Association of chronic kidney disease with postoperative outcomes: a national surgical quality improvement program (NSQIP) multi-specialty surgical cohort analysis.

BACKGROUND: Chronic kidney disease (CKD) is associated with higher incidence of major surgery. No studies have evaluated the association between preoperative kidney function and postoperative outcomes across a wide spectrum of procedures. We aimed to evaluate the association between CKD and 30-day postoperative outcomes across surgical specialties. METHODS: We selected adult patients undergoing surgery across eight specialties. The primary study endpoint was major complications, defined as death, unplanned reoperation, cardiac complication, or stroke within 30 days following surgery. Secondary outcomes included Clavien-Dindo high-grade complications, as well as cardiac, pulmonary, infectious, and thromboembolic complications. Multivariable regression was performed to evaluate the association between CKD and 30-day postoperative complications, adjusted for baseline characteristics, surgical specialty, and operative time. RESULTS: In total, 1,912,682 patients were included. The odds of major complications (adjusted odds ratio [aOR] 2.14 [95% confidence interval (CI): 2.07, 2.21]), death (aOR 3.03 [95% CI: 2.88, 3.19]), unplanned reoperation (aOR 1.57 [95% CI: 1.51, 1.64]), cardiac complication (aOR 3.51 [95% CI: 3.25, 3.80]), and stroke (aOR 1.89 [95% CI: 1.64, 2.17]) were greater for patients with CKD stage 5 vs. stage 1. A similar pattern was observed for the secondary endpoints. CONCLUSION: This population-based study demonstrates the negative impact of CKD on operative outcomes across a diverse range of procedures and patients.

INTRAVITREAL ANTI-VASCULAR ENDOTHELIAL GROWTH FACTOR PHARMACEUTICAL UTILIZATION AND FINANCIAL IMPACT IN A VALUE-BASED CARE PROGRAM.

PURPOSE: To investigate trends and the potential impact of the COVID-19 pandemic on the utilization of intravitreal antivascular endothelial growth factor (anti-VEGF) pharmaceuticals in an accountable care organization (ACO). METHODS: We retrospectively analyzed the Centers for Medicare and Medicaid Services beneficiary claims for all patients in the Houston Methodist Coordinated Care ACO registry during the years 2018, 2019, and 2020. RESULTS: Across the 3 years studied, a mean of 708 patients received anti-VEGF injections per year. The percentage of patients who received anti-VEGF injections decreased in each sequential year, with a steeper decline during the COVID-19 pandemic in the year 2020 (decrease by 0.4% from 2019 to 2020, P < 0.001; decrease by 0.2% from 2018 to 2019, P = 0.1453). The percentage of patients receiving bevacizumab of the total number of patients receiving any anti-VEGF treatment decreased (bevacizumab decreased by 6% from 2019 to 2020, P = 0.0174; decreased by 7% from 2018 to 2019, P = 0.0074). The COVID-19 pandemic did not seem to correlate with a change in the distribution of the specific anti-VEGF injection used. CONCLUSION: Despite the lower price which may correlate with value-based care, bevacizumab was the least used anti-VEGF treatment. COVID-19 correlated with a larger decrease in the utilization of all three anti-VEGF drugs.

Association between isolated abnormal coagulation profile on transfusion following major surgery: A NSQIP analysis of individuals without bleeding disorders.

INTRODUCTION: Preoperative coagulation screening for patients without bleeding disorders remains controversial. The combinatorial risk of INR, aPTT, and platelet count (PLT) abnormalities leading to bleeding requiring transfusion is not known in these patients. We examined the association between abnormal coagulation profile and the risk of transfusion following common elective surgery in patients without bleeding disorders. STUDY DESIGN AND METHODS: We utilized the National Surgical Quality Improvement Program (NSQIP) database from 2004 to 2018 to identify patients without a history of bleeding disorders undergoing common 23 major elective procedures across 10 specialties. Multivariable logistic regression was used to assess the association between coagulation profile and bleeding requiring packed red blood cell transfusion intra-/post-operatively. RESULTS: Of the 672,075 patients meeting inclusion criteria, 53.7% presented with normal coagulation profile preoperatively. Overall, 12.2% (n = 82,368) received transfusion. In the setting of normal aPTT/PLT, both Equivocal INR of 1.1-1.5 (aOR 1.41, 95% CI 1.38-1.44) and Abnormal INR of >1.5 (aOR 1.81, 95% CI 1.71-1.93) were significantly associated with an increased risk of transfusion. Equivocal (60-70) and Abnormal (>70) aPTT with normal INR/PLT did not demonstrate a comparable risk of transfusion. We observed a synergistic effect of combinatorial lab abnormalities on the risk of transfusion when both Abnormal INR/aPTT and Low PLT of <100,000 were present (aOR 5.18, 95% CI 3.04-8.84), compared to the effect of Abnormal INR/aPTT and normal/elevated PLT (aOR 1.90, 95% CI 1.48-2.45). DISCUSSION: The preoperative presence of abnormal findings in INR or PLT was significantly associated with the risk of bleeding requiring transfusion during intraoperative and postoperative periods.

