Skip main navigation

Serum Triglycerides as a Risk Factor for Cardiovascular Diseases in the Asia-Pacific Region

Originally publishedhttps://doi.org/10.1161/01.CIR.0000145615.33955.83Circulation. 2004;110:2678–2686

Abstract

Background— The importance of serum triglyceride levels as a risk factor for cardiovascular diseases is uncertain.

Methods and Results— We performed an individual participant data meta-analysis of prospective studies conducted in the Asia-Pacific region. Cox models were applied to the combined data from 26 studies to estimate the overall and region-, sex-, and age-specific hazard ratios for major cardiovascular diseases by fifths of triglyceride values. During 796 671 person-years of follow-up among 96 224 individuals, 670 and 667 deaths as a result of coronary heart disease (CHD) and stroke, respectively, were recorded. After adjustment for major cardiovascular risk factors, participants grouped in the highest fifth of triglyceride levels had a 70% (95% CI, 47 to 96) greater risk of CHD death, an 80% (95% CI, 49 to 119) higher risk of fatal or nonfatal CHD, and a 50% (95% CI, 29% to 76%) increased risk of fatal or nonfatal stroke compared with those belonging to the lowest fifth. The association between triglycerides and CHD death was similar across subgroups defined by ethnicity, age, and sex.

Conclusions— Serum triglycerides are an important and independent predictor of CHD and stroke risk in the Asia-Pacific region. These results may have clinical implications for cardiovascular risk prediction and the use of lipid-lowering therapy.

The importance of serum triglycerides as a risk factor for cardiovascular diseases is controversial. Many epidemiological studies have demonstrated a univariate association between triglycerides and cardiovascular risk, particularly in relation to coronary heart disease (CHD).1–12 However, this relationship is attenuated and often becomes nonsignificant after adjustment for major cardiovascular risk factors, particularly HDL cholesterol levels. More recently, the results of a meta-analysis showing an independent association between triglycerides and cardiovascular risk support the suggestion that at least some triglyceride-rich lipoproteins are directly atherogenic.13 While recognizing this possibility, the recent National Cholesterol Education Program–Adult Treatment Panel III (NCEP-ATPIII) guidelines imply that the evidence is insufficient and primarily view elevated triglycerides as a marker for other risks rather than as an independent risk factor.14

Previous analyses have been hindered by a number of methodological problems.5,15 One key limitation has been the potential for regression dilution bias when estimates of associations are based on a single baseline measure of triglycerides.16,17 Such analyses are likely to underestimate true associations, particularly because measures of serum triglycerides exhibit a relatively high degree of intraindividual variability.18 To date, very few studies of sufficient size describing the relationship between triglycerides and cardiovascular risk have had the ability to account for regression dilution.19

Considerable variation in population levels of serum triglycerides has been noted across the Asia-Pacific region.9,20–24 Furthermore, the epidemiology of cardiovascular diseases in much of the region is characterized by a relatively low incidence of CHD and a high incidence of stroke.25–28 Few studies have examined the association between triglycerides and cardiovascular risk in nonwhite populations9,29 or have specifically addressed stroke as an outcome of interest.30–34

The Asia Pacific Cohort Studies Collaboration (APCSC) is an individual participant data meta-analysis of prospective cohort studies conducted in a number of Asian countries, Australia, and New Zealand.35 With several hundred events recorded and with the availability of repeated measures of risk factors, allowing correction for regression dilution, the collaboration is able to produce reliable evidence on the nature and size of the associations between risk factors and cardiovascular diseases. In this report, we describe the results relating to serum triglycerides.

Methods

Participating Studies

The design of the APCSC has been reported elsewhere.35 In summary, cohort studies were eligible for inclusion and were invited to participate if (1) the study population was from Asia or the Pacific region; (2) date of birth, sex, and blood pressure for each individual were recorded at baseline; and (3) ≥5000 person-years of follow-up had been completed. For the present analyses, only those studies with baseline measures of triglycerides were included. Studies were classified as Asian if their participants were recruited from mainland China, Hong Kong, Japan, Singapore, Taiwan, or Thailand or as ANZ if their participants were recruited from Australia or New Zealand.

Measurement of Baseline Variables

Triglycerides were measured at baseline in 26 of the 43 studies included in the APCSC database in March 2003. In addition to age, sex, and blood pressure, serum total cholesterol, serum HDL cholesterol, smoking status, alcohol consumption, body mass index, diabetes status, and fasting glucose were also included in the pooled data set when available. Lipid measurements were determined from serum samples, which were obtained in 90% of participants while fasting. Information on the method of triglyceride assay was available from 24 of the 26 studies; 23 of these, contributing 91% of participants, determined triglyceride levels with enzymatic methods. Total cholesterol was measured by enzymatic methods in the same 23 studies. HDL cholesterol was assayed with a precipitation method in 12 studies, by enzymatic methods in 10, and by electrophoresis in 1; it was not measured in 3 studies.

