Cost-effectiveness of bioimpedance-guided fluid management in patients undergoing haemodialysis: the BISTRO RCT

Article history

The contractual start date for this research was in June 2016. This article began editorial review in August 2023 and was accepted for publication in March 2024. The authors have been wholly responsible for all data collection, analysis and interpretation, and for writing up their work. The Health Technology Assessment editors and publisher have tried to ensure the accuracy of the authors’ article and would like to thank the reviewers for their constructive comments on the draft document. However, they do not accept liability for damages or losses arising from material published in this article.

Copyright statement

Copyright © 2024 Zanganeh et al. This work was produced by Zanganeh et al. under the terms of a commissioning contract issued by the Secretary of State for Health and Social Care. This is an Open Access publication distributed under the terms of the Creative Commons Attribution CC BY 4.0 licence, which permits unrestricted use, distribution, reproduction and adaptation in any medium and for any purpose provided that it is properly attributed. See: https://creativecommons.org/licenses/by/4.0/. For attribution the title, original author(s), the publication source – NIHR Journals Library and the DOI of the publication must be cited.

2024 Zanganeh et al.

Aim of the bioimpedance spectroscopy to maintain renal output economic evaluation

The overarching aim of the BISTRO economic evaluation was to determine the costs, outcomes and overall cost-effectiveness of bioimpedance-guided fluid management (BGFM), compared with current fluid management (CFM). The primary (base-case) analysis was conducted from the perspective of the NHS and Personal Social Services (PSS). For each of the comparators, costs included use of healthcare resources associated with the intervention and care received in the primary care and hospital settings. Outcomes were expressed in terms of QALYs. Sensitivity analyses were carried out to present results based on different scenarios and sources of data.

Trial design and participants

The economic evaluation was embedded into the BISTRO RCT. The trial’s protocol and clinical effectiveness results are reported in detail elsewhere. 21,22 In brief, BISTRO was an open-label, pragmatic, randomised, multicentre UK wide trial of patients having incident haemodialysis, comparing current best practice in setting the post-dialytic target weight with the same assessment guided by serial bioimpedance measurements. In terms of inclusion criteria, potential participants were adult patients undergoing haemodialysis, aged > 18 years, within 3 months of commencing centre-based maintenance haemodialysis due to advanced kidney disease (CKD stage 5). Patients required evidence of RKF > 500 ml urine volume/day or residual glomerular filtration rate > 3 ml/minute/1.73 m². Participants should have entered the study on outpatient treatment. Exclusion criteria were the inability or unwillingness to give informal consent, or inability to comply with trial procedures, and either high risk of death or expected transplantation within 6 months.

Intervention

The study intervention was the incorporation of bioimpedance-derived information about body composition into the clinical assessment of fluid status of dialysis patients. In essence, the intervention was the availability of this additional information, specifically the normally hydrated weight reported by the device, which could then be used in conjunction with clinical judgement. Clinicians whose usual role was to assess fluid status were trained in the use of the fluid assessment proforma and asked to set the post-dialysis target weight to avoid excessive volume depletion, where possible. For patients in the control arm, the target weight was set using clinical judgement only. To achieve blinding, bioimpedance measurements were taken in both study groups but the results were concealed from the clinical teams and trial participants in the control group. To minimise performance bias and information bias, the bioimpedance measurements were taken independently of the fluid assessments by trained nurses. In an independent selection process, overseen by Kidney Research UK, the Fresenius body composition monitor (BCM) was selected and used for measuring bioimpedance. 22 All participating centres received bespoke training at the site during visits prior to enrolling patients, which included a standardised approach to taking bioimpedance measurements by the research nurses. Full bioimpedance data sets were downloaded on to the computers at participating renal centres and, throughout the trial, regular blinded quality control assessments of submitted readings were undertaken by the study team.

Randomisation and follow-up

Randomisation was one to one for the bioimpedance intervention and control groups, with random permuted blocks stratified by centre. All participants were followed up until the point when the first of the following events occurred: the participant became anuric, died, had a kidney transplantation, or withdrew from the study due to stopping dialysis (e.g. recovery of function), patient choice or investigator exclusion (e.g. medical condition) or reached the end of the trial with a maximum study follow-up period of 24 months. In those patients who reached the end point before 24 months, data on quality of life and health-related costs were collected while they remained in the study.

