Scientific approaches

We have developed a new method called ‘genomic neighbor typing’ that can be used to quickly determine if bacteria are likely to be resistant to antibiotics, greatly aiding timely and effective treatment. Traditional techniques take too long to impact patient care, but this new method uses genetic information to identify resistance within minutes using real-time sequencing data. For example, it can accurately detect antibiotic resistance in common bacteria like Streptococcus pneumoniae and E. coli within 10 minutes to four hours. This approach can greatly improve how quickly and accurately doctors can treat infections and minimize the inappropriate use of antibiotics that will not be effective. The method also shows promise in rapidly determining whether two infections could be linked as part of the same outbreak, and research is ongoing in this space. 

Unexpected outbreaks of new and reemerging infectious diseases challenge our ability to detect and track them. Using genome sequences from bacteria and viruses, we can track who infects whom and identify super-spreading events or transmission clusters, and the locations where they are most likely to occur. Using “digital exhaust”—the traces ordinary people leave when they do internet searches, interact with social media, and use their mobile phones—we develop new tools to identify, track, and forecast disease outbreaks. We are also actively working to leverage new and developing sources of data, such as wastewater surveillance, to help enable rapid detection and response to novel threats.

The delay between when a person gets sick with an infection and when their case gets reported to health authorities can vary from hours with prompt testing and electronic reporting to, unfortunately, days or weeks in less-resourced settings. These delays threaten our ability to use case data to plan public health interventions. Nowcasting uses statistical models to overcome these delays by estimating the current case burden from available patterns of reporting, giving decision makers timely information on trends in an epidemic. We have contributed fundamental methods to this activity, providing models that gave local health authorities daily estimates of COVID-19 incidence in the U.S. early in the pandemic. This work helped decision makers understand the extent of unfolding COVID-19 outbreak when official case counts were still increasing exponentially. Other areas include developing nowcasting methods that use historical patterns of reporting delays to estimate actual monthly malaria cases. 

HIV presents unique challenges for surveillance because diagnoses are often missed or occur years after the infection, and patients remain living with HIV for the rest of their lives. We work to improve methods used to understand HIV transmission. This includes developing new mathematical models, statistical methods, and surveillance tools to characterize HIV epidemic trends, transmission dynamics and the demographic impacts of HIV, particularly in sub-Saharan Africa. In partnership with UNAIDS, WHO, and several collaborative research groups, we develop mathematical models and software that help public health officials make informed decisions about HIV prevention and care programs. 

We work with large datasets from population genomics, insurance claims, clinical diagnostics, wastewater, and other sources to design, interpret, optimize, and assess the effectiveness of disease surveillance strategies. We use dynamic models to simulate spread of SARS-CoV-2 in the population. This approach helps us understand what factors impact spread, and evaluate and improve the effectiveness of intervention strategies. For instance, we used models to project the course of COVID-19 over multiple years, inform vaccine prioritization, design prevention strategies for health care settings, and assess the impact of variants with enhanced transmissibility and/or partial immune escape on disease spread in the general population. 

We develop mathematical models to enhance the effectiveness of malaria elimination programs. In collaboration with partners, our model showed that introduction of a hormone that disrupts the mosquito’s ability to reproduce could be a tool as effective as insecticide treated nets against malaria-transmitting mosquitoes. We examine what might happen if we target specific parts of how malaria spreads, considering the complex details of malaria transmission. We integrate different aspects of malaria biology often ignored in models of malaria prevention programs.  

Causal inference methods provide a framework to understand how well vaccines and treatments work in settings like large outbreaks where it is impractical or unethical to do clinical trials for each question. We develop methods to test vaccines with low-cost study designs (such as the “test-negative” design), to evaluate how vaccines work when given after a person has already been exposed to infection (“post-exposure prophylaxis”), and to evaluate different protocols for treating infections with antibiotics. We are developing new approaches to improving antibiotic treatment with more rigorous studies to compare outcomes of different treatments. 

When a new infectious disease threat arises, every day counts in developing, testing, manufacturing, and distributing vaccines. The G7 and global vaccine funders have committed to the “100 days mission” to have vaccines available within 100 days of a new infectious disease emergency. CCDD is a global leader in developing new approaches to speed up the testing of vaccines, from defining novel clinical trial and observational study protocols to supporting pharmaceutical companies and public health practitioners in real-world studies of vaccines against  COVID-19, mpox, and other pathogens.

Genomic epidemiology for infectious disease uses pathogen sequences to understand the patterns of how they spread, and to characterize the lineages that present the most significant threat now and increasingly those that are emerging and likely to present a challenge for public health in the future. We have applied these methods to pathogens of many different types. In addition, we analyze deep sequencing data – genome sequences of tens to thousands of individual viruses or bacteria in an infected person — to study within-host variation in samples from observational and experimental settings. For instance, deep sequencing data enabled the detection of a super-spreading event of Mycobacterium tuberculosis in a large outbreak in a remote community and demonstrated the persistence of within-host variants across transmission steps in a controlled Citrobacter rodentium mouse model. These insights also inform the development of models to infer transmission links between sequenced samples. They were a key part of the evidence that showed transmission of the Delta variant of SARS-CoV-2 from vaccinated individuals, during a large outbreak in the summer of 2021.  

