Scanlon D P, Chernew M, Sheffler S, Fendrick A M
Department of Health Policy and Administration, Pennsylvania State University, University Park 16802-6500, USA.
Jt Comm J Qual Improv. 1998 Jan;24(1):5-20. doi: 10.1016/s1070-3241(16)30355-8.
Despite the considerable attention paid to the collection of data and the publication of health plan report cards, there is no available study on the comparability of published report cards. Ratings from seven health plan report cards publicly available in fall 1996 were compared--including those produced by major periodicals, a large national employer, a nonprofit consumer group, a health maintenance organization accreditation agency, and a consortium of employers.
Dimensions of plan performance common to the seven report cards were identified. Spearman rank correlation coefficients were computed for each pair of report cards for each of the three dimensions that were evaluated. COMPARABILITY OF REPORT CARDS: Although plan ratings tended to be positively correlated as hypothesized, the magnitude of the estimated correlation coefficients varied. For example, the estimated correlation coefficient between two periodicals' overall plan ratings was 0.48. The ranges of estimated correlations were 0.18-0.70 for preventive care (among four report cards) and 0.19-0.73 for enrollee satisfaction (among three report cards).
Discrepancies in ratings may reflect methodologic issues pertaining to the sample of health plans used, plan performance measures included, and the processes by which individual measures were aggregated to construct indices and ratings. Health plan report cards may be sending mixed signals to consumers. These inconsistencies may explain why focus group studies have found that despite the widespread indication that plan performance measures would be useful, relatively few of those who had seen such information report using it in making their plan choice. Future efforts to evaluate health plans should clearly identify assumptions, methods, normative judgments, and limitations.
尽管人们对数据收集和健康计划报告卡的发布给予了相当多的关注，但目前尚无关于已发布报告卡可比性的研究。对1996年秋季公开的七份健康计划报告卡的评级进行了比较，其中包括主要期刊、一家大型全国性雇主、一个非营利性消费者团体、一个健康维护组织认证机构以及一个雇主联盟所发布的报告卡。
确定了七份报告卡共有的计划绩效维度。针对所评估的三个维度中的每一个，计算了每对报告卡之间的斯皮尔曼等级相关系数。报告卡的可比性：尽管计划评级正如假设的那样往往呈正相关，但估计的相关系数大小各不相同。例如，两份期刊的总体计划评级之间的估计相关系数为0.48。预防保健方面（在四份报告卡中）的估计相关系数范围为0.18 - 0.70，参保人满意度方面（在三份报告卡中）的估计相关系数范围为0.19 - 0.73。
评级差异可能反映了与所使用的健康计划样本、所纳入的计划绩效指标以及将各个指标汇总以构建指数和评级的过程相关的方法学问题。健康计划报告卡可能向消费者传递了混杂的信号。这些不一致性可能解释了为什么焦点小组研究发现，尽管普遍表明计划绩效指标会有用，但看过此类信息的人中相对较少有人报告在选择计划时使用了这些信息。未来评估健康计划的工作应明确确定假设、方法、规范性判断和局限性。
