Software design is a key concern in code review through which developers actively discuss and improve each software change. Nevertheless, code review is a collaborative task influenced by technical and social aspects. Consequently, these aspects can play a key role in how software design degrades. They can contribute to accelerating or reversing design degradation during the process of each single change s review. However, there is little understanding about: (i) the impact of code review and their practices on design degradation over time; and (ii) to what extent social and technical aspects are related to the reduction or increase of design degradation.We addressed these limitations driven by two goals. Our first goal is to provide a characterization of how modern code reviews impact design degradation during software maintenance. We consider various technical and socials aspects to study the code reviews impact. Our second goal is to explore the role of technical and social aspects in distinguishing and predicting (un)impactful design changes during code reviews, using machine learning techniques. A design change is considered impactful when it varies the density and diversity of degradation symptoms (i.e., code smells). These goals were addressed with two empirical studies. Our first study reports a characterization of the impact of code reviews on the evolution of degradation symptoms along each code review. We also took into consideration when there was an explicit goal or discussions around software design. Our second study reports an analysis of technical and social aspects influencing changes along all the revisions within a single code review. Then, we could observe the role of social aspects in distinguishing and predicting design impactful changes. Our results show that the majority of code reviews have little or no impact on degradation, even with explicit design discussions. Long discussions and a high rate of reviewers disagreement increase the risk of degradation. Both social and technical aspects are able to distinguish design (un)impactful changes. In summary, our results provided us with a better understanding of influential aspects that help us in deriving guidelines to mitigate degradation during code reviews. Our results also provide insights to design a new code review tool salso able to warn developers early about the harmful design impact along code reviews.