A dose-dependent association of total cholesterol with all-cause and speci c mortality: results from the National Health and Nutrition Examination Survey

Guo-dong He Guangdong Provincial People's Hospital Xiao-cong Liu Guangdong Provincial People's Hospital Lin Liu Guangdong Provincial People's Hospital Yu-ling Yu Guangdong Provincial People's Hospital Chao-lei Chen Guangdong Provincial People's Hospital Jia-yi Huang Guangdong Provincial People's Hospital Kenneth Lo Brown University Yu-qing Huang Guangdong Provincial People's Hospital Ying-qing Feng (  651792209@qq.com ) Guangdong Cardiovascular Institute https://orcid.org/0000-0003-0443-5506

Mortality linkage methods used herein can be obtained from the NCHS. The anonymized data of individuals who participated in NHANES between 1999 to 2014 were allied to the longitudinal Medicare and mortality data based on the sequence number of NHANES. The NCHS classi ed mortality from CVD as follows: codes: I00-I09 (acute rheumatic fever and chronic rheumatic heart diseases), codes: I11 (hypertensive heart disease), codes: I13 (hypertensive heart and renal disease), codes: I20-I25 (ischemic heart diseases), codes: I26-I51 (other heart diseases), codes: I60-I69 (cerebrovascular disease), and codes: C00-C99 (mortality from cancer) according to ICD-10.

Data analysis
All the data analyses covered the complex, strati ed, multistage, and cluster-sampling design (such as oversampling of some subpopulations) of NHANES based on the strata, sample weights, and primary sampling units incorporated in the NHANES data.
TC levels were divided into six groups < 120, 120-159, 160-199, 200-239, 240-279, and ≥ 280 mg/dL. The group which had the lowest mortality acted as a reference. Multivariable Cox proportional hazards model was applied to estimate the HRs (hazard ratios) and 95% CIs (con dence intervals). Cox proportional hazards regression models were utilized to understand the link between TC levels and total mortality and cause-speci c mortality based on speci c covariates: age, gender, and race (model II); model III plus education level, married, smoking, SBP, BMI, HDLC, estimated GFR, comorbidities (diabetes, hypertension, CVD, and cancer), and drug usage.
The two-piecewise linear regression model and Cox models with penalized splines were used to examine the non-linear and irregular shape of the link between baseline TC and the risk of total mortality and cause-speci c mortality using non-parametric smoothers. The survival probability of primary outcomes according to the different ranges of TC variability was calculated by using Kaplan-Meier curves for cause-speci c mortality and all-cause mortality.
To determine whether the relationship varied by age at baseline (< 65 or ≥ 60 years), sex (female/male), race (white/non-white), CVDs (yes/no), cancer (yes or no) and lipid-lowering drugs (yes or no), we performed subgroup analyses.
All analyses were administrated with SAS 9.3 software (SAS Inc), and the alpha level for signi cance in statistical analysis was set as 0.05 in this study.

Baseline characteristics
The demographic features of the study participants based on the six TC levels are listed in Table 1. The average age of the 36775 adults was 49. 24 years, and about half of the sample were women (51.45%).
Non-white race and white race adults accounted for 52.42% and 47.58%, respectively. Almost threequarters of participants self-identi ed as high school or above. Half of the sample were married. 53.38% of the sample were not current smokers. Women and higher education level persons were more likely to be with higher TC values. The levels of systolic and diastolic blood pressure increased across the TC levels. Hypertension and diabetes approximately accounted for 55% and 16%. However, only 27.04% and 8.98% were treated with antihypertensive drugs and hypoglycemic agents. Values are mean ± standardized differences or n (%). Follow-up data on mortality can be obtained from the date the survey was started (median follow-up: 96.06 ± 53.65 months). Cardiovascular, cancer and all-cause mortality showed U-curve associations after adjusting for confounding variables. Based on the restricted cubic spline analysis (Fig. 2), TC levels of 213(mg/dL), 200(mg/dL), and 218(mg/dL) were associated with the lowest all-cause, cardiovascular diseases, and cancer mortality, respectively. These were largely similar to the TC categorical analysis. There were signi cant non-linear relationships between the levels of TC and speci c and all-cause mortality.  Kaplan-Meier survival curve (Fig. 3) for all-cause, cardiovascular, and cancer mortality showed that the lowest cumulative survival rate was observed in the lowest TC level group.
We observed that the cutoff values were 145 mg/dL,143 mg/dL, and 149 mg/dL for cancer, cardiovascular, and all-cause mortality, respectively, by using a two-piecewise linear regression model (  The two-piecewise linear regression model were adjusted for age, gender, race, education level, married, smoking, body mass index, systolic blood pressure, estimated glomerular ltration rate, high density lipoprotein cholesterol, comorbidities (hypertension, diabetes, cardiovascular disease, and cancer), and medication use(antihypertensive drugs, hypoglycemic agents, antiplatelet drugs, and lipid-lowering drugs).

