OBES SURG (2014) 24:377–384 DOI 10.1007/s11695-013-1106-9

ORIGINAL CONTRIBUTIONS

Hiatal Hernia Repair in Laparoscopic Adjustable Gastric Banding and Laparoscopic Roux-En-Y Gastric Bypass: A National Database Analysis Benjamin J. S. al-Haddad & Robert B. Dorman & Nikolaus F. Rasmus & Yong Y. Kim & Sayeed Ikramuddin & Daniel B. Leslie

Published online: 4 December 2013 # Springer Science+Business Media New York 2013

Abstract Hiatal hernia (HH) repairs are commonly done concomitantly with laparoscopic Roux-en-Y gastric bypass (LRYGB) and laparoscopic adjustable gastric banding (LAGB) to decrease gastroesophageal reflux disease (GERD). There is limited evidence about the additional surgical risk these combined procedures engender. We used the United States Nationwide Inpatient Sample 2004–2009 to compare mortality risk, prolonged length of stay (PLOS), and perioperative adverse events using propensity score-matched analysis. We repeated the analysis after removing patients diagnosed with GERD. There were 42,272 weighted patients undergoing LRYGB alone representing 206,559 discharges nationally and an additional 1,945 and 9,060, respectively, undergoing LRYGB+HH repair. For LAGB, there were 10, 558 records representing 52,901 LAGB-only discharges and 1,959 representing 9,893 LAGB+ HH repair discharges. Thirty-eight percent (95 % CI: 36, 41 %) of the patients in the LRYGB-only group had GERD compared to 55 % (51, 59 %) in the LRYGB+HH repair group. Among the LAGB groups, 31 % (28, 34 %) of LAGB-only patients had GERD compared to 44 % (38, 49 %) in the LAGB+HH repair group. We find that the average treatment effect on the treated (considering the concomitant procedure as treatment and the single procedure as control) for PLOS was −0.12353

(−0.15909, −0.08797) between the LRYGB groups and −0.04353 (−0.07488, −0.01217) for the LAGB groups. We find no evidence of increased risk of perioperative adverse events among patients undergoing concomitant HH repair with LRYGB or LAGB. Patients undergoing the combined procedure appear to be at lower risk of PLOS; this may be due to surgical training norms. Keywords Hiatal hernia repair . Laparoscopic adjustable gastric band . Laparoscopic Roux-en-Y . Gastric bypass Abbreviations GERD LRYGB LAGB HH PLOS NIS ATT ICD9

Gastroesophageal reflux disease Laparoscopic Roux-en-Y gastric bypass Laparoscopic adjustable gastric banding Hiatal hernia Prolonged length of stay Nationwide Inpatient Sample Average treatment effect on the treated International Classification of Diseases version 9

Introduction B. J. S. al-Haddad : R. B. Dorman : N. F. Rasmus : Y. Y. Kim : S. Ikramuddin : D. B. Leslie (*) Department of Surgery, 420 Delaware St. SE, MMC 290, Minneapolis, MN 55455, USA e-mail: [email protected] B. J. S. al-Haddad Division of Epidemiology & Community Health, B681 Mayo Building, University of Minnesota, 420 Delaware St. SE, MMC 293, Minneapolis, MN 55455, USA

Although varying definitions of gastroesophageal reflux disease (GERD) make estimating prevalence challenging, evidence suggests that 19.8 to 42 % of the USA population may suffer from the disease [1, 2]. Obesity and hiatal hernia are both thought to be causally associated with GERD, and there is evidence that both the extent of obesity and the size of hiatal hernia exacerbate reflux symptoms [3, 4]. Relatedly, both laparoscopic adjustable gastric banding (LAGB) and

378

Roux-en-Y gastric bypass procedures may change GERD symptoms [5, 6]. Patients suffering from GERD who undergo a bariatric procedure may also undergo concomitant hiatal hernia repair [7, 8]. There are several case series which have generated some evidence about the additional perioperative risk for adverse events that this combined procedure engenders [9–14]. Despite this literature, it is still unknown whether the concomitant procedure increases risk of adverse events. We used a nationally representative database of discharge records to discover whether patients undergoing concomitant HH repair with either laparoscopic Roux-en-Y gastric bypass (LRYGB) or LAGB were at increased risk of mortality, prolonged length of stay (PLOS), or perioperative adverse events.

