MAP Centre for Urban Health Solutions, Unity Health Toronto, Toronto, Ontario, Canada.
ICES Western, London Health Sciences Research Institute, London, Ontario, Canada.
J Clin Epidemiol. 2024 Aug;172:111430. doi: 10.1016/j.jclinepi.2024.111430. Epub 2024 Jun 14.
Conducting longitudinal health research about people experiencing homelessness poses unique challenges. Identification through administrative data permits large, cost-effective studies; however, case validity in Ontario is unknown after a 2018 Canada-wide policy change mandating homelessness coding in hospital databases. We validated case definitions for identifying homelessness using Ontario health administrative databases after introduction of this coding mandate.
We assessed 42 case definitions in a representative sample of people experiencing homelessness in Toronto (n = 640) from whom longitudinal housing history (ranging from 2018 to 2022) was obtained, and a randomly selected sample of presumably housed people (n = 128,000) in Toronto. We evaluated sensitivity, specificity, positive and negative predictive values, and positive likelihood ratios to select an optimal definition, and compared the resulting true positives against false positives and false negatives to identify potential causes of misclassification.
The optimal case definition included any homelessness indicator during a hospital-based encounter within 180 days of a period of homelessness (sensitivity = 52.9%; specificity = 99.5%). For periods of homelessness with ≥1 hospital-based healthcare encounter, the optimal case definition had greatly improved sensitivity (75.1%) while retaining excellent specificity (98.5%). Review of false positives suggested that homeless status is sometimes erroneously carried forward in healthcare databases after an individual transitioned out of homelessness.
Case definitions to identify homelessness using Ontario health administrative data exhibit moderate to good sensitivity and excellent specificity. Sensitivity has more than doubled since the implementation of a national coding mandate. Mandatory collection and reporting of homelessness information within administrative data present invaluable opportunities for advancing research on the health and healthcare needs of people experiencing homelessness.
针对无家可归者进行纵向健康研究具有独特的挑战。通过行政数据进行识别可以实现大规模、具有成本效益的研究；然而，在 2018 年加拿大全国范围内的政策变更要求在医院数据库中对无家可归者进行编码之后，安大略省的病例有效性尚不清楚。在引入该编码要求后，我们使用安大略省健康行政数据库验证了用于识别无家可归者的病例定义。
我们评估了多伦多代表性的无家可归者样本（n=640）中 42 种病例定义，该样本从 2018 年到 2022 年期间获得了纵向住房史，以及多伦多随机选择的假定有住房的人样本（n=128000）。我们评估了敏感性、特异性、阳性和阴性预测值以及阳性似然比，以选择最佳定义，并将由此产生的真阳性与假阳性和假阴性进行比较，以确定潜在的分类错误原因。
最佳病例定义包括在无家可归期内的任何 180 天内，在医院就诊期间的任何无家可归指标（敏感性=52.9%；特异性=99.5%）。对于无家可归期内有≥1 次医院就诊的情况，最佳病例定义的敏感性大大提高（75.1%），同时保持了极好的特异性（98.5%）。对假阳性的审查表明，在个人脱离无家可归状态后，无家可归状态有时会错误地在医疗保健数据库中延续。
使用安大略省健康行政数据识别无家可归者的病例定义具有中等至良好的敏感性和极好的特异性。自实施国家编码要求以来，敏感性提高了两倍多。在行政数据中强制收集和报告无家可归者信息为研究无家可归者的健康和医疗需求提供了宝贵的机会。