Subgroup Analyses
Subgroup analyses according to age, gender, race, with or without CVD, with or without cancer, and lipidlowering drugs of diabetes are presented in Table 4. When analyzing a subgroup variable, age, gender, race, education level, married, smoking, body mass index, systolic blood pressure, estimated glomerular ltration rate, high density lipoprotein cholesterol, comorbidities (hypertension, diabetes, cardiovascular disease, and cancer), and medication use(antihypertensive drugs, hypoglycemic agents, antiplatelet drugs, and lipid-lowering drugs) were all adjusted except the variable itself.
Spline analyses of all-cause (A), cardiovascular (B), and cancer (C) mortality by total cholesterol levels. Adjusted for age, gender, race, education level, married, smoking, body mass index, systolic blood pressure, estimated glomerular ltration rate, high density lipoprotein cholesterol, comorbidities (hypertension, diabetes, cardiovascular disease, and cancer), and medication use(antihypertensive drugs, hypoglycemic agents, antiplatelet drugs, and lipid-lowering drugs).
Strong links between the lowest range of TC level and all-cause mortality were observed in all subgroups except for using lipid-lowering drugs. When the range of TC level was 120-159 mg/dL, signi cant associations were also observed in patients aged ≥ 65 years, male, all races, without CVD, without cancer or without using lipid-lowering drugs. When the range of TC range was 200-239 mg/dL, we only found a signi cant association in patients with CVD. However, In the range 240-279 mg/dL, there was a substantial association in patients without CVD or aged ≥ 65 years. In the groups aged < 65 years and using lipid-lowering drugs, the associations between the highest rang of TC level and all-cause mortality was signi cant. There were substantial interactions with CVD and lipid-lowering drugs in the association of TC level with all-cause mortality (P for interaction = 0.002 and, < 0.001, respectively) The signi cant associations between the lowest range of TC and cardiovascular mortality were observed in patients aged ≥ 65 years, male, white, with CVD, with cancer or without using lipid-lowering drugs. Similarly, the signi cant association was also found in patients aged < 65 years with a TC range of 200-239 mg/dL and aged < 65 years or male with the highest range of TC level. However, no signi cant interaction was observed in cardiovascular mortality. nutrition also did not recommend a reduction in dietary cholesterol [23]. Additionally, previous studies mainly focus on elderly populations. Our study showed the stronger effect size associated with TC range < 120 mg/dL at age ≥ 65 years as well as age < 65 in all-cause and cancer mortality. However, interpreting the connection between CVD mortality and cholesterol has been an intractable clinical challenge. In this study, TC range < 120 mg/dL was negatively linked to cardiovascular mortality in patients age ≥ 65 years.
In contrast, the TC range of ≥ 280 mg/dL was positively related to cardiovascular and all-cause mortality in patients age < 65 years. Since people with low cholesterol levels exhibited higher rates of all-cause and cancer mortality, we postulated that low TC may partially re ect frailty. However, for CVD mortality, aged people (age ≥ 65 years) with TC range < 120 mg/dL had a higher risk of death, and those below 65 years of age with TC range ≥ 280 mg/dL were more likely to die relative to aged people. Similarly, the association between low TC and higher occurrence rate of serious adverse cardiovascular events in men aged ≥ 70 years has also been reported [24].
TC levels were classi ed into three categories: desirable (less than 200) borderline high (between 200 to 239), and high level (≥ 240 mg/dL). However, the categories were suggested according to the interactions between IHD and TC. In this study, 213(mg/dL), 200(mg/dL), and 218(mg/dL) TC levels were linked to the lowest all-cause, cardiovascular, and cancer mortality in the spline analyses, respectively. Our data indicated that the optimal overall survival (OS) ranges were higher relative to those for IHD. The previous studies have shown a higher optimal body-mass index range for OS relative to IHD mortality [25].
Cholesterol levels may re ect the general health status but not speci c for CVD [26].
The association of low cholesterol levels with higher mortality may be explained by reverse causality.
However, a study that involved long-term follow-up of the Japanese-American population indicated that people with low cholesterol levels that last for more than 20 years were likely to have the worst all-cause mortality. It suggested that reverse causality was not likely to be fully responsible for the higher mortality linked to low cholesterol [27].
Besides, a key nding in cancer of our study suggested that low cholesterol could be linked to high cancer-associated deaths. Although a few studies of statins indicated that statin therapy increases cancer incidence [28][29][30], it was still challenging to elucidate the connection between low cholesterol and cancer disease mortality. Previous studies on liver cancer reported an increased level of cholesterol [31,32]. However, in this study, a higher rate of cancer mortality in the group with the lowest cholesterol was also signi cant, and the effects of malnutrition should be considered. This suggested that screening the low cholesterol levels of patients with cancers was needed.
A large-scale, long-term complete follow-up study evaluating the in uence of TC on all-cause and speci c mortality are clear strengths of this study, and a large sample size used can ensure statistical e ciency.
The NHANES database represented the entire US population, and this study was showing that low TC is a signi cant risk factor, not only in patients but also in the general population. Another advantage of this study is that it estimated all causes, CVD, and cancer mortality risk related to TC levels as low as < 120 mg/dL. However, this study had some limitations. Firstly, we could not understand the reason behind the link between TC and mortality risk in a cross-sectionally designed study. Secondly, we might have underestimated the risk associated with high cholesterol. Also, we could not verify that low cholesterol levels induced by statin increase mortality. Thirdly, this study was based on American participants; as such, its ndings may not apply to other populations. Although the U-curve associations can apply to other ethnic groups, the level of relative risk related to TC, as well as the optimal TC range with the lowest mortality, could differ among various ethnic groups. Study cohort.

Figure 2
Spline analyses of mortality by total cholesterol levels. Spline analyses of all-cause (A), cardiovascular (B), and cancer (C) mortality by total cholesterol levels. Adjusted for age, gender, race, education level, married, smoking, body mass index, systolic blood pressure, estimated glomerular ltration rate, high density lipoprotein cholesterol, comorbidities (hypertension, diabetes, cardiovascular disease, and cancer), and medication use(antihypertensive drugs, hypoglycemic agents, antiplatelet drugs, and lipid-lowering drugs).

Figure 3
Kaplan-Meier survival curve for all-cause (A), cardiovascular (B), and cancer (C) mortality by total cholesterol groups.