Spatial co-location patterns reveal associations among spatial features and serve as critical tools for discovering and understanding complex spatial phenomena. However, existing studies overlook the adverse effects of isolated instances on prevalence measures and fail to capture asymmetric combinatorial dependencies between spatial features, which results in the omission of valuable co-location information. To solve these problems, we propose a revised prevalence measure called correlated participation index (CPI). It reduces the influence of isolated instances while preserving the anti-monotonicity property. In addition, we propose the concept of united co-location pattern (UCP) and define distinct prevalence types to represent hierarchical and directional combinations of features, which enrich the semantic expressiveness of co-location patterns. We further develop a united co-location mining algorithm based on CPI, called UCPM-CPI, and design pruning strategies to optimize its computational efficiency. Extensive experiments on real-world datasets demonstrate the effectiveness and efficiency of the proposed algorithm.