Sodium bicarbonate to improve physical function in patients over 60 years with advanced chronic kidney disease: the BiCARB RCT

Article history

The research reported in this issue of the journal was funded by the HTA programme as project number 10/71/01. The contractual start date was in October 2012. The draft report began editorial review in February 2019 and was accepted for publication in September 2019. The authors have been wholly responsible for all data collection, analysis and interpretation, and for writing up their work. The HTA editors and publisher have tried to ensure the accuracy of the authors’ report 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 report.

Declared competing interests of authors

Peter T Donnan is a member of the New Drugs Committee of the Scottish Medicines Consortium and has received recent grants from Gilead Sciences, Inc. (Foster City, CA, USA) and Shire Pharmaceuticals Ltd (Basingstoke, UK). Alison Avenell works in the Health Services Research Unit, which is core funded by the Chief Scientific Office of the Scottish Government Health and Social Care Directorate. Paul McNamee works in the Health Economics Research Unit, which is core funded by the Chief Scientist Office of the Scottish Government Health and Social Care Directorate. Huey Chong works in the Health Economics Research Unit, which is core funded by the Chief Scientist Office of the Scottish Government Health and Social Care Directorate.

Copyright statement

© Queen’s Printer and Controller of HMSO 2020. This work was produced by Witham et al. under the terms of a commissioning contract issued by the Secretary of State for Health and Social Care. This issue may be freely reproduced for the purposes of private research and study and extracts (or indeed, the full report) may be included in professional journals provided that suitable acknowledgement is made and the reproduction is not associated with any form of advertising. Applications for commercial reproduction should be addressed to: NIHR Journals Library, National Institute for Health Research, Evaluation, Trials and Studies Coordinating Centre, Alpha House, University of Southampton Science Park, Southampton SO16 7NS, UK.

2020 Queen’s Printer and Controller of HMSO

Background and rationale for the trial

Chronic kidney disease is common; 6.1% of the English population have an estimated GFR (eGFR) of < 60 ml/minute/1.73 m² (i.e. GFR category 3–5 in the international CKD staging system),1 and rates in people aged > 70 years are five times higher than this. 2 Population-based estimates suggest a prevalence of severe CKD (eGFR 15–29 ml/minute/1.73 m², equivalent to GFR category 4) of approximately 2% in those aged ≥ 70 years,3 and approximately 20% of such patients will have a degree of metabolic acidosis (often operationalised as a serum bicarbonate level of < 22 mmol/l), with rates increasing as renal function declines. 4 As discussed in the following sections, metabolic acidosis has been associated with multiple adverse outcomes, including impaired bone health and vascular health, accelerated renal decline and impaired physical function. The extent to which acidosis is causal for these phenomena, as opposed to associations being the result of residual confounding by other unmeasured factors present in patients with CKD, remains unclear.

Review of existing trial evidence

As part of the preparation for this report, we undertook a systematic review to synthesise current trial evidence in this area. The search strategy for this systematic review is provided in Appendix 1; details of included studies are provided in Table 23 (see Appendix 2) and forest plots for the main meta-analysed outcomes are provided in Figures 19–25 (see Appendix 2). The systematic review was registered on the PROSPERO database (CRD42018112908). Evidence available prior to completion of the BiCARB trial is discussed here and has been published along with the full methods;21 meta-analyses were then repeated with the addition of results from the BiCARB trial and these results are discussed in Chapter 6.

Imperative for the current trial

Few trials to date have included many older people with CKD, despite older people being the group most likely to have CKD. CKD in older patients is almost always accompanied by multimorbidity and thus a narrow focus on measures of kidney disease alone is unlikely to reflect what is important to the patient. Any trial seeking to provide a comprehensive view of the net health gain or loss from treatment in older patients with CKD must therefore measure a range of outcomes of relevance to older people and must seek evidence of both benefit and harm, an approach that is more likely to provide appropriate evidence on which to base guidelines.

Chronic kidney disease is common and affects many older people. The accompanying acidosis may worsen the muscle weakness that affects many older people, as discussed above, with muscle weakness being a key risk factor for falls, disability, institutionalisation and premature death. 33 Only a minority of older people with CKD progress to a point where they require dialysis for renal failure. However, the effect on quality of life, and the cost burden from dialysis, are considerable; the cost burden is between £20,000 and £25,000 per patient per year, depending on modality and dialysis setting. 34 Finally, cardiovascular disease is the leading cause of hospitalisation and death in older people and is responsible itself for half to one-third of the decline in physical function seen with age.

An intervention that successfully reverses acidosis in this older population may therefore be able to simultaneously improve multiple important associated comorbidities in older people, with consequent improvements in function and quality of life, as well as potential reductions in hospitalisation and later institutionalisation.

Trial objectives

The primary objective of the BiCARB trial was to determine whether or not oral bicarbonate therapy improves physical function compared with placebo in older people with CKD and mild acidosis. The secondary objectives were to (1) determine whether or not oral bicarbonate therapy improves health-related quality of life compared with placebo; (2) compare the impact of oral bicarbonate therapy with that of placebo on biochemical markers of CKD; (3) assess whether or not use of oral bicarbonate therapy is associated with an excess of adverse events compared with placebo; (4) estimate the cost-effectiveness of using oral bicarbonate therapy compared with placebo; and (5) assess the effect of oral bicarbonate therapy compared with placebo on bone turnover and vascular health, as assessed by biochemical markers.

The protocol has been previously published by the authors in Witham et al. 35 This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly credited.

Participants

Trial intervention and comparator

The trial intervention was oral sodium bicarbonate tablets. Each tablet contained 500 mg of sodium bicarbonate (equivalent to 6 mmol of sodium and 6 mmol of bicarbonate). The initial dose dispensed was one tablet three times per day. If the serum bicarbonate level was still < 22 mmol/l at 3 months, the dose was increased to two tablets three times per day for the remaining duration of participation. The comparator was matching placebo tablets. No specific measures were used to enhance adherence beyond reminding participants to take their medication at each study contact.

Outcome measures

Sample size calculation

We based the original sample size calculation on the ability to detect a 1-point difference in the SPPB score (i.e. the primary outcome). This difference has been proposed as the minimum clinically important difference (MCID) by previous investigators. 37 Previous work with older people showed a standard deviation of 2.6 for the SPPB. To detect a 1-point difference between groups at 12 months given this standard deviation would require 143 participants per group, given a two-sided alpha of 0.05 and power of 90%.

To ensure that the trial had sufficient power for the key secondary outcome of health-related quality of life, we also estimated the sample size required to detect the MCID for the EuroQol-5 Dimensions (EQ-5D) measure. For the EQ-5D, the MCID is 0.074. 46 To detect this with a two-sided alpha of 0.05 and power of 90%, assuming a standard deviation of change of 0.2, as found in our previous studies,47,48 would require 154 participants per group.

