Hypertriglyceridemia is a lipid abnormality prevalent in 1/3 of the United States adult population. Our objective was to describe the independent contribution of hypertriglyceridemia to medical care costs. Using an observational cohort of 108,324 members of Kaiser Permanente Northwest, we analyzed the electronic medical records of those patients aged ≥18 years who had triglyceride (TG) measurements in 2008 and had been members of Kaiser Permanente Northwest for the entire year. After assigning patients to TG categories of <150, 150 to 199, 200 to 499, and ≥500 mg/dl, we compared the annual direct medical costs. To isolate the independent contribution of the TG levels, we adjusted the costs for age, gender, body mass index, blood pressure, smoking history, low-density lipoprotein and high-density lipoprotein cholesterol, and health conditions such as cardiovascular disease, diabetes, and renal disease. Of the 108,324 study subjects, 64.1% had normal TG levels (<150 mg/dl), 16.4% had borderline high levels (150 to 199 mg/dl), 18.0% had high TG levels (200 to 499 mg/dl), and 1.5% had very high TG levels (≥500 mg/dl). After adjustment, the patients with TG levels ≥500 mg/dl (severe hypertriglyceridemia) had significantly greater mean total costs ($8,567, 99% confidence interval $7,034 to $10,100) than those with levels <150 mg/dl ($6,186, 99% confidence interval $6,058 to $6,314), 150 to 199 mg/dl ($6,449, 99% confidence interval $6,196 to $6,702), or 200 to 499 mg/dl ($6,376, 99% confidence interval $6,118 to $6,634). The differences were driven by both outpatient and pharmaceutical costs. The inpatient costs were also greater for those with TG levels ≥500 mg/dl, but the difference did not reach statistical significance. In conclusion, severe hypertriglyceridemia was associated with 33% to 38% greater medical costs per annum, independent of resource-intensive conditions such as cardiovascular disease, heart failure, hypertension, and diabetes.
Given the clear association between triglyceride (TG) levels and resource-intensive conditions such as cardiovascular disease and diabetes mellitus, it is likely that patients with elevated TG concentrations would generate a considerable burden on the healthcare system. To our knowledge, however, the medical care use and costs associated with TG levels have not been described. Furthermore, isolating the independent contribution of TG levels to medical costs might help inform the debate of whether and at what level the TG concentration ought to be treated. Therefore, we sought to fully describe and compare the demographic and clinical characteristics and the annual medical care use and costs of patients with normal, borderline, high, and severe hypertriglyceridemia (SHTG).
Methods
The study site was Kaiser Permanente Northwest (KPNW), a 480,000-member group model health maintenance organization. We used an observational study design that capitalized on the comprehensive medical use data maintained by KPNW, including an electronic medical record of all patient encounters, laboratory results that were analyzed by a single regional laboratory using standardized methods, and dispenses from the pharmacies located at all clinics. The institutional review board of the Kaiser Permanente Center for Health Research reviewed and approved the present study.
The TG levels are often measured as a part of routine lipid screening conducted within KPNW. For the present analysis, we identified all 108,324 patients aged ≥18 years who had TG measurements in 2008 and had been members of KPNW for the entire year. If multiple measurements were taken, we selected the first available. We assigned the patients to TG categories of <150, 150 to 199, 200 to 499, and ≥500 mg/dl, as defined by the National Cholesterol Education Program report.
We calculated age as of the date of the TG test. Gender, race, smoking history, height, weight, and blood pressure were obtained from the electronic medical records. Other lipid values were extracted from the laboratory database, as were the fasting glucose, hemoglobin A1c (for patients with diabetes), and serum creatinine, from which we estimated the glomerular filtration rate. From the diagnoses contained in the electronic medical record, we identified the co-morbidities ( International Classification of Diseases, 9th revision, Clinical Modification codes) present during or before 2008: myocardial infarction (410.x, 412), angina and acute coronary syndrome (411.1, 411.81, 411.89), other ischemic heart disease (413.x, 414.x), stroke (430.xx–432.xx, 434.xx–436.xx, 437.1), peripheral vascular disease (440.x, 441.x, 444.x), revascularization surgery (V45.81, V45.82, V45.89), dyslipidemia (272.0, 272.1, 272.2, 272.4, 272.7), hypertension (401.x–405.x), heart failure (428.x), dysrhythmias (427.x), diabetes (250.x), chronic renal failure (585.x), depression (296.2, 296.3, 300.4, 309.1, 311), and malignant neoplasms (140.x–172.x, 174.x to 209.x). We identified the prescriptions dispensed at any point during 2008 from the pharmacy database.
