Dialyzer Reuse and Mortality Risk in Patients with End-Stage Renal Disease: A Systematic ReviewGalvao T.F.a, b · Silva M.T.a, c · Araujo M.E.A.b · Bulbol W.S.b · Cardoso A.L.M.P.d
aFaculty of Medicine, University of Brasilia, Brasilia, bGetulio Vargas University Hospital, Federal University of Amazonas, Manaus, cMinistry of Health, Department of Science and Technology, Brasilia, and dFaculty of Pharmaceutical Sciences, Federal University of Amazonas, Manaus, Brazil
Background and Aim: Robust evidence about dialyzer reuse effects on mortality is not available. Our aim was to summarize the evidence for the effectiveness of dialyzer reuse compared to single use in patients with end-stage renal disease. Methods: We searched MEDLINE, Embase, CINAHL, SciELO, LILACS, USRDS ADR, universities’ theses databases and annals of congress from major nephrology societies. Reviewers performed the study selection and data extraction independently. We used the GRADE approach to assess the quality of the evidence. Mortality was the primary outcome. Results: A total of 1,190 studies were retrieved, and 14 were included in the review (n = 956,807 patients). The disinfectants used on dialyzer reprocessing were hypochlorite, formaldehyde, glutaraldehyde, and peracetic acid. The evidence available from the studies was of very low quality. Most studies found no differences between groups. In studies with statistically significant differences, these differences were not observed in all groups and they varied by the type of disinfectant, time of observation and treatment unit. Conclusions: No significant differences were identified for the superiority or inferiority of dialyzer reuse versus single use when assessing the mortality of patients with end-stage renal disease. Studies of higher quality, including randomized clinical trials, are required to provide conclusive evidence regarding the effectiveness and safety of dialyzer reuse.
Copyright © 2012 S. Karger AG, Basel
Dialyzer reuse in patients with end-stage renal disease has been employed since 1960 . According to reports, the practice of dialyzer reuse in the USA is decreasing [2,3]. The largest US dialysis provider, Fresenius Medical Care discontinued reuse in the 2000s, in a combined own product manufacturing and renal care chain [4,5]. This trend is likely due to the availability of less expensive high-flux, single-use dialyzers, greater compatibility of the synthetic membranes, and alternative sterilization methods that avoid first-use syndrome . However, other large providers – like DaVita and smaller ones – still reuse dialyzers . In lower resource settings, dialyzer reuse is predominant [7,8].
To our knowledge, systematic reviews of this issue are not available. Narrative reviews recognize the lack of conclusive evidence regarding the effects of dialyzer reuse on mortality [5,9,10,11,12,13]. Some have noted the benefits of reducing costs and waste disposal with reuse [5,9,11,12,13], and others have highlighted the concerns for increased risks with reuse, including adverse reactions, changes in membrane permeability or potential technique errors [9,10,12,13]. However, the effectiveness of dialyzer reuse has not yet been established, and uncertainties about the effect of this practice on the morbidity and mortality of patients remain unsolved.
The objective of this study was to evaluate the effect of dialyzer reuse on the mortality of patients with end-stage renal disease compared to patients with single use of dialyzers through a systematic review of the literature.
Materials and Methods
We considered eligible controlled studies that compared patients with end-stage renal disease on hemodialysis using dialyzers reprocessed by any sterilization technique to patients with single-use dialyzers. There were no restrictions on language, length of follow-up, or publication date or status.
We searched MEDLINE, Embase, the Cumulative Index to Nursing and Allied Health Literature (CINAHL), the Latin American and Caribbean Center on Health Sciences Information (LILACS), Scientific Electronic Library Online (SciELO), United States Renal Data System 2011 Annual Data Report (USRDS ADR), universities’ theses databases and the annals of congress of major nephrology societies (the American Society of Nephrology, the Brazilian Society of Nephrology, the Canadian Society of Nephrology, the European Renal Association-European Dialysis and Transplant Association, the International Society of Nephrology, the Latin American Society of Nephrology and the World Congress of Nephrology). The search on gray literature was included to identify and avoid publication bias. We screened the references of the relevant articles to identify potentially eligible studies and contacted experts to identify those studies that could not be located or were under development. The last literature search was conducted in January 2012.
