Low protein diets for non‐diabetic adults with chronic kidney disease
Abstract
Background
Chronic kidney disease (CKD) is defined as reduced function of the kidneys present for 3 months or longer with adverse implications for health and survival. For several decades low protein diets have been proposed for participants with CKD with the aim of slowing the progression to end‐stage kidney disease (ESKD) and delaying the onset of renal replacement therapy. However the relative benefits and harms of dietary protein restriction for preventing progression of CKD have not been resolved. This is an update of a systematic review first published in 2000 and updated in 2006, 2009, and 2018.
Objectives
To determine the efficacy of low protein diets in preventing the natural progression of CKD towards ESKD and in delaying the need for commencing dialysis treatment in non‐diabetic adults.
Search methods
We searched the Cochrane Kidney and Transplant Register of Studies up to 7 September 2020 through contact with the Information Specialist using search terms relevant to this review. Studies in the Register are identified through searches of CENTRAL, MEDLINE, and EMBASE, conference proceedings, the International Clinical Trials Register (ICTRP) Search Portal and ClinicalTrials.gov.
Selection criteria
We included randomised controlled trials (RCTs) or quasi RCTs in which adults with non‐diabetic CKD (stages 3 to 5) not on dialysis were randomised to receive a very low protein intake (0.3 to 0.4 g/kg/day) compared with a low protein intake (0.5 to 0.6 g/kg/day) or a low protein intake compared with a normal protein intake (≥ 0.8 g/kg/day) for 12 months or more.
Data collection and analysis
Two authors independently selected studies and extracted data. For dichotomous outcomes (death, all causes), requirement for dialysis, adverse effects) the risk ratios (RR) with 95% confidence intervals (CI) were calculated and summary statistics estimated using the random effects model. Where continuous scales of measurement were used (glomerular filtration rate (GFR), weight), these data were analysed as the mean difference (MD) or standardised mean difference (SMD) if different scales had been used. The certainty of the evidence was assessed using GRADE.
Main results
We identified 17 studies with 2996 analysed participants (range 19 to 840). Four larger multicentre studies were subdivided according to interventions so that the review included 21 separate data sets. Mean duration of participant follow‐up ranged from 12 to 50 months.
Random sequence generation and allocation concealment were considered at low risk of bias in eleven and nine studies respectively. All studies were considered at high risk for performance bias as they were open‐label studies. We assessed detection bias for outcome assessment for GFR and ESKD separately. As GFR measurement was a laboratory outcome all studies were assessed at low risk of detection bias. For ESKD, nine studies were at low risk of detection bias as the need to commence dialysis was determined by personnel independent of the study investigators. Five studies were assessed at high risk of attrition bias with eleven studies at low risk. Ten studies were at high risk for reporting bias as they did not include data which could be included in a meta‐analysis. Eight studies reported funding from government bodies while the remainder did not report on funding.
Ten studies compared a low protein diet with a normal protein diet in participants with CKD categories 3a and b (9 studies) or 4 (one study). There was probably little or no difference in the numbers of participants who died (5 studies 1680 participants: RR 0.77, 95% CI 0.51 to 1.18; 13 fewer deaths per 1000; moderate certainty evidence). A low protein diet may make little or no difference in the number of participants who reached ESKD compared with a normal protein diet (6 studies, 1814 participants: RR 1.05, 95% CI 0.73 to 1.53; 7 more per 1000 reached ESKD; low certainty evidence). It remains uncertain whether a low protein diet compared with a normal protein intake impacts on the outcome of final or change in GFR (8 studies, 1680 participants: SMD ‐0.18, 95% CI ‐0.75 to 0.38; very low certainty evidence).
Eight studies compared a very low protein diet with a low protein diet and two studies compared a very low protein diet with a normal protein diet. A very low protein intake compared with a low protein intake probably made little or no difference to death (6 studies, 681 participants: RR 1.26, 95% CI 0.62 to 2.54; 10 more deaths per 1000; moderate certainty evidence). However it probably reduces the number who reach ESKD (10 studies, 1010 participants: RR 0.65, 95% CI 0.49 to 0.85; 165 per 1000 fewer reached ESKD; moderate certainty evidence). It remains uncertain whether a very low protein diet compared with a low or normal protein intake influences the final or change in GFR (6 studies, 456 participants: SMD 0.12, 95% CI ‐0.27 to 0.52; very low certainty evidence).
Final body weight was reported in only three studies. It is uncertain whether the intervention alters final body weight (3 studies, 89 participants: MD ‐0.40 kg, 95% CI ‐6.33 to 5.52; very low certainty evidence).Twelve studies reported no evidence of protein energy wasting (malnutrition) in their study participants while three studies reported small numbers of participants in each group with protein energy wasting. Most studies reported that adherence to diet was satisfactory. Quality of life was not formally assessed in any studies.
Authors' conclusions
This review found that very low protein diets probably reduce the number of people with CKD 4 or 5, who progress to ESKD. In contrast low protein diets may make little difference to the number of people who progress to ESKD. Low or very low protein diets probably do not influence death. However there are limited data on adverse effects such as weight differences and protein energy wasting. There are no data on whether quality of life is impacted by difficulties in adhering to protein restriction. Studies evaluating the adverse effects and the impact on quality of life of dietary protein restriction are required before these dietary approaches can be recommended for widespread use.
PICOs
Plain language summary
Low protein diets for non‐diabetic adults with chronic kidney disease
What is the issue?
Various forms of kidney disease can lead to kidney failure with affected people ultimately requiring dialysis treatment. A diet low in protein may be recommended to try to slow the progress of kidney disease to kidney failure. We still do not know whether low protein diets can slow the progress of kidney disease and delay the need to start dialysis.
What did we do?
We searched the Cochrane Kidney and Transplant Specialised Register up to 7 September 2020 for randomised controlled trials (RCT), which enrolled non‐diabetic adult patients with chronic kidney disease, not yet requiring dialysis, and which compared different dietary protein intakes, including very low (0.3 to 0.4 g/kg/day), low (0.5 to 0.6 g/kg/day) or normal protein intakes (≥ 0.8 g/kg/day) for 12 months or more.
What did we find?
We examined the evidence from 17 studies (21 data sets) with 2996 people with reduced kidney function. We found that very low protein diets compared with low or normal protein intakes probably reduce the number of people with advanced kidney failure, who progress to dialysis. When low protein diets were compared with normal protein diets, there was little of no difference in the number of people with less severe kidney failure, who progressed to dialysis. Side effects of low protein diets such as weight loss were uncommon but many studies did not report on side effects.
Conclusions
In people with advanced kidney failure, a very low protein intake probably slows the progress to kidney failure. However we need more information on the side effects of low protein diets and on whether quality of life is reduced because of difficulties in keeping to such a diet.
