Inter-cell Interference (ICI) in the uplink of modern cellular communication systems with high frequency reuse is hard to predict, as fast scheduling and link adaptation lead to a high interference fluctuation. This fluctuation poses a big challenge for link adaptation algorithms that need an accurate SINR estimate to assign suitable modulation and coding schemes for transmission. To enable ICI prediction it has been proposed to exchange scheduling decisions amongst base stations. However, if this exchange is subject to delays, the system performance decreases as scheduling decisions become suboptimal. In this paper we propose a new scheme that maintains scheduling optimality at the cost of reduced link adaptation accuracy when faced with backhaul delays. We compare both schemes and provide insights into the multi-user diversity / link adaptation accuracy trade-off that arises for non-stationary users if information exchange on the backhaul is subject to delays.