Andrew J. Spieker

COVID‐19 has posed significant challenges worldwide, with pregnant individuals identified as being at increased risk for severe illness and poor outcomes compared to the general population. This study aimed to compare the course of critical illness due to COVID‐19 of pregnant patients to nonpregnant reproductive‐age women and describe their perinatal outcomes. A retrospective cohort study was conducted including women ages 14–49 years admitted to the intensive care unit (ICU) with COVID‐19–related illness from 3/2020 to 2/2022 at Vanderbilt University Medical Center in Nashville, Tennessee. Demographic, clinical characteristics, and outcomes were compared between pregnant/postpartum and nonpregnant patients. Neonatal outcomes were also analyzed. Among 159 women, 35 (22%) were pregnant or postpartum. Pregnant patients had a lower body mass index, were more likely to be publicly insured, and were less likely to have underlying comorbidities. The groups had similar proportions of mechanical ventilation, venous thromboembolism, and death. However, pregnant patients were more likely to require extracorporeal membrane oxygenation (ECMO). The proportion of individuals who were vaccinated against COVID‐19 was low in both groups. Of 19 neonates born, the mean gestational age was 32.9 weeks, with 94.7% delivered via cesarean and 78.9% requiring NICU admission. Five intrauterine fetal demises and one neonatal death occurred. Though pregnant patients admitted to the ICU with COVID‐19 had fewer comorbidities compared to nonpregnant women, nearly one‐fifth of them required ECMO. Pregnant patients with COVID‐19 ICU admission face high risks of ECMO, preterm delivery, cesarean, and fetal loss. Improving vaccination uptake is critical, especially in this population.

Background: Human adenovirus (HAdV) is a common cause of pediatric acute respiratory illness (ARI), contributing to 5–13% of cases worldwide. Clinical manifestations vary by HAdV species and type; therefore, delineating type-specific disease presentations and understanding severity of specific HAdV types' disease may help develop targeted interventions. Methods: We conducted a multicenter, prospective study within the New Vaccine Surveillance Network to characterize HAdV types. Children <18 years old with ARI were enrolled in the emergency department or inpatient setting at 7 US children's hospitals from 1 December 2016 to 30 November 2019. Respiratory specimens were collected and tested for HAdV and other viruses. Subsequently, typing was conducted on HAdV specimens using single-plex real-time PCR assays targeting sequences in the hexon gene. Comparisons between HAdV types were performed to determine differences in characteristics and outcomes. Generalized linear mixed effects models were used to evaluate severity. Results: Among the 1843 HAdV-positive cases, 1402 specimens (76%) were typed. The most common types detected were HAdV-C1 (n = 439), HAdV-C2 (n = 393), and HAdV-B3 (n = 221). Children with HAdV-B7 (n = 78) had higher odds of severe outcomes compared to those with other HAdV types (aOR = 2.05; 95% CI: 1.24, 3.40). Symptom presentation varied across types within species B, C, and E; while all species had high frequency of upper respiratory symptoms, species B cases presented with a higher frequency of non-respiratory manifestations. Conclusions: Among children with HAdV-positive ARI, those with HAdV-B7 had higher odds of severe outcomes. These findings suggest heterogeneity in clinical presentation and severity among HAdV types, emphasizing the importance of HAdV type in future prevention and treatment strategies.

