My research interests lie principally, but not exclusively in the field of epidemiologic methods aiming to assess the determinants of social inequalities in population health outcomes and comparative effectiveness research with a specific interest in longitudinal analysis, causal inference, repeated measures, and translational epidemiology. At UCT, I used marginal structural models applied to large longitudinal data from Khayelitsha (HIV-Cohort) to assess the effectiveness of an observational, nonrandomized intervention. At Harvard, I used fixed effects methods in the context of the analysis of the components of the variance and within siblings design (observational cross-over) to evaluate the effect of a small fetoplacental ratio at birth on the risk of delivering a small for gestational age infant. Recently, using multilevel analysis, I studied the contextual effect of regional unemployment on stillbirth by geographical regions in Spain as a complement to previous studies where I evaluated the multiplicative effect of maternal education and ethnicity on the risk of delivering a stillborn.

Currently, I am studying the impact of the 2009 economic crisis on stillbirth rates among African immigrant women in Spain and evaluating the best framework to extract cancer patients comorbidity information from population-based administrative records. Also, I am developing in collaboration with colleagues from the Cancer Survival Group data-adaptive methods for model selection and evaluation based on cross-validation techniques (cvAUROC) and applying advanced causal inference methods such as targeted maximum likelihood estimation (TMLE) to study cancer outcomes.

Recently, I have programmed the implementation of TMLE for Stata statistical software users, named ELTMLE and divulgated it at the Stata Users Group Meeting in London, 2017. Together with collaborators from the Cancer Survival Group, we proposed a structural framework for population-based cancer epidemiology and evaluate the performance of double-robust estimators for a binary exposure in cancer mortality. In simulations studies, we have demonstrated that TMLE shows the best bias-variance trade-off, more precise estimates, and appropriate 95% confidence interval coverage than its competitors, supporting the use of the data-adaptive model selection strategies based on machine-learning algorithms. We applied TMLE to estimate adjusted one-year mortality risk differences in 183,426 lung cancer patients diagnosed after admittance to an emergency department versus non-emergency cancer diagnosis in England, 2006–2013. The adjusted mortality risk (for patients diagnosed with lung cancer after admittance to an emergency department) was 16% higher in men and 18% higher in women, suggesting the importance of interventions targeting early detection of lung cancer signs and symptoms.