|
|
||||||||
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
In-Depth Reviews |



* Division of Nephrology and
Department of Epidemiology and Biostatistics, University of Western Ontario, London, Ontario, and
Division of Nephrology, Humber Regional Hospital, Toronto, Ontario, Canada;
Department of Quantitative Health Sciences, Cleveland Clinic Research Foundation, Cleveland, Ohio; and || Department of Medicine, University of Illinois, Chicago, Illinois
Address correspondence to: Dr. Rita Suri, London Health Sciences Center, Kidney Clinical Research Unit, Room ELL-111, Victoria Hospital, 800 Commissioners Road East, London, Ontario, Canada, N6A 4G5. Phone: 519-685-8066; Fax: 519-685-8072; E-mail: rita.suri{at}lhsc.on.ca
| Abstract |
|---|
|
|
|---|
3 mo were included. Twenty-five articles reporting 14 unique populations with 268 patients (five to 72 per study) met inclusion criteria. Of the 14 cohorts, 13 were studied with an observational design, 10 were studied prospectively, and four had parallel control groups. Mean age ranged form 41 to 64 yr, mean time on dialysis was 2 to 11 yr, 0 to 28% of patients had diabetes, >90% had arteriovenous fistulae, and >50% were dialyzed at home. Most data were described at
12 mo of follow-up. Outcomes included quality of life, cardiovascular disease, erythropoiesis, nutritional status, hospitalizations, and vascular access failures. Reporting was too heterogeneous to allow pooling of data. Ten of 11 studies suggested improvements in blood pressure; findings for other outcomes varied. Discontinuation of DHD occurred in 0 to 57% in-center and 0 to 15% home patients. Studies of DHD are limited by small sample size, nonideal control groups, selection and dropout biases, and paucity of data on potential risks. Randomized trials with adequate statistical power are required to establish the efficacy and the safety of DHD. | Introduction |
|---|
|
|
|---|
| Materials and Methods |
|---|
|
|
|---|
Included Studies
Published full-text case series, cohort studies, and randomized, controlled trials were included when two investigators independently agreed that the article described original data for a population of five or more adults (
18 yr of age), who were receiving DHD (defined as 1.5 to 3 h/session, 5 to 7 d/wk), and reported a follow-up time of
3 mo. The location of daily dialysis could be either at home or in-center. For studies that seemed to describe the same cohort of patients, the most recent study with the most complete data was included. Articles that reported cost-effectiveness and met the above criteria were included, provided that they had original outcome data (e.g., utility measures); articles that report only absolute costs without outcomes were excluded. Abstracts were excluded because of difficulties with accurate data abstraction. A third investigator resolved disagreements with respect to study inclusion.
Finding Relevant Studies
An independent review of citations from MEDLINE (OVID 1966 to May 31, 2005) and EMBASE (OVID 1980 to May 31, 2005) bibliographic databases was conducted by two investigators. Full-text articles were retrieved for further assessment when either investigator considered a citation potentially relevant. The search strategy was pilot-tested and modified with known relevant articles in an iterative process. The search strategy included multiple synonyms for the terms dialysis, daily, quality of life, cardiovascular function, nutrition, anemia, calcium, phosphate, hospitalizations, mortality, vascular access, and outcomes (detailed strategies available upon request). Supplemental searches using the names of all authors of included articles were performed. We also hand-searched reference lists of all potentially relevant articles, reviews, and HD journals that were not indexed in MEDLINE or EMBASE (Home Hemodialysis International, Hemodialysis International, and Dialysis and Transplantation) for the last 6 yr.
Methodologic Assessment
To summarize succinctly the design characteristics of the included studies, two investigators (R.S. and G.N.) developed a methodologic assessment tool, on the basis of criteria of internal and external validity suggested for studies of therapy (2) (Table 1). Each article was rated independently using this tool, and disagreements were resolved by consensus.
|
Statistical Analyses
Characteristics of individual included studies were compiled in tabular form. Because of heterogeneity among the included studies, outcomes were not pooled using meta-analysis. For each outcome, the range of mean changes is described. Vascular access failure rates were calculated as number of events per patient year or number of events per access year on the basis of data provided in the individual studies. Modality survival, defined as the percentage of patients who continued or wished to continue DHD at the last follow-up time, was calculated censoring for death, transplantation, and relocation to another center.
| Results |
|---|
|
|
|---|
= 0.86). Reasons for excluding citations were as follows: Abstracts (13 citations), patients not on DHD (72 citations), children (one citation), fewer than five patients (four citations), follow-up <3 mo (three articles), review articles or editorials without original data (46 citations), kinetic modeling theories (31 citations), and economic articles that reported costs without outcomes (three citations). Two authors independently excluded two articles because they did not examine clinical outcomes. An additional 29 articles reported similar outcomes for the same cohorts of patients as the 29 included articles but at earlier follow-up times (3260). (References for excluded studies available on request.)
