Lifting Universal Masking in Schools - Covid-19 Incidence among Students and Staff.

BACKGROUND: In February 2022, Massachusetts rescinded a statewide universal masking policy in public schools, and many Massachusetts school districts lifted masking requirements during the subsequent weeks. In the greater Boston area, only two school districts - the Boston and neighboring Chelsea districts - sustained masking requirements through June 2022. The staggered lifting of masking requirements provided an opportunity to examine the effect of universal masking policies on the incidence of coronavirus disease 2019 (Covid-19) in schools. METHODS: We used a difference-in-differences analysis for staggered policy implementation to compare the incidence of Covid-19 among students and staff in school districts in the greater Boston area that lifted masking requirements with the incidence in districts that sustained masking requirements during the 2021-2022 school year. Characteristics of the school districts were also compared. RESULTS: Before the statewide masking policy was rescinded, trends in the incidence of Covid-19 were similar across school districts. During the 15 weeks after the statewide masking policy was rescinded, the lifting of masking requirements was associated with an additional 44.9 cases per 1000 students and staff (95% confidence interval, 32.6 to 57.1), which corresponded to an estimated 11,901 cases and to 29.4% of the cases in all districts during that time. Districts that chose to sustain masking requirements longer tended to have school buildings that were older and in worse condition and to have more students per classroom than districts that chose to lift masking requirements earlier. In addition, these districts had higher percentages of low-income students, students with disabilities, and students who were English-language learners, as well as higher percentages of Black and Latinx students and staff. Our results support universal masking as an important strategy for reducing Covid-19 incidence in schools and loss of in-person school days. As such, we believe that universal masking may be especially useful for mitigating effects of structural racism in schools, including potential deepening of educational inequities. CONCLUSIONS: Among school districts in the greater Boston area, the lifting of masking requirements was associated with an additional 44.9 Covid-19 cases per 1000 students and staff during the 15 weeks after the statewide masking policy was rescinded.

Can algorithms Replace Expert Knowledge for Causal Inference? A Case Study on Novice Use of Causal Discovery.

BACKGROUND: With growing interest in causal inference and machine learning among epidemiologists, there is increasing discussion of causal discovery algorithms for guiding covariate selection. We present a case study of novice application of causal discovery tools and attempt to validate the results against a well-established causal relationship. METHODS: As a case study, we attempted causal discovery of relationships relevant to the effect of adherence on mortality in the placebo arm of the Coronary Drug Project (CDP) dataset. We used four algorithms available as existing software implementations and varied several model inputs. RESULTS: We identified 15 adjustment sets from 17 model parameterizations. When applied to a baseline covariate adjustment analysis, these 15 adjustment sets returned effect estimates with similar magnitude and direction of bias as prior published results. When using methods to control for time-varying confounding, there was generally more residual bias than compared to expert-selected adjustment sets. CONCLUSION: Although causal discovery algorithms can perform on par with expert knowledge, we do not recommend novice use of causal discovery without the input of experts in causal discovery. Expert support is recommended to aid in choosing the algorithm, selecting input parameters, assessing underlying assumptions, and finalizing selection of the adjustment variables.

Synthesizing Subject-matter Expertise for Variable Selection in Causal Effect Estimation: A Case Study.

BACKGROUND: Causal graphs are an important tool for covariate selection but there is limited applied research on how best to create them. Here, we used data from the Coronary Drug Project trial to assess a range of approaches to directed acyclic graph (DAG) creation. We focused on the effect of adherence on mortality in the placebo arm, since the true causal effect is believed with a high degree of certainty. METHODS: We created DAGs for the effect of placebo adherence on mortality using different approaches for identifying variables and links to include or exclude. For each DAG, we identified minimal adjustment sets of covariates for estimating our causal effect of interest and applied these to analyses of the Coronary Drug Project data. RESULTS: When we used only baseline covariate values to estimate the cumulative effect of placebo adherence on mortality, all adjustment sets performed similarly. The specific choice of covariates had minimal effect on these (biased) point estimates, but including nonconfounding prognostic factors resulted in smaller variance estimates. When we additionally adjusted for time-varying covariates of adherence using inverse probability weighting, covariates identified from the DAG created by focusing on prognostic factors performed best. CONCLUSION: Theoretical advice on covariate selection suggests that including prognostic factors that are not exposure predictors can reduce variance without increasing bias. In contrast, for exposure predictors that are not prognostic factors, inclusion may result in less bias control. Our results empirically confirm this advice. We recommend that hand-creating DAGs begin with the identification of all potential outcome prognostic factors.

