Search

Background: Over the next 5 years, new breast cancer screening guidelines recommending magnetic resonance imaging (MRI) for certain patients will significantly increase the volume of imaging data to be analyzed. While this increase poses challenges for radiologists, artificial intelligence (AI) offers potential solutions to manage this workload. However, the development of AI models is often hindered by manual annotation requirements and strict data-sharing regulations between institutions., Methods: In this study, we present an integrated pipeline combining weakly supervised learning-reducing the need for detailed annotations-with local AI model training via swarm learning (SL), which circumvents centralized data sharing. We utilized three datasets comprising 1372 female bilateral breast MRI exams from institutions in three countries: the United States (US), Switzerland, and the United Kingdom (UK) to train models. These models were then validated on two external datasets consisting of 649 bilateral breast MRI exams from Germany and Greece., Results: Upon systematically benchmarking various weakly supervised two-dimensional (2D) and three-dimensional (3D) deep learning (DL) methods, we find that the 3D-ResNet-101 demonstrates superior performance. By implementing a real-world SL setup across three international centers, we observe that these collaboratively trained models outperform those trained locally. Even with a smaller dataset, we demonstrate the practical feasibility of deploying SL internationally with on-site data processing, addressing challenges such as data privacy and annotation variability., Conclusions: Combining weakly supervised learning with SL enhances inter-institutional collaboration, improving the utility of distributed datasets for medical AI training without requiring detailed annotations or centralized data sharing., Competing Interests: Competing interests: J.N.K. declares consulting services for Owkin, France, and Panakeia, UK, and has received honoraria for lectures by Bayer, Eisai, MSD, BMS, Roche, Pfizer, and Fresenius. J.N.K. and D.T. hold shares in StratifAI GmbH, Germany. S.M. declares employment and shareholding with Osimis, Belgium. No other potential conflicts of interest are declared by any of the authors. The authors received advice from the customer support team of Hewlett Packard Enterprise (HPE) when performing this study, but HPE did not have any role in study design, conducting the experiments, interpretation of the results, or decision to submit for publication., (© 2025. The Author(s).)

Objectives: The aim of this work is to evaluate the performance of deep learning (DL) models for breast cancer diagnosis with MRI., Materials and Methods: A literature search was conducted on Web of Science, PubMed, and IEEE Xplore for relevant studies published from January 2015 to February 2024. The study was registered with the PROSPERO International Prospective Register of Systematic Reviews (protocol no. CRD42024485371). The quality assessment of diagnostic accuracy studies-2 (QUADAS2) tool and the Must AI Criteria-10 (MAIC-10) checklist were used to assess quality and risk of bias. The meta-analysis included studies reporting DL for breast cancer diagnosis and their performance, from which pooled summary estimates for the area under the curve (AUC), sensitivity, and specificity were calculated., Results: A total of 40 studies were included, of which only 21 were eligible for quantitative analysis. Convolutional neural networks (CNNs) were used in 62.5% (25/40) of the implemented models, with the remaining 37.5% (15/40) hybrid composite models (HCMs). The pooled estimates of AUC, sensitivity, and specificity were 0.90 (95% CI: 0.87, 0.93), 88% (95% CI: 86, 91%), and 90% (95% CI: 87, 93%), respectively., Conclusions: DL models used for breast cancer diagnosis on MRI achieve high performance. However, there is considerable inherent variability in this analysis. Therefore, continuous evaluation and refinement of DL models is essential to ensure their practicality in the clinical setting., Key Points: Question Can DL models improve diagnostic accuracy in breast MRI, addressing challenges like overfitting and heterogeneity in study designs and imaging sequences? Findings DL achieved high diagnostic accuracy (AUC 0.90, sensitivity 88%, specificity 90%) in breast MRI, with training size significantly impacting performance metrics (p < 0.001). Clinical relevance DL models demonstrate high accuracy in breast cancer diagnosis using MRI, showing the potential to enhance diagnostic confidence and reduce radiologist workload, especially with larger datasets minimizing overfitting and improving clinical reliability., Competing Interests: Compliance with ethical standards. Guarantor: The scientific guarantor of this publication is Fiona J. Gilbert. Conflict of interest: The authors of this manuscript declare no relationships with any companies, whose products or services may be related to the subject matter of the article. Statistics and biometry: Not applicable. Informed consent: Not applicable. Ethical approval: Not applicable. Study subjects or cohorts overlap: Not applicable. Methodology: Retrospective Systematic review and meta-analysis Multicenter study, (© 2025. The Author(s).)

Background Deep learning (DL) algorithms have shown promising results in mammographic screening either compared to a single reader or, when deployed in conjunction with a human reader, compared with double reading. Purpose To externally validate the performance of three DL algorithms as mammographic screen readers in an independent UK data set. Materials and Methods Three commercial DL algorithms (DL-1, DL-2, and DL-3) were retrospectively investigated from January 2022 to June 2022 using consecutive full-field digital mammograms collected at two UK sites during 1 year (2017). Normal cases with 3-year follow-up and histopathologically proven cancer cases detected either at screening (that round or next) or within the 3-year interval were included. A preset specificity threshold equivalent to a single reader was applied. Performance was evaluated for stand-alone DL reading compared with single human reading, and for DL reading combined with human reading compared with double reading, using sensitivity and specificity as the primary metrics. P < .025 was considered to indicate statistical significance for noninferiority testing. Results A total of 26 722 cases (median patient age, 59.0 years [IQR, 54.0-63.0 years]) with mammograms acquired using machines from two vendors were included. Cases included 332 screen-detected, 174 interval, and 254 next-round cancers. Two of three stand-alone DL algorithms achieved noninferior sensitivity (DL-1: 64.8%, P < .001; DL-2: 56.7%, P = .03; DL-3: 58.9%, P < .001) compared with the single first reader (62.8%), and specificity was noninferior for DL-1 (92.8%; P < .001) and DL-2 (96.8%; P < .001) and superior for DL-3 (97.9%; P < .001) compared with the single first reader (96.5%). Combining the DL algorithms with human readers achieved noninferior sensitivity (67.0%, 65.6%, and 65.4% for DL-1, DL-2, and DL-3, respectively; P < .001 for all) compared with double reading (67.4%), and superior specificity (97.4%, 97.6%, and 97.6%; P < .001 for all) compared with double reading (97.1%). Conclusion Use of stand-alone DL algorithms in combination with a human reader could maintain screening accuracy while reducing workload. Published under a CC BY 4.0 license. Supplemental material is available for this article.

Background: Integrating artificial intelligence (AI) into mammography screening can support radiologists and improve programme metrics, yet the potential of different strategies for integrating the technology remains understudied. We compared programme-level performance metrics of seven AI integration strategies., Methods: We performed a retrospective comparative evaluation of seven strategies for integrating AI into mammography screening using datasets generated from screening programmes in Germany (n=1 657 068), the UK (n=223 603) and Sweden (n=22 779). The commercially available AI model used was Vara version 2.10, trained from scratch on German data. We simulated the performance of each strategy in terms of cancer detection rate (CDR), recall rate, and workload reduction, and compared the metrics with those of the screening programmes. We also assessed the distribution of the stages and grades of the cancers detected by each strategy and the AI model's ability to correctly localise those cancers., Findings: Compared with the German screening programme (CDR 6·32 per 1000 examinations, recall rate 4·11 per 100 examinations), replacement of both readers (standalone AI strategy) achieved a non-inferior CDR of 6·37 (95% CI 6·10-6·64) at a recall rate of 3·80 (95% CI 3·67-3·93), whereas single reader replacement achieved a CDR of 6·49 (6·31-6·67), a recall rate of 4·01 (3·92-4·10), and a 49% workload reduction. Programme-level decision referral achieved a CDR of 6·85 (6·61-7·11), a recall rate of 3·55 (3·43-3·68), and an 84% workload reduction. Compared with the UK programme CDR of 8·19, the reader-level, programme-level, and deferral to single reader strategies achieved CDRs of 8·24 (7·82-8·71), 8·59 (8·12-9·06), and 8·28 (7·86-8·71), without increasing recall and while reducing workload by 37%, 81%, and 95%, respectively. On the Swedish dataset, programme-level decision referral increased the CDR by 17·7% without increasing recall and while reducing reading workload by 92%., Interpretation: The decision referral strategies offered the largest improvements in cancer detection rates and reduction in recall rates, and all strategies except normal triaging showed potential to improve screening metrics., Funding: Vara., Competing Interests: Declaration of interests ZVF, MB, TM, VŠ, SB, and CL are current or former employees of Vara, with stock options as part of the standard compensation package. FJG, NRP, and SEH have research collaborations with Vara, ScreenPoint, Lunit, Google, Volpara, iCAD, Therapixel, CureMetrix, Sunnybrook Research Institute, and Massachusetts Institute of Technology., (Copyright © 2024 The Author(s). Published by Elsevier Ltd. This is an Open Access article under the CC BY-NC-ND 4.0 license. Published by Elsevier Ltd.. All rights reserved.)

