Association between the nonHDLc/HDLc ratio and 30-day mortality for patients with sepsis: a retrospective observational cohort study based on a large multicentre critical care database

Background: Dyslipidemia contributes to the development and progression of cardiovascular disease. The objective of this study was to investigate the association between the non-high-density lipoprotein-cholesterol-to-high-density lipoprotein-cholesterol (non-HDL-c/HDL-c) ratio and mortality for patients with sepsis. Methods: Using data from the eICU Collaborative Research Database (eICU-CRD) with high granularity data for over 200,000 admissions to ICUs monitored by eICU Programs across the United States. We identied 1680 patients with sepsis. All-cause mortality within 30-days after the date of visit to the ICU. We estimated the risk of mortality using multivariable logistic-regression model. Result: There were 115 deaths (6.85%). The probability of mortality decreased when the nonHDLc/HDLc ratio lower than the turning point ( <3.58) with a adjusted odds ratio (OR) of 0.75 (95% CI: 0.61–0.94, P=0.011) for every 1 increment of nonHDLc/HDLc ratio. With the per-SD increase in the nonHDLc/HDLc ratio, the OR for mortality was 0.36 (95% CI: 0.16–0.79, P=0.011) when nonHDLc/HDLc ratio was <3.58, while the OR was 1.56 (95% CI: 1.29–1.88, P<0.001) when nonHDLc/HDLc ratio was ≥ 3.58. Conclusion: Higher nonHDLc/HDLc ratio, even at a low level, was associated with a higher risk of 30-day mortality for patients with sepsis. The probability of mortality rose rapidly when the nonHDLc/HDLc ratio higher than the turning point (may at 3.58). the mean ± SD, median (interquartile range), or percentage; HDLc High-density lipoprotein cholesterol; BP Blood pressure; LDLc Low-density lipoprotein cholesterol; COPD, 237 for systolic BP and diastolic for for triglycerides, 468 for LDLc, and 3 (0.2%) for creatinine.

investigate association between the nonHDLc/HDLc ratio and 30-day mortality. In our observational study, we hypothesized that in patients with sepsis even high and low the nonHDLc/HDLc ratio is associated with a higher risk of all-cause mortality within 30-days after the date of visit to the ICU.

Data Source
We extracted data from eICU-CRD database, which is publicly and freely accessible to researchers, according to data usage agreement by the review board of PhysioNet (our record ID: 40859994). The eICU-CRD covers 200 859 ICU admissions of 139 367 patients in 208 US hospitals between 2014 and 2015. All data were stored automatically and retrieved electronically through the Philips Healthcare eICU programme. It includes records of demographics, physiological indices from bedside monitors, diagnosis via International Classi cation of Diseases, 9th Edition, Clinical Modi cation (ICD-9-CM) codes, and other laboratory data obtained during routine medical care. All data were deidenti ed by the eICU programme and anonymous to researchers before analysis [9]. As this research was a retrospective cohort study based on data from eICU-CRD, no ethical approval was required from our local ethics committee.

Study population
Brie y, all patients with sepsis admitted to ICU from eICU database were included. The following exclusion criteria were applied: (i) 89 or 18 years old; (ii) missing total cholesterol. In total, 1680 patients with sepsis were included. The study owchart was presented in Fig. 1.

Variables
All subject data within the rst 24 hours after admission were collected from eICU-CRD. The physiological variables, including heart rate (HR) and current blood pressure (BP), were obtained from the table apacheApsVar. The baseline characteristics such as age, gender, ethnicity weight and height were collected from the tables of patient and apachePatientResult. The laboratory indices, for total cholesterol, triglycerides, HDLc, LDLc and serum creatinine were extracted from the table lab. The comorbidities, for chronic obstructive pulmonary disease (COPD) and pneumonia were extracted from the table diagnosis.

Outcomes
The outcome of the study was all-cause mortality within 30 days after the date of visit to the ICU. In supplementary analyses, we also analyzed 60-days mortality.

