Confounding by Indication and Channeling over Time: The Risks of /VAgonists

A previously published nested case-control study, the Saskatchewan Asthma Epidemiologic Project (SAEP) spanning 1980-1987, investigated the risk of fatal or near fatal asthma and found different risks for two inhaled ^-agonists, fenoterol and salbutamol. The authors assessed whether this companson was confounded by indication because of channeling of inhaled fenoterol to more severely afflicted asthmatics. Using three subcohorts selected from a cohort of 12,301 asthmatics assembled from the computerized databases of Saskatchewan Health and followed over 7 years, the authors studied two forms of channeling and investigated whether greater asthma severity and less well-controlled disease were associated with preferential prescribing of a first prescnption of inhaled fenoterol, as opposed to inhaled salbutamol, and whether they were associated with the likelihood of a switch from inhaled salbutamol to fenoterol as well as a switch from inhaled fenoterol to salbutamol. The authors found that the initial choice between fenoterol and salbutamol was independent of the severity of the asthma and disease control, but that preferential prescribing of fenoterol occurred among users of salbutamol who showed signs of increased seventy or uncontrolled asthma. The switch from inhaled fenoterol to salbutamol was, however, minimally related to asthma severity. They conclude that the comparison between inhaled fenoterol and salbutamol in the SAEP may have been biased by indication. This study demonstrates that long-term information on medication use is essential to ensure that the results of such case-control studies are not biased by indication. Am J Epidemiol 1996;144:1161-9.