Editor: Deborah Sesok-Pizzini, MD, MBA, adjunct professor, Department of Clinical Pathology and Laboratory Medicine, Perelman School of Medicine, University of Pennsylvania, Philadelphia, and vice chair for clinical pathology and transfusion medicine, Nemours Children’s Health, Wilmington, Del.
Generating national estimates of pediatric sepsis in U.S. hospitals using clinical data
July 2026—Sepsis is a major cause of pediatric morbidity and mortality worldwide. It is difficult to measure the number of pediatric sepsis cases since most estimates are based on administrative codes, and coding practices vary by institution. Medical record review offers greater accuracy in recording pediatric sepsis diagnoses, but it is labor intensive and subject to interrater variability. Alternatively, electronic health record-based surveillance applies standardized criteria to routinely available clinical data at scale. In an effort to standardize pediatric sepsis reporting and estimate sepsis incidence, mortality, and trends in non-neonatal children, an international multispecialty task force developed the Phoenix criteria, in 2024, to update definitions of pediatric sepsis and septic shock. Although a major advancement in pediatric sepsis surveillance, Phoenix was not optimized for population surveillance. To address this gap, the authors developed and validated a pediatric sepsis event (PSE) definition aligned with the 2024 criteria but adapted for wide-scale EHR-based surveillance using routine clinical data captured in clinical data warehouses. They applied the definition to multihospital data sets to estimate the incidence, characteristics, mortality, trends, and national burden of pediatric sepsis with the intent of establishing it as a standardized framework to support local and national quality improvement initiatives for addressing the condition. To this end, the authors performed a retrospective cohort study of 3.9 million hospitalizations involving children 30 days to 17 years old. They applied their PSE definition to two EHR data sets: Epic Cosmos (245 health care systems, 2016–2023) and HCA Healthcare (146 hospitals, 2018–2023). They then used secondary data sets to assess the feasibility of implementation and face validity across different settings. The PSE algorithm was validated through medical record reviews of high-risk encounters at three geographically diverse hospitals. A PSE required presumed infection with concurrent organ dysfunction using Phoenix-derived thresholds adapted for routine EHR data. Septic shock was defined as a PSE with cardiovascular dysfunction. The authors calculated sepsis incidence, characteristics, and in-hospital mortality. Sensitivity and specificity of PSE for physician-adjudicated Phoenix sepsis were then compared with the administrative codes for severe sepsis/septic shock. Regression models were used to estimate national sepsis case counts in 2022 and temporal trends from 2016 to 2022. Among 3,926,809 pediatric hospitalizations from 2016 to 2023, 51,542 sepsis cases were identified (1.3 percent incidence), including 31,744 (61.6 percent) involving septic shock. A total of 37,405 (72.6 percent) of the sepsis cases were community onset. The in-hospital mortality was 10.1 percent, and sepsis was present in 17.8 percent of hospitalizations that resulted in death. On medical record review, the PSE definition yielded 69.9 percent sensitivity (95 percent confidence interval [CI], 58.1–79.8 percent) and 93.1 percent specificity (95 percent CI, 89.6–95.7 percent), which is higher sensitivity and similar specificity when compared with the administrative codes. The national estimates for 2022 were 18,231 sepsis cases (95 percent CI, 16,129–20,334) and 1,877 deaths (95 percent CI, 1,629–2,126). Of interest, neither estimates of sepsis cases nor deaths changed significantly from 2016 to 2022. The authors concluded that an EHR-based definition for pediatric sepsis showed strong validity compared with the physician-adjudicated Phoenix sepsis framework. Furthermore, even though the tool was developed in the United States, the PSE framework may inform similar initiatives internationally, with local validation.
Rhee C, Balamuth F, Dysart K, et al. National estimates of pediatric sepsis in US hospitals using clinical data. JAMA. doi.org/10.1001/jama.2026.3100
Correspondence: Dr. Chanu Rhee at crhee@bwh.harvard.edu
Underfilled tubes revisited: acceptability of short draws with various blood tests
Underfilling of evacuated blood tubes, or short draws, can cause errors in analytic reporting and are caused by multiple scenarios, such as air bubbles in the collection tubing, premature removal of evacuated tubes during blood collection, unsuccessful transfer of blood from a syringe to a tube, or drawing from patients with poor venous access or unexpected movements that impede blood flow. These scenarios are not uncommon in pediatric phlebotomy. Short draws may cause unacceptable biases in terms of analytical performance specifications or biological variation that result from a higher anticoagulant/blood ratio that dilutes analytes. Excess anticoagulant may interfere with chemistry tests and complete blood counts. Short draws are associated with in vitro hemolysis and release of intracellular components into plasma. Due to such factors, the Clinical Laboratory Standards Institute recommends against testing short draws, and it is standard practice for laboratories to reject these samples. Few studies have examined the impact of short draws on laboratory testing results and investigated the allowable error or desirable bias across concentration ranges that could allow for reporting some results as valid. The authors conducted a pilot study to estimate test result biases between short draws and fully filled evacuated blood tubes, or complete draws, for more commonly ordered lab tests, including coagulation, chemistry, and hematology tests; evaluate biases as percentages of total allowable error, set by the College of American Pathologists, and as percentages of what is considered desirable biases using biological variation; determine how biases change across concentration ranges; and develop a list of potential tests across a large acute care panel that can be reported despite short draws. They studied blood draw fill volume in BD Vacutainer tubes (1 mL for short draws [33–56 percent fill volume] and 1.8–3 mL for complete draws [100 percent fill volume]) from 12 healthy volunteers for three coagulation tests, 36 chemistry tests, and a CBC with differential. The study included anonymized laboratory results from 12 healthy staff volunteers at the participating location. Complete draw and short draw tubes were collected in the appropriate order of draw. The authors then analyzed the testing results and defined tests as “strong candidates” for reporting on short draws if there was no statistically significant bias between the short and complete draw results. They defined tests as “potential candidates” if there was statistically significant bias between the short and complete draw results but the absolute bias was less than 25 percent of total allowable error and less than the desirable bias from biological variation. The authors excluded, as candidates for reporting, tests for which it was not possible to determine the magnitude of absolute bias. They also excluded analytes, for reporting, that increased or decreased bias across concentration ranges. The study results showed that the following were strong candidates for reporting on short draws: two coagulation tests (INR, PT), 14 chemistry tests (AST, BILT, CK, CORT, FT4, GLUC, HbA1c, K, LAC, NH3, NT-proBNP, PHOS, TSH, urea), and 14 hematology tests (BA%, BA, EO%, EO, HCT, LY%, MCH, MCHC, MO%, NE%, NRBC%, NRBC, PLT, RDW). Nine chemistry tests (ALP, AMY, CRE, FE, FERR, LIP, MG, TRIG, URA) and two hematology tests (MCV, MPV) were potential candidates for reporting. Of note, these results applied to tests that were reported on 1-mL short blood draws in BD Vacutainer tubes using specific instruments. They may not be generalizable to labs using different evacuated tubes and instruments. The authors concluded that underfilled blood tubes may be valid collections for several coagulation, hematology, and chemistry tests. Using them may prevent unnecessary phlebotomy, such as repeat blood draws, in particular for pediatric patients. However, laboratories need to perform their own studies to determine if short draws are valid and acceptable using their specific tubes and instruments.
de Koning L, Ezra S, Cai F, et al. Underfilled tubes revisited: What blood tests can be reported on short draws? Am J Clin Pathol. doi.org/10.1093/AJCP/AQAF109
Correspondence: Dr. Lawrence de Koning at abldekon@ucalgary.ca