The search strategy used in MEDLINE (via PubMed) was: (‘renal dialysis’ [tiab] or ‘renal dialyses’ [tiab] or ‘hemodialysis’ [tw] or ‘hemodialyses’ [tw] or ‘extracorporeal dialyses’ [tw] or ‘dialysis’ [tw]) and (‘equipment reuse’ [tw] or ‘reuse’ [tw] or ‘product recycling’ [tw] or ‘disinfection’ [mesh] or ‘disinfection’ [tw] or ‘sterilization’ [mesh] or ‘sterilization’ [tw]) and (‘acetic acid’ [tw] or ‘renalin’ [tw] or ‘chlorine’ [tw] or ‘citric acid’ [tw] or ‘citrate’ [tw] or ‘formaldehyde’ [tw] or ‘methanal’ [tw] or ‘formol’ [tw] or ‘formalin’ [tw] or ‘glutaraldehyde’ [tw] or ‘glutardialdehyde’ [tw] or ‘glutarol’ [tw] or ‘hydrogen peroxide’ [tw] or ‘hydroperoxide’ [tw] or ‘peroxyacetic acid’ [tw] or ‘peroxyethanoic acid’ [tw] or proxitane [tw] or ‘peroxides’ [tw] or ‘sodium hypochlorite’ [tw] or ‘disinfectant’ [tw] or ‘disinfectants’ [tw] or ‘biocides’ [tw] or ‘membranes, artificial’ [mesh] or ‘membrane’ [tw] or ‘artificial membranes’ [tw] or ‘artificial membrane’ [tw] or ‘polymers’ [mesh] or ‘polymers’ [tw] or ‘sulfones’ [mesh] or ‘sulfones’ [tw] or ‘polysulfone’ [tw] or ‘cellulose’ [tw]). Modifications of this strategy were applied when searching the other databases.
Four pairs of researchers independently reviewed the retrieved study titles and abstracts. The full texts were obtained if it was not possible to assess the eligibility. Three researchers (M.E.A.A., M.T.S. and T.F.G.) checked all selected articles. We completed a data extraction sheet to obtain the data of interest. One author extracted the data (T.F.G.), and the other two confirmed the extraction (M.E.A.A., M.T.S.). Disagreements were resolved by a consensus between the authors.
We extracted the following variables: year, country, funding sources, study design, population, history of dialysis (incident or prevalent), adjusted variables, type of dialyzer, membrane permeability, type of disinfectant, reprocessing technique, number of reprocesses, type of healthcare unit (free-standing or hospital), type of analysis (intention-to-treat or per protocol), sample size and mortality risk. If any data were not available in the paper, we attempted to contact the corresponding author of the study.
Adjusted variables were labeled as demographics (e.g. age, gender or race), cause of renal failure, vintage (time on dialysis treatment), facility (e.g. dialyzer reuse technique, number of reuses, formaldehyde concentration, dialysis unit size, unit age, high-flux dialyzer, type of dialysate, water treatment, facility-standardized mortality ratio, facility profit status, germicide type), and comorbidities (e.g. hypertension, neoplasia).
For the USRDS ADR 2011, we collected data from ESRD Chapter 10 – Providers. Patients treated on Fresenius Medical Care and Dialysis Clinic Inc. (DCI) were considered as single use, and patients treated on DaVita were considered as reuse (≧75% patients are treated on reuse). This information was confirmed by contacting the providers and authors of studies performed on such providers.
The primary outcomes in this review included relative risks (RRs), hazard ratios (HRs), odds ratios (ORs) of mortality or standardized mortality ratios (SMRs) and the respective 95% confidence intervals (95% CIs). The threshold for significance was a two-tailed p < 0.05.
Due to the differences between the study measures and significant heterogeneity, meta-analyses were not performed.
We used the Grading of Recommendations Assessment, Development and Evaluation (GRADE) approach to assess the quality of the studies . This method rates the quality of evidence as high, moderate, low or very low. For observational studies, we evaluated: study limitations (that is, the failure to develop and apply appropriate eligibility criteria, flawed measurement of both exposure and outcome, the failure to adequately control confounding, the failure to accurately measure all known prognostic factors or an incomplete, inadequately short follow-up), imprecision, inconsistency, indirectness of evidence, publication bias, and factors that could increase the quality of the evidence (e.g. a large magnitude of effect or confounding would reduce the demonstrated effect). The quality assessment was considered when interpreting the findings.
Our literature search identified 1,190 articles (fig. 1). By screening the titles and abstracts, 980 records were excluded. We analyzed the full texts of the remaining publications, and 14 were included in the review (n = 956,807 patients).
|Fig. 1. Flowchart of the search, selection and inclusion of studies.|
All studies included in the review took place in the USA with the exception of one performed in Taiwan . Most studies were conducted in the 1980s and 1990s. In these decades, conventional dialyzers were predominantly employed; however, in most articles, the type of dialyzer used was not clearly stated (table 1).
|Table 1. Main characteristics of included studies|
The disinfectants used in reprocessing comprised hypochlorite, formaldehyde, glutaraldehyde, and peracetic acid. Five studies reported using both manual and automated techniques for dialyzer reprocessing [28,30,32,33,36]. The average number of reuses ranged from 2.54  to 15 . Some data were obtained by contacting the corresponding authors of the studies.