Outcomes

Each study reported deaths by underlying cause; some studies also reported nonfatal cardiovascular events. Twenty-four studies provided detailed information on the method of follow-up. In 19 of these studies, deaths were ascertained only by registry and/or hospital record linkage without routine follow-up visits. Nonfatal events were detected by various combinations of hospital database linkage, medical record review, and scheduled follow-up visits. Outcomes were classified according to the ninth revision of the International Classification of Diseases (ICD-9). The fatal outcomes considered in this analysis were CHD (ICD-9, 410 to 414), total stroke (ICD-9, 430 to 438), hemorrhagic stroke (ICD-9, 431.0 to 432.9), and ischemic stroke (ICD-9, 433.0 to 434.9). Two composite outcomes were also examined: fatal or nonfatal CHD and fatal or nonfatal stroke (total and for each stroke subtype).

Statistical Methods

All analyses used individual participant data and were restricted to those ≥20 years of age at study entry. For grouped analyses, individuals were classified according to approximately equal fifths of baseline triglycerides for the entire study population (≤0.7, 0.8 to 1.0, 1.1 to 1.3, 1.4 to 1.8, and ≥1.9 mmol/L). Trends in mean values of other major continuous cardiovascular risk factors across these fifths were assessed through simple linear regression, with the groups coded in rank order. Trends in percentages for binary risk factors were assessed similarly with χ2 tests for trend.36 Cox proportional-hazards regression models were used to estimate hazard ratios (HRs) for usual triglyceride level, with corresponding 95% CIs calculated by use of the floating absolute risk method.36,37 Log-linearity of triglyceride associations was explored both through fitting fifths of triglycerides as a continuous predictor and through the HR and 95% CI for a 1-SD increase in continuous usual log-transformed triglyceride level. Log-transformation of triglyceride values was performed to account for the extreme right skewness of the data.

All Cox models included systolic blood pressure, smoking status, and time-dependent age plus stratification variables for study and sex. This method allows background risk to vary by sex and between studies but assumes that there are equal relative effects of triglycerides throughout. The base model was additionally adjusted for the total-to-HDL cholesterol ratio. Some risk factors were incompletely recorded for specific individuals within the studies in which triglycerides were routinely measured. Multiple imputation methods38,39 were used to estimate these missing values (0.4%, 0.9%, and 26% values missing for smoking status, total cholesterol, and HDL cholesterol, respectively). Additional modeling with separate adjustment for total cholesterol and HDL cholesterol was performed, and further adjustment for diabetes, alcohol consumption, and body mass index was done. Sensitivity analyses restricted analyses to subgroups of participants with (1) no missing covariate values of smoking status, total cholesterol, and HDL cholesterol at baseline; (2) confirmed fasting blood samples; and (3) fasting glucose measurements available.

To assess the association of usual triglyceride level with each outcome, baseline triglyceride measurements were adjusted to account for regression dilution bias. Repeated fasting triglyceride measurements were available in ≈6% of participants from 3 studies (2 in Asia and 1 in ANZ) between 2 and 20 years after the baseline measurement. These repeated measures were used to estimate a regression dilution coefficient (1.9) with the use of a linear mixed regression model that accounted for the heterogeneity of variance between studies, within-subject correlation, and the varying time intervals between measurements.40 Repeated measures were also available from these 3 studies for total and HDL cholesterol and for systolic blood pressure. The same basic approach was used to estimate adjusted regression dilution coefficients with these covariates accounted for.

Results

Baseline Data

The 26 studies in the APCSC with data on baseline triglycerides, which include a total of 96 224 individuals with 796 671 person-years of follow-up, are summarized in Table 1. Compared with participants in the ANZ studies, the Asian cohort tended to be older (49 versus 46 years; P<0.001) and included fewer women (47% versus 51%; P<0.001).

TABLE 1. Major Characteristics of Studies

Study NameSubjects, nTriglycerides, mean (SD), mmol/LWomen, %Mean Age, yMedian Follow-Up, y
*Triglycerides confirmed as measured in fasting state.
†90% of participants confirmed fasting.
Aito Town16711.34 (0.51)56.651.015.2
Akabane*18341.24 (0.77)55.754.511.0
Anzhen 02*41411.23 (0.87)51.047.03.0
Beijing Aging16931.52 (0.81)50.369.24.8
Civil Service Workers*92881.46 (1.05)33.146.76.7
CVDFACTS55491.21 (0.82)55.647.16.0
EGAT*34911.72 (1.31)22.743.011.4
Fangshan*8231.55 (1.19)67.447.12.7
Guangzhou Occupational*15 9601.52 (1.18)36.143.87.9
Hong Kong*1951.83 (1.52)51.879.12.5
Huashan*17981.57 (1.05)52.153.32.8
Kounan Town*12201.15 (0.83)55.351.86.4
Miyama4141.29 (0.80)62.159.26.6
Ohasama19061.60 (1.02)64.858.24.1
Saitama*36231.48 (1.03)62.254.511.0
Seven Cities*63761.74 (1.00)57.157.92.7
Shigaraki37501.49 (0.99)59.557.24.4
Shirakawa*46431.27 (0.66)54.348.017.5
Singapore Heart*22951.39 (0.92)49.140.614.6
Singapore 9233031.47 (1.47)51.839.26.2
Tanno/Soubetsu*19761.35 (0.95)53.251.116.4
Xi’an16951.19 (0.53)33.744.419.7
Subtotal Asia77 6361.46 (1.04)47.048.97.0
Australian Longitudinal Study of Aging*11561.63 (0.99)48.177.94.7
Australian National Heart Foundation*91111.50 (1.17)50.643.48.3
Busselton23391.41 (0.96)54.644.217.5
Perth59821.26 (0.95)49.345.38.4
Subtotal ANZ18 5881.42 (1.07)50.546.38.4
Total96 2241.45 (1.05)47.748.47.9

