Search

Considerable variability exists between surgeons' assessments of a patient's individual pre-operative surgical risk. Surgical risk calculators are not routinely used despite their validation. We sought to compare thoracic surgeons' prediction of patients' risk of postoperative adverse outcomes versus a surgical risk calculator, the Surgical Risk Preoperative Assessment System (SURPAS). We developed vignettes from 30 randomly selected patients who underwent thoracic surgery in the American College of Surgeons' National Surgical Quality Improvement Program (NSQIP) database. Twelve thoracic surgeons estimated patients' preoperative risks of postoperative morbidity and mortality. These were compared to SURPAS estimates of the same vignettes. C-indices and Brier scores were calculated for the surgeons' and SURPAS estimates. Agreement between surgeon estimates was examined using intraclass correlation coefficients (ICCs). Surgeons estimated higher morbidity risk compared to SURPAS for low-risk patients (ASA classes 1-2, 11.5% vs. 5.1%, p=

Background The universal Surgical Risk Preoperative Assessment System (SURPAS) prediction models for postoperative adverse outcomes have good accuracy for estimating risk in broad surgical populations and for surgical specialties. The accuracy in individual operations has not yet been assessed. The objective of this study was to evaluate the Surgical Risk Preoperative Assessment System in predicting adverse outcomes for selected individual operations. Methods The SURPAS models were applied to the top 2 most frequent common procedural terminology codes in 9 surgical specialties and 5 additional common general surgical operations in the 2009 to 2018 database of the American College of Surgeons National Surgical Quality Improvement Program (ACS NSQIP). Goodness of fit statistics were estimated, including c-indices for discrimination, Hosmer-Lemeshow graphs and P values for calibration, overall observed versus expected event rates, and Brier scores. Results The total sample size was 2,020,172, which represented 29% of the 6.9 million operations in the ACS NSQIP database. Average c-indices across 12 outcomes were acceptable (≥0.70) for 13 (56.5%) of the 23 operations. Overall observed-to-expected rates were similar for mortality and overall morbidity across the 23 operations. Hosmer-Lemeshow graphs over quintiles of risk comparing observed-to-expected rates of mortality and overall morbidity were similar for 52% and 70% of operations, respectively. Model performance was better in less complex operations and those done in patients with lower preoperative risk. Conclusion SURPAS displayed accuracy in estimating postoperative adverse events for some of the 23 operations studied, but not all. In the procedures where SURPAS was not accurate, developing disease or operation-specific risk models might be appropriate.

Background Residents report that faculty preference is a significant driver of opioid prescribing practices. This study compared opioid prescribing preferences of surgical residents and faculty against published guidelines and actual practice and assessed perceptions in communication and transparency around these practices. Methods Surgical residents and faculty were surveyed to evaluate the number of oxycodone tablets prescribed for common procedures. Quantities were compared between residents, faculty, Opioid Prescribing Engagement Network guidelines, and actual opioids prescribed. Frequency with which faculty communicate prescribing preferences and the desire for feedback and transparency in prescription practices were assessed. Results Fifty-six (72%) residents and 57 (59%) faculty completed the survey. Overall, faculty preferred a median number of tablets greater than recommended by Opioid Prescribing Engagement Network in 5 procedures, while residents did so in 9 of 14 procedures. On average, across all operations, faculty reported prescribing practices compliant with Opioid Prescribing Engagement Network 56.1% of the time, whereas residents did so 47.6% of the time (P = .40). Interestingly, opioids actually prescribed were significantly less than recommended in 7 procedures. Among faculty, 62% reported often or always specifying prescription preferences to residents, while only 9% of residents noted that faculty often did so. Residents (80%) and faculty (75%) were amenable to seeing regular reports of personal opioid prescription practices, and 74% and 65% were amenable to seeing practices compared with peers. Only 34% of residents and 44% of faculty wanted prescription practices made public. Conclusion There is a disconnect between opioid prescribing preferences and practice among surgical residents and faculty. Increased transparency through individualized reports and education regarding Opioid Prescribing Engagement Network guidelines with incorporation into the electronic medical record as practice advisories may reduce prescription variability.

Background The Surgical Risk Preoperative Assessment System (SURPAS) uses eight variables to accurately predict postoperative complications but has not been sufficiently studied in emergency surgery. We evaluated SURPAS in emergency surgery, comparing it to the Emergency Surgery Score (ESS). Methods SURPAS and ESS estimates of 30-day mortality and overall morbidity were calculated for emergency operations in the 2009–2018 ACS-NSQIP database and compared using observed-to-expected plots and rates, c-indices, and Brier scores. Cases with incomplete data were excluded. Results In 205,318 emergency patients, SURPAS underestimated (8.1%; 35.9%) while ESS overestimated (10.1%; 43.8%) observed mortality and morbidity (8.9%; 38.8%). Each showed good calibration on observed-to-expected plots. SURPAS had better c-indices (0.855 vs 0.848 mortality; 0.802 vs 0.755 morbidity), while the Brier score was better for ESS for mortality (0.0666 vs. 0.0684) and for SURPAS for morbidity (0.1772 vs. 0.1950). Conclusions SURPAS accurately predicted mortality and morbidity in emergency surgery using eight predictor variables.

The purpose of this study was to develop and test personalized predictions for functional recovery after Total Knee Arthroplasty (TKA) surgery, using a novel neighbors-based prediction approach. We used data from 397 patients with TKA to develop the prediction methodology and then tested the predictions in a temporally distinct sample of 202 patients. The Timed Up and Go (TUG) Test was used to assess physical function. Neighbors-based predictions were generated by estimating an index patient’s prognosis from the observed recovery data of previous similar patients (a.k.a., the index patient’s “matches”). Matches were determined by an adaptation of predictive mean matching. Matching characteristics included preoperative TUG time, age, sex and Body Mass Index. The optimal number of matches was determined to be m = 35, based on low bias (− 0.005 standard deviations), accurate coverage (50% of the realized observations within the 50% prediction interval), and acceptable precision (the average width of the 50% prediction interval was 2.33 s). Predictions were well-calibrated in out-of-sample testing. These predictions have the potential to inform care decisions both prior to and following TKA surgery.

