- Email: [email protected]
Medication use after bariatric surgery in a managed care cohort
Surgery for Obesity and Related Diseases 4 (2008) 601– 607
Original article
Medication use after bariatric surgery in a managed care cohort Denise M. Hodo, M.P.H.a, Jennifer L. Waller, Ph.D.b, Robert G. Martindale, M.D., Ph.D.c, Donna M. Fick, Ph.D., R.N.d,e,* a
Center for Health Care Improvement, Department of Medicine, Medical College of Georgia, Georgia, Georgia’s Health Sciences University, Augusta, Georgia b Department of Biostatistics, Medical College of Georgia, Georgia’s Health Sciences University, Augusta, Georgia c Department of Surgery, Oregon Health and Sciences University, Portland, Oregon d College of Health and Human Development, Pennsylvania State University School of Nursing, University Park, Pennsylvania e Department of Psychiatry, Pennsylvania State University School of Medicine, University Park, Pennsylvania Received June 19, 2007; revised November 21, 2007; accepted January 10, 2008
Abstract
Background: Bariatric surgery has been shown to provide long-term weight loss, in addition to a significant reduction in obesity-related co-morbidities. The primary aim of this study was to describe the medication use and costs within a managed care cohort after bariatric surgery. A secondary aim was to describe the use rates for other health services after bariatric surgery. Methods: This retrospective cohort study used an administrative database from a large managed care organization to examine the health outcomes in persons 6 months before and 6 months after bariatric surgery. Results: The average number of prescription claims per person decreased after surgery, from 6.93 (SD 7.16) before to 4.88 (SD 5.84) after surgery (P ⬍.001). The average number of claims for office visits decreased from 5.52 before to 3.94 after surgery (P ⫽ .0028), and the average number of claims for outpatient visits decreased from 0.75 before to 0.40 after surgery (P ⬍.001). However, the average number of inpatient visit claims increased after bariatric surgery, from 0.04 (SD 0.31) to 0.07 (SD 0.52) claims per person (P ⫽ .04). In the preoperative period, the paid costs for pharmacy claims were an average of $221.30 (SD $341.25). After surgery, the pharmacy paid costs decreased to an average cost of $158.90 (SD $454.13). Conclusion: Within this sample, medication use and costs decreased within 6 months of bariatric surgery. Decreases were also noted in the postoperative period in several obesity-related comorbidities, office visits, emergency room visits, and outpatient visits; however, an increase occurred in inpatient stays after surgery. (Surg Obes Relat Dis 2008;4:601– 607.) © 2008 American Society for Metabolic and Bariatric Surgery. All rights reserved.
Keywords:
Bariatrics; Obesity; Gastric bypass; Outcomes research
A major public health concern currently facing the United States is the growing number of individuals who are overweight or obese [1,2]. The health risks of being over-
This study was funded by the Center for Health Care Improvement, a collaborative venture between the Medical College of Georgia and BlueCross BlueShield of Georgia. *Reprint requests: Donna M. Fick, Ph.D., A.P.R.N.-B.C., F.G.S.A., Pennsylvania State University, 307C Health and Human Development East, University Park, PA 16802-6509. E-mail: [email protected]
weight and obese are well described and include type 2 diabetes mellitus, hypertension, coronary artery disease, sleep apnea, stroke, cerebrovascular disease, gallstones, osteoarthritis, and some forms of cancer [2,3]. Those who are obese also experience economic ramifications, including approximately 77% greater medication costs relative to normal-weight individuals [4 – 6] and greater medical care costs. One study estimated that the medical care costs of obesity in the United States approach $70 billion [7]. This estimate exceeds the medical care cost burden of many other chronic diseases such as hypertension and
1550-7289/08/$ – see front matter © 2008 American Society for Metabolic and Bariatric Surgery. All rights reserved. doi:10.1016/j.soard.2008.01.008
602
D. M. Hodo et al. / Surgery for Obesity and Related Diseases 4 (2008) 601– 607
coronary heart disease. Additionally, obesity can have a negative effect on quality of life and self-image [7]. Also, the social bias against those who are obese can result in economic disadvantages in employment and educational opportunities [8]. One method of treatment proven to provide lasting weight loss and a reduction of co-morbidities is bariatric surgery [9 –12]. Bariatric surgery has been associated with improvement or resolution of obesity-related co-morbidities such as diabetes, hypertension, obstructive sleep apnea, dyslipidemia, cardiovascular disease, and gastroesophageal reflux disease (GERD) [9,13–20]. Previous studies have suggested that bariatric surgery is associated with decreased healthcare costs for the patient as a result of the decrease in co-morbid conditions after the surgery. Specifically, patients undergoing bariatric surgery have been shown to experience a decrease in prescription costs for diabetic and hypertension medications [5] and overall medical care costs [21]. However, other studies have suggested that hospitalizations can increase after bariatric surgery as a result of surgical complications. Zingmond et al. [22] found that in the year after gastric bypass surgery, substantially more hospitalizations were required for surgery-related complications relative to the numbers of elective procedures before gastric bypass. The primary aim of the present study was to describe the medication use patterns before and after surgery in a large southeastern managed care organization (MCO) cohort of patients who had undergone bariatric surgery. A secondary aim of this study was to describe the health service use patterns within the same cohort before and after bariatric surgery. Methods This retrospective cohort study used data derived from the administrative database of a large MCO, located in the southeast United States, to examine the health outcomes in persons 6 months before and 6 months after bariatric surgery. The Current Procedural Terminology, 4th edition, codes 43846 and 43847 were used to identify individuals who had undergone bariatric surgery from January 1, 2001 to December 31, 2003. The inclusion criteria were the Current Procedural Terminology, 4th edition, codes 43846 and 43847, indicating Roux-en-Y gastric bypass and 12 months of consecutive enrollment. Separate codes for the laparoscopic procedure were not introduced until 2005. Therefore, our sample included both open and laparoscopic procedures. Although the cost considerations for the 2 procedures are different, an examination of the difference in the effect of each on medication use and costs was beyond the scope of this study. At data collection, the MCO had only recently begun to provide coverage for bariatric surgery; therefore, each person included in the cohort had 1 year of continuous claims data, beginning 6 months before bariatric
surgery and ending 6 months after bariatric surgery. The Medical College of Georgia institutional review board for human subjects research approved this study. Study measures The demographic data collected included age, gender, Deyo-adapted Charlson co-morbidity index (CCI), line of service, and MCO region code (determined by the member’s place of residence). The CCI is a validated co-morbidity index that provides a measure of the burden of illness [13]. The CCI has also been validated for use with administrative databases by Deyo et al. [14] by using coding algorithms to map the International Classification of Diseases, 9th edition, clinical modification codes to the 17 CCI co-morbidity variables; this was the version of the CCI used in the present study. Quan et al. [15] found that although the CCI derived from administrative data underestimated some diagnoses relative to the CCI using chart review, the 2 showed moderate overall agreement, with a kappa score of 0.56. The MCO region codes correspond to 10 different geographic regions of the state, as well as separate codes for out-of-state and unknown residence. The lines of service provided by the healthcare insurer included a mix of health maintenance organization managed care, preferred provider organizations, fee for service, and indemnity. Medical claims data were gathered for all healthcare settings to examine the rates of use of various health services. Facility claims data included charges for inpatient hospitalization services as billed on the UB-92 forms. Provider claims data included charges for healthcare provider services as billed on the CMS 1500 forms. Pharmacy claims data included charges for outpatient prescription medications only. These included claims for refills, as well as claims for unique prescriptions without refills. Inpatient medication charges were a part of the facility claims and, thus, were not included in the analysis of prescription use and costs. Healthcare costs were determined from the MCO claims data for payments in U.S. dollars made directly to the provider. Healthcare services use included the number of claims for the following services: outpatient radiology services, outpatient surgery, visits to clinics affiliated with hospitals, home healthcare services, visits to an emergency department, laboratory work, medical equipment, and nursing home admissions. Medication use included the number of outpatient pharmacy claims for unique prescriptions and for refills. Obesity-related conditions Obesity-related conditions were selected on the basis of previous research and clinical expertise [2,3,16]. These included diabetes, sleep apnea, GERD, dyslipidemia, depression, stress incontinence, pseudotumor cerebri, congestive
D. M. Hodo et al. / Surgery for Obesity and Related Diseases 4 (2008) 601– 607 Table 1 Demographic data (n ⫽ 605) Variable Gender Male Female Age (yr) CCI score Before surgery After surgery
n
%
81 524
13.4 86.6
Mean
SD
40.44
9.31
0.42 0.54
0.75 0.88
603
els. All statistical analyses were performed using Statistical Analysis Systems, version 8.2 (SAS Institute, Cary, NC), and the overall significance level was .05. Results
SD ⫽ standard deviation; CCI ⫽ Charlson co-morbidity index.
