Dyslipidemia in the Young: From Genotype to Treatment Balder, Jan-Willem
IMPORTANT NOTE: You are advised to consult the publisher's version (publisher's PDF) if you wish to cite from it. Please check the document version below.
Document Version
Publisher's PDF, also known as Version of record
Publication date:
2018
Link to publication in University of Groningen/UMCG research database
Citation for published version (APA):
Balder, J-W. (2018). Dyslipidemia in the Young: From Genotype to Treatment. Rijksuniversiteit Groningen.
Copyright
Other than for strictly personal use, it is not permitted to download or to forward/distribute the text or part of it without the consent of the author(s) and/or copyright holder(s), unless the work is under an open content license (like Creative Commons).
Take-down policy
If you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediately and investigate your claim.
Downloaded from the University of Groningen/UMCG research database (Pure): http://www.rug.nl/research/portal. For technical reasons the number of authors shown on this cover page is limited to 10 maximum.
Download date: 29-06-2021
6
J.K. de Vries
1, J.W. Balder
1,2, M.J. Pena
3, P. Denig
3, A.J. Smit
1*Non-LDL Dyslipidemia is Prevalent in the Young, and Determined by Lifestyle Factors and Age: the Lifelines Cohort
1. Department of Vascular Medicine, University Medical Center Groningen, University of Groningen, Groningen, the Netherlands.
2. Department of Pediatrics, Section Molecular Genetics, University Medical Center Groningen, University of Groningen, Groningen, the Netherlands.
3. Department of Clinical Pharmacy and Pharmacology, University Medical Center Groningen, University of Groningen, Groningen, the Netherlands.
*Corresponding author: a.j.smit@umcg.nl
Submitted to Atherosclerosis J.K. de Vries
1, J.W. Balder
1,2, M.J. Pena
3, P. Denig
3, A.J. Smit
1*6
Non-LDL Dyslipidemia is Prevalent in the Young, and Determined by Lifestyle Factors and Age: the Lifelines Cohort
1. Department of Vascular Medicine, University Medical Center Groningen, University of Groningen, Groningen, the Netherlands.
2. Department of Pediatrics, Section Molecular Genetics, University Medical Center Groningen, University of Groningen, Groningen, the Netherlands.
3. Department of Clinical Pharmacy and Pharmacology, University Medical Center Groningen, University of Groningen, Groningen, the Netherlands.
*Corresponding author: a.j.smit@umcg.nl
Submitted to Atherosclerosis
94
Abstract
Background
Non-LDL dyslipidemia (NLD) confers cardiovascular risk and prevalence rates appear to be high in elderly populations. Small cohorts have identified several lifestyle, anthropometric, and medical factors associated with NLD.
Aims
To assess gender- and age-specific prevalence of NLD in a contemporary population cohort (n = 167,729) and to identify independent determinants of NLD, focusing on lifestyle, anthropometric, and medical factors.
Methods and results
The prevalence of NLD was assessed per 10-year age interval in adults without cardiovascular disease and without using lipid-modifying drugs from the Dutch Lifelines cohort. NLD was defined as low HDL-cholesterol, high triglycerides or high remnant cholesterol as per guideline cutoff values. Multivariable regression was used to identify factors independently associated with NLD. Determinants included age, smoking, alcohol use, physical activity, diet, BMI, diabetes mellitus (DM), chronic kidney disease, and in women, menopausal state and oral contraceptive use.
NLD occurred in 15 – 19% of women and 13 – 30% of men in this cohort, with the highest prevalence of 30% in 35 – 55 year old men. In most age groups, the prevalence in women was lower than in men. Obesity (both genders: odds ratio [OR] 5.3, 95%
confidence interval [95% CI] 5.0 – 5.7), current smoking (men: OR 1.8, 95% CI 1.7 – 1.9;
women: OR 2.2, 95% CI 2.1 – 2.3), and DM (men: OR 2.2, 95% CI 1.8 – 2.6; women: OR 2.7, 95% CI 2.3 – 3.1) were strongly associated with NLD.
Conclusions
NLD already occurs frequently at an early age. Modifiable lifestyle choices, obesity, and DM were strong determinants of NLD. Public health efforts could substantially contribute to decreasing NLD.
94
Abstract
Background
Non-LDL dyslipidemia (NLD) confers cardiovascular risk and prevalence rates appear to be high in elderly populations. Small cohorts have identified several lifestyle, anthropometric, and medical factors associated with NLD.
Aims
To assess gender- and age-specific prevalence of NLD in a contemporary population cohort (n = 167,729) and to identify independent determinants of NLD, focusing on lifestyle, anthropometric, and medical factors.
Methods and results
The prevalence of NLD was assessed per 10-year age interval in adults without cardiovascular disease and without using lipid-modifying drugs from the Dutch Lifelines cohort. NLD was defined as low HDL-cholesterol, high triglycerides or high remnant cholesterol as per guideline cutoff values. Multivariable regression was used to identify factors independently associated with NLD. Determinants included age, smoking, alcohol use, physical activity, diet, BMI, diabetes mellitus (DM), chronic kidney disease, and in women, menopausal state and oral contraceptive use.
NLD occurred in 15 – 19% of women and 13 – 30% of men in this cohort, with the highest prevalence of 30% in 35 – 55 year old men. In most age groups, the prevalence in women was lower than in men. Obesity (both genders: odds ratio [OR] 5.3, 95%
confidence interval [95% CI] 5.0 – 5.7), current smoking (men: OR 1.8, 95% CI 1.7 – 1.9;
women: OR 2.2, 95% CI 2.1 – 2.3), and DM (men: OR 2.2, 95% CI 1.8 – 2.6; women: OR 2.7, 95% CI 2.3 – 3.1) were strongly associated with NLD.
Conclusions
NLD already occurs frequently at an early age. Modifiable lifestyle choices, obesity, and DM were strong determinants of NLD. Public health efforts could substantially contribute to decreasing NLD.
6
94 95
Introduction
Development of atherosclerotic cardiovascular disease (CVD) is strongly determined by modifiable risk factors such as dyslipidemia.
4Besides elevated total cholesterol (TC) and low-density lipoprotein cholesterol (LDL-c), non-LDL dyslipidemia, including low levels of high-density lipoprotein cholesterol (HDL-c), elevated triglycerides (TG), or high remnant cholesterol (RC), is associated with increased risk of developing atherosclerotic CVD.
40, 137-139
Current guidelines provide ample recommendations for LDL-c lowering therapy in appropriate subgroups, whereas treatment of non-LDL dyslipidemia is more controversial.
40Reported prevalence rates of individual components of non-LDL dyslipidemia range between 22% and 74% in various – often small and elderly – cohorts,
140, 141but little is known about the overall prevalence rate in the general population.
Many adverse lifestyle factors are already and increasingly prevalent in young adults, increasing the burden of non-LDL dyslipidemia and adding to the development of atherosclerotic disease.
142, 143Because early recognition and treatment of non-LDL dyslipidemia and its determining factors could contribute to reducing the burden of CVD, it is of importance to ascertain the prevalence of non-LDL dyslipidemia and its determining factors in young and healthy people.
143Whereas elevated LDL-c has a stronger genetic component compared to non- LDL dyslipidemia, lifestyle choices and anthropometric and medical factors are more strongly associated with non-LDL dyslipidemia compared to LDL-c.
144Factors that have been identified as potentially associated with non-LDL dyslipidemia include age, gender, obesity, smoking, alcohol use, physical activity, dietary factors, menopausal state and oral contraceptive use for women, and a range of medical conditions like diabetes mellitus (DM) and chronic kidney disease (CKD).
137, 145-155Most of these studies, however, have looked at the associations between single factors or a small subset of these factors and individual components of non-LDL dyslipidemia. Little is known about the relative contribution of lifestyle, anthropometric, and medical factors on non-LDL dyslipidemia.
Our aim is to describe gender- and age-specific prevalence rates of non-LDL dyslipidemia in a large, contemporary population cohort without known atherosclerotic disease, and identify determinants of non-LDL dyslipidemia. We hypothesize that lifestyle, anthropometric, and medical factors are independently associated with non-LDL dyslipidemia.
6
94 95
Introduction
Development of atherosclerotic cardiovascular disease (CVD) is strongly determined by modifiable risk factors such as dyslipidemia.
