Fish Oil vs. Drugs for Hypertriglyceridemia in HIV+ Patients
Fish Oil vs. Drugs for Hypertriglyceridemia in HIV+ Patients
The Centers for AIDS Research Network of Integrated Clinical Systems (CNICS) Cohort is a prospective longitudinal observational cohort of HIV-infected patients receiving clinical care from January 1995 to the present. The CNICS data repository captures longitudinal comprehensive clinical data on the CNICS cohort from outpatient and inpatient encounters. Demographic, clinical, medication, and laboratory data are obtained from each electronic health record site and other institutional data sources. Laboratory data are uploaded directly from the Clinical Laboratory Systems at each site. Medication data are entered into the electronic health record by clinicians or prescription fill/refill data are uploaded directly from Pharmacy Systems and verified through medical record review. Patients from 5 sites were included in this analysis: University of Alabama at Birmingham, University of Washington, University of California, San Diego, University of California, San Francisco, and the University of North Carolina.
We conducted our analysis using 2 separate stages: (1) a quasi-experimental pre/post design of patients who initiated fish oil and (2) a classic comparative effectiveness new user cohort study with fish oil as the reference medication. The quasi-experimental pre/post design controls for within-subject time invariant confounders when evaluating the effectiveness of fish oil at reducing TG levels within individuals. The new user cohort design uses multivariate regression to adjust for differences in treatment group characteristics that may have a direct impact on the outcome. Furthermore, this study design can control for known biases associated with prevalent users (where differences in pretreatment disease is unknown) by requiring that pretreatment TG levels be available to adjust for differences in pretreatment level between groups. Individuals were included in the analyses if they met the following criteria: 18 years or older; initiated fish oil, fenofibrate, gemfibrozil, or atorvastatin between January 1, 2000 and December 31, 2009; had at least 1 TG result before (within 6 months) and after initiation of therapy; and were not on medications for dyslipidemia for at least 6 months before study entry. Atorvastatin was selected as the reference statin because it is one of the preferred agents for HIV-infected patients with dyslipidemia and is the statin most commonly used in the CNICS cohort. Patients were excluded if the date they initiated lipid-lowering medications or their laboratory result date were missing, if lipid-lowering medications were changed to alternative medications or additional lipid-lowering medications were added before the first posttreatment TG result, or if the posttreatment measure was >1 year after the pretreatment measure. Patients were excluded if their baseline TG value was >810 mg/dL. A level of 810 mg/dL was selected as the upper limit to maintain comparability, as fish oil and atorvastatin were not routinely used in patients with TG levels >810 mg/dL. Patients were also excluded if their baseline TG value was <150 mg/dL because some medications (specifically atorvastatin) were used to treat other conditions besides hypertriglyceridemia.
The primary outcome was the absolute change in TG level. The last TG value recorded pretreatment was compared with the first TG value recorded posttreatment. The posttreatment TG level was obtained at least 6 weeks after treatment to ensure adequate time had elapsed for the therapy to take effect. Fasting status was not always routinely available. Baseline covariates included age, sex, race, height, weight, body mass index (BMI), HIV transmission risk factors, diabetes mellitus status, most recent CD4 cell count, most recent HIV-1 RNA copies, and antiretroviral therapy (ART) at TG therapy initiation [other ART/no ART vs. protease inhibitor (PI)–based vs. non-nucleoside reverse transcriptase inhibitor (NNRTI)–based regimen].
Multivariate analyses of variance were conducted to detect a difference in the absolute change in TG values in patients using fish oil adjusted for age (as a continuous centered variable), sex, race, baseline CD4 cell count (as a continuous centered variable), PI use, and duration between TG measures (as candidate confounder variables). In the self-controlled pre/post quasi-experimental analysis, we adjusted only for age, sex, race, baseline CD4 cell counts, PI use, and duration between TG measures to have a more precise estimate of the TGs-lowering effect of fish oil. Fish oil was used as the reference medication for the comparative effectiveness cohort analysis. Linear regression analyses were used to compare medications while adjusting for baseline differences in age, sex, race, CD4 cell count, PI use, diabetes, BMI, and time between TG measures. We conducted sensitivity analyses among the subset known to be fasting. All statistical analyses were conducted using SAS Version 9.2 (SAS Institute, Inc, Cary, NC). The institutional review board of participating institutions and that of the University of Florida approved the study.