Materials and Methods Data Data was extracted from the United States Nationwide Inpatient Sample (NIS) 2004 to 2009 [15]. The NIS is a nationally representative database available from 1988 to 2010 which contains individual-level data on inpatient demographics, diagnoses, procedures, and comorbidities from hospital discharge records. The hospitals were sampled from all, “…non-Federal, short-term, general, and other specialty hospitals, excluding hospital units of institutions [15].” Veterans Affairs hospitals and short-term rehabilitation hospitals were not included. To generate the database, the hospitals were stratified on geographic region, control (public or private), location (urban or rural), teaching status, bed size, and three-digit zip code, and then a systematic random 20 % sample was taken. Because a 20 % random sample was taken, each discharge was weighted to count for around five discharges. The discharges were further weighted if the hospitals were closed for parts of the year. Variables Patients in the database diagnosed with obesity (defined using International Classification of Diseases 9 (ICD9) codes 278.00, 278.01, 278.8, and 278.1), who underwent LRYGB (44.38) or LAGB (44.95) on the day of admission and aged 19 to 85 were included in the study sample. The patients undergoing hiatal hernia repairs (HH) were defined using ICD9 codes 53.71, 53.70, 53.75, 53.69, and 53.7. It was assumed that patients coded as undergoing either bariatric procedure and HH repair on different days were miscoded and actually had both a bariatric procedure and HH repair on the same day. ICD9 codes 530.81 and 530.85 were also used to determine which patients had GERD. PLOS was defined

OBES SURG (2014) 24:377–384

as a hospital stay at or longer than the 70th percentile of stays for each procedure. Race was coded as a binary variable, where white was the reference, and other races were recoded as non-white. Hospitals were divided into three groups (100, with the last being the reference group), depending on the number of LRYGB or LAGB discharges per year. Variables coding for the presence of specific surgical complications were created for the diagnoses and ICD9 codes in Appendix. Experiencing any one or more of the events in Appendix was considered as having an adverse event. Lastly, mortality was examined separately. Analysis The data were analyzed in Stata version 12 using the appropriate discharge weights, strata, and population sampling units using the svy suite of commands [16]. Baseline age, sex, median household income for patient zip code, race, number of chronic conditions, and specific comorbidities were examined for differences between groups using 95 % confidence intervals. The main analysis focused on two comparisons. First, we compared the immediate postsurgical outcomes among patients who underwent LRYGB alone to those who underwent LRYGB+HH repair (Fig. 1). Second, we separately compared outcomes among patients who underwent LAGB alone to those who underwent LAGB+ HH repair. A propensity score-matched approach was used for analysis with the psmatch2 Stata module. This methodology attempts to recapitulate important aspects of the randomized controlled trial by matching “treated” patients (those who underwent the combined procedure) to “untreated patients” (those who only underwent LRYGB or LAGB) by their propensity to have been treated. This method required two steps. First, logistic regression was used to predict who would undergo the combined procedure based on age, sex, diabetes status, depression, hypertension, GERD, congestive heart failure, liver disease, chronic pulmonary disease, zip code income quartile, and race. The estimated regression equation was used to calculate each patient's propensity score for undergoing the combined procedure. Second, treated and untreated patients were matched for the same propensity score to undergo treatment (concomitant HH repair). This strategy attempted to mimic randomization by increasing treated and untreated patients' exchangeability. Postmatched groups were examined for balance on these covariates using standardized percentage bias estimates [17]. After matching, the differences in risk of complications and adverse events were examined by subtracting the risk in the untreated group from the risk in the treated group to produce a risk difference measure called the average treatment effect on the treated (ATT) [18]. The ATT

OBES SURG (2014) 24:377–384

379

Fig. 1 LRYGB vs. LRYGB+HH repair. The average treatment effect on the treated (ATT) and 95 % confidence intervals comparing patients who underwent LRYGB (untreated) with propensity score-matched patients

who underwent LRYGB + HH (treated). Negative ATT suggests protective effect of concomitant procedure. If neither group experienced a category of event, it was not listed here

represents the average benefit (or detriment) from treatment for those who were actually treated and can be thought of as a kind of risk difference [19]. The analysis was carried out separately for LAGB and LRYGB, and a sub-analysis was carried out for those patients not diagnosed with GERD in both groups.

Results

Table 1 Presurgical patient characteristics (LAGB)

This n takes the sampling method into account, where each unweighted observation counts for approximately five observations

a

The proportion of patients undergoing the combined procedure differed between groups. While 15.8 % of LAGB procedures had a concomitant HH repair, only 4.2 % of LRYGB included a HH repair (see Tables 1 and 2).

