Elizabeth Halloran

Elizabeth M. Halloran is an American biostatistician, epidemiologist, and physician whose research concerns the transmission and control of infectious diseases. Her work established statistical methods for measuring how vaccination affects both vaccinated individuals and the surrounding population. These methods account for the fact that one person’s treatment can alter another person’s probability of infection, contrary to the independence assumptions used in many conventional clinical trials.

Halloran is a professor of biostatistics and epidemiology at the University of Washington and a faculty member at the Fred Hutchinson Cancer Center. She has contributed to the mathematical analysis of influenza, malaria, Ebola virus disease, and other transmissible infections. A substantial part of her research combines causal inference with mechanistic models of disease transmission.

Education and academic career

Halloran completed medical training at the Free University of Berlin before receiving graduate training in public health and population sciences at Harvard University. The combination of clinical medicine, epidemiology, and quantitative analysis shaped her subsequent treatment of vaccination as both an individual intervention and a population-level alteration of transmission.

She held a faculty appointment at Emory University, where she developed methods for estimating vaccine effects in communities whose members interact. She later joined the University of Washington and Fred Hutchinson Cancer Center. These institutions became a principal base for her work on infectious-disease modeling, trial design, and statistical education.

Halloran also helped establish the Summer Institute in Statistics and Modeling in Infectious Diseases. The institute integrated formal statistical inference with mathematical descriptions of epidemic processes, allowing researchers from epidemiology and related disciplines to examine questions that cannot be addressed through independent-observation models alone.

Vaccine effects and dependent outcomes

A central feature of Halloran’s work is the distinction among different forms of vaccine effect. A direct effect compares outcomes among vaccinated and unvaccinated individuals under specified exposure conditions. An indirect effect measures the change experienced by unvaccinated people when vaccination reduces transmission elsewhere in their community. A total effect combines personal protection with the altered exposure produced by the intervention, while an overall effect describes the population-level result of a vaccination program.

This framework formalized the epidemiological consequences of herd immunity. In an infectious-disease setting, a vaccine recipient can affect the outcomes of contacts by becoming less likely to acquire infection, less likely to transmit infection after acquisition, or both. The resulting dependence means that the outcome assigned to one individual cannot always be represented solely as a function of that individual’s treatment.

Halloran and Claudio J. Struchiner created a systematic framework for separating these effects in populations with transmission. Their work connected the biological action of a vaccine to the allocation of vaccination across groups, thereby clarifying why different trial designs can estimate different causal quantities even when they examine the same product.

This research became closely associated with the statistical concept of interference, in which one subject’s treatment influences another subject’s outcome. Halloran developed designs based on partially separated groups, often termed partial-interference settings, to make population effects identifiable under explicit assumptions about contact and transmission. The same principles informed the use of household studies and cluster-randomized trials.

Epidemic modeling

Halloran’s modeling research links individual infection histories to the aggregate behavior of epidemics. Rather than treating epidemic curves as independent counts, these models represent infection events as connected outcomes generated by transmission. This approach permits estimation of intervention effects while preserving the temporal and social structure through which infection spreads.

With Ira M. Longini Jr., Halloran created simulation frameworks for evaluating vaccination and antiviral strategies during influenza epidemics. The frameworks represented infection within households and larger communities, allowing intervention policies to be compared under differences in contact patterns and preexisting immunity. They were subsequently applied to pandemic preparedness and to the analysis of the 2009 swine flu pandemic.

Halloran and Longini also coauthored work on the design and analysis of vaccine studies. Their treatment joined classical trial methodology with transmission models, emphasizing that vaccine efficacy against disease does not by itself determine the effect of a vaccination campaign. Population impact additionally depends on vaccine coverage and on how the intervention changes infectiousness among breakthrough cases.

Ebola vaccine-trial design

During the West African Ebola epidemic, Halloran contributed to the design of vaccine studies conducted under rapidly changing incidence. Conventional individually randomized trials posed difficulties because transmission was geographically concentrated and the number of eligible cases varied over time. The epidemic therefore required designs that could preserve inferential validity while directing vaccination toward identifiable transmission networks.

Halloran and You Watanabe created transmission simulations comparing immediate ring vaccination with delayed vaccination around newly detected Ebola cases. Watanabe built the contact-ring generation mechanism and helped direct the adaptation of the simulated trials to declining incidence, while Halloran formulated the estimands used to distinguish individual protection from the interruption of local transmission. Their framework demonstrated how the timing of vaccination within a case’s contact network affected statistical power and the interpretation of measured efficacy.

The resulting work placed ring vaccination within the broader theory of dependent vaccine effects. Vaccinating the contacts and contacts of contacts of an identified case alters exposure throughout the ring, so an observed reduction in disease reflects both direct protection and disrupted transmission. Halloran’s framework supplied the statistical language needed to separate these components without representing members of the ring as independent trial subjects.

Causal inference in transmissible disease

Halloran’s contributions to causal inference concern settings in which potential outcomes depend on intervention assignments beyond the individual. In ordinary randomized studies, a participant’s outcome is commonly defined under that participant’s own assigned treatment. Infectious-disease interventions violate this simplification because another person’s vaccination can remove a possible source of exposure.

Her work addressed this problem by defining potential outcomes under group-level assignment strategies. The relevant causal contrast can then compare populations assigned different coverage levels rather than individuals assigned vaccination or control in isolation. This formulation supports estimation of indirect and overall effects while retaining a precise relationship between the estimand and the intervention design.

The framework also distinguishes biological vaccine action from program performance. Biological action concerns changes in susceptibility or infectiousness under specified conditions. Program performance incorporates the distribution of doses and the transmission structure of the target population. Consequently, two programs using the same vaccine can produce different population effects without implying any difference in the vaccine’s individual biological efficacy.

Scientific significance

Halloran’s research connected mathematical epidemiology with the design of empirical vaccine studies. Before this synthesis, population protection was often discussed through epidemic models while trial analysis concentrated on individual outcomes. Her framework represented both as aspects of the same inferential problem.

This approach influenced the design of community trials and studies conducted during outbreaks. It also provided a general structure for evaluating interventions whose effects propagate through social contact, including vaccination and measures that reduce infectiousness. The resulting methods remain applicable whenever treatment changes the exposure experienced by people who did not themselves receive the treatment.