Authors' conclusions
Summary of findings
| Low protein diet versus normal protein diet for non‐diabetic adults with CKD | |||||
| Patient or population: non‐diabetic adults with CKD | |||||
| Outcomes | Anticipated absolute effects* (95% CI) | Relative effect | No. of participants | Certainty of the evidence | |
|---|---|---|---|---|---|
| Risk with normal protein diet | Risk with low protein diet | ||||
| Death (all causes) | 55 per 1,000 | 42 per 1,000 | RR 0.77 | 1680 (5) | ⊕⊕⊕⊝ |
| ESKD | 144 per 1,000 | 151 per 1,000 | RR 1.05 | 1814 (6) | ⊕⊕⊝⊝ |
| End or change in GFR | The SMD for end or change in GFR was 0.18 lower (0.75 lower to 0.38 higher) with low protein diet compared to normal protein diet | ‐ | 1680 (8) | ⊕⊝⊝⊝ | |
| *The risk in the intervention group (and its 95% CI) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). | |||||
| GRADE Working Group grades of evidence | |||||
| 1 The confidence intervals include potential for important benefits and harms 2 Important and unexplained heterogeneity present 3 The outcome reported is a surrogate outcome | |||||
| Very low protein diet versus low or normal protein diet for non‐diabetic adults with CKD | |||||
| Patient or population: non‐diabetic adults with CKD | |||||
| Outcomes | Anticipated absolute effects* (95% CI) | Relative effect | No. of participants | Certainty of the evidence | |
|---|---|---|---|---|---|
| Risk with low or normal protein diet | Risk with very low protein diet | ||||
| Death (all causes) | 39 per 1,000 | 49 per 1,000 | RR 1.26 | 681 (6) | ⊕⊕⊕⊝ |
| ESKD | 458 per 1,000 | 293 per 1,000 | RR 0.64 | 1010 (10) | ⊕⊕⊕⊝ |
| End or change in GFR | The SMD for end or change in GFR was 0.12 (0.27 lower to 0.52 higher) with very low protein diet compared to low or normal protein diet | ‐ | 456 (6) | ⊕⊕⊝⊝ | |
| *The risk in the intervention group (and its 95% CI) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). | |||||
| GRADE Working Group grades of evidence | |||||
| 1 The confidence intervals are wide and include potential for important benefits and harms 2 Serious unexplained heterogeneity 3 Outcome is a surrogate outcome 4 Unclear allocation concealment in 4 studies contributing information to analysis | |||||
| Nutritional measures for non‐diabetic adults with CKD | |||||
| Patient or population: non‐diabetic adults with CKD | |||||
| Outcomes | Anticipated absolute effects* (95% CI) | Relative effect | No. of participants | Certainty of the evidence | |
|---|---|---|---|---|---|
| Risk with normal or low protein diet | Risk with very low or low protein diet | ||||
| Final body weight: low protein versus normal protein diet | The mean final body weight 3.09 kg lower (5.02 to 1.16 lower) with low protein diet compared to normal protein diet | ‐ | 223 (2) | ⊕⊝⊝⊝ | |
| Final body weight: very low protein diet versus low protein diet | The mean final body weight 1.4 kg higher (3.40 lower to 6.21 higher) with very low protein diet compared to low protein diet | ‐ | 291 (4) | ⊕⊝⊝⊝ | |
| Protein energy wasting (malnutrition) | 4 per 1,000 | 6 per 1,000 | RR 1.31 | 2373 (15) | ⊕⊕⊝⊝ |
| *The risk in the intervention group (and its 95% CI) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). | |||||
| GRADE Working Group grades of evidence | |||||
| 1 Increased risk of bias related to incomplete outcome data and selective reporting 2 Small studies and wide confidence intervals and include potential for important benefits and harms 3 3/4 studies are unclear for allocation concealment and random sequence generation 4 Serious unexplained heterogeneity | |||||
Background
Description of the condition
Chronic kidney disease (CKD) is defined as abnormalities of the structure or function of the kidneys present for three months or more with adverse implications for health (KDIGO 2012). It is classified based on the cause, the severity of reduced kidney function as measured by the glomerular filtration rate (GFR) and the severity of albuminuria. KDIGO 2012 defined a GFR of less than 60 mL/min/1.73 m2 as indicating reduced kidney function (normal GFR in young healthy adults is approximately 125 mL/min/1.73 m2). CKD is associated with a range of complications leading to adverse health outcomes. Death (all causes) and cardiovascular death increase in individuals with GFR < 60 mL/min/1.73 m2 (Matshushita 2010). The rate of deterioration in kidney function is variable and depends on the underlying cause of CKD and is associated among other factors with elevated blood pressure, increasing levels of proteinuria and diabetes mellitus. In many people, though not all, with CKD, kidney function deteriorates progressively with people developing symptoms of uraemia. Eventually people require treatment with haemodialysis or peritoneal dialysis with some receiving kidney transplants.
Description of the intervention
The World Health Organisation (WHO) recommends that healthy adults should receive a daily protein intake of 0.8 g/kg/day. Most healthy adults in developed countries consume a diet with a protein intake exceeding 1 g/kg/day. In CKD with reduced GFR, nephrologists and dietitians have prescribed low (0.5 to 0.6 g/kg/day) or very low (0.3 to 0.4 g/kg/day) high biologic‐value protein diets aiming to reduce the rate at which GFR deteriorates and to alleviate some of the complications of advanced CKD including metabolic acidosis, bone disease and uraemic symptoms and thus delay the onset of end‐stage kidney disease (ESKD), which leads to significant reduction in quality of life. To achieve very low protein intakes, some centres prescribe vegetarian diets (Garneata 2013). Very low protein diets are frequently supplemented with essential amino acids and nitrogen free keto‐analogues of amino acids to reduce the risk of malnutrition. If sufficient calories are ingested, keto‐analogues can be converted to amino acids via urea recycling. Extensive nutritional counselling is required to ensure that participants understand how to maintain a low or very low protein diet with an adequate calorie intake (30 to 35 Kcal/kg ideal body weight/day). Compliance with a reduced protein diet is frequently assessed with measurement of the urinary urea nitrogen in 24‐hour urine collections and calculation of protein intake using the Maroni formula (6.25 X [urinary urea nitrogen x 0.03 body weight in kg]) (Maroni 1985).
How the intervention might work
Experimental studies in rats have shown that loss of nephrons leads to increased glomerular filtration in the remaining nephrons. The compensatory hyperfiltration results from increased plasma flow rates and increasing hydraulic pressure in the remaining nephrons. Eventually these haemodynamic changes lead to increased glomerular permeability with proteinuria and the development of progressive glomerulosclerosis. Long‐term studies in rats have demonstrated that compared with rats with CKD on a high protein diet, rats with CKD on a low protein diet had fewer sclerotic glomeruli and less proteinuria (Hostetter 1986). These experimental data supported the view that protein restriction in people with CKD could protect glomeruli from progressive glomerulosclerosis, slow the deterioration in kidney function and delay the onset of ESKD. In addition, protein restriction reduces uraemic symptoms associated with metabolic acidosis, CKD‐metabolic bone disease, hypertension, and fluid overload which could also delay the onset of ESKD even if the rate of kidney function deterioration measured by GFR does not change (Kasiske 1998).
Why it is important to do this review
There remains considerable controversy as to whether protein restriction does slow the rate of deterioration in kidney function in people with non‐diabetic CKD with proponents providing data to support or refute the benefit of protein restriction (Johnson 2006; Mandayam 2006). KDIGO 2012 concluded that dietary protein intake < 0.8 g/kg/day did not offer any advantage over 0.8 g/kg/day and suggested that protein restriction to 0.8 g/kg/day be limited to adults with GFR < 30 mL/min/1.73 m2. KDIGO 2012 also advised that people on any dietary protein restriction required careful monitoring of clinical and biochemical markers to avoid nutritional deficiencies.
Most of the clinical studies (both randomised controlled trials (RCTs) and observational studies) were designed to test the efficacy of reducing protein intake on surrogate kidney function outcomes, such as decline in creatinine clearance (CrCl) or changes in the reciprocal of creatinine over time. Unfortunately, changing protein intake modifies creatinine markers because reducing protein intake decreases creatinine production and changes kidney function (glomerular filtration as well as CrCl) by unidentified mechanisms. Although a few studies used methods to measure GFR using non‐creatinine measures such as 51CrEDTA clearance and I‐125 Iothalamate clearance, the results from these studies have been conflicting.
Two non‐Cochrane systematic reviews (Kasiske 1998; Pedrini 1996) have evaluated the efficacy of reduced protein diets. Pedrini 1996 reported that a low‐protein diet significantly reduced the risk of kidney failure or death. In contrast Kasiske 1998 found that dietary protein restriction reduced the rate of decline in estimated GFR by only 0.53 mL/min/year. Because the decision to commence dialysis is not based only on declining GFR but also on the presence of uraemic symptoms and nephrologists vary in their criteria for commencing dialysis, it is quite possible that those on a higher protein intake will have more uraemic symptoms and be considered for dialysis earlier than those on lower protein intakes with an equivalent rate of GFR decline but fewer uraemic symptoms. In this publication we update a Cochrane systematic review first published in 2000 (Fouque 2000b) and updated in 2006 (Fouque 2006) and 2009 (Fouque 2009). The 2009 update (Fouque 2009) reported the composite outcome of death and ESKD (dialysis initiation or renal transplantation) as the primary outcome. Overall fewer events (deaths, ESKD) were observed with very low or low protein intake compared with those occurring with low or normal protein intake suggesting that a reduced protein intake reduces the number of people who die or reach ESKD. We aimed to determine whether the addition of further RCTs would further clarify whether low or very low protein diets benefit adults with non‐diabetic CKD by delaying the onset of ESKD and/or slowing the rate of GFR decline without adverse effects on nutritional status.