Background Artificial intelligence (AI) systems can be used to identify interval breast cancers, although the localizations are not always accurate. Purpose To evaluate AI localizations of interval cancers (ICs) on screening mammograms by IC category and histopathologic characteristics. Materials and Methods A screening mammography data set (median patient age, 57 years [IQR, 52-64 years]) that had been assessed by two human readers from January 2011 to December 2018 was retrospectively analyzed using a commercial AI system. The AI outputs were lesion locations (heatmaps) and the highest per-lesion risk score (range, 0-100) assigned to each case. AI heatmaps were considered false positive (FP) if they occurred on normal screening mammograms or on IC screening mammograms (ie, in patients subsequently diagnosed with IC) but outside the cancer boundary. A panel of consultant radiology experts classified ICs as normal or benign (true negative [TN]), uncertain (minimal signs of malignancy [MS]), or suspicious (false negative [FN]). Several specificity and sensitivity thresholds were applied. Mann-Whitney U tests, Kruskal-Wallis tests, and χ 2 tests were used to compare groups. Results A total of 2052 screening mammograms (514 ICs and 1548 normal mammograms) were included. The median AI risk score was 50 (IQR, 32-82) for TN ICs, 76 (IQR, 41-90) for ICs with MS, and 89 (IQR, 81-95) for FN ICs ( P = .005). Higher median AI scores were observed for invasive tumors (62 [IQR, 39-88]) than for noninvasive tumors (33 [IQR, 20-55]; P < .01) and for high-grade (grade 2-3) tumors (62 [IQR, 40-87]) than for low-grade (grade 0-1) tumors (45 [IQR, 26-81]; P = .02). At the 96% specificity threshold, the AI algorithm flagged 121 of 514 (23.5%) ICs and correctly localized the IC in 93 of 121 (76.9%) cases, with 48 FP heatmaps on the mammograms for ICs (rate, 0.093 per case) and 74 FP heatmaps on normal mammograms (rate, 0.048 per case). The AI algorithm correctly localized a lower proportion of TN ICs (54 of 427; 12.6%) than ICs with MS (35 of 76; 46%) and FN ICs (four of eight; 50% [95% CI: 13, 88]; P < .001). The AI algorithm localized a higher proportion of node-positive than node-negative cancers ( P = .03). However, no evidence of a difference by cancer type ( P = .09), grade ( P = .27), or hormone receptor status ( P = .12) was found. At 89.8% specificity and 79% sensitivity thresholds, AI detection increased to 181 (35.2%) and 256 (49.8%) of the 514 ICs, respectively, with FP heatmaps on 158 (10.2%) and 307 (19.8%) of the 1548 normal mammograms. Conclusion Use of a standalone AI system improved early cancer detection by correctly identifying some cancers missed by two human readers, with no differences based on histopathologic features except for node-positive cancers. © RSNA, 2024 Supplemental material is available for this article .

Background Breast screening enables early detection of cancers; however, most women have normal mammograms, resulting in repetitive and resource-intensive reading tasks. Purpose To investigate if deep learning (DL) algorithms can be used to triage mammograms by identifying normal results to reduce workload or flag cancers that may be overlooked. Materials and Methods In this retrospective study, three commercial DL algorithms were investigated using consecutive mammograms from two UK Breast Screening Program sites from January 2015 to December 2017 and January 2017 to December 2018 on devices from two mammography vendors. Normal mammograms with a 3-year follow-up and histopathologically proven cancer detected at screening, the subsequent round, or in the 3-year interval were included. Two algorithm thresholds were set: in scenario A, 99.0% sensitivity for rule-out triage to a lone reader, and in scenario B, approximately 1.0% additional recall providing a rule-in triage for further assessment. Both thresholds were then applied to the screening workflow in scenario C. The sensitivity and specificity were used to assess the overall predictive performance of each DL algorithm. Results The data set comprised 78 849 patients (median age, 59 years [IQR, 53-63 years]) and 887 screening-detected, 439 interval, and 688 subsequent screening round-detected cancers. In scenario A (rule-out triage), models DL-1, DL-2, and DL-3 triaged 35.0% (27 565 of 78 849), 53.2% (41 937 of 78 849), and 55.6% (43 869 of 78 849) of mammograms, respectively, with 0.0% (0 of 887) to 0.1% (one of 887) of screening-detected cancers undetected. In scenario B, DL algorithms triaged in 4.6% (20 of 439) to 8.2% (36 of 439) of interval and 5.2% (36 of 688) to 6.1% (42 of 688) of subsequent-round cancers when applied after the routine double-reading workflow. Combining both approaches in scenario C resulted in an overall noninferior specificity (difference, -0.9%; P < .001) and superior sensitivity (difference, 2.7%; P < .001) for the adaptive workflow compared with routine double reading for all three algorithms. Conclusion Rule-out and rule-in DL-adapted triage workflows can improve the efficiency and efficacy of mammography breast cancer screening. © RSNA, 2023 Supplemental material is available for this article. See also the editorial by Nishikawa and Lu in this issue.

Accumulating evidence from retrospective studies demonstrate at least non-inferior performance when using AI algorithms with different strategies versus double-reading in mammography screening. In addition, AI algorithms for mammography screening can reduce work load by moving to single human reading. Prospective trials are essential to avoid unintended adverse consequences before incorporation of AI algorithms into UK's National Health Service (NHS) Breast Screening Programme (BSP). A stakeholders' meeting was organized in Newnham College, Cambridge, UK to undertake a review of the current evidence to enable consensus discussion on next steps required before implementation into a screening programme. It was concluded that a multicentre multivendor testing platform study with opt-out consent is preferred. AI thresholds from different vendors should be determined while maintaining non-inferior screening performance results, particularly ensuring recall rates are not increased. Automatic recall of cases using an agreed high sensitivity AI score versus automatic rule out with a low AI score set at a high sensitivity could be used. A human reader should still be involved in decision making with AI-only recalls requiring human arbitration. Standalone AI algorithms used without prompting maintain unbiased screening reading performance, but reading with prompts should be tested prospectively and ideally provided for arbitration., Competing Interests: Declaration of Competing Interest The authors declare the following financial interests/personal relationships which may be considered as potential competing interests: [Support for this meeting was from Newnham College, University of Cambridge, Cambridge, UK. F.J.G. is supported by a Cancer Research UK early detection programme grant. H.A. is Chief Scientific Officer, Preemptive Medicine and Health, Flagship Pioneering.]., (Copyright © 2023 The Authors. Published by Elsevier B.V. All rights reserved.)

Objective: To assess the variation between racial/ethnic groups in emergency department (ED) treatment of asthma for pediatric patients., Methods: This study was a cross-sectional analysis of pediatric (2-18 years) asthma visits among 6 EDs in the Upper Midwest between June 2011 and May 2012. We used mixed-effects logistic regression to assess the odds of receiving steroids, radiology tests, and returning to the ED within 30 days. We conducted a subanalysis of asthma visits where patients received at least 1 albuterol treatment in the ED., Results: The sample included 2909 asthma visits by 1755 patients who were discharged home from the ED. After adjusting for demographics, insurance type, and triage score, African American (adjusted odds ratio [aOR], 1.78; 95% confidence interval [CI], 1.40-2.26) and Hispanic (aOR, 1.64; 95% CI, 1.22-2.22) patients had higher odds of receiving steroids compared with whites. African Americans (aOR, 0.58; 95% CI, 0.46-0.74) also had lower odds of radiological testing compared with whites. Asians had the lowest odds of 30-day ED revisits (aOR, 0.26; 95% CI, 0.08-0.84), with no other significant differences detected between racial/ethnic groups. Subgroup analyses of asthma patients who received albuterol revealed similar results, with American Indians showing lower odds of radiological testing as well (aOR, 0.47; 95% CI, 0.22-1.01)., Conclusions: In this study, children from racial/ethnic minority groups had higher odds of steroid administration and lower odds of radiological testing compared with white children. The underlying reasons for these differences are likely multifactorial, including varying levels of disease severity, health literacy, and access to care.