IntroductionI commend Drs. Bellamy and Sullivan for initiating an important conversation about the role of qualifying examinations (QEs) in biostatistics graduate programs, and echo their sentiments regarding the importance of fostering an inclusive and welcoming environment for trainees.1 Bellamy and Sullivan conclude their report with a call to action: “…[to] become clear on the purpose of the QE, beyond being an arbitrary milestone along the doctoral training continuum.” They provide compelling support for their case that contemporary graduate training programs have room for growth in this respect. The major purpose of my commentary is to take specific steps toward addressing this.Characterizing the role of the QEsSince 2021, I have led the writing/evaluation of the QEs for Vanderbilt University’s graduate program in biostatistics. For me, the timing of this discussion is fortuitous and opportune as I transition into the role of Director of Graduate Studies. Our program’s QEs are a mixture of in-class and take-home problems oriented toward theory, methods, and application. At the time of this writing, our program’s student handbook asserts the following:The primary role of the comprehensive examinations is to aid the graduate faculty in assessing whether students are sufficiently prepared for the next stage of their graduate training, including advanced coursework and development of a thesis or dissertation.In some respects, this parallels what Bellamy and Sullivan articulate (and challenge) as a commonly purported argument in favor of QEs. I view the above statement of purpose as adequate in distinguishing a QE from an arbitrary milestone. On the other hand, Bellamy and Sullivan raise a legitimate question as to the optimality of the QE in achieving this goal. I believe optimality hinges upon the care put into developing the examinations and the program’s sustained communication with its students. Below, I offer recommendations that balance fairness with the high standards that our students expect us to maintain.1.Be transparent, forthcoming, and available. For several years, I have held sessions to explain the QE process in detail as early as six months prior. When providing tips, I’ve emphasized the balance between individual- and group-study. I’ve pointed students to the website that hosts several previous exams (publicly available and easy to access) so they can acclimate to the exam format. Finally, I’ve offered review sessions each year, and each year (to my delight) students have taken me up on this.2.Ensure balance in scope and difficulty. Many programs delegate exam writing across graduate faculty to ensure the burden (considerable as it is) is not unduly placed on any one person. Without some degree of centralized oversight, however, this runs the risk of an imbalanced or otherwise unreasonable examination. I share Bellamy and Sullivan’s stance on the counterproductivity of questions designed to be difficult for difficulty’s sake. I have insisted not only that each question is vetted by the instructor of the course corresponding to that question’s topic, but that each examination is vetted in its entirety by at least one other faculty member for scope, clarity, and alignment.3.Mind the modernity. As our expectations of proficiency in certain topics (e.g., computing) increase over time, the direct applicability of some historically emphasized topics in mathematical statistics wane. While it requires some creativity in the exam writing process, we can reflect these changes by periodically updating the exam’s emphasis.4.Encourage students to seek resources for accommodations. We can be sensitive to heterogeneity in learning styles. I encourage students who believe they may be eligible for accommodations to consult with the student support center on campus. Many times, students who may be eligible for accommodations may not be aware of this important service until it is brought to their attention.5.Use of the QE as an aid. Our program’s current characterization of the QE clarifies that it is not considered in isolation. As a medium-sized program, we have the luxury of being able to take an individualized approach to remediation (e.g., re-doing a problem in a time-limited open-book setting, retaking a course, completing a follow-up oral examination, or serving as a TA for a targeted course) following holistic discussions of a student’s progress (with the QE treated as a piece—rather than the whole—of the puzzle). To the extent possible, decisions should be based on a pattern of signals.The above is not an exhaustive characterization of the measures our program has taken to ensure the quality of our QEs, but is instead a sampling of the sorts of things I believe may alleviate some concerns put forth by Bellamy and Sullivan. Fundamentally, I view the QEs as a motivation for students to consolidate their understanding. I also view proficiency in the foundations of mathematical statistics as a prerequisite to many contemporary advanced topics; nevertheless, we can certainly acknowledge the importance of the foundations without endorsing the deeply disturbing viewpoints of some of their developers. I further posit that well-constructed QEs have the following potential advantages toward the goal of inclusivity:1.Though a QE may be imperfect, we must consider the unintended consequences of removing an objective indicator of performance from consideration. In so doing, we could inadvertently increase reliance on the unconscious biases we seek to avoid.2.Students taking a longer time to process and consolidate material may benefit immensely from having dedicated time to study, followed by an additional chance to demonstrate what they’ve learned. This “glass-half-full” perspective on the utility of a QE is in direct opposition to the more pessimistic...

