NDT Advance Access published online on August 1, 2008
Nephrology Dialysis Transplantation, doi:10.1093/ndt/gfn442
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
From registry data collection to international comparisons: examples of haemodialysis duration and frequency
1 French ESRD Registry REIN, Agence de la biomedecine, Saint-Denis La Plaine, France 2 University Hospital of Maastricht, Maastricht, The Netherlands 3 Finnish Registry for Kidney Disease, Helsinki, Finland 4 Norway ESRD Registry, Oslo, Norway 5 Vodnjanska Clinical Center, Macedonian ESRD Registry, Skopje, Macedonia 6 Slovenian ESRD Regsitry, Ljubljana, Slovenia 7 French-Speaking Belgium ESRD Registry, Bruxelles, Belgium 8 OEDTR, Austrian ESRD-Registry, Wels, Austria 9 The Spanish Society of Nephrology, Santander, Spain 10 Department of Medical Informatics, ERA–EDTA Registry, Academic Medical Center, University of Amsterdam, Amsterdam, The Netherlands
Correspondence and offprint requests to: Cécile Couchoud, Coordination Nationale du REIN, Agence de la Biomédecine, 1 Avenue du Stade de France, 93212 Saint Denis La Plaine, Cedex, France. Tel.: +33-1-55-93-64-67; Fax: +33-1-55-93-69-36; E-mail: cecile.couchoud{at}biomedecine.fr
 |
Abstract |
|---|
 Top Abstract IntroductionMethodsResultsDiscussionReferences |
|---|
Background. The purpose of this study was to investigate haemodialysis (HD) dose practice patterns in different European countries in the light of the European Best Practice Guidelines (EBPG) and to study the associations of patient characteristics and country with weekly dialysis duration.
Methods. Renal registries in Europe were asked to contribute to the study with individual patient data on weekly HD duration, number of HD sessions a week and last measured Kt/V. Additional items were age, sex, date of first renal replacement therapy (RRT), dry weight, height, HD modality, HD technique, diabetes status and vascular access type. Multivariate logistic regression was used to study the probability of receiving HD for <12 h per week.
Results. Seven registries contributed data on 26 136 patients on HD on 31 December 2005. Eighty-three percent of the patients received HD for at least 12 h per week as recommended by the EBPG (range 49.0–97.3% across countries). Multivariate analysis showed significant differences across countries concerning the risk of receiving <12 h. Other risk factors included age (older), sex (female), BMI (low) and duration of RRT (shorter). Diabetes was associated with longer total HD duration.
Conclusion. This study shows a great international variability in weekly HD duration and some discrepancies between current practices and the EBPG. It also points out the difficulty of obtaining and comparing Kt/V values under current registry practices.
Keywords: Best Practice Guidelines; clinical performance indicator; ESRD registry; haemodialysis dose; haemodialysis schedule
|
Introduction |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionReferences |
|---|
The aim of the QUality European STudies (QUEST) initiative is to stimulate the development of reliable databases that allow clinical results to be compared directly with the goals of the European Best Practice Guidelines (EBPG), a comparison that must underlie any quality improvement programme [1]. This study comparing clinical performance indicators between renal registries in Europe is the first report to result from the QUEST initiative.
Except for the DOPPS centre studies, comparisons of European data on haemodialysis (HD) dose measurement and its delivery have been sparse until now, and little is known about the implementation of the EBPG. Traditionally, HD dose has been quantified in terms of urea kinetics. One index of HD dose is the fractional urea clearance, commonly expressed as Kt/Vurea. Debate continues about the optimal number and length of HD sessions [2,3]. Three HD sessions a week are most common. Earlier reports showed that twice weekly schedules are sometimes used for elderly patients, women, patients with low body mass index (BMI) and patients with renal residual function (RRF) at the beginning of renal replacement therapy (RRT) [4], but they are not recommended by current international guidelines [5] or in patients with urea clearance <2 mL/min [6]. In recent years, daily HD (defined as 
5 sessions per week, day or night) was developed with the aim of improving HD outcome [7]. Session length was shortened somewhat in the 1970s, with improvement of the dialysis techniques (membrane, fistula, convective techniques, etc.), but little consideration was given to potential risks, such as impaired blood pressure control and cardiac performance, malnutrition and hyperphosphataemia. More recently, the EBPG recommended at least 12 h of HD a week [5].
