Estimation of treatment effects
Observational datasets often provide sample sizes that are far beyond the scope of any clinical trial. They also provide the opportunity to assess the effectiveness of an intervention in practice, an important consideration for areas such as surgery where routine practice may not match the standards achieved by pioneers. Hence, observational datasets provide an opportunity to assess the effectiveness of interventions and clinical practice in the absence of clinical trial data or where trial data is inconclusive. The robustness of any estimate of effectiveness rests on appropriate control for differences in patient populations.
Our research has applied GenMatch a genetic matching technique that improves on propensity score matching to compare quality of life in patients receiving cemented, cementless or hybrid Total Hip Replacement using data from collected in national Patient Reported Outcome Measures (PROMs) programme. We applyied three different matching techniques to evaluate the effectiveness of extracorporeal membrane oxygenation (ECMO) to support patients with pandemic H1N1 influenza A (pH1N1)-related ARDS using data from 192 acute hospitals in the UK. Our research has also investigated the risk-adjusted outcomes, costs and lifetime cost-effectiveness of alternative care locations following acute traumatic brain injury (RAIN) using large data collected from 67 adult critical care units in the UK.