Advice

Evidence review

Evidence review

Clinical and technical evidence

Nine publications reporting upon the findings of 7 clinical studies were included in this briefing.

The publications by Finkelstein et al. (2012), Jaber et al. (2011) and Jaber et al. (2010) all report outcomes of the FREEDOM study (Following Rehabilitation, Economics And Everyday‑Dialysis Outcome Measurements) at different time‑points (see protocol outline by Jaber et al. 2009). The study by Munshi et al. (2013) was the only relevant study that had been published in abstract form only.

The 7 clinical studies ranged from single‑centre feasibility studies to prospective multicentre cohort studies. Table 1 summarises all the selected studies with their corresponding publications and their basic characteristics.

Table 1 Summary of identified studies characteristics

Study

Population

Number of people

Haemodialysis schedule

Comparator

Design

Manufacturer funding for study

FREEDOM study interim report (Finkelstein et al. 2012)

Adult

291*

Short daily

Baseline

Prospective/

cohort

Yes

FREEDOM study interim report (Jaber et al. 2011)

Adult

235*

Short daily

Baseline

Prospective/

cohort

Yes

FREEDOM study interim report (Jaber et al. 2010)

Adult

239*

Short daily

Baseline

Prospective/

cohort

Yes

(Weinhandl, Liu et al. 2012)

Adult

11,228

Daily

Matched‑cohort

Retrospective/

matched cohort

Yes

(Goldstein et al. 2008)

Paediatric

4

Daily

Baseline

Prospective/

feasibility

Yes

(Zaritsky et al. 2014)

Adult

78

Short daily

Baseline

Unknown/

cross‑sectional

No

(Munshi et al. 2013)

Not reported

973

Unknown

Peritoneal dialysis

Unknown/

cohort

Not stated

(Kohn et al. 2010)

Adult

5

Short daily

Baseline

Unknown/

feasibility

No

(Kraus et al. 2007)

Adult

25

Daily

Baseline

Prospective/

cross‑over feasibility

Yes

* Interim outcome reports of the FREEDOM study. These are cumulative recruitment figures of the same patient cohort, with the most recent being from Finkelstein et al. (2012).

Pilot and feasibility studies

Three feasibility studies (Kohn et al. 2010; Goldstein et al., 2008; Kraus et al. 2007) and 1 cross‑sectional study (Zaritsky et al. 2014) reported on initial safety and efficacy findings using the NxStage System One in home haemodialysis. Kohn et al. (2010) studied the effectiveness of the NxStage System One for metabolic waste product removal from the blood in 5 people having short home haemodialysis sessions 6 times per week. In terms of short daily low flow rate, the NxStage System One (used 6 days per week for an average of 17.5 hours per day, over 8–16 months) was equally effective as conventional haemodialysis (used 3 times per week for 4 hours per session). This was measured by the removal of beta 2 microglobulin, phosphorus and urea nitrogen waste products from the blood (table 2).

Goldstein et al. (2008) studied the feasibility of haemodialysis with the NxStage System One in a small paediatric population (n=4) over 16 weeks of treatment. They reported no serious adverse events and concluded that frequent haemodialysis with the NxStage System One is feasible for children who would benefit from home haemodialysis (table 3).

Kraus et al. (2007) conducted a 2‑month crossover feasibility study of 25 people using the NxStage System One for in‑centre followed by at‑home haemodialysis. They concluded that the system offers a viable dialysis option for people with stage 5 CKD who are capable of home haemodialysis. There was a statistically significant increase in the number of adverse events recorded per 100 treatments in people having dialysis in‑centre compared with at home dialysis. Adverse events in more than 1% of the treatments were blood under‑heating, muscle cramping, hypotension, headache, dizziness and fatigue (see table 4).

In their cross‑sectional study, Zaritsky et al. (2014) analysed the following outcomes in 24 people having short daily haemodialysis using the NxStage System One compared with 54 people having conventional in‑centre haemodialysis:

  • Standardised Kt/V (a measure of renal dialysis adequacy).

  • The level of fibroblast growth factor 23 in the plasma.

  • Haematocrit and serum albumin levels.

  • Blood calcium, phosphorus and intact parathyroid hormone levels.

  • The use of vitamin D, calcimimetic and phosphate binder.

Most of these biochemical parameters did not differ between people using the 2 haemodialysis methods (table 5).

Large cohort studies

The FREEDOM study is a prospective, multicentre, observational cohort trial sponsored by NxStage Medical. The study includes people covered by Medicare health insurance who convert to daily home haemodialysis with the NxStage System One. The target sample size is 500 people. At the time of writing the study is reported to be completed, although full results have not yet been published.