Assuming a 10% loss to follow-up every 6 months (based on previous medication trials in frail older people47,49), we estimated that we would require 380 participants (190 per group) to ensure adequate power for the primary outcome and the EQ-5D outcome at 12 months.

Regulatory approvals

The BiCARB trial was a Clinical Trial of an Investigational Medicinal Product (CTIMP). As such, the trial was subject to approval and oversight from the Medicines and Healthcare products Regulatory Authority (EudraCT number 2011-005271-16; Clinical Trial Authorisation number 41692/0001/001-0001). Ethics approval was granted by the East of Scotland NHS Research Ethics Committee (reference number 12/ES/0023). The trial was co-sponsored by the University of Dundee and NHS Tayside (Tayside Academic Health Sciences Collaboration). The trial was registered at www.isrctn.com with the identifier ISRCTN09486651.

Participants

Randomisation and treatment allocation

Randomisation was performed using an interactive web-based randomisation, drug assignment and inventory management system [Tayside Randomisation SysTem (TRuST)] run by the Health Informatics Centre, University of Dundee. The system was run independently of the research team to preserve allocation concealment. Randomisation was performed in a 1 : 1 ratio, stratified by site, and employed a minimisation algorithm to balance male versus female sex, CKD category 4 versus category 5, and age < 75 years versus ≥ 75 years.

Participants were allocated study medication bottles (one bottle per month) containing either 500-mg sodium bicarbonate tablets or matching placebo tablets; bottles were allocated based on bottle identification numbers generated by the TRuST randomisation system.

Intervention and comparator

The trial intervention consisted of either 500-mg sodium bicarbonate tablets or matching placebo tablets (containing lactose and microcrystalline cellulose). Active and placebo tablets were manufactured and bottled by Legosan AB (Kumla, Sweden). Bottles were imported to the UK via Tayside Pharmaceuticals (Dundee, UK), which undertook quality testing and qualified person release and distributed bottles to participating sites. Study medications were held at site pharmacies under temperature-controlled conditions prior to dispensing to participants. For the first 3 months of participation, participants were instructed to take one tablet three times per day.

Outcomes measurement

Outcomes were measured at baseline and at 3, 6, 12 and 24 months. Outcomes were collected by research nurses at each site, who were masked to treatment allocation. Figure 1 shows the study processes at each visit, from screening to the end of trial participation.

FIGURE 1.

Flow of participant visits and activities through the trial. AE, adverse event; BMI, body mass index; BP, blood pressure; Con Meds, concomitant medications; IMP, Investigational Medicinal Product; KDQoL, Kidney Disease Quality of Life; U+Es, urea and electrolytes.

Data management

Trial data were collected onto paper case report forms and then entered onto a trial-specific database built using OpenClinica software V3.1.2 (OpenClinica LLC, Waltham, MA, USA). Participants were identified using a unique study identifier and data were stored on a secure, backed-up, University of Dundee server system. Source data verification was conducted for all randomised participants for age, sex, inclusion and exclusion criteria, laboratory values analysed as part of routine clinical practice, baseline medications and adverse events. Batch validation and database audit procedures were run as outlined in the trial Data Management Plan. Target error rates for the primary outcome and adverse events were set at < 0.5% and for other audited fields were set at < 2%, with corrections made until error rates fell within these limits.

Safety reporting

All adverse events (serious and non-serious) were collected at each site using adverse event logs. Adverse events were coded centrally by System Organ Class and Preferred Term using the Medical Dictionary for Regulatory Activities (MedDRA) coding dictionary version 16.1 (www.meddra.org). Given the anticipated high frequency of adverse events in this study population, serious adverse events were collected but not reported to the trial sponsor or to the regulatory authority (Medicines and Healthcare products Regulatory Agency) if they fell into the following categories:

• any new cardiovascular event

• any new diagnosis or treatment of cancer

• any death or hospitalisation as a result of a fall or fracture

• any death or hospitalisation as a result of infection

• any death or hospitalisation as a result of exacerbation of an existing medical condition

• any admission for elective or planned investigation or treatment

• death, admission or treatment for deteriorating renal function or high or low potassium concentrations.

All adverse events (including those in the above list) were presented to the independent Data Monitoring Committee (DMC) classified by MedDRA System Organ Class; prespecified outcomes of particular interest (death, worsening heart failure, fluid overload or breathlessness) were also presented. All adverse events are included in the analysis reported here.

Trial oversight committees

An independent DMC met every 6 months. The DMC comprised an experienced trials biostatistician, an academic geriatrician and an academic nephrologist. The DMC had access to unblinded data on baseline participant characteristics and adverse events. The DMC reported to the chairperson of the independent Trial Steering Committee (TSC); members were appointed by the NIHR and operated under an agreed charter.

The independent TSC was appointed by the NIHR and was chaired by an experienced triallist specialising in geriatric medicine. Other independent members of the TSC were an academic nephrologist, an academic geriatrician and a lay member with personal experience of kidney disease. The TSC met at least every 6 months over the course of the trial; additional meetings were held as required for timely decision-making. The TSC chairperson reported to the project manager at the NIHR by letter and provided minutes after each TSC meeting; the TSC also operated under an agreed charter.

Day-to-day management of the trial was performed by the Trial Management Group (TMG), comprising the lead applicant, co-applicants and Tayside Clinical Trials Unit (CTU) staff. Local investigators and research nurses at each site were invited to join all TMG meetings, which took place monthly until the end of recruitment and then every 2 months until the end of the grant funding period. Monthly teleconferences between the trial manager and research nurses were used to share best recruitment practice and to troubleshoot trial processes.

Important changes to the trial design and conduct after trial commencement

Several significant changes were made to the conduct of the trial after commencement, mostly in response to the slow recruitment rates:

• The number of sites planned was originally six; this was expanded to 27 to address the slow recruitment rates.

• A substudy at two sites (Dundee and Aberdeen) was originally planned to examine bone mineral density by dual-energy X-ray absorptiometry (DEXA) and vascular stiffness by applanation tonometry. These substudies were discontinued because of poor recruitment rates, with only two participants undergoing the substudy measurements.

• The exclusion criteria were relaxed early in the recruitment phase following a review of reasons for non-recruitment. Changes were made to reduce the lower age limit from 65 to 60 years, to allow the inclusion of those taking calcium acetate or sevelamer and to allow the inclusion of those with hypertension controlled according to home monitoring despite high office blood pressure readings. Both of the phosphate binders, calcium acetate and sevelamer, are routinely used alongside bicarbonate in clinical practice and home monitoring of blood pressure is increasingly used in clinical practice to determine the adequacy of blood pressure control. These changes were therefore deemed not to compromise the safety or scientific integrity of the trial but likely to enhance the generalisability of the results by expanding the pool of eligible participants. In addition, provision was made to include patients currently taking sodium bicarbonate if they underwent a 3-month washout period.

