Detailed analysis

A total of six publications reported results on four RCTs comparing medical with surgical treatments.^{8, 20}–²⁴ One RCT was excluded because the surgical techniques did not meet our inclusion criteria of reviewing only Nissen or Toupet methods.²⁰ Lundell et al. reported 3-year²¹ and 5-year²² follow-up data. Spechler et al. reported 2-year²⁴ and 10-year⁸ follow-up data (N.B., For Lundell and Spechler, unless otherwise noted, the longer follow-up data are presented here). Mahon et al. reported data for 1-year follow-up.²³ The three RCTs enrolled a total of 762 patients, 666 of whom had follow-up information. The studies differed in the severity of esophagitis in patients at baseline and in the procedure performed. Spechler²⁴ and Mahon²³ included patients with erosive esophagitis. Lundell included patients with no higher than grade one esophagitis at baseline.²¹ Lundell²¹ and Spechler²⁴ used open fundoplication while Mahon²³ used laparoscopic fundoplication. All three studies were graded as methodological quality B.

The three studies included patients with different medical treatments and different response to medical treatments at baseline. None of the trials enrolled patients whose symptoms were poorly controlled with medical therapy. Spechler included patients whose reflux disorder responded to non-PPI based medical treatments at enrollment.²⁴ Lundell included patients whose symptoms and esophagitis responded to omeprazole.²¹ Mahon²³ reported that the enrolled patients were “dependent” on PPIs for at least 3 months (although the definition of “dependent” was not stated). Lundell²¹ and Mahon²³ used regular intake of a PPI for the medical intervention; Spechler²⁴ used a combination of H2RAs, antacids and pro-motility agents either on a regular or as needed basis for the medical intervention (However, most of the patients used PPIs during the follow-up period, but in a nonstandardized fashion.⁸). In all three RCTs, more than 80% of the patients who actually received the interventions provided follow-up data.

A total of seven non-randomized studies compared medical with surgical treatments in patients with GERD.²⁵–³² The average follow-up duration ranged from 6 months to more than 10 years. All were graded as methodological quality C.

Four provided data on baseline response to medical therapy.^{25, 27, 28, 31}A prospective study compared patients whose GERD symptoms or esophagitis responded to PPIs and were maintained on them with a group that underwent laparoscopic fundoplication because they either had recurrent symptoms of GERD or esophagitis, did not respond to PPIs, or were unwilling to continue PPIs.³¹ A retrospective study compared all patients with GERD who were “managed nonoperatively” and were not referred for operation with all patients who underwent laparoscopic fundoplication over a 1-year period.²⁵ Another retrospective study compared patients whose GERD symptoms and esophagitis responded to H2RAs and lifestyle modification and were kept on medical management with patients whose GERD symptoms did not respond to the same therapy and therefore, underwent open fundoplication.²⁷ One open label study compared ranitidine with open fundoplication in patients with symptomatic GERD. ²⁸ Patients were selected for fundoplication or ranitidine based on the patient's and surgeon's preference. Three of the studies included patients with severe esophagitis.^{27, 28, 31}

A 1-year²⁶ and 4-year²⁹ retrospective analysis of administrative data and computerized health care records from the Tennessee Medicaid program compared the usage of prescription antireflux medications in patients with GERD treated medically with those treated surgically.

Two retrospective cohort analyses specifically examined the incidence of esophageal adenocarcinoma in medically versus surgically treated patients.^{30, 32}

The findings from both the RCTs and non-randomized comparisons have been organized by the following outcomes of interest: change in symptoms, quality-of-life (QOL), and patient satisfaction; change in esophagitis status; change in pH study results; change in lower esophageal sphincter (LES) pressure; change in medication usage status; change and follow-up information regarding Barrett's esophagus, and the incidence of esophageal adenocarcinoma. Details of these outcomes are presented in the Evidence Tables while the key points are summarized here. Adverse effects are presented under key question 3.

Change in symptoms, quality-of-life and patient satisfaction (Table 1)

Three RCTs and three non-randomized studies reported outcomes in symptoms, quality-of-life, and patient satisfaction.^{8, 22, 23, 25, 27, 28} The various studies used different methods of measuring symptom improvement and patient satisfaction. They ranged from patients' descriptions of heartburn, regurgitation, and satisfaction to structured scales like GERD-Health Related Quality of Life (GERD-HRQL), Visick Scale, and SF-36. The Spechler RCT reported significantly better Gastroesophageal Reflux Disease Activity Index (GRACI) score in the surgical (open fundoplication) arm than in the medical group after both groups discontinued all antireflux medications during the week of assessment; the medically treated patients went from having mild reflux symptoms to having mild to moderate symptoms.⁸ The difference between groups was not significant when both groups were on their usual antireflux medications. The Lundell RCT did not demonstrate a difference between omeprazole and open fundoplication in the prevalence of patients with moderate to severe heartburn at defined time points during the follow-up.²² The Mahon RCT reported that Gastrointestinal and General Well-Being Score improved more in the laparoscopic group compared with the PPI group at 1 year follow-up.²³ The Gastrointestinal well-being score went from a baseline of 31.7 to 37 in the surgical group and from 34.3 to 35.0 in the medical group (between group P<0.001). The clinical implication of the two-point difference between groups was not described.

Table 1

Medical vs. surgical treatments of GERD: Change in symptoms, QOL and satisfaction.

Lundell and Spechler also examined quality-of-life outcomes. Lundell used the Psychological General Well-Being Index (PGWB) and Gastrointestinal Symptom Rating Scale (GSRS); Spechler used Medical Outcome Short Form Health Survey (SF-36). There were no differences in quality-of-life assessment.

The one retrospective study comparing laparoscopic surgery with medical treatment reported better quality-of-life score in the surgical group.²⁵ The one retrospective study comparing open Nissen with medical treatment with more than 10 years follow-up reported more symptom improvement in the surgical than the medical group.²⁷

Change in esophagitis status (Table 2)

Two RCTs, one non-randomized comparison, and one retrospective cohort study reported on changes in esophagitis status.^{8, 21, 24, 27, 28} Spechler RCT, with a 10-year follow-up, reported no difference in the endoscopic grade of esophagitis between open fundoplication and medical treatment,⁸ although there were lower grades of esophagitis in the surgical group at 2-year follow-up.²⁴ Lundell RCT reported an increase in percentage of esophagitis in both open fundoplication and medically treated groups at 3-year follow-up.²¹ Isolauri et al. - the retrospective study reporting on change in esophagitis status - reported improvement in both open fundoplication and medically treated groups, although there was more improvement in the open fundoplication than in the medical group. At 10-year follow-up, 86% of the open fundoplication group had grade 0 compared with 46% of the medical group. Similar differences were found for the other grades.²⁷ Johansson and Tibbling - the open label comparison of ranitidine with open fundoplication, where 90% of the patients in both groups had esophagitis at baseline - reported that there was a significant improvement in the ranitidine group but not in the open fundoplication group after 8 weeks of ranitidine. Six months later, there was no further improvement in the ranitidine maintenance group, however, all the patients in the open fundoplication group had a normal endoscopic mucosa.²⁸

Table 2

Medical vs. surgical treatments of GERD: Change in esophagitis status.

Change in pH study results (Table 3)

Three RCTs and one non-randomized comparison reported changes in pH study results.^{8, 21, 24, 28}. Spechler RCT, in the 10-year follow-up, reported a non-significant lower percent-time with pH<4 in the open fundoplication group compared with the medical group, although only 10 surgical patients were evaluated (versus 38 in the medical arm).⁸ The same RCT, in the 1-year follow-up, reported that the percent-time pH<4 improved more in open fundoplication group compared to the group on symptomatic medical therapy (P<0.03).²⁴ Lundell RCT, in the 1-year follow-up, reported lower percent-time with pH<4 in both open fundoplication and medical groups compared to their baseline values.²¹ Mahon RCT, in the 3-month follow-up of laparoscopic surgery versus PPI study, reported lower percent-time with pH<4 in both groups and there was greater improvement in surgical than medical group.²³

Table 3

Medical vs. surgical treatments of GERD: Change in pH study results.