This study was designed to: (1) characterize longitudinal patient-reported outcomes (PROs) between breast cancer patients undergoing lumpectomy and mastectomy and (2) compare return to baseline scores at 3 months and 6 months postoperatively. Newly diagnosed breast cancer patients seen at an academic breast center between June 2019 and February 2021 were invited to participate in longitudinal PRO surveys at their initial clinic visit. If willing to participate, patients were emailed the validated BREAST-Q™ questionnaire at the initial clinic visit (baseline), 2 weeks after surgery, and then every 3 months for the first year. We used linear mixed models to estimate the differences in slopes over time between lumpectomy and mastectomy for each PRO measure. Pearson’s Chi-square tests with Yates’ continuity correction were used to compare proportions of patients who return to baseline PRO scores. P < 0.05 was considered significant. Of 164 patients invited to participate, 100 (61%) completed a baseline survey and were included in analyses. Mastectomy patients had significantly greater decreases in breast satisfaction (P = 0.002), psychosocial well-being (P < 0.0001), and sexual well-being (P < 0.0001) over time compared with lumpectomy patients. Both surgical groups reported a decrease in physical well-being, although the decline was more significant in lumpectomy patients (P = 0.005). At 3 months and 6 months after surgery, significantly larger proportions of lumpectomy patients returned to their baseline breast satisfaction, psychosocial well-being, and physical well-being compared with mastectomy patients. Understanding how outcomes important to patients change over the care continuum can provide opportunities for early intervention and may prevent debilitating long-term morbidities of treatment.

Background: Cluster-randomized trials allow for the evaluation of a community-level or group-/cluster-level intervention. For studies that require a cluster-randomized trial design to evaluate cluster-level interventions aimed at controlling vector-borne diseases, it may be difficult to assess a large number of clusters while performing the additional work needed to monitor participants, vectors, and environmental factors associated with the disease. One such example of a cluster-randomized trial with few clusters was the “efficacy and risk of harms of repeated ivermectin mass drug administrations for control of malaria” trial. Although previous work has provided recommendations for analyzing trials like repeated ivermectin mass drug administrations for control of malaria, additional evaluation of the multiple approaches for analysis is needed for study designs with count outcomes. Methods: Using a simulation study, we applied three analysis frameworks to three cluster-randomized trial designs (single-year, 2-year parallel, and 2-year crossover) in the context of a 2-year parallel follow-up of repeated ivermectin mass drug administrations for control of malaria. Mixed-effects models, generalized estimating equations, and cluster-level analyses were evaluated. Additional 2-year parallel designs with different numbers of clusters and different cluster correlations were also explored. Results: Mixed-effects models with a small sample correction and unweighted cluster-level summaries yielded both high power and control of the Type I error rate. Generalized estimating equation approaches that utilized small sample corrections controlled the Type I error rate but did not confer greater power when compared to a mixed model approach with small sample correction. The crossover design generally yielded higher power relative to the parallel equivalent. Differences in power between analysis methods became less pronounced as the number of clusters increased. The strength of within-cluster correlation impacted the relative differences in power. Conclusion: Regardless of study design, cluster-level analyses as well as individual-level analyses like mixed-effects models or generalized estimating equations with small sample size corrections can both provide reliable results in small cluster settings. For 2-year parallel follow-up of repeated ivermectin mass drug administrations for control of malaria, we recommend a mixed-effects model with a pseudo-likelihood approximation method and Kenward–Roger correction. Similarly designed studies with small sample sizes and count outcomes should consider adjustments for small sample sizes when using a mixed-effects model or generalized estimating equation for analysis. Although the 2-year parallel follow-up of repeated ivermectin mass drug administrations for control of malaria is already underway as a parallel trial, applying the simulation parameters to a crossover design yielded improved power, suggesting that crossover designs may be valuable in settings where the number of available clusters is limited. Finally, the sensitivity of the analysis approach to the strength of within-cluster correlation should be carefully considered when selecting the primary analysis for a cluster-randomized trial.

Background Use of neoadjuvant therapy for elderly patients with pancreatic cancer has been debatable. With FOLFIRINOX (folinic acid, 5-fluorouracil, irinotecan, oxaliplatin) or gemcitabine plus nab-paclitaxel (GnP) showing tremendous effects in improving the overall survival of patients with borderline resectable and locally advanced pancreatic cancer, there is no definitive consensus regarding the use of this regimen in the elderly. Methods This study evaluated the eligibility of elderly patients with borderline resectable or locally advanced pancreatic cancer for neoadjuvant therapy. Patients registered in the database of pancreatic cancer at the University of Colorado Cancer Center, who underwent neoadjuvant treatment between January 2011 and March 2019, were separated into three age groups (less than 70, 70–74, 75 or more years) and respective treatment outcomes were compared. Results The study included 246 patients with pancreatic cancer who underwent neoadjuvant treatment, of whom 154 and 71 received chemotherapy with FOLFIRINOX and GnP respectively. Among these 225 patients, 155 were younger than 70 years, 36 were aged 70–74 years, and 34 were aged 75 years or older. Patients under 70 years old received FOLFIRINOX most frequently (124 of 155 versus 18 of 36 aged 70–74 years, and 12 of 34 aged 75 years or more; P Conclusion The safety and efficacy of multiagent chemotherapy in patients aged 75 years or over were similar to those in younger patients. Modern multiagent regimens could be a safe and viable treatment option for clinically fit patients aged at least 75 years.

Background Postoperative complications, length of index hospital stay, and unplanned hospital readmissions are important metrics reflecting surgical care quality. Postoperative infections represent a substantial proportion of all postoperative complications. We examined the relationships between identification of postoperative infection prehospital and posthospital discharge, length of stay, and unplanned readmissions in the American College of Surgeons National Surgical Quality Improvement Program database across nine surgical specialties. Methods The 30-day postoperative infectious complications including sepsis, surgical site infections, pneumonia, and urinary tract infection were analyzed in the American College of Surgeons National Surgical Quality Improvement Program inpatient data during the period from 2012 to 2017. General, gynecologic, vascular, orthopedic, otolaryngology, plastic, thoracic, urologic, and neurosurgical inpatient operations were selected. Results Postoperative infectious complications were identified in 5.2% (137,014/2,620,450) of cases; 81,929 (59.8%) were postdischarge. The percentage of specific complications identified postdischarge were 73.4% of surgical site infections (range across specialties 63.7–93.1%); 34.9% of sepsis cases (27.4–58.1%); 26.5% of pneumonia cases (18.9%–36.3%); and 53.2% of urinary tract infections (48.3%–88.0%). The relative risk of readmission among patients with postdischarge versus predischarge surgical site infection, sepsis, pneumonia, or urinary tract infection was 5.13 (95% confidence interval: 4.90–5.37), 9.63 (8.93–10.40), 10.79 (10.15–11.45), and 3.32 (3.07–3.60), respectively. Over time, mean length of stay decreased but postdischarge infections and readmission rates significantly increased. Conclusion Most postoperative infectious complications were diagnosed postdischarge. These were associated with an increased risk of readmission. The trend toward shorter length of stay over time was observed along with an increase both in the percentage of infections detected after discharge and the rate of unplanned related postoperative readmissions over time. Postoperative surveillance of infections should extend beyond hospital discharge of surgical patients.

The severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) pandemic necessitated rapid local public health response, but studies examining the impact of social distancing policies on SARS-CoV-2 transmission have struggled to capture regional-level dynamics. We developed a susceptible-exposed-infected-recovered transmission model, parameterized to Colorado, USA‒specific data, to estimate the impact of coronavirus disease‒related policy measures on mobility and SARS-CoV-2 transmission in real time. During March‒June 2020, we estimated unknown parameter values and generated scenario-based projections of future clinical care needs. Early coronavirus disease policy measures, including a stay-at-home order, were accompanied by substantial decreases in mobility and reduced the effective reproductive number well below 1. When some restrictions were eased in late April, mobility increased to near baseline levels, but transmission remained low (effective reproductive number

Objective To assess the contribution of unknown institutional factors (contextual effects) in the de-implementation of completion axillary node dissection (cALND) in women with breast cancer. Summary background data Women included in the National Cancer Database with invasive breast carcinoma from 2012-2016 that underwent upfront lumpectomy and were found to have a positive sentinel node. Methods A multivariable mixed effects logistic regression model with a random intercept for site was used to determine the effect of patient, tumor, and institutional variables on the risk of cALND. Reference Effect Measures (REM) were used to describe and compare the contribution of contextual effects to the variation in cALND use to that of measured variables. Results By 2016, cALND was still performed in at least 50% of the patients in a quarter of the institutions. Black race, younger women and those with larger or hormone negative tumors were more likely to undergo cALND. However, the width of the 90% REM range for the contextual effects exceeded that of the measured site, tumor, time, and patient demographics, suggesting institutional contextual effects were the major drivers of cALND de-implementation. For instance, a woman at an institution with low-risk of performing cALND would have 74% reduced odds of having a cALND than if she was treated at a median-risk institution, while a patient at a high-risk institution had 3.91 times the odds. Conclusion Compared to known patient, tumor and institutional factors, contextual effects had a higher contribution to the variation in cALND use.

The Screening and Linkage to Services for Autism pragmatic trial examined the effect of family navigation on referral and evaluation for autism spectrum disorder in an urban healthcare system serving low-income families. Children aged 16–30 months with initial scores ⩾3 on the Modified Checklist for Autism in Toddlers–Revised with Follow-up (M-CHAT-R/F) were randomized to usual care ( n = 133) or navigation if indicated ( n = 142). Bilingual, bicultural autism family navigators offered navigation to intervention-group families if the completed M-CHAT-R/F indicated the need for autism spectrum disorder diagnostic evaluation. Autism family navigators also assisted families to arrange for completion of the M-CHAT-R/F interview if not done, and then offered navigation if indicated. Of 49 families for whom navigation was indicated, 45% received navigation, including care coordination, education, and practical support. In intention-to-treat analyses, the intervention did not increase referrals for evaluation but more than doubled completed autism spectrum disorder diagnostic evaluations. Intervention children were also more likely to complete the M-CHAT-R/F interview with fidelity. Research is needed to identify effective methods to improve uptake of family navigation. Family navigation may be an important tool for reducing systemic inequalities in the early identification and treatment of young children with autism spectrum disorder ( ClinicalTrials.gov , https://clinicaltrials.gov/ ; NCT02374541).Lay abstractYoung children with autism spectrum disorder can benefit from early detection and intervention to moderate autism spectrum disorder’s effect on their development. Routine screening with standardized instruments facilitates early identification. However, many screen-positive children do not receive referrals for evaluation, complete diagnostic evaluations, or engage in services. This is especially true for children of color and those from less educated or low-income families. The Screening and Linkage to Services for Autism trial evaluated the effect of family navigation on referral and evaluation for autism spectrum disorder in screen-positive toddlers. The trial was conducted in an urban, low-income population. It was designed to test “real-world” implementation of family navigation. We randomly assigned 275 children aged 16–30 months with an initial positive autism spectrum disorder screen to receive navigation if indicated or usual care. Navigation increased the number of at-risk toddlers who were evaluated for autism spectrum disorder. It also improved the quality of autism spectrum disorder screening in primary care clinics. However, many children failed to receive indicated referrals for evaluation. Navigation did not improve referral rates. Furthermore, nearly half of families who were potentially eligible for navigation could not be contacted. Other ways of structuring the navigation program may be necessary to improve families’ access to and acceptance of navigation, especially in the setting of financial, cultural, and linguistic barriers. Our trial nevertheless showed that family navigation may be an important tool for reducing systemic inequalities in the early identification and treatment of young children with autism spectrum disorder.

Objectives To use electronic health record data from the first 2 hours of care to derive and validate a model to predict hypotensive septic shock in children with infection. Design Derivation-validation study using an existing registry. Setting Six emergency care sites within a regional pediatric healthcare system. Three datasets of unique visits were designated:1)training set (five sites, April 1, 2013, to December 31, 2016),2)temporal test set (five sites, January 1, 2017, to June 30, 2018), and3)geographic test set (sixth site, April 1, 2013, to June 30, 2018). Patients Patients in whom clinicians were concerned about serious infection from 60 days to 18 years were included; those with septic shock in the first 2 hours were excluded. There were 2,318 included visits; 197 developed septic shock (8.5%). Interventions Lasso with 10-fold cross-validation was used for variable selection; logistic regression was then used to construct a model from those variables in the training set. Variables were derived from electronic health record data known in the first 2 hours, including vital signs, medical history, demographics, and laboratory information. Test characteristics at two thresholds were evaluated: 1) optimizing sensitivity and specificity and 2) set to 90% sensitivity. Measurements and main results Septic shock was defined as systolic hypotension and vasoactive use or greater than or equal to 30 mL/kg isotonic crystalloid administration in the first 24 hours. A model was created using 20 predictors, with an area under the receiver operating curve in the training set of 0.85 (0.82-0.88); 0.83 (0.78-0.89) in the temporal test set and 0.83 (0.60-1.00) in the geographic test set. Sensitivity and specificity varied based on cutpoint; when sensitivity in the training set was set to 90% (83-94%), specificity was 62% (60-65%). Conclusions This model predicted risk of septic shock in children with suspected infection 2 hours after arrival, a critical timepoint for emergent treatment and transfer decisions. Varied cutpoints could be used to customize sensitivity to clinical context.