• The TSC took the decision, in conjunction with the funder, to stop recruitment once 300 of the original target of 380 participants had been randomised. This decision was taken in view of the slowing recruitment rates; the sample size calculations were revisited prior to this decision being taken, as discussed in Re-estimation of the sample size.

• The TSC took the decision, in conjunction with the funder, to truncate follow-up once all participants had reached the primary outcome point of 12 months. This decision was taken to enable a prompt conclusion to the trial so that the results could be disseminated in a timely fashion; it was not based on an interim analysis of the results. A small number of individuals did not therefore progress to the 24-month follow-up point.

• Two extensions to the recruitment time were granted by the NIHR to compensate for the slower than anticipated recruitment rates.

Re-estimation of the sample size

To inform the decision on whether or not to terminate recruitment once 300 participants had been randomised, a revised sample size calculation was prepared by the research team and was considered by the TSC. This calculation was carried out without knowledge of treatment allocation or any follow-up data beyond baseline values and standard deviations for the trial population.

The revised sample size calculation assumed the use of a mixed-model repeated-measures analysis with two time points (12-month follow up and baseline), a standard deviation of 2.6 for the primary outcome of SPPB score, an attrition rate of 30% by 12 months and an alpha of 0.05. Assuming a within-person correlation of 0.7, 300 participants would give 85% power to detect a 1-point difference in SPPB score between groups (the MCID) at 12 months. Assuming a within-person correlation of 0.6, the same sample size would give 87% power to detect this difference.

Based on these revised power estimates, the TSC recommended that recruitment stop at 300 participants, as sufficient power to detect the MCID for the primary outcome would still be present and continuing recruitment would risk greatly prolonging the trial with little benefit in terms of statistical power.

Statistical analysis

A prespecified statistical analysis plan (SAP) was drafted, reviewed by the TMG and the independent TSC and signed off before the last visit of the last participant (see Report Supplementary Material 2).

The primary outcome (between-group difference in SPPB score at 12 months) was analysed using linear mixed models, adjusted for baseline measurements, minimisation variables (age, sex and stage of CKD) and a random effect variable for recruitment site. Prespecified subgroup analyses for the primary outcome were conducted (SPPB ≥ 10 points vs. < 10 points, CKD category 4 vs. category 5, age < 75 vs. ≥ 75 years, male vs. female, baseline bicarbonate < 18 mmol/l vs. ≥ 18 mmol/l). These factors were selected as being both of clinical interest and likely to be related to the primary outcome based on previous work. Sensitivity analyses were planned for > 80% versus ≤ 80% adherence, exclusion of those undergoing washout prior to randomisation and using multiple imputation for missing data. Multiple imputation was performed with SAS^® PROC MI v9.4 (SAS Institute Inc., Cary, NC, USA) using a Markov chain Monte Carlo method with multiple chains over 1000 iterations (SAS and all other SAS Institute Inc. product or service names are registered trademarks or trademarks of SAS Institute Inc. in the USA and other countries. ^® indicates USA registration). Predictor variables were visit, sex, age group, CKD stage and practice.

Secondary outcomes were analysed using repeated-measures models, adjusted for baseline values and minimisation variables as above. Time-to-event analyses (time to death, time to commencing renal replacement therapy) were conducted using Cox proportional hazards models adjusted for minimisation variables as above. For all analyses, a two-sided p-value of < 0.05 was taken as significant, with no adjustment for multiple testing. Analyses were performed using SAS v9.4 software. Unmasking of randomisation groups was performed only after completing the statistical analysis.

Health economic analysis

A prespecified health economic analysis plan was also drafted, reviewed by the TMG and signed off before the last visit of the last participant (see Report Supplementary Material 1).

The primary aim was to assess the cost-effectiveness of the addition of bicarbonate therapy relative to placebo from the perspective of the UK health and social care system. A cost–utility analysis was undertaken that involved estimation of the incremental costs and incremental effects [effectiveness was measured using quality-adjusted life-years (QALYs), based on responses to the EQ-5D-3L instrument]. Estimation was performed using generalised linear regression modelling, with adjustment for skewed data and for baseline differences in cost, EQ-5D-3L values and other patient characteristics (age, sex, stage of CKD). Non-parametric bootstrap methods50 were used for calculating CIs around cost and QALY differences. Cost-effectiveness acceptability curves were employed to show the probability that bicarbonate therapy was cost-effective for different values of willingness to pay per additional QALY. 51

Recruitment

A total of 300 participants were randomised into the trial between May 2013 and February 2017. Appendix 5 shows the cumulative recruitment per month throughout the trial recruitment phase (see Figures 26 and 27) and recruitment by site (see Table 23). Following discussion between the trial team, the independent DMC and TSC and the funder, recruitment was terminated at the end of February 2017 because of the very low ongoing recruitment rates. As part of this decision-making process, revised sample size/power calculations indicated that, under the proposed analysis method for the primary outcome, the trial had 87% power to detect the MCID with the recruited sample size.

Flow of participants through the trial

Figure 2 shows the flow of participants through the trial, using the format recommended by the Consolidated Standards of Reporting Trials (CONSORT). 52 Dropout rates were similar in both arms at each time point, but overall dropout rates were slightly higher than anticipated (18% at 6 months and 27% at 12 months). Once all participants had completed their 12-month (primary outcome) visit, the TSC recommended that further follow-up for the last 40 participants be truncated; the last patient visit occurred in February 2018. These participants did not therefore progress to their 18-month or 24-month visit and thus dropout after 12 months appears artefactually higher. Only four participants underwent the 3-month washout option prior to the screening visit.

FIGURE 2.

Consolidated Standards of Reporting Trials (CONSORT) flow diagram. a, Some did not attend visit but had not dropped out. A, withdrew because of adverse event; C, early study completion because of truncated follow-up; D, died; L, lost to follow-up; O, other; P, participant chose to withdraw; W, withdrawn by investigator.

Participant baseline characteristics

Participant baseline characteristics are presented in Table 2. The two groups were well balanced for most key baseline characteristics, including aetiology of renal dysfunction.

Adherence and effect of the intervention on serum bicarbonate levels

Data on adherence to the study medication are shown in Table 3. The adherence rate was moderate, with approximately 50% in both arms exceeding the threshold of 80% commonly used to denote good adherence. The mean prescribed dose of bicarbonate in the bicarbonate arm across the whole follow-up period was 1.88 g per day (compared with a maximum possible dose of 3 g per day) and the mean ingested dose of bicarbonate in the bicarbonate arm across the whole follow-up period was 1.39 g per day.