LAGB

Age (years) Female White Income quartile of zip code Weighted n a Unweighted n

LAGB+HH

Mean/proportion

95 % CI

45.33 0.77 0.67 2.75 52,901 10,558

44.52 0.76 0.63 2.65 43,946 –

46.13 0.79 0.71 2.86 61,856 –

Mean/proportion

95 % CI

49.17 0.82 0.70 2.86 9,893 1,959

47.84 0.78 0.62 2.72 6,810 –

50.50 0.86 0.79 3.00 12,976 –

380

OBES SURG (2014) 24:377–384

Table 2 Presurgical patient characteristics (LRYGB) LRYGB

LRYGB+HH

Mean/ 95 % CI proportion

Mean/ 95 % CI proportion

Age (years)

43.56

43.23

43.88

47.42

46.09 48.75

Female

0.80

0.79

0.81

0.88

0.85

White

0.64

0.60

0.68

0.60

0.51

0.68

2.54

2.68

2.52

2.40

2.64

0.90

Income quartile 2.61 of zip code 206,559 Weighted n a

176,286 236,832 9,060

5,921 12,198

Unweighted n

–

–

42,272

–

1,945

–

This n takes the sampling method into account, where each unweighted observation counts for approximately five observations

a

Considering pre-matched patient baseline characteristics (see Tables 1 and 2), patients in both the LAGB+HH and the LRYGB+HH groups were older than the patients in the LAGB and LRYGB groups, respectively. The LAGB+HH patients had an average age of 49.2 (95 % CI: 47.8, 50.5), while the LAGB patients had an average age of 45.3 years (44.5, 46.1). The average age of the LRYGB+HH patients was 47.4 years (46.1, 48.8), while the average age of the LRYGB patients was 43.6 (43.2, 43.9). The LRYGB+HH consisted of 88 % females (85, 90 %), while the LRYGB group was 80 % females (79, 81 %); the LAGB and LAGB+HH did not differ on sex. The groups did not differ with respect to proportion of white or income quartile of zip code. With respect to presurgical patient comorbidities (see Tables 3 and 4), the patients in the LAGB+HH group had a GERD prevalence of 44 % (38, 49 %) compared to 31 % (28, 34 %) in the LAGB only group. Similarly, the LRYGB+HH group had a GERD prevalence of

Table 3 Presurgical patient comorbidities (LAGB)

This n takes the sampling method into account, where each unweighted observation counts for approximately five observations

a

55 % (51, 59 %) compared to the LRYGB only of 38 % (36, 41 %). Further, 25 % (22, 27 %) of the patients in the LRYGB+HH had diabetes mellitus type II compared to 31 % (30, 32 %) in the LRYGB only group. Apart from these differences, patients did not differ on prevalences of hypertension, depression, congestive heart failure, type II diabetes mellitus with complications, hepatic or pulmonary disease. For the propensity score analysis, matching on age, sex, diabetes status, depression, hypertension, GERD, congestive heart failure, liver disease, chronic pulmonary disease, zip code income quartile, and race decreased standardized percentage bias by 63 and 73 %, and resulted in relatively balanced matching in both the LAGB and LAGB+HH, and the LRYGB and LRYGB+HH groups, respectively. Thirty percent of the patients had hospital stays of 3 days or longer in both groups; accordingly, prolonged length of stay was defined as 3 days or greater. The ATTs and 95 % confidence intervals comparing the LAGB (non-treated) to the LAGB+HH (treated) groups is illustrated in Fig. 2. Notably, the ATT was not different than 0 for total adverse events; however, the ATT was −0.04 (−0.075, −0.012) for PLOS. The patients in the LAGB+HH group had an ATT of 0.003 (0.00, 0.005) for postoperative hypotension. The ATTs were not different than 0 for the other adverse events including mortality. The ATTs comparing the LRYGB (non-treated) to the LRYGB+HH (treated) groups are illustrated in Fig. 1. The ATT of HH repair was −0.124 (−0.16, −0.088) for PLOS. Additionally, the ATT for postoperative fever was −0.00529 (−0.0097, −0.00089). The ATT for all of the other adverse events including mortality did not differ from 0. The sub-analysis for the LRYGB+HH patients without GERD found an ATT for the PLOS of