The age- and sex-adjusted mean triglyceride values were 1.45 mmol/L (95% CI, 1.44 to 1.46) for Asian participants and 1.44 mmol/L (95% CI, 1.43 to 1.46) for ANZ participants. The age-adjusted mean triglyceride value was greater in men compared with women (1.58 mmol/L [95% CI, 1.57 to 1.58] versus 1.31 mmol/L [95% CI, 1.30 to 1.32]; P<0.001). The age- and sex-adjusted levels of major cardiovascular risk factors after stratification of the study population by fifths of baseline triglycerides are summarized in Table 2. Age, body mass index, systolic blood pressure, serum total cholesterol, serum total-to-HDL cholesterol ratio, fasting serum glucose, and prevalence of current smoking and diabetes all increased with increasing triglycerides (all P<0.001). Serum HDL cholesterol decreased with increasing triglycerides (P<0.001).

TABLE 2. Distribution of Major Risk Factors by Baseline Triglycerides Fifth Adjusted by Age and Sex

Variablesn*rFifth of Triglycerides
1 (≤0.7 mmol/L)2 (0.8–1.0 mmol/L)3 (1.1–1.3 mmol/L)4 (1.4–1.8 mmol/L)5 (≥1.9 mmol/L)
P for trend, all <0.001.
*Number of participants with data available for each variable.
†Spearman’s coefficient for correlation with triglycerides.
Mean (95% CI)
    Age, y96 2240.1943.8 (43.6–44.0)47.2 (47.1–47.4)49.2 (49.0–49.3)50.4 (50.2–50.6)50.7 (50.5–50.9)
    Body mass index, kg/m282 5860.3321.9 (21.8–21.9)22.6 (22.5–22.6)23.3 (23.3–23.4)24.1 (24.1–24.2)25.2 (25.1–25.2)
    Systolic blood pressure, mm Hg96 2240.23121.4 (121.1–121.7)123.3 (123.0–123.5)125.0 (124.7–125.3)126.7 (126.5–127.0)129.7 (129.5–130.0)
    Serum total cholesterol, mmol/L95 8740.314.66 (4.65–4.67)4.95 (4.93–4.96)5.16 (5.14–5.18)5.38 (5.36–5.39)5.67 (5.66–5.69)
    Serum HDL cholesterol, mmol/L70 971−0.391.53 (1.53–1.54)1.45 (1.45–1.46)1.38 (1.37–1.38)1.30 (1.30–1.31)1.17 (1.16–1.18)
    Serum total-to-HDL cholesterol ratio70 8450.493.24 (3.21–3.28)3.67 (3.64–3.70)4.08 (4.05–4.11)4.54 (4.51–4.57)5.47 (5.44–5.50)
    Fasting blood glucose, mmol/L25 7790.135.07 (5.04–5.11)5.04 (5.01–5.07)5.14 (5.10–5.17)5.23 (5.20–5.27)5.50 (5.47–5.54)
Percentage (95% CI)
    Current smoking95 37329.7 (28.8–30.5)31.3 (30.6–32.0)32.6 (31.8–33.5)33.5 (32.6–34.3)34.1 (33.2–34.9)
    Current alcohol consumption82 92046.5 (45.3–47.8)46.7 (45.7–47.7)45.5 (44.5–46.6)45.6 (44.6–46.6)45.4 (44.4–46.4)
    History of diabetes72 8202.1 (1.8–2.4)2.1 (1.9–2.3)2.7 (2.4–3.0)3.2 (2.9–3.5)5.2 (4.8–5.6)

Outcomes

During follow-up, 5137 deaths occurred, of which 1932 (38%) were assigned an underlying cardiovascular cause (Table 3). Of these, 667 deaths were due to stroke (577 in Asia, 90 in ANZ), and 670 were due to CHD (349 in Asia, 321 in ANZ). In the Asian cohorts, stroke and CHD accounted for 41% and 25% of cardiovascular deaths, respectively; corresponding values in the ANZ cohorts were 17% and 61%. Nonfatal events were recorded in 10 studies for stroke (456 events) and 8 studies for CHD (180 events).

TABLE 3. Cardiovascular Events

Study NameNo. of SubjectsDeathsFatal or Nonfatal Events*
CVDStrokeHaem-SIsch-SU/K-SCHDStrokeHaem-SIsch-SU/K-SCHD
CVD indicates cardiovascular disease; CHD, coronary heart disease; Haem-S, hemorrhagic stroke; Isch-S, ischemic stroke; U/K-S, stroke of unknown type or subarachnoid hemorrhage.
*Only studies that reported both fatal and nonfatal outcomes contributed events to this combined end point.
Aito Town1671552451915
Akabane183435121927385161728
Anzhen 024141211001161421
Beijing Aging16931506767
Civil Service Workers92881221011
CVDFACTS55405929871412
EGAT349151161633
Fangshan82331181612
Guangzhou Occupational15 960632820821
Hong Kong19515447
Huashan17981363303156903
Kounan Town122024123722
Miyama414100000
Ohasama190716522144210275
Saitama36231205515271324
Seven Cities637635713094306612251231011
Shigaraki375029132473
Shirakawa464316372282816458731391765
Singapore Heart22955620421429727224361
Singapore 92330333622222444142633
Tanno/Soubetsu197673331610724
Xi’an169580412415235
Subtotal Asia77 6361410577229146202349547187248112193
Australian Longitudinal Study of Aging115615035842360
Australian National Heart Foundation91111121611577
Busselton233916922321711911782584175
Perth59829117421165
Subtotal ANZ18 588522901596632111782584175
Total96 2241932667244155268670664195273196368

