Disease modifies the dependency of percentiles of the phase angle distribution on age, sex, height and weight in hospitalized patients

Background & aims: Phase angle (PhA) is viewed as a holistic indicator of quantity and quality of cellularity and hydration status and has emerged as a significant predictor of patient outcome in clinical medicine. We sought to analyze the impact of hospitalization as a surrogate for disease on the distribution of PhA and its dependency on influence variables age, sex, height and weight without any assumption as to the form of PhA-distribution.

Methods: First PhA measurements obtained from 2418 women (median age 75 IQR[63; 82]) and 2541 men (median age 70 IQR[60; 79]) hospitalized in a Community General Hospital were analyzed. Multivariable quantile regression was applied for estimating percentiles P1 - P95 using parsimonious models including a dichotomous factor for sex and cubic polynomials for age (model A) and height and weight (model B) using only linear interaction terms between the four variables sex, age, height, and weight.

Results: The association of PhA was strongest with age (women r = -0.48; men r = -0.47). In each age class average PhA values of hospitalized patients were below those reported for healthy individuals. In contrast to percentiles above the median showing a monotonous decrease with age as reported from healthy individuals the lower percentiles of patients showed a marked dip-and-plateau deformation. This deformation was associated with a change in the distribution span of PhA between P1 and P95 which was narrower at young age, expanded markedly due to a persisting fraction of patients with low PhA over the age range from 50 to 80 years and became narrower again at higher age due to the decreasing fraction of patients with high PhA. These distribution patterns were the same, irrespective of using either model A or model B. Furthermore, bootstrapping confirmed the estimated form of the percentile curves.

Conclusions: Disease modifies the PhA distribution pattern resulting not only in lower PhA in patients than in healthy individuals but also in a dip-and-plateau deformation of lower PhA percentile curves for the association with age. The dip-and-plateau pattern and the narrowing of the span between P1 and P95 with older age suggest that there is a low threshold value for PhA, below which life is impossible.

Clinical trial registry: DRKS00025307, https://www.drks.de/DRKS00025307.