Of the 29 included articles, 25 were published in 1998 or later, and four were published before 1982. Dialysis practices over the last three decades have changed considerably, potentially reducing the current applicability of findings from these four earlier studies. Thus, these four studies are described only briefly in this review (36). Ultimately, 25 articles that described 14 cohorts of at least 268 unique patients were included for methodologic assessment and complete data abstraction (731).
Methodologic Assessment of Included Articles
Because several included studies described different outcomes for the same cohort of patients, the methodologic assessment was performed for each of the 14 unique cohorts (Table 1). Only one study used a randomized design; this was a randomized crossover trial (23). Of the 13 observational cohort studies, three reported enrolling separate control groups. No study reported that blinded adjudicators assessed subjective outcome measurements, such as manual blood pressure, left ventricular mass index by echocardiogram, dry body weight, utility scores, etc. The loss to follow-up was low in the majority of studies. No study met all methodologic criteria. The mean number of criteria met for all cohorts was 56% (range 11 to 71%); for cohorts with control groups, it was 66% (range 57 to 71%).
Patient and Treatment Characteristics
Table 2 summarizes the demographic and treatment characteristics of 17 cohorts of DHD patients reported in 29 studies during the period 1966 to 2004. The 14 cohorts reported in 1998 or later represent at least 268 unique patients from 21 centers in eight countries. The mean time on DHD ranged from 3 to 58 mo (731).
|
Patient preference as a result of lifestyle or other reasons seemed to be at least one reason for prescribing DHD in 12 of 14 cohorts. Six cohorts enrolled only patients who were "medically stable" or able to undergo home or out-of-hospital HD (7,15,21,24,27,31), whereas five cohorts also included patients with a specific medical indication for DHD, such as uncontrolled hypertension, intradialytic hypotension on conventional HD, malnutrition, etc. (14,16,17,19,24). One cohort was selected solely on the basis of medical indication (23), and indications were not specified for one cohort (22).
The mean age of patients ranged from 45 to 64 yr, and the majority of patients were male (range 33 to 100%). All except three cohorts were composed entirely of prevalent patients who had been on conventional HD for at least 6 mo; the mean time on conventional HD ranged form 2 to 11 yr. The percentage of patients with diabetes was 0 to 28%, and >90% of patients had arteriovenous fistulae or grafts.
Outcomes with DHD
Selected outcomes that were reported in individual studies are summarized in Table 3. Although the reported mean time on DHD ranged from 3 to 58 mo (Table 2), all studies reported continuous outcomes between 3 and 24 mo of follow-up, with the majority at 12 mo.
|
Findings varied considerably between studies with respect to most outcomes (Table 3), but two findings were relatively consistent. Decreases in systolic or mean arterial BP were reported in 10 of 11 studies (10,14,16,17,2224,27,29,31), whereas six of eight studies found no statistically significant change in phosphate or phosphate binder dose with DHD at 3 to 24 mo of follow-up (12,14,16,24,26,27). Health-related quality of life improved in some studies but not in others. Improvements were seen in hematocrit, hemoglobin, or erythropoietin dose in 7 of 11 studies (14,16,17,22,24,29,31). Albumin increased in 5 of 10 studies (11,16,18,22,29).
The single randomized trial reported similar findings as the observational studies (23). Twelve hypertensive conventional HD patients were randomly assigned in a crossover design to DHD and conventional HD, with each period lasting 6 mo. On DHD, the 24-h ambulatory blood pressure decreased by 20 mmHg (P < 0.01), while antihypertensive medication use was reduced. The left ventricular mass index decreased by 28 g/m2 (P = 0.01). Extracellular water as measured by bioimpedance was reduced by 5.1% (P = 0.02). Estimates of variance were not provided. There was no observed change in hemoglobin, erythropoietin dose, or albumin.
No study was able to evaluate mortality, and only one study evaluated hospitalization rates and length of stay (14). Hospital admission rates and mean length of stay for up to 6 yr on DHD were compared with admission rates and length of stay for the entire cohort during their 12 mo before starting DHD. The hospitalization rate decreased from 2.5 ± 2.7 per patient year during the year on conventional HD before starting DHD to 1.5 ± 9.9 per patient year on DHD (mean follow-up 18.9 mo; P = 0.002). Similarly, the mean length of stay decreased from 12.2 ± 16.1 to 8.0 ± 58.5 d per patient year (P < 0.0001). However, 22 of the 42 patients in this cohort had died by 1 yr of follow-up, and only four of 42 remained by 6 yr (14).