A modest but significant increase in serum bicarbonate concentration was seen in the intervention arm by 3 months; this difference attenuated with time and was no longer significant by 24 months, as shown in Figure 3. The treatment effect of bicarbonate supplementation across the whole follow-up period was 1.1 mmol/l (95% CI 0.6 to 1.6 mmol/l; p < 0.001); bicarbonate levels at each follow-up time point are shown later in this chapter (see Table 7).

FIGURE 3.

Change in bicarbonate concentration relative to baseline.

Primary outcome

Table 4 shows the primary outcome analysis: the difference in SPPB score at 12 months between groups. No significant effect of bicarbonate treatment was seen on the primary outcome (treatment effect –0.4 points, 95% CI –0.9 to 0.1 points; p = 0.15); analysis adjusted only for baseline SPPB score and analyses adjusted for baseline SPPB score, age, sex and CKD category gave the same result. Multiple imputation to account for missing data gave similar results (treatment effect –0.3 points, 95% CI –1.0 to 0.3 points; p = 0.29). As only four participants underwent the washout period prior to randomisation, the sensitivity analysis excluding this subgroup was not conducted.

Subgroup analyses

No significant interaction was seen in subgroup analyses (age, sex, CKD category, high vs. low baseline bicarbonate concentration, high vs. low baseline SPPB score), with all subgroups showing similar effect sizes for the primary outcome; all tests for interactions were non-significant. Details are presented in Figure 4.

FIGURE 4.

Subgroup analyses for the primary outcome.

Secondary outcomes

Renal function

No significant treatment effect was seen on renal function between the bicarbonate arm and the treatment arm on repeated-measures analysis; this was consistent whether using serum creatinine concentration, eGFR derived from serum creatinine or serum cystatin C concentration as alternative markers of renal function. There was also no significant treatment effect between arms on the urinary albumin-to-creatinine ratio. Details are presented in Table 7. For those remaining under follow-up in the trial, the rates of decline in eGFR were low, as shown in Figure 5.

TABLE 7 - Secondary outcomes: measures of renal function and associated biochemistry

IQR, interquartile range.

Repeated-measures analysis of variance adjusted for baseline values, age, sex and CKD category.

Repeated-measures analysis of variance adjusted for baseline values, age and sex.

Repeated-measures analysis of variance: log-transformed value.

Outcome	Time point	Randomised group		Treatment effecta (95% CI)	p-value
		Bicarbonate	Placebo
Bicarbonate (mmol/l), mean concentration (SD)	Baseline	20.6 (2.6) (n = 152)	20.1 (2.5) (n = 148)	1.1 (0.6 to 1.6)	< 0.001
	3 months	22.4 (2.7) (n = 137)	20.7 (3.4) (n = 133)
	6 months	22.3 (2.7) (n = 124)	21.1 (3.2) (n = 116)
	12 months	22.5 (2.6) (n = 107)	21.4 (3.9) (n = 98)
	24 months	22.9 (4.1) (n = 79)	22.5 (3.3) (n = 77)
eGFR (ml/minute/1.73 m2), mean concentration (SD)	Baseline	19.7 (6.5) (n = 152)	18.2 (6.4) (n = 148)	0.6b (–0.8 to 2.0)	0.39
	3 months	18.8 (6.4) (n = 137)	18.7 (7.6) (n = 133)
	6 months	19.0 (7.3) (n = 126)	18.6 (8.0) (n = 117)
	12 months	17.9 (7.6) (n = 112)	18.1 (7.7) (n = 101)
	24 months	19.5 (10.2) (n = 79)	18.0 (8.2) (n = 78)
Creatinine (µmol/l), mean concentration (SD)	Baseline	289 (101) (n = 152)	307 (103) (n = 148)	–8b (–28 to 13)	0.46
	3 months	305 (118) (n = 137)	309 (116) (n = 133)
	6 months	311 (138) (n = 126)	313 (120) (n = 117)
	12 months	341 (177) (n = 112)	320 (140) (n = 101)
	24 months	319 (150) (n = 79)	332 (150) (n = 78)
Cystatin C (mg/l), mean concentration (SD)	Baseline	3.11 (0.74) (n = 143)	3.14 (0.74) (n = 136)	–0.01b (–0.17 to 0.14)	0.89
	3 months	3.20 (0.88) (n = 129)	3.21 (0.85) (n = 120)
	6 months	3.21 (0.87) (n = 121)	3.21 (0.96) (n = 107)
	12 months	3.41 (1.09) (n = 103)	3.33 (1.04) (n = 93)
	24 months	3.39 (1.20) (n = 71)	3.35 (1.03) (n = 72)
Urinary albumin-to-creatinine ratio, median (IQR)	Baseline	23 (7–79) (n = 142)	22 (7–100) (n = 135)	0.32c (–0.05 to 0.70)	0.09
	3 months	32 (8–112) (n = 131)	26 (5–99) (n = 127)
	6 months	25 (7–106) (n = 113)	19 (7–76) (n = 110)
	12 months	25 (7–115) (n = 102)	21 (8–71) (n = 90)
	24 months	19 (6–91) (n = 63)	21 (5–53) (n = 69)

FIGURE 5.

Change in eGFR relative to baseline.

No difference was seen in the time to meeting criteria for a composite end point of decline in renal function, defined as a doubling of the baseline creatinine concentration, a 40% reduction in eGFR from baseline or commencement of renal replacement therapy. The hazard ratio (HR) by Cox proportional hazards modelling for reaching this composite end point, adjusted for age, sex and CKD category at baseline, was 1.03 (95% CI 0.66 to 1.63; p = 0.88).

Vascular health

No significant treatment effect was found for blood pressure, total cholesterol or N-terminal pro-B-type natriuretic peptide on repeated-measures analysis. The results are presented in Table 8 and the change in blood pressure relative to baseline in each group is shown in Figure 6.

TABLE 8 - Secondary outcomes: markers of vascular health

IQR, interquartile range; NT-pro-BNP, N-terminal pro-B-type natriuretic peptide.

Repeated-measures analysis of variance adjusted for baseline values, minimisation variables, age, sex and CKD category.

Repeated-measures analysis of variance: log-transformed value.