heart failure, nonalcoholic steatohepatitis, low back pain, degenerative joint disease, pulmonary embolus, sexual dysfunction, venous stasis ulcers, varicose veins, obesity-induced hypoventilation, and hypertension. Also included were outcomes and potential complications of bariatric surgery, including surgical wound infection, gastrointestinal bleeding, splenectomy, nausea and vomiting, and postoperative anastomotic stricture. Obesity-related conditions and outcomes after surgery were defined according to the International Classification of Diseases, 9th edition, clinical modification, codes and, where indicated, Current Procedural Terminology, 4th edition, codes. Use of respiratory aids as defined by the Health Care Financing Administration Common Procedures coding system was also included. (A complete list of codes used can be obtained by contacting D. M. Fick). Statistical analysis Four periods were created to include the 3– 6 months before surgery, the 3 months before surgery to surgery, the date of surgery to 3 months after surgery, and 3– 6 months after surgery. Descriptive statistics were calculated for the demographic variables and for the diagnoses, complications, outcomes, costs, and utilization across the four periods. To examine the differences between the 3– 6 months before surgery and the 3– 6 months after surgery in costs and utilization, a mixed model analysis of covariance (ANCOVA) was used, with gender, age, CCI, and line of business (health maintenance organization, preferred provider organization, fee for service, and indemnity) considered as covariates. The mixed model ANCOVA was used to account for the correlation between points within subject. Because of the large variability in costs and the non-normality, a natural log transformation was used in the mixed model analyses. To examine differences between the 3– 6 months before surgery and the 3– 6 months after surgery in diagnoses and complications, generalized estimating equation (GEE) modeling was used, with gender, age, and CCI considered as covariates. The underlying distribution of the dependent variable assumed in the GEE models was a binomial distribution, and the link function used was the logit. Because of low frequencies, some of the diagnoses and complications could not be examined in the GEE mod-
The total number of members enrolled in the MCO varies from year to year, as new members enroll and others discontinue coverage. The average enrollment in the MCO during the 3-year period was 1,660,712. The incidence of bariatric surgery was 15.5/100,000 for 2001, 40.0/100,000 for 2002, and 31.1/100,000 for 2003. A total of 868 patients (58.9%) did not have continuous enrollment for at least 6 months before and 6 months after surgery and were excluded from the analysis. The average costs for the day of surgery were $10,677.77 (SD $33,988.82) for facility charges, $3,301.42 (SD $1,406.55) for provider charges, and $.85 (SD $14.91) for prescription charges. A total of 605 individuals underwent bariatric surgery during the specified period and met the eligibility criteria. As shown in Table 1, most of the sample were women and had a mean age of 40 years. The most frequently reported region code was the Atlanta region. Because race and ethnicity are not collected by the MCO, we were unable to include this in our analysis. The sample had a mean CCI score of .42 (SD .75) before surgery. The average length of stay during the perioperative period was 3.05 days (SD 4.85). The utilization rates before and after surgery of specific medication classes are listed in Table 2. Psychiatric medications were the most frequently used class of medication, with 222 claims for psychiatric medications during the preoperative period. Although prescriptions for psychiatric medications did decrease in the postoperative period by 19%, this decrease was not statistically significant. However, the second most frequently used class, cardiac medications, decreased by 54% in the postoperative period, from 203 to 93 prescriptions. Both antiasthmatic and diabetes
Table 2 Rate of use by therapeutic class for study population (n ⫽ 605) Therapeutic class
Before surgery
After surgery
Analgesics Antiobesity Antiasthmatics Cardiac Central nervous system Diuretics Gastrointestinal Diabetes Other medical supply Muscle relaxants Psychiatric Skin preparations
127 (20.99) 0 (0) 48 (7.93) 203 (33.55) 36 (5.95) 104 (17.19) 98 (16.20) 86 (12.41) 18 (2.98) 55 (.09) 222 (36.69) 61 (10.08)
140 (23.14) 0 (0) 19 (3.14) 93 (15.37) 26 (4.30) 51 (8.43) 82 (13.55) 34 (5.62) 18 (2.98) 38 (6.28) 180 (29.75) 49 (8.10)
Data presented as numbers, with percentages in parentheses.
604
D. M. Hodo et al. / Surgery for Obesity and Related Diseases 4 (2008) 601– 607
Table 3 Prescription medication use and costs before and after surgery (n ⫽ 605) Outcome
3–6 mo Before surgery
3–6 mo After surgery
F value*
P value
Average No. of claims† Asthma medications Cardiac medications Diuretic medications Gastrointestinal medications Diabetes medications Psychiatric medications Mean prescription cost
0.14 (0.57) 1.03 (1.97) 0.34 (0.85) 0.34 (0.91) 0.56 (1.75) 0.96 (1.61) $221.30 ($341.25)
0.04 (0.27) 0.40 (1.23) 0.14 (0.50) 0.23 (0.68) 0.15 (0.78) 0.81 (1.72) $158.90 ($454.13)
18.03 80.02 37.01 2.74 44.13 3.39 15.21
⬍0.001 ⬍0.001 ⬍0.001 0.098 ⬍0.001 0.066 ⬍0.001
Data presented as mean value, with standard deviation in parentheses. * Mixed-model analysis of covariance used to account for correlation between points within subject, with gender, age, Charlson co-morbidity index, and line of business as covariates. † Refills included in average number of claims.
