Wright D D, Kane R L, Kronhaus A, Woolley F R, Altman D
J Community Health. 1982 Summer;7(4):284-91. doi: 10.1007/BF01318960.
This study explored the usefulness of archival data in predicting rural health care utilization. A regression model was used to see how well observed utilization for local populations could be predicted by calculating expected values in advance from age- and sex-specific national rates applied to local age and sex profiles. Although the correlation between observed and expected utilization was reasonably high (r = 0.92), an attempt was then made to improve prediction by considering other data that do not require independent collection. These archival data included indicators of historic utilization (local Medicaid payments, the percentage of births to county residents occurring in the mother's county of residence, percentage of children immunized, and infant mortality) and services already available. Observed utilization data were obtained by surveys in eight rural counties, and the predictor was tested on three additional rural communities. A predictor equation that added to the expected utilization only one variable (the percentage of births to county residents occurring in the mother's county of residence) was found to account for approximately 95% of the variance in observed utilization. This predictor is recommended for planners who need convenient, low-cost market feasibility estimates for proposed project sites and a way to establish intermediate goals or incentives during early project development.
本研究探讨了档案数据在预测农村医疗保健利用方面的有用性。使用回归模型来查看通过根据适用于当地年龄和性别分布的特定年龄和性别的全国比率预先计算期望值，能在多大程度上预测当地人群的实际医疗保健利用率。尽管观察到的利用率与预期利用率之间的相关性相当高（r = 0.92），但随后尝试通过考虑其他无需独立收集的数据来改进预测。这些档案数据包括历史利用率指标（当地医疗补助支付情况、本县居民在其母亲居住县出生的比例、儿童免疫接种率以及婴儿死亡率）和已有的服务。观察到的利用率数据通过对八个农村县的调查获得，并在另外三个农村社区对预测指标进行了测试。发现一个仅在预期利用率基础上增加一个变量（本县居民在其母亲居住县出生的比例）的预测方程能够解释观察到的利用率中约95%的方差。对于那些需要为拟议项目地点提供便捷、低成本的市场可行性估计，以及在项目早期开发阶段建立中间目标或激励措施的规划者，推荐使用这个预测指标。