Bacterial vaccines often target some but not all types of bacteria within a species. The pneumococcal conjugate vaccines are a case in point, being highly effective against some of the most important strains while leaving others untouched. We have pioneered approaches to understand how the bacterial population will evolve in response to such vaccines, to inform the design and deployment of future interventions for maximum public health benefit.

Bacteria, unlike humans, have large variation in the genes they carry, even within a species. The sum of all genes present in members of a bacterial species is known as its pangenome, while genes present in some but not all members of a species are called “accessory genes.” We observed that each accessory genes in Streptococcus pneumoniae is present at similar frequencies in different populations across space and time. This led us to posit that these genes are experiencing frequency-dependent selection that could predict how the population would adapt following the introduction of vaccines. Ongoing work is refining this theory to predict and explain ongoing changes as new vaccines are introduced, and understand which accessory genes are driving this process.   

Pathogens evolve in response to the changing landscape they face, including the interventions we deploy to treat infections and protect public health. This can happen over time scales which are short by human standards, as shown by the ongoing evolution of SARS-CoV-2 as well as the dissemination of drug resistance in bacterial and other pathogens. Understanding how pathogens adapt and change is crucial to ensure that medical interventions remain effective and to know what we can expect from pathogens in the future. Our research improves understanding of how pathogens evolve, which helps reduce antimicrobial resistance, stay ahead of how pathogens evolve as they attempt to escape vaccines, and trace disease transmission more effectively. This work is key for both pandemic preparedness and a foundational part of epidemiological science.  

One reason malaria is hard to fight is the huge variety in the genes of the malaria parasite, Plasmodium falciparum. We study the processes that create genetic diversity in disease-causing germs and how this affects control efforts. We focus especially on malaria. We are creating new ways to analyze relationships between parasites’ genes and how they spread. This includes tools using family history type analysis (“Identity By Descent”) and other new methods. We are developing techniques to understand how the patterns of different malaria parasites relate to how people move and how intense malaria transmission is. 

The COVID-19 pandemic accelerated our efforts to turn the big data of electronic medical records into understanding of SARS-CoV-2 strain evolution, immune escape, and severity, vaccine effectiveness, and long-term outcomes. Detailed medical records provide data on a scale unachievable in clinical trials, but sophisticated methods are required to ensure that we distinguish true causal mechanisms from spurious correlations in data that arise in the real world, outside clinical trials. Often applied with colleagues from CAUSALab at Harvard Chan School, these methods include target trial emulation, use of negative controls, and innovative study designs to detect and correct bias to achieve reliable causal inference. 

The MIGHTE Lab at Northeastern University, in collaboration with CCDD, utilizes several novel data sources for infectious disease modeling and forecasting, including digital traces from the internet and mobile networks. Specifically, it analyzes Google search query data from the general public, clinicians’ searches (UpToDate), tweets, hospital visit records, and participatory surveillance data to track influenza hospitalizations trends in real-time. The MIGHTE lab also incorporates weather data, population mobility patterns derived from mobile phone data, and genomic sequencing information to discover disease spreading dynamics.

By combining these alternative data streams with traditional epidemiological reports, the models aim to generate timely estimates of sharp changes in disease activity for diseases that include influenza, dengue, malaria, Zika, and COVID-19, and other emerging infectious diseases. These efforts contribute to the field of digital epidemiology, an emerging research field that aims at harnessing novel internet-based data sources which were not originally conceived to track epidemiological events, to enhance infectious disease monitoring and prediction. 

During the Ebola outbreak of 2014 and the COVID-19 pandemic, numerous ethical questions arose about the conduct of clinical trials of vaccines. Many of the ethical issues raised depend in part on quantitative considerations: how soon, and how confidently, will various study designs answer the urgent questions that can improve public health for all. In Ebola we described the ethical case for individually randomized trials in terms of the benefits these can provide for public health while respecting ethical duties to participants. In COVID-19, we made an early case in favor of permitting human challenge studies in the effort to develop vaccines more rapidly and with greater confidence and weighed in on issues such as how to treat placebo participants in randomized trials after trial end. 

Our Center has focused quantitative efforts on understanding how COVID-19 infection and mortality rates were distributed across demographic groups in the early outbreak and projecting how that might change over time. We have also focused on projecting how different vaccine allocation plans would affect rates of infection and mortality in different demographic groups, and the implications for equity, and on defining measures to operationalize health equity efforts to ensure beneficial results, not only fair distribution of countermeasures (e.g. vaccines).