The intensive care unit is a highly specialized and expensive hospital resource serving both emergency and scheduled patients. The vast majority of scheduled patients arrive from the operating theater. Therefore, the operating theater schedule has a strong impact on intensive care unit occupancy levels. Prior research focuses on the creation of a new master surgery schedule to optimize the patient flow in downstream units. In practice, however, the master surgery schedule affects a multitude of related processes, and changing it causes significant disruptions within the hospital. Hence, our approach emphasizes a centralized reallocation of scheduled surgeries while maintaining the existing master surgery schedule. We propose a mixed-integer quadratic model that optimizes the tactical surgery schedule to balance the expected day-to-day occupancy of scheduled patients in the surgical intensive care unit. Supported by two years of data from a German university hospital, we analyze three planning strategies and their impact on bed utilization in the intensive care unit. Our approach yields an improvement of 17.5% in intensive care bed utilization variability compared to a decentral approach, which is similar to current hospital practice. Additionally, we show that our approach can realize the majority of the improvement potential without the disruptions that derive from an entirely new master surgery schedule.