Objective  To establish comprehensive indicator systems for early warning of acute respiratory infection (ARI) based on multi-source data.

Methods The Delphi method was adopted for expert consultation. Twenty experts were invited to score the necessity, availability, and sensitivity of different early warning indicators, and the weights were determined based on the scoring results.

Results The two rounds of expert consultation showed the expert positive coefficients being both 100%, the coefficients of expert authority being both 0.79, and the coefficients of concordance being 0.280 and 0.487 (P < 0.05), respectively. According to the characteristics of ARI in the stages of occurrence, transmission, and epidemics, two indicator systems–the early warning indicator system for risk of ARI and the early warning indicator system for potential clustering or outbreak of ARI–were established. The former included 22 indicators in 5 categories, and the latter included 29 indicators in 6 categories. In the early warning indicator system for risk of ARI, the top 3 indicators with the highest weights were immunization coverage rate (0.052), variations of known pathogens (0.051), and variations of dominant prevalent strains (0.051). In the early warning indicator system for potential clustering or outbreak of ARI, the top 3 indicators with the highest weights were surveillance data of respiratory syndrome (0.040), positive rate of specific pathogen (0.040), and number of cases reported in online direct report system (0.040).

Conclusions The comprehensive indicator systems for early warning of ARI based on multi-source data and Delphi method established in this study have certain scientificity and practicability, which can provide support for improving the multi-point triggered early warning of infectious diseases.