To date, 3 interim reports of the FREEDOM Study have been published: Jaber et al. (2010) reports on experiences of depressive symptoms (table 6); Jaber et al. (2011) reports on experiences of restless leg syndrome (table 7) and Finkelstein et al. (2012) reports on quality of life measures (table 8).

A matched‑cohort retrospective study by Weinhandl et al. (2012) provides the only published evidence comparing survival rates between standard in‑centre and daily home haemodialysis using the NxStage System One. The authors report that in a cohort of 11,228 people with stage 5 CKD, daily home haemodialysis was associated with modest improvements in survival relative to standard three times‑weekly in‑centre haemodialysis (table 9).

The single‑centre study by Munshi et al. (2013) analysed patient survival and technique survival (the time between starting 1 renal replacement therapy and switching to another) in people using home haemodialysis compared with people using peritoneal dialysis. The authors reported the experience of 236 people having home haemodialysis using the NxStage System One, compared with 737 people having peritoneal dialysis, over 7 years (from 2004 to 2011). The study concluded that both treatment methods provided similar patient survival, but that technique survival was better for home haemodialysis with the NxStage System One compared with peritoneal dialysis (table 10).

Table 2 Summary of the Kohn et al. (2010) single‑centre feasibility study

Study component

Description

Objectives/hypotheses

To study the effectiveness of the NxStage System One for blood waste product removal in patients having short daily home haemodialysis.

Study design

A single‑centre feasibility study. The prospective or retrospective design of this study could not be verified from the publication.

Setting

USA‑based. Study duration was 8 to 16 months.

No information was provided on follow‑up schedule.

Inclusion/exclusion criteria

Inclusion criteria: patients having short home haemodialysis 6 times per week; aged 18 or older; with no residual renal function (defined as a urine output <100 mL/day).

Primary outcomes

B2M and urea clearance; phosphorus and potassium removal.

Statistical methods

Descriptive analysis, Pearson's correlation analysis, linear regression analysis.

Participants

5 patients, 4 of African‑American family origin and 1 of Chinese‑American family origin. Patient ages were not reported.

Results

Descriptive analysis:

Weekly B2M, phosphorus and urea nitrogen compared favourably with published values from short daily and 3 times‑weekly conventional haemodialysis done with machines using much higher dialysate flow rates.

To achieve this result an average of 17.5 hours of dialysis per week was needed.

Conclusions

Short daily low flow rate with NxStage System One was equally effective as 3 times‑weekly conventional haemodialysis in removing B2M, phosphorus and urea nitrogen waste products.

Abbreviations: B2M, beta 2 microglobulin.

Table 3 Summary of the Goldstein et al. (2008) prospective feasibility study

Study component

Description

Objectives/hypotheses

To demonstrate the feasibility of haemodialysis with the NxStage System One in a paediatric population during 16 weeks of treatment.

Study design

A prospective, multicentre feasibility study.

Setting

2 USA‑based centres. Follow‑up was at 2, 4, 8, 12 and 16 weeks after the initiation of NxStage System One. Study duration was 16 weeks.

Inclusion/exclusion criteria

Patients having 3‑times weekly haemodialysis for at least 2 consecutive months without a change in haemodialysis prescription; weighing >35 kg; and with no active inflammatory process, including infection or systemic inflammatory disease.

Primary outcomes

Primary outcome: feasibility of using the NxStage System One.

Secondary outcomes: impact of the NxStage System One on:

  • post‑dialysis target weight (kg)

  • standard Kt/V

  • serum electrolytes

  • blood pressure

  • serum cytokines

  • normalised protein catabolic rate nutritional

  • health‑related quality of life parameters.

Statistical methods

No formal sample size calculation was done.

Investigators only reported raw data for each trial participant, except for changes in cytokine levels, which was statistically analysed using ANOVA.

Participants

4 patients enrolled, 3 of whom had frequent haemodialysis at home and 1 of whom had in‑centre dialysis.

Weight range was 38.0–61.4 kg and body surface area range was 1.27–1.60 m2.

Causes of CKD were focal segmental glomerulosclerosis, immune complex glomerulonephritis, membranoproliferative glomerulonephritis type II and bilateral renal dysplasia.

Results

All 4 patients completed the 16‑week study period.

Changes were observed in:

  • pro‑inflammatory cytokine levels (variable changes observed)

  • reduction in blood pressure compared to pre‑treatment systolic and diastolic values

  • reduction in mean serum phosphorus.