4Besides elevated total cholesterol (TC) and low-density lipoprotein cholesterol (LDL-c), non-LDL dyslipidemia, including low levels of high-density lipoprotein cholesterol (HDL-c), elevated triglycerides (TG), or high remnant cholesterol (RC), is associated with increased risk of developing atherosclerotic CVD.
40, 137-139
Current guidelines provide ample recommendations for LDL-c lowering therapy in appropriate subgroups, whereas treatment of non-LDL dyslipidemia is more controversial.
40Reported prevalence rates of individual components of non-LDL dyslipidemia range between 22% and 74% in various – often small and elderly – cohorts,
140, 141but little is known about the overall prevalence rate in the general population.
Many adverse lifestyle factors are already and increasingly prevalent in young adults, increasing the burden of non-LDL dyslipidemia and adding to the development of atherosclerotic disease.
142, 143Because early recognition and treatment of non-LDL dyslipidemia and its determining factors could contribute to reducing the burden of CVD, it is of importance to ascertain the prevalence of non-LDL dyslipidemia and its determining factors in young and healthy people.
143Whereas elevated LDL-c has a stronger genetic component compared to non- LDL dyslipidemia, lifestyle choices and anthropometric and medical factors are more strongly associated with non-LDL dyslipidemia compared to LDL-c.
144Factors that have been identified as potentially associated with non-LDL dyslipidemia include age, gender, obesity, smoking, alcohol use, physical activity, dietary factors, menopausal state and oral contraceptive use for women, and a range of medical conditions like diabetes mellitus (DM) and chronic kidney disease (CKD).
137, 145-155Most of these studies, however, have looked at the associations between single factors or a small subset of these factors and individual components of non-LDL dyslipidemia. Little is known about the relative contribution of lifestyle, anthropometric, and medical factors on non-LDL dyslipidemia.
Our aim is to describe gender- and age-specific prevalence rates of non-LDL dyslipidemia in a large, contemporary population cohort without known atherosclerotic disease, and identify determinants of non-LDL dyslipidemia. We hypothesize that lifestyle, anthropometric, and medical factors are independently associated with non-LDL dyslipidemia.
96
Material and methods
Participants and data collection
We included adult participants from the population-based Lifelines cohort. Lifelines is a large, prospective cohort study with a three-generational design, examining the health and health-related behaviors in a representative sample of 167,729 persons of the north Netherlands population. Individuals aged 25 – 50 years old were invited to participate and, on inclusion, the participant’s partner, parents, and children were also invited to participate.
Following this design, young adults were oversampled. Recruitment lasted from 2006 – 2013. Detailed information has been published previously, and all participants provided written informed consent.
72For our study, adult participants with known atherosclerotic CVD (or when data on CVD were missing), missing lipid levels, or use of drugs that are known to alter serum lipid levels were excluded (Figure 1). Data were collected through questionnaires on: demographics;
health status, using the European Community Respiratory Health Survey (ECRHS II); diet, using a Food Frequency Questionnaire (FFQ) specifically developed for Lifelines; and on physical activity, using the Short Questionnaire to Assess Health-Enhancing (SQUASH).
72During the baseline visit height, weight and blood pressure were measured. All prescribed and over-the-counter medication was brought to the study center and classified according to the Anatomical Therapeutic Chemical (ATC) classification system by a study nurse. Blood and urine were sampled for routine chemistry panels. TC, LDL-c, and HDL-c were measured with a direct assay and TG using an enzymatic colorimetric test, and creatinine using an IDMS-traceable enzymatic method (all on Roche, Modular P, Mannheim, Germany). HbA1c was measured using a turbidimetric assay (Roche Diagnostics Nederland BV, Almere, the Netherlands).
Primary outcome: definition of non-LDL dyslipidemia and its constituents
We defined non-LDL dyslipidemia as a composite of low HDL-c (< 1.0 mmol/l for men and
< 1.2 mmol/l for women), high TG (fasting TG > 1.70 mmol/l), or high RC (calculated by subtracting LDL-c and HDL-c from TC; defined as RC > 0.7 mmol/l).
40, 137Non-fasting TG measurements were excluded from analysis. These are cutoff points used to aid in clinical decisions to initiate treatment, and we consider non-LDL dyslipidemia clinically relevant when any of the components exceed the aforementioned cutoff levels.
40Determinants
We included age, alcohol use, smoking, physical activity, dietary intake of fruit, vegetables, fish, and meat, BMI, DM, CKD, and for women, oral contraceptive use and menopausal state as determinants of non-LDL dyslipidemia. Alcohol use was self-reported, and participants
96
Material and methods
Participants and data collection
We included adult participants from the population-based Lifelines cohort. Lifelines is a large, prospective cohort study with a three-generational design, examining the health and health-related behaviors in a representative sample of 167,729 persons of the north Netherlands population. Individuals aged 25 – 50 years old were invited to participate and, on inclusion, the participant’s partner, parents, and children were also invited to participate.
Following this design, young adults were oversampled. Recruitment lasted from 2006 – 2013. Detailed information has been published previously, and all participants provided written informed consent.
72For our study, adult participants with known atherosclerotic CVD (or when data on CVD were missing), missing lipid levels, or use of drugs that are known to alter serum lipid levels were excluded (Figure 1). Data were collected through questionnaires on: demographics;
health status, using the European Community Respiratory Health Survey (ECRHS II); diet, using a Food Frequency Questionnaire (FFQ) specifically developed for Lifelines; and on physical activity, using the Short Questionnaire to Assess Health-Enhancing (SQUASH).
72During the baseline visit height, weight and blood pressure were measured. All prescribed and over-the-counter medication was brought to the study center and classified according to the Anatomical Therapeutic Chemical (ATC) classification system by a study nurse. Blood and urine were sampled for routine chemistry panels. TC, LDL-c, and HDL-c were measured with a direct assay and TG using an enzymatic colorimetric test, and creatinine using an IDMS-traceable enzymatic method (all on Roche, Modular P, Mannheim, Germany). HbA1c was measured using a turbidimetric assay (Roche Diagnostics Nederland BV, Almere, the Netherlands).
Primary outcome: definition of non-LDL dyslipidemia and its constituents
We defined non-LDL dyslipidemia as a composite of low HDL-c (< 1.0 mmol/l for men and
< 1.2 mmol/l for women), high TG (fasting TG > 1.70 mmol/l), or high RC (calculated by subtracting LDL-c and HDL-c from TC; defined as RC > 0.7 mmol/l).
40, 137Non-fasting TG measurements were excluded from analysis. These are cutoff points used to aid in clinical decisions to initiate treatment, and we consider non-LDL dyslipidemia clinically relevant when any of the components exceed the aforementioned cutoff levels.
40Determinants
We included age, alcohol use, smoking, physical activity, dietary intake of fruit, vegetables, fish, and meat, BMI, DM, CKD, and for women, oral contraceptive use and menopausal state as determinants of non-LDL dyslipidemia. Alcohol use was self-reported, and participants
6
96 97
were classified as abstainers (0 drinks/day), light drinkers (< 1 drink/day), moderate drinkers (1 to 2 drinks/day), or heavy drinkers (> 2 drinks/day). Smoking habits were self-reported;
participants were defined as non-smokers if they had never smoked and were not currently smoking; former smokers were defined as reporting smoking for at least a year, had stopped smoking, and were not currently smoking; and current smokers were defined as reporting tobacco use in the previous month or reported smoking for at least a year and did not report having quitted smoking. Sufficient physical activity was defined as self-reported moderate activity for ≥ 30 minutes/day ≥ 5 days a week, sedentary lifestyle was defined as no days per week with ≥ 30 minutes of moderate physical activity, and the remainder were defined as insufficient physical activity. Intake of fruit, fish, meat (all in servings/week), and vegetables (serving spoons/week) was self-reported and modeled as continuous variables.
Normal BMI (weight divided by height squared) was defined as BMI < 25 kg/m
2, overweight BMI 25 – 30 kg/m
2, and obese BMI ≥ 30 kg/m
2.
DM was defined as self-reported, using glucose-lowering drugs, or HbA1c ≥ 6.5%
(48 mmol/mol). CKD was defined as eGFR < 60 ml/min, calculated using the CKD-epi formula.
156In women, the use of oral contraception or hormone replacement therapy was defined as self-reported use in the month before the baseline study visit. Menopausal state was defined as self-reported menopause, or when the menstrual cycle was reported absent and the last menstruation was > 1 year before the baseline visit, or when this information was missing and the age was ≥ 63 years (equals 3 standard deviations above the mean age of menopause in the Netherlands).
