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

Study design

The trial design is a randomised, double-blind, parallel group, placebo controlled
trial, analysed by intention to treat. The treatment and follow-up will last 2 years
for each participant. Key outcomes will include measures of physical function, quality
of life, progression of renal disease, bone turnover, vascular health and adverse
events. A comprehensive health economic analysis is embedded in the trial.

Study population

The study will recruit 380 community-dwelling participants aged 60 and older with
GFR 30 mL/min/1.73 m
²
, not on renal replacement therapy, with a serum bicarbonate concentration of 22
mmol/L (the lower limit of the reference range in most UK centres). Potential participants
will be recruited from both primary care and through secondary care sites, for example,
renal clinics, Medicine for the Elderly clinics, general Medicine clinics, diabetes
clinics, hypertension clinics and cardiovascular clinics at each site. At each centre,
the local investigator and renal physicians, renal specialist nurses, geriatricians,
cardiologists and diabetologists will be assisted by the study research nurses to
identify potential participants. Where local patient registries are kept, these will
be exploited to search for potentially suitable participants.

Primary care recruitment will be carried out with the assistance of the primary care
research networks in Scotland (SPCRN) and England (CRN). Network involvement via the
Ageing and Renal UKCRN specialty groups will assist with recruitment, access to patients
and access to network research nurse time. Table 1 gives the inclusion and exclusion criteria for the trial. Potential participants
already taking oral bicarbonate therapy will have the option to enter the trial after
a 3-month washout period. Written informed consent will be obtained from all participants
prior to enrolment. Research ethics committee approval has been given by the East
of Scotland Research Ethics Committee (ref 12/ES/0023), the trial is approved by the
UK Medicines and Healthcare Regulatory Agency (EudraCT number 2011-005271-16) and
the trial is registered at www.isrctn.com (ISRCTN09486651). The trial is funded by the UK NIHR Health Technology Assessment
programme (ref: 10/71/01). The Sponsor is Tayside Academic Health Sciences Centre,
a joint initiative of the University of Dundee and NHS Tayside, and trial management
is via Tayside Clinical Trials Unit (UKCTU number 49).

Table 1. Inclusion and exclusion criteria for the BiCARB trial

Intervention

The study medication will be administered as oral tablets. Placebo and active tablets
will be identical in appearance. Bicarbonate tablets will contain 500 mg of sodium
bicarbonate (6 mmol sodium and 6 mmol bicarbonate); placebo tablets will be composed
of lactose. The Good Medical Practice (GMP)-compliant investigational medicinal product
manufacturer (Legosan AB, Kumla, Sweden) will prepare and bottle matching active and
placebo tablets for the trial. Participants will commence therapy at a dose of 500
mg of oral sodium bicarbonate or placebo three times per day; reflecting a common
starting dose in clinical practice. At 3 months, if serum bicarbonate is found to
be 22 mmol/L, the dose will be increased to 1 g three times per day or placebo. This
is equivalent to six tablets per day; larger doses are likely to worsen adherence
to therapy. This dose is higher (3 g versus 1.8 g) than that used in a previous small
RCT 13]; it is therefore likely to produce a larger increment in bicarbonate at an earlier
time point, thus enhancing the ability of our trial to demonstrate beneficial effects
on outcome parameters.

Randomisation will be via a centrally controlled web-based Good Clinical Practice
(GCP)-compliant randomisation system, run by Tayside Clinical Trials Unit (TCTU).
Investigators at each site will access the web-based system to allocate each participant
at the end of the baseline visit. Randomisation will be stratified by site. To ensure
balanced assignment across critical variables, a minimisation algorithm will be employed,
using baseline age, sex and GFR category to balance allocation across trial arms.

To ensure that medication adherence is maximised, we will employ a combination of
written information about the study medication and why taking it is important, together
with aide memoirs including BiCARB-branded mugs and fridge magnets. In the second
year of follow-up, regular telephone contact (at 15 and 21 months) will be made to
remind participants about the importance of medication adherence. Interventions incorporating
these components have been shown in a previous Cochrane review to enhance adherence
26]. If the study drug needs to be stopped or the patient wishes to stop they can remain
in the study in order to perform an intention-to treat-analysis.

Outcomes

The primary outcome for the trial is the between-group difference in change for the
Short Physical Performance Battery (SPPB) between baseline and 12 months. The SPPB
is a validated measure of lower limb function that reflects everyday activity. It
is a powerful predictor of subsequent disability, institutionalisation and death in
older people 27]–29], and has more data validating its use in older people than any other measure of physical
function. Secondary outcomes will include SPPB at other time points, along with measures
of physical function (grip strength as measure of upper limb function, and six minute
walk as a measure of endurance capacity), quality of life (generic, measured using
the EQ-5D tool, and disease-specific, measured using the KDQoL tool 30]), anthropometry, renal function and bicarbonate concentrations, markers of bone turnover,
calcium and phosphate metabolism, blood pressure and B-type natriuretic peptide. Key
outcomes are listed in Table 2.