The purpose of this study, initiated by the QUEST working group on dialysis adequacy, was to investigate HD dose practice patterns in several European countries in the light of the EBPG and to study the relation of HD dose to patient characteristics and country.
|
Methods |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionReferences |
|---|
This observational study is based on existing registry data. National and regional registries in Europe were asked to contribute individual patient HD dose data to the study, including preferably all, but at least one of the following three indicators: total weekly duration of HD, number of HD sessions per week and last measured Kt/V. Other items requested were age on 31 December 2005, sex, date of first RRT, weight, height, HD modality (full-care centre, limited-care centre, home care), HD technique (HD-low flux with UF coeff. <15 mL/mm Hg/h, HD-high flux with UF coeff.
15 mL/mm Hg/h, HD unspecified, HDF-low volume at <15 L/session, HDF-high volume at 15 L/session, HDF unspecified, or other convective techniques), diabetes status (listed as comorbid condition) and type of vascular access (native fistula, graft, untunnelled central venous catheter or tunnelled central venous catheter). When comorbid condition information was not available, diabetes status was defined by the presence of diabetic nephropathy as the primary renal disease. All patients aged 15 years and older who were receiving HD on 31 December 2005 in the participating registries were included. The QUEST working group asked 38 registries covering 29 European countries to participate in this study. Seven registries were able and willing to contribute the individual patient data requested: Austria, French-speaking Belgium, Macedonia, Finland, France, Norway and Slovenia. The data for Macedonia included 14 of 18 centres (76% of all patients) and for France 13 of 24 regions (50% of all French patients but 100% of the patients in those regions). The other registries included data for 100% of their HD patients.
Missing data included: BMI for Slovenia, HD technique for Finland and vascular access for Austria. Diabetes status was assessed by the primary renal disease for Austria and Macedonia. The percentage of missing data in the registries for which those data were available was 25% for BMI, 9% for diabetes status, 2% for HD modality, 2% for HD technique and 54% for vascular access. In addition, the number of sessions per week and their length were not available for Austria; in the other six registries, the percentage of missing data for these variables was 8% and 9%, respectively. The Kt/V was not available for Austria, Slovenia or Norway; the percentage of missing data in the other four registries was 62%.
For this analysis, the total weekly duration and the mean length of HD sessions as well as the number of sessions were categorized into groups. All data were analysed with SAS 9.1 (SAS, Inc., Cary, NC, USA). Categorical variables were described as frequencies, and continuous variables as medians with their ranges. Bivariate analyses explored the relations of patients, treatment characteristics and countries with overall and country HD dose variables. Differences between subgroups were compared by 
2 tests. Multivariate logistic regression was used to study the associations—in patients with complete data—between patient characteristics and the probability of receiving HD for <12 h per week. Because of the many missing values and the large variety of methods used to measure Kt/V, we did not perform an in-depth analysis of the Kt/V data.
|
Results |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionReferences |
|---|
Table 1 summarizes the characteristics of the 26 136 HD patients of these seven registries. The median age was 68 years (range: 15–99). The median time since first RRT was 3 years (maximum: 40 years). There were statistically significant differences across countries for all of these variables (P < 0.0001).
|
The median number of HD sessions a week was 3 (range: 1–7). More than 90% of patients had three sessions a week in all countries except Norway, where 21% of the patients had fewer than three (Table 2). More than 0.5% of the patients in Finland, France and Norway had daily HD. Patients who most frequently had fewer than three sessions a week were the youngest and oldest age categories, women, patients with lower BMI, shorter time on RRT, treated in full-care centres or with convective techniques, and with a graft or a catheter. Daily HD was more frequent for younger patients and for patients treated at home. Diabetes status was not associated with the number of sessions.
|
Median session length was 4 h (range: 1–12) and differed significantly between countries; 21% of the patients in French-speaking Belgium received short HD (defined as
3 h) whereas 2.6% of the patients in Finland were treated with long HD (defined as 6 h) (Table 3). Elderly patients, women, patients with lower BMI, shorter time on RRT, conventional HD, a graft or catheter, or who were treated at home were more frequently treated with short HD. Long HD was more frequent in middle-aged patients, patients with higher BMI, diabetes or longer time on RRT, or who were treated at home.