No change was observed in:

  • mean patient weight

  • nPCR

  • levels of serum albumin

  • levels of serum electrolytes

  • levels of haematocrit

  • HRQOL.

No adverse effects were reported.

Conclusions

Frequent haemodialysis with the NxStage System One is feasible for children who would benefit from home‑based maintenance dialysis.

Abbreviations: ANOVA, analysis of variance; CKD, chronic kidney disease; nPCR, Normalised Protein Catabolised Rate; HRQOL, Health‑Related Quality of Life; standard Kt/V, a measurement of renal dialysis adequacy.

Table 4 Summary of the Kraus et al. (2007) prospective crossover study

Study component

Description

Objectives/hypotheses

To demonstrate the safety and efficacy of home‑based daily haemodialysis compared with standard centre‑based haemodialysis with the NxStage System One.

Study design

A prospective, multicentre, open‑label, crossover, feasibility study.

Setting

Patients were recruited between February and November 2004 at 6 USA‑based sites. Follow‑up was at 2, 4 and 8 weeks of the in‑centre and home phases.

Inclusion/exclusion criteria

Inclusion criteria: life expectancy >1 year; haemodialysis at least 3 times weekly for a minimum of 3 months; an identifiable caregiver; >18 years old; vascular access capable of a minimum blood flow rate of 350 mL/min, and capable of receiving a delivered single‑pool Kt/V of 0.45 in ≤3.5 hours.

Exclusion criteria: estimated GFR>46 mL/min/1.73 m2 as estimated by 24‑hour urine in the presence of greater than 400 cm3 urine in 24 hours; liver disease; uncontrolled hypertension; symptomatic intradialytic hypotension; haemoglobin <10 g/dL; active infectious or inflammatory disease; documented noncompliance; and malignancy other than superficial skin carcinomas.

Primary outcomes

The primary efficacy end point was the ability to deliver ≥90% of the clinically prescribed dialysis volume.

The primary safety end point was the composite measure of intra‑ and inter‑dialytic adverse events, defined as any unfavourable or unintended sign, symptom or disease temporally associated with use of the device.

Secondary end points were:

  • delivered single‑pool urea Kt/V per treatment

  • KDQoL Short Form

  • successful completion of the training programme by the patient and their caregiver

  • clinical utility (defined as usability) of the NxStage System One

  • ability to achieve target net ultrafiltration volume per treatment.

Statistical methods

Formal sample size calculation (including the drop‑out of 10 patients) of 35 patients to detect a difference between treatment groups for the primary end point as well as the adverse event rates.

The primary efficacy end point was compared between the 2 treatment environments (in‑centre and home) using a generalised linear model with GEE.

Retrospective descriptive analysis of patients' medical records for baseline comparisons between in‑centre and home treatment parameters.

All significance tests calculated for p<0.05.

No adjustments were made for multiple comparisons.

Participants

Of the 32 patients enrolled, 25 completed the study.

Mean age was 51 years, ranging from 18 to 71 years; 63% were men. Patient ethnicities were reported as: 75% were white, and 19% were of black or African‑American family origin.

The primary aetiology for renal disease was diabetes, hypertension, glomerulonephritis, polycystic disease and other.

Results

Successful delivery of at least 90% of prescribed fluid volume (primary end point) was achieved in 98.5% of treatments in‑centre and 97.3% at home.

Total effluent volume as a percentage of prescribed volume was 94–100% for all study weeks.

The composite rate of intra‑ and inter‑dialytic adverse events per 100 treatments was significantly higher for in‑centre (5.30) compared with home (2.10; p=0.007).

Compared with immediately before the study, there were reductions in blood pressure, antihypertensive medications and interdialytic weight gain.

Conclusions

Daily home haemodialysis with a small, easy‑to‑use haemodialysis device was found to be a viable dialysis option for patients with stage 5 CKD capable haemodialysis that was either self‑administered or administered by a caregiver.

Abbreviations: CKD, chronic kidney disease; GEE, generalised estimated equations; KDQoL, Kidney Disease Quality of Life.

Table 5 Summary of the Zaritsky et al. (2014) non‑randomised, cross‑sectional study

Study component

Description

Objectives/hypotheses

To compare the levels of FGF23 and other biochemical variables in patients having short daily haemodialysis using the NxStage System One and in patients having conventional in‑centre haemodialysis.

Study design

A multicentre, non‑randomized cross‑sectional study. The prospective or retrospective design could not be verified from the publication.

Setting

Patients were recruited between January 2009 and January 2012 at 2 USA‑based centres. Biochemical measurements were taken immediately before dialysis treatment (48 hours after the prior treatment in the conventional group and 24 hours after the prior treatment in the short daily haemodialysis group).