LAGB

Type II diabetes mellitus Hypertension GERD Depression Congestive heart failure Hepatic disease Pulmonary disease Weighted n a Unweighted n

LAGB+HH

Proportion

95 % CI

0.28 0.52 0.31 0.18 0.02 0.06 0.17 52,901 10,558

0.26 0.50 0.28 0.16 0.01 0.04 0.15 43,946 –

0.30 0.55 0.34 0.20 0.02 0.08 0.18 61,856 –

Proportion

95 % CI

0.25 0.54 0.44 0.19 0.01 0.05 0.17 9,893 1,959

0.23 0.50 0.38 0.16 0.00 0.03 0.15 6,810 –

0.28 0.58 0.49 0.23 0.01 0.06 0.20 12,976 –

OBES SURG (2014) 24:377–384

381

Table 4 Presurgical patient comorbidities (LRYGB)

LRYGB Proportion

This n takes the sampling method into account, where each unweighted observation counts for approximately five observations

a

LRYGB+HH 95 % CI

Proportion

95 % CI

Type II diabetes mellitus

0.31

0.30

0.32

0.25

0.22

0.27

Hypertension GERD Depression Congestive heart failure Hepatic disease Pulmonary disease Weighted n a Unweighted n

0.55 0.38 0.22 0.01 0.09 0.18 206,559 42,272

0.54 0.36 0.20 0.01 0.06 0.17 176,286 –

0.57 0.41 0.24 0.01 0.11 0.19 236,832 –

0.58 0.55 0.22 0.01 0.07 0.18 9,060 1,945

0.55 0.51 0.18 0.01 0.05 0.15 5,921 –

0.61 0.59 0.25 0.02 0.09 0.21 12,198 –

−0.10724 (−0.15929, −0.05518), and the ATT for reduced risk of drainage was −0.00646 (−0.01258, −0.00034). The ATT for all of the other adverse events did not differ from 0. Among the LAGB+HH patients, the ATT for PLOS was not different than 0, but the ATT for injury to adjacent structure was −0.01361 (−0.02598, −0.00123). None of the ATTs for the other adverse outcomes differed from 0.

Conclusion

Fig. 2 LAGB vs. LAGB+HH repair. The average treatment effect on the treated (ATT) and 95 % confidence intervals comparing patients who underwent laparoscopic adjustable gastric banding (untreated) with propensity score-matched patients who underwent

laparoscopic adjustable gastric banding and hiatal hernia repair (treated). Negative ATT suggests protective effect of concomitant procedure. If neither group experienced a category of event, it was not listed here

The major objective of this study was to determine whether or not the concomitant repair of HH with LAGB or LRYGB resulted in increased morbidity and mortality in the perioperative period. Considering all the defined adverse events together, no evidence was found that patients undergoing both procedures were at increased risk.

382

Concomitant bariatric surgery with HH was found to be four times more common among patients undergoing LAGB than LRYGB. This substantial difference in rate of HH repair likely has several etiologies, including a difference in surgeon opinion about the effectiveness of each bariatric procedure for GERD resolution and a difference in training pathways for performance of LAGB and LRYGB. Some surgeons involved in training courses for LAP-BAND ® and RealizeBand ® were taught by proctors to repair HH by approximating the crura anteriorly or posteriorly [20, 21]. This training was given due to an emphasis on a higher prevalence of GERD in the absence of a repair [10]. Patients in both the LAGB+HH and the LRYGB+HH groups were older than their counterparts in the LAGB and LRYGB groups, respectively. This supports earlier evidence that older age may be a risk factor for hiatal hernia and subsequent repair [22]. Among the LRYGB patients, those undergoing the combined procedure were more likely to be women; this also appears to be true in the LAGB group. This difference may reflect true differences in HH incidence or it may attest to the greater difficulty in viewing and manipulating the anatomy of the gastroesophageal junction in males. As expected from previous studies, the LAGB+HH and LRYGB+HH groups both had higher prevalences of GERD compared to the LAGB and LRYGB groups [3, 4, 23–25]. Our major finding is the apparent absence of increased risk from the concomitant procedures in the LAGB+HH and the LRYGB+HH groups. Although an increased risk of postoperative hypotension was found in the LAGB+HH group, this difference of 0.002687 (0.000045, 0.005329) is not likely clinically significant and may be due to type I error. Similarly, the LRYGB+HH had reduced postoperative fever compared to the LRYGB only group, this difference of −0.00529 (−0.0097, −0.00089) is not likely to be clinically significant and may also be due to type I error. These results did not change importantly after removing patients diagnosed with GERD. This encouraging result suggests that there is no additional risk to patients when LRYGB or LAGB procedure is combined with HH repair. The reason for lack of risk related to HH repair cannot be discerned from these data. ICD9 codes related to HH repair may reflect a broad spectrum of tissue dissection. The repair least likely to induce tissue damage involves a single suture placed anterior to the esophagus without disrupting the phrenoesophageal membrane. A more extensive dissection for a similar-sized HH could involve a circumferential dissection and suture approximation of the crura. The intraoperative decision making algorithm and type of repair cannot be gleaned from ICD9 coding data. There are several case series in the literature addressing various types of the concomitant procedure. Frezza et al. published evidence from a case series of 21 patients

