Article type
Year
Abstract
Introduction: Fish oil (FO) may lower triglycerides (Tg) in patients with type 2 diabetes mellitus (DM2) and dyslipidemia but, as indicated by observational studies, it may deteriorate glycemic control.
Objectives: 1. To complete a Cochrane systematic review on the effects of FO supplementation in subjects with DM2. 2. To describe the problems encountered during this process.
Methods: We conducted a systematic review of RCTs of FO supplementation versus placebo in subjects with DM2. Using a combination of search strategies (MEDLINE, EMBASE, CCTR, hand-searching and expert input) updated through 9/1998 we identified 158 articles of which 63 were considered relevant based on title. 18 RCTs were included in the final analysis (11 crossover and 7 parallel group design). We independently assessed study quality, abstracted and pooled the data using a random effects model estimating the weighted mean difference for fasting glucose, glycosylated hemoglobin (gHb), total cholesterol, HDL and LDL cholesterol and Tg. To explain heterogeneity sensitivity analyses were conducted for trial design and quality, baseline Tg level, fish oil dose and duration of the intervention. A funnel plot was used to seek for publication bias.
Results: The pooled data set included 823 subjects (range 8 to 418) followed for a mean of 12 weeks (range 2 to 52). Doses of FO ranged from 1.7 to 10 g/d. 10 studies were of high quality. None of the 11 crossover RCTs reported period-specific data. Of the 8 crossover RCTs reporting gHb, 3 had a washout phase and 4 reported on the presence of carry-over effect. 10 of 15 RCTs reporting gHb, an integrative marker of 12 weeks of glycemic control, had a duration of 8 weeks or less. Meta-analysis demonstrated a significant effect of FO on lowering Tg (-49.6 mg/dL 95% CI - 63.1 to 36.1) and raising LDL cholesterol (+8.2 mg/dL 95% CI 0.6 to 15.8). No significant effect was observed on fasting plasma glucose (4.7 mg/dL 95% CI -1.5 to 10.9), gHb, total or HDL cholesterol. The effects on Tg and LDL were larger in the RCTs that recruited hypertriglyceridemic subjects and used the higher doses of FO. These studies account for the heterogeneity observed for the Tg and LDL outcomes. No significant publication bias was detected.
Discussion: 1. FO supplementation in DM2 decreases Tg without a statistically significant effect on glycemic control. 2. RCTs assessing the long term effects of FO supplementation on morbidity and mortality in DM2 are needed. 3. The crossover trial design, very prevalent in DM2 research, is frequently reported in a way that does not allow pooling, the methodology of which is still in evolution. 4. Glycosylated hemoglobin is used in crossover studies of short duration with neither adequate washout periods nor an assessment of carry-over effect.
Objectives: 1. To complete a Cochrane systematic review on the effects of FO supplementation in subjects with DM2. 2. To describe the problems encountered during this process.
Methods: We conducted a systematic review of RCTs of FO supplementation versus placebo in subjects with DM2. Using a combination of search strategies (MEDLINE, EMBASE, CCTR, hand-searching and expert input) updated through 9/1998 we identified 158 articles of which 63 were considered relevant based on title. 18 RCTs were included in the final analysis (11 crossover and 7 parallel group design). We independently assessed study quality, abstracted and pooled the data using a random effects model estimating the weighted mean difference for fasting glucose, glycosylated hemoglobin (gHb), total cholesterol, HDL and LDL cholesterol and Tg. To explain heterogeneity sensitivity analyses were conducted for trial design and quality, baseline Tg level, fish oil dose and duration of the intervention. A funnel plot was used to seek for publication bias.
Results: The pooled data set included 823 subjects (range 8 to 418) followed for a mean of 12 weeks (range 2 to 52). Doses of FO ranged from 1.7 to 10 g/d. 10 studies were of high quality. None of the 11 crossover RCTs reported period-specific data. Of the 8 crossover RCTs reporting gHb, 3 had a washout phase and 4 reported on the presence of carry-over effect. 10 of 15 RCTs reporting gHb, an integrative marker of 12 weeks of glycemic control, had a duration of 8 weeks or less. Meta-analysis demonstrated a significant effect of FO on lowering Tg (-49.6 mg/dL 95% CI - 63.1 to 36.1) and raising LDL cholesterol (+8.2 mg/dL 95% CI 0.6 to 15.8). No significant effect was observed on fasting plasma glucose (4.7 mg/dL 95% CI -1.5 to 10.9), gHb, total or HDL cholesterol. The effects on Tg and LDL were larger in the RCTs that recruited hypertriglyceridemic subjects and used the higher doses of FO. These studies account for the heterogeneity observed for the Tg and LDL outcomes. No significant publication bias was detected.
Discussion: 1. FO supplementation in DM2 decreases Tg without a statistically significant effect on glycemic control. 2. RCTs assessing the long term effects of FO supplementation on morbidity and mortality in DM2 are needed. 3. The crossover trial design, very prevalent in DM2 research, is frequently reported in a way that does not allow pooling, the methodology of which is still in evolution. 4. Glycosylated hemoglobin is used in crossover studies of short duration with neither adequate washout periods nor an assessment of carry-over effect.