We evaluated whether Denver neighborhoods with elevated rates of adult hospitalizations for laboratory-confirmed influenza had lower adult coverage with influenza vaccine. Overall vaccine coverage was low. Hospitalization rates were associated with demographic and socioeconomic characteristics. Active immunization of at-risk neighborhoods may be necessary to address disparities in influenza hospitalization rates.

Background: Formal surgical risk assessment tools have been developed to predict risk of adverse postoperative patient outcomes. Such tools accurately predict common postoperative complications, inform patients and providers of likely perioperative outcomes, guide decision making, and improve patient care. However, these are underutilized. We studied the attitudes towards and techniques of how surgeons preoperatively assess risk.Methods: 108 surgical faculty, 95 surgical residents, and 58 affiliate surgeons at a large academic tertiary referral hospital and affiliate community hospitals were emailed a 16-question survey between 8/2019-6/2020. Surveys were analyzed using descriptive statistics (frequency distributions and percentages for categorical variables, means, and standard deviations for continuous variables), and Fisher’s exact test and unpaired t-tests comparing responses by surgical attendings vs. residents.Results: Overall response rates were 50.0% for faculty surgeons, 47.4% for residents, and 36.2% for affiliate surgeons. Only 20.8% of surgeons used risk calculators most or all of the time. Attending surgeons were more likely to use prior experience and current literature while residents used risk calculators more frequently. Risk assessment tools were more likely to be used when predicting major complications and death in older patients with significant risk factors. Greatest barriers for use of risk assessment tools included time, inaccessibility, and trust in accuracy.Conclusions: Even though risk assessment tools are available, few surgeons use them as part of their routine practices. Developers of the tools need to design the tools to better promote the positive features and to avoid the burdens commonly reported by users.

Initial treatment with either neoadjuvant chemoradiation (CRT) or induction FOLFOX (5-Fluorouracil, leucovorin, and oxaliplatin) chemotherapy followed by CRT is considered standard treatment for locally advanced rectal cancer. We compared patient-reported outcomes (PRO) during CRT in patients who had received induction chemotherapy versus those who did not.We reviewed records of patients with locally advanced rectal cancer who were treated with CRT between September 2009 and October 2014, and who had completed ≥ 4 PRO assessments during treatment. Clinician- and patient-reported toxicities were collected each week during treatment. We fit binomial generalized linear models to maximum toxicity scores across all patients' visits.Of 123 patients with ≥ 4 PRO assessments, 87 (71%) patients reported a clinically meaningful PRO score of 3 or higher for diarrhea, and 91 (74%) patients reported a PRO score of ≥ 3 for urgency, during 1 or more weeks of treatment, corresponding to 'very frequent' or worse. Of 116 patients who had also completed ≥ 4 clinician-reported assessments for descriptive analysis, clinically significant diarrhea (Common Terminology Criteria for Adverse Events grade ≥ 2) was reported in 9% of patients, and grade 2 proctitis and cystitis were reported in 20% and 4%, respectively. Eighty-four (68%) patients had undergone induction chemotherapy prior to CRT. Patients who received induction chemotherapy had 68% lower odds of experiencing significant urgency (odds ratio [OR], 0.32; 95% confidence interval [CI], 0.11-0.95; P = .04), 76% lower odds of bleeding (OR, 0.24; 95% CI, 0.1-0.62; P .01), and 75% lower odds of tenesmus (OR, 0.25; 95% CI, 0.11-0.6; P .01) versus those treated with upfront CRT.Based on PROs, a high proportion of patients experienced clinically significant symptoms during pelvic CRT, with diarrhea and urgency being most commonly reported. This appears to be under-reported on clinician-reported assessments. Delivery of induction chemotherapy was associated with lower odds of experiencing urgency, bleeding, and tenesmus on PROs during subsequent CRT, with no significant impact on diarrhea and rectal pain.

Background The Epstein-Barr virus (EBV) viral glycoprotein gp350 has been proposed as a candidate antigen for an EBV vaccine. However, the proposed formulations of these vaccines have not taken into account the presence of 2 unique EBV strains (EBV-1 and EBV-2) present in areas of high incidence of the EBV-associated cancer, Burkitt lymphoma. Methods In this study, we analyze the kinetics of EBV-1 and EBV-2 infection in an asymptomatic infant cohort from Kisumu, Kenya. We also analyzed the kinetics of the antibody response against 5 EBV antigens, gp350 (IgG and IgA), VCA (IgG), EBNA-1 (IgG), EAd (IgG), and Zta (IgG). Results We observed a high frequency of coinfection with both EBV types over time, with the only observable defect in the antibody response in infants coinfected being a significantly lower level of anti-gp350 IgA at peak response. Gp350 IgA levels were also significantly lower in coinfected infants 2.5 months postinfection and at the time of coinfection. Conclusions These results suggest that anti-gp350 IgA antibodies may be important for sterilizing immunity against secondary infection. These findings have implications for the development of an efficacious EBV vaccine to prevent both EBV-1 and EBV-2 infection in a population at high risk for Burkitt lymphoma.

Background Malaria is one of the greatest recurring threats to public health in Mozambique with approximately 10 million cases and thousands of deaths reported annually. Although a malaria early warning system is being established in the country, it is focused on short-term (4–8 week) prediction windows. Increased understanding of the links between quasi-predictable interannual climate variability and malaria could lengthen the lead time for the malaria early warning system and enhance planning and actionable mitigation efforts by public health officials in the country. Methods To identify patterns of interannual variability, we did two principal component analyses of processed weekly district-level malaria incidences for the period 2010–17 in Mozambique and southern Africa regional precipitation for the period 1981–2019. We also did linear regression analyses of sea surface temperatures onto a precipitation index, and composites of various climate variables to establish links between precipitation and modes of climate variability. Findings Two dominant spatiotemporal patterns collectively account for 81% of the interannual variability of malaria in Mozambique. These patterns consisted of a primary hotspot over the central and southern part of the country and a secondary hotspot over the northern third of the country. We found these patterns to be closely related to precipitation variability driven by climate phenomena: the El Nino-Southern Oscillation and the Subtropical Indian Ocean Dipole. Interpretation Quasi-predictable global and regional climate phenomena influence regional precipitation over southern Africa. Knowledge of these associations can inform an enhancement of the malaria early warning system in Mozambique by lengthening early warning lead times. Similar analyses are possible for other countries whose climate is dominated by tropical modes of climate variability. Funding None.

