Ethnic Disparities in Use of Bariatric Surgery in the USA: the Experience of Native Americans

There is a significant economic burden associated with obesity that includes the cost of managing the condition and associated co-morbidities [1], its impact on health-related quality of life HRQoL [2] and its impact in terms of lost productivity [3]. These costs have risen as the prevalence of obesity has increased. In the USA, between 1988–1994 and 2015–2016, for example, the prevalence of obesity among adults rose from 22.9 to 39.6% (State of Obesity, 2019). In 2000–2005, the healthcare costs associated with obesity were estimated to be $210 billion per year, approximately 21% of US healthcare expenditures [4], and it was estimated that medical costs will increase by $44–66 billion per year by 2030 [5]. The experience in other developed countries while perhaps not as stark has nevertheless been similar [6]. Public health efforts to stem the rise in obesity have met with limited success [7, 8] though interventions that are targeted at the individual level have shown greater promise. Bariatric surgery appears to be a safe and cost-effective means of achieving weight loss [9]. In the USA, the uptake of bariatric surgery increased from 2.7 per 100,000 procedures in 1990 to 54.2 per 100,000 in 2008 among adults aged 18 and over [10] before plateauing through to 2012 at around 51 per 100,000 [11].

Previous work has looked at variations in expenditure by type of procedure [12], in uptake between men and women [10], between different ages and insurance status categories [13] as well as across distinct geographic areas with different levels of population need [14, 15]. This work has shown, inter alia, marked changes over time in the type of procedure that has been used in the USA (laparoscopic surgery increasing from 2.1 of procedures in 1998 to over 90% of procedures in 2008) [10]; women were consistently more likely than men to have surgery constituting around 80% from 1998 to 2012 [10, 11]; Whites were more likely to receive surgery than Blacks [16]; population need – in terms of prevalence of obesity at the state level – had little relationship with the uptake of surgery [14, 15]; there has been a reduction in complications associated with surgery over time [17]; and variations in surgery costs were related to co-morbidity, complexity and length of hospital stay [12].

Native American adults are 50% more likely to be obese than non-Hispanic Whites, and they have higher age-adjusted odds for diabetes, heart disease and stroke of 2.4, 1.3 and 1.8 than Whites, respectively [18]. Despite these facts, no published study has examined the use of bariatric surgery by Native Americans relative to other ethnic groups. This may in part be explained by their relatively small size as a proportion of the population. While the percentage of Native Americans who are obese is similar to Blacks [19], Blacks constituted 13.4% of the population in 2018, Hispanics or Latino 18.3%, Asians 5.9% and Americans Indians and Alaska Natives just 1.3% of the population [19]. Inevitably, the small number of Native Americans in surveys makes it more difficult to investigate comparative ethnic group experiences. Nevertheless, the lack of research with respect to this group gives rise to a clear gap in our understanding regarding their comparative experience.

In this paper, we report the results of an examination of the uptake of bariatric surgery between 2008 and 2016 among obese adults using the nationally representative inpatient data HCUP-NIS. We give particular emphasis to the comparative experience of Native Americans. We compare uptake between different ethnic groups controlling for a range of covariates including co-morbidity, gender, age, location and insurance status and time using data pooled on over two million episodes of inpatient care.

Table 1 presents descriptive statistics of the pooled 9-year cohort used in subsequent analyses. This eligible cohort comprised a total of 1,729,245 inpatient episodes. The percentage of White, Native American, Black, Hispanic, Asian and other ethnicities were 68.8%, 0.7%, 19.3%, 8.4%, 0.7% and 2.1%, respectively.

As can be seen from the table compared to other ethnicities, a higher percentage of inpatients of Native American ethnicity had type II diabetes, lived in rural areas (not metropolitan and micropolitan areas) and were treated in rural areas. Compared with Whites, Native Americans were younger, less likely to have private insurance, more likely to live in poor areas (first, i.e. lowest income quartile) and less likely to be treated in a teaching hospital.

In Fig. 1, the number of elective admissions for different types of procedure is shown over time. The figure clearly shows the increasing popularity of laparoscopic sleeve gastrectomy (LSG) (for a variety of reasons including evidence of efficacy, safety and expansion of insurance coverage [12]) over other types of surgery, this one type contributing to much of the overall growth in bariatric surgery since late 2011.

In Fig. 2, the breakdown of surgery by the ethnic group over time is presented, the resurgence in surgeries after 2011 and the increasing proportion of surgeries related to Blacks and Hispanics perhaps being the most notable aspects of the figure. In Fig. 3a–f (appendix 3), trends in specific types of procedure by ethnic group are shown. In brief, these show the increased use of LSG after 2011 and the decline of other types of surgery among this group. Figure 3a shows that Native Americans were slightly slower in the uptake of LSG relative to other ethnic groups.

