<p>The additive hazards model is one of the most commonly used models for regression analysis of failure time data and a great deal of literature has been established for its estimation under various situations. In this paper, we consider the situation where only interval-censored data are available and also there exists a random change point, which occurs quite often on cancer studies among others and for which there does not seem to exist an established estimation approach. For the situation, we propose a sieve maximum likelihood estimation procedure with the use of Bernstein polynomials, and the asymptotic properties of the resulting estimators are established. In addition, a simulation study is conducted to assess the empirical performance of the proposed approach and suggests that it works well in practical situations. The proposed methodology is applied to a set of real data on child mortality that motivated this study.</p>