Bioimpedance spectroscopy to maintain renal output economic evaluation methods

The economic evaluation took the form of a cost–utility analysis using, as a primary outcome, incremental costs (or cost savings) per QALY gained. The perspective of the base-case analysis was that of the UK NHS and PSS as recommended by NICE reference case for appraising health technologies. 23 Additional analyses were undertaken to explore different sources of data for particular cost categories, including broader costs, different health-related quality of life (HRQoL) instruments and value sets. The time horizon of this within-trial analysis matched the length of the BISTRO follow-up, that is, 24 months following randomisation. In line with recommendations,23 costs and QALYs accruing beyond the first year were discounted at 3.5% per annum.

Analyses were carried out in accordance with the aims and methods specified in the BISTRO trial protocol and the health economics analysis plan (HEAP). The latter detailed the objectives to be pursued and the methods to be followed for the BISTRO economic evaluation. Deviations from the protocol were minor and served the purposes of meeting the study’s objectives when obstacles were presented. The most noteworthy deviation was the additional use of Hospital Episode Statistics (HES) data when available data on resource use collected through case report forms were limited. The trial protocol and the HEAP were reviewed and approved by the BISTRO independent trial steering committee.

As specified in the study’s HEAP, a decision model would be considered to extend the time horizon beyond 2 years, if findings suggested significant differential costs and outcomes as a result of the intervention accruing beyond the trial follow-up period. Analysis of the study’s effectiveness carried out subsequently found the interventions to be equivalent (small, non-statistically significant effect of BGFM compared with CFM in terms of risk of anuria and rate of loss of RKF),21 which was not anticipated to change beyond the 2-year period. This, and the paucity of reliable evidence linking BGFM-related effects to long-term outcomes,24 suggested limited value (and caution) in extrapolating beyond the trial follow-up period.

Health outcomes

The main measure of health outcome in the BISTRO economic evaluation was QALYs. The QALY is a widely used and recommended23 metric that combines quantity (length) of life and preference-based HRQoL into a single value. HRQoL was obtained through participants’ responses to the EuroQol EQ-5 dimension-5 level (EQ-5D-5L)25 and Short-Form 12 (SF-12) instruments26 at baseline and 3-monthly thereafter, until month 24. The EQ-5D-5L consists of two parts: the EQ-5D-5L descriptive system and the EQ visual analogue scale (EQ VAS). The descriptive system asks respondents to indicate their health state by ticking boxes next to the statement that represents the level of health (no problems, slight problems, moderate problems, severe problems and extreme problems) across five dimensions (mobility, self-care, usual activities, pain/discomfort and anxiety/depression). The EQ VAS asks respondents to rate their health on a vertical visual analogue scale, where the end points are labelled ‘The best health you can imagine’ and ‘The worst health you can imagine’. 25 The SF-12 status description instrument asks respondents to answer 12 questions about their perceived status in relation to seven dimensions (physical activities, social activities, usual activities, pain, mental health, energy and fatigue, and general health). 26 Each participant’s responses to the EQ-5D-5L and SF-12 status description instruments can be translated into a single, preference-based HRQoL index score (utility value) using appropriate value sets. Although a new UK specific EQ-5D-5L value set exists,27 it has been a subject of controversy28 and, currently, NICE recommends29 the use of the Hernandez Alava et al. algorithm. 30 Therefore, for the base-case analysis, each participant’s responses to the instrument’s health status classification system were translated into a single, preference-based (utility) index score using the Hernandez Alava et al. value set for the EQ-5D-5L. 30 SF-12 was converted to the Short-Form 6 Dimensions (SF-6D) quality of life instrument and utility values31 (presented in sensitivity analysis). QALYs were calculated as the area under the curve connecting utility scores reported at different time points. 32 Deceased patients were assigned a utility of zero from the date of death.