Causal relationships between pain, medical treatments, and knee osteoarthritis: A graphical causal model to guide analyses.

OBJECTIVE: Randomized controlled trials (RCTs) are a gold standard for estimating the benefits of clinical interventions, but their decision-making utility can be limited by relatively short follow-up time. Longer-term follow-up of RCT participants is essential to support treatment decisions. However, as time from randomization accrues, loss to follow-up and competing events can introduce biases and require covariate adjustment even for intention-to-treat effects. We describe a process for synthesizing expert knowledge and apply this to long-term follow-up of an RCT of treatments for meniscal tears in patients with knee osteoarthritis (OA). METHODS: We identified 2 post-randomization events likely to impact accurate assessment of pain outcomes beyond 5 years in trial participants: loss to follow-up and total knee replacement (TKR). We conducted literature searches for covariates related to pain and TKR in individuals with knee OA and combined these with expert input. We synthesized the evidence into graphical models. RESULTS: We identified 94 potential covariates potentially related to pain and/or TKR among individuals with knee OA. Of these, 46 were identified in the literature review and 48 by expert panelists. We determined that adjustment for 50 covariates may be required to estimate the long-term effects of knee OA treatments on pain. CONCLUSION: We present a process for combining literature reviews with expert input to synthesize existing knowledge and improve covariate selection. We apply this process to the long-term follow-up of a randomized trial and show that expert input provides additional information not obtainable from literature reviews alone.

Looking under the lamp-post: quantifying the performance of contact tracing in the United States during the SARS-CoV-2 pandemic.

Contact tracing forms a crucial part of the public-health toolbox in mitigating and understanding emergent pathogens and nascent disease outbreaks. Contact tracing in the United States was conducted during the pre-Omicron phase of the ongoing COVID-19 pandemic. This tracing relied on voluntary reporting and responses, often using rapid antigen tests due to lack of accessibility to PCR tests. These limitations, combined with SARS-CoV-2's propensity for asymptomatic transmission, raise the question "how reliable was contact tracing for COVID-19 in the United States"? We answered this question using a Markov model to examine the efficiency with which transmission could be detected based on the design and response rates of contact tracing studies in the United States. Our results suggest that contact tracing protocols in the U.S. are unlikely to have identified more than 1.65% (95% uncertainty interval: 1.62-1.68%) of transmission events with PCR testing and 1.00% (95% uncertainty interval 0.98-1.02%) with rapid antigen testing. When considering a more robust contact tracing scenario, based on compliance rates in East Asia with PCR testing, this increases to 62.7% (95% uncertainty interval: 62.6-62.8%). We did not assume presence of asymptomatic transmission or superspreading, making our estimates upper bounds on the actual percentages traced. These findings highlight the limitations in interpretability for studies of SARS-CoV-2 disease spread based on U.S. contact tracing and underscore the vulnerability of the population to future disease outbreaks, for SARS-CoV-2 and other pathogens.

Preconception Periodontitis and Risk of Spontaneous Abortion in a Prospective Cohort Study.