Outcome	Time point	Randomised group		Treatment effecta (95% CI)	p-value
		Bicarbonate	Placebo
NT-pro-BNP (pg/ml), median concentration (IQR)	Baseline	5910 (1678–10,221) (n = 115)	4453 (1555–10,521) (n = 115)	0.13 (–0.18 to 0.44)b	0.42
	12 months	6809 (1651–12,691) (n = 91)	4158 (1725–9743) (n = 79)
	24 months	5062 (1748–10,357) (n = 80)	5653 (2369–13,287) (n = 72)
Total cholesterol (mmol/l), mean (SD)	Baseline	4.3 (1.0) (n = 144)	4.2 (1.1) (n = 141)	0.1 (–0.2 to 0.3)	0.58
	3 months	4.4 (1.1) (n = 135)	4.2 (1.1) (n = 125)
	6 months	4.4 (1.1) (n = 124)	4.2 (1.1) (n = 113)
	12 months	4.4 (1.2) (n = 104)	4.3 (1.2) (n = 96)
	24 months	4.4 (1.1) (n = 72)	4.4 (1.2) (n = 77)
Systolic blood pressure (mmHg), mean (SD)	Baseline	143 (18) (n = 152)	143 (18) (n = 148)	0 (–4 to 3)	0.93
	3 months	143 (20) (n = 138)	143 (19) (n = 133)
	6 months	140 (20) (n = 128)	141 (16) (n = 117)
	12 months	143 (21) (n = 114)	143 (16) (n = 103)
	24 months	143 (21) (n = 81)	142 (18) (n = 79)
Diastolic blood pressure (mmHg), mean (SD)	Baseline	75 (11) (n = 152)	73 (10) (n = 148)	1 (–1 to 3)	0.16
	3 months	75 (10) (n = 138)	73 (11) (n = 133)
	6 months	74 (12) (n = 128)	73 (11) (n = 117)
	12 months	74 (11) (n = 114)	73 (9) (n = 103)
	24 months	76 (11) (n = 81)	72 (10) (n = 79)

FIGURE 6.

Change in blood pressure relative to baseline. (a) Change in systolic blood pressure (SBP); and (b) change in diastolic blood pressure.

Adverse events

A large number of adverse events (n = 857) were recorded, as expected for a trial enrolling older patients with extensive multimorbidity. In total, 263 out of 300 (88%) participants experienced at least one adverse event. Adverse events were more frequent in the bicarbonate arm (457 vs. 400), with a notable excess of events coded as gastrointestinal (45 vs. 25), musculoskeletal (28 vs. 17), cardiac (32 vs. 19), nervous system (24 vs. 12) and respiratory (26 vs. 14). Full details are presented in Table 11.

For cardiac adverse events, no difference in the number of episodes of decompensated heart failure was seen (8 vs. 10), but myocardial infarction was more common in the bicarbonate arm (10 vs. 2) (Table 12).

Need for renal replacement therapy

In total, 66 out of 300 (22%) participants commenced dialysis or underwent renal transplantation during the trial, with no difference between the bicarbonate arm and the placebo arm (33 vs. 33; p = 1.0) (see Table 12). Time-to-event analysis showed no significant difference in the HR between groups (HR 1.22, 95% CI 0.74 to 2.02; p = 0.43) (Figure 7). Similar results were seen for time to a composite outcome of either doubling of serum creatinine concentration or commencement of renal replacement therapy (HR 1.16, 95% CI 0.73 to 1.84; p = 0.53).

FIGURE 7.

Time to commencement of renal replacement therapy. HR (adjusted for age, sex and CKD category): 1.22 (95% CI 0.74 to 2.02; p = 0.43).

Deaths

Twenty-six deaths were recorded during the trial, with a similar number of deaths in each arm (bicarbonate vs. placebo: 15 vs. 11; p = 0.45) (see Table 12). Time-to-event analysis showed no significant difference in HR between groups (HR 1.30, 95% CI 0.60 to 2.83; p = 0.51) (Figure 8).

FIGURE 8.

Time to death. HR (adjusted for age, sex and CKD category): 1.30 (95% CI 0.60 to 2.83; p = 0.51).

Meta-analysis of outcomes with the BiCARB trial data included

Figures 9–15 show the results of the meta-analyses of existing trials of bicarbonate therapy, but with the BiCARB trial results added. The increase in serum bicarbonate seen with treatment in the BiCARB trial was lower than that seen in most other trials, and the favourable effect on eGFR seen in other trials was also not seen in the BiCARB trial. Meta-analyses including BiCARB data showed no significant effect of bicarbonate treatment on weight, mid-arm muscle circumference or systolic blood pressure. Heterogeneity was high across all analyses, as shown by the high I² values.

FIGURE 9.

Meta-analysis: difference in serum bicarbonate concentration (mmol/l) (any time point). IV, instrumental variable; SD, standard deviation.

FIGURE 10.

Meta-analysis: difference in serum bicarbonate concentration (mmol/l) (1-year follow-up only). IV, instrumental variable; SD, standard deviation.

FIGURE 11.

Meta-analysis: difference in eGFR (ml/minute/1.73 m²) (any time point). IV, instrumental variable; SD, standard deviation.

FIGURE 12.

Meta-analysis: difference in eGFR (ml/minute/1.73 m²) (1-year follow-up only). IV, instrumental variable; SD, standard deviation.

FIGURE 13.

Meta-analysis: difference in systolic blood pressure (mmHg) (any time point). IV, instrumental variable; SD, standard deviation.

FIGURE 14.

Meta-analysis: difference in weight (kg) (any time point). IV, instrumental variable; SD, standard deviation.

FIGURE 15.

Meta-analysis: difference in mid-arm muscle circumference (cm) (any time point). IV, instrumental variable; SD, standard deviation.

Key findings

This pragmatic, multicentre RCT found that administration of oral sodium bicarbonate using a dose regimen similar to that currently used in UK practice did not improve physical function or quality of life or slow down deterioration of renal function compared with placebo in older people with category 4 or 5 CKD and a serum bicarbonate concentration of < 22 mmol/l. Consistent with these findings, the health economic analysis showed that treatment with bicarbonate was less cost-effective than placebo. Analyses of markers of bone and vascular health found no evidence that bicarbonate improved bone health; blood pressure was neither raised nor lowered significantly by bicarbonate therapy compared with placebo. Bicarbonate therapy was moderately well adhered to. Overall adverse event rates, including that for falls, were higher in the bicarbonate group than in the placebo group despite the two groups being balanced for most comorbidities and medications at baseline. Bicarbonate is known to cause gastrointestinal side effects, such as nausea and bloating, and this was reflected in the higher rate of gastrointestinal side effects in the intervention group. Higher rates of cardiac, respiratory and neurological adverse events were seen in the bicarbonate group, with lower rates of cancer. These differences may be the result of chance, but pooling of these data with the results from other trials of bicarbonate is required to investigate the safety profile further. The excess of cardiovascular events in the intervention group was driven by a higher rate of myocardial infarction, but the mechanism underpinning this is not clear as blood pressure was not increased in the intervention group. Fluid overload as a result of sodium load in patients with heart failure or other types of cardiovascular disease was a concern at the planning stage of the trial, but no difference in fluid overload events was seen between the two groups and decompensated heart failure events were not responsible for the observed excess of cardiovascular events.