medication prescriptions decreased by 60%, and diuretic medication prescriptions decreased by 51%. Table 3 lists the results of the ANCOVA for prescription use. Statistically significant differences were seen between the 3– 6-month period before surgery and the 3– 6-month period after surgery, with the period before surgery having significantly greater mean numbers of antiasthmatic prescriptions (P ⬍.001), cardiac prescriptions (P ⬍.001), diuretic prescriptions (P ⬍.001), and diabetic prescriptions (P ⬍.001). The average prescription costs also decreased by 28% during the postoperative period from $221.30 (SD $3441.25) to $158.90 (SD $454.13; P ⬍.001). As listed in Table 4, the mixed model ANCOVA showed statistically significant differences between the 3– 6 months before surgery and the 3– 6 months after surgery, with the period before surgery having significantly greater mean values compared with after surgery for office visits (P ⫽ .003), number of prescriptions (P ⬍.001), and outpatient visits (P ⬍.001). However, inpatient visits had a statistically significant increase during the 3– 6-month period after surgery, from an average of 0.04 (SD 0.31) to 0.07 (SD 0.52; P ⫽ .04). During the 6-month period after surgery (surgery to 3 months and 3– 6 months after surgery), 24 claims for
Table 4 Differences in average number of claims for specific health services before and after surgery (n ⫽ 605) Service
3–6-mo Preoperatively
3–6-mo Postoperatively
F value*
P value
Office visits Prescriptions Inpatient visits Emergency room Outpatient visits
5.52 (6.63) 6.93 (7.16) 0.04 (0.31) 0.11 (0.41)
3.94 (4.69) 4.88 (5.84) 0.07 (0.52) 0.09 (0.40)
8.98 41.20 4.20 0.04
0.003 ⬍0.001 0.041 0.843
0.75 (1.12)
0.40 (0.96)
23.09
⬍0.001
Data presented as mean value, with standard deviation in parentheses. * Mixed-model analysis of covariance used to account for correlation between points within subject, with gender, age, Charlson co-morbidity index, and line of business as covariates.
surgical wound infection (3.97%), 6 (0.99%) for postoperative anastomotic stricture, and 5 (0.83%) for pulmonary embolism were made within our study sample. The results of the GEE models for diagnoses, controlling for gender, age, and CCI score, are listed in Table 5. The results showed that patients were significantly less likely to have sleep apnea (odds ratio 0.35, 95% confidence interval 0.24 – 0.52), GERD (odds ratio 0.44, 95% confidence interval 0.26 – 0.75), or hypertension (odds ratio 0.53, 95% confidence interval 0.39 – 0.73) diagnoses in the 3– 6 months after surgery relative to the 3– 6 months before surgery. Patients were significantly more likely to have nausea and vomiting (odds ratio 9.36, 95% confidence interval 2.70 – 32.42) in the 3– 6 months after surgery. Because of the low
Table 5 Preoperative and postoperative rates of co-morbidities and results of generalized estimating equation model (n ⫽ 605) Outcome
3–6 mo Before surgery
Diabetes 82 Sleep apnea 128 GERD 57 Dyslipidemia 15 Depression 22 CHF 6 Low back pain 50 Degenerative joint 45 disease Hypertension 173 Nausea and 3 vomiting Pulmonary embolus 0 Wound infection 1 Postoperative 0 anastomotic stricture
3–6 mo After surgery
Odds 95% CI ratio
P value
55 49 23 22 21 2 39 34
0.71 0.35 0.44 1.68 1.06 0.26 0.87 0.92
0.43–1.17 0.24–0.52 0.26–0.75 0.82–3.42 0.56–2.01 0.04–1.76 0.55–1.38 0.56–1.49
0.1752 ⬍0.0001 0.0024 0.1538 0.8636 0.1674 0.5661 0.7196
100 24
0.53 9.36
0.39–0.73 2.70–32.42
⬍0.0001 0.0004
2 2 2
OR ⫽ odds ratio; CI ⫽ confidence interval; GERD ⫽ gastroesophageal reflux disease; CHF ⫽ congestive heart failure. Data presented as number of claims for each diagnosis.
D. M. Hodo et al. / Surgery for Obesity and Related Diseases 4 (2008) 601– 607
frequency of occurrence of surgical complications, they could not be included in the GEE models; however, the frequency of claims for each is listed in Table 5. Discussion Obesity has the ability to negatively affect all aspects of life for those who are obese. The primary aim of this study was to describe the medication use after bariatric surgery and to examine the prescription costs after surgery. Also, we sought to describe the health outcomes and service utilization patterns after bariatric surgery. Decreases in the rates of medication use ranged from a 19% decrease in psychiatric prescriptions to a 60% decrease in diabetic and antiasthmatic medications. Along with this decrease in use rates, the pharmacy paid costs decreased by 28% in the 6 months after bariatric surgery. This outcome is significant in light of the approximately 77% greater prescription costs for obese individuals relative to normal-weight individuals as has been reported elsewhere [5,6]. Similar results were found in a study by Potteiger et al. [10]. They examined the costs and use rates of diabetic and hypertension medications before and after bariatric surgery in 51 consecutive patients. The costs were measured from the perspective of the patient and were determined from national average drug prices. Their results showed a dramatic reduction of both diabetes and hypertension medications, as well as reductions in prescription costs per month. Another study of 50 consecutive bariatric surgery patients found dramatic reduction in prescription use and cost. Gould et al. [9] measured the numbers of prescriptions and costs using prices listed in an on-line pharmacy. Both of these studies were conducted with relatively small samples at 1 facility. Our study used claims data from multiple centers and multiple surgeons, providing important evidence for the effect that bariatric surgery can have on medication use and costs. A reduction in health services use in the 6 months after bariatric surgery was also found. The greatest decrease was found in the average number of claims for outpatient visits, which decreased by 47% in the postoperative period. Also significant was the 29% decrease in the average number of claims for physician office visits. Although not statistically significant, the average number of claims for emergency department visits decreased by 18% after bariatric surgery. The decreases in prescription use and costs and decreases in office visits, emergency department visits, and outpatient visits were somewhat overshadowed by the significant increase in claims for inpatient stays during the postoperative period. A comparison of the numbers of claims for inpatient hospital stays in the 3– 6 months before surgery and the 3– 6 months after surgery revealed a 75% increase. Considering that the average number of inpatient claims in the 3– 6 months after surgery was .07 per person, or a total of 42 claims of 605 individuals, along with the relatively low
605
complication rates in our sample, it is likely that the significant increase in postoperative inpatient stays resulted from a small number of people who experienced significant complications. Our results are similar to those recently by Encinosa et al. [17], who found increasing rates of hospital admissions because of surgical complications for 6 months after the initial surgery. They did not measure the use rates before surgery but examined differences in complication rates and health services use from surgery to 3 months compared with 3– 6 months after surgery. They found that use because of surgical complications did increase over time. A recent report by Zingmond et al. [12] also found a significant increase in hospital admissions during the 3 years after bariatric surgery relative to the 3 years before bariatric surgery in a cohort of 60,777 California residents. In the first year after surgery, the most common reasons for hospital admission were related to complications associated with the procedure. The number of gastrointestinal or bypass surgery-related reasons for admission was similar to those of elective procedures in the second year. In the third year after bariatric surgery, the most common reasons for admission were elective procedures, including plastic surgery. However, these results are in contrast to those by Christou et al. [18], who found a decrease in hospitalization rates during the 5 years after bariatric surgery. The study used administrative claims data from the Regie de l’assurance maladie du Québec database in Quebec, Canada. In their cohort of 1035 bariatric surgery patients, significantly fewer hospitalizations occurred compared with the nonsurgically treated control group. It is possible, however, that any initial increase in the rates of hospitalization after bariatric surgery was offset by the significant decreases in hospitalization as time progressed. Additionally, their results showed significantly lower healthcare costs (presented in Canadian dollars) for those who underwent bariatric surgery. A similar study that used data from the same database found that the initially greater hospital costs remained for 3.5 years after surgery and then decreased substantially [11]. After the 3.5-year point, patients who had undergone bariatric surgery actually had significantly fewer hospital admissions and costs relative to the nonsurgically treated control group. The cost savings after that point were so substantial that when the 2 groups were compared, the overall costs after 5 years in the surgery group were less than that of the control group, even with the costs of surgery included. Although these results are impressive, 1 potential drawback exists. Because of the differences in payor systems between the United States and Canada, the economic results they found might not be applicable to patients and payors in the United States. The patients in our study were less likely to have sleep apnea, GERD, or hypertension 6 months after bariatric surgery. However, our study did not show a decrease in several other obesity-related co-morbidities, such as diabetes, dyslipi-