Objectives
To determine the efficacy of low protein diets in preventing the natural progression of CKD towards ESKD and in delaying the need for commencing dialysis treatment in non‐diabetic adults.
Methods
Criteria for considering studies for this review
Types of studies
We included all randomised controlled trials (RCTs) and quasi‐RCTs (RCTs in which allocation to treatment was obtained by alternation, use of alternate medical records, date of birth or other predictable methods) in which those in the experimental group received a reduced protein intake for 12 months or more while those in the control group received a higher or normal protein intake. Cross‐over studies were to be considered if the starting period of the intervention was randomly allocated and each intervention was in place for at least 12 months.
Types of participants
-
Adults suffering from moderate to severe CKD, as estimated by either serum creatinine (SCr), CrCl or GFR measurement but excluding participants on peritoneal dialysis, haemodialysis or following a kidney transplant.
-
Because of the difficulty to control for confounding factors, studies of diabetic participants or children with CKD were excluded from the review though studies including small numbers of diabetic participants were included.
Types of interventions
-
Studies comparing a normal protein intake (≥ 0.8 g/kg/day) with a low protein intake (0.5 to 0.6 g/kg/day) or very low protein intake (0.3 to 0.4 g/kg/day) for 12 months or more
-
Studies comparing a low protein intake (0.5 to 0.6 g/kg/day) with a very low protein intake (0.3 to 0.4 g/kg/day) for 12 months or more
-
Studies in which participants received supplements of essential amino acids, keto‐analogues or both were included provided that the total nitrogen intake differed between treatment groups.
Types of outcome measures
Primary outcomes
-
Death (all causes)
-
ESKD as defined by the need to commence dialysis during follow up or to receive a kidney transplant during follow‐up.
Secondary outcomes
-
End of study or change in GFR
-
End of study body weight
-
End of study body mass index (BMI)
-
Development of protein energy wasting (malnutrition) as defined by the study authors
-
Quality of life.
Search methods for identification of studies
Electronic searches
We searched the Cochrane Kidney and Transplant Specialised Register up to 7 September 2020 through contact with the Information Specialist using search terms relevant to this review. The Cochrane Kidney and Transplant Specialised Register contains studies identified from several sources.
-
Monthly searches of the Cochrane Central Register of Controlled Trials (CENTRAL)
-
Weekly searches of MEDLINE OVID SP
-
Handsearching of kidney‐related journals and the proceedings of major kidney conferences
-
Searching of the current year of EMBASE OVID SP
-
Weekly current awareness alerts for selected kidney and transplant journals
-
Searches of the International Clinical Trials Register (ICTRP) Search Portal and ClinicalTrials.gov.
Studies contained in the Specialised Register are identified through search strategies for CENTRAL, MEDLINE, and EMBASE based on the scope of Cochrane Kidney and Transplant. Details of these strategies as well as a list of handsearched journals, conference proceedings and current awareness alerts are available in the Specialised Register section of information about Cochrane Kidney and Transplant.
See Appendix 1 for search terms used in strategies for this review.
Searching other resources
-
Reference lists of review articles, relevant studies and clinical practice guidelines.
-
Letters seeking information about unpublished or incomplete trials to investigators known to be involved in previous studies.
Data collection and analysis
Selection of studies
The initial review and updates to 2009 was undertaken by four authors using the search strategy described. The titles and abstracts were screened by two authors, based on the defined inclusion criteria. They discarded studies that were not relevant (i.e. studies of lipid lowering agents) although studies and reviews that could have included relevant data or information on studies was retained initially. Disagreements were resolved by discussion.
The 2018 update was undertaken by three authors (DH, EH, DF). Potentially relevant studies were initially determined by two authors from titles and abstracts. Full text articles of potentially eligible articles were reviewed for eligibility by two authors to determine which studies satisfied the inclusion criteria.
Data extraction and management
Data extraction and assessment of risk of bias were performed independently by two authors using standardised data extraction forms. Studies in languages other than English were translated before data extraction. Where more than one report of a study was identified, data were extracted from the most complete report but the remaining reports were checked for additional information. Where there were discrepancies between reports, data from the primary source were used. Any further information required from the original authors was requested by written correspondence and any relevant information obtained in this manner was included in the review. Any disagreements were resolved in consultation with the third author.
Assessment of risk of bias in included studies
The following items were assessed independently by two authors using the risk of bias assessment tool (Higgins 2011) (Appendix 2).
-
Was there adequate sequence generation (selection bias)?
-
Was allocation adequately concealed (selection bias)?
-
Was knowledge of the allocated interventions adequately prevented during the study?
-
Participants and personnel (performance bias)
-
Outcome assessors (detection bias)
-
-
Were incomplete outcome data adequately addressed (attrition bias)?
-
Are reports of the study free of suggestion of selective outcome reporting (reporting bias)?
-
Was the study apparently free of other problems that could put it at a risk of bias?
Measures of treatment effect
For dichotomous outcomes (death (all causes), requirement for dialysis, adverse effects) the risk ratios (RR) with 95% confidence intervals (CI) for individual studies were calculated and summary statistics estimated using the random effects model. Where continuous scales of measurement were used to assess the effects of treatment (GFR, weight, BMI), these data were analysed as the mean difference (MD) or standardised mean difference (SMD) if different scales had been used. Either final levels or change in levels were included in meta‐analyses of continuous scales of measurement. When both measures were provided in a study, final levels were included. Where standard deviations (SD) for changes in levels or final levels were missing and not available from triallists, these were imputed (Higgins 2011).
Unit of analysis issues
Data from cross‐over studies were to be included in the meta‐analyses if separate data for the first part of the study were available. No cross‐over studies were identified.
Dealing with missing data
We aimed to analyse available data in meta‐analyses using intention‐to‐treat (ITT) data. However, where ITT data were not provided, or additional information could not be obtained from authors, available published data were used in the analyses.
Assessment of heterogeneity
We first assessed the heterogeneity by visual inspection of the forest plot. Heterogeneity was then analysed using a Chi2 test on N‐1 degrees of freedom, with an alpha of 0.05 used for statistical significance and with the I2 test (Higgins 2003). A guide to the interpretation of I2 values was as follows.
-
0% to 40%: might not be important
-
30% to 60%: may represent moderate heterogeneity
-
50% to 90%: may represent substantial heterogeneity
-
75% to 100%: considerable heterogeneity.
The importance of the observed value of I2 depends on the magnitude and direction of treatment effects and the strength of evidence for heterogeneity (e.g. P‐value from the Chi2 test, or a confidence interval for I2) (Higgins 2011).
Assessment of reporting biases
The search strategy used aimed to reduce publication bias related to failure to publish negative results. Where there were multiple publications from the same study, the primary publications and additional reports were reviewed to identify all outcomes to reduce the risk of selective outcome reporting bias.
Data synthesis
Data were combined using random‐effects model for dichotomous and continuous data.
Subgroup analysis and investigation of heterogeneity
Subgroup analyses were planned to assess differences in results possibly related to population groups, different ways of measuring decline in GFR and to risk of bias assessment but there were too few studies in each analysis to allow meaningful subgroup analyses.
Sensitivity analysis
Where a single study differed considerably from the other studies in the meta‐analysis, this study was temporarily excluded to determine whether its removal altered the results of the meta‐analysis.