Although rates of death and renal replacement therapy were similar in both arms, it is concerning that several measures of physical function were worse in the bicarbonate arm than in the placebo arm, despite adjusting for baseline values. Differential dropout is unlikely to explain this finding as the dropout rate was similar in each arm. Although the difference in SPPB score on repeated measures was of borderline clinical significance, the degree of difference in grip strength and 6-minute walk distance is likely to be of clinical significance. 53 These findings, in conjunction with the higher adverse event rate in the bicarbonate arm, call into question not only the efficacy but also the safety of using bicarbonate in this target patient population.

Results in context

For many of the outcomes measured in the BiCARB trial it was not possible to combine the data with those from previous trials. This was particularly the case for physical function and quality of life data, which have not been a focus of outcome measurement in previous trials. As this is the first trial to examine the effects of bicarbonate supplementation on physical function measures, it is not possible to determine if the observed worsening in physical function measures with bicarbonate is a chance finding or a real effect. Existing observational data suggest a relationship between higher serum bicarbonate concentrations and better physical function,14,15 and a meta-analysis of studies testing the effect of bicarbonate on elite athletic performance found a weak, but significant, positive effect of bicarbonate on the performance of middle-distance athletes. 54 We were unable to find evidence of a biological mechanism that would convincingly explain why bicarbonate supplementation should worsen physical function.

Adding the results of the BiCARB trial to the meta-analysis findings showed no overall effect of bicarbonate on weight or mid-arm muscle circumference. Similarly, there was no significant worsening of systolic blood pressure with bicarbonate treatment, both overall and when confining analyses to 1-year outcomes. When combining all available trials in meta-analysis, overall estimates of the effect of oral bicarbonate therapy on serum bicarbonate levels and eGFR still suggest a modest benefit, but the results are highly heterogeneous, suggesting that the population studied in the BiCARB trial may respond differently for these outcomes from other populations studied to date, which differ in their predominant aetiology, age and ethnic background. It is also important to note that there was considerable heterogeneity in the length of follow-up, treatment schedule and whether or not patients, clinicians and study teams were masked to treatment allocation. Restricting analyses to a single time point (12 months’ follow-up) removes some, but not all, of this heterogeneity. In addition, measures of variance for outcomes in some trials included in the meta-analysis were much lower than would be expected, raising the possibility that standard errors had been reported rather than standard deviations, as advertised in the papers. It is not possible to verify this without access to individual participant-level data, but the use of standard errors would lead to these trials being given undue weight in the meta-analysis results. Results from these meta-analyses should be treated with caution because of these limitations.

Generalisability and limitations with regard to generalisability

Trials of oral bicarbonate therapy to date have targeted a wide range of groups: those with serum bicarbonate concentrations in the normal range as well as those with low serum bicarbonate concentrations; those with specific renal conditions (e.g. CKD of unknown origin, hypertensive nephropathy with albuminuria); and those with moderate renal disease (CKD category 3), which is not a usual target for bicarbonate therapy. This trial is distinctive in targeting older people, who make up the majority of patients with CKD in the UK and most other high-income countries. It included patients from across the UK with advanced CKD (thus reflecting the usual target group for bicarbonate therapy in practice).

The increase in serum bicarbonate concentration seen in the bicarbonate arm compared with the placebo arm was modest and it could be argued that the lack of benefit with regard to the main study outcomes is unsurprising given this modest increment in serum bicarbonate concentration. However, the increase in serum bicarbonate concentration was consistent with that seen in previous trials, including a recent trial published only in abstract form,55 albeit at the lower end of the range of responses seen. Furthermore, several trials demonstrating larger increases in serum bicarbonate concentration used an open-label design and titrated treatment to reach specific bicarbonate targets. In current UK practice, doses of 1.5–3 g per day of oral bicarbonate are commonly used and our treatment strategy reflected this practice. Although we cannot therefore rule out a beneficial effect of higher doses of bicarbonate, our results support the contention that current UK practice for bicarbonate replacement in CKD does not improve a wide range of outcomes in older patients with advanced CKD. Increasing the dose of sodium bicarbonate beyond 3 g per day risks further worsening of adherence, as bicarbonate tablets are large and difficult for older people to swallow. In addition, higher doses run the risk of increasing the rate of adverse events still further, particularly gastrointestinal side effects. Nevertheless, treatment regimens (using either bicarbonate or newer acid-binding agents) that produce greater increases in serum bicarbonate concentration could provide benefits and require testing.

The commissioning brief and trial design focused on patients with a mild degree of acidosis. Few participants with a serum bicarbonate concentration of < 18 mmol/l were enrolled, as most patients with a serum bicarbonate concentration this low are already being treated with bicarbonate. We are therefore unable to comment on the potential benefits of treating more severe levels of acidosis in this population.

Other limitations

One of the key limitations of this study is the high proportion of white participants. One previous UK trial23 that showed positive effects of oral sodium bicarbonate treatment enrolled predominantly participants of South Asian and African ethnicity – there was an insufficient number of participants to exclude a beneficial effect in these ethnic groups. The trial enrolled a preponderance of men, which may limit the generalisability of the results, given the relative under-representation of women. The study population had relatively stable CKD, with low rates of progression to end-stage kidney disease, and this was consistent with the low levels of proteinuria in the study population.

The original target for recruitment in this trial (380 participants) was not reached, despite participants being recruited from 27 UK sites. This was, in part, because of a lack of clinical equipoise; surveys of UK practitioners performed by the trial team during the trial suggested that most nephrologists were treating mild degrees of acidosis with bicarbonate already, thus reducing the pool of eligible participants. Despite this, revised sample size calculations performed to inform the decision to cease recruitment suggested that the sample size randomised (n = 300) had 87% power to detect a 1-point difference in the primary outcome of change in SPPB score. Our results exclude a 1-point improvement in the primary outcome by a wide margin and also exclude a more conservative 0.5-point improvement (posited by some researchers as the MCID for the SPPB53).

Bicarbonate levels in the placebo group increased gradually over time, which reduced the difference in bicarbonate levels between the groups. Some of this increase is likely to have been the result of regression to the mean and some is likely to have been the result of the dropout of participants who started renal replacement therapy (who were more likely to have worse renal function and worse acidosis), but some may also have been the result of individuals in the placebo group stopping the study medication and starting unblinded bicarbonate therapy as part of routine practice. However, the impact of starting unblinded bicarbonate therapy is likely to have been small as only 18 participants in the placebo group stopped the study medication to start unblinded therapy.

