Is the association between inhaled beta-agonist use and life-threatening asthma because of confounding by severity?

We have previously reported an increasing dose-response relationship between the regular use of beta-agonist inhalers and the risk of asthma death and near death among a cohort of 12,301 subjects who had been dispensed 10 or more prescriptions of asthma drugs from January 1980 to April 1987. That analysis was based solely on information obtained from linkable computerized data bases. Such an association might be explained in part by the tendency of patients with more severe asthma, that is, those at greatest risk for an adverse outcome, to use more beta-agonist medication. To further examine this potential confounding by severity, we gathered clinical information independently from the field on the 129 case patients and their 655 control patients from the matched case-control analysis of 12,301 subjects. In 68% of the control patients with a life-threatening episode and 75% of the matched control subjects, we obtained a valid questionnaire from at least one physician who had seen the patient during the previous 2 yr. Acceptable information on hospitalizations because of asthma was obtained in 87% of those hospitalized. Clinical features associated with an increased risk of fatal and near-fatal asthma were: a history of loss of consciousness or seizures during a previous asthma attack (odds ratio, 10.2; 95% Cl, 3.9 to 26.7), a history of attacks of asthma precipitated by eating certain foods (odds ratio, 5.1; 95% Cl, 2.4 to 11.1), a clinical score designed to reflect the severity of prior attacks of asthma leading to hospitalization, and prior respiratory acidosis among those in whom a blood gas determination was recorded. After adjustment for these clinical markers of asthma severity, there was no significant reduction in the odds ratio for asthma death and near death per metered-dose inhaler of fenoterol (e.g., odds ratio, 2.5; 95% Cl, 1.7 to 3.8) or albuterol (e.g., odds ratio, 2.0; 95% Cl, 1.5 to 2.7). We conclude that this further attempt at adjusting for confounding by severity did not alter the relationship between the use of beta-agonist inhalers by subjects outside of hospital and the risk of asthma death and near death.