• Jennifer L Cerully, Andrew M Parker, Lise Rybowski, Mark Schlesinger, Dale Shaller, Rachel Grob, Melissa L Finucane, and Steven C Martino.
• RAND Corporation, 4570 Fifth Avenue, Suite 600, Pittsburgh, PA, USA. jcerully@rand.org.
• J Gen Intern Med. 2019 Feb 1; 34 (2): 243-249.

BackgroundPublic reports on healthcare quality typically include complex data. To lower the cognitive burden of interpreting these data, some report designers create summary, or roll-up, measures combining multiple indicators of quality into one score. Little is known about how the availability of roll-ups affects clinician choice.ObjectiveTo determine how presenting quality scores at different levels of aggregation affects patients' clinician choices.DesignWe conducted a simulated clinician-choice experiment, randomizing participants to three versions of a public reporting website and comparing their clinician choices. One version aggregated all clinician-level quality measures into roll-ups, the second provided disaggregated (drill-down) scores only, and the third offered both roll-ups and drill-downs.ParticipantsFive hundred fifty panelists drawn from a probability-based Internet panel.Main MeasuresWe assessed the amount of effort participants exerted by tracking the length of time spent on the website and the number of concrete actions taken on the website (e.g., clicking items). We evaluated decision quality by measuring whether participants selected a clinician who performed more poorly than others and incongruence between participants' stated preferences for dimensions of quality and their chosen clinician's performance on those dimensions.Key ResultsParticipants seeing drill-downs alone (mean = 14.9) or with roll-ups (mean = 19.2) took more actions than those who saw roll-ups alone (mean = 10.5) (ps < 0.05). However, participants seeing only drill-downs made poorer choices than those who saw roll-ups alone or with drill-downs. More participants seeing drill-downs chose a clinician who was outperformed (36.3% versus 23.4% [roll-up] and 25.6% [drill-down + roll-up], ps < 0.05) and made choices incongruent with stated preferences (51.2% versus 45.6% [roll-up] and 47.5% [drill-down + roll-up], ps < 0.05). The distinction between roll-up and drill-down was somewhat stronger for sicker participants.ConclusionsOur results suggest that roll-ups in healthcare quality reports, alone or as a complement to drill-downs, can help patients make better decisions for themselves.

25 keywords...

hide…