Background
Patients with chronic kidney disease (CKD) have a significantly higher mortality compared to the general population and this increases as the estimated glomerular filtration rate (eGFR) declines [
1]. Thus, preservation of kidney function and avoidance of end-stage renal disease (ESRD) is a key therapeutic target. Among the recommendations for preservation of kidney function is the control of dietary sodium as a modifiable risk factor.
Recent international guidelines have incorporated sodium restriction into their recommendations. Kidney Disease Improving Global Outcomes (KDIGO) suggests a reduction to <2 g/day of sodium which corresponds to 5 g/day of salt for adult patients with CKD [
2]. These recommendations are based on low level evidence from studies with marked heterogeneity [
3], although the results overall suggest that modest sodium restriction should be beneficial for patients with CKD. However, there is a paucity of studies specifically addressing the effect of sodium intake in patients with advanced CKD (defined as eGFR < 30 ml/min/1.73 m
2). Studies that included individuals with advanced CKD have been limited by surrogate short-term outcomes or inclusion of non-diabetic patients [
4‐
6]. As such, in patients with advanced CKD, it remains unknown whether alterations in sodium intake impact clinically relevant outcomes, such as mortality, CKD progression and the need for dialysis.
The aim of our study was to determine if urinary sodium excretion is associated with mortality and need for renal replacement therapy in patients with advanced CKD. We hypothesized that higher levels of urinary sodium excretion (used as a surrogate for sodium intake) would be associated with adverse clinical outcomes.
Methods
Patient population and measurements
This is a retrospective cohort study using prospectively collected data on adult patients (>18 years) followed in the progressive renal insufficiency clinic at The Ottawa Hospital, a 1,150 bed academic tertiary care center serving a population of approximately 1.2 million, located in Ontario, Canada. The progressive renal insufficiency clinic is a specialty multi-disciplinary care clinic in the CKD program for patients approaching ESRD. Patients with progressive kidney disease are referred to this clinic at the discretion of primary nephrologist in anticipation of needing renal replacement therapy. Attempts are made to collect 24 h urine for sodium excretion at least twice per year. Standardised instructions are given to each patient for collection of 24 h urine.
The study included patients with eGFR < 30 ml/min/1.73 m
2 calculated by the 4 variable MDRD formula [
7], with at least two visits to the clinic and at least one 24-h urine sodium excretion (USE) measurement between January 2010 and December 2012. Patients who required any kind of renal replacement therapy prior to the second visit were excluded. The date of first visit was deemed the date of study entry.
Data collection
Data on all patients were obtained through the clinic’s comprehensive database since its inception in January 1st, 2010. The accuracy of the data in the database is verified every six months by auditing 5 % of entries and accuracy has been > 95 %. Data collected included age, gender, blood pressure (BP), body-mass index (BMI), eGFR, urine excretion of sodium, protein and creatinine, cause of CKD, comorbidities, medications, date of death and date of dialysis start or renal transplantation. Blood pressure was measured in sitting position after at least five minutes of rest. Mean arterial blood pressure (MAP) was calculated using the formula: MAP = 2/3 diastolic BP + 1/3 systolic BP. Urine sodium, urine creatinine and urine protein were all measured on the Siemens Vista 1500 analyzer using manufacturers’ reagents. If 24- h urine protein was not available, spot urine protein to creatinine ratios were converted to 24 h urine protein excretion [
8].
Blood pressure, eGFR and proteinuria were measured at each clinic visit. 24-h USE was performed periodically to monitor compliance and assist in counseling patients regarding sodium restriction.
Sodium intake for each patient was estimated from the average 24-h USE measured during the follow-up period. In order to account for inappropriate collection of 24-h urine, 24-h urine sodium excretion was corrected for expected creatinine excretion. Expected creatinine excretion was calculated by the following formulas: If male then creatinine excretion (mmol/day) = (28.2−(0.172*age))* 0.00884*weight (Kg); if female then creatinine excretion (mmol/day) = (21.9−(0.115*age))* 0.00884*weight (Kg) [
9].
