Introduction

Bariatric surgery (BS), combined with a healthy lifestyle, is considered the most effective procedure for controlling weight-associated comorbidities in severe obesity [1]. Although weight loss (WL) is the most frequently used marker in clinical practice, there is still no consensus on the best criteria to consider while determining prognosis after surgical treatment. Many studies have shown excellent results in the medium- and long-term after surgery for WL [2,3,4,5,6]; however, other studies have indicated that approximately 20–25% of patients struggle with insufficient WL [7] and some patients regain weight 2 years after surgery [8,9,10].

Postoperative WL and its maintenance are related to several factors, including age [11,12,13], sex [12], preoperative body mass index (BMI)/body composition [11, 14, 15], number of pre- and post-operative appointments with the multidisciplinary team [16,17,18], time after surgery [12], healthy lifestyle habits [19, 20], and type of surgery [13]. Additionally, socioeconomic factors, such as educational and income levels [21], ethnicity and racial disparities [22,23,24], and health insurance type [25] also influence the magnitude of WL after BS. However, the level of evidence from these studies remains unclear, especially considering the impact of the markers analyzed separately and the confounding factors in treatment outcomes [26, 27].

Since BS is a treatment option in various countries for people of different socioeconomic statuses (SES), races, and ethnicities, it is important to understand the degree of risk of or vulnerability to the social determinants of this type of treatment. A recent systematic review investigated the racial disparities, not on WL, but in postoperative adverse events, and found increased 30-day morbidity, mortality, and length of stay in black patients [28]. Thus, systematization specifically related to WL would contribute to the screening and follow-up of patients from the preoperative period and to the monitoring of WL and prevention of obesity recurrence. Therefore, this systematic review and meta-analysis aimed to investigate the association between SES and weight loss at least 12 months after BS.

Methods

Protocol and Registration

This systematic review was conducted based on the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) protocol [29]. The study protocol was registered in the International Prospective Register of Systematic Reviews (PROSPERO) (no. CRD42020150681).

Research Question

The research question was elaborated using the Population, Exposure, and Comparison Outcome (PECO) strategy, as recommended by the PRISMA protocol [29, 30], as follows: “Do socioeconomic factors influence weight loss 12 months after bariatric surgery?”.

Eligibility Criteria

Inclusion Criteria

This review included observational studies involving adults who underwent any type of BS. Studies relating WL outcomes after surgery to socioeconomic data, such as level of education, occupation, employment status, income, ethnicity/race, health insurance, and any other scale of assessing SES, were included. To reduce publication and retrieval bias, the search was not restricted by language or publication date and status.

Exclusion Criteria

The exclusion criteria were as follows: individuals aged < 18 or > 65 years; procedures for WL other than BS; self-reported WL; socioeconomic data that had not been obtained through direct interview; follow-up period < 12 months; reviews, letters, conference abstract, personal opinions, books, cross-checking information, case reports, and qualitative studies; and studies with required data being unavailable, even after several attempts to contact the authors by email.

Information Sources and Search Strategies

The search strategy was developed in accordance with the criteria established by the Peer Review of Electronic Search Strategies (PRESS) checklist [31]. Subsequently, two researchers with expertise in systematic review evaluated and contributed to its adequacy. The following databases were searched: PubMed, Embase, Lilacs, Scopus, and Web of Science. The gray literature was also partially searched using ProQuest Dissertations and Theses Global and Google Scholar (limited to the first 200 most relevant articles). The search was conducted on October 25, 2021, and updated on February 28, 2022. Additionally, manual searches were performed using selected articles in the reference lists. The details of the search terms used in each database are described in Table S1. To eliminate duplicate references and screening, the Rayyan app was used [32].

Study Selection

Two authors (M.S.M.A. and L.C.L.B.) independently selected the studies to be included in two phases. The first phase comprised screening of the articles by their titles and abstracts. The second phase involved reading the remaining articles in full and selecting eligible ones for review. The included articles were analyzed by the two authors.

