Fasting status modifies the association between triglyceride and all‐cause mortality: A cohort study

Abstract Background and Aims Both fasting and non‐fasting levels of triglyceride have been shown positively associated with all‐cause mortality. It is unknown whether fasting status modifies this association. This study aimed to address this question. Methods This study included 34,512 US adults (27,036 fasting and 7476 nonfasting participants). All‐cause mortality was ascertained by linkage to the National Death Index records. Cox proportional hazards models were used to estimate hazard ratios of triglyceride for mortality. Results This cohort was followed up for a mean of 13.0 years. During the follow‐up, 8491 all‐cause deaths were recorded. A 1‐natural‐log‐unit increase in triglyceride was associated with an 8% higher multivariate‐adjusted risk of all‐cause mortality. Interaction analyses showed that fasting status interacted with triglyceride in predicting all‐cause mortality. Sub‐analyses showed that a 1‐natural‐log‐unit increase in triglyceride was associated with a 17% higher multivariate‐adjusted risk of all‐cause mortality in the nonfasting subcohort; however, there lacked such an association in the fasting sub‐cohort. Similarly, high (200–499 mg/dL) and very high levels of triglyceride (≥500 mg/dL) were associated with higher all‐cause mortality risks compared with low normal triglyceride (<100 mg/dL) only in the nonfasting subcohort. Conclusion This study found that, compared to fasting triglyceride, nonfasting triglyceride was more sensitive in predicting all‐cause mortality. This study supports the initiatives by some guidelines to recommend the use of nonfasting triglycerides for risk assessment.


| Triglyceride classification
The baseline concentration of triglyceride in the serum was directly retrieved from the NHANES website. [11][12][13] The participants were divided into five groups according to triglycerides levels as previously reported, 3 that is, low-normal (<100 mg/dL), high-normal (100-149 mg/dL), borderline-high (150-199 mg/dL), high (200-499 mg/dL), and very high (≥500 mg/dL). This classification was based on the four categories (normal, borderline-high, high, and very high) recommended by the National Cholesterol Education Program (NCEP) Expert Panel, 16 with further categorizing the normal triglyceride levels into low-normal and high-normal as previously reported. 3

| Mortality ascertainment
Data on all-cause mortality were directly retrieved from NHANESlinked mortality files. [11][12][13] To evaluate mortality status, the National Center for Health Statistics conducted probabilistic matching to link the NHANES data with death certificate records from the National Death Index (NDI) records. Follow-up time was defined as the time (in months) from the time when the blood was drawn at the Mobile Examination Center until death, or until the end of follow-up (i.e., December 31, 2015), whichever occurred first. [11][12][13]

| Statistical analyses
Data were presented as mean and standard deviation for normally distributed continuous variables, median and interquartile range for not normally distributed continuous variables or number and percentages for categorical variables. 19 The difference in age among groups was analyzed using one-way analysis of variance (ANOVA) and the difference in triglyceride (non-normally distributed) among groups was analyzed using Kruskal-Wallis one-way ANOVA. 20 Differences among categorical variables were analyzed using Pearson's χ 2 test. Cox proportional hazards models 21 were used to calculate hazard ratios (HRs) and 95% confidence intervals (CIs) of triglyceride for mortality, with or without adjustment for confounding factors including age, sex, ethnicity, obesity, education, povertyincome ratio, lifestyle factors (physical activity, alcohol consumption, and smoking status), survey period, and clinical confounders (hypercholesterolemia, hypertension, and diabetes). Triglyceride was treated as a continuous variable (natural log-transformed) or categorical variable in the models. Sub-analyses were conducted in the fasting and nonfasting subcohorts and in the male and female subcohorts. Sensitivity analyses were conducted by further adjustment for history of heart attack (yes, no, or unknown) and stroke (yes, no, or unknown).
The null hypothesis was rejected for a two-sided p value of <0.05. All analyses were performed using SPSS version 27.0 (IBM SPSS Statistics for Windows, IBM Corporation).

| General characteristics
This study included 34,512 US adults (27,036 fasting and 7476 nonfasting participants) with a mean (standard deviation) age of 49 (19) years. Baseline characteristics of the participants are described in Tables 1 and S1-S2. People with higher triglyceride had a higher percentage of males, had less income and education, and had a higher prevalence of hypercholesterolemia and diabetes in the whole cohort (Table 1) as well as in the fasting and non-fasting subcohorts (Tables S1-S2).