Few studies have evaluated the association between periodontitis and spontaneous abortion (SAB), and all had limitations. We used data from the Pregnancy Study Online (PRESTO), a prospective preconception cohort study of 3,444 pregnancy planners in the United States and Canada (2019-2022), to address this question. Participants provided self-reported data on periodontitis diagnosis, treatment, and symptoms of severity (i.e., loose teeth) via the enrollment questionnaire. SAB (pregnancy loss at <20 weeks' gestation) was assessed via bimonthly follow-up questionnaires. Participants contributed person-time from the date of a positive pregnancy test to the gestational week of SAB, loss to follow-up, or 20 weeks' gestation, whichever came first. We fitted Cox regression models with weeks of gestation as the time scale to estimate adjusted hazard ratios (HRs) and 95% confidence intervals (CIs), and we used inverse probability of treatment weighting to account for differential loss to follow-up. We used probabilistic quantitative bias analysis to estimate the magnitude and direction of the effect of exposure misclassification bias on results. In weighted multivariable models, we saw no appreciable association between preconception periodontitis diagnosis (HR = 0.97, 95% CI: 0.76, 1.23) or treatment (HR = 1.01, 95% CI: 0.79, 1.27) and SAB. A history of loose teeth was positively associated with SAB (HR = 1.38, 95% CI: 0.88, 2.14). Quantitative bias analysis indicated that our findings were biased towards the null but with considerable uncertainty in the bias-adjusted results.

Comparative Effectiveness of Dynamic Treatment Strategies for Medication Use and Dosage: Emulating a Target Trial Using Observational Data.

BACKGROUND: Availability of detailed data from electronic health records (EHRs) has increased the potential to examine the comparative effectiveness of dynamic treatment strategies using observational data. Inverse probability (IP) weighting of dynamic marginal structural models can control for time-varying confounders. However, IP weights for continuous treatments may be sensitive to model choice. METHODS: We describe a target trial comparing strategies for treating anemia with darbepoetin in hemodialysis patients using EHR data from the UK Renal Registry 2004 to 2016. Patients received a specified dose (microgram/week) or did not receive darbepoetin. We compared 4 methods for modeling time-varying treatment: (A) logistic regression for zero dose, standard linear regression for log dose; (B) logistic regression for zero dose, heteroscedastic linear regression for log dose; (C) logistic regression for zero dose, heteroscedastic linear regression for log dose, multinomial regression for patients who recently received very low or high doses; and (D) ordinal logistic regression. RESULTS: For this dataset, method (C) was the only approach that provided a robust estimate of the mortality hazard ratio (HR), with less-extreme weights in a fully weighted analysis and no substantial change of the HR point estimate after weight truncation. After truncating IP weights at the 95th percentile, estimates were similar across the methods. CONCLUSIONS: EHR data can be used to emulate target trials estimating the comparative effectiveness of dynamic strategies adjusting treatment to evolving patient characteristics. However, model checking, monitoring of large weights, and adaptation of model strategies to account for these is essential if an aspect of treatment is continuous.

Emulating Target Trials to Avoid Immortal Time Bias - An Application to Antibiotic Initiation and Preterm Delivery.

BACKGROUND: Randomized trials in pregnancy are extremely challenging, and observational studies are often the only option to evaluate medication safety during pregnancy. However, such studies are often susceptible to immortal time bias if treatment initiation occurs after time zero of follow-up. We describe how emulating a sequence of target trials avoids immortal time bias and apply the approach to estimate the safety of antibiotic initiation between 24 and 37 weeks gestation on preterm delivery. METHODS: The Tsepamo Study captured birth outcomes at hospitals throughout Botswana from 2014 to 2021. We emulated 13 sequential target trials of antibiotic initiation versus no initiation among individuals presenting to care <24 weeks, one for each week from 24 to 37 weeks. For each trial, eligible individuals had not previously initiated antibiotics. We also conducted an analysis susceptible to immortal time bias by defining time zero as 24 weeks and exposure as antibiotic initiation between 24 and 37 weeks. We calculated adjusted risk ratios (RR) and 95% confidence intervals (CI) for preterm delivery. RESULTS: Of 111,403 eligible individuals, 17,009 (15.3%) initiated antibiotics between 24 and 37 weeks. In the sequence of target trials, RRs (95% CIs) ranged from 1.04 (0.90, 1.19) to 1.24 (1.11, 1.39) (pooled RR: 1.11 [1.06, 1.15]). In the analysis susceptible to immortal time bias, the RR was 0.90 (0.86, 0.94). CONCLUSIONS: Defining exposure as antibiotic initiation at any time during follow-up after time zero resulted in substantial immortal time bias, making antibiotics appear protective against preterm delivery. Conducting a sequence of target trials can avoid immortal time bias in pregnancy studies.

