Establishing a Multicentre Trauma Registry in India: An Evaluation of Data Completeness
Establishing a Multicentre Trauma Registry in India: An Evaluation of Data Completeness
复制标题
DOI：
10.1007/s00268-019-05039-2
复制
发表时间：
2019-10-01
影响因子：
2.6
通讯作者：
Mitra, Biswadev
中科院分区：
文献类型：
--
作者：
Shivasabesan, Gowri;O'Reilly, Gerard M.;Mitra, Biswadev
创伤登记数据的完整性对其有效使用至关重要。本研究旨在评估印度一项新的多中心创伤登记研究中缺失数据的程度，并评估数据完整性与缺失数据的潜在预测因素（特别是死亡率）之间的相关性。方法根据2016年4月19日至2017年4月30日收集的数据，确定所有成年人变量的缺失数据比例。如果初始收缩压、心率、呼吸频率或格拉斯哥昏迷量表中的任何一项缺失，则将院内生理数据定义为缺失。使用手动逐步选择的单变量logistic回归和多变量logistic回归来研究死亡率（和其他潜在的预测因子）与缺失的生理数据之间的关联。结果对登记的4466例创伤患者的资料进行了分析。在59个变量中，大多数（n = 51; 86.4%）缺失的观察值小于20%。有808例（18.1%）患者至少缺失一次首次住院生理学观察结果。住院死亡与住院期间生理数据缺失相关（校正OR 1.4; 95% CI 1.02-2.01; p = 0.04）。其他与缺失数据相关的重要因素包括：患者到达时间、医院护理、“其他”损伤地点和特定损伤机制。攻击/杀人伤害意图和胸部X线检查的发生与不缺失任何首次住院生理变量相关。结论大部分变量收集良好。医院死亡是更严重损伤的代表，与首次住院生理观察结果缺失相关。这仍然是创伤登记的一个重要限制。
Background The completeness of a trauma registry's data is essential for its valid use. This study aimed to evaluate the extent of missing data in a new multicentre trauma registry in India and to assess the association between data completeness and potential predictors of missing data, particularly mortality. Methods The proportion of missing data for variables among all adults was determined from data collected from 19 April 2016 to 30 April 2017. In-hospital physiological data were defined as missing if any of initial systolic blood pressure, heart rate, respiratory rate, or Glasgow Coma Scale were missing. Univariable logistic regression and multivariable logistic regression, using manual stepwise selection, were used to investigate the association between mortality (and other potential predictors) and missing physiological data. Results Data on the 4466 trauma patients in the registry were analysed. Out of 59 variables, most (n = 51; 86.4%) were missing less than 20% of observations. There were 808 (18.1%) patients missing at least one of the first in-hospital physiological observations. Hospital death was associated with missing in-hospital physiological data (adjusted OR 1.4; 95% CI 1.02-2.01; p = 0.04). Other significant associations with missing data were: patient arrival time out of hours, hospital of care, 'other' place of injury, and specific injury mechanisms. Assault/homicide injury intent and occurrence of chest X-ray were associated with not missing any of first in-hospital physiological variables. Conclusion Most variables were well collected. Hospital death, a proxy for more severe injury, was associated with missing first in-hospital physiological observations. This remains an important limitation for trauma registries.