| Association of circulating triglyceride with all-cause mortality
This cohort was followed up for 450,333 person-years (mean followup, 13.0 years). During the follow-up, 8491 all-cause deaths were recorded. A 1-natural-log-unit increase in triglyceride was associated with an 8% higher multivariate-adjusted risk of all-cause mortality (HR, 1.08; 95% CI, 1.04-1.13, Table 2). The interaction analysis showed that triglyceride interacted with fasting status in predicting all-cause mortality (Table 3). Subanalyses showed that, in the nonfasting sub-cohort, a 1-natural-log-unit increase in triglyceride was associated with a 17% higher multivariate-adjusted risk of all-cause mortality (HR, 1.17; 95% CI, 1.09-1.25, Table 2), whereas there lacked such an association in the fasting subcohort (Table 2). In addition, the interaction analysis showed that triglyceride interacted with sex in predicting all-cause mortality (Table S3): the positive association between triglyceride and all-cause mortality presented in females but not in males (Table S4). Ethnicity did not interact with triglyceride in predicting all-cause mortality (Table S5).
When triglyceride was treated as a categorical variable, people who had high triglyceride (HR, 1.09; 95% CI, 1.02-1.16) or very high triglyceride (HR, 1.19; 95% CI, 1.01-1.39) had higher risks of allcause mortality compared with those with low-normal triglyceride (Table 4). Subanalyses showed that triglyceride was positively associated with all-cause mortality risks in a dose-dependent manner in the non-fasting sub-cohort, whereas there lacked such an association in the fasting subcohort (Table 4).

| Sensitivity analyses
Sensitivity analyses showed that further adjustment for history of heart attack and stroke did not materially change the results (Table S6).

| DISCUSSION
Using a representative cohort of US adults, this study found that fasting status modified the association between triglyceride and allcause mortality: nonfasting triglyceride, but not fasting triglyceride, was positively associated with all-cause mortality risks after adjustment for all the tested confounders.
This study suggests that non-fasting triglyceride may be better in predicting all-cause mortality than its fasting counterpart. The results of this study are of clinical importance given that non-fasting tests are more comfortable and convenient for individuals than fasting tests. 13 In addition, fasting blood tests may be less safe for people with diabetes who can experience hypoglycemia when fasting. 22,23 Nonfasting triglyceride level was~27 mg/dL higher compared to fasting samples in general, 22,24 which increment is believed to be not clinically significant for most individuals. 22 Nonfasting samples seem to have at least the same prognostic value as fasting samples for general risk screening. 22 27 The findings of the current study support the initiatives of these guidelines 22,23,26,27 to recommend the use of nonfasting triglyceride for general screening and risk evaluation.
The ACC/AHA guideline 26    This study found that, in the whole cohort comprising the fasting and nonfasting subcohorts, triglyceride was associated with all-cause mortality. This result is consistent with two recent studies which reported similar findings. 9,10 The exact cause of death was not investigated by the current study; however, the result was consistent with previous observations that higher triglyceride was associated with higher mortality from cardiovascular disease 2 and diabetes. 11 This study found that fasting triglyceride was not associated with all-cause mortality, which is consistent with some previous reports 4,5 but inconsistent with others. [1][2][3] The lack of association in the current study is not due to a limitation in sample size, as this study included 27,036 participants with nonfasting triglyceride.
Why nonfasting triglyceride is more sensitive in predicting allcause mortality than its fasting counterpart is unknown. Out of a routine day with three meals (e.g., meals at 8 a.m., 12 noon, and 7 p.m., respectively), a person is in a fasting state for only about 5 h whereas in a non-fasting state for the remaining 19 h. Therefore, non-fasting triglyceride, which is about 27 mg/dL higher than the fasting value, 22,24 may more represent the impact of triglyceride on a person's health. It is worth noting that postprandial levels of triglyceride can be affected by many factors, for example, type of diet, time after a meal, and comorbidity. Therefore, an oral lipid tolerance test, 28 similar to the oral glucose tolerance test, is expected to have higher clinical significance compared to a random nonfasting triglyceride test. This hypothesis needs to be investigated in the future.
This study found that the positive association between triglyceride and all-cause mortality was presented in females but males. The underlying reason is unknown. The triglyceride-related gender difference has been reported previously. For example, a meta-analysis of 17 population-based prospective studies showed that high triglyceride was a risk factor for cardiovascular disease in the general population, 29 and the association seemed to be stronger in females than in males: relative risks (95% CIs) were 1.14 (1.05-1.28) and 1.37 (1.13-1.66) in males and females, respectively. 29 Consistently, the Framingham Study showed that baseline triglyceride was positively associated with the incidence of coronary heart disease in females but not in males. 30