Essays on cognitive development and medical care

This thesis consists of four self-contained papers:

Essay I

with Linuz Aggeborn

Fluoridation of the drinking water is a public policy whose aim is to improve dental health. Although the evidence is clear that fluoride is good for dental health, concerns have been raised regarding potential negative effects on cognitive development. We study the effects of fluoride exposure through the drinking water in early life on cognitive and non-cognitive ability, education and labor market outcomes in a large-scale setting. We use a rich Swedish register dataset for the cohorts born 1985-1992, together with drinking water fluoride data. To estimate the effects, we exploit intra-municipality variation of fluoride, stemming from an exogenous variation in the bedrock. First, we investigate and confirm the long-established positive relationship between fluoride and dental health. Second, we find precisely estimated zero-effects on cognitive ability, non-cognitive ability and education for fluoride levels below 1.5 mg/l. Third, we find evidence that fluoride improves later labor market outcomes, which indicates that good dental health is a positive factor on the labor market.

Essay II

I study the associations between cognitive and non-cognitive abilities and mortality using a population-wide dataset of almost 700,000 Swedish men born between 1950 and 1965. The abilities were measured at the Swedish military enlistment at age 18-20. In addition, I investigate if income and education are good proxies for the abilities. The results suggest that both cognitive and non-cognitive abilities are strongly associated with mortality, but that noncognitive ability is a stronger predictor. The associations are only partly mediated through income and education. For middle and high income earners and individuals with a college education there are no associations with mortality. However, for low income earners and individuals without a college education, both abilities are strongly associated with mortality.The associations are mainly driven by the bottom of the distributions.

Essay III

with Matz Dahlberg, Kevin Mani and Anders Wanhainen

We examine how health information affects individuals' well-being using a regression discontinuity design on data from a screening program for an asymptomatic disease, abdominal aortic aneurysm (AAA). The information provided to the individuals is guided by the measured aorta size and its relation to pre-determined levels. When comparing individuals that receive information that they are healthy with those that receive information that they are in the risk zone for AAA, we find no effects. However, when comparing those that receive information that they have a small AAA, and will be under increased surveillance, with those who receive information that they are in the risk zone, we find a weak positive effect on well-being. This indicates that the positive information about increased surveillance may outweigh the negative information about worse health.

Essay IV

I estimate the effect of SSRI antidepressants on the risk of mortality for myocardial infarction (MI) patients using Propensity Score Matching on individual health variables such as pharmaceutical drug prescription, patient history and severity of the MI. The effect of antidepressants on mortality is a heavily debated topic. MI patients have an elevated risk of developing depression, and antidepressants are among the most common treatments for depression and anxiety. However, there are indications that some classes of antidepressants may have drug-induced cardiovascular effects and could be harmful for individuals with heart problems, but there is a lack of large-scale studies using credible identification strategies. My findings indicate no increased risk of two-year mortality for MI patients using SSRI. The results are stable for several specifications and robustness checks.