Change in lower esophageal sphincter pressure (Table 4)

Two RCTs and two non-randomized studies provided information on lower esophageal sphincter (LES) pressure.^{23, 24, 28, 31} Spechler RCT reported a significant increase in LES pressure in the open fundoplication group compared with the medical group 1 year after surgery (P value, not stated).²⁴ Mahon RCT reported an increase in LES pressure in the laparoscopic surgery group compared with the PPI group 3 months after surgery (P<0.001 between groups).²³ One retrospective study reported that patients who developed Barrett's esophagus at follow-up had more defective LES pressure and more impaired esophageal peristalsis before treatment (P<0.05).³¹ Johansson and Tibbling - the open label comparison of ranitidine with open fundoplication - reported that the LES pressure had a significant increase in the open fundoplication group compared to baseline (P<0.05).²⁸

Table 4

Medical vs. surgical treatments of GERD: Change in LES pressure.

Change in medication usage status (Table 5)

One RCT⁸ and two retrospective analyses^{26, 27} (one²⁶ with a subsequent follow-up publication²⁹) reported on changes in medication usage. Spechler RCT,⁸ with the 10-year follow-up, reported that proportionately fewer patients were on PPIs after open fundoplication compared with the medically treated patients, but almost two-thirds of the surgically treated patients with available follow-up information were still on some form of antireflux medications regularly (By contrast, data from comparative studies of surgical techniques and surgical cohort studies reported that approximately 90% of the patients were off antireflux medications at 5-years follow-up.). In a retrospective cohort analysis, Isolauri et al. reported that 5% of the surgically treated patients were on H2RAs occasionally compared with 33% of the medically treated patients at more than 10 years follow-up.²⁷

Table 5

Medical vs. surgical treatments of GERD: Change in medication usage status.

In a 4-year follow-up study using administrative data and computerized health care records from the Tennessee Medicaid program, Khaitan et al. determined that the proportion of persons using prescription medications (H2RAs, PPIs, or prokinetic agents) for acid suppression was lower in the surgical group than the medical group during each year of follow-up.²⁹ Using the same database, Holzman et al. reported a marked decrease in use of GERD-related prescription pharmaceuticals in the surgical cohort the first year after surgery.²⁶

Changes and follow-up information regarding Barrett's esophagus (Table 6)

Two RCTs and three non-randomized studies provided follow-up information on Barrett's esophagus.^{8, 22, 24, 27, 28, 31}. Lundell RCT reported that there was no difference in the point prevalence of Barrett's esophagus between the two study arms at 5-year follow-up.²² Spechler RCT reported an almost six-fold increase in the incidence of esophageal adenocarcinoma in patients with Barrett's esophagus at baseline compared to patients without Barrett's esophagus⁸ (see additional details in the next section). The retrospective study by Wetscher et al. reported that 14% of medically treated patients developed Barrett's esophagus at 2-year follow-up while none of the surgically treated patients developed Barrett's at 3.5 year follow-up.³¹ Isolauri et al. reported increase in Barrett's in both study arms at follow-up of 9 to 13 years.²⁷ Johansson and Tibbling reported one patient with Barrett's at baseline was found to have only “mild basal cell hyperplasia” 6 months after open fundoplication.²⁸

Table 6

Medical vs. surgical treatments of GERD: Status of Barrett's esophagus.

Incidence of esophageal adenocarcinoma (Table 7)

One RCT and two retrospective analyses provided information on the comparative incidence of esophageal adenocarcinoma.^{8, 30, 32} Spechler RCT reported no difference in the incidence of esophageal adenocarcinoma between the medical arm and the surgical arm at follow-up ranged from 4 to 12 years (this study also reported a six-fold increase in the incidence of esophageal adenocarcinoma in patients with Barrett's at baseline compared to patients without Barrett's).⁸ However, the authors acknowledged that the study lacked sufficient statistical power to detect important differences between groups in the rate of cancer development.

Table 7

Medical vs. surgical treatments of GERD: Incidence of esophageal adenocarcinoma.

Tran et al. - a retrospective study based on data from national computerized Veterans Administration databases (1986-1990) - examined the incidence of esophageal adenocarcinoma in patients with GERD who had medical treatment, patients with GERD who had surgical treatment, and in patients without GERD.³⁰ They did not ascertain prevalence of Barrett's esophagus. The mean duration of follow-up in this study was greater than 10 years. There were 1,892 patients in the GERD medical group, 946 in the surgical group, and 5,676 in the non-GERD group. During a follow-up period of 59,439 person-years and a mean duration of 10.5 years, no patients in the non-GERD group were diagnosed with esophageal cancer. During a follow-up period of 20,115 patient-years and a mean duration of 10.6 years, there were eight cases of esophageal cancer in the GERD medical group (40/100,000 person-years). During a follow-up period of 211,156 patient-years and a mean duration of 11.8 years, there were eight cases of esophageal cancer in the GERD fundoplication group (72/100,000 person-years). The difference in incidence rates between the two GERD groups was not statistically significant. Kaplan-Meier analysis showed that there was no statistically significant difference between the cumulative rates of esophageal cancer in the surgical group versus the medical group.

Ye et al. specifically examined the question of incidence of esophageal adenocarcinoma in patients with GERD who had surgical treatment and in patients who did not have surgical treatment.³² There were no data specifically concerning Barrett's in these studies. This retrospective cohort analysis was based on data obtained from a Swedish Inpatient Registry (1965 to 1997). The unoperated GERD group had 66,965 patients and the antireflux surgery group had 11,077 patients. The incidence rate of esophageal adenocarcinoma in males in the antireflux surgery group was 37/100,000 person-years. The incidence of esophageal adenocarcinoma in males in the unoperated group was 22.4/100,000 person-years. There were no cases of esophageal adenocarcinoma in females in the antireflux surgery group. The incidence of esophageal adenocarcinoma in females in the unoperated group was 6.6/100,000 person-years. Statistical comparisons between these groups were not reported.

Comparative effectiveness of medical treatments

Key points for interclass and intraclass comparisons of medical treatments

• We focused on comparisons between PPIs and H₂ receptor antagonists (H2RAs) and on PPI intraclass differences. Three recent, good-quality meta-analyses of RCTs comparing one medication versus another provided the data on the efficacy of medical treatments. The primary studies included in the meta-analyses had a follow-up duration of no more than 1 year with the exception of one study that reported a follow-up of 5 years.
• For medical treatments, results are applicable to adults with heartburn and/or regurgitation and some degree of esophagitis, corresponding to the characteristics of patients who were enrolled in the primary RCTs included in the meta-analyses.
• PPIs were superior to ranitidine for resolution of GERD symptoms at 4 weeks. PPIs were significantly more effective than ranitidine for healing of esophagitis at 8 weeks. PPIs at a standard dose (as suggested by the manufacturers' prescribing information) or a lower dose (usually one-half of the standard dose) were better than H2RAs in maintaining healing.
• There was no comparative difference between omeprazole, lansoprazole, pantoprazole, and rabeprazole for relief of symptoms at 8 weeks.
• No significant difference was found in the comparisons of esomeprazole 40 mg with lansoprazole 30 mg or pantoprazole 40 mg for relief of symptoms at 4 weeks. There was a significant difference in favor of esomeprazole when esomeprazole 40 mg was compared to omeprazole 20 mg for symptom relief at 4 weeks. The combined risk difference in three trials was 10% (95% CI 6%, 14%); for 10 patients treated with esomeprazole 40 mg versus omeprazole 20 mg, one additional patient would be symptom-free at 4 weeks in the esomeprazole group (NNT=10).
• For maintenance medical treatment of 6 months to 1 year, PPIs at a standard dose were more effective than at a lower dose in preventing relapse of symptoms.
• For healing of esophagitis at 8 weeks, there was no comparative difference between omeprazole, lansoprazole, pantoprazole, and rabeprazole.
• Esomeprazole 40 mg and lansoprazole 30 mg were equally effective in healing of esophagitis at 8 weeks in an analysis that combined the results of three RCTs. Two trials compared esomeprazole 40 mg to pantoprazole 40 mg for esophagitis healing at 8 weeks. One reported that healing rate was higher in the esomeprazole group; the risk difference was 3% (95% CI 1%, 5%) (NNT=33). In the other, esomeprazole and pantoprazole were equally effective. Two trials compared esomeprazole 40 mg with omeprazole 20 mg, and both found a higher 8-week healing rate in the esomeprazole group, risk difference was 8%, (95% CI 5%, 11%) (NNT=13). (Comment: The clinical importance of these differences is unclear since the magnitude of difference was small and the exact dosage equivalence between various PPIs has not been established. Furthermore, whether healing of esophagitis at 8 weeks predicts maintenance of healing is uncertain.)
• For maintenance medical treatment of 6 months to 1 year, PPIs at a standard dose were more effective than at a lower dose in preventing relapse of esophagitis.