|
Figure 1 shows the distribution of the number of sessions a week and the length of a session.
|
The distribution of the total weekly HD duration differed between countries (Table 4). While most patients in each country received 12 h per week, 44% of the patients in Finland and 55% of those in Slovenia received >12 h per week. In contrast, 51% of the patients in French-speaking Belgium received <12 h a week. Among the patients with <12 h, 33% were treated three times a week for 3.5 h per session and 25% three times a week for 3 h per session.
|
Multivariate analysis of total HD duration per week with the data for 16 279 patients for whom we had complete information showed that patients from Austria, French-speaking Belgium and Norway were at a substantially higher risk of receiving dialysis for <12 h a week, compared with patients from France (reference group) (Table 5). Other risk factors for shorter dialysis duration included: age (older), sex (women), BMI (lower) and time on RRT (shorter). Patients with diabetes were less likely to have a shorter weekly duration. There was a linear trend associating age, BMI and RRT duration with a risk of low total HD duration: OR 1.33 (95% CI: 1.29–1.38) for each additional 10 years of age, OR 0.92 (95% CI: 0.915–0.93) for each additional 1 kg/m2 of BMI and OR 0.94 (95% CI: 0.93–0.95) for each additional year of RRT treatment. However, because weekly HD duration may be affected by differences in residual renal function, we performed a separate analysis of patients on RRT for >2 years. The resulting ORs were very similar: French-speaking Belgium (OR: 6.2, 95% CI: 4.9–7.8), Austria (OR: 1.4, 95% CI: 1.2–1.7) and Norway (OR: 2.7, 95% CI: 2.1–3.5).
|
|
Discussion |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionReferences |
|---|
This study of the adequacy of HD in seven European countries shows that 83% of patients receive at least 12 h HD per week, as recommended by the EBPG. It also shows, however, that there is great variability between countries and in patient subgroups for the number and length of HD sessions.
HD dose in the light of the different EBPG (EBP Guidelines II.1.3, II.5.1 and II.5.2) [5]
The EBP Guidelines recommend that ‘even if the standards of adequacy are reached, a minimum time of 4 h 3 times per week is desirable’ (evidence level: B). The proportion of patients receiving <12 h varied between countries from 3% to 51%. These international differences in weekly HD duration were partially but not totally explained by patient characteristics; they persisted after adjusting for age, sex, diabetes status, BMI and RRT duration. It is interesting to note that the Guidelines recommend that treatment time and/or frequency should be increased for elderly HD patients (evidence level: B); the opposite was observed in every country.
Six percent of the patients had 3 h or less per session although some studies suggest that session length may be an important independent determinant of patient mortality. The DOPPS study found that in Europe every additional 30 min per session on HD was associated with a 6% reduction in mortality risk (RR = 0.94; P < 0.01) [8]. With the ANZDT Registry data, Marshall et al. showed that a session length of 4.5–4.9 h was associated with the lowest mortality risk (HR 0.8, 95% CI: 0.66–0.97 compared with 4–4.4 h) even after adjustment for HD dose expressed as Kt/V [9]. Patients with <3.5 h had a higher risk of mortality (HR 1.57, 95% CI: 1.14–2.17 compared with 4–4.4 h). Similar results were found by the Japanese registry in 1997; the probability of death decreased progressively as the duration of the HD session increased, at least up to 5 h [10].
Patients receiving daily HD accounted for 32% of those with sessions shorter than 3 h. The median session length for patients on daily HD was 150 min, a result similar to those from a recent systematic review of 14 cohorts where dialysis duration varied from 1.5 to 3 h for patients treated by daily HD [7].
