Schang Laura, Hynninen Yrjänä, Morton Alec, Salo Ahti
Department of Management, London School of Economics and Political Science, London, United Kingdom; Department of Health Services Management, Ludwig-Maximilians-Universität München, Schackstraße 4, 80539 Munich, Germany.
Department of Mathematics and Systems Analysis, Systems Analysis Laboratory, Aalto University School of Science, Aalto, Finland.
Soc Sci Med. 2016 Aug;162:59-67. doi: 10.1016/j.socscimed.2016.06.026. Epub 2016 Jun 16.
Although composite indicators are widely used to inform health system performance comparisons, such measures typically embed contentious assumptions, for instance about the weights assigned to constituent indicators. Moreover, although many comparative measures are constructed as ratios, the choice of denominator is not always straightforward. The conventional approach is to determine a single set of weights and to choose a single denominator, even though this involves considerable methodological difficulties. This study proposes an alternative approach to handle incomplete information about an appropriate set of weights and about a defensible denominator in composite indicators which considers all feasible weights and can incorporate multiple denominators. We illustrate this approach for comparative quality assessments of Scottish Health Boards. The results (displayed as ranking intervals and dominance relations) help identify Boards which cannot be ranked, say, worse than 4th or better than 7th. Such rankings give policy-makers a sense of the uncertainty around ranks, indicating the extent to which action is warranted. By identifying the full range of rankings that the organizations under comparison may attain, the approach proposed here acknowledges imperfect information about the "correct" set of weights and the appropriate denominator and may thus help to increase transparency of and confidence in health system performance comparisons.
尽管综合指标被广泛用于指导卫生系统绩效比较，但这类指标通常包含有争议的假设，例如关于赋予构成指标的权重。此外，尽管许多比较指标被构建为比率形式，但分母的选择并非总是一目了然。传统方法是确定一组单一的权重并选择一个单一的分母，即便这涉及相当大的方法学难题。本研究提出了一种替代方法，用于处理综合指标中关于一组合适权重和一个合理分母的不完整信息，该方法考虑了所有可行的权重并且可以纳入多个分母。我们将此方法用于苏格兰卫生委员会的比较质量评估。结果（以排名区间和优势关系呈现）有助于识别那些无法排名的委员会，比如不能排在第4名之后或第7名之前。这样的排名让政策制定者了解排名周围的不确定性，表明有必要采取行动的程度。通过确定被比较组织可能获得的所有排名范围，本文提出的方法承认了关于“正确”权重集和合适分母的信息不完整，因此可能有助于提高卫生系统绩效比较的透明度和可信度。
