Social platforms are the predominant media through which people communicate, share opinions, and freely express themselves nowadays. Unfortunately, they are also a hub for the distribution of harmful content generated by users. Online harm spreaders comprise a multiplicity of users acting online with malicious intentions. This includes disseminating misinformation, propagating rumors or conspiracy theories, as well as making hateful comments targeting individuals or groups. Most works in the literature have addressed the problem of distinguishing a given type of harm spreader (e.g., fake news spreaders or hate speech spreaders) from regular users. In this work, a study intended to characterize users under the umbrella of online harm spreaders across multiple behavioral dimensions is presented. The statistical distribution in the values of several features in each dimension were observed and compared in order to determine which are the common traits and salient characteristics of online harm spreaders. Findings demonstrate that dimensions such as sentiments, emotions and some writing styles are differentiating factors and can serve as basis for more accurate general-purpose classification approaches for detecting spreaders.