157Statistical analysis
Baseline characteristics were stratified by gender and presented as mean (± standard deviation [SD]) for normally distributed continuous variables, median (25
th– 75
thpercentile) for skewed variables, or proportions (n, %) for categorical variables. Proportions of non- LDL-dyslipidemia were calculated per 10-year age interval. Multiple imputations using the iterative Markov chain Monte Carlo method were performed for missing data on smoking, alcohol use, dietary intake, physical activity, and menopause (maximum 10% missing data).
Ten imputed datasets were created, and the pooled results were used for further analysis.
Multivariable logistic regression analysis was used to explore independent associations between non-LDL dyslipidemia and lifestyle, anthropometric, and medical covariates.
Analyses were stratified by gender. Statistically significant associations with P-value
< 0.05 in univariate analyses were included in multivariable regression analysis. Strength of associations is reported with odds ratios (OR) and 95% confidence intervals (95% CI).
Subsequently, four regression models were created. The first model only included age groups in 10-year intervals as determinant of non-LDL dyslipidemia. Consecutively, lifestyle factors (smoking, alcohol use, physical activity and dietary factors), BMI, and medical factors
6
96 97
were classified as abstainers (0 drinks/day), light drinkers (< 1 drink/day), moderate drinkers (1 to 2 drinks/day), or heavy drinkers (> 2 drinks/day). Smoking habits were self-reported;
participants were defined as non-smokers if they had never smoked and were not currently smoking; former smokers were defined as reporting smoking for at least a year, had stopped smoking, and were not currently smoking; and current smokers were defined as reporting tobacco use in the previous month or reported smoking for at least a year and did not report having quitted smoking. Sufficient physical activity was defined as self-reported moderate activity for ≥ 30 minutes/day ≥ 5 days a week, sedentary lifestyle was defined as no days per week with ≥ 30 minutes of moderate physical activity, and the remainder were defined as insufficient physical activity. Intake of fruit, fish, meat (all in servings/week), and vegetables (serving spoons/week) was self-reported and modeled as continuous variables.
Normal BMI (weight divided by height squared) was defined as BMI < 25 kg/m
2, overweight BMI 25 – 30 kg/m
2, and obese BMI ≥ 30 kg/m
2.
DM was defined as self-reported, using glucose-lowering drugs, or HbA1c ≥ 6.5%
(48 mmol/mol). CKD was defined as eGFR < 60 ml/min, calculated using the CKD-epi formula.
156In women, the use of oral contraception or hormone replacement therapy was defined as self-reported use in the month before the baseline study visit. Menopausal state was defined as self-reported menopause, or when the menstrual cycle was reported absent and the last menstruation was > 1 year before the baseline visit, or when this information was missing and the age was ≥ 63 years (equals 3 standard deviations above the mean age of menopause in the Netherlands).
157Statistical analysis
Baseline characteristics were stratified by gender and presented as mean (± standard deviation [SD]) for normally distributed continuous variables, median (25
th– 75
thpercentile) for skewed variables, or proportions (n, %) for categorical variables. Proportions of non- LDL-dyslipidemia were calculated per 10-year age interval. Multiple imputations using the iterative Markov chain Monte Carlo method were performed for missing data on smoking, alcohol use, dietary intake, physical activity, and menopause (maximum 10% missing data).
Ten imputed datasets were created, and the pooled results were used for further analysis.
Multivariable logistic regression analysis was used to explore independent associations between non-LDL dyslipidemia and lifestyle, anthropometric, and medical covariates.
Analyses were stratified by gender. Statistically significant associations with P-value
< 0.05 in univariate analyses were included in multivariable regression analysis. Strength of associations is reported with odds ratios (OR) and 95% confidence intervals (95% CI).
Subsequently, four regression models were created. The first model only included age groups in 10-year intervals as determinant of non-LDL dyslipidemia. Consecutively, lifestyle factors (smoking, alcohol use, physical activity and dietary factors), BMI, and medical factors
98
(DM, CKD and for women menopausal state and oral contraceptive use) were added. All possible sequences were analyzed. Receiver operating curves (ROC) were created for each model, and the increase in area under the curve (AUC) was used to assess the additional value of each subsequent step. Differences in AUC were assessed with the method according to Hanley and McNeil.
158As a sensitivity analysis, the regression analyses were also performed in the non-imputed database.
Statistical analyses were performed with SPSS version 22 (IBM, Armonk, New York, USA), two-tailed P-values < 0.05 were considered statistically significant.
Figure 1. Flow chart of inclusion of participants for this analysis from the LifeLines cohort.
Results
A total of 133,721 adult participants met our inclusion criteria and were included in this analysis (Figure 1). The median age was 44 years (25
th– 75
thpercentile: 35 – 51 years); women were slightly overrepresented (59%) (see Table 1).
98
(DM, CKD and for women menopausal state and oral contraceptive use) were added. All possible sequences were analyzed. Receiver operating curves (ROC) were created for each model, and the increase in area under the curve (AUC) was used to assess the additional value of each subsequent step. Differences in AUC were assessed with the method according to Hanley and McNeil.
158As a sensitivity analysis, the regression analyses were also performed in the non-imputed database.
Statistical analyses were performed with SPSS version 22 (IBM, Armonk, New York, USA), two-tailed P-values < 0.05 were considered statistically significant.
Figure 1. Flow chart of inclusion of participants for this analysis from the LifeLines cohort.
Results
A total of 133,721 adult participants met our inclusion criteria and were included in this analysis (Figure 1). The median age was 44 years (25
th– 75
thpercentile: 35 – 51 years); women were slightly overrepresented (59%) (see Table 1).
6
98 99
Prevalence of non-LDL dyslipidemia and its components in men
In men, non-LDL dyslipidemia was found in approximately 27%, with a marked influence of age (Table 2 and Figure 2). The prevalence of non-LDL dyslipidemia rose from 13% in the youngest age group (18 – 25 years old) to approximately 30% in 35 – 55 year old, and declined to approximately 20% after the age of 65 years. The composition of non-LDL dyslipidemia remained relatively stable across the age groups (Figure 3). There were only minor changes in the relative proportion of low HDL-c, declining from 30% in the youngest age group to approximately 20% from the age of 35 years, and in the proportion of high RC increasing from 25% in the youngest to slightly more than 30% starting at the age of 35 years.
Factors contributing to non-LDL dyslipidemia in men
In men, 2.5 – 5.5% of participants had missing data on lifestyle factors, except for alcohol use, where 10.6% had missing data. After multiple imputation, alcohol use was associated with a lower chance of non-LDL dyslipidemia with lowest OR for moderate use of alcohol (OR 0.8, 95% CI 0.7 – 0.8); smoking (OR 1.8, 95% CI 1.7 – 1.9, for current smoking vs. never smoked) and physical inactivity (OR 1.4, 95% CI 1.3 – 1.6, for sedentary lifestyle vs. sufficiently active) were associated with an increased chance of non-LDL dyslipidemia. Intake of fruit, vegetables, and fish, modeled as the number of servings per week, were significantly associated with lower chance of non-LDL dyslipidemia; no significant association with non-LDL dyslipidemia was found for the use of meat. Obesity was associated with increased chance of non-LDL dyslipidemia (OR 5.3, 95% CI 5.0 – 5.7, for obesity vs. normal weight), as was the presence of DM and CKD (OR 2.2, 95% CI 1.8 – 2.6, and OR 1.9, 95% CI 1.5 – 2.4, respectively) (Table 3).
Analysis of the non-imputed cohort resulted in slightly less conservative point estimates, but not significantly different compared to the analysis in the imputed cohort (Supplementary Table 2).
The ROC curves show an AUC of 0.52 for the association of age groups with non-LDL dyslipidemia in men (Table 4). Adding lifestyle factors, anthropometrics, and medical factors on top of age increased the AUC to 0.62, 0.69, and 0.70 respectively. Each additional step was statistically significant. Rotating the sequence in which determinants were added to the model did change the relative contribution of each set of determinants, however, BMI and lifestyle factors increased the AUC in every sequence more than medical factors (Supplementary Table 1).
