1 INTRODUCTION

Adults living in rural areas in the United States are disproportionately affected by obesity; as of 2016, the estimates show obesity prevalence of 43.1% of adults in non-metropolitan statistical areas compared to 35.1% in large metropolitan statistical areas.1 Higher obesity prevalence among adults from rural areas contributes to the higher rates of chronic disease and mortality and poorer overall health and quality of life observed in rural versus urban areas.2 Given these significant disparities, there is an urgent need for increased access to, and dissemination of, evidence-based obesity treatments in rural communities.

Bariatric surgery is recognized as one of the most effective interventions for substantial weight loss among patients with moderate to severe obesity3; however, eligible residents of rural areas are 23% less likely to undergo bariatric surgery than urban counterparts.4 Further, there is a paucity of research evaluating bariatric surgery outcomes among patients living in rural communities. One study conducted by Bergmann and colleagues5 found that rural status significantly predicted bariatric surgery completion among adults evaluated in a large university hospital in West Virginia; however, the authors noted that this finding was confounded by insurance type, given that patients from rural areas were more likely to be denied access to surgery based on their insurance payer (specifically, patients living in rural areas were significantly more likely to be insured via West Virginia Medicaid than those who did not live in rural areas, and this insurance plan had not approved any patients for bariatric surgery in their clinic). The association between rural status and surgery completion was not significant after excluding patients insured under West Virginia Medicaid. Importantly, this study also found that rural status was not a significant predictor of retention at follow-up appointments nor of weight-loss outcomes at 6 or 12 months.

While the study by Bergmann and colleagues5 provided preliminary evidence that adults living in rural areas may experience comparable postsurgical outcomes to adults from urban areas, its small sample size (82 patients who completed surgery; however, analyses were conducted only with the 73 patients who completed month 6 and 50 patients who completed month 12 follow-up visits) may have limited its power to observe statistically significant differences between groups. Moreover, individuals living in rural communities face unique, multifaceted barriers to making changes in eating and activity behaviors (e.g., limited access to healthy grocery options, lack of sidewalks and recreational facilities/gyms, time restraints due to caregiving or other family responsibilities, social networks promoting unhealthy eating/activity behaviors, higher financial strain, and poverty),6 which may in turn impact postoperative weight-loss success. Indeed, one study by Mock and colleagues7 found that, among 34 patients pursuing bariatric surgery in rural Appalachia, a limited food budget was significantly associated with less weight loss at 3 months post-surgery; however, this association was no longer significant at 12 months following surgery completion. Given the limited amount of research investigating rural/urban bariatric surgery outcomes and the multifactorial influences that may impact the ability of adults in rural areas to engage in health behaviors required to enhance postsurgical weight loss, the current study aimed to evaluate demographic differences and postoperative weight-loss outcomes in a larger sample of rural and urban residents undergoing bariatric surgery.

2 MATERIALS AND METHODS

A retrospective chart review of adults who completed laparoscopic bariatric surgery was performed to evaluate differences in demographic characteristics and weight-loss outcomes in patients from rural and urban counties. Participants were 170 adult patients (ages 18+) who underwent bariatric surgery (laparoscopic Roux-en-Y gastric bypass or sleeve gastrectomy) at a large university medical center in North Central Florida between 1 July 2017 and 30 June 2018. Per standard clinic operations, follow-up appointments were scheduled for all patients at three- and six-months post-surgery. Reminder letters for upcoming appointments were sent to patients via the Electronic Health Record (EHR), or via phone call/mailed letter if patients did not have access to the messaging features of the EHR. Clinic staff completed three attempts to contact patients who did not attend 3- or 6-month follow-up appointments. This study was approved by the university Institutional Review Board.

2.1 Measures

Data collected from patients' EHR included age, race, gender, insurance type, surgery type, initial preoperative weight (2 weeks prior to surgery), postoperative weight at 3 and 6 months, and city of residence. Rural/urban classification was determined by cross-referencing each patient's city of residence with the county of its location. Participant county of residence was classified as rural or urban according to the U.S. Census Bureau's categorization of census tracts as “completely rural,” “mostly rural,” or “mostly urban”.8 In this study, “completely rural” and “mostly rural” were combined into a single category, labeled “rural,” to simplify comparison between rural versus urban counties. Insurance type was coded as “government insurance” (e.g., Medicaid, Medicare, or Medicaid/Medicare coverage provided through private-sector insurers, such as Medicare Advantage) or “private insurance” (any health insurance coverage offered by an entity other than the state or federal government) as data were collected from the EHR. The primary weight outcome was percent total weight loss (%TWL) at 3 and 6 months post-surgery, calculated from weight data from patients' EHR.