Triglycerides and Risk of CHD

There was a continuous, positive association between usual triglyceride levels and the risk of CHD that persisted after adjustment for age, sex, blood pressure, smoking, and total-to-HDL cholesterol ratio (Figure 1). Compared with those individuals grouped in the lowest fifth of usual triglyceride values, the risk of fatal CHD was 70% (95% CI, 47 to 96) greater for participants with usual triglyceride levels in the highest fifth. For the composite outcome of fatal or nonfatal CHD, the corresponding value was 80% (95% CI, 19 to 119). Each 1-SD-higher level of log-triglycerides was associated with a significantly greater risk of death resulting from CHD (HR, 1.33; 95% CI, 1.09 to 1.62) and fatal or nonfatal CHD (HR, 1.56; 95% CI, 1.20 to 2.03).

Figure 1. Association between usual triglyceride level and death caused by CHD (A), fatal or nonfatal CHD (B), death resulting from stroke (C), and fatal or nonfatal stroke (D). Analyses are adjusted by time-dependent age, systolic blood pressure, smoking status, and total-to-HDL cholesterol ratio; study and sex are included as stratification variables. Bars show 95% CIs. Probability value for linear trend across fifths of triglycerides is shown.

Triglycerides and Risk of Stroke

Overall, no association between triglycerides and risk of fatal stroke was observed (Figure 1C). However, evidence of a positive log-linear association emerged when the combined outcome of fatal or nonfatal stroke was considered (Figure 1D). Compared with those in the lowest fifth of triglycerides, the risk of fatal or nonfatal stroke among individuals in the highest fifth was increased by 50% (HR, 1.50; 95% CI, 1.29 to 1.76). On examination of stroke subtypes (Figure 2), triglycerides appeared to be associated with the risk of ischemic stroke, although, with a substantially greater number of events recorded, a significant log-linear trend across the fifths was observed only for the combined fatal and nonfatal outcome. The HR for the risk of fatal or nonfatal ischemic stroke comparing the highest and lowest fifths of triglycerides was 1.97 (95% CI,1.52 to 2.55). For the same outcome, the HR associated with each 1-SD-higher level of triglycerides was 1.35 (95% CI, 1.00 to 1.83). There was no evidence of a significant association between triglycerides and risk of hemorrhagic stroke.

Figure 2. Association between usual triglyceride level and fatal ischemic stroke (A), fatal or nonfatal ischemic stroke (B), fatal hemorrhagic stroke (C), and fatal or nonfatal hemorrhagic stroke (D). Analyses and conventions as for Figure 1.

Subgroups of Participants

Although there was an appearance of a stronger independent association between triglycerides and the risk of fatal CHD in the ANZ compared with the Asian population (Figure 3), the statistical test for heterogeneity was nonsignificant (HR for highest versus lowest fifth of triglycerides, 1.79 [95% CI, 1.50 to 2.14] versus 1.49 [95% CI, 1.20 to 1.84; P=0.09]). There was no evidence for an interaction between triglyceride level and age, sex, or diabetes status for fatal CHD.

Figure 3. HRs (and 95% CIs) comparing within subgroups risk of fatal CHD between individuals belonging to highest vs lowest fifth of usual triglyceride (TG) levels. Analyses are adjusted by study, sex, age at risk, systolic blood pressure, smoking, and total-to-HDL cholesterol ratio. *Diabetes subgroups compared in a smaller sample.

Sensitivity Analyses

In addition to adjustment for study, sex, age, systolic blood pressure, and smoking status, separate adjustment for total cholesterol, HDL cholesterol, and total-to-HDL cholesterol ratio attenuated the association between triglycerides and fatal CHD by a similar extent (≈10% lower point estimate of the HR comparing the highest to the lowest fifth of usual triglyceride levels) (Figure 4A). Limiting the analysis to individuals with a triglyceride value ≤4.4 mmol/L and adding HDL cholesterol and calculated LDL cholesterol to the model with study, sex, age, systolic blood pressure, and smoking status also had little effect (HR associated with the highest versus lowest fifth of usual triglycerides for fatal CHD, 1.69 [95% CI, 1.44 to 1.99] versus 1.96 [95% CI, 1.71 to 2.56]).

Figure 4. Sensitivity analyses for HR comparing risk of fatal CHD between individuals belonging to highest vs lowest fifth of usual triglyceride (TG) levels. A, Models including all participants; B, models including only participants without imputed covariate values; C, models including only participants with fasting triglyceride measurements; and D, models including only participants with baseline fasting blood glucose available. SBP indicates systolic blood pressure; TC, total cholesterol; and HDL-C, HDL cholesterol.