Objectives: The aims of this study were to determine differences in emergency department (ED) use by Native American (NA) children in rural and urban settings and identify factors associated with frequent ED visits., Methods: This cross-sectional, cohort study examined visits to 6 EDs: 2 rural, 2 midsize urban, and 2 large urban EDs from June 2011 to May 2012. Univariate and multiple regression analyses were conducted. Frequent ED visitors had more than 4 visits in the study period., Results: We studied 8294 NA visits (5275 patients) and 44,503 white visits (33,945 patients). Rural EDs had a higher proportion of NA patients, those below 200% of the income poverty level, and those who traveled more than 10 miles from their residence to attend the ED (all P < 0.05) compared with midsize and urban EDs. Native American patients had a high proportion of mental health diagnoses compared with whites (4.9% vs 1.9%, P < 0.001). Frequent ED visitors had greater odds of NA race, age younger than 1 year, public insurance, female sex, residence within less than 5 miles from the ED, and chronic disease., Conclusions: Native American children seem to have greater challenges compared with whites obtaining care in rural areas. Native American children were more likely to be frequent ED visitors, despite having to travel farther from their residence to the ED. Native American children visiting rural and midsize urban EDs had a much higher prevalence of mental health problems than whites. Additional efforts to provide both medical and mental health services to rural NA are urgently needed.

Emergency department (ED) utilization by American Indian (AI) children is among the highest in the nation. Numerous health disparities have been well documented in AI children, but limited information is available on parental experiences of care for AI children in the ED. Our objective was to understand parental attitudes towards ED care for AI children. Focus groups were held with AI parents/caregivers at five sites in the Upper Midwest. Traditional content analysis was used to identify themes. A total of 70 parents participated in ten focus groups. Three main themes were identified: healthcare environment, access to care, and interaction with providers. Healthcare environment issues included availability of specialists, wait times, and child-friendly areas. Transportation and financial considerations were major topics in access to care. Issues in interaction with providers included discrimination, stereotyping, and trust. This is one of the first studies to assess parent perspectives on ED use for AI children. Obtaining parental perspectives on ED experiences is critical to improve patient care and provide important information for ED providers.

Background: Children who leave the emergency department (ED) without complete evaluation or care (LWCET) have poorer outcomes in general. Previous studies have found that American Indian (AI) children have higher rates of LWCET than other racial or ethnic groups. Therefore, this study aims to examine LWCET in AI children by exploring differences by ED location and utilization patterns., Methods: This is a retrospective cohort study of five EDs in the upper Midwest between June 2011 and May 2012. We included all visits by children aged 0-17 who identified as African American (AA), AI or White. Logistic regression was used to determine differences in LWCET by race and ED location controlling for other possible confounding factors including sex, age, insurance type, triage level, distance from ED, timing of visit, and ED activity level., Results: LWCET occurred in 1.73% of 68,461 visits made by 47,228 children. The multivariate model revealed that AIs were more likely to LWCET compared to White children (Odds Ratio (OR) = 1.62, 95% Confidence Interval (CI) = 1.30-2.03). There was no significant difference in LWCET between AA and White children. Other factors significantly associated with LWCET included triage level, distance from the ED, timing of visit, and ED activity level., Conclusion: Our results show that AI children have higher rates of LWCET compared to White children; this association is different from other racial minority groups. There are likely complex factors affecting LWCET in AI children throughout the upper Midwest, which necessitates further exploration.

The objective of this study was to determine the patient characteristics and health care utilization patterns that predict frequent emergency department (ED) use (≥4 visits per year) over time to assist health care organizations in targeting patients for care management. This was a retrospective, population-based study of 13,265 Medicaid children aged 0-20 years who were attributed to a single pediatric accountable care organization for at least 2 consecutive years between June 2012 and May 2015. Year-to-year persistence as a frequent ED user was 36.3% (95% confidence interval [CI]: 33.4 to 38.4), which does not support the notion that once a frequent user, always a frequent user. Hence, interventions to reduce frequent ED use may appear to be effective when ED use would have regressed toward the mean without any intervention. At an individual patient level, predictability of frequent ED use was 0.437 (95% CI: 0.358 to 0.485) across frequent ED users of all ages compared with 0.723 (95% CI: 0.435 to 0.824) for those aged <1 year. Accordingly, this latter group may be a better target for interventions than frequent ED users generally.

Objectives: To examine emergency department (ED) visits for mental health concerns by American Indian children in a multicenter cohort. To analyze demographic and clinical factors, the types of mental health concerns, and repeat mental health visits., Study Design: Cross-sectional study of children 5-18 years old who visited 1 of 6 EDs in the Upper Midwest from June 2011 to May 2012 and self-identified as white or American Indian. Mental health visits were identified by primary diagnosis and reasons for visit and were categorized into diagnostic groups. We explored racial differences in ED visits for mental health, diagnostic groups, and repeat mental health visits. Analysis involved χ(2) tests, Cochran-Mantel-Haenszel tests, and regression models including age, triage, timing, and insurance, and their interactions with race., Results: We identified 26 004 visits of which 1545 (5.4%) were for a mental health concern. The proportion of visits for mental health differed by race and age. American Indian children had lower odds of a mental health visit for 5-10 year olds (OR, 0.40; 95% CI, 0.26-0.60), but higher odds for 11-17 year olds (OR, 1.62; 95% CI, 1.34-1.95). In the older age group, American Indian children were seen primarily for depression and trauma- and stressor-related disorders, whereas white children were seen primarily for depression and disruptive, impulse control, and conduct disorders. Repeat visits were not different by race., Conclusions: Differences were noted in mental health visits between American Indian and white children and were influenced by age. These findings warrant further investigation into care-seeking patterns and treatment for mental health in American Indian children., (Copyright © 2016 Elsevier Inc. All rights reserved.)

Context: Underscreening and problematic repeat lead testing in children., Objective: Identify proportion of underscreening for elevated blood-lead levels in children. For children who receive a lead test, measure the level of problematic repeat lead tests, defined as those with a high probability of not meeting recommended guidelines for lead testing in children measured using a combination of patients' age, test type and sequencing, days between tests, and encounter diagnosis coding., Design: A population-based retrospective cross-sectional design., Setting: All health care services organizations in the state of Minnesota that delivered health services to the defined study population., Participants: The study population was a Medicaid cohort of 12 436 children aged 0 to 18 years observed over a 1-year period., Main Outcome Measures: Proportion of eligible children not receiving at least 1 lead test; proportion of problematic repeat lead tests., Results: Thirty-five percent of children who should have received at least 1 lead test (n = 1714) during the study period did not. A total of 1856 children had at least 1 lead test and 190 had 2 or more. Fifty percent (50%) of the repeat tests were identified as problematic, representing 5.1% of the lead tests performed. Repeat tests performed in different health systems than the systems where the initial tests were performed had 5.3 times greater odds (adjusted odds ratio: 5.3 [95% confidence interval, 2.8-9.9]) of being problematic., Conclusions: The current approach to delivering mandatory lead testing across the state Medicaid population does not ensure that children are appropriately tested and has potential inefficiencies in that testing when it does take place. Use of multiple health care systems is associated with increased potential inefficiencies. Future Medicaid accountable care agreements between the state Medicaid program and participating health systems should emphasize clear population accountability for test screenings to improve patients' safety. A central queryable health resource or health information exchange may enable this.