Resource use and costs

Key healthcare resource use and costs for both comparators were obtained from two primary resources: (1) patient-level data collected within the BISTRO trial and (2) routinely collected data for care received within a hospital through HES. As per the BISTRO protocol, the latter source was used to provide more accurate information on episodes of care provided in hospital (i.e. scheduled and unscheduled inpatient admissions, critical care admissions and hospital outpatient visits). Data collected as part of the BISTRO trial were captured through case report forms (CRFs) and a client services receipt inventory for chronic kidney disease questionnaire. 33 CRFs were completed at baseline and 3-monthly thereafter until month 24 (bioimpedance and haemodialysis sessions were also completed at months 1 and 2). A table listing information about the type of data collected from CRFs and the schedule of data collection can be found in Appendix 1, Table 12. Administrative hospitalisation data (HES) for BISTRO participating sites were obtained from NHS Digital in England, Public Health in Scotland, NHS Informatics Service in Wales and by the site research teams in Northern Ireland. HES data were cleaned (e.g. cancellations and duplicates removed) and checked against randomisation date, analysis date, event type, date of 2-year follow-up and trial end date. Table 1 gives the source of data for each key resource use category in the base-case analysis.

Healthcare resources were translated into costs using unit cost values taken from up-to-date national sources, including the Unit Costs of Health and Social Care 2020 [Personal Social Services Research Unit (PSSRU)]34 and the National Schedule of NHS Costs 2019–20. 35 Monetary values throughout the study were expressed in Great British pounds using 2019–20 as the base year. Service use over the 24-month period was multiplied by unit costs to arrive at total cost for each patient in each of the alternative comparators.

Missing data

Missing data are a common occurrence in studies collecting patient-level data. The choice of analytic method to be used for accounting for missing data depends on the extent of missing information and the likely underlying mechanism through which data were missing. Descriptive analyses of missing data were carried out to investigate the patterns of missing data (through graphs) and the likely mechanism of missingness. Utility values and cost values for deceased participants were imputed by zero from the time of their death. Additionally, a cost of zero was attached to bioimpedance sessions (in the intervention group) and haemodialysis (in both groups) after transplantation or recovery of sufficient kidney function to allow the patient to stop dialysis. The remaining missing utility and costs data were imputed under the missing at random assumption, at each time point for utility and for year 1 and 2 for different cost components, using fully conditional multiple imputation by chained equations implemented through the MICE package in STATA® version 17 (StataCorp LP, College Station, TX, USA). 40 The appropriateness of using missing at random was evaluated by investigating the missing data patterns and comparing attributes with and without missing utility data at each follow-up time point; and cost components data at years 1 and 2. The multiple imputation model used covariates and utility scores collected at baseline. The imputation was conducted separately by trial arm. 41

Statistical and cost–utility analyses

In line with recommended practice for the analysis of RCT data, the intention-to-treat (ITT) principle was adopted. 42 The ITT data set comprised all randomised patients, analysed according to randomised groups, and including those deviating from protocol, switching treatment, withdrawn or lost to follow-up. The unadjusted calculated mean total per-patient values were given for the intervention and control arms alongside mean difference and 95% confidence intervals (CIs) obtained through non-parametric bootstrap (bias corrected and accelerated) methods using 1000 replications. 43 The distributions of the calculated costs and QALYs were interrogated using graphs and skewness statistics and generalised linear regression models (GLM) were used. 44 Adjustment for baseline utility and for the same covariates (Table 5) as for the primary outcome was made. Cost–utility is presented as an incremental cost-effectiveness ratio (ICER), calculated as difference in costs between comparators over difference in QALYs between comparators. 45 The use of bioimpedance was selected as the intervention and current management without bioimpedance as control. Measures of uncertainty were calculated for the mean difference estimates through bootstrap methods using 1000 replications. 43 For the base-case analysis, to account for the inherent uncertainty in the results due to sampling variation, the joint distribution of differences in cost and QALYs was derived using 5000 of non-parametric bootstrap simulations/iterations. 46 The simulated cost and QALYs pairs were depicted on a cost-effectiveness plane and were plotted as cost-effectiveness acceptability curves (CEACs). 47 A CEAC shows the probability of the bioimpedance-guided and standard fluid management options being cost-effective across a range of possible values of willingness to pay (WTP) for an additional QALY. Net monetary benefits (NMB) for each compared option, as well as incremental net monetary benefits (BGFM vs. CFM), were estimated for a range of different WTP thresholds, where a positive incremental NMB would indicate that BGFMI is cost-effective compared with CFM. Data management tasks and statistical analyses were carried out in STATA version 17. Findings of this economic evaluation were reported in accordance with the Consolidated Health Economic Evaluation Reporting Standards statement. 48