Factors associated with endometrial cancer and hyperplasia among middle-aged and older Hispanics.

OBJECTIVE: While disparities in endometrial hyperplasia and endometrial cancer are well documented in Blacks and Whites, limited information exists for Hispanics. The objective is to describe the patient characteristics associated with endometrial hyperplasia symptoms, endometrial hyperplasia with atypia and endometrial cancer, and assess factors contributing to racial/ethnic differences in disease outcomes. METHODS: This single-center, retrospective study included women aged ≥50 years with ≥ two encounters for endometrial hyperplasia symptoms, endometrial hyperplasia with atypia and endometrial cancer between 2012 and 2016. Multivariate logistic regression models evaluated the predictors of endometrial cancer and hyperplasia. RESULTS: We included 19,865 women (4749 endometrial hyperplasia symptoms, 71 endometrial hyperplasias with atypia, 201 endometrial cancers) with mean age of 60.45 years (SD 9.94). The odds of endometrial hyperplasia symptoms were higher in non-Hispanic Blacks (Odds Ratio [OR] 1.56, 95% Confidence Interval [CI] 1.20-1.72), Hispanics (OR 1.35, 95% CI 1.22-1.49), family history of female cancer (OR 1.25, 95% CI 1.12-1.39), hypertension (OR 1.24, 95% CI 1.14-1.35), and birth control use (OR 1.29, 95% CI 1.15-1.43). Odds of endometrial cancer and atypical hyperplasia increased for ages 60-64 (OR 7.95, 95% CI 3.26-19.37; OR 3.66, 95% 1.01-13.22) and being obese (OR 1.61, 95% CI 1.08-2.41; OR: 6.60, 95% CI 2.32-18.83). Odds of endometrial cancer increased with diabetes (OR 1.68, 95% CI 1.22-2.32). CONCLUSION(S): Patients with obesity and diabetes had increased odds of endometrial cancer and hyperplasia with atypia. Further study is needed to understand the exogenous estrogen effect contributing to the increased incidence among Hispanics.

Leveraging Large Language Models for Knowledge-free Weak Supervision in Clinical Natural Language Processing.

The performance of deep learning-based natural language processing systems is based on large amounts of labeled training data which, in the clinical domain, are not easily available or affordable. Weak supervision and in-context learning offer partial solutions to this issue, particularly using large language models (LLMs), but their performance still trails traditional supervised methods with moderate amounts of gold-standard data. In particular, inferencing with LLMs is computationally heavy. We propose an approach leveraging fine-tuning LLMs and weak supervision with virtually no domain knowledge that still achieves consistently dominant performance. Using a prompt-based approach, the LLM is used to generate weakly-labeled data for training a downstream BERT model. The weakly supervised model is then further fine-tuned on small amounts of gold standard data. We evaluate this approach using Llama2 on three different n2c2 datasets. With no more than 10 gold standard notes, our final BERT models weakly supervised by fine-tuned Llama2-13B consistently outperformed out-of-the-box PubMedBERT by 4.7-47.9% in F1 scores. With only 50 gold standard notes, our models achieved close performance to fully fine-tuned systems.

Multimodal prehabilitation and postoperative outcomes in upper abdominal surgery: systematic review and meta-analysis.

