Faith, Trust and Pixie Dust professional judgement in healthcare, is it “broken” and can we fix it? Professor Carl Thompson Department of Health Sciences, University of York Trust in professionals: UK. Ford J BMJ 2007;335:465 ©2007 by British Medical Journal Publishing Group Trust: USA % “high or very high” 90 80 70 60 50 40 doctors nurses 30 20 10 0 http://www.gallup.com/poll/1654/honesty-ethics-professions.aspx#4 Judgement and Variation 65% had tonsillectomy 46% for tonsillectomy 1000 11yr olds 35% “to be seen” 45% for tonsillectomy 54% “to be seen” American child health association 1934 Only 65 kids left; stop study ran out of doctors Fig 1 Accessing stroke rehabilitation Source: Archives of Physical Medicine and Rehabilitation 2010; 91:788-793 (DOI:10.1016/j.apmr.2009.11.028 ) “machines that go ping” Cognitive continuum (cf. Hammond 1996) Cameron HM, McGoogan E. J Pathol. 1981 Apr;133(4):273-83 Some problems • Heuristics or biases that result (Kahneman, Tversky et al.) – e.g. availability, overconfidence, anchoring, representativeness. • The alternative: fast and frugal (Gerd Gigerenzer) • Do our common ”solutions” make it worse? – Groupthink Nutrition (ANS) • • • • • • • • 27 doctors nurses and nutritionists Large Teaching Foundation Trust 54 (49) “cases” judged NICE guidance as a “rule” 74% rule aware, 70% using, 63% confident 1323 judgements Feed (clinicians) = 693 (Guidelines) = 270 Accuracy/sensitivity .45 (below chance) Journal of Human Nutrition and Dietetics Volume 25, Issue 5, pages 427-434, 30 MAY 2012 Initiating artificial nutrition support: a clinical judgement analysis Journal of Human Nutrition and Dietetics Volume 25, Issue 5, pages 427-434, 30 MAY 2012 DOI: 10.1111/j.1365-277X.2012.01260.x http://onlinelibrary.wiley.com/doi/10.1111/j.1365-277X.2012.01260.x/full#f2 Calibration curves of easy and difficult judgments under no time pressure/time pressure. Yang et al. BMC Medical Informatics and Decision making 2012 12:113 doi:10.1186/1472-6947-12-113 Paper vs. nurses (250 from 4 countries) Mews >4 .69 Sensitivity P (judge+ | risk+) >11 .65 >4 .57 1 – specificity (False positive rate) Can we fix it? • “Support” – Ask a friend or friends – Evidence based decision making/training? – More machines that go ping (CDSS)? • Education and Training (cognitive forcing, feedback) – Something about professional attitude to “rules”? Clinical vs. actuarial “judgements” • Neurosis vs. psychosis: rule = 70%; clinicians 50% to 67% • Dx of progressive brain dysfunction based on intellectual testing: rule = 83%; mixed bag of clinicians 63% (inexperienced) to 58% (experienced). Knowledge of rule results = 68% and 75%... • Predicting survival time: experts (no correlation); rule = modest but significant • Repeat…ad nauseum RM Dawes, D Faust, PE Meehl: Science 31 March 1989: Resistance? • Don’t know about the evidence? • Don’t know there is a problem? • Don’t recognise it (“mental healthcare discourse anyone?”) • Russian roulette and individualisation of probabilities • Don’t think its as good as ME and my recall – average drivers • “You can’t handle the truth” (certainty pays the bills) Its not “broken” but needs fettling* • If you know something about judgements and decisions (structure, context) then you can predict and improve them… it’s not r****t **i**ce! • CDSS, education and training for decision making… need cultural support • Guidance (MRC, 2008) for developing “good” interventions… we need to use them www.mrc.ac.uk/complexinterventionsguidance • Where good tools exist then implementation (science) required * A northern verb: to arrange, tidy up, put in order