606
D. M. Hodo et al. / Surgery for Obesity and Related Diseases 4 (2008) 601– 607
demia, and depression, such as been demonstrated in other studies [8,19]. It is possible that these findings could have resulted from the time frame used in the present study, as well as physician documentation practices regarding chronic diseases, such as diabetes. This analysis covered a span of 6 months before surgery through 6 months after surgery, which might not have been long enough to demonstrate a decrease in these chronic conditions in the administrative data. Conversely, it is also possible that bariatric surgery did not result in as great of a co-morbidity reduction for this cohort as has been reported elsewhere [8,19]. Christou et al. [18] found a significant reduction in co-morbidities such as cancer, diabetes, and cardiovascular diseases during a 5-year follow-up period and also found a significantly reduced mortality rate among the patients who underwent bariatric surgery, even with the inclusion of perioperative deaths. Limitations The primary limitation of this study was the relatively short follow-up period. When the data collection took place, the MCO had only recently begun offering coverage for bariatric surgery. It is likely that the short follow-up did not allow for the detection within the claims data of decreases in the comorbid conditions that typically respond to weight loss. Although administrative claims data have proved to be a rich source of data for outcomes studies, they have several associated limitations and biases. Because claims data are generated through real-world patient encounters, the information is more generalizable relative to a randomized controlled trial. However, this increase in external validity potentially introduces a decrease in the internal validity through documentation or coding errors, as well as undercoding or overcoding. Additionally, it is possible that all services and diagnoses might not be recorded on an insurance claim form, because of the limitations in the numbers of procedures and diagnoses that can be recorded [20]. Claims data are also subject to issues introduced by plan design and changes that can occur over time [21]. In our study, we attempted to control for any changes that might have occurred through the use of line of business as a covariate in the ANCOVA and GEE models. Although this probably did not account for every change in pharmacy benefits that could have occurred, because our study results are similar to those of previously published reports, it is unlikely that plan changes accounted for all the reductions in medication use and costs. Important clinical data that were unavailable included preoperative and postoperative body mass index, laboratory values, and measures of patient satisfaction with the outcomes. The sample was drawn from an administrative claims database from a large, southeastern MCO; therefore, the findings might not be generalizable to populations outside of this region and insurance provider.
Conclusion Our research suggests that bariatric surgery can have a significant effect on medication use and costs, even within a relatively short follow-up period. This finding is important, given the evidence that obese individuals use more health services and experience greater costs relative to normal-weight individuals and that prescription medication use plays a substantial role in this increase [5]. The reduction in the use of medications after bariatric surgery could translate into cost savings for the MCO, given the ever-increasing epidemic of obesity. Although a true cost-effectiveness analysis was beyond the aims and scope of this study, our research suggests that the cost savings from decreased medication use after bariatric surgery can be realized within a relatively short period. In the short term, this might not offset the cost of surgery and potentially increased hospitalization resulting from surgery; however, the long-range costs of obesity itself in terms of the greater use of resources might be far more costly to the health system than the initial costs associated with bariatric surgery.
Disclosures The authors claim no commercial associations that might be a conflict of interest in relation to this article.
References [1] Ogden CL, Carroll MD, Curtin LR, McDowell MA, Tabak CJ, Flegal KM. Prevalence of overweight and obesity in the United States, 1999 –2004. JAMA 2006;295:1549 –55. [2] Stein CJ, Colditz GA. The epidemic of obesity. J Clin Endocrinol Metab 2004;89:2522–5. [3] National Task Force on the Prevention and Treatment of Obesity. Overweight, obesity, and health risk. Arch Intern Med 2000;160:898 –904. [4] Fontaine KR, Redden DT, Wang C, Westfall AO, Allison DB. Years of life lost due to obesity. JAMA 2003;289:187–93. [5] Raebel MA, Malone DC, Conner DA, Xu S, Porter JA, Lanty FA. Health services use and health care costs of obese and nonobese individuals. Arch Intern Med 2004;164:2135– 40. [6] Sturm R. The effects of obesity, smoking, and drinking on medical problems and costs. Health Aff (Millwood) 2002;21:245–53. [7] Buchwald H. Bariatric surgery for morbid obesity: health implications for patients, health professionals, and third-party payers. J Am Coll Surg 2005;200:593– 604. [8] Buchwald H, Avidor Y, Braunwald E, et al. Bariatric surgery: a systematic review and meta-analysis. JAMA 2004;292:1724 –37. [9] Gould JC, Garren MJ, Starling JR. Laparoscopic gastric bypass results in decreased prescription medication costs within 6 months. J Gastronitest Surg 2004;8:983–7. [10] Potteiger CE, Paragi PR, Inverso NA, et al. Bariatric surgery: shedding the monetary weight of prescription costs in the managed care arena. Obes Surg 2004;14:725–30. [11] Sampalis JS, Liberman M, Auger S, Christou NV. The impact of weight reduction surgery on health-care costs in morbidly obese patients. Obes Surg 2004;14:939 – 47.
D. M. Hodo et al. / Surgery for Obesity and Related Diseases 4 (2008) 601– 607 [12] Zingmond DS, McGory ML, Ko CY. Hospitalization before and after gastric bypass surgery. JAMA 2005;294:1918 –24. [13] Charlson ME, Pompei P, Ales KL, MacKenzie CR. A new method of classifying prognostic comorbidity in longitudinal studies: development and validation. J Chron Dis 1987;40:373– 83. [14] Deyo RA, Cherkin DC, Ciol MA. Adapting a clinical comorbidity index for use with ICD-9-CM administrative databases. J Clin Epidemiol 1992;45:613–19. [15] Quan H, Parsons GA, Ghali WA. Validity of information on comorbidity derived from ICD-9-CCM administrative data. Med Care 2002; 40:675– 85. [16] Roberts RE, Deleger S, Strawbridge WJ, Kaplan GA. Prospective association between obesity and depression: evidence from the Alameda County Study. Int J Obes 2003;27:514 –21.
607
[17] Encinosa WE, Bernard DM, Chen CC, Steiner CA. Healthcare utilization and outcomes after bariatric surgery. Med Care 2006;44:706 –12. [18] Christou NV, Sampalis JS, Liberman M, et al. Surgery decreases long-term mortality, morbidity, and health care use in morbidly obese patients. Ann Surg 2004;240:416 –24. [19] Sugerman HJ, Wolfe LG, Sica DA, Clore JN. Diabetes and hypertension in severe obesity and effects of gastric bypass-induced weight loss. Ann Surg 2003;237:751– 8. [20] Birnbaum HG, Cremieux PY, Greenberg PE, LeLorier J, Ostrander J, Venditti L. Using healthcare claims data for outcomes research and pharmacoeconomic analyses. Pharmacoeconomics 1999;16:1– 8. [21] Motheral BR, Fairman MA. The use of claims databases for outcomes research: rationale, challenges, and strategies. Clin Ther 1997;19:346 – 66.
