Serum phosphorus and mortality in the Third National Health and Nutrition Examination Survey (NHANES III): effect modification by fasting

Background: Serum phosphorus levels have been associated with mortality in some but not all studies. Because dietary intake prior to measurement can affect serum phosphorus levels, we hypothesized that the association between serum phosphorus level and mortality is strongest in those who have fasted longer.

Study design: Prospective cohort study.

Setting & participants: Nationally representative sample of 12,984 participants 20 years or older in the Third National Health and Nutrition Examination Survey (1988-1994).

Factors: Serum phosphorus level, fasting duration (dichotomized as ≥ 12 or < 12 hours).

Outcomes: All-cause and cardiovascular mortality determined by death certificate data from the National Death Index.

Measurements: Serum phosphorus measured in a central laboratory and fasting duration recorded as time since food or drink other than water was consumed.

Results: Individuals fasting 12 or more hours had lower serum phosphorus levels than those fasting less than 12 hours (3.34 vs 3.55 mg/dL; P < 0.001) and higher correlation with repeat measurement (0.66 vs 0.53; P = 0.002). In multivariable-adjusted Cox regression models, the highest quartile of serum phosphorus was associated with increased mortality in participants fasting 12 or more hours (adjusted HR, 1.74; 95% CI, 1.38-2.20; reference, lowest quartile) but not in participants fasting less than 12 hours (adjusted HR, 1.08; 95% CI, 0.89-1.32; P for interaction = 0.002). Relationships were consistent using 8 hours as the fasting cutoff point or cardiovascular mortality as the outcome.

Limitations: Observational study, lack of fibroblast growth factor 23 or intact parathyroid hormone measurements.

Conclusions: Fasting but not nonfasting serum phosphorus levels were associated with increased mortality. Risk prognostication based on serum phosphorus may be improved using fasting levels.