Data Collection

Data were extracted by one author (M.S.M.A.), and all information was cross-checked by a second author (L.C.L.B.). We considered WL data obtained 12 months after surgery, and the following information was collected from all selected studies: authors, publication year, country of the study, aim of the study, type of surgery, intervals of follow-up/postoperative period of evaluation, sample size, sex, age, exposure, outcomes, and socioeconomic variables associated with successful surgery. When a study presented insufficient data, all study authors were contacted by email, at least twice.

Methodological Quality in Individual Studies

The quality of the methodology of each included study was assessed using the Newcastle–Ottawa Quality Assessment Scale (NOQAS) [33]. Two authors independently assessed the quality of each study. The NOQAS grades studies in three domains as follows: study group selection, study group comparability, and the assessment of either the outcome or exposure of interest. A maximum of nine points can be awarded, and studies with three or four points in the selection domain, one or two points in the comparability domain, and two or three points in the outcome/exposure domain were considered of good quality. Those with two stars in the selection domain, one or two stars in the comparability domain, and two or three stars in the outcome/exposure domain were considered of fair quality. Poor-quality studies were those with no or one star in the selection domain, no star in the comparability domain, and no or one star in the outcome/exposure domain [33].

Certainty of Evidence

The certainty of the evidence for studies was evaluated by the Grading of Recommendations Assessment, Development, and Evaluation (GRADE) guidelines [34], based on an explicit question, including the specifications of all important and critical outcomes. According to the methodological design used, initial certainty was defined and then analyzed according to different domains (risk of bias, inconsistency, indirectness, imprecision, and publication bias). Finally, the evidence was summarized considering its certainty and the strength of the recommendations that the studies indicated.

Data Synthesis

The main outcome was WL after BS, as shown in different ways as the total WL (TWL), percentage of total WL (%TWL), percentage of excess WL (%EWL), percentage of excess BMI loss (%EBMIL), percentage of sample with successful WL or weight gain, and BMI change.

Data were analyzed using the STATA® program version 17 (Stata Corp LLC, College Station, TX, USA; serial number: 301706385466). Data were analyzed using the STATA® program version 17 (Stata Corp LLC, College Station, TX, USA), serial number: 301706385466.

The degree of statistical heterogeneity of the studies was evaluated using the heterogeneity index (I2), applying the following cutoff points: 0–40% (not very important), 30–60% (moderate), 50–90% (substantial), and 75–100% (considerable). Standardized and unstandardized mean differences and beta coefficients were used for the meta-analysis of evaluated outcomes. To estimate statistical significance, 95% confidence intervals were calculated. The random-effects model was used through the restricted maximum likelihood, Hedges estimator, and DerSimonian–Laird method. Sensitivity analysis was conducted to estimate the effect of heterogeneity on the outcome of the meta-analysis.

Results

Literature Search

A total of 8963 records were obtained from the database search. After eliminating duplicates, 4693 records were screened through their titles and abstracts; 104 articles were selected for full-text reading, of which 64 were excluded from the analysis (Table S2). Thirteen additional records were identified from the manual search of the reference list of the fully read articles. Fifty-three original articles were included for qualitative synthesis as described in the study flowchart (Fig. 1), and 21 were included for quantitative synthesis (meta-analysis). The summary of the characteristics and results of individual studies is presented in Table 1.

Fig. 1
figure 1

Flowchart of the search and selection process of studies

Full size image
Table 1 Summary of characteristics of the included studies and socioeconomic variables associated with weight loss after bariatric surgery in order of year of publication (n = 53)
Full size table

Study Characteristics

In this review, 72,087 patients were evaluated at least 12 months after BS. Twenty studies presented results after 36 months [35,36,37, 39, 41, 43, 44, 46,47,48,49, 52, 53]. All included studies had an observational design (cohort) with 37 retrospective [35,36,37, 39, 41, 43, 44, 46,47,48,49, 52,53,54,55,56,57,58,59,60,61,62,63,64,65,66,67,68,69,70,71,72,73,74,75,76,77] and 16 prospective [38, 40, 42, 45, 50, 51, 78,79,80,81,82,83,84,85,86,87] cohorts; they were published between 1991 and 2022. The number of participants ranged from 50 [72] to 16,629 [68].

The studies included were conducted in the United States of America (USA) [35,36,37,38,39,40,41,42,43,44,45, 49,50,51,52,53,54,55,56,57,58,59,60,61, 63,64,65, 67, 69, 70, 72,73,74,75,76,77,78,79, 81,82,83,84, 86, 87], Brazil [80], Iran [79], Israel [62, 85], Germany [47], Sweden [46] and the UK [48]. One study was conducted on patients form 14 specialized centers in Singapore, Malaysia, Taiwan, Hong Kong, Japan, Korea, India, Australia, Switzerland, and the USA [66]. All studies were available in English.

In terms of surgical techniques, nine different procedures were identified, and in most studies, the sample consisted wholly or partially of patients who underwent Roux-en-Y gastric bypass [35,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50, 52, 53, 55,56,57, 59, 61,62,63,64,65, 67,68,69,70,71,72,73,74,75,76,77,78, 80, 82, 84,85,86]. Sleeve gastric surgery was performed in 20 studies [36, 37, 41, 44, 47, 53, 56, 60, 62, 65, 66, 68, 69, 71, 73, 77, 82, 83, 85]. In 23 studies, more than one surgical technique was performed [36, 37, 41, 44, 47,48,49, 53, 54, 56, 62,63,64,65, 68,69,70,71, 73, 77, 81, 82, 85].

The socioeconomic variables studied included the income level [35, 36, 39, 46,47,48, 65, 70, 78, 80, 82], insurance type [39,40,41,42, 44, 52, 54, 58, 61, 67, 73, 78, 83, 84], employment status [39, 45, 46, 52, 59, 60, 68, 79, 83], educational level [36, 39, 41, 46, 59, 67, 68, 79], and race/ethnicity [35,36,37,38,39, 41, 49,50,51,52,53, 55,56,57, 59, 60, 62,63,64,65,66,67,68,69,70, 72, 74,75,76,77, 81,82,83, 85,86,87].

Indexes used to represent SES included the following: the Distressed Communities Index, a composite ranking by zip code that quantifies socioeconomic risk called neighborhood [43]; the Hollingshead four-factor index of social status based on occupation and educational level [57]; neighborhood socioeconomic status, six aggregate census tract measures including educational level, employment status, income, and the value of occupied housing [52]; and the European Deprivation Index quintiles and Geographical Health Accessibility [71]. Nineteen studies evaluated more than one socioeconomic factor [35, 36, 39, 41, 46, 52, 57, 59, 60, 65, 67, 68, 70, 71, 78, 79, 82,83,84].

Postoperative WL was presented in different ways in the studies: TWL [42, 61, 70, 78, 83], BMI change [39, 56, 68, 72, 78, 85, 86], %BMI change [58, 74], %EBMIL [35, 43, 54, 65], percentage of the sample with success in WL [8, 38, 41, 45, 57, 59, 60, 84], %TWL [39, 44, 46, 47, 50, 53, 55, 61, 64, 66, 70, 71, 75, 77, 85, 86], percentage of participants who regained weight [80], %EWL [8, 36, 37, 39, 40, 44, 48, 49, 51, 58, 62, 63, 66, 67, 73, 74, 76, 81, 82, 86,87,88], and percentage of the sample with ongoing WL, with weight stable or weight regain [52].

Results of Individual Studies

In this review, 29 (54.7%) studies reported the influence of at least one socioeconomic factor on WL outcomes [35, 37, 39, 41, 46, 47, 49,50,51,52,53, 55, 56, 63, 65,66,67,68,69,70,71,72, 75, 80,81,82,83, 86, 87]. Among these, 22 studies (75.8%) showed an association between race/ethnicity and weight loss [37, 39, 41, 49,50,51, 53, 55, 56, 63, 65, 67, 69, 70, 72, 75, 77, 81, 82, 86, 87], three (10.3%) found association with income level [35, 46, 47], three (10.3%) found an association with educational level [41, 46, 68], and three (10.3%) found an association with employment status [46, 68, 83]. Details on exposure and outcome variables of each study are available in Table 1 and Table S3.

Methodological Quality in Studies

The average NOQAS score for all studies was 6.8 ± 1.1 (Table S4). Out of the 53 studies, 34 (64.1%) were of good quality [8, 35,36,37,38,39,40,41,42, 44, 46, 47, 49, 50, 52, 53, 55, 57, 59,60,61,62,63,64,65,66, 68,69,70, 73, 78, 82,83,84], while 10 were of fair quality [43, 45, 67, 72, 74,75,76,77, 81, 87]. Nine studies were classified as having poor quality [48, 51, 54, 56, 58, 71, 80, 85, 86]. The “comparability” parameter (adjustment for confounding factors in the analyses) presented a critical performance among studies with the lowest scores.

Results of Syntheses

Regarding race/ethnicity, Figs. 2 and 3 show that, on average, white individuals had greater WL (%EWL) compared to blacks in the first and second years of follow-up, respectively. In linear regression analysis, the results were not statistically significant due to the high heterogeneity of the studies. When performing a sensitivity analysis, I2 = 0 was considered statistically significant. All meta-analyses are presented in Figures S1–S7.

Fig. 2
figure 2

Meta-analysis of the mean difference and 95% confidence interval in terms of race/ethnicity, with the percentage of weight loss (%EWL) in patients 12 months after bariatric surgery

Full size image
Fig. 3
figure 3

Meta-analysis of the mean difference and 95% confidence interval in terms of race/ethnicity, with the percentage of weight loss (%EWL) in patients 24 months after bariatric surgery

Full size image

Certainty of Evidence

In the evaluation by the GRADE process (Table S6), all studies were rated with very low certainty of evidence and classified with serious risk of bias, inconsistency, indirectness, and imprecision. Therefore, in all studies, publication bias strongly suggested spurious effects, while no effect was observed.

Discussion

The meta-analyses performed in this study highlighted that race/ethnicity was the factor that most influenced the different markers of therapeutic success. However, there is no standard method for evaluating WL, which makes comparative analyses challenging. We compiled the main studies that investigated the association of socioeconomic factors with long-term WL after BS.

Socioeconomic variables, to some extent, affect health and lead to the development of chronic diseases, such as obesity [89]. The mechanisms involved in these associations are complex and need further clarification [2]. Studies that assess the association between SES and obesity show contradictory results, probably due to the different ways of classifying socioeconomic levels and factors related to the level of development of the countries where the studies were conducted.

Newton et al. [90], in a systematic review to summarize the evidence on the association of life course SES with obesity, found that women with low SES had a higher prevalence of obesity in developed countries. Furthermore, Purslow et al.[91], in a study on middle-aged men and women, showed that individuals with low SES had the highest risk of weight gain. This can be explained by numerous factors, from less healthy dietary patterns to psychological stress and discrimination [92]. Restricted environmental opportunity and cultural issues have been speculated as possible explanations [93]. In contrast, Dinsa et al. [94], in a systematic review, investigated the association between SES and obesity in low- and middle-income countries among men and women. In low-income countries, better SES correlated with a higher prevalence of obesity in both sexes. In middle-income countries, better SES correlated with a lower prevalence of obesity in women.

Regarding severe obesity, the likelihood of undergoing BS is lower in males and individuals with low income, low education levels, and no health insurance [88, 95, 96]. However, even among those who could not undergo BS, the impact of SES on the outcomes was complex, and one key issue was the country’s health system model. All existing health systems worldwide have numerous shortcomings regardless of the funding approach model [97] or the country’s level of development [98].

An effective healthcare system provides hypothetically equitable access to high-quality healthcare, including treatment and curative, health promotion, prevention, and rehabilitation services to all [99]. Although access to BS is equal in some countries, such as Sweden [27], therapeutic results may vary depending on the follow-up model of the healthcare system. Therefore, the degree of WL after surgery may have been influenced by the characteristics of the healthcare system of the countries where studies were conducted.

Studies associating socioeconomic factors with less postoperative WL showed discrepant results. Individuals with lower educational levels tended to lose less excess weight after BS [41, 100]. Contrastingly, Stenberg et al. [46] found that a higher educational level was also associated with less WL [46], which can be partially explained by the long working hours and poor work-life balance observed among people with higher education [101]. Regarding employment status, a possible explanation for the association between unemployment and less long-term WL after surgery is that health insurance, which allows for necessary postoperative follow-up, is unavailable to unemployed individuals [42, 62, 76]. In contrast, employed patients demonstrate greater treatment adherence, even when universal health coverage is available [102].

Regarding income levels, all studies included in this review were conducted in developed countries, which requires a more careful analysis of the results and demonstrates the need for further studies, especially in less developed countries. Therefore, people with lower incomes may experience lower postoperative WL [46, 47]. Some authors describe poor access to healthy food [103] and physical activities as possible causes of the negative effects of low income on WL, especially in women [104]. However, even among individuals with low income, it is assumed that when they are well-assisted, they can establish priorities to maintain a healthy diet and an active lifestyle.

Regarding racial disparities in BS, the meta-analysis showed that black people experienced less WL after 12 and 24 months of surgery compared to white individuals. This association was also observed by Zhao et al. [105] in a systematic review that investigated racial disparities in postoperative WL and comorbidity resolution. They found that racial minorities lost less weight than non-Hispanic white individuals; however, the factors associated with this discrepancy remain unclear. In another study, Admiraal et al. [106] conducted a meta-analysis to determine the difference in %EWL 12 to 24 months after BS in people of African and Caucasian descent. They found better WL results in patients of Caucasian descent than in patients of African descent, regardless of the type of BS. The etiology of the difference in WL between patients of African and Caucasian descent is unclear, although metabolic, behavioral, and socioeconomic factors are considered probable causes. White individuals with greater improvement in energy expenditure in response to diet and exercise WL interventions [107,108,109] are more physically active and consume a diet lower in calories than black individuals [86, 110], suggesting that economic issues related to ethnic minorities and biological and environmental factors may synergistically explain these results.

The broad investigation of the main available databases and most of the studies used in this review can be considered strengths of this review. In addition, the use of tools such as meta-analysis showed results that can be useful in the management of patients undergoing BS in the long term.

Despite the considerable amount of studies, this review has some limitations. In general, the nomenclatures of race and ethnicity are not generally standardized across studies, the classifications used for income level and educational level were different in many studies, and there are different ways of presenting the weight loss outcomes, which made it difficult to compile the results. Concerning the quantitative synthesis, it was not possible to conduct more advanced analytical techniques to investigate the source of heterogeneity, such as subgroup, sensitivity, and meta-regression analyses due to the small number of studies included in the meta-analysis. Additionally, there is an inherent limitation that all the studies have an observational design which limits the assessment of causality.

This review indicates the importance of the government guaranteeing people access to BS and follow-up. Access to postoperative consultations, safety, transport, and equal opportunities can also affect adequate health status and its long-term maintenance [111]. The standardization of presenting variables related to both postoperative WL and SES is urgently required for further analysis to be conducted.

Conclusion

Race/ethnicity is associated with WL outcomes after BS, which may be related to equality in permanent access to healthcare systems. To obtain better results, further studies are needed to address socioeconomic issues related to surgical outcomes.