2004-2005年广东省第3次死因回顾性调查样本代表性与数据质量评价

目的阐述2004—2005年广东省第3次死因回顾调查样本代表性与数据质量，为资料利用提供依据。方法通过多阶段分层整群抽样方法，对广东省13个县（市、区）2004-2005年全部死亡人群进行回顾调查，依据人口学特征、拟舍优度检验、玛叶指数等对2004-2005年广东省第3次死因回顾调查样本（其中12个调查点数据）的代表性进行评价，从死因诊断依据、死亡率、死因编码质量等方面评价数据质量。结果本次调查覆盖全省12个市、县（区）共758．5472万人口，无论是年龄构成还是性别比、地域和社会经济等指标统计学检验，还是年龄构成的拟合优度检验，调查样本与2000年广东省人口普查户籍人口相比，差异均无统计学意义（均P〉0．05）；选取调查样本10—69岁和20～79岁2个年龄组样本，计算玛叶指数为3．59，说明调查样本人口数据无年龄偏好。本次调查地区共收集2004-2005年12个调查点死亡个案99919例，平均粗死亡率为658．6／10万，10个调查点粗死亡率符合质控要求，婴儿死亡率为6．7％。，全部死亡个案的死因最高诊断单位以县及以上医疗机构为主，达62．48％，死因最高诊断依据以临床＋理化和临床诊断为主，合占85．93％；其中死于恶性肿瘤个案的诊断单位以县及以上医疗机构诊断比例最高，达89．33％，诊断依据以尸检、病理、手术和临床＋理化等合计比例达83．76％。全部调查个案的编码总体准确率达96．07％，死因不明率为1．89％，死因不准确率为1．56％，死因错误率为0．48％。结论2004—2005年广东省第3次死因回顾调查样本代表性良好，调查结果可代表全省情况，调查过程质控严格，数据完整性、可靠性和准确性指标均符合质控要求，数据质量良好。

Sample representativeness and data quality evaluation of the third death retrospective survey in Guangdong Province, 2004-2005.

Objective To describe the representativeness of sample and the quality of the data obtained from the 3rd death retrospective survey in Guangdong Province during 2004 to 2005 and provide an important reference for utilization. Methods The demographic characteristics, goodness of fit test and M- yer＇s index were used to evaluate the representativeness of sample from the 3rd death retrospective survey in Guangdong Province. Crude death rate, diagnosis basis, diagnoses of death causes and coding accuracy were applied to valuate the data quality. Results In this survey 12 cities/counties were selected, covering 7 585 472 population. Compared to the household population from the census of Guangdong Province in 2000, no significant differences were found in age structure, gender ratio, region distribution, and social economic status of the population by statistical test, nor in the age composition by using the goodness of fit test in the survey samples （ all P 〉 0. 05 ）. The Myer＇ s index was 3.59 for two selected age groups in 10 - 69 and 20 - 79 years, indicating that the distribution of age was proportional and without preferences in any age groups. A total of 99 919 death cases were collected in this survey in 2004-2005, with an average crude death rate of 658. 6/100 000, and infant mortality of 6. 7‰. Of the highest diagnosis agencies of all death cases, hospitals at or above the county level accounted for the biggest proportion （62. 48% ）. The means adopted the most for diagnosing causes of death was mainly the clinical diagnosis plus physical and chemical tests or clinical diagnosis only （85.93 % of all cases）. As the highest proportion of medical diagnosis, up to 89.33% of death of all malignant tumor cases was diagnosed by the hospitals at or above the county level, mainly based on autopsy, pathological diagnosis, surgery diagnosis, clinical diagnosis as well as physical and chemical tests （83.76%）. Of all surveyed cases, the coding accuracy rate reached 96.07%, the rate of unknown cause of death was 1.89%, the rate of inaccurate or unspecified death cause was 1.56%, and error rate of death cause was 0. 48%. Conclusion The representativeness of sample from the 3rd death retrospective survey in Guangdong during 2004-2005 was good. By strict quality control during the investigation, the integrity, reliability and accuracy of the data met the quality requirements.