The main findings of this study are a significant improvement in all lipid variables in the entire follow-up period (> 10 years), with the attenuation of the changes from 1 to 2 years post-surgery to end of observation period except for HDL and total-/HDL-cholesterol-ratio which has stable values from 5 to 10 years. Long-term weight loss was substantial, and lipid profile improved more in patients with greater weight loss. Additionally, patients who had DMII preoperatively had a greater improvement in overall changes in lipid values at 12 years. This relation has not previously been described in a long-term perspective.

In accordance with results from other long-term studies, a statistically significant improvement on weight and lipid levels was documented [12, 20,21,22,23]. Three studies have previously described a peak in both lipid changes and weight loss after 2 years, similar to nadir found in this study [12, 19, 20]. LDL, total cholesterol, and TG show a continuous attenuation towards baseline levels but still have an overall improved profile at 12 years, with 1/4 of the initial (nadir) effect sustained for LDL, 1/2 for total cholesterol, and 2/3 for TG. The relation between weight/BMI and lipid variables complies with previous research with follow-up for up to 5 years [12]. Weight loss followed the same development as lipid changes. At 12 years, %TWL was 21.2%, which corresponds with findings in other studies [12, 20, 21].

When exploring relations to other parameters, BMI and %TWL showed significant relationship to end of observation lipid levels. For changes in TG and HDL, %TWL at end of follow-up proved significant, whilst BMI did not. This might imply percentage of body mass lost is more significant than the exact BMI at the end of the follow-up. To our knowledge, none of the long-term studies (> 10 years) has examined the relationship between changes in lipid levels at the end of follow-up with other parameters. Gero et al. point out the coherence between weight loss and lipid improvement after 5 years, with a greater improvement in patients with greater weight loss [12]. These findings are extended with our study with an even longer observation period. While this study explores the relation between selected parameters and lipid values, increasing age during the follow-up period can also partially explain the gradual increase in lipid levels post nadir.

HDL and total-/HDL-cholesterol ratio changes did not follow weight development the same way as remaining lipid variables, presenting stable values from 5 to 10 years post-surgery (Figs. 2 and 3). This finding is supported by previous research [12, 20]. HDL is influenced by other factors such as physical activity and a healthier diet, which might be reasonable to assume for bariatric surgery patients [29].

Compared to guideline thresholds, there was a 12% and 81% reduction from baseline to ES study in patients with lipid levels who did not meet ESC and Norwegian guidelines. These improvements are supported by other long-term studies where dyslipidemia prevalence decreased by 46% and 52.4% 10 years after RYGB [12, 21]. The total-/HDL-cholesterol ratio improved from 4.48 (± 1.17) at baseline to 3.10 (± 0.78) at end of study (p < 0.001; Table 2), also supporting an improved CVD risk profile (Fig. 2).

For patients with preoperative DMII, there was a greater enhancement of absolute values for HDL, total cholesterol, and TG; this relation has to our knowledge never been described in a 10-year perspective. During this study, there was considerable improvement with a 59% decrease in DMII from baseline to ES. Other studies with 10–12-year follow-up time have found similar results with 51% and 54.2% remission rates [21, 23]. Kim et al. argue that DMII’s relation with metabolic syndrome can explain how this patient group can have different premises for abnormalities in lipid profile [30]. DMII remission can occur in response to relative weight loss and give changes in metabolism in patients who remain overweight. This might provide an explanation on how DMII remission can be an influential factor on end of follow-up lipid variables.

The BAROBS data consists of a patient cohort with a long observation time with nearly no loss to follow-up throughout the period, and information on the use of lipid-lowering drugs both at baseline and at the end of follow-up was available.

The greatest weakness of the study is the various use of methods for analyzing lipid values during the study period. However, an analysis between methods was performed, and the overall conclusion was that these methodological changes underestimated the true changes of LDL and HDL. For TG, there was both a slight overestimation and underestimation at different hospitals, and for total cholesterol, there was next to no divergence between methods. Thus, for all practical purposes, this has decreased the positive changes for LDL and HDL, and there should be a minor risk of overestimating results due to bias in the analytical methods used during the observation period.

Seventy-two percent of patients had LDL levels > 2.6 mmol/L at baseline, and 28% had above 3.5 mmol/L, but we have not performed any sub-group analyses to explore if there is any difference in response between patients with and without hyperlipidemia.

This study has limited information about other factors which also can influence lipid variables, such as lifestyle changes or other baseline values than those registered in the BAROBS study.

Finally, this is an observational study, with the limitations and weaknesses of this design.