OBES SURG (2014) 24:377–384

undergoing crural repair with LAGB with two postsurgical complications (wound infection and dysphagia) in the first 6 months postsurgery [9]. In a larger series, Gulkarov et al. reported lower reoperation rates for the combined procedure than for the LAGB alone, but did not report perioperative complications separately [10]. In a case series of three patients undergoing paraesophageal hernia repair with LRYGB, Kasotakis et al. reported no intra-or 30-day morbidity [11]. Salvador-Sanchis et al. report one LRYGB and giant type III hiatal hernia repair in which there are no reported intraoperative or perioperative complications [12]. Angrisani et al. published a case series examining three patients who underwent LAGB with HH repair compared to nine who only had the LAGB procedure, and only had one complication in each group in the 7 months after surgery (postoperative proximal gastric pouch dilation in a LAGB and band dislocation in a LAGB+HH) [13]. Dolan et al. only found three complications among 62 cases of LAGB+HH repair over a median follow-up of 14 months; these were pulmonary embolus, slippage requiring repositioning of the band, and persistent dysphagia requiring band removal [14]. Because of the low numbers of concomitant repairs in these studies, it is plausible that the studies with control groups may not have had adequate power to detect any effect of the joint procedure. It is also likely that patient selection for one procedure or another may bias these and our study. We could not use this NIS to study patients undergoing the sleeve gastrectomy because of the absence of an ICD9 code specifically for this procedure during the time period under study. However, we look forward to future work addressing this question in patients undergoing this procedure. An unexpected and unexplained finding was the lower risk of PLOS among patients undergoing the combined procedures. There are numerous possible explanations for this finding. Perhaps the simplest is that surgeons who were more likely to do concomitant HH repair may also discharge their patients earlier than surgeons who do not routinely perform and advocate the combined procedure. There are several limitations to this work. The most important is the imperfect counterfactual or control group. Patients undergoing LRYGB and HH repair were compared to patients only undergoing LRYGB. Since the patients only undergoing LRYGB may or may not suffer from a hiatal hernia, they are a different patient population from the LRYGB+HH in not one, but in two ways; namely, that they do not have a HH repair, but they also may not have a HH. This results in the confounding of the effect of the HH repair with the effect of the presence of the hernia itself. The same is true for the LAGB analysis. The ideal comparison would be between patients with HH who undergo LRYGB without repair to those who undergo the LRYGB+HH repair. However, such an idealized study would likely face insurmountable ethical, financial, and power challenges. As

OBES SURG (2014) 24:377–384

383

such, despite this important limitation, the present study offers valuable insight into the question at hand. Another limitation is the time period of data collection. The NIS does not follow patients after discharge, so post-discharge outcomes cannot be examined. Relatedly, it was unclear how the diagnosis of HH was made. A preoperative diagnosis of HH has influenced many surgeons to advise against LAGB surgery for weight loss, and this may also affect the preoperative planning in the extent of HH repair, depending on the practice pattern of the surgeon. Despite these limitations, the current study benefits from a nationally representative data set and is well powered to answer the central questions of interest. This study produces nationally representative evidence, suggesting that concomitant HH repair with LAGB or LRYGB may not increase risk of morbidity and mortality.

Table 5 (continued)

Acknowledgments We would like to thank Mr. Mathew Grabau, Ms. Nikki Voulgaropoulos, Mrs. Bridget Slusarek, Mrs. Elsie Waddick, Mrs. Barbara Sampson, Dr. Nathan Liu, and Dr. Aboude Nowaylati for their helpful contributions. B.J.S. al-Haddad and Y. Kim were supported by NIH MSTP grant T32 GM008244 and the Department of Surgery Research Funds. R.B. Dorman, N.F. Rasmus, S. Ikramuddin, and D.B. Leslie were supported by the Department of Surgery Research Funds.

