Levey AS, Adler S, Caggiula AQ, England BK, Greene T,
Hunsicker LG, et al
Progression of advanced renal disease in the Modification of
Diet in Renal Disease Study
Am J Kidney Dis
(May) 27:652-663 1996
In this study, the authors of the MDRD study who found no significant
effect of protein restriction on the progression of renal disease in
patients with non-diabetic renal insufficiency reanalyzed the data
correlating actual protein intake to outcome (instead of an intention
to treat analysis).
Protein intake was estimated from UUN. For individuals taking amino
acid supplement (originally assigned to the very low protein group)
the nitrogen component from the amino acids was subtracted from the
UUN to estimate dietary protein intake.
The primary outcome used in this analysis was the slope of change in
GFR using a "one slope informative censoring model where time to renal
failure is taken into account in estimation of slope". They also
looked at risk of renal failure and death.
The authors claim that lower achieved protein intake is associated
with a slower mean decline in GFR and lower risk of renal failure
compared with higher protein intake at any given time.
Comment: One needs to question whether it is appropriate to
analyze the results of a controlled randomized trial in this fashion.
Ignoring the intention to treat analysis may introduce bias to the
results. For instance patients who do not comply with the very low
protein diet and have higher protein intake are also more likely to
not comply with other aspects of their care which may impact on the
progression of their renal disease, and are not accounted for in the
analysis and make a higher protein intake appear to have a worse
affect on the progression of the renal disease.
In the Cox regression and correlational analyses the authors use mean
protein intake over entire study period as opposed to protein intake
at the start of the study or protein intake as a time dependent
covariate which can introduce bias. This flawed use of a covariate
and its effect on results are reviewed by Wolfe, et.al. (AJKD, Jan,
In addition, the lower risk of renal failure estimated for the low
protein intake diet may be related to lower levels of azotemia at a
given time for those able to achieve lower protein diets. The
authors admit that patients on the low protein diet are started on
dialysis at lower GFR's and higher creatinines but show that levels of
BUN and number of symptoms from azotemia were similar for those
achieving a low protein diet compared with higher protein intake.
Certainly, this can at least in part explain the renal survival
benefit of low protein intake which is unrelated to deterioration of
GFR. (Jeffery Fink, MD, UW Nephrology Clinical Research Training
Group, Seattle, WA)