Introduction

Socioeconomic status has been found to be a significant predictor of poor outcome in a variety of conditions, including appendicitis [1], breast cancer [2, 3], liver transplantation [4, 5], coronary artery bypass surgery [6], stroke [7], knee arthroplasty [8], and intensive care [9], among others [10].

Patients who undergo bariatric surgery are disproportionately middle aged white women with private insurance. Those who have publicly funded insurance (Medicare/Medicaid) have a decreased chance of selection for surgery (OR, 0.54) [10] and have longer length of stay, increased charges, more comorbidities and increased mortality compared to privately and self-insured patients [11, 12]. Publicly funded patients also have more medical and psychiatric complications after gastric bypass surgery than those who have private medical insurance [13]. While most reports combine Medicare and Medicaid patients under “public assistance”, funding by Medicare may not necessarily mean that the patient is economically deprived. Furthermore, some of the outcome differences in the above studies related to “public assistance”, may have been caused by differences in surgical technique, medical follow-up or initial comorbidities.

The present study was performed to provide a more controlled analysis of outcomes in bariatric patients with the lowest socioeconomic status.

Materials and Methods

The records of 830 consecutive patients with primary open gastric bypass (804) or banded sleeve gastrectomy (26) between 3/9/98 and 1/17/08 were analyzed, excluding patients with renal failure/transplants or revisions. The banded sleeve gastrectomy was a typical sleeve gastrectomy with a band of Alloderm® placed 6 cm from the GE junction using a 37F sizer to prevent subsequent dilatation of the sleeve. Surgical outcomes of the two procedures were the same (data not shown). Performance of the procedures by a single surgeon in the same hospital using the same technique with follow-up in the same office eliminated differences in surgical technique and aftercare. The patients were divided into three groups: (1) Medicaid with or without Medicare to represent socioeconomically deprived patients who could not afford any private insurance (Medicaid), (2) Medicare patients with or without Commercial coinsurance to represent a nondisadvantaged but high-risk group (Medicare), and (3) patients who had Commercial insurance or self-pay who were considered as nondisadvantaged (Commercial). A further breakdown revealed that of the 183 Medicaid patients, there were 78 who also had Medicare. Of the 77 Medicare patients, 13 had Medicare only without coinsurance. Of the 570 Commercial patients, 566 had Commercial insurance only and four were self-pay.

Results

Follow-up was imperfect despite having patients sign contracts, routine telephone calls to patients, mailed questionnaires and inquiries to offices of primary care physicians. Follow-up for deaths was supplemented by accessing deaths reported to Social Security Administration.

Table 1 provides a comparison of the Medicaid, Medicare, and Commercial groups. Preoperatively, the Medicaid and Medicare patients had a higher BMI (58.4 vs. 52.8 vs. 50.9), and more comorbid conditions. There was also an increased occurrence of preoperative thromboembolytic disease and/or cardiac disease in Medicare and Medicaid patients. The percentage of resolution of diabetes mellitus in the Medicaid patients was 76.4% of patients affected compared to 60.5% in the Medicare group and 70.5% in the Commercial group, but the preoperative incidence was greater in the Medicaid and Medicare patients, related to both age and body size.

Table 1 Comparison of initial status and outcomes
Full size table

Mortality at 30 days, 90 days, and 1 year is shown in Table 2 and cause of death is shown in Table 3. Six of the seven patients who died in the Medicaid group had BMIs > 65. Mortality increased in all groups by 3 years: Medicaid 8.0%; Medicare 3.0%; and Commercial 2.0%. However, follow-up was incomplete (about 50%) despite vigorous efforts, making interpretation difficult. There were no deaths in the 19 patients ≥65 years.

Table 2 Mortality in patients at risk at different time intervals
Full size table
Table 3 Causes of death
Full size table

Thirty-three of the 183 patients in the Medicaid group had readmissions potentially related to their surgery during the first year (18.0%). Twelve patients were admitted because of nausea, vomiting, and dehydration and three because of pancreatitis. Five patients had GI bleeding, and three had small bowel obstructions. Ten of the Medicare patients (13.0%) had readmissions, three for nausea, vomiting, and dehydration. In contrast, 59 of the 570 Commercial patients (10.4%) were admitted during their first year. Nausea, vomiting, and dehydration caused the admission of 15 patients, GI bleeding occurred in 12 patients and small bowel obstruction in six. The above data show the Medicaid group had much higher initial BMI and comorbid conditions than the Commercial group, and this was associated with higher death and complications.