Detailed analysis

We relied on three rigorously conducted systematic reviews of PPIs and H2RAs published in 2001, and 2005 for this section of the report. McDonagh and Carson from the Oregon Health and Science University Evidence-based Practice Center (OHSU EPC) assessed the comparative efficacy of different PPIs in healing esophagitis, and reducing symptoms of GERD, among adults presenting with GERD symptoms.³³ The methodological quality of this review was grade A. Caro et al. assessed the comparative efficacy of PPIs and ranitidine in healing esophagitis, and reducing symptoms of GERD, among adult outpatients with an endoscopically confirmed diagnosis of gastroesophageal reflux.³⁴ The methodological quality of this review was grade B. Donnellan et al. assessed the efficacy of PPIs (at different doses) compared to H2RAs in preventing the relapse of mucosal inflammation in adults with esophagitis as well as in preventing relapse of symptoms in adults with endoscopy negative reflux disease and esophagitis.³⁵ The methodological quality of this review was grade A.

PPIs included esomeprazole, lansoprazole, omeprazole, pantoprazole, and rabeprazole. Donnellan et al.³⁵ defined healing or standard dose as 20 mg once daily for esomeprazole, omeprazole, and rabeprazole; 30 mg once daily for lansoprazole; and 40 mg once daily for pantoprazole. A lower, maintenance dose was defined 10 mg once daily for esomeprazole, omeprazole, and rabeprazole; 15 mg once daily for lansoprazole; and 20 mg once daily for pantoprazole. Double dose was defined as double the standard (healing) dose. These definitions were not adopted by the other two meta-analyses.^{33, 34} H2RAs included cimetidine, famotidine, nizatidine or ranitidine. Acute treatment ranged in duration from 4 to 8 weeks while maintenance treatment was administered continuously for at least 6 months.

Acute treatment of symptoms

McDonagh and Carson³³ reported four RCTs with comparisons between PPIs that measured symptom relief as a primary outcome, and 13 that reported symptom relief as a secondary outcome. Symptom relief in these studies was assessed through patient diaries, investigator-elicited reports, or both. The definition of “symptom relief” varied. Evaluated symptoms included day and nighttime heartburn, dysphagia, odynophagia, pain on swallowing, or acid regurgitation. Complete resolution of heartburn was usually defined as seven consecutive days without heartburn. Quality of life and patient satisfaction were also evaluated in one study.

Fourteen trials reported the proportion of patients with complete resolution of symptoms at 4 weeks. McDonagh and Carson performed a random effects meta-analysis of data from these studies to determine an estimate of the proportion of patients who were symptom free at 4 weeks for each drug. Proportions ranged from 65% to 77%; and 95% confidence intervals overlapped, indicating the drugs are similarly efficacious for complete resolution of symptoms at 4 weeks. Risk differences in rates of complete symptom resolution at 4 weeks were also calculated in these trials. With the exception of esomeprazole 40 mg versus omeprazole 20 mg, risk differences were non-significant in rest of the comparisons. The combined data significantly favored esomeprazole 40 mg; for every 10 persons treated with esomeprazole 40 mg versus omeprazole 20 mg, one additional patient would be symptom-free at 4 weeks in the esomeprazole group. The combined data for esomeprazole 40 mg versus either lansoprazole 30 mg (risk difference 5%; 95%CI 0%, 9%) or pantoprazole 40 mg (risk difference 2%; 95%CI –11%, 7%) did not indicate a significant difference between the drugs.

Eleven studies reported the time to resolution of symptoms (defined as the absence of heartburn) either as the percentage of patients with the outcome after a given time point (1 day, 7 days, etc.) or the median number of days to resolution, or both. In one study this outcome was reported as the number of days needed for 50% and 75% of patients to achieve relief of symptoms. Another measure used was “the time to sustained resolution of heartburn”, defined as the time to the first series of 7 consecutive days without heartburn. This outcome was used only in studies on esomeprazole, so it is not possible to compare this outcome with studies of other PPIs. However, time-to-relief of heartburn was similar for all PPIs.

Caro et al.³⁴ analyzed 11 RCTs that compared a PPI with ranitidine. These RCTs reported resolution of symptoms as an outcome. Omeprazole was used in eight RCTs, lansoprazole in one, pantoprazole in one, and rabeprazole in another. Symptoms in these studies were assessed through patient diary cards, interviews, and visual analogue scales. Evaluated symptoms included day and nighttime heartburn, dysphagia, odynophagia, acid eructation, or regurgitation. PPIs were generally superior to ranitidine for resolution of GERD symptoms at 4 weeks. Because the studies used different methods of collecting and recording various symptoms at different time points, only the data on complete heartburn resolution were combined. The rate of heartburn resolution was 1.53-fold higher with PPIs compared with ranitidine (95% CI 1.37, 1.72).

Acute treatment of esophagitis (Tables 8, 9)

McDonagh and Carson³³ retrieved 13 published RCTs that compared a PPI versus a second PPI. All of the PPIs were effective at healing esophagitis. Healing rates at 4 weeks ranged from 49% to 91%, and at 8 weeks ranged from 71 % to 99%. To determine an estimate of healing rates for each drug, they combined data from the trials, using a random effects model to control for the effect of the study. Healing rates were similar and confidence intervals overlapped, indicating no significant differences among PPIs.

Table 9

Medical treatment for GERD: Esophagitis healing after 4-weeks of medical treatment.

McDonagh and Carson also calculated the percent risk difference for healing in the comparisons. Table 8 shows the differences in healing rates at 4 and/or 8 weeks for the 18 trials that provided the number healed and the total number of patients in each arm of the studies. With the exception of esomeprazole 40 mg versus omeprazole 20 mg, risk differences at 4 and 8 weeks were non-significant in rest of the comparisons. Two trials compared esomeprazole 40 mg to omeprazole 20 mg, and both found a higher healing rate in the esomeprazole group. Two studies compared esomeprazole 20 mg to omeprazole 20 mg, and found no significant difference in healing rate at 4 or 8 weeks.

Table 8

Medical Treatment for GERD: Esophagitis healing rates in trials of PPIs (Risk difference, 95% CI).