In Table 2, the likelihood of bariatric surgery among eligible inpatients is reported in a series of models controlling for different combinations of covariates. In each model, the adjusted odds ratio of surgery is significantly lower among Native Americans than Whites. In Model 3, adjusted only for clinical variables, the adjusted odds ratio of surgery among Native Americans is lowest among that of any ethnic group. In Table 3, adjusted odds ratios for a range of outcomes among those in receipt of surgery are reported by ethnic group. As can be seen, Native Americans in receipt of surgery had lower length of stay and lower costs and were less likely to be discharged to another healthcare facility.

In Table 4 (appendix 4), the full results of the logistic regression on uptake of surgery (Model 4) are reported.

Full results are reported in Table 4 in the supplement.

Bariatric surgery is a safe and cost-effective treatment of morbid obesity [9]. Its use in the USA has increased steadily since the extension of coverage by Medicare in 2011 as perioperative safety has improved and insurance coverage expanded [10, 11]. Previous studies demonstrated distinct patterns of utilization. The Andersen model (of predisposing, enabling and need factors) may be used to aid our interpretation of these patterns. The higher uptake among females that was found in other studies [13, 14] and echoed here, for example, may relate to gender differences in predisposing factors that are associated with reproduction [28] or social norms around the perception of obesity that tend to be particularly acute for females [29]. Similarly, the differences in uptake between eligible Blacks and Whites that were observed in this study and that have been related previously to insurance status [16] may be interpreted as an enabling factor, while ACCI or the presence of specific conditions may be interpretable in terms of need.

No previous study in this field has examined the comparative experience of Native Americans and potential explanatory factors in this. While Native Americans are a relatively small ethnic group in the USA, this lack of research is surprising given that Native Americans have much higher levels of obesity and obesity-related morbidity than Whites. Our study has shown that, controlling for a range of need, enabling and predisposing factors, eligible Native Americans are approximately 30% less likely to receive bariatric surgery than their White counterparts, a disparity comparable to that of Blacks. In models that controlled solely for clinical factors, the disparity experienced by Native Americans was even sharper (over 40% less likely) and greater than that experienced by Blacks. (Sensitivity analyses that included patients with BMI > 40 without co-morbid hypertension or type 2 diabetes in the eligibility criteria had no material effect on results). Compared with other ethnicities, the slower transition to laparoscopic sleeve surgery among Native Americans (now the dominant form of surgery performed) may provide further evidence of disadvantage in terms of equitable access. Our analyses show that the length of stay among Native Americans and healthcare costs are significantly shorter and lower compared to Whites; the average surgery-related episode cost was $775 less among Native Americans compared to $378 more among Blacks. These findings regarding lower hospital costs may be explained at least partly in terms of Native Americans having a shorter length of stay (approximately 1/3 of a day, Table 3) and being more likely to receive surgery outside urban centres (16.8% compared to 8.6%, Table 1). Our control for ACCI and insurance status suggests, for example, that this difference is unlikely to be related to need or insurance coverage. The fact that Native Americans (for a variety of reasons including historical discrimination) tend to be geographically concentrated and isolated [30] may explain lower uptake. Perhaps, Native Americans find it more difficult to access hospitals that provide bariatric surgery, and they may have been disproportionality affected by the concentration of services in Centers of Excellence following the decision to provide national coverage via the Centers for Medicare and Medicaid in 2006. Why Native Americans exhibit shorter lengths of stay and a lower probability of being discharged to another healthcare facility might also relate to geographic isolation though this is purely speculative. The findings are though consistent with a broader experience of disadvantage among this group in terms of socio-economic status [31], health [32] and access to healthcare [31, 33‐37]. The fact that this is the first study to examine the experience of this group in relation to bariatric surgery given their relative need lends further weight to the suggestion that they are an underserved group.

There are a number of limitations to our study. First, we are unable to link inpatient episodes over time to specific individuals or explore issues such a readmissions or changes in BMI following bariatric surgery. Second, we are similarly constrained by the data included in the survey in terms of the ability to control for potentially useful covariates. For example, while we can control for diabetes status, we cannot control for how successfully blood glucose is controlled or how long since diabetes was diagnosed. Similarly, while we can control for age and sex, we have no direct insight into the intentions of women regarding pregnancy. We can use proxy indicators related to place of residence for some covariates such as income – no such proxy exists with respect to indicators of need. Third, we observe only those surgeries performed in an inpatient setting and can make no claims in respect of non-admitted patients. These, however, are limitations imposed on us by the survey and limit the scope of our investigation.

These limitations must be balanced against the great strength of this study which is its sample size. Such a large sample size HCUP-NIS affords that allows us to explore the role of ethnicity for groups whose numbers would preclude analysis in other surveys. Because of this, it has been possible to demonstrate a previously unidentified disparity.

Our study highlights the existence of disparities in utilization of bariatric surgery among eligible Native Americans relative to Whites in the USA. The disparity is similar in magnitude to that experienced by Blacks and is evident after we control for a variety of potentially contributing factors that include insurance status. The disparity may in part relate to the geographic isolation experienced by this ethnic group. It reflects a broader experience of disadvantage experienced by Native Americans in terms of socio-economic status, health and access to healthcare. It is hoped that this study will both serve to draw attention to a previously unexamined dimension of this experience and help encourage further studies into their use of and access to bariatric services.