Base-case and sensitivity analyses

The primary analysis was complemented by a series of additional sensitivity analyses carried out to explore the impact of different sources of data (i.e. BISTRO CRFs, HES), specifications and assumptions in the cost-effectiveness analysis. The base-case and sensitivity analyses are described in Table 3.

Detail information regarding resource use data sources used in sensitivity analyses are given in Table 4.

Participants characteristics

The study recruitment process is summarised in the Consolidated Standards of Reporting Trials diagram, which has been published elsewhere. 21 Overall, 439 patients were initially recruited from 34 centres, for a maximum follow-up period of 24 months. Randomisation led to well-balanced study arms according to prespecified baseline patient characteristics (Table 5).

Data completeness

The availability of key economic data (healthcare resource use, EQ-5D-5L and SF-12) by comparator is provided in Appendix 7, Table 20. At each point in time, EQ-5D-5L data were considered incomplete if the health status classification part of the questionnaire was not completed or it was partially completed (i.e. fewer than its five domains were completed), which precluded the calculation of utility values. Similarly, resource use and costs were considered missing if an item or answer was not available (i.e. it was not recorded in the CRF or it was unavailable through HES).

As anticipated, there was a high level of completeness of HES data (only 14 missing data from HES England). As expected, the amount of missing data increased with time in trial and the level of completeness was markedly low for particular resource use data collected via CRFs (e.g. completeness of bioimpedance sessions, haemodialysis sessions, and primary and community care services dropped to 22%, 22% and 13%, respectively, at year 2: 12–24 months). In a similar fashion, completion of EQ-5D dropped to levels below 60% from month 6 onwards, with less than one-third of the expected completed questionnaires returned at the 18-month follow-up point. Overall, there were entirely complete data for all resource use categories for only 40 participants (9.15% of all participants; 22 in BGFM group, 18 in CFM group) and only for 48 participants (10.98% of all participants; 28 in BGFM group, 20 in CFM group) for all EQ-5D-5L measurements. The low availability of data over the 24 months made imputation of missing data highly necessary. Further information on data completion is given in Appendix 7.

Costs

Imputed costs for each resource use category and assessed comparator for the base-case analysis are given in Table 6. Findings show BGFM to be associated with greater costs for the cost categories of primary and community care (£201.22), bioimpedance (£185.86), critical care (£71.92), outpatient consultations (£41.32) and unscheduled inpatient admissions (£9.93). Conversely, BGFM was associated with cost savings for the cost categories of haemodialysis (−£494.48), inpatient nursing home admissions (−£54.39), outpatient nursing home visits (−£11.92) and scheduled inpatient admissions (−£4.10). Differences in costs for bioimpedance, inpatient stay and outpatient visits in nursing homes, and care received in primary and community settings were statistically significant (p < 0.00).

The total costs and the difference between the assessed options for the base-case analysis are given in Table 7. Overall, over 24 months, the mean per-patient cost with BGFM was £382 lower (non-significant, 95% CI −3319 to 2556) than that with CFM.

Health outcomes

Table 8 summarises the imputed health outcomes (EQ-5D-5L utility scores) and between group mean differences at each time point and across the follow-up period for the base-case analysis. EQ-5D utility scores from follow-up time point 12 suggest a small non-significant gain for the BGFM group.

The total QALYs and the difference between the assessed options for the base-case analysis are given in Table 9. Overall, over 24 months, there was a small, non-significant QALY gain for the BGFM group 0.043 (95% CI −0.019 to 0.105).