A priori, we categorized the patients according to their mean 24-h USE into three groups: low sodium diet (LSD) -USE < 100 mEq/day, medium sodium diet (MSD) - USE 100–150 mEq/day and high sodium diet (HSD) - USE > 150 mEq/day. We selected these cut-off points first, to determine whether patients with advanced CKD have higher mortality or faster progression of CKD if their sodium intake is higher than currently recommended (>100 mEq/day), and second, to investigate whether daily sodium intake between 100 and 150 mEq (which is simpler to achieve than <100 mEq/day for most patients) is associated with different outcomes compared to LSD and HSD.
Outcomes
The primary outcome was a composite of progression to ESRD requiring any type of renal replacement therapy (dialysis or renal transplant) and mortality.
The secondary outcome was the decline in eGFR in ml/min/1.73 m2 per year.
Statistical analyses
Continuous variables are reported as mean (± SD) and categorical variables are reported as numbers and percentages. Analysis of variance with Tukey correction was used to compare continuous variables. If Levene's test for equality of error variances was positive, the data was log transformed. Chi squared test was used to compare categorical variables. Time-to-event analyses were conducted using the Kaplan-Meier (KM) method and Cox proportional hazards. The log rank test was used with the KM method to assess for statistical significance. To better quantify individual sodium exposure and account for within person data variability, sodium intake was modeled continuously using time-dependent Cox models using the mean 24-h urine sodium excretion. Potential confounders included in the Cox models were sex, age, baseline eGFR, mean arterial blood pressure (MAP), body mass index (BMI), diabetes, log transformed 24-h proteinuria, and renin-angiotensin system (RAS) blockers (angiotensin converting enzyme inhibitors and angiotensin receptor blockers) use. MAP and log transformed 24-h proteinuria were treated as time dependent variables, and changes in their values over time were incorporated into the adjusted Cox analyses. Sensitivity analysis was performed for sodium uncorrected for expected creatinine excretion as well as after censoring for death.
All statistical analyses were performed using SAS software (Version 9.3) and Medcalc (Version 13.2). P < 0.05 was considered statistically significant.
Discussion
In this retrospective cohort study of 341 patients with advanced CKD, we did not find an association between urinary sodium excretion (USE) and combined outcome of mortality and need for renal replacement therapy (RRT) or eGFR decline. Patients with high USE were more likely to be younger, had higher BMI, and were more proteinuric. There was no association between need for RRT and the mean USE as a categorical or continuous variable. Furthermore, there was no association between USE and the eGFR decline. Our study suggests USE (a surrogate for dietary sodium) may not play a significant role in the progression of CKD to end-stage renal disease in patients with severe kidney disease. To our knowledge, this is the first report on the association of USE on the need for renal replacement therapy and eGFR declines in a diverse CKD population with eGFR < 30 ml/min/1.73 m2.
There are no randomized control trials (RCTs) examining sodium intake and long-term renal outcomes, such as progression of CKD or ESRD and thus most of the evidence comes from observational studies. There are, however, short-term RCTs showing a beneficial effect of dietary sodium restriction on proteinuria and blood pressure [
6,
10]. This can suggest that lowering sodium intake may slow down the eGFR decline and thus postpone dialysis initiation. The Modified Diet in Renal Disease (MDRD) study that included 840 non-diabetic patients with CKD (mean eGFR 32.5 ml/min/1.73 m
2) did not show any association between urine sodium and kidney failure (which was defined as initiation of dialysis or transplantation) or the composite outcome of renal failure and all-cause mortality [
5]. Neither a recent post hoc analysis of the ONTARGET and TRANSCEND trials revealed an association between sodium intake and clinically important renal outcomes in patients with cardio-vascular risk factors [
11]. A limitation of these two trials was that the sodium intake was estimated from a spot morning urine sample using the Kawasaki formula, however, the large sample size (28, 880 patients), international representation of the cohort and completeness of data retrieval add considerable robustness to the conclusions. A secondary analysis of the REIN trials [
12] involving 500 non-diabetic patients with CKD (baseline creatinine clearance 42.5 ml/min) demonstrated that each 100 mEq/g increase in urinary sodium/creatinine excretion was associated with a 1.61-fold (95 % CI, 1.15–1.24) higher risk for ESRD. However, after adjusting for baseline proteinuria this effect was significantly attenuated to 1.38-fold (95 % CI; 0.95–2.00). A recent study by McQuarrie et al [
13] showed a similar finding – they demonstrated an association between USE and the risk of requiring RRT in CKD patients, but this was not independent of albuminuria or baseline eGFR.