Background: American Indian children have high rates of emergency department (ED) use and face potential discrimination in health care settings., Objective: Our goal was to assess both implicit and explicit racial bias and examine their relationship with clinical care., Research Design: We performed a cross-sectional survey of care providers at 5 hospitals in the Upper Midwest. Questions included American Indian stereotypes (explicit attitudes), clinical vignettes, and the Implicit Association Test. Two Implicit Association Tests were created to assess implicit bias toward the child or the parent/caregiver. Differences were assessed using linear and logistic regression models with a random effect for study site., Results: A total of 154 care providers completed the survey. Agreement with negative American Indian stereotypes was 22%-32%. Overall, 84% of providers had an implicit preference for non-Hispanic white adults or children. Older providers (50 y and above) had lower implicit bias than those middle aged (30-49 y) (P=0.01). American Indian children were seen as increasingly challenging (P=0.04) and parents/caregivers less compliant (P=0.002) as the proportion of American Indian children seen in the ED increased. Responses to the vignettes were not related to implicit or explicit bias., Conclusions: The majority of ED care providers had an implicit preference for non-Hispanic white children or adults compared with those who were American Indian. Provider agreement with negative American Indian stereotypes differed by practice and respondents' characteristics. These findings require additional study to determine how these implicit and explicit biases influence health care or outcomes disparities.

Background: Racial disparities are frequently reported in emergency department (ED) care., Objectives: To examine racial differences in triage scores of pediatric ED patients. We hypothesized that racial differences existed but could be explained after adjusting for sociodemographic and clinical factors., Methods: We examined all visits to two urban, pediatric EDs between August 2009 and March 2010. Demographic and clinical data were electronically extracted from the medical record. We used logistic regression to analyze racial differences in triage scores, controlling for possible covariates., Results: There were 54,505 ED visits during the study period, with 7216 (13.2%) resulting in hospital admission. White patients accounted for 36.4% of visits, African Americans 28.5%, Hispanics 18.0%, Asians 4.1%, and American Indians 1.8%. After adjusting for potential confounders, African American (adjusted odds ratio [aOR] 1.89, 95% confidence interval [CI] 1.69-2.12), Hispanic (aOR 1.77, 95% CI 1.55-2.02), and American Indian (aOR 2.57, 95% CI 1.80-3.66) patients received lower-acuity triage scores than Whites. In three out of four subgroup analyses based on presenting complaints (breathing difficulty, abdominal pain, fever), African Americans and Hispanics had higher odds of receiving low-acuity triage scores. No racial differences were detected for patients with presenting complaints of laceration/head injury/arm injury. However, among patients admitted to the hospital, African Americans (aOR 1.47, 95% CI 1.13-1.90) and Hispanics (aOR 1.71, CI 1.22-2.39) received lower-acuity triage scores than Whites., Conclusion: After adjusting for available sociodemographic and clinical covariates, African American, Hispanic, and American Indian patients received lower-acuity triage scores than Whites., (Copyright © 2016 Elsevier Inc. All rights reserved.)

Repetition by clinicians of the same tests for a given patient is common. However, not all repeat tests are necessary for optimal care and can result in unnecessary hardship. Limited evidence suggests that an electronic health record may reduce redundant laboratory testing and imaging by making previous results accessible to physicians. The purpose of this study is to establish a baseline by characterizing repeat testing in a pediatric population and to identify significant risk factors associated with repeated tests, including the impact of using multiple health systems. A population-based retrospective cross-sectional design was used to examine initial and repeat test instances, defined as a second test following an initial test of the same type for the same patient. The study population consisted of 8760 children with 1-25 test claims over a 1-year period. The study setting included all health care service organizations in Minnesota that generated these claims. In all, 17.2% of tests met the definition of repeat test instances, with several risk factors associated with per patient repeat test levels. The incidence of repeat test instances per patient was significantly higher when patients received care from more than 1 health system (adjusted incidence rate ratio 1.4; 95% confidence interval: 1.3-1.5). Repeat test levels are significant in pediatric populations and potentially actionable. Interoperable health information technology may reduce the incidence of repeat test instances in pediatric populations by making prior test results readily accessible. (Population Health Management 2016;19:102-108).

Objective: To assess the association between the length of consistent primary care as part of an accountable care organization (attribution length) and population-level and same-hospital readmissions. Readmission studies are generally focused on same-hospital readmissions rather than readmissions to any hospital (population-level readmissions)., Study Design: A retrospective study of Medicaid claims data for 28,794 unique pediatric patients attributed to a single children's hospital between September 2013 and May 2015. Study used logistic regression to estimate the impact of attribution length on readmissions and a zero-inflated Poisson model to assess the impact of attribution length on readmission cost and readmission days., Results: The study showed attribution length was associated with a significant reduction in the population-level 30-day readmission rate from 8.9%-6.2% (P = .010) primarily by reducing readmissions that occurred at hospitals other than the discharging hospital. There was no significant reduction in the same-hospital readmission rate. Readmissions to a different hospital occurred in 37% of readmissions. Although not significant at the P = .05 level, attribution length was associated with a 44% reduction (P = .100) in 30-day readmission costs or a 5.0% reduction in the cost of an inpatient episode of care and a 53% reduction (P = .019) in readmission days., Conclusions: Consistent primary care (attribution length) may be able to reduce 30-day, pediatric Medicaid patients' readmissions at the population level. The decrease occurred primarily in readmissions to hospitals other than the discharging hospital. There was no decrease in the rate of same-hospital readmissions., (Copyright © 2016 Elsevier Inc. All rights reserved.)

Importance: Little is known about the effect of pediatric accountable care organizations (ACOs) on the use and costs of health care resources, especially in a Medicaid population., Objective: To assess the association between the length of consistent primary care (length of attribution) as part of an ACO and the use and cost of health care resources in a pediatric Medicaid population., Design, Setting, and Participants: A retrospective study of Medicaid claims data for 28,794 unique pediatric patients covering 346,277 patient-attributed months within a single children's hospital. Data were collected for patients attributed from September 1, 2013, to May 31, 2015. The effect of the length of attribution within a single hospital system's ACO on the use and costs of health care resources were estimated using zero-inflated Poisson distribution regression models adjusted for patient characteristics, including chronic conditions and a measure of predicted patient use of resources., Exposures: Receiving a plurality of primary care at an ACO clinic during the preceding 12 months (attribution to the ACO)., Main Outcomes and Measures: The primary outcome measure was the length of attribution at an ACO clinic compared with subsequent inpatient hospitalization and subsequent use and cost of outpatient and ancillary health care resources., Results: Among the 28,794 pediatric patients receiving treatment covering 346,277 patient-attributed months during the study period, continuous attribution to the ACO for more than 2 years was associated with a decrease (95% CI) of 40.6% (19.4%-61.8%) in inpatient days but an increase (95% CI) of 23.3% (2.04%-26.3%) in office visits, 5.8% (1.4%-10.2%) in emergency department visits, and 15.3% (12.5%-18.0%) in the use of pharmaceuticals. These changes in the use of health care resources combined resulted in a cost reduction of 15.7% (95% CI, 6.6%-24.8%). At the population level, the impact of consistent primary care was muted by the many patients in the ACO having shorter durations of participation., Conclusions and Relevance: These findings suggest significant and durable reductions of inpatient use and cost of health care resources associated with longer attribution to the ACO, with attribution as a proxy for exposure to the ACO's consistent primary care. Consistent primary care among the pediatric Medicaid population is challenging, but these findings suggest substantial benefits if consistency can be improved.

Background: Adverse events, diverse and often costly, commonly occur in pediatric intensive care units (PICUs). Serious safety events (SSEs) are captured through well-developed systems, typically by voluntary reporting. Less serious safety events (LSSEs), including close calls, however, occur at a higher frequency than those that result in immediate harm or death but are underestimated by standard reporting systems. LSSEs can reveal system defects and precede serious events resulting in patient or provider harm., Methods: A unique active surveillance program was created at Children's Hospitals and Clinics of Minnesota to quantify and categorize, and, ultimately reduce, LSSEs, in PICUs. Premedical college graduates without formal health care training daily canvassed the PICUs and facilitated reporting of LSSEs at the point of care. Events were recorded on a Web application and stored in a relational database management system. Events were enumerated and categorized according to distinctive characteristics (Theme Index) and real or potential harm (Harm Index)., Results: Some 1,980 PICU patients, representing 10,766 PICU patient-days in a 15-month period (June 1, 2013- August 31, 2014) experienced 2,465 LSSEs-5.4 LSSEs/ day or 0.23 LSSEs/patient-day. Such events resulted in a patient intervention 38% of the time. Some 158 quality/safety improvement projects were initiated during the observation period, 74 of which have been completed. Quality/safety information was broadcasted to providers, local leadership, and hospital management., Conclusions: LSSEs occur frequently in our PICUs. Non-health care providers can cost-effectively facilitate reporting by actively canvassing PICU providers on a daily basis and can contribute to quality/safety improvement projects and local safety culture. Reported events can serve as a focus for quality/safety improvement projects. A Web application and mobile tablet interfaces are efficient tools to record events.

