Stochastic Dual Dynamic Programming (SDDP) is currently one of the most employed methods for hydrothermal planning. All previous works on this subject are based on a hazard-decision approach, whereas reality is more closely related to a decision-hazard process. This dissonance between planning and implementation is a source of time-inconsistency, as future planned decisions under the same conditions may not be put into practice. If on the one hand the hazard-decision modeling framework allows a scenario-decomposable efficient solution methodology, on the other hand the decision-hazard structure provides a more robust (pessimistic) solution as it does not rely on anticipativity assumptions. In this work, we measure the inconsistency-gap related to the current methodology and propose an alternative approach for hydrothermal planning that utilizes an informationrevelation structure and decision process based on a decision-hazard framework, thereby approximating the planning model to realistic operational actions. Instead of relying on non-anticipativity constraints, which would prevent the scenario decomposition of each two-stage stochastic subproblem, the proposed methodology considers first-stage decisions as state variables to be optimized through the SDDP procedure. In this framework, the complexity and time required to find a solution is considerably reduced yet ensuring the decision-hazard decision structure and non-anticipativity of the first-stage decisions. Results based on the Brazilian power system indicate that this inconsistency may considerably increase generation of more expensive thermal units, leading to spikes in energy market spot prices and an increase in overall operational costs. Therefore, the proposed decision-hazard approach and augmented-state solution methodology constitute timely and relevant contributions to both industry practices and state of the art literature on the subject of hydrothermal operation planning under uncertainty.