Three studies compared esomeprazole 40 mg to lansoprazole 30 mg. In a large trial with 5,241 patients at multiple centers in the US, healing rates were higher in the esomeprazole group at 4 weeks (risk difference 4%; 95% CI 2%, 6%) and at 8 weeks (risk difference 3%; 95% CI 1%, 5%). A second, smaller trial of lansoprazole 30 mg versus esomeprazole 40 mg in patients with mostly mild to moderate esophagitis found the two to have equivalent healing rates at 8 weeks. Results at 4 weeks were not reported. The third study was conducted in patients with moderate to severe esophagitis (Los Angeles Grade C and D). At 4 weeks, the esomeprazole group had a higher healing rate, but at 8 weeks the difference was not significant. Combined estimates showed a 5% higher healing rate at 4 weeks and 3% at 8 weeks for esomeprazole 40 mg. The difference at 8 weeks was not significant using a random effects model (risk difference 3%; 95% CI 0%, 5%). Two trials compared esomeprazole 40 mg to pantoprazole 40 mg. In one trial that was rated as fair to poor quality in the McDonagh and Carson report, healing at 4 weeks was 6% greater in the esomeprazole group (95% CI 3%, 9%). At 8 weeks, the difference was smaller but statistically significant (risk difference 3%; 95% CI 1%, 5%). In the other comparison of esomeprazole 40 mg to pantoprazole 40 mg, healing rates were reported at “early” (4 to 6 weeks) and “late” (8 to10 weeks) time points. Healing rates were equivalent at early and late time points. It was not possible to pool these two studies because of differences in how results were reported. In addition, one of the studies included only patients with grade B (84%) and C (16%) esophagitis, whereas the other study enrolled patients with grade A through D.

In the systematic review by Caro et al.,³⁴ seven RCTs compared omeprazole 20 mg with ranitidine in 1,575 participants. The omeprazole group achieved a significantly higher healing rate (relative risk 1.81; 95% CI 1.54, 2.13) at 4 weeks. In the same systematic review, lansoprazole was associated with a significantly higher healing rate than ranitidine (relative risk 1.83; 95% CI 1.63, 2.08) at 4 weeks, as shown by three RCTs including 948 patients. Two RCTs compared pantoprazole versus ranitidine and rabeprazole versus ranitidine, recruiting 249 and 338 subjects, respectively. The pantoprazole and rabeprazole groups achieved significantly higher healing rates than ranitidine at 4 weeks. At 8 weeks, a similar pattern was observed in all the comparisons between PPIs and ranitidine.

Maintenance of symptom relief (Table 10)

Donnellan et al.³⁵ included 18 RCTs in a meta-analysis that compared healing versus maintenance dose of a PPI in 5,116 participants. At the end of follow-up, which ranged from 26 to 52 weeks among the RCTs, the percentage of patients experiencing relapse of symptoms was significantly lower for those treated with a healing dose of a PPI (relative risk 0.78; 95%CI 0.68, 0.88).

In the same meta-analysis, there were five RCTs comparing a healing dose of a PPI versus H2RA in about 800 participants. At the end of follow-up, which again ranged from 26 to 52 weeks, the percentage of patients experiencing relapse of symptoms was significantly lower for those treated with a healing dose of PPI (relative risk 0.48; 95%CI 0.39, 0.60).

Four RCTs compared a maintenance dose of a PPI with H2RA in 831 participants. At the end of follow-up, which ranged from 24 to 52 weeks, the percentage of patients experiencing relapse of symptoms was significantly lower for those treated with a maintenance dose of a PPI (relative risk 0.55; 95%CI 0.47, 0.65).

There were three more comparisons addressed by the systematic review: a healing dose of PPI versus a second healing dose of a PPI, which included two RCTs with 1,001 participants; lansoprazole maintenance dose versus lansoprazole 30 mg alternate days, including two RCTs with 187 subjects; and double dose of PPI versus healing dose of a PPI that used two RCTs with 347 patients. None of the three comparisons showed any significant benefit for any of the treatment modalities examined.

Meta-regression was also used to assess the effect of follow-up duration, country in which the trial was conducted, type of drug, and method of randomization on the outcomes. No significant association between any of the factors and relapse of symptoms was reported.

Maintenance of healed esophagitis (Table 11)

Twenty-two RCTs were included in a meta-analysis comparing a healing versus maintenance dose of a PPI in about 6,000 participants.³⁵ Less than one third of the participants presented initially with severe esophagitis. At the end of follow-up, which ranged from 24 to 52 weeks, the percentage of patients experiencing relapse of esophagitis was significantly lower for those treated with a healing dose of PPI (relative risk 0.63; 95%CI 0.55, 0.73).

Table 11

Medical treatments for GERD: Relapse of esophagitis on maintenance therapy.

Ten RCTs compared healing dose of a PPI with an H2RA in about 1,600 participants. Almost one third of the participants presented initially with severe esophagitis. At the end of follow-up, again from 24 to 52 weeks, the percentage of participants experiencing relapse of esophagitis was significantly lower for those treated with a healing dose of a PPI (relative risk 0.36; 95%CI 0.28, 0.46).

Six RCTs compared a maintenance dose of a PPI with an H2RA in 1,156 participants. About 30% of the participants presented initially with severe esophagitis. At the end of 24 to 52 weeks of follow-up, the percentage of patients experiencing relapse of esophagitis was significantly lower for those treated with a maintenance dose of PPI (relative risk 0.57; 95%CI 0.47, 0.69).

There were three more comparisons of interest addressed by the systematic review: a healing dose of PPI versus a second healing dose of a PPI, which included three RCTs with 1,020 participants; lansoprazole maintenance dose versus lansoprazole 30 mg alternate days, including two RCTs with 189 subjects; and esomeprazole 40 mg versus esomeprazole 20 mg that used two RCTs with 693 patients. None of the three comparisons showed any significant benefit for any of the treatment modalities examined.

Meta-regression was also used to assess the effect of follow-up duration, country in which the trial was conducted, type of drug and method of randomization on the outcomes. No significant association between any of the factors and relapse of esophagitis was reported.

Key points for comparisons of surgical techniques

• Four RCTs and four non-randomized studies compared different approaches to fundoplication. All the RCTs were graded methodological quality B. The grade of the non-randomized comparative studies ranged from B to C.
• Studies differed in their inclusion criteria, particularly in the extent to which the prior response to medical therapy was described. Only non-randomized trials explicitly included patients with an unsatisfactory response to medical treatment. However, even in such studies the definitions of “unsatisfactory” varied and were not always defined clearly.
• One RCT and one non-randomized study compared laparoscopic with open approach. Efficacy was similar for the two approaches. There was no difference in outcomes of heartburn, regurgitation, QOL, and usage of antisecretory medications. Almost 90% of patients who were followed for 5 or more years in both surgical arms reported improvement in symptoms.
• There was no difference in the efficacy for symptom relief, QOL improvement and decrease usage of antisecretory medications in the two study arms of laparoscopic total fundoplication versus partial fundoplication, laparoscopic fundoplication with division of short gastric vessels versus without, and open total fundoplication versus partial fundoplication. Notably, fewer than 50 percent of patients who had undergone surgery had follow-up at 5 or more years.

Table 12

Comparative studies evaluating the long-term outcomes of Laparoscopic Total fundoplication versus Open fundoplication.

Table 13

Comparative studies evaluating the long-term outcomes of technique variation of laparoscopic fundoplication and open fundoplication.

Table 14

Cohort studies evaluating the long-term outcomes of fundoplication.

Key points for long-term effectiveness of surgery

• In addition to the long-term data from comparative studies, nine observational studies also provided data on long-term effectiveness of surgery. These focused on patients with GERD that had been well documented by objective measures. The methodological quality of the studies ranged from grade B to C.
• All of the studies reported improvement in symptoms in 80–90% of patients at 5 or more years of follow-up.
• Only one-third of the studies provided data on QOL at 5 or more years. These suggested that QOL had either improved compared to the preoperative period or had normalized.
• None of the observational studies reported esophagitis status at follow-up.
• About one-third of the surgical studies reported data on pH status. The mean pH score had normalized; however they did not report the proportion of patients in whom esophageal pH exposure improved or normalized.
• The proportion of patients requiring regular antisecretory medications after surgery was described in about two-thirds of the studies. Approximately 80 to 90 percent of patients were off all such medications. However, fewer than 50 percent of patients who had undergone surgery were available for long-term follow-up and there was no information regarding medication use in patients who were lost to follow-up.