When analyses were restricted to the subset of participants with complete covariate data (therefore not requiring any imputations), there was little effect on the point estimates of association (HR associated with highest versus lowest fifth of usual triglyceride level for CHD death, 1.72; 95% CI, 1.46 to 2.02) (Figure 4B). Separate analyses for the subset of participants with confirmed fasting triglyceride values (Figure 4C) produced estimates of association similar to those observed for the overall fasting and nonfasting population.

Further adjustment for diabetes status, alcohol consumption, and body mass index in the smaller number of studies with such data had little effect on the HRs for CHD death (data not shown). Although adjustment for fasting glucose attenuated the point estimate of the association between triglycerides and the risk of death as a result of CHD (Figure 4D), the number of events was small, resulting in wide CIs.

Finally, there was no evidence of between-study heterogeneity in the association between triglycerides and any of the outcomes of CHD death (P=0.20), fatal or nonfatal CHD (P=0.32), hemorrhagic stroke (P=0.56), or ischemic stroke (P=0.85).

Discussion

To the best of our knowledge, this is the largest prospective analysis of the association between triglycerides and cardiovascular events published to date. The results indicate that serum triglyceride level is an independent determinant of cardiovascular risk across a broad population group within the Asia-Pacific region. The evidence is particularly strong for CHD; however, these data also indicate an independent association between triglycerides and risk of ischemic stroke.

The mixed results of previous studies are probably due to a number of factors. As previously discussed, bias resulting from regression dilution is an important limitation when associations are determined on the basis of a single baseline measurement of triglycerides. Although undesirable, decisions in clinical practice often rely on a single measurement of triglyceride level; thus, it could be argued that risk prediction tools using triglycerides should not adjust for regression dilution. However, it is appropriate to account for this potential bias when attempting to describe the true relationship between a risk factor and disease, as was the objective of the present analyses. Nevertheless, even without correction for regression dilution, triglycerides remain independently associated with CHD risk in the current study.

The lack of consistency of previous data may also partly relate to different risk profiles for fasting and postprandial states.9,41,42 However, it is likely that an important explanation for the variability of results of smaller studies relates to the marked heterogeneity in triglyceride-rich lipoproteins found in the circulation and the evidence directly implicating some but not all of these lipoproteins in atherogenesis.43 In smaller studies, the effects of any such atherogenic particles are likely to be diluted and more difficult to detect. Apart from low statistical power, variability in the nature of triglyceride-rich lipoproteins may also explain the findings in other studies of a log-linear association with CHD risk at low levels of triglycerides, with a decline in this risk at higher levels.10,44

Because of statistical correlation and complex metabolic interactions between triglyceride- and cholesterol-rich lipoproteins, most other studies have shown that the association between triglycerides and cardiovascular risk diminishes after adjustment for total cholesterol and, more importantly, HDL cholesterol. This was also observed in the present study, although the size of the attenuation was small, and the association between triglycerides and risk of cardiovascular disease remained statistically significant. It has been argued that adjustment for total cholesterol (as a surrogate for LDL cholesterol) is not appropriate because this procedure adjusts for the triglyceride-rich VLDL component of total cholesterol, which rises disproportionately as serum triglycerides increase.5 Although our results may therefore reflect an overadjustment, this would imply that the association between triglycerides and cardiovascular risk may have been underestimated. Although LDL cholesterol was not directly measured in our studies, combined adjustment for HDL cholesterol and calculated LDL cholesterol did not change the conclusions. A more substantial reduction in the HR was noted after adjustment for baseline fasting glucose, an observation also made by others.5,19 In the present analyses, the subset of participants for whom fasting glucose values were available at baseline was small, limiting our ability to draw reliable conclusions. There was, however, no evidence of a difference in the association between triglycerides and fatal CHD among individuals with and without diabetes.

The data suggest that the association between triglycerides and stroke is not uniform. There is evidence of a significant positive log-linear association with ischemic stroke but not with hemorrhagic stroke. The lack of association with fatal stroke is consistent with a relatively high proportion of hemorrhagic events that have a higher case fatality.45 However, when the combined fatal and nonfatal stroke outcome is considered, ischemic events predominate, and evidence of a positive log-linear association with triglycerides emerges. These data are consistent with a role of elevated triglycerides in the atherothrombotic process but not in terms of hemorrhagic stroke risk.

The association between triglycerides and cardiovascular disease appears consistent across participant subgroups. There was a trend suggesting a stronger association between triglycerides and fatal CHD in the ANZ compared with the Asian cohorts, although this result was not statistically significant and is likely to represent a chance finding. Unlike other studies, including the previous meta-analysis,13,46 we were unable to demonstrate any differences in the associations between triglycerides and cardiovascular risk between men and women or between different age groups.

In addition to providing unique pooled data on Asian populations and having repeated triglyceride measurements, allowing control of regression dilution bias, the present meta-analysis has a number of important advantages. The combination of data from many cohorts results in a large number of events that allow precise estimation of the association between triglycerides and cardiovascular risk both overall and within subgroups. The use of individual participant data allows accurate adjustment for covariates and provides an opportunity to reduce bias resulting from nonresponse through imputation of missing values. There are limitations, however, including differences in the method of triglyceride measurement, missing data on covariates, and incomplete information on nonfatal outcomes. Furthermore, given the limited data set available for adjustment for fasting blood glucose, the presence of some residual confounding remains a possibility. An emphasis on fatal outcomes, inclusion of only those studies that reported both fatal and nonfatal events when the composite outcomes were considered, and the various sensitivity analyses provide some reassurance that these factors are unlikely to have introduced substantial bias.

