Achieving (at least) a worst-case correctness error is essential for an underlying public-key encryption (PKE) scheme to which the Fujisaki-Okamoto (FO) transformation is applied. There are three average-case to worst-case (ACWC) transformations—denoted as \(\textsf{ACWC}_{0}\) , \(\textsf{ACWC}_{1}\) (PKC 2023), and \(\textsf{ACWC}_{2}\) (TIFS 2023)-which generically convert a PKE scheme with an average-case correctness error into one with a worst-case correctness error. However, in these ACWC transformations the \(\gamma \) -spreadness, a critical factor in determining explicit rejection ( \(\textsf{FO}^{\perp }\) ) or implicit rejection ( \(\textsf{FO}^{\not \perp }\) ), has not been established with rigorous proofs. Existing analyses of \(\gamma \) -spreadness lack rigorous proofs, include analytical flaws, or fail to achieve the tightest possible bounds. In this work, we reprove the \(\gamma \) -spreadness of ACWC-transformed PKE schemes by leveraging two key facts: the random oracle is chosen at random and the encoding mechanism used in the ACWC framework is message-hiding. Our new proofs are applied to the previous NTRU-based PKE schemes, called \(\textsf{NTRU}\text {-}\textsf{C}\) , \(\textsf{NTRU}\text {-}\textsf{B}\) , and \(\mathsf {NTRU+}\) , giving the corrected \(\gamma \) -spreadness for those PKE schemes with concrete parameters.