Detailed analysis of technique comparisons

We identified four RCTs (a total of 406 patients)³⁶–³⁹ and four non-randomized comparative studies (a total of 1,141 patients in five publications)⁴⁰–⁴⁴ of fundoplication for the treatment of GERD. One RCT³⁸ and one non-randomized comparative study⁴⁴ examined open fundoplication versus laparoscopic fundoplication. Two RCTs^{37, 39} and three non-randomized comparative studies⁴⁰–⁴³ compared two different approaches to laparoscopic fundoplication techniques: laparoscopic total versus partial, laparoscopic with versus without division of short gastric vessels. One non-randomized comparative study compared open total fundoplication versus open partial fundoplication.³⁶ More than 80% of the subjects had follow-up data from 5 to 11 years in the RCTs. All the RCTs were graded methodological quality B. Approximately 60%⁴¹ and 20%⁴⁰ of subjects had follow-up of 5 to 6 years, respectively, in two prospective non-randomized comparative studies. The methodological grades in non-randomized comparative studies ranged from B to C.

All studies were conducted in an academic setting. All patients had symptoms of chronic GERD, and all had at least one GERD-related diagnostic test performed in the preoperative period. The studies differed in their inclusion criteria, particularly in the extent to which patients' prior response to medical therapy was described. Only one RCT³⁸ stated clearly patients were treated with a PPI but the specific definition of response or the dose used was not reported. Two RCTs^{36, 38} reported consecutive patient enrollment. Two RCTs excluded patients with severe esophageal motility disorders.^{37, 39}

Among non-randomized comparative studies, three reported recurrent GERD symptoms in the patients despite 15 to 24 months of treatment with PPIs.^{40, 41, 44} One study reported consecutive patient enrollment.⁴⁴ One study reported symptom recurrence shortly after medication withdrawal as an indication for surgery.⁴⁴ Three studies of laparoscopic partial fundoplication included patients with poor esophageal motility.^{40, 41, 43} One study excluded 47% of patients from analysis for multiple reasons.⁴³

Symptoms and quality of life

Symptoms and quality of life were assessed using a structured interview or validated quality-of-life scale instrument. Among RCTs, almost 90% of patients who were followed for 5 or more years in both surgical arms reported improvement in heartburn or regurgitation. Three RCTs³⁷–³⁹ reported improved quality-of-life in both treatment arms. One RCT reported significant improvement in quality of life compared to pre-operative period but no significant differences between surgical treatments.³⁸

One non-randomized study reported a significantly higher proportion of patients with improved GERD symptoms in the total fundoplication group, compared with the patients in the partial fundoplication group.⁴³ In the rest of the non-randomized studies, 80 to 90% of patients with follow-up data reported improvement in heartburn or regurgitation in both arms at 5 or more years. Two non-randomized comparative studies provided data on improved quality of life with mean score comparable to that of healthy individuals, but provided no information on the proportion of patients with improvement.^{40, 41}

Need for antisecretory medications

Two RCTs^{37, 39} reported that approximately 90% of patients were off PPIs or off all antisecretory medications in both treatment arms at 5-year follow-up. One RCT reported more subjects were off PPIs in the laparoscopic fundoplication arm compared with the open fundoplication arm.³⁸ Approximately 90% of patients were off PPIs or off all antisecretory medications in both treatment arms of two non-randomized comparative studies^{41, 44}at 5-year follow-up.

Esophageal acid exposure and manometry findings

One RCT reported the mean pH score returned to normal and the mean LES pressure improved in both treatment arms but provided no information on the proportion of patients with improvement.³⁸ Other RCTs did not provide data on esophageal acid exposure at long-term follow-up. Three non-randomized comparative studies reported long-term data on pH status and esophageal manometric studies,^{40, 41, 43} and one reported a significantly higher proportion of patients who had normalized pH status with laparoscopic total fundoplication compared to patients with partial fundoplication (72% versus 44%).⁴³

Healing of esophagitis

No studies provided long-term endoscopic data.

Barrett's esophagus and adenocarcinoma

Two RCTs included patients with Barrett's esophagus (6%³⁶ and 13%³⁸). None of these patients developed dysplasia and/or adenocarcinoma in the columnar-lined esophagus at 5 or more years of follow-up. No long-term follow-up data were provided in a non-randomized comparative study where 52% of the patients had Barrett's esophagus in the preoperative period.⁴¹

Detailed analysis of long-term effectiveness of surgery

In addition to the comparative studies, five cohort studies⁴⁵–⁴⁹ of laparoscopic fundoplication and four cohort studies⁵⁰–⁵³ of open fundoplication provided data on long-term efficacy of surgical treatment of GERD. These nine cohort studies enrolled a total of 1,752 patients. Except for two studies that reported close to 100% follow-up for 6 years⁴⁹ and 20 years⁵¹, the rest of the observational studies had fewer than 50% of patients at follow-up of 5 to 6 years. The methodological grades in these studies ranged from B to C.

All studies were conducted in an academic setting. All patients had symptoms of chronic GERD, and all had at least one GERD-related diagnostic test performed in the preoperative period. The studies differed in their inclusion criteria, particularly in the extent to which patients' prior response to medical therapy was described.

Three studies reported recurrent GERD symptoms despite 6 to 18 months of treatment with PPIs or antisecretory medications.^{45, 48, 50} One study included patients whose symptoms were controlled adequately on PPIs, but chose to undergo surgery in preference to long-term medical therapy.⁴⁵ Four studies^{47, 49, 50, 52} enrolled consecutively operated patients and one⁵² included patients who underwent primary antireflux procedures.

Symptoms and quality of life

All cohort studies reported improvement in reflux symptoms. One study reported that preoperative response to PPI correlated significantly with response to surgery.⁴⁵

Need for antisecretory medications

All cohort studies (except for two studies^{52, 53} which did not provide data) reported that 80 to 90% of the subjects were off PPIs or off all antisecretory medications.

Esophageal acid exposure and manometry findings

Four cohort studies^{45, 48, 50, 53} reported data on pH status, and in two,^{50, 53} the proportion of patients who had normalized was also mentioned. One study reported significant improvement in mean esophageal acid exposure score and manometric score compared to preoperative period.⁴⁵

Healing of esophagitis

No cohort studies provided long-term endoscopic data.

Barrett's esophagus and adenocarcinoma

Thirteen percent of patients in one cohort study had Barrett's esophagus and none developed dysplasia or adenocarcinoma at 6-year follow-up.⁴⁶

Comparative endoscopic studies (comparison with sham) and cohort studies (Tables 15, 16)

Key points for comparative endoscopic studies (comparison with sham) and cohort studies

Table 15

Endoscopic treatments: Sham controlled trials.

• Three endoscopic procedures are available in the US: EndoCinch™ Suturing System, NDOplicator™, and Stretta™. A fourth (Enteryx™) was voluntarily removed from the market due to safety concerns during final preparation of this report. We elected to include the data pertaining to Enteryx™ since it was the method used in one of the only two sham-controlled trials and because of the relatively large number of reports, which allowed for a better understanding of how various endpoints in the endoscopic studies correlated with one another.
• There were only two sham-controlled trials: one using Enteryx™ and the other Stretta™. There were 14 cohort studies. The longest mean follow-up was one year for the sham-controlled trials and 27 months for the cohort studies. Both sham-controlled trials were given a methodological grade of B. The methodological grade of the cohort studies ranged from B to C. No studies directly compared one endoscopic procedure with another.
• The inclusion and exclusion criteria in most of the studies limited applicability to patients with well-defined GERD who have a small hiatal hernia and either mild to moderate or no esophagitis.
• Stretta™ was more effective than sham in improving symptoms of reflux and QOL at 6 months. Enteryx™ was more effective than sham in improving symptoms of reflux at 3 months. QOL at 6 months improved in both the Enteryx™ and sham groups. Between group comparison was not described.
• Enteryx™ and sham both reported improvement in QOL at 6 months.
• Healing of esophagitis has not been consistently demonstrated in endoscopically treated patients. Esophagitis improved in some patients and worsened in others in the various reports.
• Improvement of esophageal pH exposure compared with sham could not be demonstrated for either Enteryx™ or Stretta™. Uncontrolled studies of all the endoscopic procedures suggest improvement or normalization in pH in some patients, but there were insufficient data to determine the magnitude of improvement relative to one another or the correlation of pH changes with other outcome measures.
• Significantly more patients treated with Enteryx™ were able to discontinue PPIs compared with sham while there was no significant difference in patients treated with Stretta™. The proportion of patients who were freed from regular use of any antisecretory agents (PPIs, H2RAs, or antacids) was no different between sham and Stretta™.

