When a health shock hits a location, the healthcare infrastructure needs to be adjusted to meet the increased demand. This may be a challenge in developing countries because of limited hospital capacity. In this study, we examine the consequences of health shocks induced by air pollution in a megacity in the developing world: Sao Paulo, Brazil. Using daily data from 2015-2017, and an instrumental variable approach based on wind speed, we provide evidence that exposure to particulate matter (PM10) causes an increase in pediatric hospitalizations for respiratory diseases, which in turn leads to a decrease in hospital admissions for elective care – phimosis surgery and epilepsy-related procedures such as video-EEG (electroencephalograph) monitoring. Importantly, emergency procedures such as appendectomy and bone fracture repair are not affected. While strained Sao Paulo hospitals seem to absorb the increased demand induced by poor air quality, our results imply that the common practice of using health outcomes unrelated to pollution as "placebo tests" in studies on the effects of air pollution might be inadequate in settings with limited healthcare infrastructure. This is often the case in developing countries, where severe pollution is also ubiquitous, but also happens in deprived areas in the developed world.