Twice-weekly schedules are generally not recommended (evidence level: B), and indeed only 4% of patients had such a schedule, to which should be added the 0.4% with only one session. Almost half of them were older than 75 years and 59% had had RRT for <1 year. No information on their residual renal function was available for this study. These results are quite similar to the finding by Hanson et al. in 1995 [4] that 4% of patients in the USRDS database had a twice-weekly schedule. In their study, older age (OR 1.02 per year, P < 0.001) and time since ESRD onset (OR 0.92 per year, P = 0.002) were predictors of such a schedule. Note that in the 1995 USRDS study, patients with this schedule had a mean weekly HD duration of 6.2 h.
HD dose and patient characteristics
In our study, risk factors for shorter dialysis duration included: age (older), sex (female), BMI (low), time on RRT (shorter) and no diabetes. Those findings are consistent with other studies. In the ANZDT study, patients with <4.5 h x 3 per week were likely to be older (median age: 65.4 year versus 58.1), female (45.9% versus 28.3%), smaller (median BMI 24.9 versus 27.8 kg/m2) and without diabetes (29.1% versus 41.9%) [9]. In the DOPPS study, the independent predictors of a treatment time <4 h x 3 were age (OR 1.2 per additional 10 years, P < 0.001), female sex (OR 1.59, P < 0.001), less time on dialysis (OR 0.93 per year increase, P < 0.001) and not having diabetes (OR 0.88, P < 0.05) [8].
Nevertheless some studies based on Kt/V values suggest that HD dose should not be reduced in these patients. The HEMO study suggests that outcome in women is better when they are treated with higher doses of dialysis (19% reduction in mortality) [11]. In that study, however, the mean time for a session was 3 h in the standard dose group and 3.5 h in the high-dose group for three times weekly schedules. The proportion of patients in the high-dose group with at least 12 h a week is unknown. Two studies suggest that increasing HD doses for small patients may improve survival [12,13].
RRT duration was less than a year in 52% of the patients with <12 h HD a week. This may reflect a strategy of progressive treatment initiation. This observation can only be verified in an incident cohort, however, to avoid survival bias related to prevalent patients.
Geographical variations
We observed low doses (<12 h a week) in three countries, but their underlying practice patterns differed. In Norway, the percentage of patients with low-dose HD is due mainly to the high percentage of patients with only two sessions a week, while in French-speaking Belgium, it is related to the high percentage of patients who have sessions of <4 h (52%). We hypothesized that the time needed to reach the HD centre might play a role in the lower number of sessions a week in some regions of Norway, but this was ruled out as an explanation in view of the geographical distribution of the centres with the highest percentages of patients with two sessions a week (personal communication from the head of the Norwegian Renal Registry). In addition, in Finland, where the geographical distribution is similar, the proportion of patients with fewer than three sessions is only 8%. Another explanation may be the organisation of dialysis care in Norway, where many centres are functioning at or above capacity. Patients in Norway are quickly transferred to transplant programmes: the percentage of prevalent patients with ESRD who had renal transplants in 2005 was 71.1%, the highest in Europe [14]. The counterpart of such a large investment in transplantation may be a lower investment in dialysis programmes, as suggested by the high frequency of catheter use (48.4%). On the other hand, the percentage of catheter use is also high in French-speaking Belgium (42.1%). The percentage of patients older than 75 years (incidence of ESRD over 75 years in 2005: 857.5 pmp, prevalence: 2836 pmp) is also high there—the highest in Europe. Austria and Norway have in common their large proportion of patients treated with convective HD. This is not the case for French-speaking Belgium. Nevertheless, introducing the HD technique in the model did not change the ORs for the countries: French-speaking Belgium (OR: 6.3, 95% CI: 5.3–7.5), Austria (OR: 1.3, 95% CI: 1.2–1.5) and Norway (OR: 2.3, 95% CI: 2.0–2.8). The relation between the HD dose and country in our study should nonetheless be interpreted with caution because of the lack of information about residual renal function. We cannot rule out the possibility that dialysis is started at a higher GFR in some centres and that these patients were initially prescribed HD twice weekly. Nevertheless, restricting the model to patients with an RRT duration of at least 2 years did not change the ORs.
