The employment of mobile devices as the data consumption node in the medical domain (known as mHealth) is gaining widespread adoption since mobile devices facilitate remote and ubiquitous access to medical data. Today, it is a common phenomenon to see medical practitioners who own multiple mobile devices such as smart phones and tablets and expect to experience application consistency across the multiple devices. However, this expectation is hampered by the fact that mobile devices rely on wireless communication mediums which can experience sporadic disconnections. What is even challenging is the presence of the CAP theorem which states that considering the following three properties of a distributed system: consistency, availability, and partition tolerance, only two of the properties can be achieved simultaneously. In an ongoing research collaboration with the Geriatrics Ward at the City Hospital, Saskatoon, Canada, we deployed a reliable mHealth architecture that enables healthcare practitioners to use their n-mobile devices to access medical records. We proposed a brokerage platform that synchronizes the medical data on the multiple devices with careful consideration to the CAP theorem. Our proposed mHealth architecture is evaluated and the result in the real-world shows high support for scalability, real-time medical data propagation, and high capacity offline storage.