On the Need to Revitalize Descriptive Epidemiology.

Nearly every introductory epidemiology course begins with a focus on person, place, and time, the key components of descriptive epidemiology. And yet in our experience, introductory epidemiology courses were the last time we spent any significant amount of training time focused on descriptive epidemiology. This gave us the impression that descriptive epidemiology does not suffer from bias and is less impactful than causal epidemiology. Descriptive epidemiology may also suffer from a lack of prestige in academia and may be more difficult to fund. We believe this does a disservice to the field and slows progress towards goals of improving population health and ensuring equity in health. The severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) outbreak and subsequent coronavirus disease 2019 pandemic have highlighted the importance of descriptive epidemiology in responding to serious public health crises. In this commentary, we make the case for renewed focus on the importance of descriptive epidemiology in the epidemiology curriculum using SARS-CoV-2 as a motivating example. The framework for error we use in etiological research can be applied in descriptive research to focus on both systematic and random error. We use the current pandemic to illustrate differences between causal and descriptive epidemiology and areas where descriptive epidemiology can have an important impact.

Choice of Outcome in COVID-19 Studies and Implications for Policy: Mortality and Fatality.

In this brief communication, we discuss the confusion of mortality with fatality in the interpretation of evidence in the coronavirus disease 2019 (COVID-19) pandemic, and how this confusion affects the translation of science into policy and practice. We discuss how this confusion has influenced COVID-19 policy in France, Sweden, and the United Kingdom and discuss the implications for decision-making about COVID-19 vaccine distribution. We also discuss how this confusion is an example of a more general statistical fallacy we term the "Missing Link Fallacy."

The Expert Next Door: Interactions With Friends and Family During the COVID-19 Pandemic.

The coronavirus disease 2019 (COVID-19) pandemic thrust the field of public health into the spotlight. For many epidemiologists, biostatisticians, and other public health professionals, this caused the professional aspects of our lives to collide with the personal, as friends and family reached out with concerns and questions. Learning how to navigate this space was new for many of us and required refining our communication style depending on context, setting, and audience. Some of us took to social media, utilizing our existing personal accounts to share information after sorting through and summarizing the rapidly emerging literature to keep loved ones safe. However, those in our lives sometimes asked unanswerable questions, or began distancing themselves when we suggested more stringent guidance than they had hoped for, causing additional stress during an already traumatic time. We often had to remind ourselves that we were also individuals experiencing this pandemic and that our time-intensive efforts were meaningful, relevant, and impactful. As this pandemic and other public health crises continue, we encourage members of our discipline to consider how we can best use shared lessons from this period and to recognize that our professional knowledge, when used in our personal lives, can promote, protect, and bolster confidence in public health.

Causal and Associational Language in Observational Health Research: A Systematic Evaluation.

We estimated the degree to which language used in the high-profile medical/public health/epidemiology literature implied causality using language linking exposures to outcomes and action recommendations; examined disconnects between language and recommendations; identified the most common linking phrases; and estimated how strongly linking phrases imply causality. We searched for and screened 1,170 articles from 18 high-profile journals (65 per journal) published from 2010-2019. Based on written framing and systematic guidance, 3 reviewers rated the degree of causality implied in abstracts and full text for exposure/outcome linking language and action recommendations. Reviewers rated the causal implication of exposure/outcome linking language as none (no causal implication) in 13.8%, weak in 34.2%, moderate in 33.2%, and strong in 18.7% of abstracts. The implied causality of action recommendations was higher than the implied causality of linking sentences for 44.5% or commensurate for 40.3% of articles. The most common linking word in abstracts was "associate" (45.7%). Reviewers' ratings of linking word roots were highly heterogeneous; over half of reviewers rated "association" as having at least some causal implication. This research undercuts the assumption that avoiding "causal" words leads to clarity of interpretation in medical research.

