This paper examines why a larger share of COVID-19 deaths occurs among young and middle-aged adults in developing countries than in high-income countries. Using novel data at the country, city, and patient levels, we investigate the drivers of this gap in terms of the key components of the standard Susceptible-Infected-Recovered framework. We obtain three main results. First, we show that the COVID-19 mortality age gap is not explained by younger susceptible populations in developing countries. Second, we provide indirect evidence that higher infection rates play a role, showing that variables linked to faster COVID-19 spread such as residential crowding and labor informality are correlated with younger mortality age profiles across cities. Third, we show that lower recovery rates in developing countries account for nearly all of the higher death shares among young adults, and for almost half of the higher death shares among middle-aged adults. Our evidence suggests that lower recovery rates in developing countries are driven by a higher prevalence of preexisting conditions that have been linked to more severe COVID-19 complications, and by more limited access to hospitals and intensive care units in some countries.

Those of us who came into the hospital service as junior doctors in the late 1940s were told that we could only expect apprentice conditions as our prospects when we became consultants were lavishly attractive. Many of us now have young families (long postponed, of course). We have lost a third of our salary in real terms over the past four years. The

Society of Anaesthesiologists' (ASA) grade system (ASA I, healthy patient; ASA II, mild systemic disease; ASA III, severe, nonincapacitating systemic disease; ASA IV, severe, incapacitating systemic disease which is a constant threat to life; and ASA V, moribund patient, not expected to live for 24 hours with or without surgery) abnormal values were found in 1-5% of patients in grade I, 17% in grade II, 35% in grade III, and 48% in grades IV and V.' In addition, most abnormal values in patients in grade II were minor and not clinically important. Therefore patients in grades I and II do not need routine estimations of urea and electrolyte concentrations whereas those in grades III, IV, and V do, as the results are much more likely to be abnormal.I therefore suggest that this grading system, which could be easily learnt by junior medical staff, would be more appropriate than age as an indicator of the need for biochemical screening preoperatively.