Table 2. Outcome measures and measurement points

Adverse events and medication changes will be sought at each study visit, as will
data on hospitalisation, death (all-cause, cardiovascular and renal), commencement
of renal replacement therapy, fractures, primary care and secondary care outpatient
visits via participant health-utilisation diaries. Adherence to medication will be
assessed by tablet count on bottles returned to the trial pharmacy. Detailed information
on medical history, concomitant medications (including renal and cardiovascular medications
such as renin-angiotensin system blockers), social and functional history will be
collected at baseline. Participant flow through the trial is outlined in Fig. 1.

Statistical analysis

Statistical analysis will be performed by intention-to-treat and reported in accordance
with the CONSORT statement (www.equator-network.org). The primary analysis will be a between-group comparison of 12 month outcomes, adjusted
for baseline values of the outcome under test via regression analysis. Further adjustment
for age, sex and GFR category will also be performed. Secondary analyses will comprise
multivariate repeated measures analysis using all available time points, adjusted
for baseline differences. A per-protocol secondary analysis will be performed to compare
adverse events, and time-to-event analyses for death or onset of renal replacement
therapy will be performed using Cox proportional hazards regression models. A two-sided
P value of 0.05 will be taken as significant for each outcome. The primary outcome
will be analysed at 12 months rather than 24 months in order to strike a balance between
excessive dropout (expected in this group of older, multi-morbid participants) and
sufficient time for the intervention to show an effect.

Pre-specified subgroup analyses will include the following: participants with a SPPB
score of 10 (denoting a frailer subgroup) versus those with SPPB ?10; participants
with a baseline serum bicarbonate concentration of 18 mmol/L versus those with serum
bicarbonate ?18 mmol/L; GFR category (4 versus 5); sex, and age 75 years versus age
?75 years. Missing data for the primary analysis will be handled using multiple imputation,
provided that the assumption of missing at random is met. Missing data in the repeated
measures analyses will be handled using generalised estimating equation approaches
to maximise the use of participants with incomplete data.

Data analysis of health economic data

The economic analysis will estimate the healthcare costs and quality of life associated
with provision of bicarbonate therapy relative to usual care (that is, usual healthcare
management without the addition of bicarbonate therapy) over the time period of the
trial (12 months for the main analysis). It will determine the magnitude of the difference
in healthcare costs and quality of life between the two treatments. Healthcare costs
will include the type and duration of hospital admissions, frequency of visits to
hospital for outpatient attendances, and other visits to/from relevant healthcare
professionals (for example, general practitioners, nurse practitioners, and physiotherapists)
at baseline and over 12 months. Quality of life will be measured with the EQ-5D instrument
over an equivalent time period, and responses converted to a single summary score
using published UK quality of life weights, to produce a Quality Adjusted Life Year,
(‘QALY’) value. In addition to QALYs, secondary analyses will include the measurement
of well-being and life satisfaction. Cost effectiveness acceptability curves will
be employed to show the probability that bicarbonate therapy is cost effective for
different values of willingness to pay per additional QALY 31]. These will be developed using the quality of life gain and life expectancy change
estimates from the ANCOVA and repeated measures analysis. Cost differences will be
modelled using generalized linear modelling, with appropriate adjustment for skew
and other baseline differences between groups. In addition, non-parametric bootstrap
methods will be used for calculating confidence intervals around cost and QALY differences.
Sensitivity analysis will be undertaken for uncertain parameters, such as alternative
centre specific costs per bed day and alternative age-adjusted quality of life weights.

A detailed statistical analysis plan will be agreed on before the end of data entry
and before the treatment code is broken.

Sample size calculation

We plan to randomize a total of 380 participants: 190 to bicarbonate and 190 to placebo.
The clinically important difference (CID) for the SPPB is a 1-point difference. Assuming
a SD of 2.6 as found in previous work 32], we would require 143 patients per group to give 90 % power to detect this difference
at a two-sided alpha of 0.05. For the EQ-5D, the minimum CID is 0.074 33]. To detect this with two-sided alpha?=?0.05 and power of 90 %, assuming a SD of change
of 0.2, as found in our previous studies, our study requires 154 patients per group.
We anticipate a dropout rate of 10 % per 6 months, due to the high rates of expected
death and illness in this multi-morbid, older group of participants; the calculated
sample size accounts for this dropout.

Progression to dialysis in a previous trial of bicarbonate therapy 13] was lower in patients allocated to bicarbonate (7 %) than in the control arm (33
%). The proposed sample size will have 90 % power to successfully detect a more conservative
difference of 7 % in the bicarbonate arm and 18 % in the control arm at 2 years. Assuming
a 10 % loss to follow-up every 6 months (a pessimistic estimate, based on previous
medication trials in frail older people in our centre), we would require 380 patients
to ensure adequate power for the primary outcome and the EQ-5D at 12 months. This
rate of attrition would leave 250 evaluable patients at 2 years, which will give additional
power to detect smaller changes using repeated measures analyses.