探索倾向得分匹配法（PSM）在血吸虫病防治（简称血防）健康教育知识、态度和行为（KAP）模式分析中的应用。收集2018年4月于四川省凉山州喜德县2个村（红莫镇桃源村为干预村,李子乡洛乃格村为对照村）开展的16岁及以上村民血吸虫病行为改变交流（BCC）干预试点项目基线问卷调查结果,应用PSM对干预村和对照村人群的年龄、性别、学历和知识、态度及行为得分进行匹配。使用SPSS 25.0进行数据统计分析,组间构成比采用χ²检验进行比较,得分采用t检验比较。采用Person相关性分析、线性回归分析对知识、态度及行为变量间的相关性进行分析。基线调查结果显示,共收集到619份有效问卷。干预村有319份有效问卷,基本情况为：男性202人,女性117人;40~59岁人群占比最多（145人,45.5%）;文盲学历人群占比最多（129人,40.4%）。对照村有300份有效问卷,基本情况为：男性159人,女性141人;40~59岁占比最多（139人,46.3%）;小学学历人群占比最多（168人,56%）。干预村抽取的120份有效问卷经PSM匹配后,对照村中成功匹配到102例问卷对象,匹配后干预村和对照村问卷对象数量接近。匹配前的干预村和对照村人群性别、年龄、学历差异均有统计学意义（均P < 0.01）,匹配后的干预村和对照村人群性别、年龄、学历差异均无统计学意义（均P > 0.05）。匹配前的干预村人群知识、态度和行为得分分别为6.5 ± 2.0、3.4 ± 0.9、11.7 ± 0.9,对照村人群知识、态度和行为得分分别为8.1 ± 2.9、3.2 ± 0.7、10.0 ± 2.3,差异均有统计学意义（均P < 0.05）。匹配后的干预村人群知识、态度和行为得分分别为7.0 ± 1.5、3.3 ± 0.9、11.3 ± 0.9,对照村人群知识、态度和行为得分分别为7.8 ± 2.4、3.4 ± 0.7、11.4 ± 1.4。其中,匹配后的干预村和对照村人群态度和行为得分差异均无统计学意义（均P > 0.05）,尽管匹配后的干预村和对照村人群知识得分差异仍有统计学意义（P < 0.01）,但二者的知识得分差值由匹配前的1.6下降至匹配后的0.8。Person相关性分析结果显示,研究对象的知识、态度、行为3个变量均为正相关（r > 0）。回归分析结果显示,知识正向影响态度,知识正向影响行为,态度正向影响行为,知识-态度-行为路径存在部分中介效应。在根据血防知识、态度和行为模式对血吸虫病流行地区的人群进行积极干预的同时,需结合恰当的统计方法进行结果分析,以针对性制定干预策略。

To explore application of the propensity score matching (PSM) method in knowledge, attitude and practice(KAP) analysis of health education on schistosomiasis control. Baseline questionnaire survey of the schistosomiasis intervention pilot project behavior change communication (BCC) was performed among villagers aged 16 and above in April 2018 in two villages in Xide County, Liangshan Prefecture, Sichuan province (Taoyuan Village of Hongmo Town served as the intervention village and Luonaige Village of Plums township as the control village). The survey results were analyzed using the PSM method to score the matching of age, gender, educational background and knowledge, attitude and practice of the surveyed villages in the two selected villages. Statistical analysis was performed using SPSS 25.0 software. The composition ratio was compared between the groups by chi-square test, and the scores were analyzed by t-test. Person correlation analysis and linear regression analysis were used to analyze the correlations between knowledge, attitude and practice. A total of 619 valid questionnaires were collected during the baseline survey. There were 319 valid questionnaires in the intervention village, including 202 males and 117 females, with the majority villagers at age 40-59 (145, 45.5%) and in illiteracy (129, 40.4%). There were 300 valid questionnaires in the control village, including 159 males and 141 females, with the majority villagers at age 40-59 (139, 46.3%) and having primary school education (168, 56%). The 120 valid questionnaires selected from the intervention village were successfully matched with 102 subjects from the control village by the PSM method, and the number of subjects was close to that of the control villages. There were significant differences in gender, age and educational background between the intervention and the control villages before matching (P < 0.01); however, these differences became insignificant after matching (P > 0.05). The scores of knowledge, attitudes and practices in the intervention village before matching were 6.5 ± 2.0, 3.4 ± 0.9 and 11.7 ± 0.9, respectively, while those in the control village were 8.1 ± 2.9, 3.2 ± 0.7 and 10.0 ± 2.3, respectively, being statistically significant between the scores of two villages (P < 0.05). After matching, the scores of knowledge, attitude and practice in the intervention village were 7.0 ± 1.5, 3.3 ± 0.9 and 11.3 ± 0.9, respectively, and those in the control village were 7.8 ± 2.4, 3.4 ± 0.7 and 11.4 ± 1.4, respectively. There were no significant differences in scores of attitude and practice between the intervention village and the control village after matching (P > 0.05). Although the difference in knowledge score between the intervention and the control villages after matching remained statistically significant(P < 0.01), the difference in the knowledge scores decreased from 1.6 before matching to 0.8 after matching. Results of Person correlation analysis showed that the three variables knowledge, attitude and practice were all positively correlated(r > 0). Regression analysis showed that knowledge had a positive impact on attitude, and practices, while attitudes also positively impacted practices. The knowledge-attitude-practice path has mediating effects. When active intervention is being conducted in schistosomiasis endemic areas based on the KAP approach, proper statistical methods may be integrated to analyze the control results, and thereby to pertinently formulate intervention strategy.