2.2 Statistical analyses

This study's sample size of 170 patients yielded a statistical power of 0.90 to detect a 10% difference in %TWL between groups at 3 and 6 months, assuming a between-group standard deviation of 10%. This threshold was selected as it has been defined by the National Heart, Lung, and Blood Institute (NHLBI) as a “clinically meaningful” weight-loss threshold for adults with obesity, and has been associated with positive changes in health such as reductions in triglycerides, blood glucose, and blood pressure.9 Descriptive statistics were calculated for patient demographic variables. Chi-square tests (using Fisher's exact P when expected cell counts were less than 5), and independent samples t-tests were used to investigate whether there were differences between rural and urban patients in demographic characteristics and weight at baseline (i.e., preoperative) and postoperative visits.

Multiple imputation (using 10 imputed datasets) was used in order to derive unbiased parameter estimates for missing weight data from patients who did not attend 3- and 6-month postoperative visits. Multiple imputation analysis offers a significant improvement over traditional approaches to address missing data and employs the assumption that missing values are Missing at Random (MAR).10 As return at follow-up visits may be related to postoperative outcomes, the MAR assumption may be violated (resulting in data Missing Not at Random [MNAR]); however, it has been argued that multiple imputation is an improvement over traditional methods of handling missing data even under MNAR scenarios.11 Following the existing recommendations,10 our imputation model used all available variables in the dataset (i.e., age, race, gender, baseline and postoperative weight outcomes, insurance type, surgery type, rural/urban status, and attendance at 3- and 6-month follow-up appointments).

3 RESULTS

Of the 170 patients included in the current study, 52 (30.6%) resided in rural counties and 118 (69.4%) in urban counties. Overall, patients were an average (mean ± SD) of 43.7 ± 12.0 years old and had preoperative body mass index (BMI) of 47.1 ± 8.1 kg/m2; 148 (87.1%) identified as female and 94 (55.3%) identified as White or Caucasian, 65 (38.2%) as Black or African American, 11 (6.5%) as Other, 161 (94.7%) as non-Hispanic/Latino, and 9 (5.3%) as Hispanic/Latino. Patient demographics and baseline characteristics by rural/urban status are displayed in Table 1. Compared to patients from urban counties, patients from rural counties had lower baseline BMI, were more likely to identify their race as White and ethnicity as Hispanic/Latino, and were more likely to undergo sleeve gastrectomy than Roux-en-Y procedures; however, there were no statistically significant differences between patients from rural and urban counties in terms of age, sex, or insurance type.

TABLE 1. Participant baseline characteristics Characteristic Rural n = 52 Urban n = 118 p Age in years, M (SD) 45.4 (12.2) 42.9 (11.8) 0.218 Body mass index (kg/m2), M (SD) 45.2 (6.8) 47.9 (8.5) 0.045 Body weight (kg), M (SD) 122.9 (22.5) 136.3 (29.9) 0.002 Sex, n (%) Female 43 (82.7%) 105 (89.0%) 0.261 Male 9 (17.3%) 13 (11.0%) Ethnicity, n (%) Hispanic/Latino 9 (17.3%) 0 (0.0%) <0.001 Non-Hispanic 43 (82.7%) 118 (100.0%) Race, n (%) White or Caucasian 38 (73.1%) 56 (47.5%) 0.002 African American or other/Multiple 14 (26.9%) 62 (52.5%) Surgery type, n (%) Roux-en-Y 14 (26.9%) 54 (45.8%) 0.021 Sleeve gastrectomy 38 (73.1%) 64 (54.2%) Insurance type, n (%) Government 30 (57.7%) 69 (58.5%) 0.052 Private insurance 22 (42.3%) 49 (41.5%)

