1 / 15

Evaluation methods – where can predictive risk models help?

Evaluation methods – where can predictive risk models help?. Adam Steventon Nuffield Trust 8 July 2013. The problem with observational studies. Intervention patients. Eligible patients. Source: Steventon et al (2012). Solutions, 1) before-after study. Solutions, 2) regression adjustment.

vernon-kidd
Download Presentation

Evaluation methods – where can predictive risk models help?

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Evaluation methods – where can predictive risk models help? Adam Steventon Nuffield Trust 8 July 2013

  2. The problem with observational studies Intervention patients Eligible patients Source: Steventon etal (2012)

  3. Solutions, 1) before-after study

  4. Solutions, 2) regression adjustment Y = f(age, number of chronic conditions, prior emergency admissions, intervention status)

  5. Solutions, 3) Matched controls Intervention patients Matched controls Eligible patients Source: Steventon etal (2012)

  6. How to select matched controls Propensity score (Rosenbaum and Rubin 1983) -Predictive risk of receiving the intervention Prognostic score (Hansen 2008) - Predictive risk of experiencing the outcome (e.g. emergency hospitalisation), in the absence of the intervention Genetic matching (Sekhon and Grieve 2012) - computer-intensive search algorithm

  7. Advantages / disadvantages Disadvantage – only allows for observed variables But Matching as ‘data pre-processing’ – reduces dependence of estimated intervention effects on regression model specification Intuitive? Good for routine monitoring – once controls found, data can be updated

  8. Overcoming regression to the mean using a control group Start of intervention

  9. Overcoming regression to the mean using a control group Start of intervention

  10. Overcoming regression to the mean using a control group Start of intervention

  11. Overcoming regression to the mean using a control group Start of intervention

  12. Solutions, 4) regression discontinuity Winningthe next election Fraction of votes awarded to Democrats in the previous election Source: Lee and Lemieux(2009)

  13. What is being done at the moment?Telehealth studies in Pubmed, 2006-2012 Source: Steventon,Krief and Grieve (work in progress)

  14. References Lee DS, Lemieux T. Regression discontinuity designs in economics. 2009. Available from: http://www.nber.org/papers/w14723.pdf?new_window=1 Sekhon JS, Grieve RD. A matching method for improving covariate balance in cost-effectiveness analyses. Health economics 2012;21:695–714. Rosenbaum P, Rubin D. The central role of the propensity score in observational studies for causal effects. Biometrika 1983;70:41–55. Hansen BB. The prognostic analogue of the propensity score. Biometrika 2008;95:481–8. Steventon A, Bardsley M, Billings J, Georghiou T, Lewis GH. The role of matched controls in building an evidence base for hospital-avoidance schemes: a retrospective evaluation. Health services research 2012;47:1679–98.

  15. adam.steventon@nuffieldtrust.org.uk

More Related