Background: COVID-19 has greatly attacked China, spreading in the whole world. Articles were posted on many official WeChat accounts to transmit health information about this pandemic. The public also sought related information via social media more frequently. However, little is known about what kinds of information satisfy them better. This study aimed to explore the characteristics of health information dissemination that affected users’ information behavior on WeChat. Methods: Two-wave data were collected from the top 200 WeChat official accounts on the Xigua website. The data included the change in the number of followers and the total number of likes on each account in a 7-day period, as well as the number of each type of article and headlines about coronavirus. It was used to developed regression models and conduct content analysis to figure out information characteristics in quantity and content. Results: For nonmedical institution accounts in the model, report and story types of articles had positive effects on users’ following behaviors. The number of headlines on coronavirus positively impacts liking behaviors. For medical institution accounts, report and science types had a positive effect, too. In the content analysis, several common characteristics were identified. Conclusions: Characteristics in terms of the quantity and content in health information dissemination contribute to users’ information behavior. In terms of the content in the headlines, via coding and word frequency analysis, organizational structure, multimedia applications, and instructions—the common dimension in different articles—composed the common features in information that impacted users’ liking behaviors.

BACKGROUND Since the novel coronavirus broke out in December 2019 in Wuhan, Hubei, it has completely spread in China. In this situation, articles were posted on many WeChat official accounts to transmit health information about this pandemic. The public also sought related information via social media more frequently. However, little is known about what kinds of information satisfy them better. OBJECTIVE This study aimed to explore the characteristics of health information dissemination that affected users’ behavior on WeChat. METHODS Two-wave data were collected from the top 200 WeChat official accounts on the Xigua website. The data, including the change in the number of followers and the total number of likes on each account in a 7-day period, were selected as dependent variables. The number of each type of articles and headlines about coronavirus served as independent variables. Multiple and simple linear regression models were used to conduct data analysis. In addition, the contents of the headlines with more than 10,000 likes were examined to further explore their characteristics. RESULTS The top 200 official accounts could be classified into three groups, namely, nonmedical institution, medical institution, and individual accounts. After filtering data, the sample size of these groups was 124. The multiple linear regression model of nonmedical and medical institution accounts had adjusted R2 of 0.355 and 0.452, respectively. For nonmedical institution accounts, the simple linear regression model had an adjusted R2 of 0.317. The other results were insignificant, and an ideal model for them could not be developed. However, R2 indicated an acceptable fit. For nonmedical institution accounts, report and story types of articles were significant, as indicated by the multiple linear regression model (B=2.724, P=.007; B=14.875, P=.003), and both types were identified to have positive effects on behavior. For the simple linear regression model, the number of headlines on coronavirus was observed to have a positive effect on behavior (B=3.084, P<.001). For medical institution accounts, reports and science types were significant, as revealed by the multiple linear regression model (B=4.381, P=.009; B=31.564, P<.001). They also had a positive effect. CONCLUSIONS Different characteristics in health information dissemination contribute to users’ behavior. For nonmedical institution accounts, report and story types positively influenced the change in the number of followers. For medical institution accounts, report and science types exerted positive effects. In headlines, organizational structure, manner of description, and multimedia applications contributed to unexpected likes. CLINICALTRIAL