The impact of multimodal prehabilitation on postoperative complications in upper abdominal surgeries is understudied. This review analyzes randomized trials on multimodal prehabilitation with patient and hospital outcomes. MEDLINE, Embase, CINAHL, and Cochrane CENTRAL were searched for trials on prehabilitation before elective (non-emergency) abdominal surgery. Two reviewers independently screened studies, extracted data, and assessed study quality. Primary outcomes of interest were postoperative pulmonary complications (PPCs) and all-cause complications; secondary outcomes included hospital and intensive care length of stay. A meta-analysis with random-effect models was performed, and heterogeneity was evaluated with I-square and Cochran's Q test. Dichotomous variables were reported in log-odds ratio and continuous variables were presented as mean difference. Ten studies (total 1503 patients) were included. Odds of developing complications after prehabilitation were significantly lower compared to various control groups (- 0.38 [- 0.75- - 0.004], P = 0.048). Five studies described PPCs, and participants with prehabilitation had decreased odds of PPC (- 0.96 [- 1.38- - 0.54], P < 0.001). Prehabilitation did not significantly reduce length of stay, unless exercise was implemented; with exercise, hospital stay decreased significantly (- 0.91 [- 1.67- - 0.14], P = 0.02). Multimodal prehabilitation may decrease complications in upper abdominal surgery, but not necessarily length of stay; research should address heterogeneity in the literature.

Bloodstream infections after solid organ transplantation: clinical epidemiology and antimicrobial resistance (2016-21).

Background: Solid organ transplant (SOT) recipients are at risk of bloodstream infections (BSIs) with MDR organisms (MDROs). Objectives: To describe the epidemiology of BSI in the year after several types of SOT, as well as the prevalence of MDRO infections in this population. Methods: We conducted a single-centre, retrospective study of kidney, liver, heart, and multi-organ transplantation patients. We examined BSIs ≤1 year from SOT and classified MDRO phenotypes for Staphylococcus aureus, enterococci, Enterobacterales, Pseudomonas aeruginosa and Candida spp. We compared BSI characteristics between SOT types and determined risk factors for 90 day mortality. Results: We included 2293 patients [1251 (54.6%) kidney, 663 (28.9%) liver, 219 (9.6%) heart and 160 (7.0%) multi-organ transplant]. Overall, 8.5% of patients developed a BSI. BSIs were most common after multi-organ (23.1%) and liver (11.3%) transplantation (P < 0.001). Among 196 patients with BSI, 323 unique isolates were recovered, 147 (45.5%) of which were MDROs. MDROs were most common after liver transplant (53.4%). The most frequent MDROs were VRE (69.8% of enterococci) and ESBL-producing and carbapenem-resistant Enterobacterales (29.2% and 27.2% of Enterobacterales, respectively). Mortality after BSI was 9.7%; VRE was independently associated with mortality (adjusted OR 6.0, 95% CI 1.7-21.3). Conclusions: BSI incidence after SOT was 8.5%, with a high proportion of MDROs (45.5%), especially after liver transplantation. These data, in conjunction with local antimicrobial resistance patterns and prescribing practices, may help guide empirical antimicrobial selection and stewardship practices after SOT.

Extraction of Radiological Characteristics From Free-Text Imaging Reports Using Natural Language Processing Among Patients With Ischemic and Hemorrhagic Stroke: Algorithm Development and Validation.

BACKGROUND: Neuroimaging is the gold-standard diagnostic modality for all patients suspected of stroke. However, the unstructured nature of imaging reports remains a major challenge to extracting useful information from electronic health records systems. Despite the increasing adoption of natural language processing (NLP) for radiology reports, information extraction for many stroke imaging features has not been systematically evaluated. OBJECTIVE: In this study, we propose an NLP pipeline, which adopts the state-of-the-art ClinicalBERT model with domain-specific pretraining and task-oriented fine-tuning to extract 13 stroke features from head computed tomography imaging notes. METHODS: We used the model to generate structured data sets with information on the presence or absence of common stroke features for 24,924 patients with strokes. We compared the survival characteristics of patients with and without features of severe stroke (eg, midline shift, perihematomal edema, or mass effect) using the Kaplan-Meier curve and log-rank tests. RESULTS: Pretrained on 82,073 head computed tomography notes with 13.7 million words and fine-tuned on 200 annotated notes, our HeadCT_BERT model achieved an average area under receiver operating characteristic curve of 0.9831, F1-score of 0.8683, and accuracy of 97%. Among patients with acute ischemic stroke, admissions with any severe stroke feature in initial imaging notes were associated with a lower probability of survival (P<.001). CONCLUSIONS: Our proposed NLP pipeline achieved high performance and has the potential to improve medical research and patient safety.

A Neuro-Informatics Pipeline for Cerebrovascular Disease: Research Registry Development.