Multiple methods for detecting asthma encounters are used today in public surveillance, quality reporting, and clinical research. Failure to detect asthma encounters can make it difficult to measure the scope and effectiveness of hospital or community-based interventions important in comparative effectiveness research and accountable care. Given the pairing of asthma with certain respiratory conditions, the objective of this study was to develop and test an asthma detection algorithm with specificity and sensitivity using 2 criteria: (1) principal discharge diagnosis and (2) asthma diagnosis code position. A medical record review was conducted (n=191) as the gold standard for identifying asthma encounters given objective criteria. The study team observed that for certain principal respiratory diagnoses (n=110), the observed odds ratio that encounters were for asthma when asthma was coded in the second or third code position was not significantly different than when asthma was coded as the principal diagnosis, 0.36 (P=0.42) and 0.18 (P=0.14), respectively. In contrast, the observed odds ratio was significantly different when asthma was coded in the fourth or fifth positions (P<.001). This difference remained after adjusting for covariates. Including encounters with asthma in 1 of the 3 first positions increased the detection sensitivity to 0.84 [95% confidence interval (CI): 0.76-0.92] while increasing the false positive rate to 0.19 [95% CI: 0.07-0.31]. Use of the proposed algorithm significantly improved the reporting accuracy [0.83 95%CI:0.76-0.90] over use of (1) the principal diagnosis alone [0.55 95% CI:0.46-0.64] or (2) all encounters with asthma 0.66 [95% CI:0.57-0.75]. Bed days resulting from asthma encounters increased 64% over use of the principal diagnosis alone. Given these findings, an algorithm using certain respiratory principal diagnoses and asthma diagnosis code position can reliably improve asthma encounter detection for population-based health impact measurement.

Background: In the critical care unit, complexity of care can contribute to both medical errors and increased costs, particularly when clinicians are forced to rely on memory. Checklists can be used to improve safety and reduce cost. A number of omission-related adverse events in 2010 prompted the development of a checklist to reduce the possibility of similar future events., Methods: The PICU Safety Checklist was implemented in the pediatric ICU (PICU) at Children's Hospitals and Clinics of Minnesota. During a 21-month period, the checklist was used to prompt the care team to address quality and safety items during rounds. The initial checklist was paper, with two subsequent versions being incorporated into the electronic medical record (EMR)., Results: The daily safety checklist was successfully implemented in the PICU. Work-flow improvements based on regular multidisciplinary feedback led to more consistent use of the checklist. Improvements on all quality and safety metrics were identified, including invasive device use, medication costs, antibiotic and laboratory test use, and compliance with standards of care. Staff satisfaction rates were > 80% for safety, communication, and collaboration., Conclusion: By using a daily safety checklist in the pediatric critical care unit, we improved quality and safety, as well as the collaborative culture among all clinicians. Incorporating the checklist into the EMR improved compliance and accountability, ensuring its application to all patients. Clinicians now often individually address many checklist items outside the formal rounding process, indicating that the checklist content has become part of their usual practice. A successful implementation showing tangible clinical improvements can lead to interest and adoption in other clinical areas within the institution.

Objectives: Accountable care puts pressure on hospitals to manage care episodes. Initial length of stay (ILOS) and readmission risk are important elements of a care episode and measures of care quality. Understanding the association between these two measures can guide hospital efforts in managing care episodes. This study was designed to explore the association between ILOS and readmission risk in a cohort of pediatric asthma patients., Materials and Methods: The sample cohort (n = 4965) consisted of all asthma patients discharged from Children's Hospitals and Clinics of Minnesota (CHC MN) from January 2008 through August 2012. Asthma discharges included cases with a principal diagnosis of asthma or certain respiratory cases with asthma listed as a secondary diagnosis. Multiple logistic regression was used to test associations, adjusting for covariates., Results: Adjusting for covariates, we found no significant association between ILOS and readmission (OR: 1.04 [95% CI: 0.98-1.10]). Analyzing ILOS categorically by length of stay, one-day stays did not have a significantly higher readmission risk (OR:1.27 [95% CI: 0.87-1.85]) than two-day stays, which had the lowest observed readmission risk. Risk increased as ILOS exceeded two days but was not significantly different by day. We found no association when comparing the difference in actual versus expected ILOS and readmission risk (shorter than expected OR: 1.13 [95% CI: 0.74-1.71]; longer than expected OR: 0.97 [95% CI: 0.69-1.38])., Conclusions: Attempts to prolong ILOS would dramatically increase costs with little impact on readmissions. For example, increasing one-day visits to two-day visits would increase hospital patient days 38% (1870 d) in this cohort while decreasing total readmissions by 3.8% [95% CI: 3.6-4.0%]. Understanding the mechanisms that impact readmissions is essential in evaluating cost-effective approaches to improving patient outcomes and lowering the cost of care.

Objective: The objective of this study was to examine the association of race and language on laboratory and radiological testing in the pediatric emergency department (ED)., Methods: This retrospective, case-cohort study examined laboratory and radiological testing among patients discharged home from 2 urban, pediatric EDs between March 2, 2009, and March 31, 2010., Results: There were 75,254 visits among 49,164 unique patients, of whom 31.0% had laboratory and 30.5% had radiological testing. African American (adjusted odds ratio [aOR], 0.93; confidence interval [CI], 0.89-0.98; P = 0.004) and biracial racial categories (aOR, 0.91; CI, 0.86-0.98; P = 0.007) were associated with decreased odds of laboratory testing compared with non-Hispanic whites. Similarly, Native American (aOR, 0.82; CI, 0.73-0.94), African American (aOR0.81; CI, 0.72-0.81), biracial (aOR, 0.82; CI, 0.77-0.88), Hispanic (aOR.76; CI, 0.72-0.81), and "other" (aOR, 0.84; CI, 0.73-0.97) racial categories were each associated with lower odds of radiological testing compared with non-Hispanic whites. Subgroup analysis of visits with a final diagnosis of fever and upper respiratory tract infection, conditions for which there were few treatment protocols, confirmed the racial differences. Subgroup analysis in visits for head injury, for which there is an established evaluation protocol, did not find a lower odds of laboratory or radiological testing by race compared with non-Hispanic whites., Conclusions: Racial disparities in laboratory and radiological testing were present in pediatric ED visits. No racial differences were seen in the radiological and laboratory charges in the head injury subgroup, suggesting that evaluation algorithms can ameliorate racial disparities in pediatric ED care.

Objective: To determine the interrater reliability and construct validity of 3 separate assessment tools for assessing trainee skills in pediatric airway endoscopy simulation., Design: An Objective Structured Assessment of Technical Skills (OSATS) was developed in which examinees were asked to name and assemble the airway foreign body instruments and retrieve a foreign body from an infant airway mannequin. Each examinee's performance was assessed in a blinded fashion by 3 pediatric otolaryngology faculty at separate residency programs using 3 assessment tools: (1) objective quantifiable measures list (eg, assists needed, forceps openings, foreign body drops), (2) 15-point OSATS checklist, and (3) Global Rating Index for Technical Skills (GRITS). Setting Otolaryngology residency program. Subjects Examinees (medical students, n = 3; otolaryngology residents, n = 17; pediatric otolaryngology faculty, n = 3) and raters (n = 3). Main Outcome Measures Interrater reliability and construct validity. Results Anonymized split-screen videos simultaneously capturing each examinee's instrument handling and the endoscopic videos were created for all 23 examinees. Nineteen videos were chosen for review by 3 raters. The interrater reliability as measured by the intraclass correlation for objective quantifiable measures ranged from 0.46 to 0.98. The intraclass correlation coefficient was 0.95 for the 15-point OSATS checklist and 0.95 for the GRITS; both showed a high degree of construct validity with scores correlating with previous experience. Conclusion Assessment tools for skills assessments must have high interrater reliability and construct validity. When assessing trainee skills in pediatric airway foreign body scenarios, the 15-point OSATS checklist developed by this group or the GRITS meets these criteria.