Original article
Medication use after bariatric surgery in a managed care cohort Denise M. Hodo, M.P.H.a, Jennifer L. Waller, Ph.D.b, Robert G. Martindale, M.D., Ph.D.c, Donna M. Fick, Ph.D., R.N.d,e,* a
Center for Health Care Improvement, Department of Medicine, Medical College of Georgia, Georgia, Georgia’s Health Sciences University, Augusta, Georgia b Department of Biostatistics, Medical College of Georgia, Georgia’s Health Sciences University, Augusta, Georgia c Department of Surgery, Oregon Health and Sciences University, Portland, Oregon d College of Health and Human Development, Pennsylvania State University School of Nursing, University Park, Pennsylvania e Department of Psychiatry, Pennsylvania State University School of Medicine, University Park, Pennsylvania Received June 19, 2007; revised November 21, 2007; accepted January 10, 2008
Abstract
Background: Bariatric surgery has been shown to provide long-term weight loss, in addition to a significant reduction in obesity-related co-morbidities. The primary aim of this study was to describe the medication use and costs within a managed care cohort after bariatric surgery. A secondary aim was to describe the use rates for other health services after bariatric surgery. Methods: This retrospective cohort study used an administrative database from a large managed care organization to examine the health outcomes in persons 6 months before and 6 months after bariatric surgery. Results: The average number of prescription claims per person decreased after surgery, from 6.93 (SD 7.16) before to 4.88 (SD 5.84) after surgery (P ⬍.001). The average number of claims for office visits decreased from 5.52 before to 3.94 after surgery (P ⫽ .0028), and the average number of claims for outpatient visits decreased from 0.75 before to 0.40 after surgery (P ⬍.001). However, the average number of inpatient visit claims increased after bariatric surgery, from 0.04 (SD 0.31) to 0.07 (SD 0.52) claims per person (P ⫽ .04). In the preoperative period, the paid costs for pharmacy claims were an average of $221.30 (SD $341.25). After surgery, the pharmacy paid costs decreased to an average cost of $158.90 (SD $454.13). Conclusion: Within this sample, medication use and costs decreased within 6 months of bariatric surgery. Decreases were also noted in the postoperative period in several obesity-related comorbidities, office visits, emergency room visits, and outpatient visits; however, an increase occurred in inpatient stays after surgery. (Surg Obes Relat Dis 2008;4:601– 607.) © 2008 American Society for Metabolic and Bariatric Surgery. All rights reserved.
Keywords:
Bariatrics; Obesity; Gastric bypass; Outcomes research
A major public health concern currently facing the United States is the growing number of individuals who are overweight or obese [1,2]. The health risks of being over-
This study was funded by the Center for Health Care Improvement, a collaborative venture between the Medical College of Georgia and BlueCross BlueShield of Georgia. *Reprint requests: Donna M. Fick, Ph.D., A.P.R.N.-B.C., F.G.S.A., Pennsylvania State University, 307C Health and Human Development East, University Park, PA 16802-6509. E-mail: [email protected]
weight and obese are well described and include type 2 diabetes mellitus, hypertension, coronary artery disease, sleep apnea, stroke, cerebrovascular disease, gallstones, osteoarthritis, and some forms of cancer [2,3]. Those who are obese also experience economic ramifications, including approximately 77% greater medication costs relative to normal-weight individuals [4 – 6] and greater medical care costs. One study estimated that the medical care costs of obesity in the United States approach $70 billion [7]. This estimate exceeds the medical care cost burden of many other chronic diseases such as hypertension and
1550-7289/08/$ – see front matter © 2008 American Society for Metabolic and Bariatric Surgery. All rights reserved. doi:10.1016/j.soard.2008.01.008
602
D. M. Hodo et al. / Surgery for Obesity and Related Diseases 4 (2008) 601– 607
coronary heart disease. Additionally, obesity can have a negative effect on quality of life and self-image [7]. Also, the social bias against those who are obese can result in economic disadvantages in employment and educational opportunities [8]. One method of treatment proven to provide lasting weight loss and a reduction of co-morbidities is bariatric surgery [9 –12]. Bariatric surgery has been associated with improvement or resolution of obesity-related co-morbidities such as diabetes, hypertension, obstructive sleep apnea, dyslipidemia, cardiovascular disease, and gastroesophageal reflux disease (GERD) [9,13–20]. Previous studies have suggested that bariatric surgery is associated with decreased healthcare costs for the patient as a result of the decrease in co-morbid conditions after the surgery. Specifically, patients undergoing bariatric surgery have been shown to experience a decrease in prescription costs for diabetic and hypertension medications [5] and overall medical care costs [21]. However, other studies have suggested that hospitalizations can increase after bariatric surgery as a result of surgical complications. Zingmond et al. [22] found that in the year after gastric bypass surgery, substantially more hospitalizations were required for surgery-related complications relative to the numbers of elective procedures before gastric bypass. The primary aim of the present study was to describe the medication use patterns before and after surgery in a large southeastern managed care organization (MCO) cohort of patients who had undergone bariatric surgery. A secondary aim of this study was to describe the health service use patterns within the same cohort before and after bariatric surgery. Methods This retrospective cohort study used data derived from the administrative database of a large MCO, located in the southeast United States, to examine the health outcomes in persons 6 months before and 6 months after bariatric surgery. The Current Procedural Terminology, 4th edition, codes 43846 and 43847 were used to identify individuals who had undergone bariatric surgery from January 1, 2001 to December 31, 2003. The inclusion criteria were the Current Procedural Terminology, 4th edition, codes 43846 and 43847, indicating Roux-en-Y gastric bypass and 12 months of consecutive enrollment. Separate codes for the laparoscopic procedure were not introduced until 2005. Therefore, our sample included both open and laparoscopic procedures. Although the cost considerations for the 2 procedures are different, an examination of the difference in the effect of each on medication use and costs was beyond the scope of this study. At data collection, the MCO had only recently begun to provide coverage for bariatric surgery; therefore, each person included in the cohort had 1 year of continuous claims data, beginning 6 months before bariatric
surgery and ending 6 months after bariatric surgery. The Medical College of Georgia institutional review board for human subjects research approved this study. Study measures The demographic data collected included age, gender, Deyo-adapted Charlson co-morbidity index (CCI), line of service, and MCO region code (determined by the member’s place of residence). The CCI is a validated co-morbidity index that provides a measure of the burden of illness [13]. The CCI has also been validated for use with administrative databases by Deyo et al. [14] by using coding algorithms to map the International Classification of Diseases, 9th edition, clinical modification codes to the 17 CCI co-morbidity variables; this was the version of the CCI used in the present study. Quan et al. [15] found that although the CCI derived from administrative data underestimated some diagnoses relative to the CCI using chart review, the 2 showed moderate overall agreement, with a kappa score of 0.56. The MCO region codes correspond to 10 different geographic regions of the state, as well as separate codes for out-of-state and unknown residence. The lines of service provided by the healthcare insurer included a mix of health maintenance organization managed care, preferred provider organizations, fee for service, and indemnity. Medical claims data were gathered for all healthcare settings to examine the rates of use of various health services. Facility claims data included charges for inpatient hospitalization services as billed on the UB-92 forms. Provider claims data included charges for healthcare provider services as billed on the CMS 1500 forms. Pharmacy claims data included charges for outpatient prescription medications only. These included claims for refills, as well as claims for unique prescriptions without refills. Inpatient medication charges were a part of the facility claims and, thus, were not included in the analysis of prescription use and costs. Healthcare costs were determined from the MCO claims data for payments in U.S. dollars made directly to the provider. Healthcare services use included the number of claims for the following services: outpatient radiology services, outpatient surgery, visits to clinics affiliated with hospitals, home healthcare services, visits to an emergency department, laboratory work, medical equipment, and nursing home admissions. Medication use included the number of outpatient pharmacy claims for unique prescriptions and for refills. Obesity-related conditions Obesity-related conditions were selected on the basis of previous research and clinical expertise [2,3,16]. These included diabetes, sleep apnea, GERD, dyslipidemia, depression, stress incontinence, pseudotumor cerebri, congestive
D. M. Hodo et al. / Surgery for Obesity and Related Diseases 4 (2008) 601– 607 Table 1 Demographic data (n ⫽ 605) Variable Gender Male Female Age (yr) CCI score Before surgery After surgery
n
%
81 524
13.4 86.6
Mean
SD
40.44
9.31
0.42 0.54
0.75 0.88
603
els. All statistical analyses were performed using Statistical Analysis Systems, version 8.2 (SAS Institute, Cary, NC), and the overall significance level was .05. Results
SD ⫽ standard deviation; CCI ⫽ Charlson co-morbidity index.