Potential Risks of DHD
Potential risks of DHD include increased blood loss through the dialyzer and vascular access, as well as an increased risk for vascular access failure as a result of more frequent cannulation. The issue of blood loss was addressed in six cohorts. One study reported increased intravenous iron requirements with DHD (27). In another, the authors speculated that increased frequency of blood tests led to increased blood losses, but blood loss was not measured directly, and there was no change in iron dose or ferritin in this study (13). Four studies reported no change (14,17,24) or increase (29) in serum transferrin saturation or ferritin, but iron dose requirements were not specified.
Quantitative data on vascular access events were reported for seven cohorts and are summarized in Table 4. Definitions and outcomes varied considerably between studies. To facilitate describing of the results, we defined "access dysfunction" as any intervention to help to salvage the access and "permanent access failure" as any event that resulted in requirement of new access placement. A statistically significant decrease in access dysfunction was found for one cohort while on DHD compared with their time on conventional HD (16); 22 of the 72 patients in this study had survived for at least 12 mo. In another cohort, there were decreased permanent access failures for arteriovenous fistulae, compared with patients who were on conventional HD (21). Conversely, the remaining five cohorts reported no statistically significant differences in vascular access dysfunction or permanent failures with DHD compared with conventional HD at a mean follow-up of 3 to 24 mo (7,14,17,25,29). There was a trend to increased access dysfunction for arteriovenous fistulae in two of these cohorts (7,14).
|
| Discussion |
|---|
|
|
|---|
We comprehensively summarized the current published evidence on DHD. Although each of the included studies reported significant, often substantial, improvement in at least one outcome with DHD, when reviewed together, the overall findings for individual outcomes were variable. The beneficial effect of DHD on hypertension seemed to be relatively consistent across studies, as did the absence of any effect of DHD on phosphate control. However, health-related quality of life, erythropoiesis, and nutritional status improved significantly in some studies but not in others. Measures of left ventricular hypertrophy improved in all four studies that evaluated this outcome, but to what extent these improvements were due to reductions in extracellular water versus actual reductions in cardiac structural mass is unclear.
The included studies varied considerably with respect to patient population; dialysis time, frequency, and location; definitions of similar outcomes; methods of outcome assessment; follow-up times; and methodologic design. Moreover, the delivered dialysis dose was not described in almost half of the studies and may have differed between studies. This heterogeneity of previous studies of DHD likely is one factor contributing to the inconsistency of results observed.
Previous studies of DHD have also had several methodologic limitations, potentially resulting in the irregular findings. Studies in this review had a median sample size of 23. Whether null findings in some studies were due to lack of statistical power or the absence of any effect of the therapy is uncertain. Conversely, the improvements that were observed in some studies may have been exaggerated as a result of the use of nonideal control groups. Outcomes on DHD were compared with a concurrent control group for only four cohorts (713,21,2325), only one study of which used a randomized, crossover design (23). The rest were prepost case series with analyses of changes in parameters from a baseline measurement on conventional HD to follow-up measurements after initiation of DHD. In these studies, because the comparative evaluations were at different times without a concurrent control group, confounding as a result of period effects, increased medical attention and treatment, and placebo effects cannot be ruled out. In the three studies with nonrandomized control subjects, there was evidence of dissimilarities in baseline characteristics between daily and conventional HD patients. In two of these cohorts, the DHD patients were treated at home, whereas the control group received dialysis in-center (713,21). Home HD patients are a select group, generally characterized by exceptional compliance, motivation, and social support and lower mortality risk compared with in-center patients after adjustment for comorbid factors (61). Because of all of these limitations, it is difficult to interpret the findings from DHD studies.
The utility of findings from previous studies is unclear. First, the patients who underwent DHD in these studies are not adequately representative of the general HD population. Most patients were male and had been on dialysis for >2 yr. Approximately one fourth or fewer had diabetes, >90% had arteriovenous fistulae or grafts, and almost half were selected to be "stable" or able to undergo home HD. Thus, whether the general HD population would achieve similar results as demonstrated in these studies is not known. Second, likely as a result of sample size constraints, the outcomes examined in previous studies have been limited to health-related quality of life or intermediate physiologic outcomes. No study has examined mortality, and only one study assessed the effect of DHD on hospitalization. Although this study suggests that hospitalization rates and length of stay may be reduced by 30 to 40% with DHD over 6 yr, interpretation of these data is limited by possible informative censoring and dropout biases.
The risks of DHD are also uncertain. Small sample size may have impaired the detection of infrequent yet clinically significant adverse events in some studies; in others, it is difficult to determine whether risks were evaluated properly. The relative risk for vascular access events was evaluated in seven of 14 cohorts: Two reported decreases with DHD (16,21), whereas two suggested a trend toward increased arteriovenous fistulae events (7,14). All of these studies were subject to the methodologic limitations described above. With respect to blood loss with DHD, the requirement for intravenous iron supplementation was reported in only one study and found to be increased (27). Finally, the potential for patient fatigue and burnout is concerning, particularly with in-center DHD, for which the median discontinuation rate was found to be 41% by 3 to 24 mo of follow-up.