In the United States, Hanson and colleagues found large and statistically significant regional variations in the prescription of twice-weekly HD, ranging from 1.5% to 5.1% for prevalent patients and 3.1% to 10.2% for incident patients [4]. These variations are substantially smaller than those in our study, from 1.7% in Macedonia to 20.8% in Norway. The DOPPS study showed differences in the mean treatment time between Europe (232–235 min in DOPPS I and II), Japan (244–240 min) and the United States (211–221 min) [8]. In our study, when we considered only the patients treated three times a week the mean weekly duration was 245 min and varied from 221 min in French-speaking Belgium to 267 min in Slovenia. Our study is the first to show that these variations are also present within Europe (P < 0.0001).
Kt/V
This study points out the difficulty in obtaining the standard HD dose indicator, which is the amount of urea removal, expressed as Kt/V, from the current registries. This indicator was not available in three registries and in the remaining four, it was missing for 62% of patients. Thus, although it is considered standard in recommendations and publications, Kt/V does not seem to be collected routinely by ESRD registries. The reason for this is not clear. Does it reflect low use of Kt/V in daily practice? Is it because of the controversy about the optimal measurement method? Or is it because nephrologists do not consider it a useful (or the main) indicator for assessing their prescriptions?
Strengths and limitations of this study
The results presented here describe current practices in the participating registries (7 of 38 in Europe, i.e. 18%). In view of the great variability seen in this study, we cannot consider them to be representative of the non-participating registries. In view of the geographical distribution of the various countries included (Scandinavian, Mediterranean, West European and Central European countries), however, the study certainly indicates that the variability in Europe is substantial.
In this step of the QUEST initiative, registries and nephrologists were not asked to change or harmonize their practices. This study was based on existing data to evaluate the ‘natural’ implementation of the guidelines and the feasiblity of international comparisons with existing registry data. This analysis shows real-life practices, in contrast to clinical research studies such as DOPPS or standardized databases such as Fresenius Medical Care, where a protocol imposes the data items and the methods. The latter present clearer and more robust results about the impact of HD dose on outcomes, for example [8,15], but they do not describe current day-to-day practices or the implementation of guidelines in centres not participating in these studies. These approaches have very different and, we think, complementary aims.
This study shows that a full description of those real-life practices and the use of registry data for international or even national comparisons depend highly on methodology. More specifically, and perhaps controversially, to reach those goals, data collection must be further standardized and quality control procedures must be implemented within registries (to ensure exhaustive recording of patients and events and completeness of the data). Standardization of indicators is now being addressed within the NephroQUEST project [16]. Because the achievement of clinical measure targets is associated with the improvement of RRT patient outcomes [17], the development of registries towards sources of important high quality clinical data is an important step. They will allow nephrologists and stakeholders to improve the quality of RRT care by comparing such information to standards of care, such as existing guidelines, and to the performance of other centres (benchmarking).
We cannot rule out the possibility of positive selection bias. The participating registries may be the ones with results indicative of better compliance with guidelines. The inclusion of 100% of the patients in a geographic area limits the possibility of positive bias at the patient level. In Macedonia, only 14 of 18 centres participated in the patient study, covering 76% of all patients; data were not available for other four centres. In France, the inclusion of only 13 regions was due to the progressive expansion of the REIN registry and not to the willingness or unwillingness of specific centres to participate in this study. On the other hand, because regions are not required to participate in the REIN registry, self-selection by the most ‘receptive’ regions for epidemiological and evaluation studies is possible. In the participating French regions, 100% of the patients were included. Nonetheless, the problem of selection, if any, is more acute because the French patients accounted for 65% of the total cohort. However, the associations between risk factors and low HD dose were similar, with or without the French data (data not shown). The great variability seen between the French regions may also have reinforced the variability seen in this study.
Given the possible presence of survival bias because this study was based on prevalent patients, the association between HD duration and patient characteristics should be interpreted cautiously.
To evaluate HD dose, we chose two indicators that are easy to define and a standard indicator recommended in the guidelines. As we pointed out above, Kt/V data could not be analysed. This first study had no information available on middle-molecule removal or other indicators of dialysis adequacy (blood pressure control, osteodystrophia, anaemia, etc.).