Cost–utility results

Results for the base-case analysis are presented in Table 10. Over 24 months BGFM was associated with a slightly lower total per-patient cost, giving an estimated total per-patient cost-saving of £382 (non-statistically significant, 95% CI −£3319 to £2556). In terms of QALYs gained, BGFM appeared to be slightly more effective than CFM, resulting in a gain of 0.043 QALYs (non-statistically significant, 95% CI −0.019 to 0.105). On average, BGFM was found to be less costly and more effective than CFM.

Figure 1 depicts the results of 5000 bootstrap replications plotted on the cost-effectiveness plane. Each point represents a pair of incremental cost and incremental effectiveness estimates for the comparison between BGFM and CFM. Approximately 90% of the simulated pairs are located in the east half of the plane, indicating that BGFM is likely to be more effective than CFM. Simulated estimates are fairly split between the north and south halves of the plane, with 44% being located in the north half and 56% in the south. Overall, 48% of the simulated estimates are located in the south-east quadrant, indicating that BGFM is less costly and more effective than CFM. Of the remaining simulated estimates, 42% are in the north-east quadrant, 8% in the south-west and 2% in the north-west.

FIGURE 1.

Cost-effectiveness plane depicting the distribution of simulated cost and QALY pairs.

The probability that BGFM is cost-effective at different WTP thresholds, representing the (hypothetical) amount decision-makers may be willing to pay for an additional QALY, is shown in Figure 2. At £0, the probability that BGFM is cost-effective is 57%. At £20,000 per QALY, this rises to 76% and, at £30,000 per QALY, it further increases to 83% (see Figure 2).

FIGURE 2.

Cost-effectiveness acceptability curve showing the probability of BGFM being cost-effective at different values of WTP for a QALY.

Incremental NMBs, which measure the difference in NMB between alternative interventions, are plotted in Figure 3. At the WTP threshold of £20,000 per additional QALY the incremental NMB is £1091, and it increases to £1519 at the WTP threshold of £30,000 per additional QALY. A positive incremental NMB indicates that, at a particular WTP threshold, BGFM is cost-effective.

FIGURE 3.

Incremental net monetary benefit (NMB) across different WTP values for a QALY. iNMB, incremental net monetary benefit.

Overall, the findings suggest that the BGFM is likely to be cost-effective, although the incremental cost savings and QALY gains were marginal.

Additional sensitivity analyses

Findings of sensitivity analyses can be found in Table 11, with more detailed information given in Appendix 8 (Tables 21 and 22) and Appendix 9 (Tables 23 and 24). Findings were largely similar to the base-case analysis, showing that, under different assumptions, BGFM is likely to be slightly less costly and modestly more effective than CFM. As an exception, a sensitivity analysis that took into account only 40 patients (9.15% of the sample) with complete data across all resource use categories and EQ-5D-5L measurements, showed an ICER of £16,780 per QALY, suggesting that the BGFM is more costly and more effective than CFM. Complete case analysis (with 90.85% missing data) was carried out for completeness. In this analysis, all cost categories apart from total outpatient visits cost and outpatient nursing home visits presented a small, non-significant increase of cost with BGFM group at 24 months (see Appendix 8, Table 21). However, findings of this analysis should be considered with caution, as they are calculated on the basis of a small number of complete data. Overall, all sensitivity analyses showed that use of bioimpedance is likely to be cost-effective at a cost-effectiveness threshold of £20,000–30,000 per QALY in the population studied.

Patient and public involvement

The BISTRO trial was supported by patient and public involvement (PPI) from inception and design to delivery and dissemination of the results. It was co-led by a patient with lived experience of dialysis and expertise in research on devices who was a funded co-applicant, employed by NIHR Devices for Dignity. This co-applicant was a full member of the trial management group and led the patient advisory group that supported trial design, delivery and dissemination. He designed all the patient-facing communications. Deciding the optimal amount of fluid removal on dialysis is a complex clinical decision and measuring the primary outcome, RKF is a complex procedure. PPI was key in ensuring that the patient perspective was considered in all aspects of the trial, which resulted in excellent adherence to trial procedures.

