With the rapid development of e-health and telemedicine, previous studies have explored the relationship between physician-patient communication and patient satisfaction; however, there is a paucity of research on the influence of the characteristics of patient communication on the characteristics of physician feedback. Based on the communication accommodation theory, as well as the computer-mediated communication theory and media richness theory, this study aimed to explore how characteristics of patient communication influence characteristics of physician feedback in online health communities. We employed a crawler software to download the communication data between 1652 physicians and 105,325 patients from the Good Doctor platform, the biggest online health community in China. We built an empirical model using this data and employed a multilevel model to test our hypotheses using Stata and Python software. The results indicate that the amount of patients' rendered information positively influences the physicians' text (α = 0.123, t = 33.147, P < .001) and voice feedback (β = 0.201, t = 40.011, P < .001). Patients' hope for help signals and the provision of their electronic health records weaken the effect of the amount of patients' rendered information on physicians' text feedback (α = -0.040, t = -24.857, P < .001; α = -0.048, t = -15.784, P < .001), whereas, it strengthened the effect of the amount of patients' rendered information on physicians' voice feedback (β = 0.033, t = 14.789, P < .001; β = 0.017, t = 4.208, P < .001). Moreover, the occurrence of high-privacy diseases strengthened the effect of the amount of patients' presented information on physicians' text and voice feedback (α = 0.023, t = 4.870, P < .001; β = 0.028, t = 4.282, P < .001). This research contributes to the development of computer-mediated communication theories and sheds light on service delivery in the online health community.