First you choose what type of desired specification you want to meet and what type of controller you want to use. What type of sensors are available to provide the feedback. For example if you want to have a state feedback type of controller, then you need to see how many states are avilable for feedback. In case you want a optimization based PID type of controller you should check if there is  any atucator constraints. If you go for a Sliding mode controller you need to know the upper bound of your parameter uncertainty. A reduced order model as suggested by  Mr. Igor Škrjanc will be a very good option to preceed.

Try to eliminate non dominent poles from the transfer function & then the reduced order transfer function of the plant can be connected to P,PI or PID controller and then to get desired controller output Kp KI or KD can be calculated by classical methods like pole placement and pole zero cancellation method.

As suggested by Deepak, after eliminating  the non dominant  pole you can try with optimal controller also.

model reduction methods are the best way to simplified the system to 2nd or 1st order and then using classical control methods like PID or LQR