Equality, diversity and inclusion

BioImpedance Spectroscopy to maintain Renal Output was an inclusive trial and the proportions of patients from minority groups are reported in the report of the primary findings21 where they are compared with contemporary data reported to the UK Renal Registry.

Principal findings

Compared with CFM, the addition of bioimpedance measurements was associated with slightly lower costs and slightly higher QALYs. Apart from an additional analysis that used only the small number of complete data sets available, which found costs to be marginally higher in BGFM, sensitivity analyses also lent support to this option being slightly less costly and more effective than CFM. It is observed that the 95% CIs for incremental costs and incremental QALYs were wide both in base-case and sensitivity analyses. This indicated uncertainty around the estimate of the incremental costs and QALYs. Although the CIs for the ICERs were wide, BGFM remains likely to be cost-effective, as the simulated estimates were largely in the south-east (48%) and north-east (42%) quadrants and the probability of BGFM being cost-effective at the WTP threshold of £20,000 per QALY is 76%.

Bioimpedance spectroscopy to maintain renal output effectiveness findings versus cost-effectiveness findings

The results need to be considered jointly with findings of the BISTRO effectiveness analysis. The latter found that the primary outcome (time to anuria) despite a hazard ratio of 0.74, did not differ significantly between the BGFM and the CFM (control) groups, with a 95% CI that spanned one. The primary outcome event rate was half that predicted prior to trial design and the difference between the normally hydrated weight and the target weight set by clinicians did not differ by group. Findings of the BISTRO economic evaluation, on the other hand, suggest that the intervention is likely to be cost-effective, being slightly less costly and more effective than CFM. In interpreting these results, it is useful to keep a number of considerations in mind. First, the economic evaluation seeks to answer a question about BGFM’s cost and effectiveness, which has a broader evaluative space. Given this, the primary outcome of the economic evaluation differs from that of the clinical study: it is a composite measure of differences in costs per difference in QALYs (with the latter being a broader measure, combining time in a health state and generic HRQoL), and hence there is always a possibility that findings about effectiveness will differ in direction from findings about cost-effectiveness. Second, the BISTRO RCT was not powered to estimate costs, QALYs or cost per QALY ratios with a desired precision. With this in mind, the question of interest is whether, despite appearing to be cost-effective, an intervention should be rejected for not reaching significance levels using common rules of inference. Such inference, it has been argued, is irrelevant for decision-making: if the aim is to maximise health benefits from a given budget, one should select the alternative that is shown to be the most cost-effective. 50 In this particular case, failing to do so would mean that an average cost-saving of approximately £380 per patient (which equates to as much as £9 million in total) and a slight increase of 0.04 QALYs would be forfeited.

Key strengths and limitations

This is the first study evaluating the cost-effectiveness of BGFM using patient-level data collected within a pragmatic, RCT. Strengths of the BISTRO trial, in which this economic evaluation is embedded, include a larger sample size compared with studies on the topic,5–7 a 24-month follow-up period and frequent collection of data useful for the economic analysis (resource use and HRQoL). Additionally, data collected as part of the BISTRO trial were complemented by detailed data on care provided in hospital from HES (e.g. inpatient admission, critical care admissions, outpatient appointments), which offered more granular and complete information. Given this, BISTRO provides detailed new evidence on a range of relevant cost components. A number of sensitivity analyses were carried out to generate results on the basis of different sources of data (i.e. CRFs, HES), cost categories and HRQoL instruments and value sets. Finally, methods used throughout the economic analysis, including the base-case and additional analyses, were in line with ‘good practice’ recommendations and requirements for the allocation of healthcare resources in the UK. 23,45,51

