The association between the changes in high-density lipoprotein (HDL) cholesterol and the risk of cardiovascular (CVD) or cerebrovascular hospitalization among patients with type 2 diabetes remains unclear. We conducted a retrospective observational cohort study of 30,067 members of the Kaiser Permanente Northwest and Georgia regions, who had type 2 diabetes and 2 HDL cholesterol measurements 6 to 24 months apart in 2001 to 2006. We followed up the cohort for ≤8 years (through 2009) to determine whether the change in HDL cholesterol was associated with subsequent CVD hospitalization. We examined the HDL cholesterol change continuously and by 3 categories: HDL cholesterol increased ≥6.5 mg/dl, decreased ≥6.5 mg/dl, or remained within ±6.4 mg/dl. The Cox regression models were adjusted for the baseline HDL cholesterol and demographic and clinical risk factors. During a mean follow-up of 55.8 ± 23.8 months, 3,023 patients (10.1%) experienced a CVD hospitalization. After multivariate adjustment, each 5 mg/dl of baseline HDL cholesterol was significantly associated with a 6% lower CVD hospitalization risk (hazard ratio 0.94 per 5 mg/dl, 95% confidence interval 0.92 to 0.95, p <0.0001) and each 5-mg/dl increase in HDL cholesterol was associated with a 4% CVD risk reduction (hazard ratio 0.96, 95% confidence interval 0.94 to 0.99, p <0.003). In the categorical analysis, a ≥6.5-mg/dl HDL cholesterol decrease was associated with an 11% increased CVD risk (hazard ratio 1.11, 95% confidence interval 1.00 to 1.24, p = 0.047) and a ≥6.5-mg/dl increase was associated with an 8% CVD risk reduction (hazard ratio 0.92, 95% confidence interval 0.84 to 1.01, p = 0.077) relative to those with stable HDL cholesterol. In conclusion, our results add to the growing body of evidence that increasing the HDL cholesterol levels might be an important strategy for CVD risk reduction. The prevention of HDL cholesterol decreases could be equally important.
The lifetime risk of cardiovascular disease (CVD) is as great as 87% among patients with type 2 diabetes. Low-density lipoprotein (LDL) cholesterol reduction with statins remains a priority for reducing the CVD risk. However, even in tightly controlled clinical trials of intensive statin therapy, ≥10% of the subjects experienced a major cardiovascular event. Although glycemic control is the cornerstone of diabetes management, intensive antihyperglycemic therapy has not been shown to further aid in CVD risk reduction. Thus, treatment targets other than LDL cholesterol and hemoglobin A1c need consideration. One such target is high-density lipoprotein (HDL) cholesterol. The American Diabetes Association has noted that targeting HDL cholesterol has intuitive appeal (especially when the level is <40 mg/dl) but acknowledges that the evidence base of HDL cholesterol therapy is lacking. Epidemiologic evidence suggests that low HDL cholesterol is associated with increased CVD risk. A low HDL cholesterol is present in about ½ of patients with type 2 diabetes, and it has been theorized that normalizing HDL cholesterol would reduce the CVD mortality rate by 23% to 42%. However, the clinical research to date has not clearly demonstrated that increasing the HDL cholesterol level will reduce CVD events, perhaps because safe and effective agents that substantially increase the HDL cholesterol level remain elusive. We undertook a longitudinal observational cohort study to determine whether changes in HDL cholesterol were associated with a subsequent risk of CVD events in a population-based sample.
Methods
Kaiser Permanente Northwest and Kaiser Permanente Georgia are integrated healthcare delivery systems serving approximately 475,000 and 235,000 members, centered in the Portland, Oregon and Atlanta, Georgia metropolitan areas, respectively. Both organizations maintain electronic medical records and other electronic databases that capture nearly 100% of their members’ medical care use. Kaiser Permanente Northwest and Kaiser Permanente Georgia also use similarly designed diabetes registries, with entrance criteria that includes an inpatient or outpatient diagnosis of diabetes, the receipt of an antihyperglycemic drug, or a fasting glucose value >125 mg/dl. Combining the data, we identified 33,205 subjects with data entered into the registries before 2004 and had 2 HDL cholesterol measurements 6 to 24 months apart after registry entry and during 2001 to 2006. We excluded 2,690 patients (8.1%) who did not have ≥6 months of health plan eligibility after the second HDL cholesterol measurement. We also excluded 448 subjects (1.3%) who had implausible values of HDL cholesterol changes (>25 mg/dl), for a final study sample of 30,067.