6
98 99
Prevalence of non-LDL dyslipidemia and its components in men
In men, non-LDL dyslipidemia was found in approximately 27%, with a marked influence of age (Table 2 and Figure 2). The prevalence of non-LDL dyslipidemia rose from 13% in the youngest age group (18 – 25 years old) to approximately 30% in 35 – 55 year old, and declined to approximately 20% after the age of 65 years. The composition of non-LDL dyslipidemia remained relatively stable across the age groups (Figure 3). There were only minor changes in the relative proportion of low HDL-c, declining from 30% in the youngest age group to approximately 20% from the age of 35 years, and in the proportion of high RC increasing from 25% in the youngest to slightly more than 30% starting at the age of 35 years.
Factors contributing to non-LDL dyslipidemia in men
In men, 2.5 – 5.5% of participants had missing data on lifestyle factors, except for alcohol use, where 10.6% had missing data. After multiple imputation, alcohol use was associated with a lower chance of non-LDL dyslipidemia with lowest OR for moderate use of alcohol (OR 0.8, 95% CI 0.7 – 0.8); smoking (OR 1.8, 95% CI 1.7 – 1.9, for current smoking vs. never smoked) and physical inactivity (OR 1.4, 95% CI 1.3 – 1.6, for sedentary lifestyle vs. sufficiently active) were associated with an increased chance of non-LDL dyslipidemia. Intake of fruit, vegetables, and fish, modeled as the number of servings per week, were significantly associated with lower chance of non-LDL dyslipidemia; no significant association with non-LDL dyslipidemia was found for the use of meat. Obesity was associated with increased chance of non-LDL dyslipidemia (OR 5.3, 95% CI 5.0 – 5.7, for obesity vs. normal weight), as was the presence of DM and CKD (OR 2.2, 95% CI 1.8 – 2.6, and OR 1.9, 95% CI 1.5 – 2.4, respectively) (Table 3).
Analysis of the non-imputed cohort resulted in slightly less conservative point estimates, but not significantly different compared to the analysis in the imputed cohort (Supplementary Table 2).
The ROC curves show an AUC of 0.52 for the association of age groups with non-LDL dyslipidemia in men (Table 4). Adding lifestyle factors, anthropometrics, and medical factors on top of age increased the AUC to 0.62, 0.69, and 0.70 respectively. Each additional step was statistically significant. Rotating the sequence in which determinants were added to the model did change the relative contribution of each set of determinants, however, BMI and lifestyle factors increased the AUC in every sequence more than medical factors (Supplementary Table 1).
100
Table 1. Baseline characteristics stratified by gender.
Men (n = 54,198) Women (n = 79,523)
Age (median, IQR) 44 (35 – 51) 44 (35 – 50)
DM (n, % yes) 1,012 (1.9%) 1,346 (1.7%)
HbA1c (%, mean, SD) 5.51 (0.41) 5.49 (0.39)
Hypertension (n, % yes) 11,617 (21.4%) 9,061 (11.4%)
Antihypertensive medication (n, % yes) 4,321 (8.0%) 7,438 (9.4%)
SBP / DBP (mean, SD) 130/76 (14/9) 122/72 (15/9)
CKD (n, % yes) 408 (0.8%) 664 (0.8%)
eGFR (ml/min/1.73 m2, mean, SD) 98.5 (14.7) 96.7 (15.0)
Menopause (n,% yes) n/a 15,455 (21.7%)
Oral contraceptive use (n,% yes) n/a 27,753 (34.9%)
BMI (kg/m2, mean, SD) 26.2 (3.6) 25.7 (4.6)
BMI-group
normal (n,%) 21,319 (39.3%) 41,073 (51.7%)
overweight (n,%) 25,678 (47.4%) 26,068 (32.8%)
obese (n,%) 7,185 (13.3%) 12,360 (15.5%)
Smoking
never (n,%) 23,428 (45.2%) 37,423 (49.3%)
former (n,%) 15,632 (30.2%) 22,768 (30.0%)
current (n,%) 12,623 (24.4%) 15,740 (20.7%)
Alcohol
none (n,%) 5,096 (10.5%) 21,002 (27.9%)
light (n,%) 21,738 (44.9%) 40,953 (54.4%)
moderate (n,%) 14,027 (28.9%) 11,073 (14.7%)
heavy (n,%) 7,597 (15.7%) 2,262 (3.0%)
Physical activity
sufficient (n,%) 24,342 (47.5%) 36,508 (48.8%)
insufficient (n,%) 24,796 (48.4%) 33,831 (45.2%)
sedentary (n,%) 2,115 (4.1%) 4,442 (5.9%)
Fruit servings / week (median, IQR) 5.0 (2.5-13.0) 6.5 (2.5-13.0) Vegetable serving spoons / week (median, IQR) 10 (7.5-13.5) 13 (7.5-13.5)
Fish servings / week (median, IQR) 0.6 (0.3-1.0) 0.6 (0.3-1.0)
Meat servings / week (median, IQR) 4.5 (4.5-6.5) 4.5 (2.5-6.5)
Abbreviations: IQR, interquartile range; DM, diabetes mellitus; SD, standard deviation; SBP, systolic blood pressure; DBP, diastolic blood pressure; CKD, chronic kidney disease; eGFR, estimated glomerular filtration rate; BMI, body mass index; n/a, not applicable.
100
Table 1. Baseline characteristics stratified by gender.
Men (n = 54,198) Women (n = 79,523)
Age (median, IQR) 44 (35 – 51) 44 (35 – 50)
DM (n, % yes) 1,012 (1.9%) 1,346 (1.7%)
HbA1c (%, mean, SD) 5.51 (0.41) 5.49 (0.39)
Hypertension (n, % yes) 11,617 (21.4%) 9,061 (11.4%)
Antihypertensive medication (n, % yes) 4,321 (8.0%) 7,438 (9.4%)
SBP / DBP (mean, SD) 130/76 (14/9) 122/72 (15/9)
CKD (n, % yes) 408 (0.8%) 664 (0.8%)
eGFR (ml/min/1.73 m2, mean, SD) 98.5 (14.7) 96.7 (15.0)
Menopause (n,% yes) n/a 15,455 (21.7%)
Oral contraceptive use (n,% yes) n/a 27,753 (34.9%)
BMI (kg/m2, mean, SD) 26.2 (3.6) 25.7 (4.6)
BMI-group
normal (n,%) 21,319 (39.3%) 41,073 (51.7%)
overweight (n,%) 25,678 (47.4%) 26,068 (32.8%)
obese (n,%) 7,185 (13.3%) 12,360 (15.5%)
Smoking
never (n,%) 23,428 (45.2%) 37,423 (49.3%)
former (n,%) 15,632 (30.2%) 22,768 (30.0%)
current (n,%) 12,623 (24.4%) 15,740 (20.7%)
Alcohol
none (n,%) 5,096 (10.5%) 21,002 (27.9%)
light (n,%) 21,738 (44.9%) 40,953 (54.4%)
moderate (n,%) 14,027 (28.9%) 11,073 (14.7%)
heavy (n,%) 7,597 (15.7%) 2,262 (3.0%)
Physical activity
sufficient (n,%) 24,342 (47.5%) 36,508 (48.8%)
insufficient (n,%) 24,796 (48.4%) 33,831 (45.2%)
sedentary (n,%) 2,115 (4.1%) 4,442 (5.9%)
Fruit servings / week (median, IQR) 5.0 (2.5-13.0) 6.5 (2.5-13.0) Vegetable serving spoons / week (median, IQR) 10 (7.5-13.5) 13 (7.5-13.5)
Fish servings / week (median, IQR) 0.6 (0.3-1.0) 0.6 (0.3-1.0)
Meat servings / week (median, IQR) 4.5 (4.5-6.5) 4.5 (2.5-6.5)
Abbreviations: IQR, interquartile range; DM, diabetes mellitus; SD, standard deviation; SBP, systolic blood pressure; DBP, diastolic blood pressure; CKD, chronic kidney disease; eGFR, estimated glomerular filtration rate; BMI, body mass index; n/a, not applicable.
