Lamiere_edit.qxp 8/2/08 12:40 Page 58
Dialysis
The Status of Peritoneal Dialysis in Europe – An Update
a report by
Norbert Lameire
Emeritus Professor of Medicine, Renal Division, University Hospital, Ghent
We recently described some of the medical and non-medical reasons why begin dialysis, the investigators asked patients to be randomised to HD or
the utilisation of peritoneal dialysis (PD) in Europe has been highly PD. Of 773 patients who fulfilled entry criteria, 735 had a preference for
variable from country to country and why its utilisation has fluctuated either HD or PD (95%) and refused to participate; only 38 provided
with time.
1
In those countries where PD was relatively popular in the consent for random assignment, making the results of the study difficult
1990s, a steady decline or at least no further growth has been noted.
2
to interpret. Of those with a preference, 52% preferred to start dialysis
This first paper reviewed the literature until 2005.
1
The purpose of this on HD and 48% preferred PD. The latter results are striking compared
paper is to update the former one with references up to July 2007. with the 8–10% of patients on PD in many European countries and the
US and is highly suggestive that many patients are often not given a
In the years after its introduction in 1978,
3
continuous ambulatory choice regarding modality. In fact, a randomised trial of sufficient power
peritoneal dialysis (CAPD) as a home-based renal replacement therapy comparing survival in HD and PD will remain a utopian aspiration.
7
(RRT) was quickly and widely applied. By the mid- to late 1980s, PD
utilisation was more than 35% in Canada and 15% in the US. Other A very recent observational study from the Dutch End-Stage Renal Disease
countries, including the UK, Hong Kong and Mexico, had much higher PD (ESRD) Registry (RENINE) compared the survival of Dutch HD and PD
utilisation rates, and worldwide the growth rate of PD outstripped that of patients.
8
Definite treatment assignment at day 91 and an intention-to-
haemodialysis (HD). However, even in the first years after its introduction, treat analysis, censoring for transplantation, was performed. Cox
wide differences in the utilisation of PD as a dialysis modality became regression models using 16,643 patients adjusted for age, gender, primary
apparent. Some countries, including Japan, Germany, Belgium and renal disease, centre of dialysis and year of start of RRT were used. These
France, never had more than 10% of prevalent dialysis patients on PD. also included several interaction terms. To account for time dependency,
Over the last few years, particularly in the US, Canada and some the analysis was stratified into three time periods: >3–6, >6–15 and >15
European countries, a decline in the utilisation of PD has been observed.
2
months. For the first period, the mortality hazard ratio (HR) of PD
Both medical and non-medical (mostly financial and reimbursement) compared with HD patients was significantly lower for 40-year-old non-
reasons explain why these wide differences between European countries diabetics, but increased with age and presence of diabetes for 70-year-old
in the utilisation of PD have persisted and why, at least in some countries, patients with diabetes as primary renal disease. The HRs of the second
the already low prevalence of PD patients is further declining. A major period were generally higher. After 15 months, the HR was 0.86 (95% CI
medical reason in many countries for limited PD utilisation is concern over 0.74–1.00) for 40-year-old non-diabetics and 1.42 (95% CI 1.23–1.65) for
the lower survival rate of the PD patient compared with the patient 70-year-old patients with diabetes as primary renal disease. It thus appears
treated by HD. that the survival advantage for Dutch PD compared with HD patients
decreases over time, with age and in the presence of diabetes as primary
Comparative Survival Data – Peritoneal disease. A similar interaction between PD patient survival and presence of
Dialysis versus Haemodialysis diabetes and younger age was found in earlier studies.
9-11
It is currently well-known that numerous methodological biases hamper
a direct comparison of PD and HD outcomes.
4
The many reasons why an In observational studies, precise control of prescription and delivery of
apparent survival benefit of HD over PD may be misleading has been the dialysis is limited and the demonstration of a causal relationship between
subject of a recent series of short papers in the Seminars of Dialysis.
5
Only care and outcome is not possible. Ideally, data should be extensive and
one randomised trial compared HD and PD in terms of patient survival.
6
collected prospectively. In this way, registries have the advantage of
In this study, after explaining the two modalities to patients about to including large populations. However, the limited potential to control
possible confounding factors limits the clinical relevance of results. For
example, under-reporting of co-morbid conditions in registries may
Norbert Lameire is Emeritus Professor of Medicine at the
produce false results.
12
Renal Division of University Hospital Ghent in Belgium. He is
also Doctor Honoris Causa at the Kaunas Medical Academy
in Lithuania and an Honorary Member of the International
Other limitations can be seen in previous studies comparing modes of
Society of Nephrology (ISN). Dr Lameire’s research interests
include basic research and clinical topics, particularly clinical
dialysis. For instance, most published studies ignored outcomes during
acute renal failure, peritoneal dialysis and organisational and the first three months of dialysis. However, a recent study
13
demonstrated
economic aspects of chronic renal replacement therapy and
that, for incident dialysis patients, this period is critical in terms of choice
transplantation. He graduated from the University of Ghent,
Faculty of Medicine in 1965.
of dialysis modality and mortality. On the other hand, due to differences
in patients and healthcare systems between countries, it is questionable
E: norbert.lameire@ugent.be
whether studies can be generalised.
14
58 © TOUCH BRIEFINGS 2007