BACKGROUND: Although stroke is well recognized as a critical disease, treatment options are often limited. Inpatient stroke encounters carry critical information regarding the mechanisms of stroke and patient outcomes; however, these data are typically formatted to support administrative functions instead of research. To support improvements in the care of patients with stroke, a substantive research data platform is needed. OBJECTIVE: To advance a stroke-oriented learning health care system, we sought to establish a comprehensive research repository of stroke data using the Houston Methodist electronic health record (EHR) system. METHODS: Dedicated processes were developed to import EHR data of patients with primary acute ischemic stroke, intracerebral hemorrhage (ICH), transient ischemic attack, and subarachnoid hemorrhage under a review board-approved protocol. Relevant patients were identified from discharge diagnosis codes and assigned registry patient identification numbers. For identified patients, extract, transform, and load processes imported EHR data of primary cerebrovascular disease admissions and available data from any previous or subsequent admissions. Data were loaded into patient-focused SQL objects to enable cross-sectional and longitudinal analyses. Primary data domains (admission details, comorbidities, laboratory data, medications, imaging data, and discharge characteristics) were loaded into separate relational tables unified by patient and encounter identification numbers. Computed tomography, magnetic resonance, and angiography images were retrieved. Imaging data from patients with ICH were assessed for hemorrhage characteristics and cerebral small vessel disease markers. Patient information needed to interface with other local and national databases was retained. Prospective patient outreach was established, with patients contacted via telephone to assess functional outcomes 30, 90, 180, and 365 days after discharge. Dashboards were constructed to provide investigators with data summaries to support access. RESULTS: The Registry of Neurological Endpoint Assessments among Patients with Ischemic and Hemorrhagic Stroke (REINAH) database was constructed as a series of relational category-specific SQL objects. Encounter summaries and dashboards were constructed to draw from these objects, providing visual data summaries for investigators seeking to build studies based on REINAH data. As of June 2022, the database contains 18,061 total patients, including 1809 (10.02%) with ICH, 13,444 (74.43%) with acute ischemic stroke, 1221 (6.76%) with subarachnoid hemorrhage, and 3165 (17.52%) with transient ischemic attack. Depending on the cohort, imaging data from computed tomography are available for 85.83% (1048/1221) to 98.4% (1780/1809) of patients, with magnetic resonance imaging available for 27.85% (340/1221) to 85.54% (11,500/13,444) of patients. Outcome assessment has successfully contacted 56.1% (240/428) of patients after ICH, with 71.3% (171/240) of responders providing consent for assessment. Responders reported a median modified Rankin Scale score of 3 at 90 days after discharge. CONCLUSIONS: A highly curated and clinically focused research platform for stroke data will establish a foundation for future research that may fundamentally improve poststroke patient care and outcomes.

Deep learning-based NLP data pipeline for EHR-scanned document information extraction.

Objective: Scanned documents in electronic health records (EHR) have been a challenge for decades, and are expected to stay in the foreseeable future. Current approaches for processing include image preprocessing, optical character recognition (OCR), and natural language processing (NLP). However, there is limited work evaluating the interaction of image preprocessing methods, NLP models, and document layout. Materials and Methods: We evaluated 2 key indicators for sleep apnea, Apnea hypopnea index (AHI) and oxygen saturation (SaO2), from 955 scanned sleep study reports. Image preprocessing methods include gray-scaling, dilating, eroding, and contrast. OCR was implemented with Tesseract. Seven traditional machine learning models and 3 deep learning models were evaluated. We also evaluated combinations of image preprocessing methods, and 2 deep learning architectures (with and without structured input providing document layout information), with the goal of optimizing end-to-end performance. Results: Our proposed method using ClinicalBERT reached an AUROC of 0.9743 and document accuracy of 94.76% for AHI, and an AUROC of 0.9523 and document accuracy of 91.61% for SaO2. Discussion: There are multiple, inter-related steps to extract meaningful information from scanned reports. While it would be infeasible to experiment with all possible option combinations, we experimented with several of the most critical steps for information extraction, including image processing and NLP. Given that scanned documents will likely be part of healthcare for years to come, it is critical to develop NLP systems to extract key information from this data. Conclusion: We demonstrated the proper use of image preprocessing and document layout could be beneficial to scanned document processing.

Pancreatic head vs pancreatic body/tail cancer: Are they different?

