Non-orthogonal multiple access (NOMA) is expected to provide high spectral efficiency (SE) and massive connectivity in future wireless networks. On the other side, backscatter communications (BC) is an emerging technology towards battery-free transmission in future wireless networks by leveraging ambient radio frequency (RF) waves that enable communications among wireless devices. This paper proposes a new optimization framework to maximize the SE of the NOMA-BC network. In particular, we simultaneously optimize the transmit power of the base station and reflection coefficient of the backscatter device in each cell under the assumption of imperfect decoding of successive interference cancellation. The SE optimization problem is coupled on multiple variables which makes it very difficult to solve. Thus, we apply a decomposition method and KKT conditions to get an efficient solution. Simulation results demonstrate the superiority of the proposed NOMA-BC framework over the benchmark NOMA without BC framework.