No information was provided on follow‑up schedule.

Inclusion/exclusion criteria

Inclusion criteria: aged 18–80 years; having had conventional maintenance dialysis and short daily haemodialysis for > 3 months.

No information was provided on exclusion criteria.

Primary outcomes

Impact of the NxStage System One on:

  • haematocrit levels

  • std Kt/V

  • albumin, calcium and phosphorus levels

  • iPTH

  • FGF23

  • vitamin D, calcimimetic and phosphate binder use.

Statistical methods

No formal sample size calculation was done.

The authors have a descriptive analysis of the outcomes, and did Spearman correlation and linear regression.

No adjustment for multiple comparisons was made.

Participants

54 patients having conventional haemodialysis and 24 having short daily haemodialysis with the NxStage System One, aged 60.8±2.6 and 45.3±4.7 years (p<0.05) respectively.

Median dialysis treatment length was 2.4 years (interquartile range 1.2, 4.1) for the people having conventional haemodialysis and 2.8 years (1.1, 7.6) for people using the NxStage System One.

Results

No statistical difference was seen in Kt/V, haematocrit, calcium, phosphorus, and PTH concentrations.

Statistical differences were seen in serum albumin (lower in the conventional haemodialysis group) and plasma FGF23 values (lower in the short daily haemodialysis cohort).

No association was seen between length of dialysis and FGF23 levels in either group.

Phosphorus levels correlated with FGF23 concentrations in both groups (r=0.42, p<0.01 and r=0.52, p<0.01).

Conclusions

FGF23 levels were significantly lower in patients undergoing short daily haemodialysis with the NxStage System One. Other biochemical parameters were not affected by the modality of treatment.

Abbreviations: FGF23, fibroblast growth factor 23; iPTH; intact‑parathyroid hormone.

Table 6 Summary of the Jaber et al. (2010) interim report of the FREEDOM prospective cohort study

Study component

Description

Objectives/hypotheses

To compare changes in depressive symptoms and post‑dialysis recovery time in patients before and after changing their treatment from in‑centre haemodialysis, peritoneal dialysis or kidney transplant, with home haemodialysis using the NxStage System One.

Study design

Prospective, multicentre, observational cohort study.

Setting

Patients were recruited between January 2006 and December 2008 at 28 USA‑based sites. Follow‑up was at 4 and 12 months after initiation of short daily haemodialysis.

Inclusion/exclusion criteria

Inclusion criteria: diagnosis of stage 5 CKD needing dialysis; Medicare as the primary payer; candidate for daily haemodialysis (defined as dialysis 6 or more times per week); ability to understand and willingness to sign an informed consent statement and a Health Insurance Portability and Accountability Act of 1996 compliant authorization statement.

Exclusion criteria: current use of the NxStage System One haemodialysis device; previous enrolment in this study; current enrolment in another investigational drug or device trial which might impact the outcome measures planned in this study; likelihood of not surviving the training period.

Primary outcomes

Interim report from the FREEDOM study presenting the secondary outcomes of depressive symptoms (as measured by the BDI) and post‑dialysis recovery time as measured by a special study question: "How long does it take you to recover from a dialysis session and resume your normal, usual activities?"

Statistical methods

  • Formal sample size calculation (30% drop‑out rate) of 25 and 55 patients (for BDI score and time to recovery respectively) performed for the 12‑month time point to achieve 80% power with type I error of 0.05 (based on a mean improvement of 8±10 units and 100±150 minutes respectively).

  • Per protocol (patients with 12‑months' follow‑up completion only.

  • ITT analysis (all patients).

  • Sensitivity analysis.

  • Multiple comparisons adjustment*.

Participants

239 patients were enrolled (ITT cohort) and 128 completed the study (per‑protocol cohort). Mean age was 52 years, 64% were men, 55% had an arteriovenous fistula (the others had arteriovenous grafts or central venous catheters) and 90% transitioned from in‑centre haemodialysis (the others transitioned from peritoneal dialysis, kidney transplant or were new to RRT).

Results

Per protocol cohort (n=128): significant decrease in mean BDI score (11.2 [95% CI 9.6 to 12.9] vs 7.8 [95% CI 6.5 to 9.1]; p<0.001).

Significant decrease in mean post‑dialysis recovery time over 12 months (476 [95% CI 359 to 594] vs 63 minutes [95% CI 32 to 95] p<0.001).

ITT analysis (n=239) showed similar results. The results remained significant after sensitivity analysis.

