The association between life-course socio-professional trajectories and mortality has already been discussed in the literature. However, these socio-professional trajectories may be subject to informative censoring due to death. This loss to follow-up which is related to an individual’s survival, should not be ignored and thus, it is of interest to model jointly these professional trajectories and their survival. The main focus has been made on continuous, binary or ordinal variables while much less attention has been paid to nominal categorical data. Therefore, an extension to the joint modelling of longitudinal nominal data and survival under a likelihood-based approach is proposed. A generalized linear mixed model is considered for modelling the longitudinal nominal data, in addition to two cause-specific proportional hazards model for the survival competing risks data. The association between longitudinal and survival outcomes is captured by assuming a multivariate Gaussian distribution for the joint distribution of the random effects of two sub-models. The proposed joint model provides a robust framework for estimating longitudinal membership probabilities, accounting for informative censoring caused by individual’s death. Simulations are carried out to assess the performance of this joint model comparing with the results of the separate longitudinal and competing risks analysis. A disadvantage of joint models is that they are computationally intensive. To overcome this problem, an approach mimicking a meta-analysis strategy of individual participant data is suggested. The relevance of this approach is then illustrated on a large sample of the French salaried population, which contains all employment records between 1976 and 2002.