OBJECTIVES: The OAK study is a randomized phase III trial of atezolizumab vs. docetaxel in prior treated NSCLC. The analysis population is the primary analysis population (425 subjects per arm, datacut: 7 Jul 2016). This work evaluates the rate of adverse events leading to hospitalizations per month, as well as the total length of hospitalization stay per patient normalized with respect to the time on treatment per patient.

METHODS: A negative binomial model is applied to the number of hospitalizations per patient, using time on treatment (plus a 30-day safety window) as offset. Similarly, a negative binomial model is applied to the total hospitalization duration per patient. One subject in the atezolizumab arm, hospitalized for over 200 days was considered an outlier; hence, excluded from the analysis.

RESULTS: The monthly rate of hospitalization, normalized with respect to the time on treatment per patient in the docetaxel arm is 0.14; whereas, the monthly rate of hospitalizations in the atezolizumab arm is 0.08, with a 40% reduction in the rate of hospitalizations (p-value = 0.0004). The mean length of hospitalization stay for the docetaxel arm is 10.63 days for the docetaxel arm and additional 1.3 days on average for the atezolizumab arm (p-value = 0.005).

CONCLUSIONS: atezolizumab statistically significantly reduced the rate of hospitalizations due to adverse events compared to docetaxel. In addition, after being hospitalized the length of stay in the atezolizumab arm is shorter than the length of stay for the docetaxel arm.