Bae J, Shade J, Abraham A, Abraham B, Peterson L, Schneider EB, Magnuson TH, Schweitzer MA, Steele KE. Effect of Mandatory Centers of Excellence Designation on Demographic Characteristics of Patients Who Undergo Bariatric Surgery. JAMA Surg. 2015;150(7):644-648. doi:10.1001/jamasurg.2015.74
From February 21, 2006, through September 24, 2013, the Centers for Medicare & Medicaid Services (CMS) required, via the National Coverage Determination manual, that bariatric surgery be performed only in hospitals that had been designated as a Center of Excellence (COE). The effect of this certification requirement on access to bariatric surgery has been reported only anecdotally.
To investigate whether the COE certification requirement proved to be a barrier to patients’ access to bariatric surgical procedures.
Design, Setting, and Participants
Using the National Inpatient Sample, we retrospectively identified patients who underwent bariatric surgery from January 1, 2006, through December 31, 2011.
Main Outcomes and Measures
Logistic regression and χ2 tests were used to examine differences in patients’ sociodemographic characteristics over time.
A total of 134 227 bariatric surgical patients were identified. The proportion of the population who were older than 64 years increased from 2.9% in 2006 to 7.0% in 2011 (P < .001) and there was a decrease in the proportion of patients who were 49 years and younger (P < .001). The percentage of female patients who underwent bariatric surgery decreased from 80.4% to 78.1% (P < .001) and the percentage of patients who were classified as black, Hispanic, or Asian or Pacific Islander increased from 12.3% to 15.1% (P < .001), 9.7% to 12.5% (P < .001), and 0.3% to 0.4% (P < .001), respectively. The proportion of patients with Medicare increased from 8.5% to 16.3% (P < .001) and those with Medicaid from 6.6% to 11.8% (P < .001). The percentage of patients with private insurance declined from 72.4% to 63.3% (P < .001). The proportion of patients in the lowest income quartile increased from 20.7% to 22.9% (P < .001) while those in the highest income quartile decreased from 25.8% to 23.9% (P < .001).
Conclusions and Relevance
The COE certification requirement by CMS did not appear to limit access to bariatric surgery. Future studies should determine whether CMS’s recent (2013) change in policy (ie, removing the mandatory COE certification for bariatric surgical insurance coverage) might sacrifice patient safety without addressing the real cause of limited access to health care.
During the past decade, the prevalence of morbid obesity has increased by 70% in the United States.1 A study from 2012 found that one-third of US adults are obese. Obesity is correlated with higher incomes among men and lower incomes among women and peaks in prevalence around middle age.2 Now a major public health concern, morbid obesity (defined as a body mass index, calculated as weight in kilograms divided by height in meters squared, of greater than 40) is associated with numerous comorbidities and high health care costs.3- 6 Bariatric surgical procedures have proved to be the most effective treatment for morbid obesity.7- 9 However, the frequency of bariatric surgical procedures performed in the United States has recently declined.10 Socioeconomic disparity in access to bariatric surgery has been suggested as a contributing factor to this decline.11,12
In 2006, the Centers for Medicare & Medicaid Services (CMS) implemented a requirement that each hospital performing bariatric surgical procedures be designated as a Center of Excellence (COE).13- 15 Requirements for the COE qualification addressed the following 10 factors: clinical pathways and standardized operating procedures; bariatric nurses, physician extenders, and a program coordinator; surgical volume and outcomes; appropriate equipment and instruments; long-term patient follow-up and outcomes data; an institutional commitment to excellence; a designated medical director; consultative services; surgeon dedication and qualified call coverage; and patient support groups. Within each of these areas are multiple specifications, ranging from a minimum of 125 qualifying bariatric surgical procedures in the year before COE certification to an established location for the center’s patients who underwent bariatric surgery.16Quiz Ref ID The decision to limit coverage was based on data suggesting that positive surgical outcomes were more frequent in hospitals that performed a high volume of procedures yearly (>125 cases); restricting patients to such hospitals was hoped to increase patient safety.13,17- 19 In 2013, the CMS overturned its policy on COE facilities owing to increasing evidence that the surgical outcomes for COE facilities and non-COE facilities were not different.13
Several studies have questioned the benefit of the National Coverage Determination (NCD) restriction to COE facilities. In 2010, Livingston10 calculated the distance to travel to a COE bariatric program vs a non-COE bariatric program and concluded that the COE restriction increased the travel distance for patients with Medicare, thereby reducing access to care. Livingston10 further argued that such patients were sicker, less mobile, and less able to travel great distances to receive bariatric medical care.