The total direct, inpatient, outpatient, and pharmacy medical costs were calculated for each subject included in the present study during 2008. We based our costing method on the procedures developed and validated by the Kaiser Permanente Center for Health Research. For outpatient costs, this method creates standard costs for office visits by specialty/department and type of clinician (physician vs physician assistant or nurse practitioner). The number of visits per department per clinician type was then multiplied by the appropriate unit cost. The pharmaceutical costs were determined from the retail prices within the service area. Hospitalizations were assigned to the diagnosis-related groups according to the primary reason for the hospitalization. The average daily rate per diagnosis-related group was then multiplied by the length of stay. The costs for medical services incurred at facilities not owned by KPNW were determined by the amount paid by KPNW to the nonplan provider. These methods ensured that although the costs reported in the present study might be specific to KPNW, they have approximated the charges a nonmember would be billed if these same services were purchased from KPNW.
We compared the demographic and clinical characteristics, health conditions, and the use of selected pharmaceutical agents across the TG categories using analysis of variance. Because of the large sample size and numerous comparisons, we required a p value of <0.001 to indicate statistical significance. We used simple ordinary least-squares regression analysis to estimate the independent contribution of TG levels to the inpatient, outpatient, pharmaceutical, and total costs while controlling for age, gender, body mass index, blood pressure, smoking history, and health conditions. Previous research in this setting has demonstrated that ordinary least-squares regression analysis predicts the costs at least as well as more sophisticated techniques. The cost data were not normally distributed. Log transformation normalized the data but did not affect the significance level of the variables or the performance of the models. Therefore, for ease of interpretation, we have reported the mean costs adjusted for these factors, calculated using the LSMEANS options in PROC GLM of Statistical Analysis Systems, version 8.2 (SAS Institute, Cary, North Carolina). We had full access to the data and take responsibility for its integrity, and all have read and agreed to the report as written.
Results
The characteristics of the 108,324 study subjects are listed in Table 1 , stratified by TG level (normal, <150 mg/dl; borderline high, 150 to 199 mg/dl; high, 200 to 499 mg/dl; and SHTG, ≥500 mg/dl). Table 2 lists the proportion of patients in each TG category with resource-intensive medical conditions. The use of pharmaceutical agents is listed in Table 3 .
| Variable | TG Category (mg/dl) | |||
|---|---|---|---|---|
| <150 | 150–199 | 200–499 | ≥500 | |
| Patients (n) | 69,442 (64%) | 17,788 (16%) | 19,503 (18%) | 1,591 (2%) |
| Age, years | 58.2 ± 14.2 ⁎ | 58.9 ± 13.1 | 57.8 ± 12.6 ⁎ | 53.9 ± 11.3 |
| Men | 46% | 49% | 53% | 68% |
| Nonwhite | 11% ⁎ | 10% † | 9% † | 11% ⁎ † |
| Ever smoked | 46% | 50% | 53% | 59% |
| Body mass index (kg/m 2 ) | 29.6 ± 6.9 | 32.4 ± 7.0 | 33.0 ± 6.7 ⁎ | 33.0 ± 6.3 ⁎ |
| Systolic blood pressure (mm Hg) | 127 ± 14 | 130 ± 13 | 131 ± 13 ⁎ | 132 ± 13 ⁎ |
| Diastolic blood pressure (mm Hg) | 74 ± 8 | 76 ± 8 | 77 ± 9 | 78 ± 9 |
| Total cholesterol (mg/dl) | 184 ± 39 | 194 ± 42 | 204 ± 45 | 240 ± 70 |
| Low-density lipoprotein cholesterol (mg/dl) | 113 ± 34 ⁎ | 117 ± 37 | 113 ± 39 ⁎ | 105 ± 42 |
| High-density lipoprotein cholesterol (mg/dl) | 51 ± 14 | 43 ± 10 | 40 ± 9 | 34 ± 9 |
| Triglycerides (mg/dl) | 95 ± 29 | 172 ± 14 | 275 ± 68 | 830 ± 617 |
| Fasting glucose (mg/dl) | 98 ± 21 | 105 ± 30 | 111 ± 38 | 132 ± 65 |
| Hemoglobin A1c (%) | 6.9 ± 1.4% | 7.2 ± 1.5% | 7.4 ± 1.6% | 8.1 ± 2.1% |
| Glomerular filtration rate (ml/min) | 89 ± 26 | 86 ± 26 ⁎ | 87 ± 28 ⁎ | 96 ± 41 |
⁎ Not statistically significant against data also followed by asterisk.