Background Defining a “high risk” surgical population remains challenging. Using the Surgical Risk Preoperative Assessment System (SURPAS), we sought to define “high risk” groups for adverse postoperative outcomes. Materials and Methods We retrospectively analyzed the 2009-2018 American College of Surgeons National Surgical Quality Improvement Program database. SURPAS calculated probabilities of 12 postoperative adverse events. The Hosmer Lemeshow graphs of deciles of risk and maximum Youden index were compared to define “high risk.” Results Hosmer-Lemeshow plots suggested the “high risk” patient could be defined by the 10th decile of risk. Maximum Youden index found lower cutoff points for defining “high risk” patients and included more patients with events. This resulted in more patients classified as “high risk” and higher number needed to treat to prevent one complication. Some specialties (thoracic, vascular, general) had more “high risk” patients, while others (otolaryngology, plastic) had lower proportions. Conclusions SURPAS can define the “high risk” surgical population that may benefit from risk-mitigating interventions.

BACKGROUNDPrognostic biomarkers are needed to identify patients at high-risk for severe COVID-19. Galectin-3 is known to drive neutrophil infiltration and release of pro-inflammatory cytokines contributing to airway inflammation.METHODSIn this prospective cohort, we assessed galectin-3 levels in 156 hospitalized patients with confirmed COVID-19. COVID-19 patients were diagnosed as eithercritical(>50% lung damage) ormoderate(severe outcome, and anon-severe outcomeif they were discharged and none of the former occurred.RESULTSElevated serum galectin-3 was significantly higher in critical patients compared to moderate ones (35.91 ± 19.37 ng/mL vs. 25 ± 14.85 ng/mL, psevere outcomeincluding IMV and/or in-hospital death, presented higher galectin-3 levels (41.17 ng/mL [IQR 29.71 – 52.25] vs. 23.76 ng/mL [IQR 15.78 – 33.97] compared to those of a non-severe outcome, pCONCLUSIONCirculating galectin-3 levels can be used to predict severe outcomes in COVID-19 patients, including the requirement of mechanical ventilation and/or death, regardless of the initial severity of the disease.

Malaria is among the greatest public health threats in Mozambique, with over 10 million cases reported annually since 2018. Although the relationship between seasonal trends in environmental parameters and malaria cases is well established, the role of climate in deviations from the annual cycle is less clear. To investigate this and the potential for leveraging inter‐annual climate variability to predict malaria outbreaks, weekly district‐level malaria incidence spanning 2010–2017 were processed for a cross‐analysis with climate data. An empirical orthogonal function analysis of district‐level malaria incidence revealed two dominant spatiotemporal modes that collectively account for 81% of the inter‐annual variability of malaria: a mode dominated by variance over the southern half of Mozambique (64%), and another dominated by variance in the northern third of the country (17%). These modes of malaria variability are shown to be closely related to precipitation. Linear regression of global sea surface temperatures onto local precipitation indices over these variance maxima links the leading mode of inter‐annual malarial variability to the El Niño‐Southern Oscillation, such that La Niña leads to wetter conditions over southern Mozambique and, therefore, higher malaria prevalence. Similar analysis of spatiotemporal patterns of precipitation over a longer time period (1979–2019) indicate that the Subtropical Indian Ocean Dipole is both a strong predictor of regional precipitation and the climatic mechanism underlying the second mode of malarial variability. These results suggest that skillful malaria early warning systems may be developed that leverage quasi‐predictable modes of inter‐annual climate variability in the tropical oceans.

Background The South African Triage Scale (SATS) is a validated in-hospital triage tool that has been innovatively adopted for use in the prehospital setting by Western Cape Government (WCG) Emergency Medical Services (EMS) in South Africa. The performance of SATS by EMS providers has not been formally assessed. The study sought to assess the validity and reliability of SATS when used by WCG EMS prehospital providers for single-patient triage. Methods This is a prospective, assessment-based validation study among WCG EMS providers from March to September 2017 in Cape Town, South Africa. Participants completed an assessment containing 50 clinical vignettes by calculating the three components — triage early warning score (TEWS), discriminators (pre-defined clinical conditions), and a final SATS triage color. Responses were scored against gold standard answers. Validity was assessed by calculating over- and under-triage rates compared to gold standard. Inter-rater reliability was assessed by calculating agreement among EMS providers’ responses. Results A total of 102 EMS providers completed the assessment. The final SATS triage color was accurately determined in 56.5%, under-triaged in 29.5%, and over-triaged in 13.1% of vignette responses. TEWS was calculated correctly in 42.6% of vignettes, under-calculated in 45.0% and over-calculated in 10.9%. Discriminators were correctly identified in only 58.8% of vignettes. There was substantial inter-rater and gold standard agreement for both the TEWS component and final SATS color, but there was lower inter-rater agreement for clinical discriminators. Conclusion This is the first assessment of SATS as used by EMS providers for prehospital triage. We found that SATS generally under-performed as a triage tool, mainly due to the clinical discriminators. We found good inter-rater reliability, but poor validity. The under-triage rate of 30% was higher than previous reports from the in-hospital setting. The over-triage rate of 13% was acceptable. Further clinically-based and qualitative studies are needed. Trial registration Not applicable.

Background With cardiovascular disease increasing, substantial research has focused on the development of prediction tools. We compare deep learning and machine learning models to a baseline logistic regression using only ‘known’ risk factors in predicting incident myocardial infarction (MI) from harmonized EHR data. Methods Large-scale case-control study with outcome of 6-month incident MI, conducted using the top 800, from an initial 52 k procedures, diagnoses, and medications within the UCHealth system, harmonized to the Observational Medical Outcomes Partnership common data model, performed on 2.27 million patients. We compared several over- and under- sampling techniques to address the imbalance in the dataset. We compared regularized logistics regression, random forest, boosted gradient machines, and shallow and deep neural networks. A baseline model for comparison was a logistic regression using a limited set of ‘known’ risk factors for MI. Hyper-parameters were identified using 10-fold cross-validation. Results Twenty thousand Five hundred and ninety-one patients were diagnosed with MI compared with 2.25 million who did not. A deep neural network with random undersampling provided superior classification compared with other methods. However, the benefit of the deep neural network was only moderate, showing an F1 Score of 0.092 and AUC of 0.835, compared to a logistic regression model using only ‘known’ risk factors. Calibration for all models was poor despite adequate discrimination, due to overfitting from low frequency of the event of interest. Conclusions Our study suggests that DNN may not offer substantial benefit when trained on harmonized data, compared to traditional methods using established risk factors for MI.

