Spillover effects

Spillover effects are the impacts interventions have on non-recipients that are connected to recipients through social or geographic proximity (i.e., “herd effects”, “indirect effects”). Interventions that produce spillover effects have a greater population-level health impact and higher cost-effectiveness than those that do not and thus may be more promising for infectious disease control and elimination. While spillover effects of vaccines have been estimated for decades, we have spearheaded research to extend methods of spillover estimation to other types of interventions. To learn more, see our systematic review and methods synthesis in the International Journal of Epidemiology and our article on spillover effects of WASH interventions in American Journal of Epidemiology.


In 2019, the Lancet Commission on malaria eradication called for a global effort to eradicate malaria worldwide by 2050, and several countries in southern Africa have set goals to eliminate malaria locally by 2030. In settings approaching malaria elimination, transmission is low, and cases cluster in geographic “hot spots”. We are investigating the effectiveness of reactive, focal chemotherapy and vector control interventions delivered in the neighborhood around malaria cases (K01AI141616). The study will analyze data from three randomized trials in Namibia, Eswatini, and Zambia.

In addition, with support from the Stanford Center for Innovation in Global Health, we are investigating whether intermittent preventive treatment for malaria in pregnancy influences child growth in early life. We will use causal mediation analyses to investigate pathways through which the intervention does or does not influence child growth. This research is conducted in partnership with the Infectious Diseases Research Collaboration in Kampala, Uganda (details here).

Household flooring

Recent water, sanitation, and hygiene (WASH) intervention trials found modest or no impacts on soil-transmitted helminth infections or diarrhea, suggesting that interventions did not block all transmission pathways. For example, there was evidence that WASH interventions did not interrupt transmission pathways through soil. In low-resource settings, many household floors are made of compacted soil. Children’s high levels of exposure to and consumption of soil may increase their risk of soil-transmitted helminth infections and other enteric pathogen infections. We and others have led observational studies that found that children in households with concrete floors have a lower prevalence of diarrhea and soil-transmitted helminth infection than those in households with soil floors (e.g., here, here). However, these findings may be strongly confounded by household wealth. In collaboration with the International Centre for Diarrhoeal Disease Research, we are conducting a randomized trial to rigorously measure whether installing concrete floors in homes with soil floors reduces household fecal contamination, child soil-transmitted helminth infection, and child diarrhea in a low-resource in rural Bangladesh (R01HD108196; trial registration here). With support from the Stanford University Woods Institute, we are investigating alternative “green” cements that are produced with fewer greenhouse gas emissions than traditional cement and that may produce the same health benefits (details here).

Climate change and health

Climate change will lead to other profound shifts in the environment that will likely influence transmission of infectious diseases. We are investigating the association between climatic and environmental variables and diarrhea, soil-transmitted helminth infection, and other enteric pathogen infections among young children in rural Bangladesh. In addition, we have projected the potential change in prevalence in these diseases under possible climate change scenarios. See preprints here and here.

Soil-transmitted helminths

Though a large body of observational evidence suggests that WASH interventions can reduce soil-transmitted helminth (STH) incidence, few randomized trials have evaluated this question, and even fewer studies have used cutting edge diagnostics with high sensitivity and specificity. We have conducted both observational and randomized studies in Bangladesh to investigate whether water, sanitation, handwashing (WASH), and nutrition interventions reduce STH infections. In addition, we compared the performance of a new molecular STH diagnostic to the traditional microscopy-based approach. More recently, we have investigated whether finished household flooring (e.g., cement floors) is associated with lower STH and Giardia infection in young children in rural Bangladesh. To learn more, see our papers in PLoS NTDs papers on WASH and diagnostics and our Lancet Global Health paper on flooring.

Child growth

Children’s growth trajectories in early life influence their long-term health and development. In particular, stunting, a form of linear growth faltering, is associated with increased risk of illness, impaired cognitive development, and mortality. We conducted individual participant meta-analyses of over 35 longitudinal cohort studies of child growth from 15 low- and middle-income countries from 1990 to 2015. We found that the majority of stunting onset occurs from birth through 3 months of age, with lower incidence at later ages. Our findings suggest that prenatal, early postnatal, and possibly pre-conception interventions are needed to prevent most linear growth faltering. To learn more, see our preprints on stunting, wasting, and risk factors associated with child growth faltering.


School-located influenza vaccination programs have the potential to increase influenza vaccination among school-age children, who are responsible for the majority of influenza transmission. We are conducting an evaluation of Shoo the Flu, a city-wide program offering free influenza vaccination to elementary schoolchildren in Oakland, California since 2014. This evaluation is measuring the impact of Shoo the Flu on influenza vaccination coverage, influenza illness, influenza hospitalization, and school absenteeism. In addition to estimating impacts of the program on schoolchildren, we are measuring community-wide spillover effects. To learn more, see our papers in PLoS Medicine, Vaccine, and our medRxiv preprint.


Our research leverages data science tools to increase transparency and reproducibility in epidemiology. We regularly publish analysis datasets and replication scripts. The data science best practices for transparency and reproducibility are summarized in a lab manual, which our team is continuously updating. In addition, we have developed an internal replication process that increases reproducibility by identifying and resolving errors prior to publication.