Methods
Data Source
The Centers for AIDS Research Network of Integrated Clinical Systems (CNICS) Cohort is a prospective longitudinal observational cohort of HIV-infected patients receiving clinical care from January 1995 to the present. The CNICS data repository captures longitudinal comprehensive clinical data on the CNICS cohort from outpatient and inpatient encounters. Demographic, clinical, medication, and laboratory data are obtained from each electronic health record site and other institutional data sources. Laboratory data are uploaded directly from the Clinical Laboratory Systems at each site. Medication data are entered into the electronic health record by clinicians or prescription fill/refill data are uploaded directly from Pharmacy Systems and verified through medical record review. Patients from 5 sites were included in this analysis: University of Alabama at Birmingham, University of Washington, University of California, San Diego, University of California, San Francisco, and the University of North Carolina.
Cohort Subjects
We conducted our analysis using 2 separate stages: (1) a quasi-experimental pre/post design of patients who initiated fish oil and (2) a classic comparative effectiveness new user cohort study with fish oil as the reference medication. The quasi-experimental pre/post design controls for within-subject time invariant confounders when evaluating the effectiveness of fish oil at reducing TG levels within individuals. The new user cohort design uses multivariate regression to adjust for differences in treatment group characteristics that may have a direct impact on the outcome. Furthermore, this study design can control for known biases associated with prevalent users (where differences in pretreatment disease is unknown) by requiring that pretreatment TG levels be available to adjust for differences in pretreatment level between groups. Individuals were included in the analyses if they met the following criteria: 18 years or older; initiated fish oil, fenofibrate, gemfibrozil, or atorvastatin between January 1, 2000 and December 31, 2009; had at least 1 TG result before (within 6 months) and after initiation of therapy; and were not on medications for dyslipidemia for at least 6 months before study entry. Atorvastatin was selected as the reference statin because it is one of the preferred agents for HIV-infected patients with dyslipidemia and is the statin most commonly used in the CNICS cohort. Patients were excluded if the date they initiated lipid-lowering medications or their laboratory result date were missing, if lipid-lowering medications were changed to alternative medications or additional lipid-lowering medications were added before the first posttreatment TG result, or if the posttreatment measure was >1 year after the pretreatment measure. Patients were excluded if their baseline TG value was >810 mg/dL. A level of 810 mg/dL was selected as the upper limit to maintain comparability, as fish oil and atorvastatin were not routinely used in patients with TG levels >810 mg/dL. Patients were also excluded if their baseline TG value was <150 mg/dL because some medications (specifically atorvastatin) were used to treat other conditions besides hypertriglyceridemia.
Assessment of Outcome and Other Baseline Covariates
The primary outcome was the absolute change in TG level. The last TG value recorded pretreatment was compared with the first TG value recorded posttreatment. The posttreatment TG level was obtained at least 6 weeks after treatment to ensure adequate time had elapsed for the therapy to take effect. Fasting status was not always routinely available. Baseline covariates included age, sex, race, height, weight, body mass index (BMI), HIV transmission risk factors, diabetes mellitus status, most recent CD4 cell count, most recent HIV-1 RNA copies, and antiretroviral therapy (ART) at TG therapy initiation [other ART/no ART vs. protease inhibitor (PI)–based vs. non-nucleoside reverse transcriptase inhibitor (NNRTI)–based regimen].
Statistical Analyses
Multivariate analyses of variance were conducted to detect a difference in the absolute change in TG values in patients using fish oil adjusted for age (as a continuous centered variable), sex, race, baseline CD4 cell count (as a continuous centered variable), PI use, and duration between TG measures (as candidate confounder variables). In the self-controlled pre/post quasi-experimental analysis, we adjusted only for age, sex, race, baseline CD4 cell counts, PI use, and duration between TG measures to have a more precise estimate of the TGs-lowering effect of fish oil. Fish oil was used as the reference medication for the comparative effectiveness cohort analysis. Linear regression analyses were used to compare medications while adjusting for baseline differences in age, sex, race, CD4 cell count, PI use, diabetes, BMI, and time between TG measures. We conducted sensitivity analyses among the subset known to be fasting. All statistical analyses were conducted using SAS Version 9.2 (SAS Institute, Inc, Cary, NC). The institutional review board of participating institutions and that of the University of Florida approved the study.
Source...