In summary, these analyses provide robust evidence that serum triglyceride levels are associated with the risk of developing cardiovascular diseases independently of other major measured risk factors, including HDL cholesterol. Whether this translates to important improvements in the ability to predict individual risk of cardiovascular disease when triglyceride measurements are incorporated into risk algorithms is unknown, as is the specific role of therapy aimed at lowering serum triglyceride levels. The clinical implications of these findings warrant further investigation.

Appendix

Writing Committee

A. Patel, F. Barzi, K. Jamrozik, T.H. Lam, H. Ueshima, G. Whitlock, M. Woodward.

Executive Committee

D. Gu, T.H. Lam, S. MacMahon, W. Pan, I. Suh, H. Ueshima, C. Lawes, A. Rodgers, M. Woodward.

Statistical Analyses

F. Barzi, V. Parag, M. Woodward.

Principal APCSC Investigators

S. Ameratunga, G. Andrews, N. Aoki, R. Broadhurst, L.Q. Chen, Z.M. Chen, S.K. Chew, S.R. Choudhury, H. Christensen, A. Dobson, X.H. Fang, J.L. Fuh, M. Fujishima, X.F. Duan, G. Giles, D.F. Gu, T. Hashimoto, Y. He, D. Heng, S.C. Ho, M. Hobbs, Z. Hong, H. Horibe, A. Hozawa, M.S. Huang, K. Hughes, Y. Imai, H. Iwamoto, R. Jackson, K. Jamrozik, S.H. Jee, C.Q. Jiang, M. Kagaya, I.S. Kim, Y. Kita, Y. Kiyohara, M.W. Knuiman, N. Kubo, T.H. Lam, J. Lee, S.C. Li, Y.H. Li, Z.Z. Li, L.S. Liu, S. MacMahon, H. Maegawa, Y. Matsutani, K. Nakachi, M. Nakamura, R. Norton, A. Nozaki, T. Ohkubo, A. Okayama, W.H. Pan, S. Saitoh, K. Sakata, G.L. Shan, K. Shimamoto, P. Sritara, I. Suh, A. Tamakoshi, H. Tanaka, Z. Tang, H. Ueshima, T.A. Welborn, G. Whitlock, J. Woo, X.G. Wu, Z.L. Wu, Z.S. Wu, J.X. Xie, T. Yamada, Q.D. Yang, C.H. Yao, S.X. Yao, X.H. Yu, H.Y. Zhang, B. Zhou, B.F. Zhou, J. Zhou.

*See the Appendix for a list of contributors.

This project has received grants from the National Health and Medical Research Council of Australia, the Health Research Council of New Zealand, and the US National Institutes of Health and an unrestricted educational grant from Pfizer Inc.

Footnotes

Correspondence to Dr Anushka Patel, Asia-Pacific Cohort Studies Collaboration Secretariat, The George Institute for International Health, University of Sydney, PO Box M201, Missenden Rd, Camperdown, Sydney NSW 2050, Australia. E-mail