Utility of a Device Briefing Tool to Improve Surgical Safety.

INTRODUCTION: Clear communication around surgical device use is crucial to patient safety. We evaluated the utility of the Device Briefing Tool (DBT) as an adjunct to the Surgical Safety Checklist. METHODS: A nonrandomized, controlled pilot of the DBT was conducted with surgical teams at an academic referral center. Intervention departments used the DBT in all cases involving a surgical device for 10 wk. Utility, relative advantage, and implementation effectiveness were evaluated via surveys. Trained observers assessed adherence and team performance using the Oxford NOTECHS system. RESULTS: Of 113 individuals surveyed, 91 responded. Most respondents rated the DBT as moderately to extremely useful. Utility was greatest for complex devices (89%) and new devices (88%). Advantages included insight into the team's familiarity with devices (70%) and improved teamwork and communication (68%). Users found it unrealistic to review all device instructional materials (54%). Free text responses suggested that the DBT heightened awareness of deficiencies in device familiarity and training but lacked a clear mechanism to correct them. DBT adherence was 82%. NOTECHS scores in intervention departments improved over the course of the study but did not significantly differ from comparator departments. CONCLUSIONS: The DBT was rated highly by both surgeons and nurses. Adherence was high and we found no evidence of "checklist fatigue." Centers interested in implementing the DBT should focus on devices that are complex or new to any surgical team member. Guidance for correcting deficiencies identified by the DBT will be provided in future iterations of the tool.

Spillover benefit of pre-exposure prophylaxis for HIV prevention: evaluating the importance of effect modification using an agent-based model.

We developed an agent-based model using a trial emulation approach to quantify effect measure modification of spillover effects of pre-exposure prophylaxis (PrEP) for HIV among men who have sex with men (MSM) in the Atlanta-Sandy Springs-Roswell metropolitan area, Georgia. PrEP may impact not only the individual prescribed, but also their partners and beyond, known as spillover. We simulated a two-stage randomised trial with eligible components (≥3 agents with ≥1 HIV+ agent) first randomised to intervention or control (no PrEP). Within intervention components, agents were randomised to PrEP with coverage of 70%, providing insight into a high PrEP coverage strategy. We evaluated effect modification by component-level characteristics and estimated spillover effects on HIV incidence using an extension of randomisation-based estimators. We observed an attenuation of the spillover effect when agents were in components with a higher prevalence of either drug use or bridging potential (if an agent acts as a mediator between ≥2 connected groups of agents). The estimated spillover effects were larger in magnitude among components with either higher HIV prevalence or greater density (number of existing partnerships compared to all possible partnerships). Consideration of effect modification is important when evaluating the spillover of PrEP among MSM.

Emulating Target Trials to Improve Causal Inference From Agent-Based Models.

Agent-based models are a key tool for investigating the emergent properties of population health settings, such as infectious disease transmission, where the exposure often violates the key "no interference" assumption of traditional causal inference under the potential outcomes framework. Agent-based models and other simulation-based modeling approaches have generally been viewed as a separate knowledge-generating paradigm from the potential outcomes framework, but this can lead to confusion about how to interpret the results of these models in real-world settings. By explicitly incorporating the target trial framework into the development of an agent-based or other simulation model, we can clarify the causal parameters of interest, as well as make explicit the assumptions required for valid causal effect estimation within or between populations. In this paper, we describe the use of the target trial framework for designing agent-based models when the goal is estimation of causal effects in the presence of interference, or spillover.

Will Podcasting and Social Media Replace Journals and Traditional Science Communication? No, but...