Conclusions

Daily haemodialysis with the NxStage System One was associated with long‑term improvement in depressive symptoms and post‑dialysis recovery time.

Abbreviations: BDI, Beck Depression Inventory; ITT, intention to treat; RRT, renal replacement therapy.

* According to the original study protocol, multiple comparisons adjustment would be performed. This could not be subsequently verified in the interim reports' statistical methods and results description.

Table 7 Summary of the Jaber et al. (2011) interim report of the FREEDOM prospective cohort study

Study component

Description

Objectives/hypotheses

To compare changes in RLS and sleep disturbances in patients before and after conversion to the NxStage System One.

Study design

A prospective, multicentre, observational, cohort study.

Setting

Patients were recruited between January 2006 and December 2009 at 28 USA‑based sites. Follow‑up was 4 and 12 months after initiation of short daily haemodialysis.

Inclusion/exclusion criteria

Inclusion criteria: diagnosis of stage 5 CKD needing dialysis; Medicare as the primary payer; candidate for daily haemodialysis (defined as dialysis 6 or more times per week); ability to understand and willingness to sign an informed consent statement and a Health Insurance Portability and Accountability Act of 1996 compliant authorization statement.

Exclusion criteria: current use of the NxStage System One haemodialysis device; previous enrolment in this study; current enrolment in another investigational drug or device trial which might impact the outcome measures planned in this study; likelihood of not surviving the training period.

Primary outcomes

Interim report from the FREEDOM study, presenting the secondary outcomes of: presence and severity of RLS, assessed at enrolment then at 4 and 12 months using the IRLS Study Group Rating Scale (version 2.2); and presence and severity of sleep disturbances.

Statistical methods

  • Formal sample size calculation (30% drop‑out rate) of 43 patients done for the 4‑ and 12‑month time points to achieve 80% power with a type I error of 0.05.

  • The sample size was calculated on the basis of a mean IRLS global score improvement of 6±9 (SD) units.

  • Per protocol and ITT analysis.

  • Sensitivity analysis.

  • Multiple comparisons adjustment*.

Participants

235 patients with stage 5 kidney disease having daily home haemodialysis with the NxStage System One.

Mean age was 52 years, 65% were men, 66% were white, 43% had diabetes, and 55% had an arteriovenous fistula.

Results

Per‑protocol analysis: mean IRLS score improved significantly.

ITT analysis: decline in the percentage of patients reporting RLS (35% vs 26%), and moderate‑to‑severe RLS (59% vs 43%).

Similar and sustained 12‑month improvement in several scales of the sleep survey.

Conclusions

Short daily home haemodialysis was associated with long‑term improvement in the prevalence and severity of RLS and sleep disturbances.

Abbreviations: ITT, intention to treat; RLS, restless legs syndrome; IRLS, International Restless Leg Scale.

* According to the original study protocol multiple comparisons adjustment would be performed. This could not be subsequently verified in the interim reports' statistical methods and results description.

Table 8 Summary of the Finkelstein et al. (2012) interim report of the FREEDOM prospective cohort study

Study component

Description

Objectives/hypotheses

To compare changes in quality of life measures in patients before and after conversion to the NxStage System One device.

Study design

A prospective, multicentre, observational, cohort study.

Setting

Patients were recruited between January 2006 and December 2009 at 33 USA‑based sites. Follow‑up was done 4 and 12 months after initiation of short daily haemodialysis.

Inclusion/exclusion criteria

Inclusion criteria: diagnosis of stage 5 CKD needing dialysis; Medicare as the primary payer; candidate for daily haemodialysis (defined as dialysis 6 or more times per week); ability to understand and willingness to sign an informed consent statement and a Health Insurance Portability and Accountability Act of 1996 compliant authorization statement.

Exclusion criteria: current use of the NxStage System One haemodialysis device; previous enrolment in this study; current enrolment in another investigational drug or device trial which might impact the outcome measures planned in this study; likelihood of not surviving the training period.

Primary outcomes

Interim report of the FREEDOM study, presenting the secondary outcomes of health‑related quality of life as measured by the SF‑36 health survey.

Statistical methods

  • Formal sample size calculation (30% drop‑out rate) of 153 patients was done at 4 and 12 months to achieve 80% power with type I error of 0.05.

  • As‑treated analysis (patients with 12 months' follow‑up completion only) and total cohort analysis (all patients).

  • Sensitivity analysis.

  • Multiple comparisons adjustment*.

Participants

312 patients were enrolled and 291 completed the survey. Mean age was 53 years, 66% were men, 70% were white, 45% had diabetes, 90% had hypertension, 27% had a history of congestive heart failure and 58% had an arteriovenous fistula.