Little is known about the relation between socioeconomic factors and health outcomes in adults and adolescents with congenital heart defects (CHD). Population-level data from the Colorado CHD surveillance system from 2011 to 2013 was used to examine the association between area deprivation and outcomes including hospitalizations, emergency department visits, cardiac procedures, all-cause and cardiac-related mortality, and major adverse cardiac events. Socioeconomic context was measured by the Area Deprivation Index at census tract level. Missing race/ethnicity was imputed using the Bayesian Improved Surname Geocoding algorithm. Generalized linear models were utilized to examine health disparities across deprivation quintiles after adjusting for insurance type, race/ethnicity, age, gender, urbanicity, and CHD severity in 5,748 patients. Cases residing in the most deprived quintile had 51% higher odds of inpatient admission, 74% higher odds of emergency department visit, 41% higher odds of cardiac surgeries, and 45% higher odds of major adverse cardiac events compared with cases in the least deprived quintile. Further, rates of hospitalizations, emergency department admissions, and cardiac surgeries were elevated in the most deprived compared with the least deprived quintile. Mortality was not significantly different across quintiles. In conclusion, findings suggest significant health equity issues for adolescent and adults with CHD based on area-based deprivation.

Background The Surgical Risk Preoperative Assessment System accurately predicts postoperative complications in elective surgery using only 8 preoperative variables, but its performance in emergency surgery has not been evaluated. We hypothesized that the Surgical Risk Preoperative Assessment System accurately predicts postoperative complications in emergency surgery and compared its performance to that of the American College of Surgeons Surgical Risk Calculator. Methods We calculated the Surgical Risk Preoperative Assessment System and the American College of Surgeons Surgical Risk Calculator risk estimates in a random sample of 1,010 emergency surgery cases from the American College of Surgeons National Surgical Quality Improvement Program 2014 to 2017 database. Risk estimates were compared with known outcomes. Analyses included the Hosmer-Lemeshow goodness of fit graphs and P values, c-indexes, and Brier scores. Results Results were similar between the Surgical Risk Preoperative Assessment System and the American College of Surgeons Surgical Risk Calculator for mortality, urinary tract infection, cardiac, venous thromboembolism, and renal complications. The American College of Surgeons Surgical Risk Calculator underestimated morbidity compared with the Surgical Risk Preoperative Assessment System (16.04% American College of Surgeons Surgical Risk Calculator vs 24.88% Surgical Risk Preoperative Assessment System vs 24.3% observed). Both calculators overestimated readmission (7.67% American College of Surgeons Surgical Risk Calculator vs 5.18% Surgical Risk Preoperative Assessment System vs 4.1% observed). Conclusion Both calculators predicted mortality, urinary tract infection, cardiac, venous thromboembolism, and renal complications well, but readmissions relatively poorly. The Surgical Risk Preoperative Assessment System estimated overall morbidity accurately, while the American College of Surgeons Surgical Risk Calculator underestimated this risk.

The burden of influenza in rural areas is largely unstudied. Rural populations may be vulnerable yet isolated from circulating virus. Laboratory-confirmed influenza hospitalizations in rural Colorado census tracts over eight influenza seasons were inconsistently distributed across seasons. Rural rates were, on average, lower than urban rates. Race, ethnicity, poverty, health insurance coverage, and distance from a hospital accounted for rate differences. Our interpretation is: 1) influenza regularly circulates in urban areas and inconsistently spreads to rural areas, 2) demographic and socioeconomic factors drive morbidity in exposed populations, and 3) public health interventions targeting high-risk urban census tracts may be beneficial.

Key Points Question Can machine learning approaches applied to harmonized electronic health record data identify patients at risk of 6-month incident atrial fibrillation with greater accuracy than standard risk factors? Findings This diagnostic study used electronic health record data from more than 2 million individuals to classify patients diagnosed with incident atrial fibrillation within a 6-month period, comparing several approaches to data management. A strategy that included the use of the 200 most common electronic health record features, random oversampling, and a single-layer neural network provided optimal classification of 6-month incident atrial fibrillation; however, this model was only marginally better than a logistic regression model with age, sex, and known risk factors for atrial fibrillation. Meaning Machine learning approaches applied to electronic health record data hold promise for predicting clinical outcomes, such as incident atrial fibrillation, but this model was not substantially more accurate than a logistic regression model with standard risk factors., This diagnostic study examines machine learning approaches applied to harmonized electronic health record data for identifying risk of atrial fibrillation., Importance Atrial fibrillation (AF) is the most common sustained cardiac arrhythmia, and its early detection could lead to significant improvements in outcomes through the appropriate prescription of anticoagulation medication. Although a variety of methods exist for screening for AF, a targeted approach, which requires an efficient method for identifying patients at risk, would be preferred. Objective To examine machine learning approaches applied to electronic health record data that have been harmonized to the Observational Medical Outcomes Partnership Common Data Model for identifying risk of AF. Design, Setting, and Participants This diagnostic study used data from 2 252 219 individuals cared for in the UCHealth hospital system, which comprises 3 large hospitals in Colorado, from January 1, 2011, to October 1, 2018. Initial analysis was performed in December 2018; follow-up analysis was performed in July 2019. Exposures All Observational Medical Outcomes Partnership Common Data Model–harmonized electronic health record features, including diagnoses, procedures, medications, age, and sex. Main Outcomes and Measures Classification of incident AF in designated 6-month intervals, adjudicated retrospectively, based on area under the receiver operating characteristic curve and F1 statistic. Results Of 2 252 219 individuals (1 225 533 [54.4%] women; mean [SD] age, 42.9 [22.3] years), 28 036 (1.2%) developed incident AF during a designated 6-month interval. The machine learning model that used the 200 most common electronic health record features, including age and sex, and random oversampling with a single-layer, fully connected neural network provided the optimal prediction of 6-month incident AF, with an area under the receiver operating characteristic curve of 0.800 and an F1 score of 0.110. This model performed only slightly better than a more basic logistic regression model composed of known clinical risk factors for AF, which had an area under the receiver operating characteristic curve of 0.794 and an F1 score of 0.079. Conclusions and Relevance Machine learning approaches to electronic health record data offer a promising method for improving risk prediction for incident AF, but more work is needed to show improvement beyond standard risk factors.