6
100 101
Table 2. Outcome of interest: percentages of non-LDL-dyslipidemia and its components, stratified by gender and age. Age-groups18 – 2525 – 3535 – 4545 – 5555 – 6565 – 75> 75all Men (n)3,43410,42816,33914,4796,3932,69942654,198 non-LDL-dyslipidemia (%)449 (13.1%)2,441 (23.4%)4,935 (30.2%)4,351 (30.1%)1,647 (25.8%)550 (20.4%)86 (20.2%)14,459 (26.7%) low HDL-c (%)224 (6.5%)1,091 (10.5%)1,882 (11.5%)1,328 (9.2%)486 (7.6%)154 (5.7%)28 (6.6%)5,193 (9.6%) high TG (%)286 (8.3%)1,839 (17.6%)4,092 (25.0%)3,765 (26.0%)1,379 (21.6%)448 (16.7%)65 (15.3%)11,874 (21.9%) high RC (%)141 (4.1%)1,138 (10.9%)2,700 (16.5%)2,463 (17.0%)849 (13.3%)273 (10.1%)37 (8.7%)7,601 (14.0%) HDL-c (mmol/l, SD)1.34 (0.29)1.30 (0.30)1.28 (0.31)1.33 (0.32)1.38 (0.34)1.41 (0.35)1.40 (0.36)1.32 (0.32) TG (mmol/l, SD)1.00 (0.64)1.25 (0.85)1.44 (1.07)1.47 (1.05)1.37 (0.85)1.26 (0.69)1.18 (0.54)1.36 (0.97) RC (mmol/l, SD)0.28 (0.25)0.38 (0.36)0.45 (0.45)0.46 (0.44)0.42 (0.37)0.38 (0.30)0.35 (0.25)0.42 (0.40) Age-groups18 – 2525 – 3535 – 4545 – 5555 – 6565 – 75> 75all Women (n)6,74214,28224,18821,1759,1593,40057779,523 non-LDL-dyslipidemia (%)1,138 (16.9%)2,371 (16.6%)3,578 (14.8%)3,263 (15.4%)1,624 (17.7%)642 (18.9%)97 (16.8%)12,713 (16.0%) low HDL-c (%)864 (12.8%)1,914 (13.4%)2,526 (10.4%)1,625 (7.7%)560 (6.1%)219 (6.4%)34 (5.9%)7,742 (9.7%) high TG (%)317 (4.7%)656 (4.6%)1,511 (6.2%)2,129 (10.1%)1,253 (13.7%)507 (14.9%)71 (12.3%)6,444 (8.1%) high RC (%)191 (2.8%)436 (3.1%)911 (3.8%)1,286 (6.1%)690 (7.5%)267 (7.9%)54 (9.4%)3835 (4.8%) HDL-c (mmol/l, SD)1.52 (0.35)1.53 (0.36)1.59 (0.37)1.68 (0.41)1.74 (0.43)1.72 (0.42)1.73 (0.43)1.62 (0.40) TG (mmol/l, SD)0.93 (0.41)0.89 (0.45)0.93 (0.55)1.05 (0.58)1.16 (0.62)1.21 (0.58)1.21 (0.57)1.00 (0.55) RC (mmol/l, SD)0.29 (0.20)0.26 (0.22)0.27 (0.25)0.31 (0.27)0.34 (0.28)0.35 (0.26)0.34 (0.27)0.29 (0.25) Abbreviations: LDL; low-density lipoprotein; HDL, high-density lipoprotein; TG, triglycerides; RC, remnant cholesterol; SD, standard deviation.
6
100 101
Table 2. Outcome of interest: percentages of non-LDL-dyslipidemia and its components, stratified by gender and age. Age-groups18 – 2525 – 3535 – 4545 – 5555 – 6565 – 75> 75all Men (n)3,43410,42816,33914,4796,3932,69942654,198 non-LDL-dyslipidemia (%)449 (13.1%)2,441 (23.4%)4,935 (30.2%)4,351 (30.1%)1,647 (25.8%)550 (20.4%)86 (20.2%)14,459 (26.7%) low HDL-c (%)224 (6.5%)1,091 (10.5%)1,882 (11.5%)1,328 (9.2%)486 (7.6%)154 (5.7%)28 (6.6%)5,193 (9.6%) high TG (%)286 (8.3%)1,839 (17.6%)4,092 (25.0%)3,765 (26.0%)1,379 (21.6%)448 (16.7%)65 (15.3%)11,874 (21.9%) high RC (%)141 (4.1%)1,138 (10.9%)2,700 (16.5%)2,463 (17.0%)849 (13.3%)273 (10.1%)37 (8.7%)7,601 (14.0%) HDL-c (mmol/l, SD)1.34 (0.29)1.30 (0.30)1.28 (0.31)1.33 (0.32)1.38 (0.34)1.41 (0.35)1.40 (0.36)1.32 (0.32) TG (mmol/l, SD)1.00 (0.64)1.25 (0.85)1.44 (1.07)1.47 (1.05)1.37 (0.85)1.26 (0.69)1.18 (0.54)1.36 (0.97) RC (mmol/l, SD)0.28 (0.25)0.38 (0.36)0.45 (0.45)0.46 (0.44)0.42 (0.37)0.38 (0.30)0.35 (0.25)0.42 (0.40) Age-groups18 – 2525 – 3535 – 4545 – 5555 – 6565 – 75> 75all Women (n)6,74214,28224,18821,1759,1593,40057779,523 non-LDL-dyslipidemia (%)1,138 (16.9%)2,371 (16.6%)3,578 (14.8%)3,263 (15.4%)1,624 (17.7%)642 (18.9%)97 (16.8%)12,713 (16.0%) low HDL-c (%)864 (12.8%)1,914 (13.4%)2,526 (10.4%)1,625 (7.7%)560 (6.1%)219 (6.4%)34 (5.9%)7,742 (9.7%) high TG (%)317 (4.7%)656 (4.6%)1,511 (6.2%)2,129 (10.1%)1,253 (13.7%)507 (14.9%)71 (12.3%)6,444 (8.1%) high RC (%)191 (2.8%)436 (3.1%)911 (3.8%)1,286 (6.1%)690 (7.5%)267 (7.9%)54 (9.4%)3835 (4.8%) HDL-c (mmol/l, SD)1.52 (0.35)1.53 (0.36)1.59 (0.37)1.68 (0.41)1.74 (0.43)1.72 (0.42)1.73 (0.43)1.62 (0.40) TG (mmol/l, SD)0.93 (0.41)0.89 (0.45)0.93 (0.55)1.05 (0.58)1.16 (0.62)1.21 (0.58)1.21 (0.57)1.00 (0.55) RC (mmol/l, SD)0.29 (0.20)0.26 (0.22)0.27 (0.25)0.31 (0.27)0.34 (0.28)0.35 (0.26)0.34 (0.27)0.29 (0.25) Abbreviations: LDL; low-density lipoprotein; HDL, high-density lipoprotein; TG, triglycerides; RC, remnant cholesterol; SD, standard deviation.
102
Prevalence of non-LDL dyslipidemia and its components in women
In women, non-LDL dyslipidemia occurred in 16%, which was relatively stable across the age groups (Table 2 and Figure 2). The contribution of the individual components of the composite score changed considerably across the age groups (Figure 3). Low HDL-c was the defining factor in more than 60% until the age of 35 years, after which a steep decline to approximately 20% at the age of 55 years occurred. The percentage of high TG increased from around 20% until the age of 35 years to approximately 50% starting at 55 years of age.
High RC increased steadily from about 15% in the youngest age group to 25% in the oldest age group.
Factors contributing to non-LDL dyslipidemia in women
In women, data on lifestyle factors were missing in 2.3% – 6.0%. After multiple imputations, alcohol use was associated with a lower chance of non-LDL dyslipidemia (OR 0.5, 95% CI, 0.5 – 0.6, for moderate alcohol use vs. abstainers). Smoking was associated with increased chance of non-LDL dyslipidemia (OR 2.2, 95% CI 2.1 – 2.3, for current smoking vs. never smoking). Physical activity was moderately associated with non-LDL dyslipidemia (OR 1.2, 95% CI 1.1 – 1.3, for sedentary vs. sufficient physical activity). Weekly intake of fruit, vegetables, and fish were significantly associated with a lower chance of non-LDL dyslipidemia, however the use of meat was not associated with non-LDL dyslipidemia.
Non-LDL dyslipidemia was associated with obesity (OR 5.3, 95% CI 5.0 – 5.6, for obese vs. normal weight). The presence of DM (OR 2.7, 95% CI 2.3 – 3.1) and CKD (OR 1.5, 95% CI 1.2 – 1.8) were associated with non-LDL dyslipidemia, as was the use of oral contraceptives (OR 1.5, 95% CI 1.5 – 1.6) and menopausal state (OR 1.2, 95% CI 1.1 – 1.3) (Table 3). Observed differences between the imputed and non-imputed cohorts were again negligible (Supplementary Table 2).