Regarding retention, 133 (78.2%) patients attended the 3-month and 105 (61.8%) patients attended the six-month postoperative visits. Of the 52 participants residing in rural areas, 11 (21.2%) did not attend the 3-month postoperative visit, and 22 (42.3%) did not attend the six-month follow up visit. In comparison, 26 of the 118 patients residing in urban counties (22.0%) did not attend the 3-month postoperative visit, and 43 (36.4%) did not attend the six month follow up visit. Logistic regression models demonstrated no significant association between rural/urban status and attendance at follow-up visits at three months (Wald χ2(1) = 0.02, p = 0.899) nor six months (Wald χ2(1) = 0.06, p = 0.469). Moreover, no differences were observed among patients who did or did not attend three or six-month postoperative visits in terms of baseline age, BMI, race/ethnicity, or insurance type. There was a significant difference between patients who did and did not attend three-month follow up in terms of insurance type (χ2(1) = 4.89, p = 0.027), such that a higher percentage of participants who attended the three-month clinic visit had private insurance (58.3% among those who attended vs. 37.8% among who did not) and a lower percentage had government insurance (41.7% among those who attended vs. 62.2% among those who did not). Additionally, there was a significant difference between patients who did and did not attend the three-month postoperative visit with regards to surgery type (χ2 (1) = 3.89, p = 0.049), such that a lower percentage of patients who attended three-month follow-up underwent the Roux-en-Y procedure (36.1% of those who attended vs. 54.1% of those who did not), and a higher percentage of patients who received sleeve gastrectomy attended the 3-month follow-up (63.9% of those who attended vs. 45.9% of those who did not). These differences between groups were no longer present at six-months (ps > 0.05).

Across all participants, average (mean ± SE) %TWL was 16.5 ± 0.3% at three months and 22.5 ± 0.6% at six months. Figure 1 displays imputed estimates of %TWL at each time point by rural/urban status. Patients from both rural and urban counties showed significant reductions in %TWL at three months (Rural: −17.68 ± 0.6%, t (802.69) = −29.52, p < 0.001; Urban: −15.95 ± 0.4%, t (556.64) = −29.16, p < 0.001) and 6 months (Rural: −23.33 ± 1.24%, t (38.57) = −18.82, p < 0.001; Urban: −22.16 ± 0.7%, t (521.36) = −32.35, p < 0.001). At three months, patients from rural areas demonstrated a significantly higher %TWL compared to patients from urban areas (−1.73 ± 0.73%; t (610.16) = −2.37, p = 0.018). The magnitude of the difference in %TWL between groups was attenuated, but remained significant after controlling for surgery type (−1.50 ± 0.74%; t (112.27) = −2.02, p = 0.046). After further controlling for insurance type and race/ethnicity, there were similar magnitudes of effects observed for three-month changes in %TWL; however, these effects were no longer statistically significant (−1.70 ± 0.86%; p = 0.069). At six months, statistically significant differences in %TWL between groups were not observed (p = 0.451), with the same pattern observed after controlling for surgery type, insurance type, and race/ethnicity (p = 0.580).

Postoperative weight loss for patients from rural (n = 52) and urban (n = 118) counties. *p < 0.05

4 DISCUSSION

The purpose of the current study was to describe and compare demographic characteristics of patients from rural versus urban counties who underwent bariatric surgery, and to explore potential differences in post-operative weight-loss outcomes between these groups. At baseline, patients from rural counties had a lower baseline BMI, were more likely to identify as White and Hispanic/Latino, and were more likely to undergo sleeve gastrectomy than Roux-en-Y surgical procedures. Of note, the finding that patients from rural counties had a lower baseline BMI was unexpected given previous literature reporting that adults from rural areas have higher BMIs and obesity prevalence than adults living in urban areas.1, 2 This may suggest selection bias such that rural patients who presented for bariatric surgery may not represent the larger population of rural adults with obesity.

Patients from both rural and urban areas experienced significant reductions in %TWL at 3 and 6 months. At 3 months, patients from rural areas demonstrated significantly greater reductions in %TWL compared to their urban peers; however, this association was no longer statistically significant after controlling for insurance type and race/ethnicity, suggesting that differences in insurance type (a marker for broader social–economic status) and race/ethnicity between the adults from rural and urban counties included in this study may partially explain differences in weight-loss outcomes at 3 months. This is consistent with previous research that has shown that non-Hispanic White adults typically achieve larger weight losses following bariatric surgery than non-Hispanic African American adults and Hispanic adults.12, 13 Consistent with the results of Bergmann and colleagues,5 rural/urban status was not associated with attendance at 3- and 6-month postoperative follow-up appointments, and there were no significant differences in %TWL between adults from rural and urban counties at 6 months. This similar pattern of results was observed despite differences in methods used to code rurality (the current study used Census Bureau classification while the study by Bergmann and colleagues used Rural–Urban Commuting Area [RUCA] codes,5, 8 which may be better able to assess within-county variability),14 supporting the robustness of this finding.