Results

Total cohort analysis (n=291): both the physical and mental summary scores improved, as did all 8 individual domains of the SF‑36.

As‑treated cohort analysis (n=154): similar improvements with the exception of the role‑emotional domain.

The results remained significant after sensitivity analysis.

Conclusions

At‑home short daily haemodialysis with NxStage System One was associated with long‑term improvements in various physical and mental health‑related quality of life measures.

* According to the original study protocol multiple comparisons adjustment would be performed. This could not be subsequently verified in the interim reports' statistical methods and results description.

Table 9 Summary of the Weinhandl et al. (2012) retrospective matched‑cohort study

Study component

Description

Objectives/hypotheses

To assess mortality in patients having daily home haemodialysis with the NxStage System One and a matched‑cohort having 3 times‑weekly in‑centre haemodialysis.

Study design

A retrospective, multicentre, observational, matched‑cohort study.

Setting

Patients were recruited over 2005–08 across an unspecified number of sites in the USA. Matched in‑centre patients were identified from the US Renal Data System database (637,109 records). Daily haemodialysis patients were identified from a registry of NxStage System One users maintained by NxStage Medical (2816 records). Follow‑up was done until December 2008.

Inclusion/exclusion criteria

Matching method:

  • Medicare as primary payer.

  • To have begun RRT during the 6 preceding months.

  • Subsequently matching according to 17 factors: age, cumulative hospital days, cumulative EPO dose, BMI, transplant waitlist registration, congestive heart failure, CKD duration, race, cancer, primary CKD cause, cerebrovascular disease, peripheral vascular disease, other cardiovascular disease, diabetes, atherosclerotic heart disease, sex and dual Medicare/Medicaid eligibility.

  • For each daily haemodialysis patient, 5 three times‑weekly in‑centre patients with matching characteristics were selected.

Primary outcomes

Mortality rate.

Statistical methods

No formal sample size calculation was done.

  • Descriptive analysis: match quality was assessed with standardised differences, with differences <10% indicating similarity.

  • Survival analysis: Kaplan–Meier method with both ITT and as‑treated analyses.

  • Cox regression, with and without model‑based adjustment for factors already included in the matching algorithm.

  • Discontinuation analysis: discontinuation of follow‑up before December 2008.

  • No multiple comparisons adjustment was done.

  • Sensitivity analysis: performed on the ITT all‑cause mortality HR with adjustment for an unmeasured hypothetical factor U.

Participants

1863 patients on daily haemodialysis with NxStage System One; 9365 on in‑centre thrice weekly dialysis.

Patients having daily haemodialysis were younger (mean age 52.2 versus 62.6 years), more frequently wait‑listed for kidney transplant (35.0% versus 14.2%), had lower incidence of congestive heart failure (26.9% versus 44.6%), and had a lower proportion of people of African‑Caribbean family origin (26.5% versus 38.8%).

Results

  • All survival estimates between the 2 cohorts were statistically significant in favour of daily home haemodialysis.

  • The cumulative incidence of death was 19.2% among the patients having home haemodialysis and 21.7% among the in‑centre patients respectively.

  • ITT analysis: home haemodialysis was associated with a 13% lower risk for all‑cause mortality than in‑centre haemodialysis (HR=0.87, 95% CI 0.78 to 0.97).

  • Cause‑specific mortality HRs were 0.92 (95% CI 0.78 to 1.09) for cardiovascular disease, 1.13 (95% CI 0.84 to1.53) for infection, 0.63 (95% CI 0.41 to 0.95) for cachexia/dialysis withdrawal, 1.06 (95% CI 0.81 to 1.37) for other specified cause, and 0.59 (95% CI 0.44 to 0.79) for unknown cause.

  • As‑treated analysis: similar findings to ITT.

  • No differences were detected from subgroup analyses.

  • Discontinuation analysis: 626 (39.5%) and 1355 (17.1%) events occurred in daily home haemodialysis and matched in‑centre patients respectively.

  • Reason‑specific discontinuation HRs for daily home haemodialysis versus matched in-centre patients were 10.4 (95% CI 8.9 to 12.3) for change in dialytic modality, 1.06 (95% CI 0.89 to 1.25) for kidney transplant, and 0.92 (95% CI 0.68 to 1.26) for cessation of Medicare primary payer status.

  • In patients having daily home haemodialysis, significant predictors of change in dialytic modality included CKD duration at haemodialysis initiation (HR 0.98 per year), dual Medicare/Medicaid eligibility (HR 1.67), and any hospitalisation during the 3 months preceding initiation (HR 1.51).

