Comment on Ellenberg and Morris: The role of statisticians in vaccine surveillance
Bibliographic record
Abstract
Ellenberg and Morris 1 illustrate nicely the similarities and contrasts between the HIV and COVID-19 pandemics, and the roles of statisticians in related research.The article ends with the exciting news of emergency use authorization for the two new mRNA-based vaccines that appear extraordinarily effective against COVID-19.This news, and the fact that there are now several vaccines either approved or close to being so in several countries, 2 represents one of the most significant differences between the two pandemics.The delivery and use of coronavirus vaccines present another challenge in which biostatisticians will be implicated in an unprecedented way.We are at the beginning of perhaps the largest vaccine distribution, and therefore the largest vaccine surveillance project, ever undertaken.The worldwide scope and rapidity of the distribution effort means that any research on early vaccines needs to be similarly rapid and of a massive scope, and there are a number of novel aspects of this vaccination effort in which statisticians will be important contributors.Vaccine surveillance typically works through surveillance of spontaneous adverse events collected by self-report (VAERS), and more focused epidemiologic studies to assess potential causal associations between a specific vaccine regimen and an adverse event.3 Observational studies of vaccine efficacy can assess real-world effectiveness against endpoints studied in trials, such as severity, but also important endpoints that were not studied in trials, such as the degree to which the vaccines can prevent transmission.What is novel about the Coronavirus pandemic and vaccine effort?The vaccine effort already includes 10 different vaccines, approved in multiple countries.2 The size of the effort, and its rapidity (over 160 million doses administered already, 4 with over 10 billion doses and 10 different vaccines promised by the end of 2021 2 ) is unprecedented.Finally, vaccines are being administered following a range of dosing strategies.Both the Pfizer and Moderna vaccines were approved for two-dose regimens with the second dose delivered in a fixed time frame.Several countries are considering delaying the second dose, or even single-dose regimens for these vaccines.4 While there is some evidence that single-dose regimens are effective, and that delaying the second dose does not affect vaccine efficacy, these strategies have not been studied in large-scale randomized trials.Observational studies will be the only way we can assess the safety and efficacy of the various vaccines, and the various different regimens, on a large scale.3,5,6 What are the challenges in conducting such surveillance?First, and foremost, systematic data collection are critical.Studies of vaccine safety and effectiveness depend on knowing dosing dates, type of vaccine used, and both history and follow-up.On a large scale, this is only feasible via linkage of accurate vaccine information to health administrative or electronic health record data.7 This effort will require coordination across jurisdictions and across data sources.Despite the massive scale of the vaccine effort, there still will be rare adverse events that will require very large sample sizes to rule out important associations, and any one country may have insufficient sample size, and/or insufficient variation in vaccine type, to detect differences.Distinguishing between real and spurious associations will likely require data on the worldwide scale.This will certainly require multidatabase efforts involving multiple countries and health care providers.Ensuring that data are standardized across jurisdictions, and analyses are coordinated across databases, will require input from statisticians throughout the process.Measurement and recording of outcomes, study design, assessment and control of confounding, and analysis to assess representativeness and transportability should all be considered.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.038 | 0.179 |
| Meta-epidemiology (narrow) | 0.002 | 0.003 |
| Meta-epidemiology (broad) | 0.004 | 0.004 |
| Bibliometrics | 0.002 | 0.003 |
| Science and technology studies | 0.006 | 0.015 |
| Scholarly communication | 0.007 | 0.017 |
| Open science | 0.011 | 0.003 |
| Research integrity | 0.090 | 0.126 |
| Insufficient payload (model declined to judge) | 0.008 | 0.010 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".