BACKGROUND: The impact of pancreatic tumor location on patient survival has been studied in large national data-based analyses which yielded controversial results. AIM: To explore if pancreatic head cancer (PHC) and pancreatic body/tail cancer (PBTC) have different overall survival (OS), molecular signature and response to chemotherapy. METHODS: We retrospectively queried patient records from July 2016 to June 2020 in our institution. Patient demographics, cancer stage on diagnosis, tumor location, somatic mutations, treatment, and survival are recorded and analyzed. A test is considered statistically significant if the P value was < 0.05. RESULTS: We reviewed 101 patients with complete records, among which 67 (66.34%) were PHC and 34 (33.66%) were PBTC. More PHC were diagnosed at younger age [61.49 vs 68.97, P = 0.010], earlier stages (P = 0.006) and underwent surgical resection (P = 0.025). There were no significant differences among all mutations and pathways studied except for TP53 mutations (37.0% in PHC vs 70.0% in PBTC, P = 0.03). OS was not statistically different between PHC and PBTC (P = 0.636) in the overall population and in subgroups according to surgical resection status or stages. In terms of response to chemotherapy, chemotherapy regimens (FOLFIRINOX-based vs gemcitabine-based) didn't impact disease free interval in those who had surgical resection in either PHC (P = 0.546) or PBTC (P = 0.654), or the duration of response to first line palliative treatment in those with advanced disease in PHC (P = 0.915) or PBTC (P = 0.524). CONCLUSION: Even though PHC and PBTC have similar poor OS and response to chemotherapy, the different presentations and molecular profiles indicate they are different diseases. Utilization of molecular profiling to develop targeted therapy for individualization of treatment is needed.

The Association between Serum Albumin and Post-Operative Outcomes among Patients Undergoing Common Surgical Procedures: An Analysis of a Multi-Specialty Surgical Cohort from the National Surgical Quality Improvement Program (NSQIP).

While studies have demonstrated an association between preoperative hypoalbuminemia and adverse clinical outcomes, the optimal serum albumin threshold for risk-stratification in the broader surgical population remains poorly defined. We sought define the optimal threshold of preoperative serum albumin concentration for risk-stratification of adverse post-operative outcomes. Using the American College of Surgeons National Surgical Quality Improvement Program Database, we identified 842,672 patients that had undergone a common surgical procedure in one of eight surgical specialties. An optimal serum albumin concentration threshold for risk-stratification was determined using receiver-operating characteristic analysis. Multivariable logistic regression analysis was used to evaluate the odds of adverse surgical events; a priori defined subgroup analyses were performed. A serum albumin threshold of 3.4 g/dL optimally predicted adverse surgical outcomes in the broader cohort. After multivariable analysis, patients with hypoalbuminemia had increased odds of death within 30 days of surgery (odds ratio [OR] 2.01, 95% confidence interval [CI] 1.94-2.08). Hypoalbuminemia was associated with greater odds of primary adverse events among patients with disseminated cancer (OR 2.03, 95% CI 1.88-2.20) compared to patients without disseminated cancer (OR 1.47, 95% CI 1.44-1.51). The standard clinical threshold for hypoalbuminemia is the optimal threshold for preoperative risk assessment.

Determining Cause of Death During Coronavirus Disease 2019 Pandemic.

Controversy exists whether the cause of death due to severe acute respiratory syndrome coronavirus 2 is directly related to the infection or to underlying conditions. The purpose of this study is to assess the relationship of severe acute respiratory syndrome coronavirus 2 infection with the cause of death in hospitalized patients. DESIGN: Retrospective observational study; deidentified discharge summaries of deceased patients were reviewed by two intensivists and classified as coronavirus disease 2019-related (caused by severe acute respiratory syndrome coronavirus 2) or coronavirus disease 2019-unrelated (not caused by severe acute respiratory syndrome coronavirus 2 or indeterminate) deaths. For classification disagreement, a separate group of three intensivists reviewed the discharge summaries and arbitrated to determine the cause of death. SETTING: Single-center study performed at the University of Texas Medical Branch. PATIENTS: All adult patients (> 18 yr) admitted from March 10, 2020, to October 22, 2020, with positive severe acute respiratory syndrome coronavirus 2 test results who expired during their hospitalization were identified. INTERVENTIONS: None. MEASUREMENTS AND MAIN RESULTS: Patient demographics, comorbidities, prescribed medications, and ventilatory support data were collected. Comparison between groups was performed using t test and chi-square test. During the study period, 1,052 patients were admitted within 14 days of severe acute respiratory syndrome coronavirus 2-positive test results, of whom 100 expired during the hospitalization. Deceased patients were predominantly male and older than 65 years. Obesity (body mass index ≥ 30 kg/m2) was present in 41%, and common comorbidities included hypertension (47%), diabetes (30%), and heart failure (20%). Death was classified as directly caused by severe acute respiratory syndrome coronavirus 2 in 85% and not caused by severe acute respiratory syndrome coronavirus 2 in 5%. An indeterminate cause of death in 10% was due to insufficient information or an atypical presentation. The observed interrater agreement on the cause of death classification was 81%. CONCLUSIONS: In this single-center study, the majority of deaths in severe acute respiratory syndrome coronavirus 2-positive hospitalized patients were related to a typical or atypical presentation of coronavirus disease 2019 disease.