Strengths of the study include an objective assessment of weight outcomes and the use of multiple imputation to address missing weight data from patients who did not attend 3- and 6-month postoperative visits. Despite these strengths, this study has several limitations. First, the sample was predominantly non-Hispanic White, and all patients had health insurance. Thus, this sample is not representative of the overall rural population (although it may represent the portion of the population living in rural areas who seek bariatric surgery, a disparity in access highlighted by Bergmann and colleagues).5 Another limitation of this study is the examination of early weight outcomes at 3 and 6 months. Given that evidence from large-scale clinical trials suggests that weight-loss trajectories do not typically differentiate until after the 6-month mark,15 it is possible that our analyses did not capture important differences between groups during the critical long-term period following surgery completion. Findings from the broader literature, however, have demonstrated that early postoperative weight-loss outcomes (i.e., 1 to 6 months post-surgery) serve as a significant predictor of long-term outcomes.16, 17 Thus, while future studies should examine longer term outcomes, the examination of surgical outcomes during this earlier postsurgical period still holds value for understanding potential rural/urban differences.

A further limitation of this study includes our broad categorization of insurance status as “government insurance” or “private insurance.” Although this was a pragmatic consideration (minimizing the number of groups to allow for observation of associations between insurance type and weight outcomes/postoperative follow-up), we were unable to include more specific details about individual insurance plans (e.g., preferred provider organizations vs. health maintenance organizations, or differences in point of service and/or cost sharing) that may have allowed for further understanding of these relationships. For example, Gasoyan and colleagues18 found that insurance plan is a stronger predictor of bariatric surgery utilization over simple payor type. Moreover, although we distinguished between government insurance and private insurance, there may still be considerable variation within this category (i.e., difference between Medicare Advantage Plans, Medicaid vs. Medicare). Similarly, the use of the Census Bureau classification of rural versus urban has been noted to be over-simplified, and likely does not consider variability within counties that may have an impact on the observed associations.14, 19 Thus, future research should consider specific insurance plan types and alternative definitions of coding rurality (such as the RUCA codes14, 19 used by Bergmann and colleagues).5 Finally, 21.8% of weight data was missing at 3 months, and 38.2% at 6 months; although we attempted to control for this with multiple imputation, additional research should aim to replicate these findings in samples with less attrition.

Overall, the results from the current study found that patients from both rural and urban counties who completed bariatric surgery achieved significant reductions in weight at 3 and 6 months post-surgery. While patients from rural areas demonstrated significantly better weight-loss outcomes at 3 months post-surgery, there were no significant differences between groups by 6 months. Taken together with the results from Bergmann and colleagues,5 the results suggest that adults from rural areas pursuing bariatric surgery have the potential to realize surgical weight-loss outcomes comparable to those of their urban counterparts. Future studies should investigate rural/urban differences in bariatric outcomes in larger samples with longer postoperative follow-up periods, and in patients who are uninsured and from minority racial/ethnic backgrounds. Further understanding of potential differences in post-bariatric surgery weight-loss outcomes may enhance initiatives designed to reduce health disparities faced by patients undergoing bariatric surgery.

ACKNOWLEDGMENTS

We would like to thank Center for Integrative Cardiovascular and Metabolic Disease (CICMD) at the University of Florida for providing support for publication of this article.

CONFLICT OF INTEREST

The authors declare that they have no conflict of interest.

AUTHOR CONTRIBUTION

All authors participated in this research. Kathryn Ross, Viviana Bauman, and Andreana Apostolopoulos jointly designed the study. Viviana Bauman and Andreana Apostolopoulos completed the literature searches. Andreana Apostolopoulos and Gwendolyn Hasse acquired the chart review data. Viviana Bauman and Kathryn Ross performed the statistical analyses. Viviana Bauman, Andreana Apostolopoulos, and Thomas Parkman drafted the manuscript. All authors edited the tables, figures, and, manuscript, and approved the final versions.

