In this paper we consider distributed cognitive radio networks (CRNs) with channel aggregation (CA) scheme where there are multiple wireless channels and multiple secondary users (SUs). SUs in CRNs with CA can transmit bits of data through more than one idle channels which are not used by primary users (PUs). SUs decide whether to access idle channels or not based on the access probability (AP), and we call users who decide to access active SUs. Each active SU uses the sequential sensing policy to sense randomly chosen L channels at the beginning of each access frame (AF) and accesses all the idle channels among the sensed channels during the rest of the AF. We analyze the average number of bits transmitted successfully by an SU per AF, call it the throughput performance of an SU. We derive a closed form of the throughput performance of an SU as a function of AP and L. By maximizing throughput, we propose a joint optimal access and sensing policy, that is, we provide the optimal value of AP a* and the optimal number of sensing channels (NSC) L*. We validate our analysis and investigate the throughput performance of the proposed optimal policy using numerical and simulation studies. Based on our investigation we further propose a near-optimal simplified policy.