Background: Risk factors for severe respiratory syncytial virus (RSV) illness include early infancy, premature birth, and underlying medical conditions. However, the clinical significance of respiratory viral co-detection is unclear. We compared the clinical outcomes of young children with RSV-only detection and those with RSV viral co-detection. Methods: We conducted active, population-based surveillance of children with medically attended fever or respiratory symptoms at 7 US medical centers (1 December 2016–31 March 2020). Demographic and clinical data were collected through parental interviews and chart abstractions. Nasal swabs, with or without throat swabs, were systematically tested for RSV and 6 other common respiratory virus groups. We compared clinical outcomes, including hospitalization, and among those hospitalized, length of stay, intensive care unit admission, supplemental oxygen use, and intubation, between children aged <2 years with RSV-only detection and those with RSV co-detection. Results: We enrolled 18 008 children aged <2 years. Of 17 841 (99.1%) tested for RSV, 5099 (28.6%) were positive. RSV was singly detected in 3927 children (77.0%) and co-detected in 1172 (23.0%). RSV co-detection with parainfluenza virus or adenovirus was associated with significantly lower odds of hospitalization (adjusted odds ratio, 0.56; 95% confidence interval [CI]: .33–.95; P = .031) and supplemental oxygen use (adjusted odds ratio, 0.66; 95% CI: .46–.95; P = .026), respectively, than RSV-only detection. For all other comparisons, we did not identify a significant association between RSV co-detection and worse clinical outcomes. Conclusions: Co-detection of RSV with another respiratory virus was not significantly associated with worse clinical outcomes compared with RSV-only detection.

Targeted maximum likelihood estimation (TMLE) is an increasingly popular framework for the estimation of causal effects. It requires modeling both the exposure and outcome but is doubly robust in the sense that it is valid if at least one of these models is correctly specified. In addition, TMLE allows for flexible modeling of both the exposure and outcome with machine learning methods. This provides better control for measured confounders since the model specification automatically adapts to the data, instead of needing to be specified by the analyst a priori. Despite these methodological advantages, TMLE remains less popular than alternatives in part because of its less accessible theory and implementation. While some tutorials have been proposed, none address the case of a time‐to‐event outcome. This tutorial provides a detailed step‐by‐step explanation of the implementation of TMLE for estimating the effect of a point binary or multilevel exposure on a time‐to‐event outcome, modeled as counterfactual survival curves and causal hazard ratios. The tutorial also provides guidelines on how best to use TMLE in practice, including aspects related to study design, choice of covariates, controlling biases and use of machine learning. R‐code is provided to illustrate each step using simulated data ( https://github.com/detal9/SurvTMLE). To facilitate implementation, a general R function implementing TMLE with options to use machine learning is also provided. The method is illustrated in a real‐data analysis concerning the effectiveness of statins for the prevention of a first cardiovascular disease among older adults in Québec, Canada, between 2013 and 2018.

Receiver operating characteristic (ROC) analysis is a tool to evaluate the capacity of an ordinal or numeric measure to distinguish between states, often employed in the evaluation of diagnostic tests. Assumptions regarding the nature of the data or the structure of the ROC curve can improve efficiency of estimation for the ROC curve itself or its associated area under the curve (a measure of stochastic ordering closely related to the target parameter of a Mann-Whitney U test). Despite recent interest in methods to conduct comparisons by way of stochastic ordering, ROC modeling strategies are not widely known but may be appreciated by practitioners of these methods. The major goals of this manuscript are to (1) provide an overview of nonparametric, parametric, and semiparametric frameworks for ROC curve estimation, (2) guide the practice of ROC analysis by offering considerations regarding methodological trade-offs, and (3) supply sample R code to further guide implementation. We utilize simulations to demonstrate various features of the methods we discuss. As an illustrative example, we analyze data from a recent cohort study in order to compare responses to SARS-CoV-2 vaccination between solid organ transplant recipients and healthy individuals.

