连续保健视角下不同权重法构建妇幼基本卫生服务综合指标比较:以吉林省国家卫生服务调查为例

Comparing Different Weights to Construct Composite Indicators of Maternal and Child's Basic Health Services from the Prospective of Continuum of Care:Based on Data from the National Health Services Survey 2008 and 2013 in Jilin Province

目的从连续保健的视角出发利用不同权重法构建妇幼基本卫生服务综合指标,基于吉林省国家卫生服务调查数据对综合指标进行比较。方法依据联合国"倒计时"和生命挽救工具,基于吉林省2008年和2013年两次国家卫生服务调查数据筛选指标,分别采用等权重法、分服务包等权重法、流行病学权重法和主成分权重法构建综合指标,计算克隆巴赫系数检验综合指标的内部一致性,采用随机区组设计的方差分析比较综合指标得分差异,计算Spearman相关系数比较综合指标的等级相关性,通过妇幼健康结局的相关分析检验综合指标的效度。结果最终共筛选出13项干预措施指标,克隆巴赫系数大于0.7,具有较好的内部一致性。方差分析得到4种综合指标得分差异有统计学意义(F=8.14,P<0.01)。等权重法、主成分权重法和分服务包等权重法两两间均存在较强的等级相关,Spearman相关系数均超过0.9,流行病学权重法与等权重法、主成分权重法和分服务包等权重法的Spearman相关系数分别为0.697、0.671和0.818。流行病学权重法综合指标对5岁以下儿童死亡率和生长迟缓发生率具有一定的预测价值(R^2=0.168,P<0.05;R2=0.398,P<0.01);等权重法、分服务包等权重法和主成分权重法综合指标仅对5岁以下儿童生长迟缓发生率具有预测价值(R^2=0.238,R^2=0.304,R2=0.232,P均<0.01)。结论流行病学权重法构建妇幼基本卫生服务综合指标具有其合理性,对妇幼健康结局具有一定预测价值。

Objective To construct composite indicators of maternal and child＇s basic health services using different weights from the prospective of continuum of care and to compare them based on data from the Na-tional Health Services Survey 2008 and 2013 in Jilin Province.Methods The study selected indicators based on the countdown and the Lives Saved Tool（Li ST）,considering the data availability from the 2008 and 2013 National Health Surveys in Jilin.Equal weights,equal weights for different areas,epidemiology weights,and principal component weights were used to construct different composite indicators.Cronbach＇s alpha reliability coefficient was used to test the internal consistency.Analysis of variance of randomized block design was used to test the differences.The Spearman＇s correlation coefficient was calculated to compare the rank correlation.The maternal mortality,children under 5 years old mortality rate,and children under 5 years old stunting incidence were used as the outcome indicators to test the validity using correlation analysis.Results Finally 13 indicators were selected,with the Cronbach＇s alpha reliability coefficient value exceeding 0.7,which means acceptable internal consistency of these indicators.Analysis of variance showed that the scores were not significantly the same（F=8.14,P〈0.01）.The rank correlations among equal weights,principal component weights,and equal weights for different areas composite indicators were quite strong,and the Spearman＇s correlation coefficient between either two were all above 0.9,while the epidemiology weights composite indicator got smaller Spearman＇s correlation coefficient with equal weights,principal component weights,and equal weights for different areas composite indicators（0.697,0.671,and 0.818,respectively）.The composite indicator using epidemiology weights showed significant negative correlation with children under 5 years old mortality rate and children under 5 years old stunting incidence（R^2=0.168,P〈0.05;R^2=0.398,P〈0.01,respectively）.The composite indicators using equal weights,equal weights for different areas and principal component weights only showed significant negative correlation with children under 5 years old stunting incidence（R^2=0.238,R^2=0.304,R^2=0.232,P〈0.01 for all,respectively）.Conclusion The composite indicator using epidemiology objective weights has some predictive value for maternal and child health outcomes.