  • Of patients having daily home haemodialysis who changed dialytic modality, 96.9% initiated in‑centre haemodialysis and 3.1% initiated peritoneal dialysis.

  • Sensitivity analysis: only a strong unmeasured confounder (for example with HR=0.8 and prevalence difference ≥60%) could alone undo the observed association of daily home haemodialysis with mortality.

Conclusions

Relative to standard 3‑times weekly in‑centre haemodialysis, daily home haemodialysis was associated with modest improvements in survival.

Abbreviations: CI, confidence interval; CKD, chronic kidney disease; HR, hazard ratio; ITT, intention to treat; RRT, renal replacement therapy.

Table 10 Summary of the Munshi et al. (2013) single‑centre cross‑sectional study

Study component

Description

Objectives/hypotheses

To compare patient and technique survival amongst patients having home dialysis with NxStage System One and peritoneal dialysis.

Study design

No information on prospective/retrospective status, single‑centre, cross‑sectional study.

Setting

Patients were recruited between January 2004 and December 2011 across an unspecified number of sites in the USA. No information was provided on follow‑up schedule.

Inclusion/exclusion criteria

No information as this was presented in abstract form and not a full publication.

Primary outcomes

Patient survival rate and technique survival.

Statistical methods

  • No formal sample size calculation was done.

  • Patient and technique survival rates were calculated using the actuarial method and standard error and statistical significance were calculated using Greenwood's formula. 8‑year aggregate data were calculated as cumulative proportion surviving.

  • Calculations for technique survival were performed with and without using transplant as technique failure.

Participants

236 patients having home haemodialysis with the NxStage System One and 737 patients having peritoneal dialysis.

Results

Patient survival was similar for home haemodialysis and peritoneal dialysis (8‑years: 40.3±5.2% and 38.0±3.5% for home haemodialysis and peritoneal dialysis respectively, p = 0.71).

8‑year technique survival was lower for peritoneal dialysis than home haemodialysis, whether transplant was considered a technique failure (12.7±2.5% and 4.2±0.7% for home haemodialysis and peritoneal dialysis respectively, p<0.001) or not (36.9±5.0% and 13.1±1.7% for home haemodialysis and peritoneal dialysis respectively, p<0.001).

Conclusions

Both home dialysis therapies provided similar patient survival rates. Although shown to be better than in previously published data, technique survival with peritoneal dialysis was lower than that for home haemodialysis dialysis.

During the preparation of this briefing, 2 ongoing trials and 1 unpublished completed trial using the NxStage System One were identified:

  • NCT00667511: Comparing nocturnal haemodialysis and short daily haemodialysis using the NxStage System One.

  • NCT01062984: Comparing continuous venovenous hemofiltration versus continuous venovenous haemodialysis in acute kidney injury.

  • NCT00633308: Phosphorus and calcium removal during long haemodialysis treatment sessions.

Costs and resource consequences

In the UK, 1080 people used home haemodialysis in 2011‑2012 (Fogarty and Cullen, 2013) and this represents the population who could consider the NxStage System One as a treatment option within the NHS. The system is currently available in a number of NHS trusts, although no information on the exact number of hospitals using NxStage System One could be obtained from the manufacturer.

No change to the organisation or delivery of current treatment services would be needed. People wishing to undertake home haemodialysis using the NxStage System One will be required to take active responsibility for its use and for in‑depth training to use the system.

No published evidence on resource consequences of the NxStage System One was identified in the systematic review of evidence. Economic analyses of the NxStage System One are planned to be included in the full results of the FREEDOM study which, at the time of writing, have yet to be published.

Strengths and limitations of the evidence

The evidence considered in this briefing ranged from small feasibility studies to large prospective and retrospective high quality cohort studies. No randomised controlled trials were identified.

The majority of health‑related quality of life evidence was presented in the published interim analysis reports of the FREEDOM study, provided by Finkelstein et al. (2012), Jaber et al. (2011) and Jaber et al. (2010). Strengths of the FREEDOM study were the repeated measures at 4 and 12 months after starting to use the NxStage System One, its multicentre prospective design, the use of validated health questionnaires such as the SF‑36 and Beck Depression Inventory health surveys, the sample size calculations for the primary and secondary outcomes, and the prospective multicentre study design. The findings of the study were strengthened because the intention‑to‑treat, per protocol and sensitivity analyses all provided the same results. However, the observational study design could introduce potential bias and confounders.