Postoperative atelectasis/pneumonia Postoperative vomiting Diarrhea following gastrointestinal surgery Postoperative small bowel obstruction/ileus or complication of anastomosis of gastrointestinal tract Postoperative hypotension Postoperative stroke Cardiac arrest/insufficiency during or resulting from a procedure Phlebitis or thrombophlebitis from procedure

Conflict of interest B.J.S. al-Haddad has no conflict of interest. R.B. Dorman has no conflict of interest. N.F. Rasmus has no conflict of interest. Y. Kim has no conflict of interest. S. Ikramuddin has no conflict of interest. D.B. Leslie has no conflict of interest.

Appendix

Table 5 Variables coding for the presence of specific surgical complications were created for the diagnoses and ICD9 codes Complication

ICD9 code(s)

Injury to adjacent structures, accidental puncture, or laceration complicating surgery Retained foreign body Hemorrhage complicating procedure Esophageal injury Splenic injury

998.2

Postoperative hematoma Postoperative seroma (non-infected)

998.4 998.11 530.4 865.00, 865.01, 865.02, 865.03, 865.04, 865.09, 865.10, 865.11, 865.12, 865.13, 865.14, 865.19 998.12 998.13

Disruption of operative wound Disruption of wound unspecified Disruption of internal operation (surgical) wound Disruption of external operation (surgical) wound

998.3 998.3 998.31 998.32

Complication

ICD9 code(s)

Persistent postoperative fistula Delayed wound healing Postoperative infection

998.6 998.83 998.5

Postoperative infected seroma Postoperative abscess/infection Postoperative urinary retention, urinary tract infection and or renal failure Pneumothorax

998.51 998.59 997.5

Hemothorax Pleural effusion Postoperative pulmonary edema Adult respiratory distress syndrome following surgery Transfusion-related acute lung injury

Acute post-hemorrhagic anemia Postoperative stroke Postoperative shock Postoperative fever Unspecified complication of procedure, not elsewhere classified Blood transfusion Placement of chest tube Exploratory laparotomy Reopening of recent laparotomy site for control of hemorrhage, exploration, incision of hematoma Drainage of intraperitoneal abscess or hematoma Reclosure of postoperative disruption of abdominal wall Percutaneous drainage of abdomen Removal of foreign body from peritoneal cavity Endogastroduodenoscopy Pulmonary embolism Deep vein thrombosis Traumatic pneumothorax and hemothorax

Injury to heart and lung Injury to other and unspecified organs

512.82, 512.84, 512.89, 512.1 511.89 511.9 518.4 518.5 518.7 997.3 564.3 564.4 997.4

458.29 997.02 997.1 997.2 285.1 997.02 998 998.89 998.9 99.04 34.04 54.11 54.12 54.19 54.61 54.91 54.92 45.13 415.1 453.4 860, 860.0, 860.1, 860.2, 860.3, 860.4, 860.5 861, 861.0, 861.1, 861.2, 861.3 862, 862.0, 862.1, 862.2, 862.3, 862.8, 862.9

384

OBES SURG (2014) 24:377–384

Table 5 (continued) Complication

ICD9 code(s)

Injury to gastrointestinal tract

863, 863.0, 863.1, 863.2, 863.4, 863.5, 863.8, 863.9 864, 864.0, 864.1 866, 866.0, 866.1 867, 867.0, 867.1, 867.2, 867.3, 867.4, 867.5, 867.6, 867.7, 867.8, 867.9 868, 868.0, 868.1 869

Injury to liver Injury to kidney Injury to pelvic organs

Injury to other intra-abdominal organs Internal injury to unspecified or ill-defined organs