Detailed analysis

Three endoscopic therapies are currently available in the United States while a fourth (Enteryx™) was voluntarily recalled in September 2005:

1.

The Stretta™ procedure (Curon Medical, Freemont, CA), which involves application of radiofrequency energy to the lower esophageal sphincter

2.

EndoCinch™ Suturing System (EndoCinch™) (Bard, Murray Hill, NJ), which involves creation of a submucosal plication in the region of the gastric cardia using a device that allows for sutures to be placed endoscopically.

3.

NDO Plicator™ (NDO Plicator ™, NDO Surgical, Mansfield, MA), which involves creation of a transmural plication in the region of the gastric cardia.

4.

Enteryx™ (Boston Scientific, Natick, MA), which involves injection of a biopolymer into the lower esophageal sphincter.

Our literature search identified a total of 30 studies on endoscopic procedures (see Evidence Table 2 for details). Five studies⁵⁴–⁵⁸ included data from patients already reported in 10 other studies^{57, 59}–⁶⁷ from this group of 30. As a result, data from these 10 studies were not included in this review unless they provided unique information (two studies^{57, 65}) that was not included in the five studies, but each patient was analyzed only once. In final tabulation, data from 22 studies on endoscopic procedures are summarized below. Three of them were RCTs.^{6, 7, 68} Three of them were non-randomized comparisons.^{65, 69, 70} Fourteen were cohort studies. One was a survey⁷¹ and one was a post hoc analysis⁵⁸ of data from two other studies.^{57, 67} Additional data from unpublished studies that have been presented in preliminary form were not included in the primary analysis but are discussed in Appendix F.

Two sham-controlled studies have been published in final form: one for Stretta^{TM 7} and the other for Enteryx^TM.6 Four other comparative trials were identified. One RCT⁶⁸ compared two different configurations of EndoCinch™ sutures. Two other non-randomized studies^{69, 70} compared EndoCinch™ with laparoscopic fundoplication. A fourth non-randomized study⁶⁵ compared the Stretta™ procedure with laparoscopic fundoplication. We did not identify a RCT comparing any of the endoscopic techniques directly with fundoplication or continued (or intensified) medical therapy. No studies directly compared the endoscopic procedures to one another.

Studies were generally of short duration, reporting outcomes at three, six and 12 months with only a few reports describing follow-up as long as 24 to 27 months.^{54, 55, 72} The largest studies (other than a survey study of 558 respondents⁷¹) included fewer than 150 subjects,⁵⁴ while the two sham-controlled trials included only 64 subjects each.^{6, 7}.

Both of the sham-controlled studies were rated methodologic quality B. Their main deficiencies were small size and short duration. Effectiveness of blinding was not assessed in either study. Many of the uncontrolled studies and non-randomized comparative studies were rated methodologic quality C. They had important limitations in their design, analysis (such as lack of an intention-to-treat analysis), and/or reporting. The corresponding device manufacturer provided direct funding for almost all the reports or supported the investigators.

Patient characteristics were similar in all of the endoscopic trials. In general, all the studies defined GERD by symptoms, findings at endoscopy, pH studies, and response to medications. Approximately half of the studies excluded patients who had severe esophagitis (Savary Miller grade >2 or Los Angeles grade >C). All of the studies excluded patients with a hiatal hernia exceeding a certain size. Studies of Stretta™, EndoCinch™, and NDO Plicator™ generally excluded patients with a hiatal hernia >2cm while two of the Enteryx™ studies permitted inclusion of patients with a hiatal hernia as large as 5 cm.^{6, 54} Patients with severe esophageal motility disturbances were excluded from all of the studies.

As noted in the methods section, we focused on the five major important outcomes in treating GERD: symptoms, change in esophagitis grade, improvement in esophageal pH exposure, reduction in the need for antisecretory medications, and improvement in quality of life. Details of these outcomes are presented in the evidence tables while the key points are summarized here. Adverse effects are presented separately. Because of the short-term follow-up, there are no data from which to assess other important objectives such as prevention of esophageal stricture, development of Barrett's esophagus or esophageal adenocarcinoma, or possible long-term adverse effects from therapy.

Patients in the two sham studies had PPI-responsive GERD who had no more than a small hiatal hernia and only mild to moderate esophagitis. Stretta™ was more effective than sham in improving symptoms of reflux and QOL at 6 months.⁷ Symptoms at follow-up were assessed while patients were off their medications. Information on symptoms while on medications was not described. Enteryx™ was more effective than sham in improving symptoms of reflux at 3 months.⁶ Enteryx™ and sham both reported improvement in QOL at 6 months compared with baseline. Between group comparison was not described. Significantly more patients treated with Enteryx™ were able to discontinue PPIs compared with sham while there was no significant difference in patients treated with Stretta™. The proportion of patients who were freed from regular use of any antisecretory agents (PPIs, H2RAs or antacids) was no different between Stretta™ and sham, and was not reported in the Enteryx™ study. An improvement in esophageal pH exposure could not be demonstrated when compared to sham for either Enteryx™ or Stretta™. Both studies had small number of patients and short duration of follow-up.

Almost all of the uncontrolled and non-randomized endoscopic studies reported symptom improvement compared to baseline (see Table 16). Normalization of esophageal pH exposure was observed in 25 to 50% of patients. Except for one study on the Stretta™ procedure,⁵⁸ which found that responders (defined by improvement in quality of life) had improved distal esophageal exposure time compared with nonresponders, none of the other studies showed data on correlation of pH changes with other outcome measures.

Table 16

Endoscopic treatments: Uncontrolled and non-randomized studies.

Almost all the uncontrolled studies showed a reduction in the need for PPIs. The proportion of patients who did not require any antisecretory medications was reported infrequently but was in the range of only 25 to 40% in studies in which it was described.

The effect of the endoscopic procedures on healing of esophagitis was unclear. Two uncontrolled studies of Enteryx™ showed a worsening of esophagitis in approximately one-third of patients;^{54, 73} the same studies also reported improvement in approximately 10% of the patients.

The durability of benefits for any of the procedures is unclear because most of the studies had short-term follow-up. One study on EndoCinch™ reported improvement in symptoms relative to baseline at 24 months.⁷² However, two studies on EndoCinch™ with follow-up of 12 and 18 months, respectively, reported loss of plications in the majority of patients.^{74, 75}

Only three studies compared an endoscopic procedure with laparoscopic fundoplication: two with EndoCinch™,^{69, 70} and one with Stretta^TM.65 The two studies of EndoCinch^{TM 69,70} concluded that patients who had fundoplication were more satisfied with their results. One of them reported that significantly fewer patients in the fundoplication group used PPIs or motility drug compared with patients in the EndoCinch™ group at follow-up of 8 months.⁶⁹ The study on Stretta^{TM 65} reported that fewer patients in the fundoplication group required PPIs, but both procedures improved symptoms and quality of life. No studies directly compared the endoscopic procedures to one another.

Detailed analysis

We reviewed studies comparing fundoplication with medical treatment in patients with Barrett's esophagus that reported the incidence of adenocarcinoma as an outcome. Only one RCT addressed this question. Parrilla et al. reported data from 113 patients with Barrett's esophagus from 1982 to 2000.⁸² (Ortiz et al.⁸³ reported earlier data on 59 patients from 1982-1993 apparently based on the same trial). The authors reported that multiple biopsies were taken systematically during endoscopy: “the first immediately above the squamo-columnar junction, the next immediately below, and the rest every 1–2 cm in a caudal direction in the various quadrants of the circumference.”⁸³ Both macroscopic and histological data were presented. In the medical treatment arm, the median follow-up was 5 years (range 1–18), two patients developed adenocarcinoma of the esophagus. In the surgical treatment arm, the median follow-up was 6 years (range 1–18), also two patients developed adenocarcinoma. The rate of malignancy was 1/129 patient-years (0.8% per year) for the medical group and 1/203 patient-years (0.5% per year) for the surgical group. The difference was not significant.

