<p>Internet of Things (IoT) is a networking solution for providing smart services across many domains. Ever since the emergence of IoT, providing adequate security for its ecosystem has been a persistent challenge, thereby making IoT networks and infrastructures highly susceptible to various network anomalies and cyber-attacks. Despite the effectiveness of machine learning based intrusion detection systems (IDSs) in detecting network anomalies and cyber-attacks, they are vulnerable to concept drift phenomenon that degrades the accuracy of the model over time. Over the recent decade, several approaches have been proposed to address the effects of concept drift on anomaly detection methods. The aim of this paper is to provide a systematic review of concept drift approaches for anomaly detection in IoT networks. Thus, it sought to examine how stream learning has improved detection of cyber-attacks in real-time by reviewing recent studies that incorporate concept drift techniques for anomaly detection in IoT environment. The review process includes an extensive literature search of relevant scientific studies that satisfy the pre-defined inclusion criteria, follow by the analysis and thorough synthesis of findings from the selected literatures. The study examines and identifies the trending concept drift approaches including the detection methods and adaptation strategies in IoT environment. It also reveals the common datasets and evaluation methods employed in the field. The review implication for future research direction is highlighted in the study, hence the need for further studies to advance the security of IoT systems to achieve the long-time vision of smart computing for automatic service delivery.</p>