The ROC curves show an AUC of 0.51 for the association of age groups with non-LDL dyslipidemia in women (Table 4). Adding lifestyle factors, anthropometrics, and medical factors on top of age increased the AUC to 0.62, 0.71, and 0.72, respectively. Each step was statistically significant. Rotating the sequence of determinant addition again showed a larger contribution of BMI and lifestyle factors compared to medical factors (Supplementary Table 1).
Discussion
Our study shows that clinically relevant non-LDL dyslipidemia occurs in 13 – 30% of men and 15 – 19% of women without atherosclerotic cardiovascular disease in the Netherlands. Non-
102
Prevalence of non-LDL dyslipidemia and its components in women
In women, non-LDL dyslipidemia occurred in 16%, which was relatively stable across the age groups (Table 2 and Figure 2). The contribution of the individual components of the composite score changed considerably across the age groups (Figure 3). Low HDL-c was the defining factor in more than 60% until the age of 35 years, after which a steep decline to approximately 20% at the age of 55 years occurred. The percentage of high TG increased from around 20% until the age of 35 years to approximately 50% starting at 55 years of age.
High RC increased steadily from about 15% in the youngest age group to 25% in the oldest age group.
Factors contributing to non-LDL dyslipidemia in women
In women, data on lifestyle factors were missing in 2.3% – 6.0%. After multiple imputations, alcohol use was associated with a lower chance of non-LDL dyslipidemia (OR 0.5, 95% CI, 0.5 – 0.6, for moderate alcohol use vs. abstainers). Smoking was associated with increased chance of non-LDL dyslipidemia (OR 2.2, 95% CI 2.1 – 2.3, for current smoking vs. never smoking). Physical activity was moderately associated with non-LDL dyslipidemia (OR 1.2, 95% CI 1.1 – 1.3, for sedentary vs. sufficient physical activity). Weekly intake of fruit, vegetables, and fish were significantly associated with a lower chance of non-LDL dyslipidemia, however the use of meat was not associated with non-LDL dyslipidemia.
Non-LDL dyslipidemia was associated with obesity (OR 5.3, 95% CI 5.0 – 5.6, for obese vs. normal weight). The presence of DM (OR 2.7, 95% CI 2.3 – 3.1) and CKD (OR 1.5, 95% CI 1.2 – 1.8) were associated with non-LDL dyslipidemia, as was the use of oral contraceptives (OR 1.5, 95% CI 1.5 – 1.6) and menopausal state (OR 1.2, 95% CI 1.1 – 1.3) (Table 3). Observed differences between the imputed and non-imputed cohorts were again negligible (Supplementary Table 2).
The ROC curves show an AUC of 0.51 for the association of age groups with non-LDL dyslipidemia in women (Table 4). Adding lifestyle factors, anthropometrics, and medical factors on top of age increased the AUC to 0.62, 0.71, and 0.72, respectively. Each step was statistically significant. Rotating the sequence of determinant addition again showed a larger contribution of BMI and lifestyle factors compared to medical factors (Supplementary Table 1).
Discussion
Our study shows that clinically relevant non-LDL dyslipidemia occurs in 13 – 30% of men and 15 – 19% of women without atherosclerotic cardiovascular disease in the Netherlands. Non-
6
102 103
LDL dyslipidemia occurs already at a young age and shows a different pattern between men and women. The prevalence of non-LDL dyslipidemia peaked in men in young adulthood, whereas the prevalence in women was relatively constant across age groups. The factors contributing to non-LDL dyslipidemia were similar for both genders and included obesity, DM, current smoking, alcohol use and physical activity.
The difference in prevalence rates between men and women could at least partially be explained by differences in their baseline characteristics. Women more often had a normal body weight (52% vs. 39%) and consumed more fruit and vegetables compared to men, while men were more often current smoker (24% vs. 21%) and moderate or heavy drinker (45% vs. 18%). No substantial differences between men and women were observed in the prevalence of DM and CKD.
Comparison of our observed prevalence rates to previous research is hampered by heterogeneity in populations and cutoff points to define low HDL-c and high TG, cultural differences, and the paucity of population level studies of high RC. The Pan-European survey, using no gender-specific cutoff points in patients with dyslipidemia attending specialist care, found a prevalence of low HDL-c in 34% of men and 39% of women and elevated TG in 57% of men and 48% of women.
140These higher prevalence rates are to be expected since they reflect an elderly and more diseased population compared to our Lifelines cohort. In the Third National Health and Nutrition Examination Survey (NHANES III), the prevalence of low HDL-c (35% men, 39% women) and high TG (35% men, 25% women) were also higher compared to our results;
159however, the obesity rate was substantially higher in NHANES.
Other studies in Mexico, China and Korea also found higher prevalence rates of low HDL-c and high TG.
141, 160, 161In our study, high RC was defined by the level at which cardiovascular risk increased significantly in Danish cohorts.
137The prevalence of high RC in our study was smaller compared to those studies, which can be explained by the inclusion of older patients and patients with atherosclerotic disease in the Danish cohorts.
We found that age is an important factor for the prevalence of non-LDL dyslipidemia, in particular when looking at the individual components. Patterns similar to our findings were also observed in Mexican men and women, despite higher overall prevalence of low HDL-c and high TG.
141On the other hand, different patterns across the age groups were observed in two Asian studies.
160, 161It is not clear what the impact of ethnic differences might be.
Because the data were collected in different time periods, a generational effect could be considered. Although such a generational-consistent dyslipidemic pattern over time was not identified in the NHANES database between 1988-1994 and 1999-2002,
162it could play a role for the more recent generation included in the Lifelines cohort.
6
102 103
LDL dyslipidemia occurs already at a young age and shows a different pattern between men and women. The prevalence of non-LDL dyslipidemia peaked in men in young adulthood, whereas the prevalence in women was relatively constant across age groups. The factors contributing to non-LDL dyslipidemia were similar for both genders and included obesity, DM, current smoking, alcohol use and physical activity.
The difference in prevalence rates between men and women could at least partially be explained by differences in their baseline characteristics. Women more often had a normal body weight (52% vs. 39%) and consumed more fruit and vegetables compared to men, while men were more often current smoker (24% vs. 21%) and moderate or heavy drinker (45% vs. 18%). No substantial differences between men and women were observed in the prevalence of DM and CKD.
Comparison of our observed prevalence rates to previous research is hampered by heterogeneity in populations and cutoff points to define low HDL-c and high TG, cultural differences, and the paucity of population level studies of high RC. The Pan-European survey, using no gender-specific cutoff points in patients with dyslipidemia attending specialist care, found a prevalence of low HDL-c in 34% of men and 39% of women and elevated TG in 57% of men and 48% of women.
140These higher prevalence rates are to be expected since they reflect an elderly and more diseased population compared to our Lifelines cohort. In the Third National Health and Nutrition Examination Survey (NHANES III), the prevalence of low HDL-c (35% men, 39% women) and high TG (35% men, 25% women) were also higher compared to our results;
159however, the obesity rate was substantially higher in NHANES.
Other studies in Mexico, China and Korea also found higher prevalence rates of low HDL-c and high TG.
141, 160, 161In our study, high RC was defined by the level at which cardiovascular risk increased significantly in Danish cohorts.
137The prevalence of high RC in our study was smaller compared to those studies, which can be explained by the inclusion of older patients and patients with atherosclerotic disease in the Danish cohorts.
We found that age is an important factor for the prevalence of non-LDL dyslipidemia, in particular when looking at the individual components. Patterns similar to our findings were also observed in Mexican men and women, despite higher overall prevalence of low HDL-c and high TG.
141On the other hand, different patterns across the age groups were observed in two Asian studies.
160, 161It is not clear what the impact of ethnic differences might be.
Because the data were collected in different time periods, a generational effect could be considered. Although such a generational-consistent dyslipidemic pattern over time was not identified in the NHANES database between 1988-1994 and 1999-2002,
162it could play a role for the more recent generation included in the Lifelines cohort.
104
Table 3. Regression coefficients, determinants of non-LDL dyslipidemia.