While the differences in bariatric surgical outcomes between COE facilities and non-COE facilities remained uncertain, the requirement became highly controversial for impeding access to care.13 As a result, CMS decided to expand the coverage to hospitals that were not COE certified for bariatric procedures in 2013.13,20- 22 The question remains as to whether the implementation of the COE requirement for bariatric surgical coverage created the unintended adverse effect of reducing access to health care in the name of improving quality.
By examining the sociodemographic trends associated with relevant bariatric surgical procedures, we investigated whether coverage determination, established in 2006 and continued through 2013, was truly a barrier to accessing such procedures.
We used data from January 1, 2006, through December 31, 2011, obtained from the Nationwide Inpatient Sample (NIS), which was developed by the Healthcare Cost and Utilization Project.23 The most recent NIS database, created in 2011, included approximately 20% of all US hospital discharges and was designed to be representative of approximately 97% of all hospital discharges in the United States.23 This data set contains representative inpatient clinical and nonclinical data, which allows for the exploration of year-specific differences in bariatric surgical procedures, in-hospital complications, and patient characteristics. Institutional review board approval was waived by Johns Hopkins Medicine because the NIS is publicly available data.
Morbidly obese inpatients who underwent bariatric surgery were identified by the International Classification of Diseases, Ninth Revision (ICD-9), and by diagnosis-related group codes. Diagnoses for morbid obesity were confirmed with ICD-9 code 278.01. Morbidly obese patients who underwent bariatric surgery were identified via ICD-9 codes 43.89, 44.68, 44.95, 44.31, 44.38, and 44.39 and diagnosis-related group code 288. We excluded patients with ICD-9 codes for malignant neoplasm of the esophagus, stomach, and small intestine, including the duodenum, pancreas, and unspecified sites (codes 150, 151, 152, 157, and 199). We also excluded patients with ICD-9 codes for gastric ulcer, duodenal ulcer, and peptic ulcer without specification of site (codes 531-533) as a primary diagnosis.
Key patient characteristics of the analysis were age, age category, sex, annual income, type of insurance, comorbidity, and race/ethnicity. Age was treated as a continuous variable. The age category consisted of the following 4 groups: 18 to 34 years, 35 to 49 years, 50 to 64 years, and older than 64 years. Annual income was also stratified into 4 income quartiles: less than $36 000, $36 000 to $44 999, $45 000 to $59 999, and $60 000 or greater. Income was stratified based on the reported quartile of income levels. The types of insurance were categorized into Medicare, Medicaid, private insurance, self-pay, no charge, and other. We grouped patients by their number of comorbidities (0, 1, or 2). There were no patients with more than 2 comorbidities in this sample. We also used the Charlson Comorbidity Index, which calculates the probability of mortality in a given period based on existing comorbidities.24 Race/ethnicity was categorized into the following groups: white, black, Hispanic, Asian or Pacific Islander, Native American, and other.
Statistical analysis was conducted using Stata, version 12 (StataCorp LP). We set α = .05 for all analyses. Population proportions of bariatric surgical procedures were calculated by year. Multiple χ2 tests and unadjusted logistic regressions were conducted to test for differences in patient characteristics across time.