'Summary of findings' tables
We presented the main results of the review in 'Summary of findings' tables. These tables present key information concerning the quality of the evidence, the magnitude of the effects of the interventions examined, and the sum of the available data for the main outcomes (Schünemann 2011a). The 'Summary of findings' tables also include an overall grading of the evidence related to each of the main outcomes using the GRADE (Grades of Recommendation, Assessment, Development and Evaluation) approach (GRADE 2008). The GRADE approach defines the quality of a body of evidence as the extent to which one can be confident that an estimate of effect or association is close to the true quantity of specific interest. The quality of a body of evidence involves consideration of within‐trial risk of bias (methodological quality), directness of evidence, heterogeneity, precision of effect estimates and risk of publication bias (Schünemann 2011b). We presented the following outcomes in the 'Summary of findings' tables.
-
Death in low protein versus normal protein diet groups
-
ESKD in low protein versus normal protein diet groups
-
End or change in GFR low protein versus normal protein diet
-
Death in very low protein versus higher protein diet groups
-
ESKD in very low versus low protein diet groups
-
End or change in GFR in very low protein diet versus higher protein diet
-
Adverse effects ‐ weight loss in very low protein diet versus higher protein diets.
Results
Description of studies
Results of the search
The first version of this review included five studies (six reports; 1494 participants) (Fouque 2000b). Subsequent updates of the review published in 2006 and 2009 included eight studies (18 reports; 1524 participants) (Fouque 2006) and 10 studies (30 reports; 2000 participants) (Fouque 2009) respectively.
For the 2018 update search (to 2 March 2018) we identified seven new studies with 24 reports (ESGCMCRF 1 1990; ESGCMCRF 2 1990; Bergstrom 1986; Garneata 2013; MDRD Feasibility A 1989; MDRD Feasibility B 1989; Meloni 2004; Milovanov 2009; Teplan 1998) and 109 additional reports of eight previously included studies. We excluded 34 reports.
In a further search to 7 September, 2020, we did not identify any new studies for inclusion but did find four new reports of three included studies (Garneata 2013; MDRD 1 1989; MDRD 2 1989), four new excluded studies (Di Iorio 2018; Kopple 1968; Milovanova 2018; Saxena 2017) with six reports, and one additional report of an already excluded study (Teplan 2006). Therefore 17 studies (21 data sets; 163 reports; 3100 randomised participants, 2996 analysed participants) were included in the update of this review (Figure 1).
Study flow diagram
One ongoing study was identified and will be assessed in a future update (NCT01418508). However the estimated completion date was December 2014 and no publication of the results has been identified.
Included studies
Two previously included studies (MDRD 1 1989; MDRD 2 1989; Rosman 1 1984; Rosman 2 1984) and two newly identified studies (ESGCMCRF 1 1990; ESGCMCRF 2 1990; MDRD Feasibility A 1989; MDRD Feasibility B 1989) were divided into two studies each because they included two groups of participants with different mean GFRs, who received different protein intakes (low protein intake versus normal protein intake or very low protein intake versus low protein intake). Teplan 1998 (published only as an abstract) was a three‐arm study comparing very low (0.4 g/kd/day), low (0.6 g/kg/day) and a restricted diet (0.8 to 1.0 g/kg/day); data could not be extracted and was not included in any of the meta‐analyses. Therefore in this systematic review we considered there to be 21 separate studies.
Nine studies (Bergstrom 1986; ESGCMCRF 1 1990; Locatelli 1989; MDRD 1 1989; MDRD Feasibility A 1989; Meloni 2004; Milovanov 2009; Rosman 1 1984; Williams 1991), in which most participants had CKD category 3a or 3b (KDIGO 2012), compared a prescribed low protein diet (0.5 to 0.6 g/kg/day) with a normal protein diet (≥ 1 g/kg/day). A tenth study (Cianciaruso 2008a), which included participants with CKD category 4 and randomised participants to a low protein diet (0.55 g/kg/day) or to the WHO recommended normal protein intake (0.8 g/kg/day), was included in the meta‐analyses with studies comparing low with normal protein intake. Protein intake was estimated from urinary urea nitrogen measurements (Maroni 1985). The mean calculated protein intake was 0.68 g/kg/day (range 0.49 to 0.85 g/kg/day) for the low protein intervention and 1.0 g/kg/day (range 0.61 to 1.54 g/kg/day) for the normal or free protein diet group. No data on calculated protein intake were available for Rosman 1 1984 since data on urea excretion were only provided graphically.
Eight studies (Di Iorio 2003; ESGCMCRF 2 1990; Garneata 2013; Chauveau 1986; Malvy 1999; MDRD 2 1989; MDRD Feasibility B 1989; Mircescu 2007), in which participants had CKD stage 4/5 compared prescribed very low protein diets (0.3 to 0.4 g/kg/day with keto‐analogues) with low protein diets (0.58 to 0.65 g/kg/day). The mean calculated protein intake for the participants who received a very low protein diet was 0.4 g/kg/day (range 0.29 to 0.5 g/kg/day) and it was 0.64 g/kg/day (range 0.56 to 0.79 g/kg/day) for participants receiving the low protein diet. No data on calculated protein intake were available for Chauveau 1986 since data on urea excretion were not provided. Ihle 1989 and Rosman 2 1984, in which participants had CKD category 4 and which compared very low protein diets (0.4 g/kg/day) with normal protein diets, were included with the eight studies comparing very low with low protein intakes. Actual protein intake could not be calculated for these two studies.
Seven studies were multicentre studies (ESGCMCRF 1 1990; ESGCMCRF 2 1990; Locatelli 1989; MDRD 1 1989; MDRD 2 1989; MDRD Feasibility A 1989; MDRD Feasibility B 1989), two studies involved two sites (Malvy 1999; Williams 1991) and the remainder were single centre studies. Participant numbers ranged from 19 to 840 with an age range of 15 to 75 years. Glomerulopathies accounted for CKD in between 23% (Williams 1991) and 60% (Meloni 2004) of participants; the types of kidney disease included were not reported in Bergstrom 1986, while Milovanov 2009 included only participants with lupus nephritis or other vasculitides. Six diabetic nephropathy participants (three in each group) were included in Di Iorio 2003. ESGCMCRF 1 1990 and ESGCMCRF 2 1990 reported that fewer than 10% of diabetic participants were included among the 554 participants assessed for inclusion in the study though it was unclear whether any diabetic participants were included in the 336 randomised participants. No diabetic participants were included in the remaining studies.
We chose to include ESGCMCRF 1 1990 and ESGCMCRF 2 1990 although there were known to be some participants from one centre, who were included in these studies and in Locatelli 1989. We were unable to obtain an exact number of participants included in both studies. However it appeared that only about 30 participants were included in both studies, which would be only 8.9% of 336 randomised participants in ESGCMCRF 1 1990 and ESGCMCRF 2 1990 and 6.6% of 456 randomised participants in Locatelli 1989 (personal communication from Professor Norbert Gretz).
Mean duration of follow up ranged from 12 to 50 months.
Excluded studies
In the 2009 update, there were 53 excluded studies (67 reports). In the 2018 update we identified a total of 71 excluded studies (101 reports). Based on the Cochrane recommendations for dealing with excluded studies, we limited the excluded studies to randomised controlled trials (RCT) and removed all non‐randomised studies from excluded studies. Therefore in the 2018 update we excluded 29 studies. Of these 13 studies investigated ineligible interventions for this review, two studies included an ineligible population such as dialysis participants or participants with diabetes mellitus and in 14 studies the duration of follow‐up was less than one year. In 2020 we identified four new excluded studies (five reports) and an additional report of one already excluded study. Two new studies were excluded for ineligible interventions and two for a duration of follow‐up of less than one year.
Risk of bias in included studies
Risk of bias graph: review authors' judgements about each risk of bias item presented as percentages across all included studies.
Risk of bias summary: review authors' judgements about each risk of bias item for each included study.
Allocation
Random sequence generation was considered at low risk of bias in 11 studies (Cianciaruso 2008a; ESGCMCRF 1 1990; ESGCMCRF 2 1990; Garneata 2013; Locatelli 1989; MDRD 1 1989; MDRD 2 1989; MDRD Feasibility A 1989; MDRD Feasibility B 1989; Meloni 2004; Williams 1991) and unclear in the remaining studies.