Statistical analysis
Continuous variables are described as means ± SD or medians (interquartile ranges) and categorical data are presented as number and percentage. The difference according to the tertiles of the nonHDLc/HDLc ratio was compared using one-way analysis of variance (ANOVA) for continuous data and Chi-squared tests for categorical variables.
We applied a generalized additive model (GAM) to investigate dose-response relationships between the nonHDLc/HDLc ratio and 30-day mortality (Fig. 1). We applied logistic-regression model to estimate the association between the nonHDLc/HDLc ratio and 30-day mortality. The results were presented as odds ratios (ORs) with their 95% con dence intervals (95% CIs). Crude regression estimates are presented, as well as estimates adjusted for covariates. We selected these confounders on the basis of their associations with the outcomes of interest or a change in effect estimate of more than 10%. [10] Adjusted for the following potential confounders: age, sex, ethnicity, serum creatinine, systolic blood pressure, pneumonia and chronic obstructive pulmonary disease.
We further applied two-piece-wise linear regression model to examine the threshold effect of the nonHDLc/HDLc ratio on mortality ( Table 3). The turning point of the nonHDLc/HDLc ratio was determined using "exploratory" analyses, which is to move the trial turning point along the pre-de ned interval and pick up the one which gave maximum model likelihood. We also conducted log likelihood ratio test comparing one-line linear regression model with two-piece-wise linear model. As described in previous analyses [11,12]. Adjusted for age, sex, ethnicity, serum creatinine, systolic blood pressure, pneumonia and chronic obstructive pulmonary disease; CI, con dence interval; OR, odds ratio.
To examine the robustness of our results, we conducted strati ed analyses according to covariates. Dummy variables were used to indicate missing covariate values. The two-sided alpha level was set at 0.05. All the statistical analysis was performed using the EmpowerStats (www.empowerstats.com, X&Y solutions, Inc. Boston MA) and R software version 3.6.1 (http://www.r-project.org).

Baseline characteristics
A total of 1680 patients with sepsis were included in the study. The median age of all patients was 67 years (IQR 57-77 years). 816 patients (48.6%) were female. Table 1 compares the demographic, vital signs, laboratory and comorbiditise of patients by tertiles of the nonHDLc/HDLc ratio. Compared with subjects in the lowest tertile of the nonHDLc/HDLc ratio, those in the highest tertile were younger, had higher body mass index (BMI) and heart rate (all P < 0.05).

30-Day Mortality
There were 115 (6.85%) patients died within 30 days after the date of visit to the Emergency department.
Using the generalized additive model, the non-linear association between nonHDLc/HDLc ratio and 30day mortality was detected ( Table 3). The linear regression model and a two-piece-wise linear regression model were compared, and the P value for the log-likelihood ratio test is < 0.001. This result demonstrates that the two-piece-wise linear regression model should be used to t the model. Table 4, strati ed analyses were conducted with strati cation by age, sex, ethnicity, serum creatinine, COPD, pneumonia, systolic BP, diastolic BP and heart rate. These analyses both show the same trend as the main analysis. We performed sensitivity analyses to test the robustness of our results. Dummy variables were used to indicate missing covariate values. Similar results were obtained after considering the impact of missing data, data not shown.

