Context: In a prior study we observed that adjustment of serum aldosterone for age, sex and sodium status led to confirmation of our hypothesis of bimodality of serum aldosterone in low-renin hypertension; the hypothesis was not statistically supported without this adjustment. Here we aim to further characterize the effect of this statistical adjustment on the normality of the distribution of serum aldosterone.
Methods: We analyzed data from 1486 normotensive men and women, aged 29 to 85, in the sixth examination cycle of the Framingham Offspring Study. We used independent two sample t-tests, regression of natural-log-transformed values of aldosterone on age, sex, and urinary sodium: creatinine ratio, the dip test for unimodality, dotplots, and Q-Q plots to study the normality of the distribution of aldosterone concentrations with and without statistical adjustment for the covariates listed above.
Results: The proportion of variability of serum aldosterone concentrations explained by age, sex, and urine sodium:creatinine ratio was 0.91%,1.3%, and 8%, respectively (p > 0.005). The distribution of adjusted serum aldosterone was unimodal (dip test: 2.26 ± 0.17 ng/dL, p = 0.997); when unadjusted, the distribution was not unimodal (dip test: 2.26 ± 0.50 ng/dL, p < 0.00005). The dotplot and Q-Q plot also showed a more normal distribution of serum aldosterone concentrations after adjustment.
Conclusion: Adjustment of serum aldosterone for age, sex, and urine sodium:creatinine ratio leads to a more normal aldosterone distribution. Further study is needed to determine whether the use of this adjustment in testing for primary aldosteronism may improve the accuracy of clinical testing