Allocation concealment was considered to be at low risk of bias in nine studies (Cianciaruso 2008a; ESGCMCRF 1 1990; ESGCMCRF 2 1990; Garneata 2013; Locatelli 1989; MDRD 1 1989; MDRD 2 1989; MDRD Feasibility A 1989; MDRD Feasibility B 1989) and assessed as unclear in the remaining studies with insufficient information available to permit judgement.
Blinding
All 20 studies were open‐label studies and so were considered at high risk for performance bias.
Detection bias (outcome assessment) was recorded separately for GFR and for the need to commence dialysis. All studies which reported this outcome were assessed to be at low risk for detection bias for GFR measurement as the outcomes were laboratory‐based and unlikely to be influenced by lack of blinding. Nine studies were assessed as at low risk of bias for need to commence dialysis as they provided information to indicate that the onset of ESKD with the start of dialysis treatment was assessed independently of the study investigators (Chauveau 1986; Cianciaruso 2008a; Garneata 2013; Malvy 1999; MDRD Feasibility A 1989; MDRD Feasibility B 1989; Mircescu 2007; Rosman 1 1984; Rosman 2 1984). One study was at high risk (Ihle 1989) and the remaining studies did not provide any information on how the onset of ESKD was assessed.
Incomplete outcome data
Five studies were considered at high risk of attrition bias as more than 10% of participants were lost to follow up or excluded from the analysis (Bergstrom 1986; Chauveau 1986; Ihle 1989; Locatelli 1989; Williams 1991). Eleven studies (Cianciaruso 2008a; Di Iorio 2003; Garneata 2013; MDRD Feasibility A 1989; MDRD Feasibility B 1989; MDRD 1 1989; MDRD 2 1989; Meloni 2004; Mircescu 2007; Rosman 1 1984; Rosman 2 1984) were assessed as at low risk of detection bias and in the remaining four studies, it was unclear how many participants were lost to follow‐up.
Selective reporting
Studies were considered to be at high risk of bias if data were provided in a format, which could not be entered into the meta‐analyses or if the study did not provide data on death, requirement for dialysis or the nutritional status of the participants. We assessed 11 studies at high risk of selective reporting (ESGCMCRF 1 1990; ESGCMCRF 2 1990; Ihle 1989; Locatelli 1989; Malvy 1999; MDRD 1 1989; MDRD 2 1989; MDRD Feasibility A 1989; MDRD Feasibility B 1989; Meloni 2004; Teplan 1998). Eight studies were assessed as low risk of selective reporting (Cianciaruso 2008a; Chauveau 1986; Di Iorio 2003; Garneata 2013; Mircescu 2007; Rosman 1 1984; Rosman 2 1984; Williams 1991) and two studies were judged as unclear.
Other potential sources of bias
We assessed eight studies (Cianciaruso 2008a; ESGCMCRF 1 1990; ESGCMCRF 2 1990; Garneata 2013; MDRD 1 1989; MDRD 2 1989; MDRD Feasibility A 1989; MDRD Feasibility B 1989) to be at low risk of potential bias as they were funded by educational or philanthropic organisations and the remaining fourteen studies were considered as unclear as there was insufficient information to permit judgement regarding funding sources.
Effects of interventions
See: Summary of findings 1 Low protein diet versus normal protein diet for non‐diabetic adults with chronic kidney disease (CKD); Summary of findings 2 Very low protein diet versus low or normal protein diet for non‐diabetic adults with chronic kidney disease (CKD); Summary of findings 3 Nutritional measures for non‐diabetic adults with chronic kidney disease (CKD)
Low protein diets versus normal or free protein diet
Eleven studies compared low protein diets (0.55 to 0.6 g/kg/day) with a normal or free protein diet (0.8 to 1 g/kg/day). Nine studies enrolled participants with category 3a and 3b CKD while one study (Cianciaruso 2008a) enrolled participants with CKD category 4 and 5.
Death (all causes)
Five of 10 studies reported data on death (all causes), which could be included in a meta‐analysis. The certainty of the evidence was considered as moderate (summary of findings Table 1) because of imprecision. Thus a low protein intake probably leads to little or no difference in death between participants, who received a lower protein diet, and those receiving a normal or free protein diet (Analysis 1.1 (5 studies, 1680 participants): RR 0.77, 95% CI 0.51 to 1.18; I2 = 0%). There were 13 fewer deaths per 1000 in the low protein group (27 fewer to 10 more deaths). Five studies did not report this outcome.
End‐stage kidney disease
Six of 10 studies reported data on this outcome, which could be included in a meta‐analysis. The certainty of the evidence was considered as low (summary of findings Table 1) because of imprecision and moderate heterogeneity. A low protein diet may make little or no difference in the number of participants who reached ESKD compared with a normal protein diet (Analysis 1.2 (6 studies, 1814 participants): RR 1.05, 95% CI 0.73 to 1.53; I2 = 62%). There were 7 participants more per 1000 reaching ESKD in the low protein group (39 fewer to 76 more) compared with the normal protein group. Removal of Cianciaruso 2008a, which included participants with CKD category 4, did not influence the heterogeneity. Exclusion of ESGCMCRF 1 1990 reduced this heterogeneity (I2 = 26%). ESGCMCRF 1 1990 did not report information which allowed us to determine whether there was detection bias or selective reporting in this study so it is possible that increased risk of bias in these domains in this study contributed to the heterogeneity. Four studies (Bergstrom 1986, MDRD Feasibility A 1989; Meloni 2004; Milovanov 2009) did not report on this outcome.
End or change in GFR
Eight of 10 studies reported data on this outcome, which could be included in a meta‐analysis. Studies used different methods to express the final GFR or the change in GFR. Because of imprecision, use of a surrogate outcome and substantial heterogeneity (Summary of findings table 1), the certainty of the evidence was considered to be very low. It therefore remains uncertain whether a low protein diet influences the final or change in GFR compared with a normal protein diet (Analysis 1.3; 8 studies, 1680 participants): SMD ‐0.18, 95% CI ‐0.75 to 0.38; I2 = 96%). Heterogeneity was reduced by exclusion of the studies by Locatelli 1989 and Meloni 2004 but it was not clear why these studies should provide data that differed from other studies. ESGCMCRF 1 1990 did not report this outcome while Rosman 1 1984 only reported the data as medians, which could not be included in the meta‐analysis.
Very low protein diet versus low protein diet
Eight studies compared a very low protein diet (0.3 to 0.4 g/kg/day) with a low protein diet (0.58 to 0.65 g/kg/day) while two studies compared a very low protein diet with a normal protein diet. All studies enrolled participants with CKD category 4.
Death (all causes)
Six of 10 studies reported data for this outcome, which could be included in a meta‐analysis. Because of imprecision in the results, the certainty of the evidence was considered to be moderate (summary of findings Table 2). Thus a very low protein intake probably leads to little or no difference in death between participants, who received a lower protein diet, and those receiving a normal or free protein diet (Analysis 2.1 (6 studies, 681 participants): RR 1.26, 95% CI 0.62 to 2.54; I2 = 0%). There were 10 more deaths per 1000 in the very low protein group (15 fewer to 60 more) compared with the normal or low protein group. Four studies (Di Iorio 2003; ESGCMCRF 2 1990; Ihle 1989; MDRD Feasibility B 1989) did not report data for this outcome.
End‐stage kidney disease
All ten studies reported data on this outcome, which could be included in a meta‐analysis. Because of moderate heterogeneity, the certainty of the evidence was considered to be moderate (summary of findings Table 2). Thus a very low protein diet probably reduces the number of participants, who reach ESKD, compared with a low or normal protein intake (Analysis 2.2 (10 studies, 1010 participants): RR 0.64, 95% CI 0.49 to 0.85; I2 = 56%). There were 165 fewer participants per 1000, who reached ESKD with a very low protein diet compared with a low or normal protein diet (69 to 233 fewer). Exclusion of MDRD Feasibility B 1989 reduced heterogeneity slightly though it remains unclear why this study's results differed from those of the other studies. Exclusion of Ihle 1989 and Rosman 2 1984, which compared a very low protein intake with a normal protein intake, did not influence the degree of heterogeneity.