† Not statistically significant against data also followed by cross.
| Variable | TG Category (mg/dl) | |||
|---|---|---|---|---|
| <150 | 150–199 | 200–499 | ≥500 | |
| Myocardial infarction | 7% ⁎ | 8% † | 8% † | 8% ⁎ † |
| Angina/acute coronary syndrome | 5% ⁎ | 6% † | 6% † | 5% ⁎ † |
| Other ischemic heart disease | 14% ⁎ | 15% † | 16% † | 15% ⁎ † |
| Stroke | 7% ⁎ | 7% ⁎ | 7% ⁎ | 7% ⁎ |
| Peripheral vascular disease | 4% ⁎ | 5% ⁎ † | 5% † | 4% ⁎ † |
| Coronary artery bypass grafting | 4% ⁎ | 5% ⁎ | 4% ⁎ | 5% ⁎ |
| Percutaneous transluminal coronary angioplasty | 4% ⁎ | 4% † | 5% † | 5% ⁎ † |
| Any cardiovascular disease | 20% ⁎ | 22% † | 23% † | 22% ⁎ † |
| Dyslipidemia | 47% | 64% | 70% | 87% |
| Hypertension | 46% | 59% | 63% ⁎ | 65% ⁎ |
| Heart failure | 6% ⁎ | 7% † | 7% † | 7% ⁎ † |
| Dysrhythmia | 14% ⁎ | 15% ⁎ | 14% ⁎ | 13% ⁎ |
| Diabetes | 19% | 29% | 36% | 51% |
| Chronic renal failure | 7% | 9% | 10% ⁎ | 11% ⁎ |
| Depression | 26% | 30% | 32% ⁎ | 34% ⁎ |
| Malignant neoplasms ‡ | 11% ⁎ | 12% ⁎ | 11% ⁎ | 10% ⁎ |
⁎ Not statistically significant against data also followed by asterisk.
† Not statistically significant against data also followed by cross.
| Variable | TG Category (mg/dl) | |||
|---|---|---|---|---|
| <150 | 150–199 | 200–499 | ≥500 | |
| Antihyperlipidemic agents | ||||
| Fibrates | 1% | 2% | 5% | 34% |
| Statins | 38% | 51% ⁎ | 53% † | 54% ⁎ † |
| Ezetimibe | 1% | 2% ⁎ | 2% ⁎ | 2% ⁎ |
| Bile sequestrants | 1% ⁎ | 1% † | 1% † | 1% ⁎ † |
| Antihypertensive agents | ||||
| Angiotensin-converting enzyme inhibitors | 26% | 33% | 37% | 42% |
| Angiotensin receptor blockers | 8% | 10% ⁎ | 11% ⁎ | 13% |
| Diuretics | 24% | 32% ⁎ | 35% ⁎ | 33% ⁎ |
| β Blockers | 25% | 33% | 37% | 38% |
| Calcium channel blockers | 10% ⁎ | 12% † | 13% † | 11% ⁎ † |
| Antihyperglycemic agents | ||||
| Metformin | 10% | 17% | 22% | 36% |
| Sulfonylureas | 6% | 11% | 15% | 23% |
| Insulin | 5% | 7% | 8% | 17% |
| Thiazolidinediones | 1% | 1% ⁎ | 1% ⁎ | 2% |
| Antidepressant agents | 24% | 28% | 31% ⁎ | 34% ⁎ |
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
