In the chapters of this thesis, I empirically investigate four distinct research questions, spanning from the detrimental impact of very early shocks - at birth and even before birth - on children medium-term health, to the mediation role of social interactions in altering the effect of weather conditions on the spread of the Sars-CoV-2, and to ED nurses' leniency in assigning priority and the impact of waiting time on health. Although each chapter focuses on different questions, they all investigate different instances of the (unintended) consequences of individuals' actions on others - with a focus on vulnerable categories, and they all share an approach oriented at unveiling policyrelevant causal relations in natural experiment settings. In the first chapter, joint with Simone Ferro and Alessandro Palma, I investigate the health effects of a relatively higher prenatal exposure to Particulate Matter. Although the detrimental effects of air pollution on health are already well documented, in our study, we try to innovate on two relevant aspects. First, by matching birth certificates to individual consumption of subsidised pharmaceutical and hospitalisation records of Tuscany, our data allows going beyond the effects detected at birth to investigate for the first time medium-term morbidity at the individual level. Second, by documenting a permanent health loss associated with higher in-utero exposure to air pollution in a setting characterised by non-extreme levels of air pollution, our findings challenge the existing consensus on whether such levels of concentration of PM should be considered healthy. In the second chapter, I first show that Emergency Department patients arriving just after a shift change have a substantially lower probability of being assigned a higher priority with respect to patients arriving just before a shift change. As such distortion in the assignment of priority will only affect them by altering their relative position in the waiting list, I employ this quasi-experimental variation in the assigned priority to investigate the immediate and the longer-term consequences of waiting time on visits' outcomes and patients' future demand for emergency health care. I find that in the short-run, patients who are assigned a lower priority (and thus who presumably wait less time before being examined by a physician) are more likely to leave the ED before being visited. Furthermore, in the longer run, they are significantly less likely to return to the ED with the same condition. In the third chapter, Simone Ferro and I investigate the role of weather conditions on the spread of the Sars-CoV-2 in the US while taking into account the mediation role of social interactions. Our findings suggest that the endogenous response of social interaction to weather conditions play a major role in shaping the overall total effect of weather on the spread of the virus. This may contribute to rationalise the paradox between the results of laboratory experiments showing that the virus is susceptible to temperature, and the fact that the seasonal increases in temperature did not slow down the spread of the virus. In the fourth and final chapter, jointly with Gabriel Facchini and Matilde Machado, I investigate the relationship between being born by cesarean section and health outcomes for children. Such relation is confounded by different dimensions of selection in the choice of birth mode and to overcome the issue we employ an individual specific, purely exogenous measure of crowding at the precise time of admission at the maternity ward. We find that higher crowding leads to a significant decrease in the total probability of cesarean section. This effect is larger for patients with a higher ex-ante probability of delivering via cesarean section. We then use such relation to identify a causal relationship between the delivery method and the consumption of drugs for children up to age three. We find a positive and significant effect on the consumption of antiinfectives (antibiotics in particular) from the first year of life on - both looking at the quantity and at an indicator for abnormal use of this type of pharmaceuticals. This points to an adverse effect on children health, which is in line with the medical literature results on the importance of delivery method on the development of the immune system.