References

  • 1 Assmann G, Schulte H, Funke H, et al. The emergence of triglycerides as a significant independent risk factor in coronary artery disease. Eur Heart J. 1998; 19 (suppl M): M8–M14.MedlineGoogle Scholar
  • 2 Avins AL, Browner WS. Improving the prediction of coronary heart disease to aid in the management of high cholesterol levels: what a difference a decade makes. JAMA. 1998; 279: 445–449.CrossrefMedlineGoogle Scholar
  • 3 Bainton D, Miller NE, Bolton CH, et al. Plasma triglyceride and high density lipoprotein cholesterol as predictors of ischaemic heart disease in British men. Br Heart J. 1992; 68: 60–66.CrossrefMedlineGoogle Scholar
  • 4 Cambien F, Jacqueson A, Richard JL, et al. Is the level of serum triglyceride a significant predictor of coronary death in “normocholesterolemic” subjects? Am J Epidemiol. 1986; 124: 624–632.CrossrefMedlineGoogle Scholar
  • 5 Criqui MH, Heiss G, Cohn R, et al. Plasma triglyceride level and mortality from coronary heart disease. N Engl J Med. 1993; 328: 1220–1225.CrossrefMedlineGoogle Scholar
  • 6 Frost PH, Davis BR, Burlando AJ, et al. Serum lipids and incidence of coronary heart disease: findings from the Systolic Hypertension in the Elderly Program (SHEP). Circulation. 1996; 94: 2381–2388.CrossrefMedlineGoogle Scholar
  • 7 Gaziano JM, Hennekens CH, O’Donnell CJ, et al. Fasting triglycerides, high-density lipoprotein, and risk of myocardial infarction. Circulation. 1997; 96: 2520–2525.CrossrefMedlineGoogle Scholar
  • 8 Haim M, Benderly M, Brunner D, et al. Elevated serum triglyceride levels and long-term mortality in patients with coronary heart disease: the Bezafibrate Infarction Prevention (BIP) Registry. Circulation. 1999; 100: 475–482.CrossrefMedlineGoogle Scholar
  • 9 Iso H, Naito Y, Sato S, et al. Serum triglycerides and risk of coronary heart disease among Japanese men and women. Am J Epidemiol. 2001; 153: 490–499.CrossrefMedlineGoogle Scholar
  • 10 Jeppesen J, Hein HO, Suadicani P, et al. Triglyceride concentration and ischemic heart disease: an eight-year follow-up in the Copenhagen Male Study. Circulation. 1998; 97: 1029–1036.CrossrefMedlineGoogle Scholar
  • 11 Stampfer MJ, Krauss RM, Ma J, et al. A prospective study of triglyceride level, low-density lipoprotein particle diameter, and risk of myocardial infarction. JAMA. 1996; 276: 882–888.CrossrefMedlineGoogle Scholar
  • 12 Austin MA, Rodriguez BL, McKnight B, et al. Low-density lipoprotein particle size, triglycerides, and high-density lipoprotein cholesterol as risk factors for coronary heart disease in older Japanese-American men. Am J Cardiol. 2000; 86: 412–416.CrossrefMedlineGoogle Scholar
  • 13 Hokanson JE, Austin MA. Plasma triglyceride level is a risk factor for cardiovascular disease independent of high-density lipoprotein cholesterol level: a meta-analysis of population-based prospective studies. J Cardiovasc Risk. 1996; 3: 213–219.CrossrefMedlineGoogle Scholar
  • 14 Expert Panel on Detection Evaluation and Treatment of High Blood Cholesterol in Adults. Executive summary of the Third Report of the National Cholesterol Education Program (NCEP) Expert Panel on Detection, Evaluation, and Treatment of High Blood Cholesterol in Adults (Adult Treatment Panel III). JAMA. 2001; 285: 2486–2497.CrossrefMedlineGoogle Scholar
  • 15 Austin MA. Plasma triglyceride and coronary heart disease. Arterioscler Thromb. 1991; 11: 2–14.CrossrefMedlineGoogle Scholar
  • 16 Clarke R, Shipley M, Lewington S, et al. Underestimation of risk associations due to regression dilution in long-term follow-up of prospective studies. Am J Epidemiol. 1999; 150: 341–353.CrossrefMedlineGoogle Scholar
  • 17 MacMahon S, Peto R, Cutler J, et al. Blood pressure, stroke, and coronary heart disease, part 1: prolonged differences in blood pressure: prospective observational studies corrected for the regression dilution bias. Lancet. 1990; 335: 765–774.CrossrefMedlineGoogle Scholar
  • 18 Jacobs DR Jr, Barrett-Connor E. Retest reliability of plasma cholesterol and triglyceride: the Lipid Research Clinics Prevalence Study. Am J Epidemiol. 1982; 116: 878–885.CrossrefMedlineGoogle Scholar
  • 19 Egger M, Davey Smith G, Pfluger D, et al. Triglyceride as a risk factor for ischemic heart disease in British men: effect of adjusting for measurement error. Atherosclerosis. 1999; 143: 275–284.CrossrefMedlineGoogle Scholar
  • 20 Bhuripanyo K, Tatsanavivat P, Matrakool B, et al. A prevalence survey of lipids abnormalities of rural area in Amphoe Phon, Khon Kaen. J Med Assoc Thai. 1993; 76: 101–108.MedlineGoogle Scholar
  • 21 Chadha SL, Gopinath N, Shekhawat S. Urban-rural differences in the prevalence of coronary heart disease and its risk factors in Delhi. Bull WHO. 1997; 75: 31–38.MedlineGoogle Scholar
  • 22 Pan WH, Chiang BN. Plasma lipid profiles and epidemiology of atherosclerotic diseases in Taiwan: a unique experience. Atherosclerosis. 1995; 118: 285–295.CrossrefMedlineGoogle Scholar
  • 23 Park Y, Lee H, Koh CS, et al. Community-based epidemiologic study on atherosclerotic cardiovascular risk factors. Diabetes Res Clin Pract. 1996; 34 (suppl): S65–S72.CrossrefMedlineGoogle Scholar
  • 24 Tao S, Li Y, Xiao Z, et al. Serum lipids and their correlates in Chinese urban and rural populations of Beijing and Guangzhou. Int J Epidemiol. 1992; 21: 893–903.CrossrefMedlineGoogle Scholar