Patient characteristics by year are shown in the Table. Quiz Ref IDThe mean age of patients increased from 43.5 to 44.8 years (P < .001). The proportions of patients in the younger population groups (18-34 years and 35-49 years) significantly decreased from 22.5% to 21.5% (P < .001) and from 42.4% to 39.5% (P < .001), respectively. On the other hand, the proportion of the population who was older than 64 years increased from 2.9% to 7.0% (P < .001). The female population proportion significantly decreased from 80.4% to 78.1% (P < .001).Quiz Ref ID The proportion of patients in the lowest income class, with an annual income of less than $36 000, increased from 20.7% to 22.9% (P < .001). However, the proportion of patients in the highest income class, with an annual income of $60 000 or more, decreased from 25.8% to 23.9% (P < .001). The percentage of patients with Medicare increased from 8.5% to 16.3% (P < .001) and the percentage with Medicaid increased from 6.6% to 11.8% (P < .001). In contrast, the proportion of patients with private insurance decreased from 72.4% to 63.3% (P < .001). Also, the proportion of patients who were classified as self-pay decreased from 7.8% to 4.2% (P < .001). Quiz Ref IDA significant decrease was observed in the percentage of patients who were white, from 74.9% to 69.7% (P < .001). Conversely, an increase in the proportion of patients who were black was observed, from 12.3% to 15.1% (P < .001). The Hispanic population proportion increased from 9.7% to 12.5% (P < .001). The proportion of patients who were Asian or Pacific Islander increased slightly from 0.3% to 0.4% (P < .001).
Socioeconomic disparities in access to bariatric surgery have been a problem for almost a decade. The CMS has attributed this problem to their policy discouraging the use of non-COE facilities.13 Center of Excellence facilities have been shown to provide a higher quality of health care than non-COE facilities. Using the MarketScan Commercial Claims and Encounter Database, Kwon et al14 showed that inpatient mortality, 90-day reoperations, complications, and readmissions decreased from 2003 through 2009 while costs remained the same. Livingston25 also suggested that patients from non-COE facilities were 18% more likely to have postoperative complications because lower procedure volumes were associated with more complications. Birkmeyer et al16 built on this study, showing that serious complications were inversely associated with procedure volumes.
Centers for Medicare & Medicaid Services status may be a marker for a higher risk of adverse surgical outcomes because Medicare and Medicaid patients tend to be older or have multiple medical comorbidities.13 There is evidence that COE certification is effective in reducing the effect of such risk factors. A study by Nguyen et al22 showed significantly decreased complications and mortality rates after NCD implementation for patients insured under Medicare and Medicaid while the surgery volume remained the same. Flum et al15 added to this by finding a reduction in 90-day mortality, serious complications, and readmission among patients using CMS services after NCD; however, they attributed the results to changes in bariatric surgical methods.
In contrast with those who showed the advantages of the COE policy, some researchers have argued that there are no differences between COE and non-COE facilities. Dimick et al21 showed that there were no differences in outcomes between COE facilities and non-COE facilities. Their article suggested that expansion of coverage to non-COE facilities could enhance access without sacrificing patient safety. Furthermore, the authors attributed the improved outcomes to advancement in bariatric surgical procedures rather than implementation of the NCD.15,22 Unfortunately, to our knowledge, no studies have evaluated the direct effect of the NCD on access to health care or surgical outcomes.
Our data analysis showed that disparities in access to bariatric surgical procedures were actually reduced during the implementation of the NCD requiring COE facilities. First, the proportion of patients older than 65 years increased from 2006 through 2011. Second, the proportion of patients who underwent bariatric surgery who were male increased. Although the prevalence of morbid obesity is similar between the sexes, women undergo bariatric surgery more frequently than men.24 The reason for this disparity is unclear.
Third, the disparity among the income classes improved, with the proportion of low-income patients who underwent bariatric surgery increasing significantly. Barriers to access to the procedure for low-income patients had been a concern, especially because obesity is more prevalent in this population. The COE requirement limited bariatric surgery to designated locations, which potentially restricted access. Because only high-volume hospitals could become COE certified, it was feared that patients with limited resources would have difficulty accessing those hospitals. However, our analysis by income level, as identified by zip code, did not support this notion.