REFERENCES

1Hales CM, Fryar CD, Carroll MD, Freedman DS, Aoki Y, Ogden CL. Differences in obesity prevalence by demographic characteristics and urbanization level among adults in the United States, 2013-2016. J Am Med Assoc. 2018; 319(23): 2419- 2429. https://doi.org/10.1001/jama.2018.7270 2Garcia MC, Rossen LM, Bastian B, et al. Potentially excess deaths from the five leading causes of death in metropolitan and nonmetropolitan counties - United States, 2010-2017. MMWR Surveill Summ. 2019; 68: 1- 11. https://doi.org/10.15585/mmwr.ss6810a1 3Buchwald H. The evolution of metabolic/bariatric surgery. Obes Surg. 2014; 24(8): 1126- 1135. https://doi.org/10.1007/s11695-014-1354-3 4Douthit N, Kiv S, Dwolatzky T, Biswas S. Exposing some important barriers to health care access in the rural USA. Publ Health. 2015; 129(6): 611- 620. https://doi.org/10.1016/j.puhe.2015.04.001 5Bergmann KL, Cox SJ, Tabone LE. Influence of a rural environment on patient access and outcomes for bariatric surgery. Surg Obes Relat Dis. 2017; 13(4): 632- 636. https://doi.org/10.1016/j.soard.2016.11.009 6Dixon BN, Piervil E, Eastman A, Ross KM. A social-ecological review of the rural versus urban obesity disparity. Health Behav Policy Rev. 2019; 6: 378- 394. https://doi.org/10.14485/HBPR.6.4.6 7Mock K, Cox S, Bias S, et al. Limited food budget affects weight loss outcomes after bariatric surgery in an Appalachian population. Surg Obes Relat Dis. 2016; 12(7 Suppl):S170. https://doi.org/10.1016/j.soard.2016.08.305 8Ratcliffe M, Burd C, Holder K, Fields A. Defining rural at the U.S. Census Bureau; 2016. https://www2.census.gov/geo/pdfs/reference/ua/Defining_Rural.pdf. Updated December, 2016. 9 National Heart, Lung, and Blood Institute. Managing overweight and obesity in adults: Systematic evidence review from the obesity expert panel. 2013. https://www.nhlbi.nih.gov/sites/default/files/media/docs/obesity-evidence-review.pdf 10Acock AC. Working with missing values. J Marriage Fam. 2005; 67(4): 1012- 1028. https://doi.org/10.1111/j.1741-3737.2005.00191.x 11Schafer JL, Graham JW. Missing data: our view of the state of the art. Psychol Methods. 2002; 7(2): 147- 177. https://doi.org/10.1037/1082-989X.7.2.147 12Wallace AE, Young-Xu Y, Hartley D, Weeks WB. Racial, socioeconomic, and rural-urban disparities in obesity-related bariatric surgery. Obes Surg. 2010; 20(10): 1354- 1360. https://doi.org/10.1007/s11695-009-0054-x 13Byrd AS, Toth AT, Stanford FC. Racial disparities in obesity treatment. Curr Obes Rep. 2018; 7(2): 130- 138. https://doi.org/10.1007/s13679-018-0301-3 14Hall SA, Kaufman JS, Ricketts TC. Defining urban and rural areas in U.S. Epidemiologic studies. J Urban Health. 2006; 83(2): 162- 175. https://doi.org/10.1007/s11524-005-9016-3 15Courcoulas AP, King WC, Belle SH, et al. Seven-year weight trajectories and health outcomes in the longitudinal assessment of bariatric surgery (LABS) study. JAMA Surg. 2018; 153(5): 427- 434. https://doi.org/10.1001/jamasurg.2017.5025 16Mor A, Sharp L, Portenier D, Sudan R, Torquati A. Weight loss at first postoperative visit predicts long-term outcome of Roux-en-Y gastric bypass using Duke weight loss surgery chart. Surg Obes Relat Dis. 2012; 8(5): 556- 560. https://doi.org/10.1016/j.soard.2012.06.014 17Manning S, Pucci A, Carter NC, et al. Early postoperative weight loss predicts maximal weight loss after sleeve gastrectomy and Roux-en-Y gastric bypass. Surg Endosc. 2015; 29(6): 1484- 1491. https://doi.org/10.1007/s00464-014-3829-7 18Gasoyan H, Halpern MT, Tajeu G, Sarwer DB. Impact of insurance plan design on bariatric surgery utilization. Surg Obes Relat Dis. 2019; 15(10): 1812- 1818. https://doi.org/10.1016/j.soard.2019.07.022 19Hart LG, Larson EH, Lishner DM. Rural definitions for health policy and research. Am J Public Health. 2005; 95(7): 1149- 1155. https://doi.org/10.2105/AJPH.2004.042432