We defined the first HDL cholesterol measurement as the baseline value and calculated the change in HDL cholesterol by subtracting the baseline value from the second measurement (6 to 24 months later). We examined the HDL cholesterol change as both a continuous and a categorical measure, using a cutpoint of ±6.5 mg/dl for categorization that optimized the discrimination and predictive power. The patients were followed up for ≤8 years from the date of the second HDL cholesterol measurement until they died, left the health plan for other reasons, or until December 31, 2009. The outcome of interest was hospitalization with a primary discharge diagnosis of coronary artery disease ( International Classification of Diseases, 9th revision, Clinical Modification [ICD-9-CM] codes 410.x, 411.x, 413.x, 414.x), hemorrhagic stroke (ICD-9-CM codes 430.x, 431.x, 432.x), ischemic stroke (ICD-9-CM codes 434.x, 435.x), or other acute but ill-defined cerebrovascular disease (ICD-9-CM code 436.x) after the second HDL cholesterol measurement.
We first estimated the Cox regression models of the interval to CVD hospitalization using only the baseline and HDL cholesterol change as predictors in the model. Next, to assess the independent association of HDL cholesterol change with CVD risk, we adjusted for the baseline values of age, gender, race, duration of diabetes, body mass index, systolic blood pressure, hemoglobin A1c, LDL cholesterol, and triglycerides; history of cardiovascular disease, cerebrovascular disease, heart failure, chronic kidney disease, neuropathy, retinopathy, and depression; and the use of angiotensin-converting enzyme inhibitors, angiotensin receptor blockers, β blockers, other antihypertensive agents, metformin, sulfonylureas, thiazolidinediones, insulin, other antihyperglycemic agents, statins, fibrates, niacin, and other antihyperlipidemic agents. We did not adjust for the site because of its high correlation with race. Including the mean interval between the baseline and follow-up HDL cholesterol measures in the models did not affect the results. The body mass index data were missing for 10% of subjects, the blood pressure data for 9%, the hemoglobin A1c data for 3.5%, LDL cholesterol data for 5%, and triglyceride data for 8%. Thus, we imputed the missing data by multiple imputation using the MI and MIANALYZE procedures in SAS, and all analyses were conducted using SAS, version 9.1 (SAS Institute, Cary, North Carolina).
Results
Of the 30,067 study subjects, the HDL cholesterol levels remained within ±6.4 mg/dl for 61.4% (n = 18,449), increased by ≥6.5 mg/dl for 21.6% (n = 6,488), and decreased by ≥6.5 mg/dl for 17.0% (n = 5,130). Statistically significant differences were found in the demographic and clinical characteristics between the groups for a number of covariates ( Table 1 ). The baseline co-morbidities among the 3 groups are listed in Table 2 , and pharmaceutical use is listed in Table 3 . During a mean follow-up of 4.5 ± 2.1 years, a total of 3,023 subjects (10.1%) experienced CVD hospitalization. The results of the Cox regression models of the baseline and change in HDL cholesterol (both as a continuous and categorical variable) as predictors of CVD hospitalization, with and without adjustment for other covariates, are listed in Table 4 .
| Variable | HDL Cholesterol Increased ≥6.5 mg/dl (n = 6,488) | HDL Cholesterol Remained at ±6.4 mg/dl (n = 18,449) | HDL Cholesterol Decreased ≥6.5 mg/dl (n = 5,130) |
|---|---|---|---|
| % Kaiser Permanente Georgia ⁎ † | 33% | 31% | 37% |