In conclusion, this study shows a great variability in the HD dose currently received and some discrepancies between current practices and the EBPG. This variability also raises the question of an adaptation of recommendations for different patient subgroups. This first study under the umbrella of the QUEST initiative shows the usefulness of registries for comparing current practices when they have the data available. It also shows their value in monitoring guideline implementation, evaluating clinical performance indicators in various patient subgroups and observing the spread of new strategies. Finally, this study shows the need to standardize and define data to meet these goals.
|
Acknowledgments |
|---|
We thank all the participating registries, all the nephrologists and other professionals participating in the registries, the ERA-EDTA registry coordination and the QUEST adequacy group members.
Conflict of interest statement. None declared.
|
References |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionReferences |
|---|
- Jager K, Zoccali C. Quality European Studies (QUEST)—a step forward in the quality of RRT care. Nephrol Dial Transplant (2005) 20:2005–2006.
[Free Full Text] - Locatelli F, Manzoni C. Duration of dialysis sessions—Was Hegel right? Nephrol Dial Transplant (1999) 14:560–563.
[Free Full Text] - Locatelli F, Buoncristiani, Canaud B, et al. Dialysis dose and frequency. Nephrol Dial Transpl (2005) 20:285–296.
[Abstract/Free Full Text] - Hanson J, Hulbert-Shearon T, Ojo A, et al. Prescription of twice-weekly hemodialysis in the USA. Am J Nephrol (1999) 19:625–633.[CrossRef][Web of Science][Medline]
- European Best Practice Guidelines Expert Group on HD. Haemodialy- sis adequacy. Nephrol Dial Transpl (2002) 17(suppl_7):S16–S31.
- NKF-KDOQI. Clinical practice guidelines; update 2000. Am J Kidney Dis (2001) 37(S1):S5–S6.[Web of Science][Medline]
- Suri R, Nesrallah G, Mainra R, et al. Daily hemodialysis: as systematic review. Clin J Am Soc Nephrol (2006) 1:33–42.
[Abstract/Free Full Text] - Saran R, Bragg-Gresham J, Levin N, et al. Longer treatment time and slower ultrafiltration in hemodialysis: associations with reduced mortality in the DOPPS. Kidney Int (2006) 69:1222–1228.[CrossRef][Web of Science][Medline]
- Marshall M, Byrne B, Kerr P, et al. Associations of hemodialysis dose and session length with mortality risk in Australian and New Zealand patients. Kidney Int (2006) 69:1229–1236.[CrossRef][Web of Science][Medline]
- Shinzato T, Nakai S, Akba T, et al. Survival in long-term haemodialysis patients: results from the annual survey of the Japanese Society for Dialysis Therapy. Nephrol Dial Transplant (1997) 12:884–888.
[Abstract/Free Full Text] - Depner T, Daugirdas J, Greene T, et al. Dialysis dose and the effect of sex and body size on outcome in the HEMO study. Kidney Int (2004) 65:1386–1394.[CrossRef][Web of Science][Medline]
- Lowrie E, Zhensheng L, Ofsthun N, et al. Body size, dialysis dose and death risk relationships among hemodialysis patients. Kidney Int (2002) 62:1891–1897.[CrossRef][Web of Science][Medline]
- Port F, Ashby V, Dhingra R. Dialysis dose and body mass index are strongly associated with survival in hemodialysis patients. J Am Soc Nephrol (2002) 13:1061–1066.
[Abstract/Free Full Text] - ERA-EDTA. Registry 2005 Annual Report. Academic Medical Centre, Amsterdam, The Netherlands, May 2004. Available at http://www.era-edta-reg.org/index.jsp.
- Lowrie E, Ofsthun N, Lazarus J. Measurement of dialyser clearance, dialysis time, and body size: death risk relationships among patients. Kidney Int (2004) 66:2077–2084.[CrossRef][Web of Science][Medline]
- Jager KJ, Zoccali C. Quality of care in end-stage renal disease: the importance of comparing ‘apples with apples’. Nephrol Dial Transplant (2008) 23:1116.
[Free Full Text] - Rocco M, Frankfield D, Hopson S, et al. Relationship between clinical performance measures and outcomes among patients receiving long-term hemodialysis. Ann Intern Med (2006) 145:512–519.
[Abstract/Free Full Text]
Accepted in revised form: 9. 7.08
CiteULike 
Connotea 
Del.icio.us What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||