In order to help determine whether these differences were related to economic factors or severity of the initial disease, patients were matched by computer (SAS® System Release 9.1, Cary, NC, USA) between the Medicaid and Commercial groups for age and BMI (Table 4). Only patients with follow-up data for 1 year were included in this particular analysis. Of the 140 patients per group, the matched Medicaid patients had more comorbidities both at onset and after 1 year. However, the decrease in BMI and percentage of resolution of comorbidities was similar. For example, 75% of patients with diabetes in the Medicaid group resolved compared to 74.1% in the Commercial group. Twenty of the matched Medicaid patients (14.3%) compared to ten of the matched Commercial patients (7.1%) had admissions during the first year. Admission for nausea, vomiting, and dehydration were more common in the Medicaid patients (six patients vs. two patients).

Table 4 Comparison of matched groups
Full size table

Discussion

The influence of socioeconomic status in morbidly obese patients has been reported before. Carbonell et al. [11] examined the results in 5,876 gastric bypass operations performed in 137 hospitals in 2000. Medicare and Medicaid patients were indicated as one or the other being the primary carrier, but they did not separate the two groups by financial status. Medicare and Medicaid insured patients had longer length of stay, higher charges, more comorbidities and postoperative morbidity and mortality compared to patients with insurance or who paid for the operation themselves. Age, gender and race all played a factor in in-hospital mortality. In the males, in-hospital mortality was 1.7% compared to 0.36% in the females. In our study, gender also had an influence on mortality in Medicaid patients at 1 year (male 3.3% vs. female 0.9%). Their data also reported that Medicaid patients had an in-hospital mortality of 1.70% whereas Medicare had a hospital mortality of 2.00%. Our study focused on Medicaid patients because this ensured that they were financially unable to pay for their own medical care. In contrast, many Medicare patients are financially well off and are receiving Medicare only because of their age [14].

Flum et al. [15] used Medicare claims’ history to evaluate mortality among Medicare beneficiaries undergoing bariatric surgery. Of 16,155 patients undergoing bariatric procedures, the 30-day mortality was 2% and the 1 year mortality was 4.6%. Male gender and older age were associated with a higher death rate. This is similar to our experience with Medicaid but not Medicare patients where Medicare patients had a mortality of 1.6% at 1 year.

Martin et al. [13] examined the effect of preoperative insurance status on results following gastric bypass and showed that patients with public funding had a greater risk of complications. The only factor that was associated with postoperative complications by univariate analysis was lower socioeconomic status. Like our study, the BMI was higher in the publicly funded group (52.2 vs. 47.6). Only two patients died, and these were in the publicly funded group. Dallal et al. [16] reported that both Medicare and Medicaid patients had a prolonged length of stay after bariatric surgery which was not seen in our study.

In contrast to some of the other studies, Livingston and Langert [17] found that poverty itself was not a risk factor for bariatric surgery. Significant factors in their analysis of 25,428 patients were age, male sex, electrolyte disorders, and congestive heart failure. Durkin et al. [18] also reported that financial status did not predict weight loss after bariatric surgery which is consistent with our findings.

In our study, the Medicaid patients had a significantly higher BMI and more severe comorbid conditions than the Medicare or Commercial groups. This is most likely related to poor access to this life-saving procedure. Santry et al. [10] and Livingston and Ko [19] both demonstrated that lower income and public insurance were associated with decreased odds of selection for bariatric surgery. We feel that this under-representation is caused by several factors including prejudice against economically deprived individuals, decreased payment to physicians and hospitals for the care of Medicaid patients (many practices will not take Medicaid patients) and a marked decrease in the ability to obtain approval for their surgery from the various Medicaid agencies. As examples, Medicaid policies vary between states but are extremely restrictive, sometimes requiring BMI > 50 regardless of comorbidities, physician-supervised diet for up to 12 months and prolonged times for review.

Conclusions

Our study is unique in that it eliminates potential differences in hospital, surgical technique and follow-up. Socioeconomic status had no influence on short-term outcome in medically similar patients. The increased risk for inferior outcome in Medicaid patients appears to be related primarily to an inability to be approved for surgical care by Medicaid providers and denial of care because of poor reimbursement. Changes in Medicaid policy to allow treatment before the disease becomes far advanced will improve survival and long-term costs.