| Age (years) | 60.7 ± 12.5 | 60.8 ± 12.5 | 61.4 ± 12.7 |
| Patients aged ≥65 years | 37% | 38% | 39% |
| Men ⁎ † ‡ | 50% | 54% | 44% |
| African-American ⁎ † | 15% | 14% | 19% |
| Current smoker | 11% | 11% | 10% |
| Duration of diabetes (years) | 5.4 ± 4.3 | 5.4 ± 4.3 | 5.6 ± 4.5 |
| Body mass index (kg/m 2 ) † ‡ | 33.0 ± 7.2 | 33.6 ± 7.5 | 33.1 ± 7.5 |
| Mean systolic blood pressure (mm Hg) | 134 ± 16 | 135 ± 16 | 135 ± 16 |
| Mean diastolic blood pressure (mm Hg) | 77 ± 9 | 77 ± 9 | 76 ± 10 |
| Hemoglobin A1c (%) ⁎ † ‡ | 7.3 ± 1.7 | 7.4 ± 1.6 | 7.6 ± 1.8 |
| Estimated glomerular filtration rate (ml/min/1.73 m 2 ) ⁎ † | 82 ± 30 | 82 ± 30 | 78 ± 30 |
| Total cholesterol (mg/dl) ⁎ † ‡ | 193 ± 43 | 184 ± 40 | 187 ± 45 |
| Low-density lipoprotein cholesterol (mg/dl) ‡ | 108 ± 34 | 106 ± 33 | 107 ± 36 |
| Triglycerides (mg/dl) † | 202 ± 201 | 194 ± 159 | 203 ± 198 |
| High-density lipoprotein cholesterol (mg/dl) | |||
| Prebaseline measure ⁎ † ‡ | 41 ± 12 | 43 ± 11 | 54 ± 14 |
| Baseline measure ⁎ † ‡ | 52 ± 13 | 43 ± 11 | 42 ± 13 |
| Mean change ⁎ † ‡ | 11.0 ± 4.0 | 0.3 ± 3.5 | −11.3 ± 4.1 |
| Percentage of change ⁎ † ‡ | 29.3 ± 14.2% | 1.1 ± 8.9% | −21.7 ± 7.6% |
| Months between measures † ‡ | 17.9 ± 4.5 | 18.0 ± 4.5 | 18.5 ± 4.4 |
⁎ Group with HDL cholesterol increasing ≥6.5 mg/dl differed from group with HDL cholesterol decreasing ≤6.5 mg/dl, p <0.001.
† Group with HDL cholesterol remaining at ±6.4 mg/dl differed from group with HDL cholesterol decreasing ≤6.5 mg/dl, p <0.001.
‡ Group with HDL cholesterol increasing ≥6.5 mg/dl differed from group with remaining at ±6.4 mg/dl, p <0.001.
| Variable | HDL Cholesterol Increased ≥6.5 mg/dl | HDL Cholesterol Remained at ±6.4 mg/dl | HDL Cholesterol Decreased ≥6.5 mg/dl |
|---|---|---|---|
| Myocardial infarction ⁎ | 12% | 11% | 10% |
| Angina pectoris | 8% | 8% | 8% |
| Other ischemic heart disease ⁎ | 26% | 24% | 23% |
| Revascularization ⁎ † | 10% | 9% | 7% |
| Any of previously listed coronary heart disease | 34% | 32% | 32% |
| Cerebrovascular disease ⁎ | 11% | 10% | 12% |
| Renal disease (glomerular filtration rate <60 ml/min) ⁎ † | 28% | 26% | 32% |
| Heart failure ‡ | 13% | 12% | 15% |
| Neuropathy † | 23% | 22% | 24% |
| Retinopathy ⁎ † | 28% | 26% | 33% |
| Depression † ‡ | 26% | 24% | 27% |
⁎ Group with HDL cholesterol increasing ≥6.5 mg/dl differed from group with HDL cholesterol decreasing ≤6.5 mg/dl, p <0.001.
† Group with HDL cholesterol remaining at ±6.4 mg/dl differed from group with HDL cholesterol decreasing ≤6.5 mg/dl, p <0.001.
‡ Group with HDL cholesterol increasing ≥6.5 mg/dl differed from Group with HDL cholesterol remaining at ±6.4 mg/dl, p <0.001.
| Variable | HDL Cholesterol Increased ≥6.5 mg/dl | HDL Cholesterol Remained at ±6.4 mg/dl | HDL Cholesterol Decreased ≥6.5 mg/dl |
|---|---|---|---|
| Antihypertensive agents (%) | |||
| Angiotensin-converting enzyme inhibitors/angiotensin receptor blockers ⁎ | 55% | 58% | 57% |
| β Blockers | 30% | 30% | 31% |
| Other antihypertensive agents † ‡ | 42% | 41% | 46% |
| Antihyperglycemic agents (%) | |||
| Metformin ‡ | 36% | 37% | 33% |
| Sulfonylureas ⁎ | 38% | 42% | 40% |
| Thiazolidinediones | |||
| Pioglitazone | 4% | 3% | 3% |
| Rosiglitazone | <1% | <1% | <1% |
| Any thiazolidinedione | 4% | 3% | 3% |
| Other oral antihyperglycemic agents | 1% | 1% | 1% |
| Insulin † ‡ | 20% | 19% | 24% |
| Antihyperlipidemic agents (%) | |||
| Statins | 47% | 48% | 47% |
| Fibrates ⁎ † | 6% | 3% | 4% |
| Niacin | <1% | <1% | <1% |
| Other lipid agents | 1% | 1% | 1% |
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