Men Women
OR (95% CI) P-value OR (95% CI) P-value Age 18 – 25 0.491 (0.438 – 0.550) < 0.001 1.379 (1.273 – 1.495) < 0.001 Age 25 – 35 0.809 (0.801 – 0.816) < 0.001 1.183 (1.113 – 1.257) < 0.001
Age 35 – 45 1.000 1.000
Age 45 – 55 0.982 (0.929 – 1.037) 0.509 1.097 (1.035 – 1.162) 0.002
Age 55 – 65 0.801 (0.743 – 0.862) < 0.001 1.389 (1.269 – 1.520) < 0.001 Age 65 – 75 0.589 (0.523 – 0.663) < 0.001 1.613 (1.453 – 1.790) < 0.001
Age > 75 0.557 (0.414 – 0.750) < 0.001 1.224 (0.965 – 1.554) 0.096
Diabetes mellitus 2.162 (1.832 – 2.551) < 0.001 2.668 (2.309 – 3.083) < 0.001 Chronic kidney disease 1.887 (1.486 – 2.396) < 0.001 1.496 (1.234 – 1.813) < 0.001
Menopause n/a 1.194 (1.116 – 1.279) < 0.001
Oral contraceptive n/a 1.530 (1.460 – 1.604) < 0.001
Body mass index-group
normal 1.000 1.000
overweight 2.698 (2.562 – 2.842) < 0.001 2.486 (2.365 – 2.603) < 0.001 obese 5.303 (4.960 – 5.669) < 0.001 5.322 (5.046 – 5.614) < 0.001 Smoking
never 1.000 1.000
former 1.186 (1.126 – 1.249) < 0.001 1.048 (0.995 – 1.104) 0.076
current 1.784 (1.689 – 1.884) < 0.001 2.183 (2.072 – 2.300) < 0.001 Alcohol
none 1.000 1.000
light 0.793 (0.789 – 0.796) < 0.001 0.720 (0.687 – 0.755) < 0.001 moderate 0.762 (0.706 – 0.824) < 0.001 0.549 (0.509 – 0.589) < 0.001
heavy 0.887 (0.805 – 0.995) 0.002 0.659 (0.583 – 0.746) < 0.001
Physical activity
sufficient 1.000 1.000
insufficient 1.164 (1.112 – 1.218) < 0.001 1.130 (1.082 – 1.181) < 0.001 sedentary 1.431 (1.288 – 1.590) < 0.001 1.159 (1.065 – 1.260) 0.001 Diet
Fruit (per serving/week) 0.990 (0.986 – 0.993) < 0.001 0.994 (0.990 – 0.997) < 0.001 Vegetable (per serving) 0.993 (0.990 – 0.997) < 0.001 0.996 (0.993 – 0.999) 0.014 Fish (per serving/week) 0.965 (0.937 – 0.995) 0.020 0.936 (0.909 – 0.965) < 0.001 Meat (per serving/week) 1.002 (0.991 – 1.012) 0.757 0.996 (0.986 – 1.006) 0.425
104
Table 3. Regression coefficients, determinants of non-LDL dyslipidemia.
Men Women
OR (95% CI) P-value OR (95% CI) P-value Age 18 – 25 0.491 (0.438 – 0.550) < 0.001 1.379 (1.273 – 1.495) < 0.001 Age 25 – 35 0.809 (0.801 – 0.816) < 0.001 1.183 (1.113 – 1.257) < 0.001
Age 35 – 45 1.000 1.000
Age 45 – 55 0.982 (0.929 – 1.037) 0.509 1.097 (1.035 – 1.162) 0.002
Age 55 – 65 0.801 (0.743 – 0.862) < 0.001 1.389 (1.269 – 1.520) < 0.001 Age 65 – 75 0.589 (0.523 – 0.663) < 0.001 1.613 (1.453 – 1.790) < 0.001
Age > 75 0.557 (0.414 – 0.750) < 0.001 1.224 (0.965 – 1.554) 0.096
Diabetes mellitus 2.162 (1.832 – 2.551) < 0.001 2.668 (2.309 – 3.083) < 0.001 Chronic kidney disease 1.887 (1.486 – 2.396) < 0.001 1.496 (1.234 – 1.813) < 0.001
Menopause n/a 1.194 (1.116 – 1.279) < 0.001
Oral contraceptive n/a 1.530 (1.460 – 1.604) < 0.001
Body mass index-group
normal 1.000 1.000
overweight 2.698 (2.562 – 2.842) < 0.001 2.486 (2.365 – 2.603) < 0.001 obese 5.303 (4.960 – 5.669) < 0.001 5.322 (5.046 – 5.614) < 0.001 Smoking
never 1.000 1.000
former 1.186 (1.126 – 1.249) < 0.001 1.048 (0.995 – 1.104) 0.076
current 1.784 (1.689 – 1.884) < 0.001 2.183 (2.072 – 2.300) < 0.001 Alcohol
none 1.000 1.000
light 0.793 (0.789 – 0.796) < 0.001 0.720 (0.687 – 0.755) < 0.001 moderate 0.762 (0.706 – 0.824) < 0.001 0.549 (0.509 – 0.589) < 0.001
heavy 0.887 (0.805 – 0.995) 0.002 0.659 (0.583 – 0.746) < 0.001
Physical activity
sufficient 1.000 1.000
insufficient 1.164 (1.112 – 1.218) < 0.001 1.130 (1.082 – 1.181) < 0.001 sedentary 1.431 (1.288 – 1.590) < 0.001 1.159 (1.065 – 1.260) 0.001 Diet
Fruit (per serving/week) 0.990 (0.986 – 0.993) < 0.001 0.994 (0.990 – 0.997) < 0.001 Vegetable (per serving) 0.993 (0.990 – 0.997) < 0.001 0.996 (0.993 – 0.999) 0.014 Fish (per serving/week) 0.965 (0.937 – 0.995) 0.020 0.936 (0.909 – 0.965) < 0.001 Meat (per serving/week) 1.002 (0.991 – 1.012) 0.757 0.996 (0.986 – 1.006) 0.425
6
104 105
Table 4. Area under the ROC curve.
Men Women
AUC P-value AUC P-value
Step 1: age groups 0.524 (0.523
–
0.526) 0.505 (0.503–
0.507)Step 2 + lifestyle factors 0.616 (0.610
–
0.621) < 0.001 0.619 (0.614–
0.624) < 0.001 Step 3 + anthropometrics 0.693 (0.688–
0.698) < 0.001 0.711 (0.706–
0.716) < 0.001 Step 4 + medical factors 0.695 (0.690–
0.700) 0.015 0.719 (0.715–
0.724) 0.037 Abbreviations: AUC, area under curve; ROC, receiver operating characteristic.Determinants of non-LDL dyslipidemia
Our study examined lifestyle factors (including alcohol use, smoking, physical activity and diet), obesity, and medical factors (DM, CKD, postmenopausal state and oral contraceptive use) as determinants of non-LDL dyslipidemia. Including all these factors, it seems that particularly lifestyle factors play an important role in non-LDL dyslipidemia. We observed an independent protective association of moderate alcohol use compared to abstinence with non-LDL dyslipidemia, despite strong confounding of alcohol use by other lifestyle factors.
This observation is in line with previous studies.
147, 149, 151, 152, 163Smoking appeared to have a dose-response relationship with non-LDL dyslipidemia, where current smoking conveyed a larger risk of non-LDL dyslipidemia compared to former and never smoking. This seems to support the literature that the adverse effects of tobacco exposure are reversible.
150Because the percentage of never smokers was the highest in the youngest age group (> 50%, data not shown), public health efforts to maintain this healthy behavior can be expected to have beneficial effects on non-LDL dyslipidemia levels in the population. Surprisingly, sedentary lifestyle conferred only small risks of non-LDL dyslipidemia when adjusting for other determinants. This seems at odds with recent reports showing a beneficial effect of moderate-to-vigorous physical activity on HDL-c and TG in a small sample.
164In that study, activity was measured using an accelerometer compared to the Lifelines’ questionnaire- based data. The self-reported physical activity level in our analysis may be an overestimation of the true level, and this could explain why we did not observe a protective association of physical activity with non-LDL dyslipidemia. In light of the recently shown effect of sedentary lifestyle on mortality,
165the high proportion of sedentary lifestyle (4.9%) and insufficient physical activity (43.8%) is alarming. Consumption of fruit, vegetables and fish, modeled as serving sizes, showed protective associations with non-LDL dyslipidemia. This is of interest because the average consumption in this cohort was substantially lower than recommended amounts, and a considerable effect on non-LDL dyslipidemia could be anticipated if consumption were increased to guideline recommendations.
406
104 105
Table 4. Area under the ROC curve.