heart failure, nonalcoholic steatohepatitis, low back pain, degenerative joint disease, pulmonary embolus, sexual dysfunction, venous stasis ulcers, varicose veins, obesity-induced hypoventilation, and hypertension. Also included were outcomes and potential complications of bariatric surgery, including surgical wound infection, gastrointestinal bleeding, splenectomy, nausea and vomiting, and postoperative anastomotic stricture. Obesity-related conditions and outcomes after surgery were defined according to the International Classification of Diseases, 9th edition, clinical modification, codes and, where indicated, Current Procedural Terminology, 4th edition, codes. Use of respiratory aids as defined by the Health Care Financing Administration Common Procedures coding system was also included. (A complete list of codes used can be obtained by contacting D. M. Fick). Statistical analysis Four periods were created to include the 3– 6 months before surgery, the 3 months before surgery to surgery, the date of surgery to 3 months after surgery, and 3– 6 months after surgery. Descriptive statistics were calculated for the demographic variables and for the diagnoses, complications, outcomes, costs, and utilization across the four periods. To examine the differences between the 3– 6 months before surgery and the 3– 6 months after surgery in costs and utilization, a mixed model analysis of covariance (ANCOVA) was used, with gender, age, CCI, and line of business (health maintenance organization, preferred provider organization, fee for service, and indemnity) considered as covariates. The mixed model ANCOVA was used to account for the correlation between points within subject. Because of the large variability in costs and the non-normality, a natural log transformation was used in the mixed model analyses. To examine differences between the 3– 6 months before surgery and the 3– 6 months after surgery in diagnoses and complications, generalized estimating equation (GEE) modeling was used, with gender, age, and CCI considered as covariates. The underlying distribution of the dependent variable assumed in the GEE models was a binomial distribution, and the link function used was the logit. Because of low frequencies, some of the diagnoses and complications could not be examined in the GEE mod-
The total number of members enrolled in the MCO varies from year to year, as new members enroll and others discontinue coverage. The average enrollment in the MCO during the 3-year period was 1,660,712. The incidence of bariatric surgery was 15.5/100,000 for 2001, 40.0/100,000 for 2002, and 31.1/100,000 for 2003. A total of 868 patients (58.9%) did not have continuous enrollment for at least 6 months before and 6 months after surgery and were excluded from the analysis. The average costs for the day of surgery were $10,677.77 (SD $33,988.82) for facility charges, $3,301.42 (SD $1,406.55) for provider charges, and $.85 (SD $14.91) for prescription charges. A total of 605 individuals underwent bariatric surgery during the specified period and met the eligibility criteria. As shown in Table 1, most of the sample were women and had a mean age of 40 years. The most frequently reported region code was the Atlanta region. Because race and ethnicity are not collected by the MCO, we were unable to include this in our analysis. The sample had a mean CCI score of .42 (SD .75) before surgery. The average length of stay during the perioperative period was 3.05 days (SD 4.85). The utilization rates before and after surgery of specific medication classes are listed in Table 2. Psychiatric medications were the most frequently used class of medication, with 222 claims for psychiatric medications during the preoperative period. Although prescriptions for psychiatric medications did decrease in the postoperative period by 19%, this decrease was not statistically significant. However, the second most frequently used class, cardiac medications, decreased by 54% in the postoperative period, from 203 to 93 prescriptions. Both antiasthmatic and diabetes
Table 2 Rate of use by therapeutic class for study population (n ⫽ 605) Therapeutic class
Before surgery
After surgery
Analgesics Antiobesity Antiasthmatics Cardiac Central nervous system Diuretics Gastrointestinal Diabetes Other medical supply Muscle relaxants Psychiatric Skin preparations
127 (20.99) 0 (0) 48 (7.93) 203 (33.55) 36 (5.95) 104 (17.19) 98 (16.20) 86 (12.41) 18 (2.98) 55 (.09) 222 (36.69) 61 (10.08)
140 (23.14) 0 (0) 19 (3.14) 93 (15.37) 26 (4.30) 51 (8.43) 82 (13.55) 34 (5.62) 18 (2.98) 38 (6.28) 180 (29.75) 49 (8.10)
Data presented as numbers, with percentages in parentheses.
604
D. M. Hodo et al. / Surgery for Obesity and Related Diseases 4 (2008) 601– 607
Table 3 Prescription medication use and costs before and after surgery (n ⫽ 605) Outcome
3–6 mo Before surgery
3–6 mo After surgery
F value*
P value
Average No. of claims† Asthma medications Cardiac medications Diuretic medications Gastrointestinal medications Diabetes medications Psychiatric medications Mean prescription cost
0.14 (0.57) 1.03 (1.97) 0.34 (0.85) 0.34 (0.91) 0.56 (1.75) 0.96 (1.61) $221.30 ($341.25)
0.04 (0.27) 0.40 (1.23) 0.14 (0.50) 0.23 (0.68) 0.15 (0.78) 0.81 (1.72) $158.90 ($454.13)
18.03 80.02 37.01 2.74 44.13 3.39 15.21
⬍0.001 ⬍0.001 ⬍0.001 0.098 ⬍0.001 0.066 ⬍0.001
Data presented as mean value, with standard deviation in parentheses. * Mixed-model analysis of covariance used to account for correlation between points within subject, with gender, age, Charlson co-morbidity index, and line of business as covariates. † Refills included in average number of claims.
