Bad smells are symptoms that something may be wrong in the information system design or source code. Although bad smells have been widely studied, we still lack an in-deep analysis about how they appear more or less frequently in specific information systems domains. The frequency of bad smells in a domain of information systems can be useful, for instance, to allow software developers to focus on the more relevant bad smells of a certain domain. Moreover, developers of new bad smell detection tools could take information about domains into consideration to improve the tool detection rates. In this paper, we investigate code smells more likely to appear in four specific information systems domains: accounting, e-commerce, health, and restaurant. Our analysis relies on 52 information systems mined from GitHub. We identified bad smells with two detection tools, PMD and JDeodorant. Our findings suggest that Comments is a domain-independent bad smell since they uniformly appear in all investigated domains. On the other hand, Large Class and Long Method can be considered domain-sensitive bad smells since they appear more frequently in accounting systems. Although less frequent in general, Long Parameter List and Switch Statements also appear more in health and e-commerce systems, respectively, than in other domains.