Men Women
AUC P-value AUC P-value
Step 1: age groups 0.524 (0.523
–
0.526) 0.505 (0.503–
0.507)Step 2 + lifestyle factors 0.616 (0.610
–
0.621) < 0.001 0.619 (0.614–
0.624) < 0.001 Step 3 + anthropometrics 0.693 (0.688–
0.698) < 0.001 0.711 (0.706–
0.716) < 0.001 Step 4 + medical factors 0.695 (0.690–
0.700) 0.015 0.719 (0.715–
0.724) 0.037 Abbreviations: AUC, area under curve; ROC, receiver operating characteristic.Determinants of non-LDL dyslipidemia
Our study examined lifestyle factors (including alcohol use, smoking, physical activity and diet), obesity, and medical factors (DM, CKD, postmenopausal state and oral contraceptive use) as determinants of non-LDL dyslipidemia. Including all these factors, it seems that particularly lifestyle factors play an important role in non-LDL dyslipidemia. We observed an independent protective association of moderate alcohol use compared to abstinence with non-LDL dyslipidemia, despite strong confounding of alcohol use by other lifestyle factors.
This observation is in line with previous studies.
147, 149, 151, 152, 163Smoking appeared to have a dose-response relationship with non-LDL dyslipidemia, where current smoking conveyed a larger risk of non-LDL dyslipidemia compared to former and never smoking. This seems to support the literature that the adverse effects of tobacco exposure are reversible.
150Because the percentage of never smokers was the highest in the youngest age group (> 50%, data not shown), public health efforts to maintain this healthy behavior can be expected to have beneficial effects on non-LDL dyslipidemia levels in the population. Surprisingly, sedentary lifestyle conferred only small risks of non-LDL dyslipidemia when adjusting for other determinants. This seems at odds with recent reports showing a beneficial effect of moderate-to-vigorous physical activity on HDL-c and TG in a small sample.
164In that study, activity was measured using an accelerometer compared to the Lifelines’ questionnaire- based data. The self-reported physical activity level in our analysis may be an overestimation of the true level, and this could explain why we did not observe a protective association of physical activity with non-LDL dyslipidemia. In light of the recently shown effect of sedentary lifestyle on mortality,
165the high proportion of sedentary lifestyle (4.9%) and insufficient physical activity (43.8%) is alarming. Consumption of fruit, vegetables and fish, modeled as serving sizes, showed protective associations with non-LDL dyslipidemia. This is of interest because the average consumption in this cohort was substantially lower than recommended amounts, and a considerable effect on non-LDL dyslipidemia could be anticipated if consumption were increased to guideline recommendations.
40106
We observed a clear association between obesity as well as DM and non-LDL dyslipidemia, in line with previous studies.
145, 149, 159However, our results mitigate the prominent role of obesity in the prevalence of non-LDL dyslipidemia after inclusion of other relevant determinants. The prevalence of DM was only 1.8% in our cohort and the prevalence of CKD was 0.8%, partly due to our exclusion of people that used lipid-lowering drugs. The DM patients in our analysis were on average younger, healthier and had less cardiovascular risk factors compared to the excluded DM patients (data not shown). This could have led to an underestimation of the association of DM with non-LDL dyslipidemia. Although the same could be true for CKD, having CKD was weakly associated with non-LDL dyslipidemia. The pathophysiology of lipid abnormalities in CKD is only roughly known, and an independent effect of CKD on non-LDL dyslipidemia besides other medical, anthropometric and lifestyle factors has not been reported previously.
146Contrary to the expected increase of HDL-c levels with the use of sex hormones,
148a high prevalence of low HDL-c was observed in young women in our cohort. Post-menopausal state was associated with a higher prevalence of high TG.
147Our results emphasize that multiple factors need to be taken into account when addressing non-LDL dyslipidemia. The apparent large contribution of modifiable lifestyle factors to non-LDL dyslipidemia and the high prevalence of suboptimal lifestyle factors in this cohort open up possibilities to reduce the risk of atherosclerotic cardiovascular disease.
143,166, 167
Our findings strongly support lifestyle modification to lower non-LDL dyslipidemia
and thus the risk of cardiovascular disease before novel pharmacological strategies are considered.
168-170Strengths and limitations
We used a large, contemporary population cohort with protocolled data collection using validated questionnaires and assessments.
72Measurement of lipid levels was performed on fresh samples minimizing artificial effects. Part of our data is questionnaire-based, which is limited with recall bias and the potential to provide socially desirable answers. Also, excluding lipid-lowering drug use may have introduced a selection bias, which may have underestimated the prevalence of non-LDL dyslipidemia. Up to 10% of missing data on lifestyle was imputed; however, no significant changes in ORs were observed between the imputed and non-imputed datasets. Finally, the AUC in our models did not exceed 0.72, indicating that unmeasured determinants play a role in non-LDL dyslipidemia.
106
We observed a clear association between obesity as well as DM and non-LDL dyslipidemia, in line with previous studies.
145, 149, 159However, our results mitigate the prominent role of obesity in the prevalence of non-LDL dyslipidemia after inclusion of other relevant determinants. The prevalence of DM was only 1.8% in our cohort and the prevalence of CKD was 0.8%, partly due to our exclusion of people that used lipid-lowering drugs. The DM patients in our analysis were on average younger, healthier and had less cardiovascular risk factors compared to the excluded DM patients (data not shown). This could have led to an underestimation of the association of DM with non-LDL dyslipidemia. Although the same could be true for CKD, having CKD was weakly associated with non-LDL dyslipidemia. The pathophysiology of lipid abnormalities in CKD is only roughly known, and an independent effect of CKD on non-LDL dyslipidemia besides other medical, anthropometric and lifestyle factors has not been reported previously.
146Contrary to the expected increase of HDL-c levels with the use of sex hormones,
148a high prevalence of low HDL-c was observed in young women in our cohort. Post-menopausal state was associated with a higher prevalence of high TG.
147Our results emphasize that multiple factors need to be taken into account when addressing non-LDL dyslipidemia. The apparent large contribution of modifiable lifestyle factors to non-LDL dyslipidemia and the high prevalence of suboptimal lifestyle factors in this cohort open up possibilities to reduce the risk of atherosclerotic cardiovascular disease.
143,166, 167
Our findings strongly support lifestyle modification to lower non-LDL dyslipidemia
and thus the risk of cardiovascular disease before novel pharmacological strategies are considered.
168-170Strengths and limitations
We used a large, contemporary population cohort with protocolled data collection using validated questionnaires and assessments.
72Measurement of lipid levels was performed on fresh samples minimizing artificial effects. Part of our data is questionnaire-based, which is limited with recall bias and the potential to provide socially desirable answers. Also, excluding lipid-lowering drug use may have introduced a selection bias, which may have underestimated the prevalence of non-LDL dyslipidemia. Up to 10% of missing data on lifestyle was imputed; however, no significant changes in ORs were observed between the imputed and non-imputed datasets. Finally, the AUC in our models did not exceed 0.72, indicating that unmeasured determinants play a role in non-LDL dyslipidemia.
6
106 107
Figure 2. Prevalence of non-LDL dyslipidemias and secondary outcomes for men (top) and women (bottom).
Abbreviations: LDL, low-density lipoprotein; HDL-c, high-density lipoprotein cholesterol; TG, triglycerides; RC, remnant cholesterol.
6
106 107
Figure 2. Prevalence of non-LDL dyslipidemias and secondary outcomes for men (top) and women (bottom).
Abbreviations: LDL, low-density lipoprotein; HDL-c, high-density lipoprotein cholesterol; TG, triglycerides; RC, remnant cholesterol.
108
Figure 3. Individual components of non-LDL dyslipidemia for men (blue) and women (orange).
Abbreviations: Rc, remnant cholesterol; TG, triglycerides; HDLc, high-density lipoprotein cholesterol.
108
Figure 3. Individual components of non-LDL dyslipidemia for men (blue) and women (orange).
Abbreviations: Rc, remnant cholesterol; TG, triglycerides; HDLc, high-density lipoprotein cholesterol.
6
108 109
Conclusion
Non-LDL dyslipidemia occurs in approximately a quarter of the LifeLines population, already at an early age and with profound differences per gender. Lifestyle choices and their effects, mainly obesity, smoking, alcohol use and DM were found to be strongly associated with non-LDL dyslipidemia. We urge for a strong public health effort to support a healthy lifestyle, and emphasize the importance of early lifestyle management especially in males for the prevention of CVD.
Funding and disclosures
This research has been supported by an unrestricted grant from MSD. Conflicts of interest:
none.
6
108 109