medication prescriptions decreased by 60%, and diuretic medication prescriptions decreased by 51%. Table 3 lists the results of the ANCOVA for prescription use. Statistically significant differences were seen between the 3– 6-month period before surgery and the 3– 6-month period after surgery, with the period before surgery having significantly greater mean numbers of antiasthmatic prescriptions (P ⬍.001), cardiac prescriptions (P ⬍.001), diuretic prescriptions (P ⬍.001), and diabetic prescriptions (P ⬍.001). The average prescription costs also decreased by 28% during the postoperative period from $221.30 (SD $3441.25) to $158.90 (SD $454.13; P ⬍.001). As listed in Table 4, the mixed model ANCOVA showed statistically significant differences between the 3– 6 months before surgery and the 3– 6 months after surgery, with the period before surgery having significantly greater mean values compared with after surgery for office visits (P ⫽ .003), number of prescriptions (P ⬍.001), and outpatient visits (P ⬍.001). However, inpatient visits had a statistically significant increase during the 3– 6-month period after surgery, from an average of 0.04 (SD 0.31) to 0.07 (SD 0.52; P ⫽ .04). During the 6-month period after surgery (surgery to 3 months and 3– 6 months after surgery), 24 claims for
Table 4 Differences in average number of claims for specific health services before and after surgery (n ⫽ 605) Service
3–6-mo Preoperatively
3–6-mo Postoperatively
F value*
P value
Office visits Prescriptions Inpatient visits Emergency room Outpatient visits
5.52 (6.63) 6.93 (7.16) 0.04 (0.31) 0.11 (0.41)
3.94 (4.69) 4.88 (5.84) 0.07 (0.52) 0.09 (0.40)
8.98 41.20 4.20 0.04
0.003 ⬍0.001 0.041 0.843
0.75 (1.12)
0.40 (0.96)
23.09
⬍0.001
Data presented as mean value, with standard deviation in parentheses. * Mixed-model analysis of covariance used to account for correlation between points within subject, with gender, age, Charlson co-morbidity index, and line of business as covariates.
surgical wound infection (3.97%), 6 (0.99%) for postoperative anastomotic stricture, and 5 (0.83%) for pulmonary embolism were made within our study sample. The results of the GEE models for diagnoses, controlling for gender, age, and CCI score, are listed in Table 5. The results showed that patients were significantly less likely to have sleep apnea (odds ratio 0.35, 95% confidence interval 0.24 – 0.52), GERD (odds ratio 0.44, 95% confidence interval 0.26 – 0.75), or hypertension (odds ratio 0.53, 95% confidence interval 0.39 – 0.73) diagnoses in the 3– 6 months after surgery relative to the 3– 6 months before surgery. Patients were significantly more likely to have nausea and vomiting (odds ratio 9.36, 95% confidence interval 2.70 – 32.42) in the 3– 6 months after surgery. Because of the low
Table 5 Preoperative and postoperative rates of co-morbidities and results of generalized estimating equation model (n ⫽ 605) Outcome
3–6 mo Before surgery
Diabetes 82 Sleep apnea 128 GERD 57 Dyslipidemia 15 Depression 22 CHF 6 Low back pain 50 Degenerative joint 45 disease Hypertension 173 Nausea and 3 vomiting Pulmonary embolus 0 Wound infection 1 Postoperative 0 anastomotic stricture
3–6 mo After surgery
Odds 95% CI ratio
P value
55 49 23 22 21 2 39 34
0.71 0.35 0.44 1.68 1.06 0.26 0.87 0.92
0.43–1.17 0.24–0.52 0.26–0.75 0.82–3.42 0.56–2.01 0.04–1.76 0.55–1.38 0.56–1.49
0.1752 ⬍0.0001 0.0024 0.1538 0.8636 0.1674 0.5661 0.7196
100 24
0.53 9.36
0.39–0.73 2.70–32.42
⬍0.0001 0.0004
2 2 2
OR ⫽ odds ratio; CI ⫽ confidence interval; GERD ⫽ gastroesophageal reflux disease; CHF ⫽ congestive heart failure. Data presented as number of claims for each diagnosis.
D. M. Hodo et al. / Surgery for Obesity and Related Diseases 4 (2008) 601– 607
frequency of occurrence of surgical complications, they could not be included in the GEE models; however, the frequency of claims for each is listed in Table 5. Discussion Obesity has the ability to negatively affect all aspects of life for those who are obese. The primary aim of this study was to describe the medication use after bariatric surgery and to examine the prescription costs after surgery. Also, we sought to describe the health outcomes and service utilization patterns after bariatric surgery. Decreases in the rates of medication use ranged from a 19% decrease in psychiatric prescriptions to a 60% decrease in diabetic and antiasthmatic medications. Along with this decrease in use rates, the pharmacy paid costs decreased by 28% in the 6 months after bariatric surgery. This outcome is significant in light of the approximately 77% greater prescription costs for obese individuals relative to normal-weight individuals as has been reported elsewhere [5,6]. Similar results were found in a study by Potteiger et al. [10]. They examined the costs and use rates of diabetic and hypertension medications before and after bariatric surgery in 51 consecutive patients. The costs were measured from the perspective of the patient and were determined from national average drug prices. Their results showed a dramatic reduction of both diabetes and hypertension medications, as well as reductions in prescription costs per month. Another study of 50 consecutive bariatric surgery patients found dramatic reduction in prescription use and cost. Gould et al. [9] measured the numbers of prescriptions and costs using prices listed in an on-line pharmacy. Both of these studies were conducted with relatively small samples at 1 facility. Our study used claims data from multiple centers and multiple surgeons, providing important evidence for the effect that bariatric surgery can have on medication use and costs. A reduction in health services use in the 6 months after bariatric surgery was also found. The greatest decrease was found in the average number of claims for outpatient visits, which decreased by 47% in the postoperative period. Also significant was the 29% decrease in the average number of claims for physician office visits. Although not statistically significant, the average number of claims for emergency department visits decreased by 18% after bariatric surgery. The decreases in prescription use and costs and decreases in office visits, emergency department visits, and outpatient visits were somewhat overshadowed by the significant increase in claims for inpatient stays during the postoperative period. A comparison of the numbers of claims for inpatient hospital stays in the 3– 6 months before surgery and the 3– 6 months after surgery revealed a 75% increase. Considering that the average number of inpatient claims in the 3– 6 months after surgery was .07 per person, or a total of 42 claims of 605 individuals, along with the relatively low
605
complication rates in our sample, it is likely that the significant increase in postoperative inpatient stays resulted from a small number of people who experienced significant complications. Our results are similar to those recently by Encinosa et al. [17], who found increasing rates of hospital admissions because of surgical complications for 6 months after the initial surgery. They did not measure the use rates before surgery but examined differences in complication rates and health services use from surgery to 3 months compared with 3– 6 months after surgery. They found that use because of surgical complications did increase over time. A recent report by Zingmond et al. [12] also found a significant increase in hospital admissions during the 3 years after bariatric surgery relative to the 3 years before bariatric surgery in a cohort of 60,777 California residents. In the first year after surgery, the most common reasons for hospital admission were related to complications associated with the procedure. The number of gastrointestinal or bypass surgery-related reasons for admission was similar to those of elective procedures in the second year. In the third year after bariatric surgery, the most common reasons for admission were elective procedures, including plastic surgery. However, these results are in contrast to those by Christou et al. [18], who found a decrease in hospitalization rates during the 5 years after bariatric surgery. The study used administrative claims data from the Regie de l’assurance maladie du Québec database in Quebec, Canada. In their cohort of 1035 bariatric surgery patients, significantly fewer hospitalizations occurred compared with the nonsurgically treated control group. It is possible, however, that any initial increase in the rates of hospitalization after bariatric surgery was offset by the significant decreases in hospitalization as time progressed. Additionally, their results showed significantly lower healthcare costs (presented in Canadian dollars) for those who underwent bariatric surgery. A similar study that used data from the same database found that the initially greater hospital costs remained for 3.5 years after surgery and then decreased substantially [11]. After the 3.5-year point, patients who had undergone bariatric surgery actually had significantly fewer hospital admissions and costs relative to the nonsurgically treated control group. The cost savings after that point were so substantial that when the 2 groups were compared, the overall costs after 5 years in the surgery group were less than that of the control group, even with the costs of surgery included. Although these results are impressive, 1 potential drawback exists. Because of the differences in payor systems between the United States and Canada, the economic results they found might not be applicable to patients and payors in the United States. The patients in our study were less likely to have sleep apnea, GERD, or hypertension 6 months after bariatric surgery. However, our study did not show a decrease in several other obesity-related co-morbidities, such as diabetes, dyslipi-