End or change in GFR
Six of 10 studies reported data on this outcome, which could be included in a meta‐analysis. Because of substantial heterogeneity, use of a surrogate outcome and a high risk of bias in some included studies, the certainty of the evidence was considered to be very low (summary of findings Table 2). It therefore remains uncertain whether a low protein diet influences the final or change in GFR compared with a normal protein diet (Analysis 2.3 (6 studies, 456 participants): SMD 0.12, 95% CI ‐0.27 to 0.52; I2 = 68%). Heterogeneity was reduced with the exclusion of MDRD 2 1989 and MDRD Feasibility B 1989. Both of these studies were at low risk of bias for selection bias unlike the other studies included in the analysis but the reason for the heterogeneity remains unclear. Garneata 2013 reported the data on GFR as medians with 95% confidence intervals and these data could not be included in the meta‐analysis. The study found a significantly higher GFR in the very low protein group compared with the low protein group after 15 months (P < 0.01). Three studies did not report the outcome (ESGCMCRF 2 1990; Malvy 1999; Rosman 2 1984).
Other outcomes
Most studies did not discuss adverse events. However all the studies reported that body weight, BMI and mid arm circumferences were measured though most studies did not provide numerical data that could be included in meta‐analyses.
Most studies reported on dietary adherence and measured this at one to three monthly intervals using urine nitrogen excretion to calculate protein intake and/or by dietary recall or interviews, facilitated by dietitians. The differences between prescribed protein intakes and actual protein intakes are shown in Table 1 and Table 2. While most studies reported that adherence to diet was satisfactory, studies of participants with CKD 3a and 3b tended to have larger differences between actual protein intakes and prescribed protein intakes (Table 1). ESGCMCRF 1 1990 and ESGCMCRF 2 1990 reported large SD for actual protein intakes because of the wide variation among participants in adherence to diet. Two studies (Chauveau 1986; Locatelli 1989) specifically reported difficulty in maintaining dietary adherence with the low protein diet. Garneata 2013 excluded people who were not able to agree to adhere to protein restriction. No study formally assessed quality of life.
| Study | Difference in prescribed protein intake | Difference in actual protein intake | Difference between prescribed and actual protein intake |
|---|---|---|---|
| 0.4 g/kg/day | 0.12 g/kg/day | 0.28 g/kg/day | |
| 0.45 g/kg/day | 0.21 g/kg/day | 0.24 g/kg/day | |
| 0.25 g/kg/day | 0.17 g/kg/day | 0.08 g/kg/day | |
| 0.2 g/kg/day | 0.2 g/kg/day | 0 g/kg/day | |
| 0.4 g/kg/day | 0.18 g/kg/day | 0.22 g/kg/day | |
| 0.625 g/kg/day | 0.19 g/kg/day | 0.435 g/kg/day | |
| 0.72 g/kg/day | 0.4 g/kg/day | 0.32 g/kg/day | |
| 0.4 g/kg/day | 0.87 g/kg/day | 0.47 g/kg/day | |
| 0.4 g/kg/day | 0.45 g/kg/day | 0.05 g/kg/day |
| Study | Difference in prescribed protein intake | Difference in actual protein intake |
|---|---|---|
| 0.3 g/kg/day | 0.21 g/kg/day | |
| 0.3 g/kg/day | 0.29 g/kg/day | |
| 0.3 g/kg/day | 0.29 g/kg/day | |
| 0.3 g/kg/day | 0.21 g/kg/day | |
| 0.295 g/kg/day | 0.22 g/kg/day | |
| 0.3 g/kg/day | 0.3 g/kg/day | |
| 0.3 g/kg/day | 0.27 g/kg/day |
Body weight
Seven studies in total reported data on end of study body weight. The data were subgrouped according to protein intakes. Because of small numbers, imprecision and a high risk of bias, the certainty of the evidence was considered to be very low (summary of findings Table 3).
In two studies which compared low protein diets with normal protein diets, the certainty of the evidence was considered very low because of increased risk of bias and imprecision, and therefore it is uncertain whether a low protein diet reduces the final body weight (Analysis 3.1.1 (2 studies, 223 participants): MD ‐3.09 kg, 95% CI ‐5.02 to ‐1.16; I2 = 0%). Cianciaruso 2008a reported final body weights as a percentage of the baseline; actual weights could not be calculated as baseline weights were provided separately for men and women. At 12 months, the final body weights were 99.8% or more of baseline weights.
In four studies which compared very low protein diets with low protein diets, the certainty of the evidence was considered very low because of increased risk of bias and imprecision, and therefore it is uncertain whether a low protein diet reduces the final body weight (Analysis 3.1.2 (4 studies, 291 participants): MD 1.4 kg, 95% CI 3.40 to 6.21; I2 = 56%).
Thus it is uncertain whether the intervention alters final body weight. The data for MDRD 1 1989; MDRD 2 1989 are shown in Table 3 as data were reported separately for men and women so could not be added to the meta‐analysis. Three studies reported that body weight dropped during the first few months of commencing a low protein diet but that subsequently weight stabilised (Ihle 1989; Malvy 1999; Meloni 2004).
| MDRD Study 1 | MDRD Study 2 | |||||||
|---|---|---|---|---|---|---|---|---|
| Usual protein diet | Low protein diet | Low protein diet | Very low protein diet | |||||
| Final body weight (kg) | N | Final body weight (kg) | N | Final body weight (kg) | N | Final body weight (kg) | N | |
| Men | 88.5 ± 14.6 | 179 to 183 | 83.2 ± 12.8 | 165 to 170 | 79.6 ± 11.5 | 74 to 77 | 79.3 ± 10.9 | 69 to 71 |
| Women | 72.2 ± 14.9 | 98 to 105 | 69.3 ± 13.7 | 107 to 115 | 65.9 ± 11.9 | 49 to 51 | 65.0 ± 14.3 | 49 to 52 |
Protein energy wasting (malnutrition)
Fifteen studies made reference to protein energy wasting. Of these 12 studies reported no evidence of malnutrition while three studies reported small numbers of participants with protein energy wasting in both groups (Analysis 3.2 (15 studies, 2373 participants): RR 1.31, 95% CI 0.42 to 4.13; I2 = 0%; low certainty evidence).
Body mass index
Four studies reported on this outcome. Two studies (ESGCMCRF 1 1990; Meloni 2004) comparing a low protein with a normal protein diet, found no difference in BMI between groups. Two studies (ESGCMCRF 2 1990; Garneata 2013) comparing very low protein with a low protein diet, also demonstrated no difference between diet groups. Studies were not combined in a meta‐analysis as three studies (ESGCMCRF 1 1990; ESGCMCRF 2 1990; Garneata 2013) provided the data as medians and ranges.
Discussion
Summary of main results
For this update we identified six additional studies to provide a total of 17 studies (21 data sets) with 3100 participants (2996 analysed participants) so we were able to report separately on the outcomes of death (all causes), numbers with ESKD, and final or change in GFR. We could also report data separately for studies which compared low with normal protein intakes and those which compared very low with low protein intakes. We sub‐divided four studies which compared different protein intakes in participants with different stages of CKD so that 20 studies were included in this review.
Ten studies, mainly evaluating participants with CKD categories 3a and 3b, compared a low prescribed protein diet (0.55 to 0.6 g/kg/day) with a normal protein diet (0.8 to ≥ 1.0 g/kg/day). The difference in calculated protein intake was 0.32 g/kg/day between the intervention groups. A low protein diet compared with a normal protein diet probably makes little or no difference in the numbers of participants who died (moderate certainty evidence) and may make little or no difference in the number of participants who progressed to ESKD (low certainty evidence). It remains uncertain whether a low protein diet compared with a normal protein intake impacts on the final or change in GFR (very low certainty evidence) (summary of findings Table 1).