Unfortunately, the data as reported in previous studies were not sufficient to allow sample size calculations for a randomized trial. As mentioned, the majority of studies examined only surrogate or self-reported outcomes. Moreover, although we were able to gather estimates of the magnitude of effects that were observed with DHD on surrogate and self-reported outcomes, no study presented the variability of the change scores. Thus, sample size calculations for a randomized trial would require several assumptions and statistical simulations in addition to using the data presented here.
Limitations of this review should be appreciated. First, we were not able to pool the data, given its substantial heterogeneity. Second, to limit the scope of the review, we did not include studies that evaluated costs of DHD. However, these data have been reviewed previously (62,63). Third, we did not include abstracts because of difficulties with accurate data abstraction. Finally, although our methodologic assessment tool was based on accepted criteria for evaluating studies of therapy (2), the tool itself has not previously been validated. Because of this limitation, we did not compare scores from individual studies. Rather, we developed and used this tool simply as a means to summarize succinctly the characteristics of the included studies.
The ESRD population has significantly impaired quality of life compared with the general population (6466) and experiences tremendous morbidity. Moreover, the 20% annual mortality rate in North America has not changed in more than a decade despite advances in medical care (67). For several physiologic reasons (68), DHD holds promise as a therapy that could improve the health of this population. Unfortunately, our review indicates that previous studies of DHD have been limited by inconsistent findings, small sample size, nonideal control groups, selection and dropout biases, the absence of hard outcomes, and inadequate assessment of potential risks. Thus, the available evidence does not support the current widespread implementation of DHD, particularly in-center DHD. A recent review resulted in similar conclusions regarding nocturnal HD (69). Despite the methodologic limitations of previous studies, however, the substantial improvements that have been demonstrated in physiologic intermediate outcomes, as well as in health-related quality of life, cannot be ignored, especially given that no other therapy for this population has had such widespread dramatic effects. Further research in the way of large, methodologically rigorous studies is warranted. In the United States and Canada, two initiatives are now under way to address this need: The Quotidian Hemodialysis Registry (70) and the Frequent Hemodialysis Network randomized trials of frequent HD (http://grants.nih.gov/grants/guide/rfa-files/RFA-DK-03005.html). The latter is sponsored by the US National Institutes of Health, in conjunction with the Centers for Medicare and Medicaid Services. It is hoped that other governments will provide funding for additional studies of frequent HD to define better the role of these therapies in the treatment of ESRD.
| Appendix 1: Description of Health-Related Quality-of-Life Instruments |
|---|
|
|
|---|
Kidney Disease Quality of LifeShort Form (KDQOL-SF)
This instrument includes the 36 items of the SF-36, as well as 43 additional kidney diseasespecific items in 12 subscales (73,74). The long form of the instrument was originally developed to assess health-related quality of life in the HEMO study (75).
Nottingham Health Profile
Similar to the SF-36, this is a self-administered, generic instrument with 46 items that measures physical, social, and emotional health problems and their impact on functioning (76).
Time Trade-Off (TTO)
The TTO is a widely used method of measuring utility (77). The interviewer asks the respondent to value health states in terms of duration of life in a state of perfect health that would be equivalent to some period in a particular health condition, for example, the patients own health state.
Health Utilities Index (HUI-3)
The HUI-3 is also a widely used instrument that assesses utility (78). The interviewer-administered version consists of 13 to 39 questions distributed among eight attributes: (1) vision, (2) hearing, (3) speech, (4) ambulation, (5) dexterity, (6) emotion, (7) cognition, and (8) pain. Utility is assessed using a health status classification system and a preference-based scoring formula.
| Footnotes |
|---|
| References |
|---|
|
|
|---|
This article has been cited by other articles:
![]() |
C. Couchoud, J. Kooman, P. Finne, T. Leivestad, O. Stojceva-Taneva, J. B. Ponikvar, F. Collart, R. Kramar, A. de Francisco, K. J. Jager, et al. From registry data collection to international comparisons: examples of haemodialysis duration and frequency Nephrol. Dial. Transplant., August 1, 2008; (2008) gfn442v1. [Abstract] [Full Text] [PDF] |
||||
![]() |
P. Komenda, M. Copland, L. Er, O. Djurdjev, and A. Levin Outcomes of a provincial home haemodialysis programme--a two-year experience: establishing benchmarks for programme evaluation Nephrol. Dial. Transplant., August 1, 2008; 23(8): 2647 - 2652. [Abstract] [Full Text] [PDF] |
||||
![]() |
A. X. Garg, T. Greene, and N. W. Levin A well-conducted randomized trial that establishes no benefit of therapy is an important medical advance Nephrol. Dial. Transplant., January 1, 2008; 23(1): 52 - 55. [Full Text] [PDF] |
||||
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |