The present study contributes to the still limited knowledge on the long-term (i.e., ≥10 years) HRQoL development following bariatric surgery and its pre-operative predictors. Compared to control patients, surgical patients experienced greater 15-year improvements in mental, physical, social, and obesity-related domains of HRQoL (after adjustment for baseline differences between the groups). The difference in the long-term change was large for obesity-related problems, but otherwise the differences were small or trivial. While HRQoL was initially worse in the surgery than in the control group due to self-selection of surgical treatment, surgical patients had better physical and obesity-related HRQoL at the 15-year follow-up. In the surgery group, patients with pre-existing diabetes experienced smaller reductions in social interaction limitations and depression than those without diabetes. Albeit those with a history of psychiatric disorder reported lower HRQoL in all domains of functioning than those without such history over the 15-year study period, there was no clear evidence for differences in the long-term improvements between the two groups.

Our results regarding the 15-year changes in HRQoL were mostly similar to those reported earlier by Karlsson et al. on a subset of SOS participants over 10 years of follow-up [20]. In line with that study [20] and a study by Kolotkin et al. [13], obesity-related HRQoL showed the greatest within-group (baseline vs. 15 years in the surgery group) and between-group (surgery vs. usual care) improvement after 15 years of follow-up. Obesity-related HRQoL denotes how bothered the respondent is by their obesity in a broad range of social activities (e.g., private gatherings at home, going to restaurants, trying on and buying clothes, intimate relations), and the observed improvement in the surgery group is probably a reflection of the 15% reduction in BMI at the 15-year follow-up. Although the respective change in BMI was minor in patients receiving usual care, they still reported a small, long-term reduction in obesity-related problems, potentially reflecting the effects of aging and accumulated life-experiences on how these social situations are perceived. In both treatment groups, women experienced more obesity-related problems than men over the entire study period. This observation could be related to the higher BMI inclusion criteria for women (BMI ≥ 38 kg/m2) compared to men (BMI ≥ 34 kg/m2) in this study, or to women’s higher likelihood to encounter weight-based stigma and discrimination [4, 31].

Despite the improvement in perceived health not being as notable as the reduction in obesity-related problems in surgical patients, it is noteworthy that the direction of change was opposite in the two treatment groups. While perceived health improved in the surgery group, it deteriorated in the control group over 15 years. These differences are likely to reflect the well-documented effects of bariatric surgery on the prevention and resolution of obesity-related co-morbidities [7,8,9,10].

Mental health status demonstrated small improvements 15 years after bariatric surgery as indicated by the scales measuring anxiety, depression, and overall mood. Whereas depression and mood improved more following surgical than non-surgical treatment (between-group differences being small in size), the alleviation of anxiety symptoms was roughly similar in both treatment groups in accordance with earlier findings [20]. Epidemiological evidence implies that although anxiety disorders are chronic, there is a natural decrease in the prevalence rates with older age [32]. This was also evident in the present study as older age was related to lower and decreasing levels of anxiety over the study period in the surgery and control groups. It is thus possible that aging has influenced the course of anxiety symptoms in both treatment groups. The surgery group also showed a small long-term reduction in health-related limitations in social life, but in the control group, the corresponding change was negligible.

An important aspect to consider regarding the effect of surgical vs. non-surgical obesity treatment on subsequent HRQoL outcomes is the fact that the SOS participants who initially had worse HRQoL were more likely to select (as a co-decision with the physician) bariatric surgery than non-surgical treatment. This has at least two possible consequences. On the one hand, surgical patients possessed a higher likelihood of HRQoL improvement than control patients, while on the other hand, HRQoL had to improve in surgical patients to reach a HRQoL level similar to control patients. Although we controlled for these baseline differences in the analyses, it is still possible that they influence the observed effects of surgery (vs. usual care) on HRQoL development.

We further extended Karlsson et al.’s study [20] by examining whether long-term HRQoL changes in the surgery group varied according to sociodemographic characteristics, pre-operative health status or surgery type. Pre-existing diabetes emerged as one pertinent factor: surgical patients with baseline diabetes experienced smaller improvements in depression and health-related limitations in social life than those without diabetes. In contrast, the 5-year prospective study of 7000 gastric bypass patients from the Scandinavian Obesity Surgery Register noted that physical, social and obesity-related HRQoL (as measured by the RAND Short Form-36 and the OP scale) improved slightly more in patients with pre-operative diabetes [19]. While these conflicting results might be partly attributable to differences in the primary surgery type (100% vs. 13% of patients underwent gastric bypass) and follow-up time (5 vs. 15 years) between the two studies, the impact of pre-operative diabetes on long-term HRQoL outcomes clearly requires further research.

Additionally, we observed that obesity-related problems decreased more following gastric bypass than other surgery types in line with the fact that this surgical method is associated with the largest long-term weight loss [10]. It is, however, noteworthy that the present analyses did not reveal other significant differences between the three surgery types in terms of long-term changes in mental, physical and social domains of functioning. A recent systematic review on 5-year post-operative psychological outcomes also concluded that the most common surgery types (gastric bypass, sleeve gastrectomy, laparoscopic adjustable gastric banding) did not have significantly different effects on depression and anxiety outcomes [33].

While patients treated medically for depression prior to gastric bypass experienced less improvement in HRQoL than patients without such treatment in the Scandinavian Obesity Surgery Register [19], we did not detect significant differences in 15-year HRQoL changes between surgical patients with and without pre-existing psychiatric disorder. Thus, in our study, having a history of psychiatric disorder did not appear to impede long-term HRQoL improvements after bariatric surgery. Nonetheless, those with such history reported considerably poorer mental, physical, social and obesity-related HRQoL over the whole study period. Our previous findings from the SOS study also highlight the relevance of pre-existing psychiatric conditions in surgical patients, since pre-operative health care visits for self-harm or mental disorders and psychiatric drug use predicted higher risk of subsequent non-fatal self-harm and suicide [34].

American Society for Metabolic and Bariatric Surgery recently issued a position statement on pre-operative health optimization prior to metabolic and bariatric surgery. The position statement highlights that for patients with pre-existing psychiatric conditions, it is essential that their symptoms are adequately managed and that a behavioral health provider monitors and supports the patient after surgery [35]. Our present observations regarding different domains of HRQoL together with previous evidence [11] imply that regular psychosocial monitoring and support is important after bariatric surgery (for all patients, but particularly for those with prior psychiatric problems) due to possible setbacks in positive HRQoL development after the first or second post-operative year.

The main strengths of our study include the assessment of key HRQoL dimensions ten times over 15 years of follow-up in a relatively large sample of surgical and control patients, as well as the use of diverse information on the pre-operative situation to identify potential risk groups for long-term adverse HRQoL outcomes. Both within-group (baseline vs. 15 years) and between-group (surgery vs. usual care) changes were examined in this matched, non-randomized study to provide a detailed assessment of HRQoL development. Future research is, however, required to examine how post-operative changes in health status associate with long-term HRQoL changes to further inform clinical practice. Certain limitations of the present study should also be considered. Due to the non-randomized study design, we cannot eliminate the possibility that the choice of bariatric surgery or usual care affect the observed differences in HRQoL development between the two treatment groups. As is common in the long-term prospective studies with regular follow-up visits, there was non-participation over time. Surgical and control patients who did not attend 15-year follow-up tended to have lower HRQoL at baseline than those who attended, but these differences were mostly trivial in magnitude. Moreover, it should be noted that this pattern of non-participation was similar in the surgery and control groups.