There were two pooled analyses that compared the reported rates of esophageal adenocarcinoma in patients managed medically with patients who underwent surgery^{84, 85}. A review by Corey et al. included cohorts and trials published from 1966 through October 2001.⁸⁵ Sixteen publications had at least 12 months of follow-up. Medical patient-years totaled 4906. Surgical patient-years totaled 4678. The incidence rate of esophageal adenocarcinoma in patients with Barrett's esophagus who received medical treatment was 5.3 (95%CI, 3.6–7.8)/1000 patient-years. The incidence rate of esophageal adenocarcinoma in patients with Barrett's esophagus who received surgical treatment was 3.8 (95%CI, 2.4–6.1)/1000 patient-years. The incidence rate of esophageal adenocarcinoma in the subgroup of patients with Barrett's esophagus who received medical treatment in the most recent 5 year analyzed (1996-2001) was 4.3 (95%CI, 2.6–5.8)/1000 patient-years. This subanalysis was done because of the advances in medical therapy in recent years with the addition of proton pump inhibitors to GERD treatment. There was no significant difference in rate of esophageal adenocarcinoma per patient-year between the surgical and medical groups or between the surgical groups and the medical therapy group in the last 5 years.

The study by Bammer et al.⁸⁴ included a retrospective review of 21 articles regarding the cancer risk for patients with Barrett's treated medically; it was 1 in 144.7 patient-years. A review of 19 articles including unpublished data showed the cancer risk in patients after antireflux surgery was 1 in 294.4 patient-years. The mean follow-up for medical surveillance was 2.7 years; the mean follow-up for antireflux surgical surveillance was 4.0 years. The authors stated that patient groups were not homogeneous in terms of age, length of symptom duration, presence of Barrett esophagus and dysplasia at baseline, and other possible risk factors for Barrett cancer. Statistical comparative data were not presented in this study. The authors concluded that the data do not prove that surgery is superior to medical treatment in the prevention of Barrett cancer, but they show a strong tendency for surgery to be the better treatment to prevent progression and the development of Barrett carcinoma.

Detailed analysis

We did not find any epidemiologic studies describing the setting in which most surgical and endoscopic antireflux procedures are performed. Most of the published literature of surgical and endoscopic therapies were written by authors affiliated with academic institutions.

Judging by the characteristics of patients who were included in the treatment studies, the vast majority of patients who eventually underwent surgical therapies exhibited the typical symptoms of GERD, including heartburn and regurgitation. Primary complaints of dysphagia, water brash, chest pain, respiratory symptoms, and bloating were much less frequent.⁸⁶–⁸⁹ Patients usually were symptomatic for several years before undergoing surgery.^{87, 90}–⁹² The majority of patients who underwent laparoscopic therapy reported a moderate level of symptoms as measured by Visick, DeMeester, or other symptom scale.^{90, 91, 93, 94}

In terms of patient demographic data, approximately half of the surgical patients were in the 40 to 60 years age bracket.^{92, 95, 96} The sex distribution was approximately equal. The range of the weight data for the few studies that provided these data include BMI 19 to ≥35 and weight 86–87 Kg with one study⁹² reporting less than 70 kg to more than 90 kg. Race or smoking status were rarely reported.

Objective testing for GERD in the observational surgical studies varied widely. In some studies, pH studies were undertaken only in the subset of patients presenting with atypical symptoms or those unresponsive to PPI therapy, but without evidence of esophagitis on endoscopy or reflux on barium swallow.⁹¹–⁹³ More than 40 percent of patients had at least the equivalent of Savary-Miller grade 1 esophagitis in studies for which it was reported.^{86, 89, 90, 92, 93, 96}–⁹⁸ Preoperative hiatal hernias were common in both patients who underwent open or laparoscopic treatment with more than 40 percent of all patients with evidence of a hiatal hernia in studies for which it was reported.^{86, 87, 89, 91, 92, 96}–⁹⁸

Endoscopic studies had strict entry criteria. Most patients did not have severe esophagitis (Savary Miller grade >2 or Los Angeles grade >C). All but two studies excluded patients with hiatal hernia > 2 to 3 cm; the two exceptions permitted inclusion of patients with hiatal hernia up to 3 to 5 cm.^{6, 54} Patients with severe esophageal motility disturbances were excluded from all the studies.

Detailed analysis

There were two studies that examined whether factors that influenced treatment outcome differed for those treated medically versus surgically.^{82, 99} In an observational study of 28 patients with low LES resting pressure, those who declined surgery for medical therapy (H2RAs or omeprazole) had worse symptoms and were less likely to have discontinued all antireflux medications than those who opted for open Nissen fundoplication after an average of 31 months post therapy.⁹⁹ In one RCT involving 101 patients with Barrett's esophagus, rates of esophagitis were greater in patients treated medically (H2RAs and/or omeprazole) than in patients treated with Nissen fundoplication after a median of 5 years. There was no significant difference between the two groups in rates of Barrett's esophagus, length of Barrett's segment, progression to dysplasia, or adenocarcinoma at follow-up.⁸²

Factors that influenced the outcome of medical therapy

A meta-analysis of five RCTs involving a total of 1,154 patients found an increased risk of esophagitis during maintenance therapy on omeprazole, ranitidine, or placebo in patients with the following associations: 1) increased pretreatment severity of esophagitis; 2) younger age; 3) non-smoking status; and 4) moderate/severe regurgitation. Body mass index, sex, alcohol consumption, concomitant NSAID therapy, disease duration, pretreatment severity of heartburn, and residual symptoms at healing were non-significant factors for relapse.¹⁰⁰.

Four studies evaluated specific patient characteristics on outcomes of medical therapy.^{99, 101}–¹⁰³ Two evaluated the risk of medical therapy failure based on esophageal manometry.^{99, 103} One of these (a cohort study with 55 patients) reported that the likelihood of stopping all antireflux medicines was reduced in patients with low LES pressure.⁹⁹ The likelihood of stopping all medical therapy was not influenced by esophageal body motility or esophageal pH pattern. A single arm, open trial of omeprazole involving 128 patients with mild esophagitis reported that the likelihood of relapse/persistence was higher in patients with a structurally incompetent LES.¹⁰³ The risk was highest in patients with LES incompetence and poor esophageal peristalsis. An RCT conducted at 19 centers examined the effects of ranitidine versus placebo on 232 patients.¹⁰¹ Smoking status had no effect on esophageal healing between or within treatment groups. Although there was no difference in symptomatic improvement regardless of the intervention group, antacid use was reduced significantly among smokers on ranitidine versus smokers on placebo (P<0.05). One cohort study found no difference between elderly and non-elderly for esophageal healing and heartburn resolution after ranitidine therapy.¹⁰²

Factors that influenced the outcome of fundoplication (Table 17)

There were 10 case-control and 15 cohort studies of laparoscopic antireflux surgery studies that analyzed the influence of specific patient characteristics on outcomes after surgery. There were no randomized controlled or direct comparison trials. Variables examined included study setting, race, sex, obesity, aerophagia, weight, symptoms, preoperative response to antisecretory medications, severity of acid reflux, preoperative LES incompetence and pressure, esophagitis severity, presence of hiatal hernia, and psychological traits. Outcomes assessed included symptoms; pH status; whether the patients were off PPIs or all medications; quality of life or patient satisfaction, which included response of whether patient would undergo surgery again; and global success or failure. Evidence for effectiveness of therapies based on patient characteristics was limited, and almost all studies that we identified were graded as methodologic quality B or C. Results are summarized below.

