探索丹麦全国性永久性养老院居住评估的同时效度 - 使用两个行政登记册的横断面数据分析。
Exploring the concurrent validity of the nationwide assessment of permanent nursing home residence in Denmark - A cross-sectional data analysis using two administrative registries.
机构信息
The Research Unit and Section of General Practice, Institute of Public Health, University of Copenhagen, Øster Farimagsgade 5, Post box 2099, 1014, Copenhagen K, Denmark.
The Research Unit and Section of General Practice, Institute of Public Health, University of southern Denmark, Odense, Denmark.
出版信息
BMC Health Serv Res. 2017 Aug 29;17(1):607. doi: 10.1186/s12913-017-2535-2.
BACKGROUND
Many register studies make use of information about permanent nursing home residents. Statistics Denmark (StatD) identifies nursing home residents by two different indirect methods, one based on reports from the municipalities regarding home care in taken place in a nursing home, and the other based on an algorithm created by StatD. The aim of the present study was to validate StatD's nursing home register using dedicated administrative municipality records on individual nursing home residents as gold standard.
METHODS
In total, ten Danish municipalities were selected. Within each Danish Region, we randomly selected one municipality reporting to Stat D (Method 1) and one not reporting where instead an algorithm created by StatD was used to discover nursing home residents (Method 2). Method 1 means that municipalities reported to Stat D whether home care has taken place in a nursing home or in a private home. Method 2 is based on an algorithm created by Stat D for the municipalities where Method 1 is not applicable. Our gold standard was the information from the local administrative system in all ten selected municipalities. Each municipality provided a list with all individuals > 65 years living in a nursing home on January 1st, 2013 as well as the central personal number. This was compared to the list of individuals >65 living in nursing home facilities in the same ten municipalities on January 1st, 2013 retrieved from StatD.
RESULTS
According to the data received directly from the municipalities, which was used as our gold Standard 3821 individuals were identified as nursing home residents. The StatD register identified 6,141 individuals as residents. Additionally, 556 of the individuals identified by the municipalities were not identified in the StatD register. Overall sensitivity for the ten municipalities in the StatD nursing home register was 0.85 (95% CI 0.84-0.87) and the PPV was 0.53 (95% CI 0.52-0.54). The municipalities for which nursing home status was based on the StatD algorithm (method 2) had a sensitivity of 0.84 (95% CI 0.82-0.86) and PPV of 0.48 (95% CI 0.46-0.50). Both slightly lower than the reporting municipalities (method 1) where the sensitivity was 0.87(95% CI 0.85-0.88) and the PPV was 0.57 (95% CI 0.56-0.59). Additionally, the sensitivity and PPV of the Stat D register varied heavily among the ten municipalities from 0.51 (95% CI 0.43-0.59) to 0.96 (95% CI 0.95-0.98) and PPV correspondingly, from 0.14 (95% CI: 0.11-0.17) to 0.73 (95% CI 0.69-0.77).
CONCLUSIONS
The overall PPV of StatD nursing home register was low and differences between municipalities existed. Even in countries with extensive nation-wide registers, validating studies should be conducted for outcomes based on these registers.
背景
许多注册研究利用永久性疗养院居民的信息。丹麦统计局（StatD）通过两种不同的间接方法识别疗养院居民，一种基于有关在疗养院进行家庭护理的报告，另一种基于 StatD 创建的算法。本研究的目的是使用专门的行政市记录验证 StatD 的疗养院登记，作为金标准。
方法
总共选择了十个丹麦市。在每个丹麦地区内，我们随机选择一个向 Stat D 报告的市（方法 1）和一个没有报告的市，而是使用 StatD 创建的算法来发现疗养院居民（方法 2）。方法 1 意味着市向 Stat D 报告是否在疗养院或私人住宅中进行了家庭护理。方法 2 基于 StatD 为不适用于方法 1 的市创建的算法。我们的金标准是所有十个选定市的当地行政系统中的信息。每个市都提供了一份在 2013 年 1 月 1 日居住在疗养院的所有>65 岁的个人名单，以及中央个人号码。这与在 2013 年 1 月 1 日从 StatD 检索到的居住在相同十个市的>65 岁的疗养院设施中的个人名单进行了比较。
结果
根据直接从市收到的数据，这些数据被用作我们的金标准，确定了 3821 人作为疗养院居民。StatD 登记册确定了 6141 人作为居民。此外，在市确定的 556 人中，在 StatD 登记册中没有发现。十个市的 StatD 疗养院登记册的总体敏感性为 0.85（95%CI 0.84-0.87），阳性预测值为 0.53（95%CI 0.52-0.54）。基于 StatD 算法（方法 2）确定疗养院身份的市的敏感性为 0.84（95%CI 0.82-0.86）和阳性预测值为 0.48（95%CI 0.46-0.50）。均略低于报告市（方法 1），敏感性为 0.87（95%CI 0.85-0.88），阳性预测值为 0.57（95%CI 0.56-0.59）。此外，StatD 登记册的敏感性和阳性预测值在十个市之间差异很大，范围从 0.51（95%CI 0.43-0.59）到 0.96（95%CI 0.95-0.98）和相应的阳性预测值，从 0.14（95%CI：0.11-0.17）到 0.73（95%CI 0.69-0.77）。
结论
StatD 疗养院登记册的总体阳性预测值较低，并且市之间存在差异。即使在拥有广泛全国性登记册的国家，也应针对基于这些登记册的结果进行验证研究。
Zotero 插件