A composite event-triggered model predictive control (MPC) framework is designed for constrained discrete systems affected by external disturbances with slow variations. The controller has two parts. The disturbance compensation controller part is constructed to suppress the dominant of disturbances, and the model predictive controller part with constraint tightening method is designed to calculate the optimal input as well as deal with residual disturbances. The event-triggered mechanism is designed to decrease the frequency for solving the optimization, and the novel triggering condition is proposed to avoid the failure of triggering condition caused by the discontinuity of measurement in discrete systems. The feasibility and stability of overall systems are analyzed.