606
D. M. Hodo et al. / Surgery for Obesity and Related Diseases 4 (2008) 601– 607
demia, and depression, such as been demonstrated in other studies [8,19]. It is possible that these findings could have resulted from the time frame used in the present study, as well as physician documentation practices regarding chronic diseases, such as diabetes. This analysis covered a span of 6 months before surgery through 6 months after surgery, which might not have been long enough to demonstrate a decrease in these chronic conditions in the administrative data. Conversely, it is also possible that bariatric surgery did not result in as great of a co-morbidity reduction for this cohort as has been reported elsewhere [8,19]. Christou et al. [18] found a significant reduction in co-morbidities such as cancer, diabetes, and cardiovascular diseases during a 5-year follow-up period and also found a significantly reduced mortality rate among the patients who underwent bariatric surgery, even with the inclusion of perioperative deaths. Limitations The primary limitation of this study was the relatively short follow-up period. When the data collection took place, the MCO had only recently begun offering coverage for bariatric surgery. It is likely that the short follow-up did not allow for the detection within the claims data of decreases in the comorbid conditions that typically respond to weight loss. Although administrative claims data have proved to be a rich source of data for outcomes studies, they have several associated limitations and biases. Because claims data are generated through real-world patient encounters, the information is more generalizable relative to a randomized controlled trial. However, this increase in external validity potentially introduces a decrease in the internal validity through documentation or coding errors, as well as undercoding or overcoding. Additionally, it is possible that all services and diagnoses might not be recorded on an insurance claim form, because of the limitations in the numbers of procedures and diagnoses that can be recorded [20]. Claims data are also subject to issues introduced by plan design and changes that can occur over time [21]. In our study, we attempted to control for any changes that might have occurred through the use of line of business as a covariate in the ANCOVA and GEE models. Although this probably did not account for every change in pharmacy benefits that could have occurred, because our study results are similar to those of previously published reports, it is unlikely that plan changes accounted for all the reductions in medication use and costs. Important clinical data that were unavailable included preoperative and postoperative body mass index, laboratory values, and measures of patient satisfaction with the outcomes. The sample was drawn from an administrative claims database from a large, southeastern MCO; therefore, the findings might not be generalizable to populations outside of this region and insurance provider.
Conclusion Our research suggests that bariatric surgery can have a significant effect on medication use and costs, even within a relatively short follow-up period. This finding is important, given the evidence that obese individuals use more health services and experience greater costs relative to normal-weight individuals and that prescription medication use plays a substantial role in this increase [5]. The reduction in the use of medications after bariatric surgery could translate into cost savings for the MCO, given the ever-increasing epidemic of obesity. Although a true cost-effectiveness analysis was beyond the aims and scope of this study, our research suggests that the cost savings from decreased medication use after bariatric surgery can be realized within a relatively short period. In the short term, this might not offset the cost of surgery and potentially increased hospitalization resulting from surgery; however, the long-range costs of obesity itself in terms of the greater use of resources might be far more costly to the health system than the initial costs associated with bariatric surgery.
Disclosures The authors claim no commercial associations that might be a conflict of interest in relation to this article.
References [1] Ogden CL, Carroll MD, Curtin LR, McDowell MA, Tabak CJ, Flegal KM. Prevalence of overweight and obesity in the United States, 1999 –2004. JAMA 2006;295:1549 –55. [2] Stein CJ, Colditz GA. The epidemic of obesity. J Clin Endocrinol Metab 2004;89:2522–5. [3] National Task Force on the Prevention and Treatment of Obesity. Overweight, obesity, and health risk. Arch Intern Med 2000;160:898 –904. [4] Fontaine KR, Redden DT, Wang C, Westfall AO, Allison DB. Years of life lost due to obesity. JAMA 2003;289:187–93. [5] Raebel MA, Malone DC, Conner DA, Xu S, Porter JA, Lanty FA. Health services use and health care costs of obese and nonobese individuals. Arch Intern Med 2004;164:2135– 40. [6] Sturm R. The effects of obesity, smoking, and drinking on medical problems and costs. Health Aff (Millwood) 2002;21:245–53. [7] Buchwald H. Bariatric surgery for morbid obesity: health implications for patients, health professionals, and third-party payers. J Am Coll Surg 2005;200:593– 604. [8] Buchwald H, Avidor Y, Braunwald E, et al. Bariatric surgery: a systematic review and meta-analysis. JAMA 2004;292:1724 –37. [9] Gould JC, Garren MJ, Starling JR. Laparoscopic gastric bypass results in decreased prescription medication costs within 6 months. J Gastronitest Surg 2004;8:983–7. [10] Potteiger CE, Paragi PR, Inverso NA, et al. Bariatric surgery: shedding the monetary weight of prescription costs in the managed care arena. Obes Surg 2004;14:725–30. [11] Sampalis JS, Liberman M, Auger S, Christou NV. The impact of weight reduction surgery on health-care costs in morbidly obese patients. Obes Surg 2004;14:939 – 47.
D. M. Hodo et al. / Surgery for Obesity and Related Diseases 4 (2008) 601– 607 [12] Zingmond DS, McGory ML, Ko CY. Hospitalization before and after gastric bypass surgery. JAMA 2005;294:1918 –24. [13] Charlson ME, Pompei P, Ales KL, MacKenzie CR. A new method of classifying prognostic comorbidity in longitudinal studies: development and validation. J Chron Dis 1987;40:373– 83. [14] Deyo RA, Cherkin DC, Ciol MA. Adapting a clinical comorbidity index for use with ICD-9-CM administrative databases. J Clin Epidemiol 1992;45:613–19. [15] Quan H, Parsons GA, Ghali WA. Validity of information on comorbidity derived from ICD-9-CCM administrative data. Med Care 2002; 40:675– 85. [16] Roberts RE, Deleger S, Strawbridge WJ, Kaplan GA. Prospective association between obesity and depression: evidence from the Alameda County Study. Int J Obes 2003;27:514 –21.
607
[17] Encinosa WE, Bernard DM, Chen CC, Steiner CA. Healthcare utilization and outcomes after bariatric surgery. Med Care 2006;44:706 –12. [18] Christou NV, Sampalis JS, Liberman M, et al. Surgery decreases long-term mortality, morbidity, and health care use in morbidly obese patients. Ann Surg 2004;240:416 –24. [19] Sugerman HJ, Wolfe LG, Sica DA, Clore JN. Diabetes and hypertension in severe obesity and effects of gastric bypass-induced weight loss. Ann Surg 2003;237:751– 8. [20] Birnbaum HG, Cremieux PY, Greenberg PE, LeLorier J, Ostrander J, Venditti L. Using healthcare claims data for outcomes research and pharmacoeconomic analyses. Pharmacoeconomics 1999;16:1– 8. [21] Motheral BR, Fairman MA. The use of claims databases for outcomes research: rationale, challenges, and strategies. Clin Ther 1997;19:346 – 66.