Discussion
This study showed that higher nonHDLc/HDLc ratio, even at a low level, was associated with a higher risk of 30-day mortality among patients with sepsis in ICU. The major nding was that the probability of mortality decreased when the nonHDLc/HDLc ratio lower than the turning point (< 3.58) with a adjusted OR of 0.75 (95% CI: 0.61-0.94, P = 0.011) for every 1 increment of nonHDLc/HDLc ratio. To our knowledge, this is the rst study to report the relation between the nonHDLc/HDLc ratio and 30-day mortality in adult ICU patients with sepsis.
Most studies investigating the relation between levels of LDL-C and the risk of all cause mortality have found no association [13][14][15] or an inverse association [16][17][18]. You et al. [19] included 356 patients with intracranial hemorrhage (mean follow-up = 0.22 years) and found that the LDL-C/HDL-C ratio was negatively correlated with all-cause mortality, they suggested that the LDL-C/HDL-C ratio should be controlled above 2.96. Liu et al. [20] recruited 3250 stroke patients (mean follow-up = 1.00 years) and found a negative relationship between the LDL-C/HDL-C ratio and all-cause mortality. The ndings of the above studies suggest that the relationship between the LDL-C/HDL-C ratio and all-cause mortality may be negative, and the proposed optimal range of the LDL-C/HDL-C ratio has been inconsistent. These con icting results can be attributed to differences in the study populations, follow-up durations, and endpoint events. Consequently, the relationship between the nonHDLc/HDLc ratio and all-cause mortality are still unclear, which prompted us to conduct the current study.
We used non-HDL cholesterol rather than LDL cholesterol because in our database had measured TC and HDL cholesterol, from which non-HDL cholesterol can be calculated by subtraction. LDL cholesterol was directly measured in only 72% of patients. Further, the most commonly used estimation method, i.e. the Friedewald equation, can be inaccurate [21]. That non-HDL and LDL cholesterol were correlated in studies with data on both variables (r = 0.93) [4]. Non-HDL cholesterol predicts CHD risk at least as well as LDL cholesterol [22] because it includes cholesterol in LDL, lipoprotein(a), intermediate-density lipoprotein, very-low-density lipoprotein and lipoprotein remnants, and is thus a simple measure of cholesterol content within all atherogenic lipoproteins.
Also, a recent study in a Chinese Hypertension Registry study of 6941 hypertensive patients aged 65 years or older who were not treated with lipid-lowering drugs, during a median follow-up of 1.72 years, 157 all-cause deaths occurred, a U-shaped relationship between the LDL-C/HDL-C ratio and all-cause mortality was found. The optimal range of the LDL-C/HDL-C ratio was 1.67-2.10 [8], similar to our results that mortality in relation to the nonHDLc/HDLc ratio was U-shaped.
Possible explanations for our ndings. The association between low levels of nonHDLc and an increased risk of all cause mortality could be explained by reverse causation. Debilitation and illness have been hypothesised to cause a decrease in levels of cholesterol [23,24]. Nonetheless, cholesterol-related risk is more complex, and involves the interplay of several factors such as cholesterol particle concentration, reverse cholesterol transport and triglyceride-rich lipoproteins, to mention a few [25]. The U shaped association between the nonHDLc/HDLc ratio and mortality might be similar to the obesity paradox, which is largely explained by methodological issues, including reverse causation [26].
The current study is based on a ICU data, could be important for understanding what is a "normal and healthy" the nonHDLc/HDLc ratio in patients with sepsis (that is, when the focus is not limited to myocardial infarction and atherosclerotic cardiovascular disease). Explored the threshold the nonHDLc/HDLc ratio level where risk of death signi cantly increases is a high priority in patients with sepsis.

Study limitations
One limitation of this study is inherent to the observational nature of the study design which lends itself subject to limitations that should be considered including confounding by indication. Our ndings are hypothesis-generating and do not imply causality. In our analysis, we adjusted for likely confounders, including age, sex, ethnicity, serum creatinine, systolic blood pressure, pneumonia and COPD. Despite this adjustment, it is still possible that some amount of unmeasured confounding remains. Additional limitations of our study include missing data for some variables. Nonetheless, we used contemporary methods to deal with missing data to minimize bias.
Another limitation relates to the fact that the diagnoses were based on the ICD-9 coding which the responsible physician found relevant, and we did not have information concerning causes of death. Since we are examining mortality over a short period after the date of visit to the ICU, we did not nd it bene cial to distinguish between cardiovascular and non-cardiovascular death.
Furthermore, we lacked information about interventions during the initial stabilization, which may have in uencedthe nonHDLc/HDLc ratio levels and survival. It is noteworthy that the potential resulting from interventions would bias toward to the null and thus result in an underestimation of the association betweenthe nonHDLc/HDLc ratio level and mortality.
Finally, we also acknowledge that as our participants were patients referred for any reason to the emergency department, which limits the generalizability of the ndings to other population.

Conclusion
Using data from a retrospective cohort study, we identi ed 1680 patients with sepsis referred for any reason to the ICU. This study identi es a nonlinear dose-response relationship between the nonHDLc/HDLc ratio and 30-day mortality. The probability of mortality rose rapidly when the nonHDLc/HDLc ratio higher than the turning point (may at 3.58).