Ten studies, which evaluated participants with CKD 4 or 5, compared a prescribed very low protein diet (0.3 to 0.4 g/kg/day) with a low protein diet (0.58 to 0.65 g/kg/day) (eight studies) or with a normal protein diet (two studies). The difference in calculated protein intake was 0.25 g/kg/day between the intervention groups. A very low protein intake compared with a low protein intake probably makes little or no difference to death but it probably reduces the number of participants, who reach ESKD (moderate certainty evidence). It remains uncertain whether a very low protein diet compared with a low or normal protein intake influences the final or change in GFR (very low certainty evidence) (summary of findings Table 2).
Fifteen studies reported on the numbers of participants with protein energy wasting (malnutrition); 12 studies had no participants with protein energy wasting while three studies reported small numbers in both treatment groups. Only eight studies provided numerical data for body weight although most studies reported that weight was measured. Only three of the 15 studies reported any evidence of protein energy wasting. No study formally assessed quality of life in the participants. Most studies reported that adherence to diet was satisfactory though studies of participants with CKD 3a and 3b tended to have smaller differences between actual protein intakes, as measured by urinary nitrogen excretion, and prescribed protein intakes.
Overall completeness and applicability of evidence
For this review we identified 16 studies (reported as 20 studies) which evaluated the efficacy and safety of low protein diets in non‐diabetic CKD. Several studies, particularly the older and smaller studies, were of low methodological quality. The primary outcomes of this review (death and ESKD) were not reported in 10 and 4 studies respectively. Numerical data on weight difference and protein energy wasting were provided in few studies though all the studies reported that participants' body weight, BMI and mid arm circumference were measured. Adherence was reported in all the studies and was measured at one to three monthly intervals utilising urine nitrogen excretion for calculated protein intake or dietary recall. While most studies reported satisfactory adherence, difficulty in maintaining adherence was reported in two studies. No studies reported any assessment of quality of life although one study commented that quality of life would be improved if participants were not restricted in their dietary protein intake. Quality of life is significantly reduced in patients who require dialysis. Quality of life should be assessed with dietary interventions, aimed at delaying the onset of ESKD, to confirm that any impact of diet on quality of life is minimal compared with the impact of dialysis treatment.
Although Bergstrom 1986, ESGCMCRF 1 1990 and ESGCMCRF 2 1990 were reported as full text journal articles, these articles only provided preliminary results and our updated search and contact with authors did not identify a publication of the full results for the studies. Most studies were small with only five studies enrolling more than 100 participants in each treatment group. Although we identified six studies not previously included in this review, only Garneata 2013 was a large, well reported and high quality study. Only two new studies (Garneata 2013; Milovanov 2009) and the full text publication of Cianciaruso 2008a were published since the 2009 update. The other four new studies identified for this update had been published before 2009.
Quality of the evidence
For all the studies included in the review, full length journal articles were available. However included studies were commonly reported incompletely and were of poor methodological quality and this may reflect pre‐2001 CONSORT (Consolidated Standards of Reporting Trials) practices in the older studies (www.consort-statement.org). Random sequence generation and allocation concealment were considered at low risk of bias in 11 and nine studies respectively. All studies were considered at high risk for performance bias as they were open‐label studies. We assessed bias for outcome assessment (detection bias) for GFR and ESKD separately. As the outcome measurement for GFR measurement was a laboratory outcome all studies were assessed at low risk of detection bias. We felt it important to include a separate assessment of bias for outcome assessment of ESKD as this outcome is more likely to be at risk of detection bias. Eight of 16 studies, reporting data on this outcome, were at unclear or high risk for detection bias; the other eight studies were at low risk of bias since the need to commence dialysis was determined by personnel independent of the trial investigators. Where outcome assessment for the need to commence dialysis is not blinded, the time of dialysis commencement may be influenced by the physicians' knowledge of the treatment groups (Kasiske 1998). Five studies were assessed at high risk of attrition bias with eleven studies at low risk. Eleven studies were at high risk for reporting bias as they did not include data which could be included in a meta‐analysis.
The overall certainty of the evidence using GRADE (GRADE 2011a; GRADE 2011b) was assessed as moderate, low or very low for different outcomes (summary of findings Table 1; summary of findings Table 2). The certainty of the evidence for death was assessed as moderate for studies comparing low with normal protein diets and for studies comparing very low with low or normal protein diets. The certainty of the evidence for ESKD was low for studies comparing low with normal protein diets and moderate for studies comparing very low with low or normal protein intake. The certainty of the evidence for end or change in GFR and body weight was assessed as very low.
Potential biases in the review process
A comprehensive search of the Cochrane Kidney and Transplant’s Specialised Register was performed for this review thus reducing the possibility that potential eligible studies were omitted from the review. Eligible studies published after the last search date or published in conference proceedings not routinely searched could have been missed. The review was completed independently by at least two authors who participated in all steps of the review, which limited the risk of errors in determining study eligibility, in data extraction, in risk of bias assessment and in data synthesis.
Agreements and disagreements with other studies or reviews
The 2009 update of this review found that reduced protein intake in CKD participants reduced the number of participants, who died or reached ESKD (Fouque 2009). The benefit was primarily seen in the subgroup of studies comparing very low protein diets with low or normal protein diets. In this update with additional studies, we were able to report separately on death and ESKD. We confirmed that the reduction in the number of participants reaching ESKD was limited to studies comparing very low protein diet with low or normal protein diets (moderate certainty evidence).
Two other systematic reviews have evaluated the efficacy of low protein diets in participants with CKD. Five RCTs, including 1413 participants with non‐diabetic CKD, were reviewed by Pedrini 1996. Dietary protein restriction compared with a normal protein intake reduced the risk of the combined outcome of death and ESKD (RR 0.67, 95% CI 0.50 to 0.89). All five RCTs are included in our updated review and include two large studies (Locatelli 1989; MDRD 1 1989; MDRD 2 1989). This study evaluated the number of participants reaching ESKD but did not evaluate change in GFR.
Kasiske 1998 evaluated 13 RCTs (1919 participants) including four studies of patients with diabetic CKD. As in this systematic review, the Modification of Diet in Renal Disease study (MDRD 1 1989; MDRD 2 1989) was treated as two studies. The difference in dietary protein intake between the intervention and control groups was 0.33 ± 0.26 g/kg/day. In the pooled results, the authors found that dietary protein restriction reduced the rate of decline in estimated GFR by only 0.53 mL/min/year (95% CI 0.08 to 0.98). They concluded that though there was a decline in GFR with protein restriction, the magnitude of this effect was relatively weak. This review evaluated changes in GFR but not the number of participants, who died or reached ESKD. The authors pointed out that their results were in keeping with the major findings of the MDRD study, which showed little benefit of protein restriction on the number of participants reaching ESKD or on GFR in participants with GFR below 30 mL/min/1.73 m2 (MDRD 2 1989).
The use of low protein diets in participants with CKD varies between countries and within countries. Few nephrologists in the USA or Canada prescribe dietary therapy for participants with CKD (Kalantar‐ Zadeh 2016) following the negative results of the MDRD study (MDRD 1 1989; MDRD 2 1989) while low protein diets are more commonly prescribed in Europe (Bellizzi 2016). International (KDIGO 2012) and national guidelines (Wright 2011) now recommend protein intakes of 0.75 to 0.8 g/kg/day for adults with GFR ≤ 30 mL/min/ 1.73 m2. These recommendations are in line with the recommended daily intake (RDI) for the general population. The average protein intake in adults in developed countries is approximately twice the RDI so guidelines suggest that participants with excess protein intakes reduce their intake to RDI levels since a high protein intake may accelerate the decline of kidney function in CKD (KDIGO 2012; Johnson 2013).
Risk of bias graph: review authors' judgements about each risk of bias item presented as percentages across all included studies.
Risk of bias summary: review authors' judgements about each risk of bias item for each included study.
