Using routine data to complement and enhance the results of randomised controlled trials
Lewsey J D, Leyland A H, Murray G D, Boddy F A
Record ID 32000000906
The project was undertaken to explore how routinely assembled hospital data might complement or supplement RCTs to evaluate medical interventions: - in contexts where RCTs are not feasible for defining the context and design of an RCT - for assessing whether the benefits indicated by RCTs are achieved in wider clinical practice.
Authors' results and conclusions:
The uses of routine data in these contexts had strengths and weaknesses. The SAH study suggested a means of assessing outcomes and survival rates following haemorrhage, which could have value in informing the design of more precise trials and in evaluating changes in outcome following the introduction of new treatments such as embolisation. However, the potential of the data was not realised because their scope and content were insufficient. For example, lack of data on the time of onset of symptoms and patients conditions at hospital admission made it difficult to establish the link between timing of surgery and the outcome, and there was insufficient information on patients conditions at discharge to enable a comparison of outcomes.
The prostatectomy study was able to address questions not answered by RCT literature because the large number of cases it included allowed exploration of subgroup effects. The data indicated that younger patients and those with previous hospital admissions for cardiovascular, respiratory or ischaemic disease were more likely to have TURP, suggesting that these may influence treatment decisions. However, the risk of re-operation was higher in patients who initially underwent TURP, and, although mortality at 90 days was higher in patients who had OPEN initially, the difference seen from the routine data was not significant at 1 and 5 years. The records for this study were more satisfactory than for the SAH study. However, lack of data on the severity and complexity of patients conditions limited the potential of the data.
The study of coronary revascularisation supported findings of the earlier meta-analysis, but with more prolonged follow-up and a broader population. Of the three studies, the data for this study were the most satisfactory, although lack of precise information on the complexity and severity of patients conditions made it difficult to establish the full extent of subgroups. Patients who had an initial PTCA were more likely to require re-intervention than those who had CABG and, as expected, there was a lower rate of death and myocardial infarction (MI) in the RCT-like subgroup than in patients excluded from this sample. Using the routine data, the rates of death and MI at 1 year were significantly higher in patients who had an initial CABG, whereas this difference was not significant in the RCTs, but the difference was not significant in both at 5 years. A Bayesian comparison of the two interventions illustrated that Bayesian analyses can provide a link between RCTs, which are unbiased by design but may not reflect real populations, and routine data, which reflect reality but may be biased. This can facilitate better evaluations of outcomes associated with new technologies.
In general, linked data have value in two main ways. First, they relate to complete populations of cases and might thus clarify issues relating to patient selection. Second, by linking episodes of care to each other and to deaths, it is possible to gain information about prior medical histories, longer-term outcomes and the place of treatment, providing a context for more focused RCTs and multicentre comparisons of new techniques and their outcomes. Both of these are probably a prerequisite for comparable work; however, the shortcomings in the content and quality of current data limit these applications. Indeed, three further intended studies laparoscopic versus open cholecystectomy, modes of treatment for breast cancer, and colorectal surgery by specialist versus generalist surgeons proved impossible to undertake because of inadequacies in the routine data.
English language abstract:
An English language summary is available
- Randomized Controlled Trials as Topic
NIHR Health Technology Assessment programme
NIHR Journals Library, National Institute for Health Research, Evaluation, Trials and Studies Coordinating Centre, Alpha House, University of Southampton Science Park, Southampton SO16 7NS, UK
2009 Queen's Printer and Controller of HMSO