Most studies were designed as retrospective cohorts that used patient records available from the clinical databases of health insurance companies, mainly Medicare (table 2). Only one study had a sample size of less than 1,000 patients . Five studies were prospective, including four cohorts [4,28,33,36] and one single crossover . In 5 studies, the patients included were not on renal replacement therapy before the data collection period (incident population) [4,27,31,35,36]. Most studies used an intention-to-treat model to analyze mortality, although this information was not available or clearly stated in some studies. The adjustment models varied widely across studies.
|Table 2. Analytical features of the studies|
The evidence was classified as very low quality using the GRADE approach (table 3). The evidence from studies was rated down from low (observational design) to very low. Studies failed on limitations and inconsistency across study methods and results. Because a formal meta-analysis was not possible, statistical heterogeneity could not be assessed. Publication bias was not detected, but it is worth mentioning that studies sponsored by manufacturers of disinfectants showed no significant results [30,32,34,35], whereas studies sponsored by manufacturers of dialyzers showed a reduced risk of mortality when comparing single dialyzer use with reuse [4,37] (table 4).
|Table 3. Quality of evidence profile for mortality outcome, adapted from GRADE |
|Table 4. Mortality risks of reuse compared to single use|
The mortality risk results are presented in table 4 and were obtained as the RRs, HRs or ORs of reuse (exposed) versus single use (nonexposed). However, 3 studies [4,36,37] calculated such risks by single use (nonexposed) related to reuse (exposed). The risk results were obtained from the available adjusted data. One study calculated SMRs from different providers using traditional SMR calculation method .
The majority of the studies revealed no statistically significant differences in mortality between dialyzer single use and reuse. The most comprehensive provider level comparison, the USRDS ADR 2011 (245,538 patients treated in 2009, adjusted for the same confounding), found similar results in SMR of patients treated by providers that employ only single use (Fresenius and DCI) and patients treated on DaVita, considered as reuse . However, patients on reuse (treated on DaVita) had a statistically significant slight reduction in SMR (SMR 0.97; 95% CI 0.99–0.96).
Even in studies where a statistically significant difference was identified, it was not observed in all groups and varied by the type of disinfectant, the length of follow-up and the health care facility (free standing or hospital) [4,27,28,29,30]. In one study, the single use of dialyzers significantly increased the risk of death compared to reuse, but different inclusion criteria were applied for patients in the reuse and single use groups, resulting in a statistically significant healthier patient profile in the reuse group .
Economic savings, along with less waste disposal, are the main reasons to reuse dialyzers [5,9,11,12,13]. Most countries of lower income employ reuse in 100% patients in renal replacement therapy [7,8]. In developing countries that do not reprocess dialyzers, reuse is highlighted as an alternative to handle with health care delivery costs [39,40]. Conversely, dialyzer reuse is prone to increase morbidity due to risks of infection and pyrogenic reaction, but no evidence of such associations is available [41,42]. To ensure safety of the reprocessing technique, renal replacement therapy providers should follow existing recommended practices and guidelines [43,44]. Such standards are central to infection control and quality guarantee in renal care with dialyzer reuse.
We decided to study mortality to assess the effectiveness of dialyzer reuse on patients’ health since it is a patient-centered and hard outcome. End-stage renal disease morbidity can be assessed from heterogeneous, wide-variant methods. Renal function and clearance markers, length of hospital stay, and immunological response are the most common ones [13,25,45]. But it is worth to say that there is a rising concern in using surrogate outcomes in medical research, as such outcomes can be misleading .
The effectiveness of dialyzer reuse was not confirmed by the studies included in this review. Some studies identified statistically significant difference in mortality between dialyzer reuse and single use in some groups analyzed, but such differences were not consistent across all groups. The USRDS universal reporting system, the most ample provider-level comparison, found a slightly reduced mortality in patients treated by a provider that reuses dialyzers in most patients.
Morbidity, in contrast, could be influenced by reuse technique risks, such as pyrogenic reactions and infections. Since these risks could be controlled by dialysis facilities staff, training is an important way to mitigate the hazard. Team approach, staff education, and communication were shown to be vital to the success of dialyzer reuse .
A major limitation of this review is the observational design that was used in all included studies. The definition of nonrandom groups undermines the comparability of subjects. Important prognostic factors, such as age and the primary cause of chronic disease, can influence the allowance of reuse or single use of dialyzers or switching for patients on renal replacement therapy . The heterogeneous distribution of risk factors between the groups reduces the validity of the results. To allow confidence for decision-making process, our search was extensive and quite sensitive, including databases of indexed papers and governmental and other nonindexed literature (gray literature).