The majority of participants who commenced renal replacement therapy were lost to follow-up at the point of commencing renal replacement therapy. The costs of ongoing renal replacement therapy for these participants could not be directly captured as part of the main health economics analysis, and quality of life after starting renal replacement therapy was also not ascertained for participants who dropped out of the trial. Renal replacement therapy costs were included in one of the economic analyses but other associated costs for patients who dropped out were not captured. The same number of participants commenced renal replacement therapy in each arm (33 in each arm), with the time to commencement of therapy being slightly shorter in the bicarbonate arm. The economic analysis is consistent with this finding, with overall costs higher in the bicarbonate arm. It is possible that unmeasured costs associated with, but not attributable to, renal replacement therapy (e.g. additional inpatient stays, additional adverse events) could influence the economic analysis. However, the fact that more adverse events were seen during the trial in the bicarbonate arm suggests that any unmeasured events would tend to accentuate, rather than reverse, the results of the health economic analysis as presented. Similarly, small differences between the groups in the time spent avoiding the commencement of renal replacement therapy are likely to have large effects on costs. However, the analysis of time to commencement of renal replacement therapy suggests that participants in the bicarbonate group started renal replacement sooner, and the addition of these renal replacement costs reinforced, rather than overturned, the results of the main economic analyses.

Strengths

The key strengths of this trial were its comparatively large size; adequate follow-up time; participant, clinician and researcher masking; and broad inclusion criteria. In contrast to almost all previous trials, our use of a placebo control reduced the opportunities for bias, particularly with respect to decision-making around the timing of commencement of renal replacement therapy. An additional key strength was the broad range of outcome measures examined, with a particular focus on physical function and quality of life. These are the outcomes that older people report are the most important to them, and this focus is of particular importance in this group of patients with extensive multimorbidity. A narrow focus on a single disease – even in patients with advanced CKD – is inappropriate in this group, and considering physical function and quality of life enables an assessment of the overall benefit of treatment to patients in a way that organ-specific measures do not.

Implications for health care

Bicarbonate therapy is currently in widespread use to treat mild degrees of acidosis in patients with stage 4 or 5 CKD. This is largely based on observational data suggesting an association between acidosis and a range of deleterious outcomes, including accelerated decline in renal function and adverse bone health, physical function and vascular health. There is little trial evidence to support current practice, a state of affairs that is acknowledged in guidelines. 31,32 Our results suggest that, at least in this predominantly male and white population of older patients with CKD category 4 and 5, 1.5–3 g per day of oral bicarbonate did not produce any health benefits and may be associated with net harms compared with placebo. Although other indications for the control of acidosis exist (e.g. high potassium concentrations), evidence from the current trial suggests that the additional cost, treatment burden and side effects of oral bicarbonate therapy may not justify its use in older people with advanced CKD and mild acidosis.

Suggestions for future research

A number of other trials of bicarbonate therapy are currently in progress or are in the process of being published. 55–57 These trials have targeted a range of CKD severities (CKD category 3b–5) and a range of entry serum bicarbonate concentrations (< 21 mmol/l, > 18 mmol/l and 20–25 mmol/l); in two of the trials, a strategy of dose adjustment to keep the serum bicarbonate concentration at > 24 mmol/l has been employed. None of these trials targets older people as a specific group. We recommend that the key research priority should be to combine data from these trials once they are available. We therefore make the following suggestions for further research:

• An individual participant meta-analysis should be conducted, examining the effects of bicarbonate therapy on physical function, quality of life, renal function and progression to renal replacement therapy, anthropometric measurements and bone and vascular health.

• Importantly, such a meta-analysis should also seek to pool adverse events, particularly cardiovascular events, and to identify the characteristics of those most likely to respond to bicarbonate therapy.

• The results from the BiCARB trial call into question the usefulness of bicarbonate therapy in other groups in whom it is currently used routinely. Depending on the results of meta-analyses, it may be necessary to formally test the effectiveness of bicarbonate therapy in other groups with CKD, for example younger patients.

• Alternative methods to manage acidosis in advanced CKD should be tested. A pilot study of dose titration to target would be one way to approach low bicarbonate levels once it is clearer which subgroups (if any) are more likely to benefit. Novel methods of managing acidosis, such as use of non-absorbed hydrochloric acid binders, also require testing.

A final recommendation is that clinical trials in disease areas that affect predominantly older people should be designed and executed in such a way that older people are well represented and should use outcome measures that are important to older people. 58 The BiCARB trial shows how this can be successfully achieved in the field of CKD; there is a need to ensure that similar outcomes and methods are used to answer other questions within the field of CKD, but also more widely in other organ-specific fields of clinical practice. The evidence base underpinning the design and delivery of trials that are appropriate for older people is very limited; research is needed on how best to design trials in older people, how to recruit and retain older people successfully and the use of outcomes that are relevant to older people with multimorbidity.

Local investigators

Aimun Ahmed, Michael Almond, Gowrie Balasubramaniam, Kolitha Basnayake, Deepak Bhatnagar, Coralie Bingham, Anthony Chan, Alex Crowe, Gabor Cserep, Neill Duncan, Helen Eddington, David Goldsmith, Manivarma Kamalnathan, Adam Kirk, Kostantinos Koutroutsos, Stewart Lambie, Madhavan Menon, Biswa Mishra, Sandip Mitra, Jim Moriarty, Noshaba Naz, Johann Nicholas, Tom Picket, Satyanarayana Reddy, Kate Shiell, Paul Stevens, Wai Tse, Martin Wilkie and Georgia Winnett.

Research nurses

Abiola Ali, Rashid Almasarwah, Cecilio Andujar, Alexandra Bailey, Joyce Banda, Janet Bendle, Janet Blood, Karen Borwick, Judith Brade, Fiona Brailsford, Paula Brassey, Margaret Brunton, Georgina Butt, Teresa Byrne, Christine Catley, Karen Chalmers, Sally Chapman, Houda Chea, Zdenka Cipinova, Hazel Clark, Elizabeth Clarke, Laura Cockayne, Viv Colclough, Jacqueline Colnet, Carolyn Corbett, Joanna Cox, Linda Crighton, Sirjana Devkota, Michelle Dorrington, Louese Dunn, Aidan Dunphy, Mary Dutton, Gillian Eaglestone, Dana Foster, Paul Frattaroli, Viji George, Karen Hallett, Deborah Harrison, Audrey Hau, Lesley Haydock, Linda Hill, Martin Hogan, Wanda Ingham, Yvonne Jackson, Linda Johnson, Laura Johnstone, Juliet Jones, Simon Kaye, Melanie Kershaw, Deepsi Khatiwada, Hayley King, Rosemary Kirk, Sarah Knight, Praveen Kunjukunju, Beverley Lane, Tara Leedham, Fiona Leslie, Mandy McAndrew, Alexandra McCarrick, Julie McEntee, Fiona McNeill, Louise Morby, Judith Muir, Estelle Nambela, Laura O’Keeffe, Faith Okhuoya, Lesley Patience, Pamela Paton, Marrissa Plaza, Sonia Raj, Nayyar Rajinder, Angelo Ramos, Caroline Renton, Richard Rye, Anessa Sameja, Belinda Schaefer, Leanne Scott, Lorraine Shah-Goodwin, Isla Smith, Christina Summersgill, Justyna Szklarzewicz, Sandra Toth, Kate Trivedi, Jane Turner, Adan Usmani, Michael Villaruel, Lynn Vinall, Thomas Walters, Joyce Ward, Jane Watkins, Lynn Watkins, Stephanie Whittaker, Joanne Wilcox, Frances Williams, Diane Winstance-Smith, Jackie Wooding and Andrea Young.

