Adverse events (AE) are harmful outcomes during medical care. The severity and frequency of these events serve as study endpoints in clinical trials, crucial for evaluating treatment safety. Patients may encounter multiple adverse events concurrently, and the recorded data exhibit diverse structures due to varying durations and characteristics of both short-term and long-term AEs. Moreover, AE severity may fluctuate over time due to disease progression or treatment response. Most current analyses focus solely on a single AE, neglecting severity information and failing to distinguish adequately between short-term and long-term AEs. In response, we propose an efficient joint model to assess treatment effects on multiple AE occurrences. This model comprehensively considers AE severities and correlations while effectively addressing structural differences between short-term and long-term AEs. Through simulation studies, this method has demonstrated high accuracy in parameter estimation.