Most of the observed effects in selected cohort studies may be due to immortal time bias (also termed as survivor selection bias or survival bias) . This bias denotes a length of time in the follow-up period of a cohort during which the outcome under study could not have occurred. The presence of immortal time bias is proven to underestimate the risk measures and overestimate the effect, in this review, mortality [49,50]. Included studies that used databases as source of information are susceptible to immortal time bias, since most databases only consider hemodialysis individuals who survived more than 90 days of treatment. This ‘immortality’ time was not possible to be assessed in such studies.
Some techniques can be used to improve the comparability of the research groups and to avoid immortal time bias, such as restriction of categories, matching, and time-dependent covariate analysis. Statistical adjustments using linear, logistic and Cox regression can also avoid confounding effects. However, it is recognized that statistical adjustment cannot control all confounding . Studies included in this review employed different adjustment models, but most results were small to moderate in size (ORs or HRs between 0.6 and 1.6), which asks for caution when interpreting results .
Comorbidities and other confounding factors arising from end-stage renal disease hamper the investigation of the influence of dialyzer reuse on patient mortality. In addition to preexisting comorbidities, end-stage renal disease patients have been hypothesized to have higher risk for developing cardiovascular disease, diabetes and cancer [38,52].
Most studies were conducted in past decades, and conventional dialyzer use was predominant. Currently, synthetic and more biocompatible membranes are employed more often in hemodialysis therapy. First-use syndrome and other adverse effects attributed to the conventional dialyzers are less frequent . Synthetic dialyzers can also lead to different degrees of biological responses, and robust evidence of the superiority of synthetic membranes versus conventional dialyzers is not available [54,55].
Membrane permeability could represent an additional confounding factor. Based on the biological plausibility of the greater elimination of potentially harmful molecules, membrane permeability is believed to influence patient prognosis . Two randomized clinical trials have shown nonsignificant lower risks of mortality in patients on hemodialysis using high-flux rather than low-flux dialyzers [57,58]. The first trial randomized 1,846 patients treated with standard or high dose of dialysis using a low-flux or high-flux dialyzer, and found no significant differences in the dose or flux assignment (high dose vs. standard dose: RR = 0.96; 95% CI: 0.84–1.10; high-flux vs. low-flux: RR = 0.92; 95% CI: 0.81–1.05) . The more recent clinical trial assigned 738 incident hemodialysis patients to either low-flux or high-flux membranes, and found a nonsignificant mortality reduction with high-flux membrane, compared to low-flux ones (HR = 0.76; 95% CI: 0.56–1.04) . Cohort studies that analyzed the permeability and biocompatibility of the membranes could not identify significant differences between groups either [59,60,61]; however, one French cohort study that investigated only permeability found better survival in patients on high-flux dialysis therapy over a long time period (RR = 0.62; 95% CI: 0.43–0.91) .
There is no evidence to support an effect on mortality of dialyzer reuse compared to single use in patients with end-stage renal disease. The studies included in this review assessed more than 900,000 patients without detecting significant and consistent differences in mortality risks when comparing reuse with single use. The inconclusive data regarding the effectiveness and safety of dialyzer reuse has not prevented this practice, which continues in many countries. Better methodological quality in selected studies and randomized clinical trials are required to assess the effectiveness and safety of dialyzer reuse.
We would like to thank our collaborators Alcineide Lima Magalhães, Carolina Dalene Silva, Débora Freire Galvão, Eric Oliveira Jarude Thomaz, Joice Costa de Oliveira, Joyce Fernandes, Laila Cristina Alves Rojas, Nádia Maria Soares Bezerra and Victor Braule Pinto Marques for helping in the study selection process.
This research was funded by a public research agency (Brazilian National Research Council).
The authors declare they have no conflicts of interest.
Tais Freire Galvao
Universidade de Brasilia, Campus Universitario
Faculdade de Medicina, Asa Norte
Brasilia, DF 70910-900 (Brazil)
Tel. +55 61 3107 1894, E-Mail firstname.lastname@example.org
Received: December 19, 2011
Accepted: January 14, 2012
Published online: February 18, 2012
Number of Print Pages : 10
Number of Figures : 1, Number of Tables : 4, Number of References : 62
American Journal of Nephrology
Vol. 35, No. 3, Year 2012 (Cover Date: March 2012)
Journal Editor: Bakris G. (Chicago, Ill.)
ISSN: 0250-8095 (Print), eISSN: 1421-9670 (Online)
For additional information: http://www.karger.com/AJN