References 1. Delaney BC. Review article: prevalence and epidemiology of gastrooesophageal reflux disease. Aliment Pharmacol Ther. 2004;20:2–4. PMID:15575863. 2. Locke III GR, Talley NJ, Fett SL, et al. Prevalence and clinical spectrum of gastroesophageal reflux: a population-based study in Olmsted County, Minnesota. Gastroenterology. 1997;112(5):1448– 56. PMID:9136821. 3. Jones MP, Sloan SS, Rabine JC, et al. Hiatal hernia size is the dominant determinant of esophagitis presence and severity in gastroesophageal reflux disease. Am J Gastroenterol. 2001;96(6): 1711–7. PMID:11419819. 4. Smith S, Edwards C, Goodman G. Symptomatic and clinical improvement in morbidly obese patients with gastroesophageal reflux disease following Roux-en-Y gastric bypass. Obes Surg. 1997;7:479–84. PMID:9730504. 5. Friedenberg FK, Xanthopoulos M, Foster GD, et al. The association between gastroesophageal reflux disease and obesity. Am J Gastroenterol. 2008;103(8):2111–22. PMID:18796104. 6. Tolonen P, Victorzon M, Niemi R, et al. Does gastric banding for morbid obesity reduce or increase gastroesophageal reflux? Obes Surg. 2006;16(11):1469–74. LA – English. PMID:17132413. 7. Hallerbäck BJ. Laparoscopy in the gastroesophageal junction. Int Surg. 1995;80(4):307–10. PMID:8740674. 8. Näslund E, Granström L, Melcher A, et al. Gastro-oesophageal reflux before and after vertical banded gastroplasty in the

treatment of obesity. Eur J Surg. 1996;162(4):303–6. PMID: 8739417. 9. Frezza EE, Barton A, Wachtel MS. Crural repair permits morbidly obese patients with not large hiatal hernia to choose laparoscopic adjustable banding as a bariatric surgical treatment. Obes Surg. 2008;18(5):583–8. PMID:18317857. 10. Gulkarov I, Wetterau M, Ren CJ, et al. Hiatal hernia repair at the initial laparoscopic adjustable gastric band operation reduces the need for reoperation. Surg Endosc. 2008;22(4):1035–41. PMID:18080712. 11. Kasotakis G, Mittal SK, Sudan R. Combined treatment of symptomatic massive paraesophageal hernia in the morbidly obese. JSLS: J Soc Laparoendosc Surg/Soc Laparoendosc Surg. 2011;15(2):188–92. PMID:21902973. 12. Salvador-Sanchis JL, Martinez-Ramos D, Herfarth A, et al. Treatment of morbid obesity and hiatal paraesophageal hernia by laparoscopic Roux-en-Y gastric bypass. Obes Surg. 2010;20(6):801– 3. PMID:18773250. 13. Angrisani L, Iovino P, Lorenzo M, et al. Treatment of morbid obesity and gastroesophageal reflux with hiatal hernia by Lap-Band. Obes Surg. 1999;9(4):396–8. PMID:10484300. 14. Dolan K, Finch R, Fielding G. Laparoscopic gastric banding and crural repair in the obese patient with a hiatal hernia. Obes Surg. 2003;13(5):772–5. PMID:14627475. 15. Healthcare Cost and Utilization Project (HCUP). HCUP Nationwide Inpatient Sample (NIS). 16. StataCorp. Stata Statistical Software: release 12. Release 12. Software SS, editor. 2011. 17. Rosenbaum PR, Rubin DB. Constructing a control group using multivariate matched sampling methods that incorporate the propensity score. Amer Stat. 1985;39(1):33–8. 18. Guo SY, Fraser MW. Propensity score analysis. 2010. 19. Morgan SL, Winship C. Counterfactuals and causal inference: methods and principles for social research. 2007 20. Gordon C, Kang JY, Neild PJ, et al. The role of the hiatus hernia in gastro-oesophageal reflux disease. Aliment Pharmacol Ther. 2004;20(7):719–32. PMID:15379832. 21. Ethicon Endo Surgery. Realize. [Accessed 2013 Sep 19]. Available from: http://www.realize.com/ 22. Allergan Inc. LapBand. 2013 [Accessed 2013 Sep 19]. Available from: http://www.lapband.com/ 23. Wilson LJ, Ma W, Hirschowitz BI. Association of obesity with hiatal hernia and esophagitis. Am J Gastroenterol. 1999;94(10):2840–4. doi:10.1111/j.1572-0241.1999.01426.x. 24. El-Serag HB, Graham DY, Satia JA, et al. Obesity is an independent risk factor for GERD symptoms and erosive esophagitis. Am J Gastroenterol. 2005;100(6):1243–50. PMID:15929752. 25. Khouzam RN, Akhtar A, Minderman D, et al. Echocardiographic aspects of hiatal hernia: a review. J Clin Ultrasound. 2007;35(4):196– 203. PMID:17354243.