Co-occurrence characteristics of cardiometabolic abnormalities among young and middle-aged adults: a study of 19,341 health examination records with at least one abnormal finding

Cardiometabolic diseases have become a major public health concern among young and middle-aged populations, while the way these abnormalities occur together in health examination populations has not been examined with methods that separate co-occurrence from the frequency of each abnormality. This study retrospectively analyzed the demographic characteristics and health examination data of 19,341 individuals who underwent physical examinations at the health management center of a tertiary hospital in Yunnan Province from April 2025 to April 2026, including 4,446 young adults aged 18–35 years and 14,895 middle-aged adults. Only records of individuals with one or more flagged abnormality were included in the exported file, so the number of people examined is unknown and every proportion below refers to records with at least one abnormality and is an upper bound for the corresponding proportion among all examinees. Data were exported from the health examination system and organized for analysis. All estimation was done in R version 4.5.3. Associations between pairs came from log-linear models fitted to the sixty-three observable response patterns, with the cell for no abnormality treated as unobserved. Each pairwise estimate from the model was compared to an odds ratio calculated among records positive for at least one of the other four abnormalities. The Venn package was used to show the overlap. 44.04% of the young group and 56.68% of the middle-aged group had two or more abnormalities. The proportion of records with four or more abnormalities in the middle-aged group (7.82%) was approximately three times higher than that in the young group (2.41%). In the young group, fatty liver disease combined with hyperuricemia was the most common pairwise co-occurrence pattern (1,150 cases, 25.87%). In the middle-aged group, fatty liver disease combined with the diabetes risk label gave the largest pairwise count (2,647 cases, 17.77%). A likelihood ratio test rejected independence between the six abnormalities in both groups, with statistics of 869.19 and 3,033.86 on 15 degrees of freedom. Fatty liver disease had the strongest pairwise associations. Its odds ratios for hypercholesterolemia, hypertriglyceridemia and hyperuricemia were 4.50, 4.23 and 3.79 in the young group. For fatty liver disease with hyperuricemia, the crude estimate was 0.67, compared with 4.06 after taking the export criterion into account. The two lipid labels rarely co-occurred on the same record: 7 co-occurrences against 124.4 expected in the young group, 67 against 665.3 in the middle-aged group. A second-order model reproduced the distribution of the number of abnormalities, and no higher-order term was needed.