Tayside Clinical Trials Unit support staff

Hasithi Bandara, Stephanie Haenicke, Furrah Hussain, Eva Lahnsteiner, Emma McKenzie and Andrew McKenzie.

Contributions of authors

Miles D Witham (https://orcid.org/0000-0002-1967-0990) (Professor of Trials for Older People; lead applicant) led the design and management of the trial, contributed to the analysis and drafted the report.

Margaret Band (https://orcid.org/0000-0001-7275-8468) (Senior Trial Manager) contributed to the design and management of the trial and contributed to writing and critical revision of the report.

Huey Chong (https://orcid.org/0000-0002-0768-1844) (Health Economist) conducted the health economic analysis and contributed to writing and critical revision of the report.

Peter T Donnan (https://orcid.org/0000-0001-7828-0610) (Professor of Biostatistics) contributed to the design and management of the trial, had a major role in the statistical analysis and contributed to critical revision of the report.

Geeta Hampson (https://orcid.org/0000-0001-8465-5247) (Consultant in Chemical Pathology and Metabolic Medicine; co-applicant) contributed to the initial design of the trial, management of the trial, analysis, and writing and critical revision of the report.

May Khei Hu (https://orcid.org/0000-0001-8346-1109) (Systematic Reviewer) led the systematic review and contributed to writing and critical revision of the report.

Roberta Littleford (https://orcid.org/0000-0003-4868-0132) (Assistant CTU Director) contributed to the design and management of the trial and critical revision of the report.

Edmund Lamb (https://orcid.org/0000-0002-5154-7351) (Consultant Biochemist; co-applicant) contributed to the initial design of the trial, management of the trial, analysis and critical revision of report.

Philip A Kalra (https://orcid.org/0000-0001-7652-1572) (Consultant Nephrologist; co-investigator) contributed to the design and management of the trial and critical revision of the report.

Gwen Kennedy (https://orcid.org/0000-0002-9856-3236) (Lead for Analytic Core Services) led the biochemistry analyses and contributed to writing and critical revision of the report.

Paul McNamee (https://orcid.org/0000-0002-4540-8718) (Professor of Health Economics; co-applicant) contributed to the initial design of the trial and management of the trial, led and supervised the health economic analysis and contributed to writing and critical revision of the report.

Deirdre Plews (https://orcid.org/0000-0002-7478-7913) (Trial Manager) contributed to the design and management of the trial and critical revision of the report.

Petra Rauchhaus (https://orcid.org/0000-0003-4994-155X) (Trial Statistician) contributed to management of the trial, led the statistical analysis and contributed to writing and critical revision of the report.

Roy L Soiza (https://orcid.org/0000-0002-1397-4272) (Consultant Geriatrician; co-applicant) contributed to the initial design of the trial, management of the trial, the systematic review, the main analysis and critical revision of the report.

Deepa Sumukadas (https://orcid.org/0000-0002-6832-3063) (Consultant Geriatrician; co-applicant) contributed to the initial design of the trial, management of the trial, analysis and critical revision of the report.

Graham Warwick (https://orcid.org/0000-0002-4178-015X) (Consultant Nephrologist; co-investigator) contributed to the design and management of the trial and critical revision of the report.

Alison Avenell (https://orcid.org/0000-0003-4813-5628) (Professor of Health Services Research; co-applicant) contributed to the initial design of the trial, management of the trial, analysis and critical revision of the report.

Publications

Witham MD, Band MM, Littleford RC, Avenell A, Soiza RL, McMurdo ME, et al. Does oral sodium bicarbonate therapy improve function and quality of life in older patients with chronic kidney disease and low-grade acidosis (the BiCARB trial)? Study protocol for a randomized controlled trial. Trials 2015;16:326.

Witham MD, Lamb EJ. Should chronic metabolic acidosis be treated in older people with chronic kidney disease? Nephrol Dial Transplant 2016;31:1796–802.

Hu MK, Soiza RL, Witham MD. Oral bicarbonate therapy in chronic kidney disease: a systematic review and meta-analysis of randomised controlled trial. J Clin Med 2019;8:ii:E208.

The BiCARB study group. Clinical and cost-effectiveness of oral sodium bicarbonate therapy for older patients with chronic kidney disease and low-grade acidosis (BiCARB): a pragmatic randomised, double-blind, placebo-controlled trial. BMC Med 2020;18:91.

Data-sharing statement

All data requests should be addressed to the corresponding author or the trial sponsor for consideration. Access to anonymised data may be granted following review.

Patient data

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 it is 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.

Disclaimers

This report presents independent research funded by the National Institute for Health 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, NETSCC, the HTA programme or the Department of Health and Social Care. If there are verbatim quotations included in this publication the views and opinions expressed by the interviewees are those of the interviewees and do not necessarily reflect those of the authors, those of the NHS, the NIHR, NETSCC, the HTA programme or the Department of Health and Social Care.

Search strategy

1. exp BICARBONATES [MESH] (24,176)

2. exp Sodium bicarbonate [MESH] (4387)

3. Bicarbonates [mp] (21,707)

4. 1 or 2 or 3 (24,458)

5. Chronic kidney disease [mp] (43,554)

6. Renal insufficiency, chronic [MESH] (17,753)

7. CKD [mp] (24,465)

8. Kidney failure, chronic [MESH] (89,084)

9. 5 or 6 or 7 or 8 (129,377)

10. 4 and 9 (826)

11. Limit 10 to (English language and RCT) (77)

Meta-analysed outcomes, before reporting of the BiCARB trial results

FIGURE 19.

Serum bicarbonate concentration (mmol/l) (any time point). IV, instrumental variable; SD, standard deviation.

FIGURE 20.

Serum bicarbonate concentration (mmol/l) (1-year follow-up only). IV, instrumental variable; SD, standard deviation.

FIGURE 21.

Estimated GFR (ml/minute/1.73 m²) (any time point). IV, instrumental variable; SD, standard deviation.

FIGURE 22.

Estimated GFR (ml/minute/1.73 m²) (1-year follow-up only). IV, instrumental variable; SD, standard deviation.

FIGURE 23.

Systolic blood pressure (mmHg) (any time point). IV, instrumental variable; SD, standard deviation.

FIGURE 24.

Weight (kg) (any time point). IV, instrumental variable; SD, standard deviation.

FIGURE 25.

Mid-arm muscle circumference (cm) (any time point). IV, instrumental variable; SD, standard deviation.