or colonised roommates and prior room occupants. Studies have linked colonisation pressure to increased risks of multidrug-resistant organisms (MDROs) such as MRSA, vancomycin-resistant enterococci, and Clostridioides difficile. Although traditional antimicrobial susceptibility testing has limitations in tracking microbial relatedness, genomic and surveillance data reinforce the role of in-hospital transmission. The study\u2019s MRSA-specific model similarly identified colonisation pressure as a significant predictor.<\/p>\n\n\n\n<p>Beyond clonal spread, plasmid-mediated interspecies transmission contributes to the dissemination of resistant organisms, including carbapenem-resistant Enterobacterales. Furthermore, antibiotic use exerts a \u201cherd effect\u201d: ward-level prescribing patterns and antibiotic exposure of prior patients have been linked to increased risk of C. difficile and other resistant infections in subsequent patients. This study adds to this knowledge by connecting antipseudomonal beta-lactam use in prior room occupants to higher HOB risk in following patients, supporting the idea that antibiotic stewardship must consider not only individual patients but also the collective ward environment.<\/p>\n\n\n\n<p>Nosocomial infections also reflect hospital care quality. Staffing ratios and nursing workload impact infection rates, with understaffing linked to poorer outcomes. While the studied hospital adheres to staffing standards, simple patient-to-nurse ratios may inadequately capture workload complexity, as patient care needs vary widely. Nursing burnout, prevalent among up to half of nursing staff globally, has also been associated with increased HAIs. This study found the average daily number of HCWs interacting with patients to be a key predictive and causal feature, suggesting that organisational factors, like consistency in care teams or dedicated infection wards, may be actionable strategies to reduce infection risk.<\/p>\n\n\n\n<p>Source:&nbsp;<a href=\"https:\/\/jamanetwork.com\/journals\/jamanetworkopen\/fullarticle\/2835883\" target=\"_blank\" rel=\"noreferrer noopener\">JAMA<\/a><br \/>Image Credit: iStock&nbsp;<br \/>&nbsp;<\/p>\n\n\n\n<h4 class=\"wp-block-heading\">References:<\/h4>\n\n\n\n<p>Vazquez-Guillamet MC, Zhang J, Bewley A et al. (2025)&nbsp;<a href=\"https:\/\/jamanetwork.com\/journals\/jamanetworkopen\/fullarticle\/2835883\" target=\"_blank\" rel=\"noreferrer noopener\">Integrating Nonindividual Patient Features in Machine Learning Models of Hospital-Onset Bacteremia<\/a>. JAMA Netw Open. 8(7):e2518815.&nbsp;<\/p>\n","protected":false},"excerpt":{"rendered":"<p>Using Nonpatient Features in Machine Learning for Hospi [&hellip;]<\/p>\n","protected":false},"author":3,"featured_media":0,"comment_status":"closed","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":[],"categories":[24,23],"tags":[],"_links":{"self":[{"href":"https:\/\/csccm.org.cn\/index.php?rest_route=\/wp\/v2\/posts\/28811"}],"collection":[{"href":"https:\/\/csccm.org.cn\/index.php?rest_route=\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/csccm.org.cn\/index.php?rest_route=\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/csccm.org.cn\/index.php?rest_route=\/wp\/v2\/users\/3"}],"replies":[{"embeddable":true,"href":"https:\/\/csccm.org.cn\/index.php?rest_route=%2Fwp%2Fv2%2Fcomments&post=28811"}],"version-history":[{"count":1,"href":"https:\/\/csccm.org.cn\/index.php?rest_route=\/wp\/v2\/posts\/28811\/revisions"}],"predecessor-version":[{"id":28812,"href":"https:\/\/csccm.org.cn\/index.php?rest_route=\/wp\/v2\/posts\/28811\/revisions\/28812"}],"wp:attachment":[{"href":"https:\/\/csccm.org.cn\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=28811"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/csccm.org.cn\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=28811"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/csccm.org.cn\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=28811"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}