The digital world in which we live is changing rapidly. The evolving media environment is having a direct impact on traditional forms of communication and knowledge translation in public health and epidemiology. Openly accessible digital media can be used to reach a broader and more diverse audience of trainees, scientists, and the lay public than can traditional forms of scientific communication. The new digital landscape for delivering content is vast, and new platforms are continuously being added. In this article, we focus on several, including Twitter and podcasting, and discuss their relevance to epidemiology and science communication. We highlight 3 key reasons why we think epidemiologists should be engaging with these mediums: 1) science communication, 2) career advancement, and 3) development of a community and public service. Other positive and negative consequences of engaging in these forms of new media are also discussed. The authors of this commentary are all engaged in social media and podcasting for scientific communication, and we reflect on our experiences with these mediums as tools to advance the field of epidemiology.

Concerns About the Special Article on Hydroxychloroquine and Azithromycin in High-Risk Outpatients With COVID-19.

In May 2020, the Journal published an opinion piece by a member of the Editorial Board, in which the author reviewed several papers and argued that using hydroxychloroquine (HCQ) + azithromycin (AZ) early to treat symptomatic coronavirus disease 2019 (COVID-19) cases in high-risk patients should be broadly applied. As members of the Journal's Editorial Board, we are strongly supportive of open debate in science, which is essential even on highly contentious issues. However, we must also be thorough in our examination of the facts and open to changing our minds when new information arises. In this commentary, we document several important errors in the manuscript, review the literature presented, and demonstrate why it is not of sufficient quality to support scale up of HCQ + AZ, and then discuss the literature that has been generated since the publication, which also does not support use of this therapy. Unfortunately, the current scientific evidence does not support HCQ + AZ as an effective treatment for COVID-19, if it ever did, and even suggests many risks. Continuing to push the view that it is an essential treatment in the face of this evidence is irresponsible and harmful to the many people already suffering from infection.

Disseminated Effects in Agent-Based Models: A Potential Outcomes Framework and Application to Inform Preexposure Prophylaxis Coverage Levels for HIV Prevention.

Preexposure prophylaxis (PrEP) for prevention of human immunodeficiency virus (HIV) infection may benefit not only the person who uses it but also their uninfected sexual risk contacts. We developed an agent-based model using a novel trial emulation approach to quantify disseminated effects of PrEP use among men who have sex with men in Atlanta, Georgia, from 2015 to 2017. Model components (subsets of agents connected through partnerships in a sexual network but not sharing partnerships with any other agents) were first randomized to an intervention coverage level or the control group; then, within intervention components, eligible agents were randomized to receive or not receive PrEP. We calculated direct and disseminated (indirect) effects using randomization-based estimators and report corresponding 95% simulation intervals across scenarios ranging from 10% coverage in the intervention components to 90% coverage. A population of 11,245 agents was simulated, with an average of 1,551 components identified. When comparing agents randomized to no PrEP in 70% coverage components with control agents, there was a 15% disseminated risk reduction in HIV incidence (risk ratio = 0.85, 95% simulation interval: 0.65, 1.05). Persons not on PrEP may receive a protective benefit by being in a sexual network with higher PrEP coverage. Agent-based models are useful for evaluating possible direct and disseminated effects of HIV prevention modalities in sexual networks.

A biologist's guide to model selection and causal inference.

A goal of many research programmes in biology is to extract meaningful insights from large, complex datasets. Researchers in ecology, evolution and behavior (EEB) often grapple with long-term, observational datasets from which they construct models to test causal hypotheses about biological processes. Similarly, epidemiologists analyse large, complex observational datasets to understand the distribution and determinants of human health. A key difference in the analytical workflows for these two distinct areas of biology is the delineation of data analysis tasks and explicit use of causal directed acyclic graphs (DAGs), widely adopted by epidemiologists. Here, we review the most recent causal inference literature and describe an analytical workflow that has direct applications for EEB. We start this commentary by defining four distinct analytical tasks (description, prediction, association, causal inference). The remainder of the text is dedicated to causal inference, specifically focusing on the use of DAGs to inform the modelling strategy. Given the increasing interest in causal inference and misperceptions regarding this task, we seek to facilitate an exchange of ideas between disciplinary silos and provide an analytical framework that is particularly relevant for making causal inference from observational data.