目的：研究单纯性高血压的直接医疗费用和门诊费用对住院费用的替代影响，并据此对卫生政策制定者和高血压患者提出关于管理高血压的建议。方法： 使用中国北方某市新型农村合作医疗（以下简称“新农合”）的高血压患者2008年至2013年的报销数据，运用描述性统计分析高血压患者的门诊费用以及根据就诊情况分成3组（单纯性高血压、高血压并发症、其他疾病）后单纯性高血压的直接医疗费用，进一步用t检验比较组间差异。运用固定效应两部模型分析门诊费用对住院费用的替代效应，进一步采用Logistic回归分析估算门诊次数和门诊报销费用对住院费用的具体替代影响。结果： 高血压患者2008年门诊次均费用为283.49元，2013年上升至370.93元；分成3组后，单纯性高血压患者2008年门诊次均费用为449.79元，2013年上升至582.53元；门诊总费用中，单纯性高血压的比例占45.73%；住院总费用中，单纯性高血压的比例占9.29%。通过固定效应两部模型计算边际效应，门诊次数的增加或者门诊报销费用的增加都能显著引起住院费用的减少，如2013年门诊次数增加1%，则门诊总费用增加647.89万元，住院费用减少3 986.51万元。门诊-住院费用替代比在2010年及以后稳定在5左右。结论： 卫生决策者和高血压患者应该重视门诊服务对住院服务的替代作用，尤其是针对单纯性高血压的患者，其主要直接医疗费用在门诊而不是住院。

Objective：To study direct medical cost of hypertensive patients without complicated diseases， to analyze substitution impacts of outpatient expenditure on inpatient expenditure, and then to make recommendations to health policy makers and hypertensive patients for managing hypertension. Methods: The claims data of hypertensive patients insured by the rural new cooperative medical scheme in a city in northern China from 2008 to 2013 were analyzed, descriptive statistics employed to analyze direct medical cost of hypertensive patients and hypertensive patients without complicated diseases which was one of the three groups divided by medical treatment conditions (hypertension without complicated diseases, hypertension with complicated diseases, and other diseases), and the differences between groups compared by t test. The fixed effect two-part model was employed to analyze the substitution effect of outpatient services on inpatient services, and specific effects of outpatient times and reimbursement expenses on the inpatient expenditure were calculated by Logistic regression analysis. Results: For the hypertensive patients, the average cost of outpatients was 283.49 Yuan in 2008 and rose to 370.93 Yuan in 2013. After being divided into three groups, for hypertensive patients without complicated diseases, the average cost of outpatient was 449.79 Yuan in 2008 and rose to 582.53 Yuan in 2013. In the total cost of the outpatient, 45.73% was for the hypertensive patients without complicated diseases; in the total cost of the inpatient, only 9.29% was for the hypertensive patients without complicated diseases. By calculating the marginal effect of parameter estimation through the fixed effect two-part model, the inpatient cost could be significantly reduced by increasing the visit times of outpatient utilization or the cost of outpatient reimbursement. For example, in 2013, a 1% rise of outpatient visit times increased outpatient expenditures by 6.48 million Yuan, which could reduce inpatient expenditures by 39.86 million Yuan. The substituted ratio of outpatient cost on inpatient cost had been around 5 in 2010 and later. Conclusion: The health policymakers and hypertensive patients should pay more attentions to the substitution of outpatient cost for inpatient cost, especially for the hypertensive patients without complicated diseases whose direct medical cost was mainly due to outpatient rather than inpatient costs.