To understand the implications of elevated serum phosphorus on clinically important outcomes in the CKD population.
Study Design and Population
This is a retrospective cohort study using and inception cohort of CKD patients from a large HMO in the North Western United States. To ensure that this was truly an inception cohort, patients (with no prior renal replacement therapy) had to have 2 successive eGFR estimations below 60 ml/min preceded by a normal eGFR. A phosphorus value after the second low eGFR was required for enrollment.
Intervention or Observation
Patients were followed for up to 5 years for the outcomes of mortality, cardiovascular mortality, cardiovascular hospitalization or requirement for renal replacement therapy. Updated phosphorus levels and eGFR were use to create a time-varying model of the effects of phosphorus levels and renal function.
Primary End-Point, Secondary End-Points
The primary end point was all cause mortality. Secondary end points were cardiovascular mortality, cardiovascular hospitalization and requirement for renal replacement therapy.
2,122 patients fulfilled the enrollment criteria out of 23,204 who had low eGFRs Main exclusions were lack of previously normal eGFR (~13,000) and lack of phosphorus measurement (~7,800). A complete data set was available for 930 patients. The completed cases were younger, more likely to be male and had a higher disease burden of diabetes mellitus and cardiovascular disease.
While the crude death rates increased slightly with the level of baseline phosphorus, this was not statistically significant. The relationship between phosphorus and all cause mortality rate was non-linear, and not higher at higher phosphorus levels. There was strong evidence of confounding by level of renal function. Phosphorus level was not associated with cardiovascular mortality or cardiovascular hospitalizations.
The rate of renal replacement therapy increased with increasing phosphorus level (HR 1.69: 95% CI 1.29 – 2.21), and the hazard ratio was increased when eGFR was removed from the model (HR 3.9: 95% CI 3.19 – 4.77). However, the authors caution that the effect may be exaggerated since there were more phosphorus values available as renal function declined presenting the risk of confounding by propensity of testing.
The strength of the study is that it does represent a true inception cohort, thus eliminating survival bias. However the retrospective nature of the study leaves it vulnerable to missing or unidentified confounders. The authors note, for example that they could not include PTH in their model because of a sparsity of measurements.
The length of follow-up is less than 5 years for this cohort, which may not be sufficient time for the effects of any variable, e.g. phosphorus to impact on survival. The number of events for several of the secondary outcomes was insufficient for the authors to be conclusive in their lack of associations between phosphorus and the outcome.
Impact on Practice
There are old data on low phosphate diets, and animal data suggesting that phosphate overload is a factor influencing progressing of renal disease. The associative data here are provocative and suggest the need for a randomized clinical trial of aggressive phosphate control in an inception cohort of early CKD to ascertain whether this influences renal outcomes.