More than half of the patients involved in our report had diabetes. The Finn Diane study that followed 2,807 type I diabetics with mean baseline eGFR > 80 ml/min/1.73 m
2 over 10 years, showed an inverse association of urine sodium and ESRD – macroalbuminuric patients with the lowest sodium excretion had the highest cumulative incidence of ESRD. [
14] The post-hoc analysis of RENAAL and IDNT yielded different results [
15]. The authors reported a significantly lower risk for doubling of serum creatinine or ESRD in the lowest tertile of sodium intake if the patients were on angiotensin receptor blocker treatment, compared to the control group treated with non-RAS antihypertensive agents. Of note, sodium intake was not randomized and there were significant differences in the study population characteristics such as lower eGFR and higher urinary albumin-to-creatinine ratio in the highest tertile of sodium intake (for which the analysis was not adjusted).
In our CKD study population, 58 % of patients (LSD and MSD group) had sodium intake within or close to the recommended target with HSD group represented a common salt intake of the general population in western and particularly in Asian countries [
16]. In spite of clear separation of the groups, we did not find an effect of lower sodium excretion resulting in less mortality and need for ESRD. It is also possible that sodium intake does not play a significant role in CKD progression once renal disease is very advanced. Another possible factor is salt sensitivity that as a mechanism may be different in CKD populations to the general population. Finally, marked dietary sodium restriction may have adverse consequences, including activation of the sympathetic and RAAS systems, reduction in insulin sensitivity, and increased LDL cholesterol [
17,
18]. Indeed, the relationship between sodium intake and mortality in the general population is J-curved [
19], and it is possible that a similar relationship exists for renal disease progression; although the nadir may vary in different patient populations.
Our study has two strengths: first, the data were obtained from a CKD patient population who received treatment based on current guidelines. Second, although it is an observational study, the data was prospectively collected and data entry was verified periodically with > 95 % of data being accurate.
The major study limitation was short follow-up period. Although there is a trend to more ESRD and faster GFR decline with higher sodium intake, it is possible that follow-up was not long enough to demonstrate statistically significant impact on ESRD and GFR decline. The second limitation is the observational nature of the study, thus causal relationships cannot be discerned, and residual confounding could impact the results. Although we calculated the average of urinary sodium excretion for each patient, the number of measurements per patient was low (mean 3.4 USE measurements/patient), and therefore we cannot account for changes in sodium intake that occurred before or after urinary sodium testing. Moreover, there are day-to-day variability in sodium intake and excretion [
20], which may also affect the accuracy of sodium intake estimation. The timing of initiating diuretics and increasing their dose may also impact sodium excretion. At our institution, diuretics are usually initiated (or dose changed) at the visit and urine collection is done on a subsequent visit thus minimizing the impact of change in diuretic dose on sodium excretion. In addition we could not address the issue of mortality with high sodium intake because of low number of deaths in our population. Finally, In our study RAS blockers were used more often in patients who were in the MSD and HSD group. Thus in our study, overweight and diabetic patients would have benefited less from RAS blockers.
Acknowledgements
We would like to acknowledge Nicole Page, Suzanne Jackson and Courtney Blondin for their help with collecting and verifying the clinical data used in this study.