Despite these strengths, this study has certain limitations. First, BISTRO was not able to recruit to the initial target within the funding period, despite a funded extension and our attempt to compensate for this by extending the follow-up period. This would have resulted in a larger sample size for both the effectiveness and cost-effectiveness analyses. Second, the level of missing data was high for some resource use categories collected via CRFs, largely as a result of COVID disruptions and a significant dropout rate during the trial. Third, limited data on use of medications were collected through CRFs (at baseline only, to facilitate statistical adjustment), which precluded the inclusion of costs of medications (chiefly antihypertensives) provided outside the hospital setting (medications costs are incorporated in inpatient admission recorded in HES data). Additionally, costs beyond the NHS and PSS were limited to the opportunity cost of time devoted by unpaid carers. However, given no differential effectiveness of BGFM and CFM, it is not anticipated that medication use and broader costs accruing outside the NHS (e.g. time off work) attributable to BGFM and CFM would differ in a systematic way across the compared options. It is also worth noting that data completeness for the EQ-5D data was markedly reduced at 6 months and, as a result, the value of total QALYs over the 24-month period is influenced, at least to some extent, by the results of multiple imputation. Reassuringly, the direction and pattern of EQ-5D results generated by multiple imputation were in agreement with those of the complete case analysis based on available observations only (see Appendix 9, Table 23). It must be also noted that the results of this analysis are applicable to the patient population of interest in BISTRO (see Trial design and participants) and may not be generalisable to other populations.

Future research

Ongoing analysis as part of the trial is exploring the association between primary outcomes and longer-term survival. Should an important link be established, it would be informative to determine whether and how this might affect longer-term costs and benefits associated with BGFM.

CRediT contribution statement

Mandana Zanganeh (https://orcid.org/0000-0003-3934-6044): Conceptualisation, Data curation (lead), Formal analysis, Investigation, Methodology, Project administration, Validation, Writing – original draft, Writing – reviewing and editing (supporting). John Belcher (https://orcid.org/0000-0001-6844-0932): Data curation (supporting), Investigation, Project administration, Writing – reviewing and editing (supporting). James Fotheringham (https://orcid.org/0000-0002-8980-2223): Investigation, Methodology, Writing – reviewing and editing (supporting). David Coyle (https://orcid.org/0000-0003-0441-2996): Funding acquisition, Investigation, Project administration, Writing – reviewing and editing (supporting). Elizabeth J Lindley (https://orcid.org/0000-0003-4957-4489): Funding acquisition, Investigation, Writing – reviewing and editing (supporting). David F Keane (https://orcid.org/0000-0002-4744-4106): Investigation, Writing – reviewing and editing (supporting). Fergus J Caskey (https://orcid.org/0000-0002-5199-3925): Funding acquisition, Investigation, Writing – reviewing and editing (supporting). Indranil Dasgupta (https://orcid.org/0000-0002-7448-2677): Funding acquisition, Investigation, Writing – reviewing and editing (supporting). Andrew Davenport (https://orcid.org/0000-0002-4467-6833): Funding acquisition, Investigation, Writing – reviewing and editing (supporting). Ken Farrington (https://orcid.org/0000-0001-8862-6056): Funding acquisition, Investigation, Writing – reviewing and editing (supporting). Sandip Mitra (https://orcid.org/0000-0003-0389-636X): Funding acquisition, Investigation, Writing – reviewing and editing (supporting). Paula Ormandy (https://orcid.org/0000-0002-6951-972X): Funding acquisition, Investigation, Writing – reviewing and editing (supporting). Martin Wilkie (https://orcid.org/0000-0003-1059-6453): Funding acquisition, Investigation, Project administration, Writing – reviewing and editing (supporting). Jamie H Macdonald (https://orcid.org/0000-0002-2375-146X): Funding acquisition, Investigation, Writing – reviewing and editing (supporting). Ivonne Solis-Trapala (https://orcid.org/0000-0002-1264-1396): Funding acquisition, Investigation, Visualisation, Writing – reviewing and editing (supporting). Julius Sim (https://orcid.org/0000-0002-1816-1676): Funding acquisition, Investigation, Project administration, Visualisation, Writing – reviewing and editing (supporting). Simon J Davies (https://orcid.org/0000-0001-5127-4755): Conceptualisation, Funding acquisition, Investigation, Project administration, Supervision, Visualisation, Writing – reviewing and editing (supporting). Lazaros Andronis (https://orcid.org/0000-0001-7998-7431): Conceptualisation, Formal analysis, Funding acquisition, Investigation, Methodology, Project administration, Supervision, Validation, Writing – reviewing and editing (lead).