Comparison 1: Low protein diet versus normal protein diet, Outcome 1: Death (all causes)
Comparison 1: Low protein diet versus normal protein diet, Outcome 2: ESKD
Comparison 1: Low protein diet versus normal protein diet, Outcome 3: End or change in GFR
Comparison 2: Very low protein diet versus low or normal protein diet, Outcome 1: Death (all causes)
Comparison 2: Very low protein diet versus low or normal protein diet, Outcome 2: ESKD
Comparison 2: Very low protein diet versus low or normal protein diet, Outcome 3: End or change in GFR
Comparison 3: Nutritional measures, Outcome 1: Final body weight
Comparison 3: Nutritional measures, Outcome 2: Protein energy wasting
| Low protein diet versus normal protein diet for non‐diabetic adults with CKD | |||||
| Patient or population: non‐diabetic adults with CKD | |||||
| Outcomes | Anticipated absolute effects* (95% CI) | Relative effect | No. of participants | Certainty of the evidence | |
|---|---|---|---|---|---|
| Risk with normal protein diet | Risk with low protein diet | ||||
| Death (all causes) | 55 per 1,000 | 42 per 1,000 | RR 0.77 | 1680 (5) | ⊕⊕⊕⊝ |
| ESKD | 144 per 1,000 | 151 per 1,000 | RR 1.05 | 1814 (6) | ⊕⊕⊝⊝ |
| End or change in GFR | The SMD for end or change in GFR was 0.18 lower (0.75 lower to 0.38 higher) with low protein diet compared to normal protein diet | ‐ | 1680 (8) | ⊕⊝⊝⊝ | |
| *The risk in the intervention group (and its 95% CI) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). | |||||
| GRADE Working Group grades of evidence | |||||
| 1 The confidence intervals include potential for important benefits and harms 2 Important and unexplained heterogeneity present 3 The outcome reported is a surrogate outcome | |||||
| Very low protein diet versus low or normal protein diet for non‐diabetic adults with CKD | |||||
| Patient or population: non‐diabetic adults with CKD | |||||
| Outcomes | Anticipated absolute effects* (95% CI) | Relative effect | No. of participants | Certainty of the evidence | |
|---|---|---|---|---|---|
| Risk with low or normal protein diet | Risk with very low protein diet | ||||
| Death (all causes) | 39 per 1,000 | 49 per 1,000 | RR 1.26 | 681 (6) | ⊕⊕⊕⊝ |
| ESKD | 458 per 1,000 | 293 per 1,000 | RR 0.64 | 1010 (10) | ⊕⊕⊕⊝ |
| End or change in GFR | The SMD for end or change in GFR was 0.12 (0.27 lower to 0.52 higher) with very low protein diet compared to low or normal protein diet | ‐ | 456 (6) | ⊕⊕⊝⊝ | |
| *The risk in the intervention group (and its 95% CI) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). | |||||
| GRADE Working Group grades of evidence | |||||
| 1 The confidence intervals are wide and include potential for important benefits and harms 2 Serious unexplained heterogeneity 3 Outcome is a surrogate outcome 4 Unclear allocation concealment in 4 studies contributing information to analysis | |||||
| Nutritional measures for non‐diabetic adults with CKD | |||||
| Patient or population: non‐diabetic adults with CKD | |||||
| Outcomes | Anticipated absolute effects* (95% CI) | Relative effect | No. of participants | Certainty of the evidence | |
|---|---|---|---|---|---|
| Risk with normal or low protein diet | Risk with very low or low protein diet | ||||
| Final body weight: low protein versus normal protein diet | The mean final body weight 3.09 kg lower (5.02 to 1.16 lower) with low protein diet compared to normal protein diet | ‐ | 223 (2) | ⊕⊝⊝⊝ | |
| Final body weight: very low protein diet versus low protein diet | The mean final body weight 1.4 kg higher (3.40 lower to 6.21 higher) with very low protein diet compared to low protein diet | ‐ | 291 (4) | ⊕⊝⊝⊝ | |
| Protein energy wasting (malnutrition) | 4 per 1,000 | 6 per 1,000 | RR 1.31 | 2373 (15) | ⊕⊕⊝⊝ |
| *The risk in the intervention group (and its 95% CI) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). | |||||
| GRADE Working Group grades of evidence | |||||
| 1 Increased risk of bias related to incomplete outcome data and selective reporting 2 Small studies and wide confidence intervals and include potential for important benefits and harms 3 3/4 studies are unclear for allocation concealment and random sequence generation 4 Serious unexplained heterogeneity | |||||
| Study | Difference in prescribed protein intake | Difference in actual protein intake | Difference between prescribed and actual protein intake |
|---|---|---|---|
| 0.4 g/kg/day | 0.12 g/kg/day | 0.28 g/kg/day | |
| 0.45 g/kg/day | 0.21 g/kg/day | 0.24 g/kg/day | |
| 0.25 g/kg/day | 0.17 g/kg/day | 0.08 g/kg/day | |
| 0.2 g/kg/day | 0.2 g/kg/day | 0 g/kg/day | |
| 0.4 g/kg/day | 0.18 g/kg/day | 0.22 g/kg/day | |
| 0.625 g/kg/day | 0.19 g/kg/day | 0.435 g/kg/day | |
| 0.72 g/kg/day | 0.4 g/kg/day | 0.32 g/kg/day | |
| 0.4 g/kg/day | 0.87 g/kg/day | 0.47 g/kg/day | |
| 0.4 g/kg/day | 0.45 g/kg/day | 0.05 g/kg/day |
| Study | Difference in prescribed protein intake | Difference in actual protein intake |
|---|---|---|
| 0.3 g/kg/day | 0.21 g/kg/day | |
| 0.3 g/kg/day | 0.29 g/kg/day | |
| 0.3 g/kg/day | 0.29 g/kg/day | |
| 0.3 g/kg/day | 0.21 g/kg/day | |
| 0.295 g/kg/day | 0.22 g/kg/day | |
| 0.3 g/kg/day | 0.3 g/kg/day | |
| 0.3 g/kg/day | 0.27 g/kg/day |
| MDRD Study 1 | MDRD Study 2 | |||||||
|---|---|---|---|---|---|---|---|---|
| Usual protein diet | Low protein diet | Low protein diet | Very low protein diet | |||||
| Final body weight (kg) | N | Final body weight (kg) | N | Final body weight (kg) | N | Final body weight (kg) | N | |
| Men | 88.5 ± 14.6 | 179 to 183 | 83.2 ± 12.8 | 165 to 170 | 79.6 ± 11.5 | 74 to 77 | 79.3 ± 10.9 | 69 to 71 |
| Women | 72.2 ± 14.9 | 98 to 105 | 69.3 ± 13.7 | 107 to 115 | 65.9 ± 11.9 | 49 to 51 | 65.0 ± 14.3 | 49 to 52 |
| Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size |
| 1.1 Death (all causes) Show forest plot | 5 | 1680 | Risk Ratio (M‐H, Random, 95% CI) | 0.77 [0.51, 1.18] |
| 1.2 ESKD Show forest plot | 6 | 1814 | Risk Ratio (M‐H, Random, 95% CI) | 1.05 [0.73, 1.53] |
| 1.3 End or change in GFR Show forest plot | 8 | 1680 | Std. Mean Difference (IV, Random, 95% CI) | ‐0.18 [‐0.75, 0.38] |
| Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size |
| 2.1 Death (all causes) Show forest plot | 6 | 681 | Risk Ratio (M‐H, Random, 95% CI) | 1.26 [0.62, 2.54] |
| 2.2 ESKD Show forest plot | 10 | 1010 | Risk Ratio (M‐H, Random, 95% CI) | 0.64 [0.49, 0.85] |
| 2.3 End or change in GFR Show forest plot | 6 | 456 | Std. Mean Difference (IV, Random, 95% CI) | 0.12 [‐0.27, 0.52] |
| Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size |
| 3.1 Final body weight Show forest plot | 6 | 514 | Mean Difference (IV, Random, 95% CI) | ‐0.46 [‐3.93, 3.01] |
| 3.1.1 Low protein versus normal protein diet | 2 | 223 | Mean Difference (IV, Random, 95% CI) | ‐3.09 [‐5.02, ‐1.16] |
| 3.1.2 Very low protein diet versus low protein diet | 4 | 291 | Mean Difference (IV, Random, 95% CI) | 1.40 [‐3.40, 6.21] |
| 3.2 Protein energy wasting Show forest plot | 15 | 2373 | Risk Ratio (M‐H, Random, 95% CI) | 1.31 [0.42, 4.13] |
