目的: 了解黑龙江省佳木斯市肾综合征出血热（HFRS）疫情特征，预测其发病趋势，为制定HFRS防控措施提供依据。方法: 采用描述流行病学方法对HFRS流行特征进行分析，率的比较用χ²检验；运用SPSS 22.0和Eview 10.0软件对2004－2021年HFRS月发病率建立自回归差分移动平均（ARIMA）最优模型，预测2022年月发病率。结果: 共报告HFRS 5 772例，年均发病率为13.15/10万；5－7月为HFRS春夏季小高峰，10－12月为秋冬季大高峰。各县（市、区）HFRS年均发病率同江和抚远市较高；发病年龄主要集中在15~69岁人群，35~39岁年龄组构成比最高；病例主要以青壮年男性为主；职业以农民最多，占69.53%；男女性别比例为3.65∶1。ARIMA（1，1，1）（1，1，1）₁₂为短期预测佳木斯市HFRS发病率的最优模型，贝叶斯信息准则（BIC）值为-0.879，残差序列Ljung-Box Q检验为白噪声（Q=15.867，P=0.322），残差序列自相关系数和偏自相关系数均落在95%置信区间内。应用该模型预测2022年HFRS月发病率，结果显示有小幅升高趋势，但仍处于较低水平。结论: 2004－2021年佳木斯市HFRS发病率总体呈下降趋势，呈季节性双峰分布；边境市年均发病率较高；男性高于女性，职业以农民为主。ARIMA（1，1，1）（1，1，1）₁₂为最优模型，可用于预测佳木斯市HFRS短期发病趋势，为针对性防控提供科学依据。

Objective: To analyze the epidemiological characteristics of hemorrhagic fever with renal syndrome (HFRS) in Jiamusi, Heilongjiang, China and predict the incidence trend, so as to provide a basis for formulating HFRS prevention and control measures. Methods: Descriptive epidemiology method was used to analyze the epidemiological characteristics, and the Chi-square test was used to compare the rates; SPSS 22.0 and Eview 10.0 softwares were used to establish an optimal autoregressive integrated moving average (ARIMA) model for the monthly incidence rate of HFRS from 2004 to 2021, and the monthly incidence rate in 2022 was predicted. Results: A total of 5 772 cases of HFRS were reported, with an average annual incidence rate of 13.15/100 000; the small peak of HFRS in spring and summer occurred from May to July, and the large peak in autumn and winter occurred from October to December. The average annual incidence rate of HFRS in Tongjiang and Fuyuan were higher than that in other cities, counties, and districts. The onset age was mainly 15-69 years old, and the proportion of 35-39 years old was the highest. The cases were mainly young males and occurred most in farmers (69.53%) compared with other occupations, with a male-to-female ratio of 3.65:1. ARIMA (1, 1, 1) (1, 1, 1)₁₂ was the optimal model for short-term prediction of the incidence rate of HFRS in Jiamusi, with BIC=-0.879. The residual sequence was determined to be white noise by the Ljung-Box Q test (Q=15.867, P=0.322), and the autocorrelation coefficient and partial autocorrelation coefficient of the residual sequence fell within 95% confidence interval. The monthly incidence rate of HFRS in 2022 predicted by the model showed a slight upward trend, but was still at a low level. Conclusions: From 2004 to 2021, the incidence rate of HFRS in Jiamusi city showed a downward trend and a seasonal bimodal pattern. The average annual incidence rate is higher in border cities; the incidence rate is higher in males than in females, and mainly in farmers. ARIMA (1, 1, 1) (1, 1, 1)₁₂ is determined as the optimal model, which can be used to predict the short-term incidence trend of HFRS in Jiamusi, providing a scientific basis for targeted prevention and control.