Background: Guidelines state that all hospitalized children with suspected or confirmed influenza receive prompt treatment with influenza-specific antivirals. We sought to determine the frequency of, and factors associated with, antiviral receipt among hospitalized children. Methods: We conducted active surveillance of children presenting with fever or respiratory symptoms from 1 December 2016 to 31 March 2020 at 7 pediatric medical centers in the New Vaccine Surveillance Network. The cohort consisted of children hospitalized with influenza A or B confirmed by clinical or research testing. The primary outcome was frequency of antiviral receipt during hospitalization. We used logistic regression to obtain adjusted odds ratios (aORs) and 95% confidence intervals (CIs) for factors associated with antiviral receipt. Results: A total of 1213 children with laboratory-confirmed influenza were included. Overall, 652 children (53.8%) received an antiviral. Roughly 63.0% of children received clinical influenza testing. Among those with clinical testing, 67.4% received an antiviral. Factors associated with higher odds of antiviral receipt included hematologic (aOR = 1.76; 95% CI = 1.03–3.02) or oncologic/immunocompromising (aOR = 2.41; 95% CI = 1.13–5.11) disorders, prehospitalization antiviral receipt (aOR = 2.34; 95% CI = 1.49–3.67), clinical influenza testing (aOR = 3.07; 95% CI = 2.28–4.14), and intensive care unit admission (aOR = 1.53; 95% CI = 1.02–2.29). Symptom duration >2 days was associated with lower odds of antiviral treatment (aOR = 0.40; 95% CI = .30–.52). Antiviral receipt varied by site with a 5-fold difference across sites. Conclusions: Almost half of children hospitalized with influenza did not receive antivirals. Additional efforts to understand barriers to guideline adherence are crucial for optimizing care in children hospitalized with influenza.

Background: Observationally, family and social support are important for optimal diabetes self-management; however, interventions targeting family/social support have not consistently been effective. A novel, diabetes-specific family functioning typology offers the opportunity to classify types of baseline family functioning to determine for whom family interventions may be effective. Purpose: We examined the effects of an intervention by baseline type of family functioning post hoc, to inform differential benefit from interventions. Methods: Participants were randomized to enhanced treatment as usual or a 9-month, mobile phone-delivered, family-focused, self-care support intervention. Adults with type 2 diabetes (N = 318) who participated in the randomized clinical trial (RCT) and provided baseline data were included. We determined participants’ diabetes-specific family functioning types at baseline using a validated, survey-administered, typology assessment tool. We investigated the associations between type and engagement (eg, attending coaching sessions and responding to text messages) and psychosocial (eg, well-being and diabetes distress) and glycemic outcomes at mid- and post-treatment. Results: Despite overall high engagement, there was variability across types in engagement and effects. Want More Involvement benefited the most; Satisfied with Low Involvement showed early improvements that waned; Collaborative & Helpful were highly engaged but derived minimal benefits from the intervention; and Critically Involved benefitted the least and may have experienced some harm. Conclusions: We demonstrated the utility of a novel diabetes-specific family functioning typology to explain variability in response to a family-focused intervention. Findings from this work answer the calls for systems-level consideration in precision behavioral medicine and drive hypothesis generation for future, tailored interventions. Registration: The larger RCT is registered with ClinicalTrials.gov (NCT04347291).

Background: Influenza is associated with significant global morbidity and mortality, with vaccination being the primary preventive strategy. Despite recommendations, influenza vaccine uptake among healthcare providers (HCPs) remains suboptimal, especially in the Eastern Mediterranean. We aimed to assess the attitudes and practices of HCPs in Jordan regarding seasonal influenza vaccination and assess sources of variation thereof. Methods: We conducted a cross-sectional survey study among a sample of HCPs practicing in Jordan (12/29/2020–04/26/2021). Participants completed a questionnaire assessing demographics, influenza vaccination history, attitudes, and practices. We used logistic regression to evaluate factors related to vaccine receipt and reasons for non-vaccination. We used proportional odds models to evaluate factors related to HCP recommendations and to compare opinions on influenza vaccination between ever- and never-vaccinated HCPs. Results: Of 305 survey initiates, 206 HCPs (67.5%) comprised the analytic sample. The median age was 35 years; 61.2% were women, and 43.7% were pharmacists. Over half (52.9%) never received an influenza vaccine; however, older age and self-identifying as a physician were associated with higher odds of having ever received the influenza vaccine. The main reasons for non-vaccination were related to the misassessment of risks and benefits. Prior receipt of influenza vaccination was strongly associated with odds of recommending vaccination (or = 10.5; 95% CI = [5.38–20.3]; p<0.001). The COVID-19 pandemic reportedly enhanced influenza vaccine acceptance among 48.5% of HCPs surveyed. Conclusions: Low influenza vaccine uptake among healthcare providers in Jordan is related to misassessment of risks and benefits. Enhancing attitudes and confidence through tailored education is crucial to overcoming hesitancy and promoting sustained improvements in vaccination attitudes and practices among HCPs in Jordan.