Background Clinicians and patients lack an evidence-based framework by which to judge individual-level recovery following total knee arthroplasty (TKA) surgery, thus impeding personalized treatment approaches for this elective surgery. Our study aimed to develop and validate a reference chart for monitoring recovery of knee flexion following TKA surgery. Methods Retrospective analysis of data collected in routine rehabilitation practice for patients following TKA surgery. Reference charts were constructed using Generalized Additive Models for Location Scale and Shape. Various models were compared using the Schwarz Bayesian Criterion, Mean Squared Error in 5-fold cross validation, and centile coverage (i.e. the percent of observed data represented below specified centiles). The performance of the reference chart was then validated against a test set of patients with later surgical dates, by examining the centile coverage and average bias (i.e. difference between observed and predicted values) in the test dataset. Results A total of 1173 observations from 327 patients were used to develop a reference chart for knee flexion over the first 120 days following TKA. The best fitting model utilized a non-linear time trend, with smoothing splines for median and variance parameters. Additionally, optimization of the number of knots in smoothing splines and power transformation of time improved model fit. The reference chart performed adequately in a test set of 171 patients (377 observations), with accurate centile coverage and minimal average bias (

Background The objective of this study was to develop an algorithm for identifying surgical site infections (SSIs) using independent variables from electronic health record data and outcomes from the American College of Surgeons National Surgical Quality Improvement Program to supplement manual chart review. Methods We fit 3 models to data from patients undergoing operations at the University of Colorado Hospital between 2013 and 2015: a similar model reported previously in the literature, a comprehensive model with 136 possible predictors, and a combination of those. All models used a generalized linear model with a lasso penalty. Several techniques for handling imbalance in the outcome were also used: Youden's J statistic to optimize the probability cutoff and sampling techniques combined with Youden's J. The models were then tested on data from patients undergoing operations during 2016. Results Two hundred thirty of 6,840 patients (3.4%) had an SSI. The comprehensive model fit to the full set of training data performed the best, achieving 90% specificity, 80% sensitivity, and an area under the receiver operating characteristic curve of 0.89. Conclusions We identified a model that accurately identified SSIs. The framework presented can be easily implemented by other American College of Surgeons National Surgical Quality Improvement Program-participating hospitals to develop models for enhancing surveillance of SSIs.

Background Prehospital (ambulance) care can reduce morbidity and mortality from trauma. Yet, there is a dearth of effective evidence-based interventions and implementation strategies. Emergency Medical Services Traumatic Shock Care (EMS-TruShoC) is a novel bundle of five core evidence-based trauma care interventions. High-Efficiency EMS Training (HEET) is an innovative training and sensitization program conducted during clinical shifts in ambulances. We assess the feasibility of implementing EMS-TruShoC using the HEET strategy, and feasibility of assessing implementation and clinical outcomes. Findings will inform a main trial. Methods We conducted a single-site, prospective cohort, multi-methods pilot implementation study in Western Cape EMS system of South Africa. Of the 120 providers at the study site, 12 were trainers and the remaining were eligible learners. Feasibility of implementation was guided by the RE-AIM (reach, effectiveness, adoption, implementation, and maintenance) framework. Feasibility of assessing clinical outcomes was assessed using shock indices and clinical quality of care scores, collected via abstraction of patients’ prehospital trauma charts. Thresholds for progression to a main trial were developed a priori. Results The average of all implementation indices was 83% (standard deviation = 10.3). Reach of the HEET program was high, with 84% learners completing at least 75% of training modules. Comparing the proportion of learners attaining perfect scores in post- versus pre-implementation assessments, there was an 8-fold (52% vs. 6%) improvement in knowledge, 3-fold (39% vs. 12%) improvement in skills, and 2-fold (42% vs. 21%) increase in self-efficacy. Clinical outcomes data were successfully calculated—there were clinically significant improvements in shock indices and quality of prehospital trauma care in the post- versus pre-implementation phases. Adoption of HEET was good, evidenced by 83% of facilitator participation in trainings, and 100% of surveyed stakeholders indicating good programmatic fit for their organization. Stakeholders responded that HEET was a sustainable educational solution that aligned well with their organization. Implementation fidelity was very high; 90% of the HEET intervention and 77% of the implementation strategy were delivered as originally planned. Participants provided very positive feedback, and explained that on-the-job timing enhanced their participation. Maintenance was not relevant to assess in this pilot study. Conclusions We successfully implemented the EMS-TruShoC educational intervention using the HEET training strategy in a single-site pilot study conducted in a low-resource international setting. All clinical outcomes were successfully calculated. Overall, this pilot study suggests high feasibility of our future, planned experimental trial.

Infectious diseases often present as coinfections that may affect each other in positive or negative ways. Understanding the relationship between two coinfecting pathogens is thus important to understand the risk of infection and burden of disease caused by each pathogen. Although coinfections with Plasmodium falciparum and Plasmodium vivax are very common outside Africa, it is yet unclear whether infections by the two parasite species are positively associated or if infection by one parasite suppresses the other. In this study, we use bivariate Poisson lognormal models (BPLM) to estimate covariate-adjusted associations between the incidence of infections (as measured by the force of blood-stage infections, molFOI) and clinical episodes caused by both P. falciparum and P. vivax in a cohort of Papua New Guinean children. A BPLM permits estimation of either positive or negative correlation, unlike most other multivariate Poisson models. Our results demonstrated a moderately positive association between P. falciparum and P. vivax infection rates, arguing against the hypothesis that P. vivax infections protect against P. falciparum infections. Our findings also suggest that the BPLM is only useful for counts with suitably large means and overdispersion.

