Balance measures for propensity score methods: a clinical example on beta‐agonist use and the risk of myocardial infarction