METHODS: Data were from the baseline surveys for chronic diseases among residents ages ≥18 years old in 4 districts/counties of Changsha between 2013 and 2014. All surveys adopted multi-stage random sampling to select samples. The Bootstrap resampling method was used to randomly select 1 000 repeated samples with replacement to obtain robust estimate of subgroup prevalence rates. Hypertension prevalence was calculated by using the data from both questionnaire interview and on-site measurement. Using the results of on-site measurement as the golden standard, the accuracy of questionnaire interview and 95% uncertainty interval were estimated. Multivariate logistic regression was used to investigate the influential factors for the underestimated hypotension prevalence based on questionnaire interview.
RESULTS: The hypertension prevalence from on-site measurement among the residents in the 4 districts/counties of Changsha was significantly higher than that from questionnaire interview (prevalence ratio: 1.26-2.31). Taking the results of on-site measurement as the golden standard, the sensitivity of questionnaire interview on hypotension prevalence range from 41.76% to 74.83% among the 4 districts/counties, and the specificity fell between 98.51% and 99.46%. The underestimation in questionnaire interview was more likely to occur in the youngest age group (18-34 years old), males, and residents were at lower levels of education in all 4 districts/counties.
CONCLUSIONS: Compared with the on-site measurement, questionnaire interview significantly under-estimate the hypertension prevalence, suggesting that the on-site measurement method should be firstly considered in epidemiological surveys for hypertension prevalence. If only the questionnaire method can be used to collect data due to conditions, it is necessary to make corresponding corrections to the questionnaire results with reference to relevant research evidence.
目的: 采用现场检测评价问卷询问所得高血压患病率的准确性,并探讨其影响因素。方法: 资料来自2013至2014年长沙市4个区/县针对≥18岁常住居民的慢性病基线调查资料,4个区/县均采用多阶段随机抽样选择样本。为获取稳定的亚组患病率,利用Bootstrap再抽样方法对样本数据进行1 000次重复抽样,计算基于复杂抽样的问卷询问和现场实测两种方式所得高血压患病率。以现场实测结果为金标准,估算问卷询问结果的准确性及95%不确定区间。采用多因素logistic回归探讨问卷询问方法相对于现场实测低估高血压患病率的影响因素。结果: 长沙市4个区/县居民现场实测的高血压患病率均明显高于问卷询问的高血压患病率(患病率比值:1.26~2.31)。以现场实测结果作为金标准,4个区/县问卷询问所得高血压患病报告的灵敏度为41.76%~74.83%,特异度为98.51%~99.46%。4个区/县问卷询问方式所得的高血压患病率在18~34岁、男性居民和文化程度偏低者中出现低估的可能性更高。结论: 相对于现场实测,问卷询问所得的高血压患病率明显低估,提示流行病学调查应采用现场实测方式估计高血压患病率。若因条件限制仅能采用问卷询问方法收集数据,则需要参照相关研究证据对问卷询问结果进行相应校正。.