  • 25 Chen D, Roman GC, Wu GX, et al. Stroke in China (Sino-MONICA-Beijing study) 1984–1986. Neuroepidemiology. 1992; 11: 15–23.CrossrefMedlineGoogle Scholar
  • 26 Khor GL. Cardiovascular epidemiology in the Asia-Pacific region. Asia Pacific J Clin Nutr. 2001; 10: 76–80.CrossrefMedlineGoogle Scholar
  • 27 Kimura Y, Takishita S, Muratani H, et al. Demographic study of first-ever stroke and acute myocardial infarction in Okinawa, Japan. Intern Med. 1998; 37: 736–745.CrossrefMedlineGoogle Scholar
  • 28 Sekikawa A, Kuller LH, Ueshima H, et al. Coronary heart disease mortality trends in men in the post World War II birth cohorts aged 35–44 in Japan, South Korea and Taiwan compared with the United States. Int J Epidemiol. 1999; 28: 1044–1049.CrossrefMedlineGoogle Scholar
  • 29 He Y, Lam TH, Li LS, et al. Triglyceride and coronary heart disease mortality in a 24-year follow-up study in Xi’an, China. Ann Epidemiol. 2004; 14: 1–7.CrossrefMedlineGoogle Scholar
  • 30 Lindenstrom E, Boysen G, Nyboe J. Influence of total cholesterol, high density lipoprotein cholesterol, and triglycerides on risk of cerebrovascular disease: the Copenhagen City Heart Study. BMJ. 1994; 309: 11–15.CrossrefMedlineGoogle Scholar
  • 31 Njolstad I, Arnesen E, Lund-Larsen PG. Body height, cardiovascular risk factors, and risk of stroke in middle-aged men and women: a 14-year follow-up of the Finnmark Study. Circulation. 1996; 94: 2877–2882.CrossrefMedlineGoogle Scholar
  • 32 Simons LA, McCallum J, Friedlander Y, et al. Risk factors for ischemic stroke: Dubbo Study of the Elderly. Stroke. 1998; 29: 1341–1346.CrossrefMedlineGoogle Scholar
  • 33 Tanne D, Koren-Morag N, Graff E, et al. Blood lipids and first-ever ischemic stroke/transient ischemic attack in the Bezafibrate Infarction Prevention (BIP) Registry: high triglycerides constitute an independent risk factor. Circulation. 2001; 104: 2892–2897.CrossrefMedlineGoogle Scholar
  • 34 Wannamethee SG, Shaper AG, Ebrahim S. HDL-cholesterol, total cholesterol, and the risk of stroke in middle-aged British men. Stroke. 2000; 31: 1882–1888.CrossrefMedlineGoogle Scholar
  • 35 Asia Pacific Cohort Studies Collaboration. Determinants of cardiovascular disease in the Asia Pacific region: protocol for a collaborative overview of cohort studies. CVD Prevent. 1999; 2: 281–289.Google Scholar
  • 36 Woodward M. Epidemiology: Study Design and Data Analysis. 2nd ed. Boca Raton, Fla: Chapman and Hall/CRC Press; 2004.Google Scholar
  • 37 Easton DF, Peto J, Babiker AG. Floating absolute risk: an alternative to relative risk in survival and case-control analysis avoiding an arbitrary reference group. Stats Med. 1991; 10: 1025–1035.CrossrefMedlineGoogle Scholar
  • 38 Shafer JL. Analysis of Incomplete Multivariate Data. London, UK: Chapman & Hall; 1997.Google Scholar
  • 39 Barzi F, Woodward M. Imputations of missing values in practice: results from imputations of serum cholesterol in 28 cohort studies. Am J Epidemiol. 2004; 160: 34–45CrossrefMedlineGoogle Scholar
  • 40 Rosner B, Spiegelman D, Willet W. Correction of logistic regression relative risk estimates and confidence intervals for measurement error: the case of multiple covariates measured with error. Am J Epidemiol. 1990; 132: 734–745.CrossrefMedlineGoogle Scholar
  • 41 Ebenbichler CF, Kirchmair R, Egger C, et al. Postprandial state and atherosclerosis. Curr Opin Lipidol. 1995; 6: 286–290.CrossrefMedlineGoogle Scholar
  • 42 Karpe F, Steiner G, Uffelman K, et al. Postprandial lipoproteins and progression of coronary atherosclerosis. Atherosclerosis. 1994; 106: 83–97.CrossrefMedlineGoogle Scholar
  • 43 Mack WJ, Krauss RM, Hodis HN. Lipoprotein subclasses in the Monitored Atherosclerosis Regression Study (MARS): treatment effects and relation to coronary angiographic progression. Arterioscler Thromb Vasc Biol. 1996; 16: 697–704.CrossrefMedlineGoogle Scholar
  • 44 Assmann G, Schulte H, von Eckardstein A. Hypertriglyceridemia and elevated lipoprotein(a) are risk factors for major coronary events in middle-aged men. Am J Cardiol. 1996; 77: 1179–1184.CrossrefMedlineGoogle Scholar
  • 45 Thrift AG, Dewey HM, Macdonell RA, et al. Incidence of the major stroke subtypes: initial findings from the North East Melbourne Stroke Incidence Study (NEMESIS). Stroke. 2001; 32: 1732–1738.CrossrefMedlineGoogle Scholar
  • 46 Hellerstein MK. Carbohydrate-induced hypertriglyceridemia: modifying factors and implications for cardiovascular risk. Curr Opin Lipidol. 2002; 13: 33–40.CrossrefMedlineGoogle Scholar

eLetters(0)

eLetters should relate to an article recently published in the journal and are not a forum for providing unpublished data. Comments are reviewed for appropriate use of tone and language. Comments are not peer-reviewed. Acceptable comments are posted to the journal website only. Comments are not published in an issue and are not indexed in PubMed. Comments should be no longer than 500 words and will only be posted online. References are limited to 10. Authors of the article cited in the comment will be invited to reply, as appropriate.

Comments and feedback on AHA/ASA Scientific Statements and Guidelines should be directed to the AHA/ASA Manuscript Oversight Committee via its Correspondence page.