Improvement science and research methods seminar May 2021: Dr Vanessa Botan on Propensity score matching

The May 2021 Improvement science and research methods seminar featured Dr Vanessa Botan who presented on ‘An introduction to propensity score matching’, explaining the use of propensity score matching in observational studies to reduce the effects of confounding. Vanessa is a Post-Doctoral Research Associate in Medical Statistics. Her background is multidisciplinary and diverse. She has developed expertise in statistical data analyses and research methods across various fields including biomedical sciences, neuroscience, and psychology. Her ongoing academic interests include medical statistics and intervention-based research. She is working on two major NIHR funded studies and a number of other projects at CaHRU.

Vanessa’s presentation explained that propensity score matching is used as a statistical technique to address selection bias in the context of an observational study and therefore to estimate the effects of a treatment, an intervention, or a program in the absence of a randomized controlled experiment. She described the different types of matching including one-to-one (1:1) or pair matching, many-to-one (M:1) matching and full matching, matching with or without replacement, greedy, optimal and nearest neighbour matching designs.

Finally she showed how she was using this technique to analyse data in the Medical Student First Responders (MSFRs) study investigating the epidemiology and outcomes of Community First Responders who are training to be doctors. The study is a substudy of a larger NIHR funded study on Community First Responders. She used one to one matching using nearest neighbour matching with a calliper, where the absolute difference in the propensity scores of matched subjects was below a prespecified threshold (the calliper distance). Matching was done based on rurality and on time of call. The “treatment” sample of 1303 MSFRs was matched to 1303 CFRs, and 1303 non-CFR calls by ambulance service staff and these compared for important outcomes. The findings are shown in the slides below and a recording of the seminar can be seen here.

Future seminars are listed below and past seminars can be accessed via the events page here:

Loading...

Taking too long?

Reload document

| Open in new tab

Download