With the rapid growth of internet of thing (IoT) wireless devices, cooperative spectrum sensing (CSS) has emerged as a promising solution to leverage the spatial diversity of multiple IoT sensing nodes (SNs) for spectrum availability. However, the cooperative paradigm also incurs increased cooperative costs between each SN and the fusion center (FC), leading to decreased cooperative efficiency and achievable throughput, especially in large-scale cognitive IoTs (CIoTs). To address these challenges, we present a sequential detection with feedback information (SD-FI) approach in this paper. To achieve this objective, we propose a two-way CSS model that formulates an optimization problem of Bayes cost in a quickest detection framework with feedback. To solve this optimization problem, we derive the structure of the optimal local decision rule from the local decision function and determine the optimal detection threshold in conjunction with the cost function. Following the optimal threshold pair, we implement the optimal SD-FI and theoretically demonstrate the uniqueness of the optimal threshold and optimal sensing time. Simulation results demonstrate superiority of our proposed approach in terms of cooperative performance (i.e., detection performance and Bayes cost) and sample size. Notably, even with limited sensing time, our proposed SD-FI exhibits high throughput, highlighting its effectiveness in enhancing spectrum availability and utilization in cognitive IoTs.