Background: Although reports of reduced nosocomial infections (NI) in very low birth weight infants have been published, the durability of these gains and changes in secondary outcomes, and clinical practices have less often been published., Methods: This was a retrospective, observational study of NI reduction in very low birth weight infants at two hospital campuses. The intervention began in 2005 with our renewed quality improvement efforts to reduce NI. We compared outcomes before (2000-2005) and after (2006-2009) the intervention by using univariate and multiple regression analyses., Results: We reduced NI by 50% comparing 2000-2005 to 2006-2009 (23.6% vs 11.6%, P < .001). Adjusting for covariates, the odds ratio for NI was 0.33 (confidence interval, 0.26 - 0.42, P < .001) in the more recent era. NI were lower even in infants with birth weight 501-1000 grams (odds ratio = 0.38, confidence interval, 0.29 - 0.51, P < .001). We also reduced bronchopulmonary dysplasia (30.2% vs 25.5%, P = .001), median days to regain birth weight (9 vs 8, P = .04), percutaneously placed central venous catheter use (54.8% vs 43.9%, P = .002), median antibiotic days (8 vs 6, P = .003), median total central line days (16 vs 15, P = .01), and median ventilator days (7 vs 5, P = .01). We sustained improvements for three years., Conclusions: Quality improvement efforts were associated with sustained reductions in NI, bronchopulmonary dysplasia, antibiotic use, central line use, and ventilator days.

Background: Gastroschisis patients are commonly small for gestational age (SGA, birth weight [BW] < 10th centile). However, the extent, symmetry and causes of that growth restriction remain controversial., Methods: We compared BW, crown-heel length (LT), occipitofrontal circumference (OFC) and ponderal index (PI) in 179 gastroschisis cases and 895 matched controls by univariate and multiple regression. Fetal ultrasounds (N = 80) were reviewed to determine onset of growth restriction. Placental histology was examined in 31 gastroschisis patients whose placental tissue was available and in 29 controls., Results: Gastroschisis cases weighed less than controls (BW = 2400 ± 502 g vs. 2750 ± 532 g, p < 0.001) and their BW frequency curve was shifted to the left, indicating lower BW as a group compared to controls (p < 0.001 by Kolmogorov-Smirnov test). BW differences varied from -148 g at 33 weeks to -616 g at 38 weeks gestation. Intrauterine growth restriction was symmetric with gastroschisis patients having a shorter LT (45.7 ± 3.3 vs. 48.4 ± 2.7 cm, p < 0.001), smaller OFC (31.9 ± 1.9 vs. 32.9 ± 1.6 cm, p < 0.001), but larger ponderal index (2.51 ± 0.37 vs. 2.40 ± 0.16, p < 0.001) compared to controls. Gastroschisis patients had a similar reduction in BW (-312 g, 95% confidence interval [CI] = -367, -258) compared to those with chromosomal abnormalities (-239 g, CI = -292, -187). Growth deficits appeared early in the second trimester and worsened as gestation increased. Placental chorangiosis was more common in gastroschisis patients than controls, even after removing all SGA patients (77% vs. 42%, p = 0.02)., Conclusions: Marked, relatively symmetric intrauterine growth restriction is an intrinsic part of gastroschisis. It begins early in the second trimester, and is associated with placental chorangiosis.

Objective: Quality improvement collaboratives (QICs) can improve short-term outcomes, but few have examined their long-term results. This study evaluated the changes in treatment practices and outcomes associated with participation in multiple sequential QICs., Design and Methods: This retrospective, 9-year, pre-post study of very low birth weight infants, we assessed treatment and outcomes from the 8 NICUs of the Reduce Lung Injury (ReLI) group of a QIC sponsored by the Vermont Oxford Network (VON). We analyzed data from 1998 (pre-ReLI), 2001 (last ReLI year), and 2006 (5 years after ReLI) by using univariate and multiple regression., Results: A total of 4065 very low birth weight infants were treated in ReLI NICUs in 1998, 2001, and 2006. From 1998 to 2006, the ReLI group decreased delivery room intubation (70% vs 52%; adjusted odds ratio [aOR]: 0.2 [95% confidence interval (CI): 0.2-0.3]; P < .001), conventional ventilation (75% vs 62%; aOR: 0.3 [95% CI: 0.2-0.4]; P < .001), and postnatal steroids for BPD (35% vs 10%; aOR: 0.09 [95% CI: 0.07-0.1]; P < .001). They increased the use of nasal continuous positive airway pressure (57% vs 78%; aOR: 3.3 [95% CI: 2.7-3.9]; P < .001). BPD-free survival remained unchanged (68% vs 66%; aOR: 0.9 [95% CI: 0.7-1.1]; P = .16), the BPD rate increased (25% vs 29%; aOR: 1.3 [95% CI: 1.1-1.6]; P = .017), survival to discharge increased (90% vs 93%; aOR: 1.5 [95% CI: 1.1-2.2]; P < .001), and nosocomial infections decreased (18% vs 15%; aOR: 0.8 [95% CI: 0.6-0.99]; P = .045)., Conclusions: Participation in VON-sponsored QICs was associated with sustained implementation of potentially better respiratory practices, increased survival, and reduced nosocomial infections. The BPD-free survival rate did not change, and the BPD rate increased. Implemented changes endured for at least 5 years after the QIC.

Objective: The objective of the study was to determine factors predicting outcome in newborns with gastroschisis., Methods: A retrospective analysis of 155 consecutive cases admitted from 1 January 1990 to 31 December 2007 was performed. Prenatal ultrasound findings were available for 89 of 155 (57%) patients and were compared with final outcome. Both univariate and multiple regression analyses were used., Results: All patients survived to discharge home. The primary outcome measure was length of stay. Multiple regression identified 4 factors associated with length of stay: (1) gestational age (P = .004), (2) nonelective silo (P < .001), (3) gastrointestinal (GI) complication (intestinal atresia, perforation, or resection) (P < .001), and (4) non-GI anomaly (P = .029). Non-GI anomalies occurred in 17 of 155 (11%) patients and tended to increase the risk of a nonelective silo or GI complication (59% vs 39%, P = .190). Dilated bowel (>10 mm) on prenatal ultrasound was associated with GI complications (22% vs 3%, P = .010). However, 78% of patients with dilated bowel on prenatal ultrasound did not have a GI complication. The absence of dilated bowel on prenatal ultrasound accurately predicted the absence of GI complications in 97% of cases., Conclusion: Prematurity, nonelective silo, GI complications, and non-GI anomalies predict the short-term outcome of newborns with gastroschisis. Prenatal ultrasound serves primarily to predict the absence of GI complications.

Objective: The objective of this study was to compare the primary and secondary outcomes of very low birth weight infants before and after participation in the Breathsavers Group of the Vermont Oxford Network-sponsored Neonatal Intensive Care Quality Collaborative., Methods: Hospitals that participated in the Breathsavers Group contributed clinical data on the outcomes of their very low birth weight infants to the Vermont Oxford Network using standardized clinical definitions, data forms, and inclusion criteria. Outcomes from the last year of the collaborative, 2003, were compared with those from the baseline year, 2001. Models for treatment practices and outcomes measures were adjusted for within-hospital correlation (clustering) and standard risk factors that were present at birth., Results: Bronchopulmonary dysplasia dropped significantly in 2003 compared with the baseline year. Survival improved but not significantly. In addition, severe retinopathy of prematurity, severe intraventricular hemorrhage, and supplemental oxygen at discharge dropped significantly. The use of conventional ventilation at any time during the initial hospitalization, postnatal steroids, and time to first dose of surfactant all decreased significantly. The use of nasal continuous positive airway pressure at any time during hospitalization increased. The use of high-frequency ventilation, delivery room intubation, and surfactant at any time during hospitalization did not change., Conclusions: The Breathsavers Group improved both clinical care processes and clinical outcomes during the Neonatal Intensive Care Quality Collaborative.

Objective: The objective of this study was to describe development and implementation of potentially better practices to reduce bronchopulmonary dysplasia in very low birth weight infants (birth weight: 501-1500 g)., Methods: Results of Breathsavers Group meetings, conference calls and critically appraised topic summaries were used to construct potentially better practices. Implementation plans and experiences were reported by participants and collated., Results: The Breathsavers Group developed 13 potentially better practices, based on published evidence and expert opinion. Participants determined which potentially better practices to implement and implementation methods. Participating NICUs implemented an average of 5 potentially better practices (range: 3-9). The Breathsavers Group also developed a resource kit, identified common obstacles to implementation, and initiated research to define bronchopulmonary dysplasia better., Conclusions: Multiinstitutional collaboration facilitated development and implementation of potentially better practices to reduce bronchopulmonary dysplasia.

