Misinformation is the spread of false information, especially on social media, regardless of whether there is intent to mislead. Disinformation, fake news and deepfakes are a subset of misinformation. Disinformation is deliberately misleading or manipulating facts. Fake news is purposefully crafted, totally fabricated information that mimics the form of mainstream news. Deepfakes are convincingly altering an image or video and manipulating it to misrepresent someone as doing or saying that was not done or said. Misinformation is distributed via popular social media networks like YouTube, Twitter, Facebook, WhatsApp and Instagram. The spurious information can be related to health (e.g., COVID Vaccines), politics, political party, propaganda, celebrities, conspiracies, news, religion, discrimination, commercial products, etc. In recent years, researchers have used many different machine learning and deep learning multimodal techniques to identify misinformation based on the content format and using additional data like user profiles and community groups. In addition, social media platforms also introduced many platform intervention policies, e.g., Facebook and YouTube. With this understanding, more than 200 relevant papers are reviewed to study various misinformation detection methods and the datasets used for evaluation. The findings of this meta-analysis can be used not only to depict the current mitigation methods discussed in the literature but also to prescribe appropriate method recommendations based on the misinformation domain and the content form on social media. For example, multimodal analysis can leverage the listed methods that classify text, image, audio and video.