We consider an epidemic model for the dynamics of an infectious disease that incorporates a nonlinear function h.I/, which describes the recovery rate of infectious individuals. We show that in spite of the simple structure of the model, a backward bifurcation may occur if the recovery rate h.I/ decreases and the velocity of the recovery rate dh.0/ dI is below a threshold value in the beginning of the epidemic. These functions would represent a weak reaction or slow treatment measures because, for instance, of limited allocation of resources o sparsely distributed populations. This includes commonly used functionals, as the monotone saturating Michaelis-Menten, and non monotone recovery rates, used to represent a recovery rate limited by the increasing number of infected individuals. We are especially interested in control policies that can lead to recovery functions that avoid backward bifurcation. Figure 4. Backward bifurcation with recruitment rate and death rate proportional to the population size, for two different treatments.. Case A) N D 2, 0.7644987 <ˇ< 0.85, D 0.2,˛D 0.00001, r D 1.2, v D 0.8, w D 5. Case A') N D 2, 1.475 <ˇ< 2.2125, D 0.2,˛D 0.00001, r D 2.2, v D 0.8, w D 5. Case B) N D 0.6, 0.0022 <ˇ< 0.005, D 0.001, r D 0.001,˛D 25, ı D 0.00001,Case B') N D 0.6, 0.003335 <ˇ< 0.005 D, D 0.001, r D 0.001,˛D 25, ı D 0.1.