Objective: Nosocomial bloodstream infections (NBIs) are associated with serious morbidity and prolonged length of stay (LOS) in very low birth weight (VLBW) infants. However, the marginal costs and excess LOS associated with these infections have never been measured in different birth weight (BW) categories after adjustment for many of the potentially confounding demographic variables, comorbidities, and treatments. The objective of this study was to measure the marginal cost and excess LOS caused by NBIs in surviving VLBW infants in different BW categories., Methods: This retrospective study examined data previously collected as part of the Neonatal Intensive Care Quality Improvement Collaborative 2000 and the Vermont Oxford Network clinical outcomes database. Univariate analyses and multiple regression were used to examine the effect of NBIs on hospital costs and LOS. Seventeen neonatal intensive care units that participated in the Neonatal Intensive Care Quality Improvement Collaborative 2000 submitted both clinical and financial data on their VLBW infants who were born from January 1, 1998, to December 31, 1999. This study included data from both university and community hospitals., Results: NBIs occurred in 19.7% of 2809 patients included in this study. NBI was associated with significantly increased treatment costs for infants with BW 751 to 1500 g. The marginal costs of NBIs, as estimated by multiple regression, varied from 5875 dollars for VLBW infants with a BW of 401 to 750 g to 12,80 dollars for those with BW of 751 to 1000 g. LOS was significantly increased in all BW categories. The excess LOS estimated by multiple regression varied from 4 days in VLBW infants with a BW of 1001 to 1251 g to 7 days in those with a BW of 751 to 1000 g., Conclusions: NBIs are associated with increased hospital treatment costs and LOS but by varying amounts depending on the BW. Preventing a single NBI could reduce the treatment costs of a VLBW infant by at least several thousand dollars. These savings are a greater percentage of the total treatment costs in VLBW infants with BW 1001 to 1500 g than in smaller infants.

Objective: Adherence to basic quality improvement principles enhances the implementation of potentially better practices (PBPs) and requires extensive planning and education. Even after PBPs have been identified and acknowledged as desirable, effective implementation of these practices does not occur easily. The objective of this study was to identify and assess implementation strategies that facilitate quality improvements in the respiratory care of extremely low birth weight infants., Methods: The 9 members of the Neonatal Intensive Care Quality Improvement Collaborative Year 2000 Reducing Lung Injury focus group identified 9 PBPs in a evidence-based manner to decrease chronic lung disease in extremely low birth weight newborns. Each site implemented several or all PBPs based on a site-specific selection process. Each site was asked to submit 1 or more examples of experiences that highlighted effective implementation strategies. This article reports these examples and emphasizes the principles on which they are based., Results: The 9 participating institutions implemented a total of 57 PBPs (range: 1-9; median: 5). Including previous implementation, the 9 participating institutions implemented a total of 70 of a possible 81 PBPs before or during the study period (range: 5-9; median: 8). We report 7 approaches that facilitated PBP implementation: information availability, feedback, perseverance, collaboration, imitation, recognition of implementation complexity, and tracking of process indicators., Conclusions: Quality improvement efforts are enhanced by identifying and then implementing PBPs. In our experience, implementation of these PBPs can be difficult. Implementation strategies, such as those identified in this article, can improve the chances that quality improvement efforts will be effective.

Objective: To make measurable improvements in the quality and cost of neonatal intensive care using a multidisciplinary collaborative quality improvement model., Design: Interventional study. Data on treatment costs were collected for infants with birth weight 501 to 1500 g for the period of January 1, 1994 to December 31, 1997. Data on resources expended by hospitals to conduct this project were collected in a survey for the period January 1, 1995 to December 31, 1996., Setting: Ten self-selected neonatal intensive care units (NICUs) received the intervention. They formed 2 subgroups (6 NICUs working on infection, 4 NICUs working on chronic lung disease). Nine other NICUs served as a contemporaneous comparison group., Patients: Infants with birth weight 501 to 1500 g born at or admitted within 28 days of birth between 1994 and 1997 to the 6 study NICUs in the infection group (N = 2993) and the 9 comparison NICUs (N = 2203); infants with birth weight 501 to 1000 g at the 4 study NICUs in the chronic lung disease group (N = 663) and the 9 comparison NICUs (N = 1007)., Interventions: NICUs formed multidisciplinary teams which worked together to undertake a collaborative quality improvement effort between January 1995 and December 1996. They received instruction in quality improvement, reviewed performance data, identified common improvement goals, and implemented "potentially better practices" developed through analysis of the processes of care, literature review, and site visits., Main Outcome Measures: Treatment cost per infant is the primary economic outcome measure. In addition, the resources spent by hospitals in undertaking the collaborative quality improvement effort were determined., Results: Between 1994 and 1996, the median treatment cost per infant with birth weight 501 to 1500 g at the 6 project NICUs in the infection group decreased from $57 606 to $46 674 (a statistical decline); at the 4 chronic lung disease hospitals, for infants with birth weights 501 to 1000 g, it decreased from $85 959 to $77 250. Treatment costs at hospitals in the control group rose over the same period. There was heterogeneity in the effects among the NICUs in both project groups. Cost savings were maintained in the year following the intervention. On average, hospitals spent $68 206 in resources to undertake the collaborative quality improvement effort between 1995 and 1996. Two thirds of these costs were incurred in the first year, with the remaining third in the second year. The average savings per hospital in patient care costs for very low birth weight infants in the infection group was $2.3 million in the post-intervention year (1996). There was considerable heterogeneity in the cost savings across hospitals associated with participation in the collaborative quality improvement project., Conclusion: Cost savings may be achieved as a result of collaborative quality improvement efforts and when they occur, they appear to be sustainable, at least in the short run. In high-cost patient populations, such as infants with very low birth weights, cost savings can quickly offset institutional expenditures for quality improvement efforts.

Objective: To determine whether a course of low-dose indomethacin therapy, when initiated within 24 hours of birth, would decrease ductal shunting in premature infants who received prophylactic surfactant in the delivery room., Design: Ninety infants, with birth weights of 600 to 1250 gm, were entered into a prospective, randomized, controlled trial to receive either indomethacin, 0.1 mg/kg per dose, or placebo less than 24 hours and again every 24 hours for six doses. Echocardiography was performed on day 1 before treatment and on day 7, 24 hours after treatment. A hemodynamically significant patent ductus arteriosus (PDA) was confirmed with an out-of-study echocardiogram, and the nonresponders were treated with standard indomethacin or ligation., Results: Forty-three infants received indomethacin (birth weight, 915 +/- 209 gm; gestational age, 26.4 +/- 1.6 weeks; 25 boys), and 47 received placebo (birth weight, 879 +/- 202 gm; gestational age, 26.4 +/- 1.8 weeks; 22 boys) (P = not significant). Of 90 infants, 77 (86%) had a PDA by echocardiogram on the first day of life before study treatment; 84% of these PDAs were moderate or large in size in the indomethacin-treated group compared with 93% in the placebo group. Nine of forty indomethacin-treated infants (21%) were study-dose nonresponders compared with 22 (47%) of 47 placebo-treated infants (p < 0.018). There were no significant differences between both groups in any of the long-term outcome variables, including intraventricular hemorrhage, duration of oxygen therapy, endotracheal intubation, duration of stay in neonatal intensive care unit, time to regain birth weight or reach full caloric intake, incidence of bronchopulmonary dysplasia, and survival. No significant differences were noted in the incidence of oliguria, elevated plasma creatinine concentration, thrombocytopenia, pulmonary hemorrhage, or necrotizing enterocolitis., Conclusion: The prophylactic use of low doses of indomethacin, when initiated in the first 24 hours of life in low birth weight infants who receive prophylactic surfactant in the delivery room, decreases the incidence of left-to-right shunting at the level of the ductus arteriosus.

