Concentrating on a class of hybrid discrete-time filtering problems that are modulated by a Markov chain, this work aims to reduce the complexity of the underlying problems. Since the Markov chain has a large state space, the solution of the problem relies on solving a large number of filtering equations. Exploiting the hierarchical structure of the system, it is noted that the transition probability matrix of the Markov chain can be viewed as a nearly decomposable one. It is shown that a reduced system of filtering equations can be obtained by aggregating the states of each recurrent class into one state. Extensions to inclusion of transient states and nonstationary cases are also treated.