Fourth, access for ethnic minorities (black, Hispanic, and Asian or Pacific Islander) improved. Obesity is more prevalent in black and Hispanic populations than in any other ethnic group. In addition, ethnic disparities in access to health care are well known. However, our study found that use of bariatric procedures by minorities increased from 2006 through 2011, suggesting that the COE policy did not impair access.
Quiz Ref IDPerhaps most important, the proportion of patients with Medicare increased significantly. This increase can be attributed to CMS’s decision to cover bariatric surgical procedures in 2006. Despite the fact that the coverage was limited to COE facilities, the proportion of patients with Medicare dramatically increased, which suggests that the COE requirement for coverage was not the underlying cause of the anecdotally reported access problem.
There were several limitations associated with this study. First, because we retrospectively analyzed a large national database, erroneous or missing data may have been present owing to miscoding and mistakes in data entry. Second, we have no clear mechanism for separating the effect of the NCD from that of accreditation because both were implemented at approximately the same time. Furthermore, there were relatively few COE facilities at the start of our study in 2006. The number of centers increased for the duration of the study; this increase may have progressively ameliorated the effect of the NCD during that period. Third, we were unable to test for the difference between COE and non-COE facilities because the NIS database did not provide this information.
A fourth limitation was that only inpatients who underwent bariatric surgery were sampled in our data. Our results, therefore, were not representative of all patients who underwent bariatric surgery in the United States, which would require both inpatient and outpatient data. Nevertheless, our analysis carries significance because 80% of US bariatric surgical procedures are inpatient procedures.26 Finally, the NIS, although a sample of many hospitals and all discharges in the United States, omits some data that are relevant to our investigation. In particular, it fails to display long-term results, such as success in weight loss, comorbid disease resolution, readmission rates, and patient experience. Therefore, we were unable to assess the long-term effectiveness of bariatric surgical procedures performed at COE-certified facilities.
Our study also included multiple strengths. The use of the NIS afforded us a large sample size with a diverse population that is generalizable to the bariatric surgical patient population in the United States. In addition, the period from which the data for our study originated gave it several strengths. First, we analyzed data from the most recent (2011) sample, which gave our study current relevance. We also were able to examine data beginning with the exact year in which the NCD policy was implemented. Thus, we investigated changes over time in various factors associated with bariatric surgery during the years in which the policy was in place.
Our findings do not support the hypothesis that the NCD policy restricting patients who underwent bariatric surgery to COE facilities reduced access to care or increased disparities. Indeed, the decision by CMS to revoke the COE restriction, based largely on anecdote, may have produced unintended harms by decreasing the importance of COE standards. Future studies are needed to clarify the precise effect of the NCD on quality and access to care for patients who undergo bariatric surgery.
Accepted for Publication: May 20, 2015.
Corresponding Author: Kimberly E. Steele, MD, PhD, The Johns Hopkins Center for Bariatric Surgery, Johns Hopkins Bayview Medical Center, 4940 Eastern Ave, Baltimore, MD 21224 (email@example.com).
Published Online: May 20, 2015. doi:10.1001/jamasurg.2015.74.
Author Contributions: Mr Bae had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: Bae, A. Abraham, B. Abraham, Peterson, Magnuson, Schweitzer, Steele.
Acquisition, analysis, or interpretation of data: Bae, Shade, A. Abraham, Peterson, Schneider, Schweitzer, Steele.
Drafting of the manuscript: Bae, A. Abraham.
Critical revision of the manuscript for important intellectual content: All authors.
Statistical analysis: Bae, A. Abraham, B. Abraham, Schneider.
Obtained funding: Bae.
Administrative, technical, or material support: Bae, B. Abraham, Peterson, Schweitzer.
Study supervision: Peterson, Schneider, Magnuson, Steele.
Conflict of Interest Disclosures: Dr Schweitzer reports being a site inspector for the American College of Surgeons. No other disclosures were reported.