Observational studies like the Dialysis Outcomes Practice Patterns Study (DOPPS) allow scientists to see practice variations among countries which can point the way towards better patient care, the conference heard.

Unit medical director of Michigan Dialysis Services and assistant professor of Medicine at University of Michigan, USA, Dr Sean Leavey explained the aim of DOPPS and discussed how it stacked up against other types of trials.

"For some years we have known that there are differences in mortality for haemodialysis patients in different countries," he said.

"And we're all dissatisfied with the survival and poor outcomes overall for dialysis patients.

"So, it's important to know if there are international differences in survival. It's important because, if we can understand what are the international differences in practices that influence patient survival, then we can perhaps do something to improve the lives of our patients," he stressed.

The prospective observational study DOPPS began in 1996 and is currently running in seven countries. It began in the USA, then spread through Europe, including the UK, to Japan. It will soon expand to 13 countries.

The countries currently in the study account for 80% of the world's haemodialysis population.

DOPPS collects baseline data and follows it up every three months.

"The purpose of DOPPS is not to compare countries but to identify the variations to get clues to better management of patients," Dr Leavey said.

"DOPPS collects major patient outcomes in terms of survival, in terms of hospitalisation, and cause-specific hospitalisation, in terms of vascular access and in terms of quality of life," he said.

A major advantage of an observational study like DOPPS was that it allowed medicine to test some parameters which could not be tested in a clinical-trial setting because of feasibility or ethical limitations.

The main disadvantage was that observational studies were not that strong in terms of causal influence. They could, however, describe associations and generate hypotheses, and they could infer causation, he explained.

"But they can also address, in a timely fashion, a lot of hypotheses that might not ever be tested in randomised clinical trials," Dr Leavey said.

Dr Leavey explained:

		"If an observational study shows a important effect that's biologically plausible;
		and if the study is prospective, so that an exposure precedes an outcome;
		and if the study can illustrate a dose response between an exposure and an outcome;
		and if a number of observational studies in different populations come up with the same observations;
		then you have a situation, while not evidence to the level of the clinical trial, is evidence that is sufficient to influence practice patterns."

To explain how DOPPS data could be used he described observations he had made about the effects of body size on haemodialysis-patient survival.

Contrary to what one might expect, mild to moderate obesity, as expressed by body mass index, improves survival in most USA and European cohorts of haemodialysis patients.

But why this should be the case is unclear.

"More energy reserves might help patients withstand the cumulative stresses of end-stage renal disease. Perhaps cardiovascular disease and other risks factors just outweigh the obesity risk factor. Or perhaps there is still some selection bias we haven't accounted for," he said.

"Observational studies describe the actual treatment practices. They can test numerous hypotheses and they can address the questions that are not accessible to be addressed by clinical trials," Dr Leavey continued.

"Findings can lead to new trials and, while correlation and association from observational research does not prove causation. If you multiply, adjust for confounding factors, take the necessary adjustments into account, and you see consistency across different studies, then you begin to get data which can provide reasonable inferences for practising physicians," he concluded.

If you would like more information about DOPPS then email info@nephronline.org

(back)