Trends and Factors Associated with Nebulized Therapy Prescription in Older Adults with Chronic Obstructive Pulmonary Disease from 2008 to 2015.

Background: Medical management of patients with chronic obstructive pulmonary disease (COPD) includes nebulized therapy as an option for inhalational drug delivery. A broad variety of short- and long-acting bronchodilators and inhaled corticosteroids in the nebulized form are available. Despite this, limited information exists on the pattern and predictors of nebulized prescription. We examined the trend and factors associated with prescription of nebulized therapy among Medicare beneficiaries with COPD. Methods: A retrospective cross-sectional study of 5% Medicare beneficiaries with COPD (n = 66,032) who were enrolled in parts A, B, and D and received nebulized prescription from 2008 to 2015 was conducted. This sample has shown to be representative of the entire fee-for-service Medicare population. The primary outcome was a prescription of nebulized medications. Reliever nebulized medications included short-acting beta agonist (SABA), short-acting muscarinic agents (SAMAs), and a combination of SABA and SAMA, while maintenance nebulized medications included long-acting beta agonists, long-acting muscarinic agents, and corticosteroid solutions as well as combinations of these agents. The secondary outcome was prescription of other inhaler respiratory medications not administered with a nebulizer. Results: Overall, 38.9% patients were prescribed nebulized medication and their prescription significantly declined from 42.4% in 2008 to 35.1% in 2015, majority of which was related to decreased prescriptions of nebulized relievers. Factors associated with the prescription of nebulized medications include female gender (odds ratio [OR] = 1.06; 95% confidence interval [CI] = 1.02-1.09), dual eligibility or low-income subsidy beneficiaries (OR = 1.49; CI = 1.44-1.53), hospitalization for COPD in the previous year (OR = 1.29; CI = 1.25-1.34), home oxygen therapy (OR = 2.29; CI = 2.23-2.36), pulmonary specialist visit (OR = 1.24; CI = 1.20-1.27), and moderate (OR = 1.61; CI = 1.57-1.65) or high (OR = 1.52; CI = 1.46-1.59) severity of COPD. Conclusion: Between 2008 and 2015, prescriptions for nebulized therapy for COPD declined among Medicare beneficiaries, probably related to increase in use of maintenance non-nebulized medications.

School-based human papillomavirus vaccination program for increasing vaccine uptake in an underserved area in Texas.

OBJECTIVE: Compare the effectiveness of community-based HPV-related education and onsite school-based vaccination versus community-based education only for increasing HPV vaccine uptake in a rural, medically underserved area. METHODS: Our cohort included 2307 Rio Grande City Consolidated Independent School District (RGCISD) middle school students from 3 schools enrolled in August 2016 and followed until April 2018. Using a quasi-experimental design, this study implemented an onsite school-based vaccination program and physician-led education on HPV and HPV vaccines for parents/guardians, school nurses/staff, and pediatric/family providers in the surrounding community (15-mile radius of RGCCISD) at 1 middle school ("intervention school"), and education-only for the remaining 2 schools ("comparison schools"). The Centers for Disease Control and Prevention's HPV-related educational materials supplemented the education. HPV vaccine status was obtained from school immunization records and the project's contracted vaccine vendor. HPV vaccine initiation and completion rates were compared pre and post intervention and between the intervention and comparison schools. Logistic regression was used to compare the odds of newly initiating/completing vaccination between the intervention and comparison schools. RESULTS: At baseline, the intervention school had lower HPV vaccine initiation and completion rates than the comparison schools (20.00% and 8.70% vs 28.97% and 14.56%). Post intervention, the intervention school had higher initiation and completion rates than the comparison schools (53.67% and 28.36% vs 41.56% and 20.53%). Students from the intervention school were over 3.6-times more likely to newly initiate/complete the HPV vaccinations than students from the comparison schools. CONCLUSION: The school with on-site vaccination events and community-based education had a higher adolescent HPV vaccination rate compared to schools that received community-based education only.