Glied S
School of Public Health, Columbia University, New York, NY 10032, USA.
Am J Public Health. 1996 Dec;86(12):1723-8. doi: 10.2105/ajph.86.12.1723.
This study attempted to assess (1) the accuracy of estimates of the indirect cost of illness and death computed with the human capital (forgone earnings) method and (2) the sensitivity of these estimates to key assumptions and parameters.
The study used data from the annual Current Population Surveys of 1964 through 1988 to compare the earnings experience of cohorts of White men aged 18 through 65 with predictions made with the human capital method. The study then assessed the sources and magnitude of the observed differences.
Predictions of forgone earnings can be as much as 18% greater or 20% smaller than actual earnings, under identical assumptions, depending on the data used. While in most cases errors are quite small, alternative, equally plausible estimates of forgone earnings may differ by as much as 50%. Estimates differed mainly because of (1) the cross section chosen to make the predictions and (2) assumptions about future earnings growth. However, other factors, such as cohort size, also contributed to variation.
Researchers and policymakers should be very careful in making and interpreting estimates of the indirect cost of illness and death.
本研究试图评估：（1）采用人力资本（收入损失）法计算的疾病和死亡间接成本估计值的准确性；（2）这些估计值对关键假设和参数的敏感性。
本研究使用了1964年至1988年年度当前人口调查的数据，比较18至65岁白人男性队列的收入经历与用人力资本法做出的预测。然后，本研究评估了观察到的差异的来源和程度。
在相同假设下，根据所使用的数据，收入损失预测值可能比实际收入高出18%或低20%。虽然在大多数情况下误差很小，但收入损失的其他同样合理的估计值可能相差多达50%。估计值存在差异主要是因为：（1）用于进行预测的横截面；（2）关于未来收入增长的假设。然而，其他因素，如同龄组规模，也导致了差异。
研究人员和政策制定者在进行和解释疾病和死亡间接成本估计时应非常谨慎。