Acknowledgements

Data-sharing statement

The trial data are available to investigators under the conditions of a data-sharing agreement. This will include group- and individual-level fully anonymised data. Applications should be made to the corresponding author.

Ethics statement

The study had UK Integrated Research Ethics approval: 206213 (date of approval: 18 August 2016) and all participants gave their written consent.

Information governance statement

Keele University is committed to handling all personal information in line with the UK Data Protection Act (2018) and the General Data Protection Regulation (EU GDPR) 2016/679. Under the Data Protection legislation, Keele University is the Data Controller, and you can find out more about how we handle personal data, including how to exercise your individual rights and the contact details for our Data Protection Officer here: https://www.keele.ac.uk/legalgovernancecompliance/legalandinformationcompliance/informationgovernance/ or by e-mail dpo@keele.ac.uk.

Disclosure of interests

Full disclosure of interests: Completed ICMJE forms for all authors, including all related interests, are available in the toolkit on the NIHR Journals Library report publication page at https://doi.org10.3310/JYPR4287.

Primary conflicts of interest: None of the authors report conflicts of interest related to bioimpedance research, nor do they receive research funding from or have financial interests in Fresenius MC, manufacturers of the BCM device (which was selected for this trial via an independent panel overseen by Kidney Research UK). James Fotheringham is a member and vice-chair of the NICE Technology Assessment Board.

Patient data statement

This work uses data provided by patients and collected by the NHS as part of their care and support. Using patient data is vital to improve health and care for everyone. There is huge potential to make better use of information from people’s patient records, to understand more about disease, develop new treatments, monitor safety and plan NHS services. Patient data should be kept safe and secure, to protect everyone’s privacy, and it’s important that there are safeguards to make sure that they are stored and used responsibly. Everyone should be able to find out about how patient data are used. #datasaveslives You can find out more about the background to this citation here: https://understandingpatientdata.org.uk/data-citation.

Department of Health and Social Care disclaimer

This publication presents independent research commissioned by the National Institute for Health and Care Research (NIHR). The views and opinions expressed by authors in this publication are those of the authors and do not necessarily reflect those of the NHS, the NIHR, MRC, NIHR Coordinating Centre, the Health Technology Assessment programme or the Department of Health and Social Care.

This article was published based on current knowledge at the time and date of publication. NIHR is committed to being inclusive and will continually monitor best practice and guidance in relation to terminology and language to ensure that we remain relevant to our stakeholders.

Trial registration

This trial is registered as ISCCTN Number: 11342007.

Funding

This article presents independent research funded by the National Institute for Health and Care Research (NIHR) Health Technology Assessment programme as award number 14/216/01 (NIHR136142).

About this article

The contractual start date for this research was in June 2016. This article began editorial review in August 2023 and was accepted for publication in March 2024. The authors have been wholly responsible for all data collection, analysis and interpretation, and for writing up their work. The Health Technology Assessment editors and publisher have tried to ensure the accuracy of the authors’ article and would like to thank the reviewers for their constructive comments on the draft document. However, they do not accept liability for damages or losses arising from material published in this article.

This article was published based on current knowledge at the time and date of publication. NIHR is committed to being inclusive and will continually monitor best practice and guidance in relation to terminology and language to ensure that we remain relevant to our stakeholders.

Copyright

Copyright © 2024 Zanganeh et al. This work was produced by Zanganeh et al. under the terms of a commissioning contract issued by the Secretary of State for Health and Social Care. This is an Open Access publication distributed under the terms of the Creative Commons Attribution CC BY 4.0 licence, which permits unrestricted use, distribution, reproduction and adaptation in any medium and for any purpose provided that it is properly attributed. See: https://creativecommons.org/licenses/by/4.0. For attribution the title, original author(s), the publication source – NIHR Journals Library and the DOI of the publication must be cited.