The quality of life outcomes reported in the FREEDOM interim reports were secondary outcomes but this was nevertheless important information for people having renal replacement therapy. This is evidenced by the comments reported from people attending the 2013 UK Home Dialysis Summit, who said that their experience of home haemodialysis was not given adequate consideration in the reporting of data on dialysis (Home Dialysis Summit 2013). It was suggested that patient experience measures should be given priority over more traditional measures of health‑related outcomes. This highlighted the unmet need for patient‑reported outcomes measures and patient‑reported experience measures. Patient‑reported outcomes may reflect the social benefits of home dialysis and support the current NHS focus on patient‑centred care and decision‑making.

The findings of the Weinhandl et al. (2012) study were strengthened because the sensitivity analysis verified the observed result on mortality. Nevertheless, although the matching procedure attempted to minimize all factors that could affect the reliability of the result, it could not account for all the possible differences between people who were using the NxStage System One and those who were using in‑centre haemodialysis. As an example of this limitation in matching the patients across the 2 treatment groups, people using the NxStage System One were younger, more likely to be listed for kidney transplant, less likely to have congestive heart failure, and less likely to belong to minority ethnic groups than people who were not offered home‑dialysis. The age difference could be explained by the matching method used, allowing people with up to 10 years age difference to be considered of similar age. Non‑randomised studies and registries are at risk of selection bias, and this was a limitation for all of the included studies. People using the NxStage System One were likely to be younger, fitter and have fewer comorbidities than people having in‑hospital dialysis.

Most studies reported in this briefing used a short daily haemodialysis schedule, and as a result it was difficult to directly compare the performance of NxStage System One to conventional in‑centre treatment schedules.

Apart from the Weinhandl et al. (2012) and Munshi et al. (2013) studies, the studies described in this briefing reported results made on baseline comparisons only. The FREEDOM study aimed to compare the number of hospitalisations for people using NxStage System One with a group of 500 people using in‑centre haemodialysis. These data will be valuable, but the full results of this trial were not available at the time of publication.

The NxStage System One is intended for use in both adults and children, yet very little clinical evidence exists for the safety and efficacy of the system in children. The only feasibility study available (Goldstein et al. 2008) was of limited value as it included only 4 patients. Additionally, Stronach et al. (2013a) reported their UK‑based experience on a small case series of 3 children with dialyser‑induced thrombocytopaenia when using the gamma‑sterilized PUREMA dialyser that was part of the standard NxStage circuit. These adverse reactions were resolved in all 3 people by changing the dialyser and circuit within the NxStage System One.

Although improved outcomes were reported in the 2 feasibility studies which investigated the transition from in‑centre dialysis to home haemodialysis with NxStage System One, the designs of these studies were unclear. Specifically, in the study by Goldstein et al. (2008), a reduction in blood levels of phosphorus and cytokines was reported in users of the NxStage System One, but it was unclear whether the blood samples for these measurements were taken before or after a haemodialysis session. The study by Kraus et al. (2007) reported that at least 90% of the prescribed fluid volume was successfully delivered in 98.5% of treatments using in‑centre haemodialysis compared with 97.3% of treatments using home haemodialysis. However, it was unclear whether the prescribed volume was the same between the 2 settings.

The matched cohort study (Weinhandl et al. 2012) used data from the NxStage registry so there was potentially some overlap between their patient cohort and the rest of the studies. 

Recruitment and logistical challenges, presented not only in home haemodialysis but in other studies investigating alternative dialysis strategies in general, make it difficult to conduct large randomised clinical trials (Abdel‑Kader and Unruh 2012). For example, in the Frequent Hemodialysis Network Nocturnal Trial, the randomisation of treatment assignment was an obstacle to recruitment because most people interested in this study wanted to be dialysed at home (Rocco et al. 2011). Additionally, the large number of variables in alternative dialysis regimens – such as dialysis session duration and frequency, diffusion and convection, blood and dialysate flow rates and dialysis prescription – means that large randomised clinical trials would be needed to provide reliable results. The subject of establishing the best research approach to evaluate alternative dialysis strategies is still a matter of ongoing debate (Van Biesen and Lameire 2013).

No studies were identified where the NxStage System One was compared with home haemodialysis using conventional equipment. Therefore, patient benefits reported in the studies included in this briefing could potentially be provided by any form of home haemodialysis. Frequent dialysis regimens can also be performed using conventional home haemodialysis equipment and so patient benefits related solely to frequency of dialysis would apply to both NxStage System One and conventional equipment.

Lastly, 4 of these 7 clinical studies were funded by the manufacturer and this introduced the potential of bias in the reporting of outcomes.