Spatial approximations simplify the geometric shape of complex spatial objects. Hence, they have been employed to alleviate the evaluation of costly computational geometric algorithms when processing spatial queries. For instance, spatial index structures employ them to organize spatial objects in tree structures (e.g., the R-tree). We report experiments considering two real datasets composed of ∼1.5 million regions and ∼2.7 million lines. The experiments confirm the performance benefits of spatial approximations and spatial index structures. However, we also identify that a second processing step is needed to deliver the final answer and often requires higher processing time than the step that uses index structures only. It leads to the interest in studying how spatial approximations can be efficiently used to improve both steps. This paper presents a systematic review on this topic. As a result, we provide an overview and comparison of existing approaches that propose, evaluate, or make use of spatial approximations to optimize the performance of spatial queries. Further, we characterize them and discuss some future trends.