Background Recurrent infections are associated with neurodevelopmental (ND) delay in infants, but the primary drivers are poorly understood. Leveraging an infant cohort from rural Guatemala designed to evaluate the effects of post-natal Zika virus on ND (DMID 16-0057), we evaluated the association between cumulative illness and ND delay and stunting. Methods Infants enrolled at 0-3 months of age underwent weekly at-home surveillance for caregiver-reported syndromic illness, including cough, fever and vomiting/diarrhea for a 12-month period. Anthropometric assessments and ND testing by Guatemalan psychologists using the Mullen Scales of Early Learning (MSEL) were perforrmed at 12-15 months of age. Multivariable generalized linear regression models were used to test associations between syndromic illness in infancy, 12-15-month MSEL Early Learning Composite (ELC) Score, and stunting (height-for-age < -2 SD) at 12-15 months. Results The cohort (n=425) had a mean enrollment age of 1.3 months; 202 (48%) were female, 387 (91%) self-reported a literate mother, and 301 (71%) were breastfeeding at study completion. Infants had reported illness for a median of 16 weeks during the surveillance period; cough was reported most frequently (median=11 weeks, range=0-37 weeks). Lower maternal education (p=0.007) and literacy (p=0.002) as well as infant age (p=0.007) and male gender (p=0.004) were associated with MSEL ELC Score Conclusion In a cohort of Guatemalan infants, cumulative fever and cough episodes were significantly associated with lower MSEL ELC Score, whereas there was no association with diarrhea/vomiting. In this low-resource community, these findings highlight the potential negative ND consequences of febrile illness and persistent cough in the first year of life. NIAID Contract HHSN272201300015I Task Order HHSN27200013 (Co-PIs: FMM and EJA). Disclosures Molly Lamb, PhD, BioFire (Grant/Research Support) Evan J. Anderson, MD, Sanofi Pasteur (Scientific Research Study Investigator)

Aim To identify factors associated with completion of Advance Care Planning (ACP) by patients seen in primary care in developed countries. We hypothesized that the quality of primary care is associated. Method We analysed respondent reported individual and healthcare utilization factors associated with the completion of ACP activities from the 2014 Commonwealth Fund International Health Policy Survey of Older Adults in 11 Countries. The primary outcome is the combined number of ACP activities completed. ACP activities included discussion of treatment preferences, documentation of healthcare wishes, or documentation of a surrogate decision maker. A quality of primary care index was calculated. Results Respondents averaged 69 years old. Most were women, graduated high school, rated their income as average or higher, and rated their health as good or better. A minority reported multimorbidity, accessed the emergency department or hospital, or were informal caregivers. Out of 25530 survey respondents, 13409 (53%) reported completion of any ACP activity; 11579 (45%) had discussed treatment preferences. Generalized linear mixed model results suggest that hospitalization (rate ratio [RR] 1.18), multimorbidity (RR 1.16), informal caregiving (RR 1.13), higher education level (RR 1.14), income (RR 1.05), access to higher quality primary care (RR 1.04) and ED visits (RR 1.04) were associated with higher rates of ACP activities. Male gender (RR 0.85) and higher perceived health status (RR 0.96) were associated with lower rates. Conclusions In this international study, individuals with greater interaction with the healthcare system through hospitalization, multimorbidity, access to quality primary care and informal caregiving reported more ACP activities.

Background Using the American College of Surgeons National Surgical Quality Improvement Program (NSQIP) complication status of patients who underwent an operation at the University of Colorado Hospital, we developed a machine learning algorithm for identifying patients with one or more complications using data from the electronic health record (EHR). Methods We used an elastic-net model to estimate regression coefficients and carry out variable selection. International classification of disease codes (ICD-9), common procedural terminology (CPT) codes, medications, and CPT-specific complication event rate were included as predictors. Results Of 6840 patients, 922 (13.5%) had at least one of the 18 complications tracked by NSQIP. The model achieved 88% specificity, 83% sensitivity, 97% negative predictive value, 52% positive predictive value, and an area under the curve of 0.93. Conclusions Using machine learning on EHR postoperative data linked to NSQIP outcomes data, a model with 163 predictors from the EHR identified complications well at our institution.

Objective To derive and validate a model of risk of septic shock among children with suspected sepsis, using data known in the electronic health record at hospital arrival. Study design This observational cohort study at 6 pediatric emergency department and urgent care sites used a training dataset (5 sites, April 1, 2013, to December 31, 2016), a temporal test set (5 sites, January 1, 2017 to June 30, 2018), and a geographic test set (a sixth site, April 1, 2013, to December 31, 2018). Patients 60 days to 18 years of age in whom clinicians suspected sepsis were included; patients with septic shock on arrival were excluded. The outcome, septic shock, was systolic hypotension with vasoactive medication or ≥30 mL/kg of isotonic crystalloid within 24 hours of arrival. Elastic net regularization, a penalized regression technique, was used to develop a model in the training set. Results Of 2464 included visits, septic shock occurred in 282 (11.4%). The model had an area under the curve of 0.79 (0.76-0.83) in the training set, 0.75 (0.69-0.81) in the temporal test set, and 0.87 (0.73-1.00) in the geographic test set. With a threshold set to 90% sensitivity in the training set, the model yielded 82% (72%-90%) sensitivity and 48% (44%-52%) specificity in the temporal test set, and 90% (55%-100%) sensitivity and 32% (21%-46%) specificity in the geographic test set. Conclusions This model estimated the risk of septic shock in children at hospital arrival earlier than existing models. It leveraged the predictive value of routine electronic health record data through a modern predictive algorithm and has the potential to enhance clinical risk stratification in the critical moments before deterioration.

Background National estimates of influenza burden may not reflect state‐level influenza activity, and local surveillance may not capture the full burden of influenza. Methods To provide state‐level information about influenza burden, we estimated excess pneumonia and influenza (P&I) and respiratory and circulatory (R&C) hospitalizations and deaths in Colorado from local hospital discharge records, death certificates, and influenza virus surveillance using negative binomial models. Results From July 2007 to June 2016, influenza was associated with an excess of 17 911 P&I hospitalizations (95%CI: 15 227, 20 354), 30 811 R&C hospitalizations (95%CI: 24 344, 37 176), 1,064 P&I deaths (95%CI: 757, 1298), and 3828 R&C deaths (95%CI: 2060, 5433). There was a large burden of influenza A(H1N1) among persons aged 0‐64 years, with high median seasonal rates of excess hospitalization among persons aged 0‐4 years. Persons aged ≥65 years experienced the largest numbers and highest median seasonal rates of excess hospitalization and death associated with influenza A (H3N2). The burden of influenza B was generally lower, with elevated median seasonal rates of excess hospitalization among persons aged 0‐4 years and ≥65 years. Conclusions These findings complement existing influenza surveillance. Periodic state‐level estimates of influenza disease burden may be useful for setting state public health priorities and planning prevention and control initiatives.