The purpose of this study was to determine whether decreased Fc gamma RIII expression on the PMN of extremely low birth weight infants (ELBW) is due to decreased receptor synthesis or increased receptor shedding from the PMN surface. 42 ELBW, 12 larger infants and 14 adults were enrolled. Plasma and total cellular Fc gamma RIII were measured by ELISA, and PMN Fc gamma RIII expression was measured by flow cytometry. ELBW PMN plasma membrane expression of Fc gamma RIII as measured by log mean channel fluorescence (5.00 +/- 1.98 vs. 10.68 +/- 1.61, p < 0.050) and plasma Fc gamma RIII levels were both lower (7.5 +/- 6.1 vs. 82.4 +/- 64.8 nM, p < 0.05) than in adult controls. In follow-up studies, 14 ELBW (age = 29 +/- 14 days, range = 14-56 days) increased PMN expression of Fc gamma RIII (p < 0.001) but not plasma Fc gamma RIII. ELBW had lower total PMN-associated Fc gamma RIII than adults (2.3 +/- 0.9 vs. 6.8 +/- 2.2 ng/10(6) PMN, p = 0.006). ELBW's PMN produce less Fc gamma RIII than adults' PMN, and expression of this receptor is developmentally regulated.

Objective: We characterized the natural history and true mortality of congenital diaphragmatic hernia (CDH) in newborn patients by identifying all infants born with this condition in a fixed geographic region over a 2-year period. We examined this population to determine the frequency of intrauterine diagnosis, the outcome of prenatally diagnosed infants, and the impact of deaths in infants with an unsuspected diagnosis (the "hidden mortality") on the overall outcome of this condition., Design: This was a retrospective population survey of all infants born with CDH in Minnesota between June 1988 and June 1990., Setting: All Minnesota birth and death records were reviewed to identify patients with the diagnosis of CDH. A separate survey of all level 3 intensive care nurseries was conducted and the record of each identified patient was reviewed. Extracorporeal membrane oxygenation was available throughout the study period., Main Outcome Measure: Survival to hospital discharge and short-term morbidity were examined for each patient., Results: Survival was 60% (29/48). Eleven of 19 deaths occurred in patients born prematurely and/or with coexisting major anomalies. Eight percent (4/48) of patients died within the first hour of life prior to diagnosis (hidden mortality). Intrauterine diagnosis of CDH was made in 15 patients. Survival was 60% (9/15) in infants whose conditions were diagnosed in utero, a rate identical to that for infants whose conditions were diagnosed in the postnatal period (61% [20/33]). There was no relationship between age at fetal diagnosis and mortality., Conclusions: The hidden mortality of CDH was low. Almost half of the total mortality for CDH was associated with coexisting, additional anomalies. Patients who were not offered extracorporeal membrane oxygenation owing to prematurity, other major anomalies, or birth at a center that did not offer extracorporeal membrane oxygenation accounted for 84% (16/19) of deaths. These data will be useful for determining the impact of new therapeutic strategies on the mortality of CDH.

Nosocomial infections increase neonates' morbidity, hospital costs, and mortality. These infections occur most commonly in very low birth weight infants, who frequently required plastic intravascular catheters and parenteral nutrition. Diagnosis often relies on a combination of laboratory tests and nonspecific clinical signs. Criteria for diagnosing nosocomial infections have been published by the Centers for Disease Control (CDC) and should be used to standardize the identification of cases. Initial antibiotic therapy depends on (1) the bacterial species most likely to cause infection, (2) antibiotic resistance patterns in one's own hospital, (3) the patient's clinical condition, and (4) previous antibiotic therapy. Antibiotic coverage of both gram-positive and gram-negative bacteria is necessary. Following laboratory identification of the infecting organism and the antibiotic susceptibility results, the patient should be reevaluated and definitive therapy prescribed. Multiple antibiotics may be needed as definitive therapy if (1) the infecting organism is likely to develop resistant mutants during therapy (e.g., Pseudomonas species), (2) higher bactericidal serum activity is required than can usually be achieved with a single agent (e.g., enterococci, Listeria), (3) the patient is neutropenic or otherwise severely immunocompromised, or (4) blood cultures are persistently positive for bacteria despite appropriate therapy with a single agent. Attempts to prevent nosocomial bacteremias by routinely administering prophylactic vancomycin may hasten the development of vancomycin-resistant, coagulase-negative staphylococci or enterococci and should be avoided.

Little information is available regarding the effect of surfactant on outcome for infants born at or before 26 weeks of gestation. We addressed this issue by reviewing records of 310 infants born at gestational ages of 23 through 26 weeks who were admitted to our nursery from 1986, when surfactant was introduced, through 1990. Surfactant was administered to 154 infants (5 during a single-dose prevention study, 25 during a multiple-dose prevention study, 124 while receiving a Food and Drug Administration treatment investigational new drug); 156 infants were not treated with surfactant. Seventy-three percent of the treated infants survived, compared with 55% of the nontreated infants. Increased survival occurred at all gestational ages between 23 and 26 weeks but were greatest in infants born at 23 and 24 weeks. At follow-up, no differences in neurologic outcome were detected between surfactant-treated and nontreated infants. We conclude that surfactant use in extremely premature infants improves survival rates without increasing the proportion of impaired survivors.

Limits of viability of extremely premature infants have recently been addressed both in Europe and the United States. These reports, which demonstrate frequent adverse outcome of infants born before 26 weeks of gestation, have not considered the impact of surfactant therapy. We reviewed records of 445 infants born between 23 and 36 weeks gestation who were admitted to our nursery following the availability of surfactant treatment in 1986 through 1992. Two hundred and eighty-five infants were treated with surfactant (Survanta, Ross Laboratories) as part of controlled, prospective trials or as routine treatment under Food and Drug Administration approval. One hundred and fifty-six infants were unable to be treated with surfactant, as either they received placebo therapy during prospective trials or were born prior to approval of routine surfactant use in the United States. Four additional infants born following the commercial availability of surfactant did not receive surfactant therapy. Survival of untreated infants was 56% compared to 75% in treated infants (P < 0.001). Infants born at all gestational ages between 23 and 26 weeks had an increased likelihood of survival as a result of surfactant treatment. No differences in neurologic outcome between surfactant treated and non-treated infants were demonstrated at subsequent follow-up. We conclude that survival of extremely premature infants is improved following surfactant therapy and that subsequent neurologic outcome is not compromised as a result of this therapy.

Routine ACT tests cannot distinguish between prolonged blood clotting due to heparin effect or acquired abnormalities of the coagulation system after a loading dose of heparin. The purpose of this study was to examine an ACT test that inactivates heparin with Heparinase allowing for ACT assessment with and without heparin effect (HR-ACT with/without Heparinase, HemoTec, Inc.). The HR-ACT values were compared with the standard OR procedure that employed the Hemochron ACT. Twenty pediatric patients undergoing cardiopulmonary bypass for repair of cardiac defects were examined. All comparative ACT values were obtained from the same blood sample. Five sampling times were examined: 1) A baseline ACT was obtained before heparin had been administered; 2) A pre bypass ACT after a single heparin dose; 3) On bypass; 4) A post protamine ACT at the conclusion of surgery; and 5) In the Intensive Care Unit (PICU), 1 hour post protamine. The HemoTec HR-ACT with Heparinase and HR-ACT tests differentiated clotting time results that reflected coagulation status without the heparin effect. It identified those patients on bypass who were less than 5 kg, with prolonged ACTs that were due in part to hemodilution despite efforts at hemoconcentration.

A critical appraisal of four cohort studies examining the relationship between Ureaplasma urealyticum and chronic lung disease (CLD) of prematurity is presented. Three studies were concurrently conducted, but the fourth was conducted 4 years later when surfactant replacement was a widespread practice. Although infants were enrolled in all studies soon after birth before they had developed CLD, there were differences in patients population, the definition of colonization with U. urealyticum, neonatal management, and the definition of CLD of prematurity. Despite the differences, all four studies found an association between colonization and development of CLD of prematurity. A combined estimate of relative risk for the four studies was 1.91 (95% confidence interval, 1.54-2.37). When infants were categorized into groups by birth weight, the association was not observed in infants who weighed > 1,250 g. The association was also not observed in infants who weighed < 750 g, but the risk of CLD of prematurity in the uncolonized control group was already 82%. Because the cohort study design allows for the possibility that one or more additional factors associated with U. urealyticum may be the true cause(s) of CLD of prematurity, there is strong but not definitive evidence that U. urealyticum causes CLD of prematurity.