Table 17

Summary of studies that evaluated patient characteristics as modifying factors of fundoplication outcome.

One study examined the rates of laparoscopic antireflux surgery failure in 445 patients identified through a nationwide registry in Sweden who were treated at low- or high-volume hospitals.⁹⁵ There was no significant increase in the rates of laparoscopic antireflux surgery failure in patients between these two settings. Nevertheless, the learning curve for surgeons performing laparoscopic fundoplication has been documented in several studies.¹⁰⁴–¹⁰⁸ Two studies found that outcomes of laparoscopic fundoplication performed by a relatively inexperienced surgeon could be improved by assistance from a more experienced surgeon.^{106, 107}

Patient characteristics

BMI

Nine studies (with a total of 2,219 patients) analyzed the association of body mass index (BMI) or weight on surgical outcomes.^{86, 92, 94}–^{96, 109, 116}–¹¹⁸ One reported aggregate BMI data stratified by low- versus high-volume hospitals.⁹⁵ Three studies did not report the data for the weight of the patients. Five studies reported no effect of weight or BMI on symptomatic outcomes.^{86, 92, 94, 96, 109} One reported worse heartburn scores for the normal weight group compared with the overweight or obese groups (P=0.001).¹¹⁷ By contrast, another study reported obese patients had more recurrences of symptoms than the normal or overweight patients (P=0.03).¹¹⁸ Three studies reported no effect of BMI on patient satisfaction or the ability to return to work.^{92, 96, 117} Increasing BMI correlated with patient satisfaction in one Swedish study regardless of the hospital's volume of laparoscopic procedures.⁹⁵ Two studies reported that BMI was not associated with global success or failure of surgery.^{96, 116}

Psychological profile

Four studies (with a total of 388 patients) evaluated an association between a history of psychological disorders or certain psychological traits at outcomes of laparoscopic fundoplication. Psychological conditions reported in three studies included major depression, anxiety disorders, sexual abuse, and functional gastrointestinal dysfunction.^{93, 119, 120}

Absence of psychiatric disorders was associated with greater improvement in GERD-HRQL symptom and total score, and patient satisfaction.^{119, 120} Another study reported psychiatric history was significantly associated with surgical failure.⁹³ The study by Kamolz et al.¹²¹ examined the impact of a patient's personality traits on temporary dysphagia and found that a correlation between dysphagia and a patient's “internal locus of control.” For patients with strong internal control (ie, high expectations that they can control their health-related outcomes), there was less dysphagia. This was compared to patients with a predominantly fatalistic external control (a belief that their health outcomes were beyond their control), which was highly correlated with dysphagia.¹²¹ It is unclear if there was overlap of patients for the two studies by Kamolz et al.^{120, 121}

Clinical data

Preoperative response to acid-suppression therapy

Four studies (with a total of 592 patients) examined the effect of the patient's preoperative response to antisecretory medications on outcomes of laparoscopic fundoplication, as measured by patient satisfaction, QOL, and global success or failure.^{45, 86, 89, 93} Two studies reported that patients with a complete or partial response to PPI therapy at baseline had a better symptomatic response.^{45, 86} One study in which 79% of patients had a complete symptomatic response to PPI therapy also reported good outcomes by Visick score.⁸⁹ Similarly, another study reported that non-response to a PPI was predictive of surgical failure.⁹³

Esophagitis

Seven studies (with a total of 1,413 patients) evaluated the relationship between baseline esophagitis and surgical outcomes.^{86, 89, 91}–^{94, 109} Six of the seven found no difference in the rates of treatment failure between those who had esophagitis and those who did not, as measured by differences between groups in GERD symptoms, medication use, or satisfaction. One found slightly higher rates of dysphagia postoperatively in patients with esophagitis preoperatively, but not of other symptoms or outcomes measured.⁹¹

Five studies (with a total of 553 patients) examined the association between the presence of severe esophagitis (defined as circumferential erosions, strictures and Barrett's esophagus) and the outcomes of laparoscopic fundoplication.^{86, 87, 89, 93, 109} Outcomes were considered to be poor in patients who had symptomatic relapse/persistence, acid reflux, or reliance on medical therapy. Four of five studies found no difference in the likelihood of poor outcomes in patients with and without severe esophagitis.^{86, 89, 93, 109} Only one small study found an increased risk of acid reflux in patients with baseline severe esophagitis.⁸⁷

Esophageal pH

Nine studies (with a total of 1,307 patients) analyzed the influence of preoperative esophageal pH exposure on the outcomes of laparoscopic fundoplication.^{45, 86, 87, 89, 92}–^{94, 97, 109} Outcomes considered included relapse/persistence of symptoms, acid reflux, reliance on medical therapy, satisfaction, or a combination of factors. Four studies explicitly reported that pH testing was performed while the patient was off all acid-modifying medications.^{45, 86, 87, 94} Of these, two reported better results in patients with increased preoperative acid reflux;^{86, 94} while one study reported no outcome differences based on preoperative 24-hour pH study results;⁴⁵ and another reported an increased risk of an abnormal postoperative DeMeester score with increasing preoperative DeMeester scores.⁸⁷

LES competence/pressure

Nine studies (with a total of 1,279 patients) examined the risk of surgical failure, including relapse/persistence of symptoms, acid reflux, reliance on medications and a combination of factors, based on LES incompetence (abnormally low pressure and short length at rest) or low LES resting pressure.^{45, 86}–^{88, 93, 94, 97, 116, 122} Eight did not find an increased risk of surgical failure with preoperative LES incompetence or a low resting pressure.^{45, 86, 88, 93, 94, 97, 116, 122} One found that low preoperative LES pressure was a risk factor for increased postoperative acid reflux.⁸⁷ Another study found that preoperative increased LES pressure was associated with increased risk of new-onset postoperative dysphagia.¹⁰⁹ The same study found that LES competence was also a risk factor for postoperative dysphagia.

Esophageal motility

Four studies (with a total of 361 patients) examined various esophageal motility parameters on outcomes of laparoscopic fundoplication.^{87, 89, 90, 96} Two studies (with a total of 156 patients) did not find an increased risk of laparoscopic fundoplication failure with preoperative esophageal body hypomotililty/low amplitude contractions as assessed by symptomatic relapse/persistence or 24-hour pH studies.^{87, 90} Three studies of distal esophageal segment dysmotility (either hyper- or hypomotility) involving 244 patients found no effect on laparoscopic treatment outcomes, including symptomatic relapse/persistence and acid reflux.^{87, 89, 96} One study involving 124 patients found that nonspecific spastic disorders of the esophagus increased the risk of postoperative heartburn, regurgitation, and antireflux medication requirement after laparoscopic fundoplication.⁹⁶ Nonetheless, patients in this study with spastic esophagus still reported improvement equal to those without spastic disorder.

Hiatal hernia

Seven studies (with a total of 1,038 patients) analyzed the influence of a hiatal hernia on surgical outcomes.^{86, 87, 89, 92}–^{94, 109} Six studies examined the effect of hernias >2cm.^{86, 87, 89, 92, 94, 109} None found differences in surgical failure rates. The one study that examined the effect of large hiatal hernias >3 cm reported a significant increase in risk of symptomatic relapse and persistence of symptoms.⁹³

Factors that influenced the outcomes of endoscopic treatments

Four cohort studies examined the association of various patient characteristics with the outcomes of the endoscopic procedures.^{54, 57, 73, 76} No consistent associations were observed.

The study on Enteryx™ by Schumacher et al.⁷³ analyzed age as a predictor for outcomes in 76 patients and reported that age less than 48 years was predictive of a treatment response (defined as a decrease in the PPI dosage) but not for symptom score improvement. Other variables examined included study center, BMI, sex, PPI dose at baseline, baseline esophagitis, investigator experience, and injection volume. None of these factors were associated with any of the outcomes examined.

Another study on Enteryx™ by Cohen et al. found a non-significant trend toward improved response among patients with greater residual implant volume.⁵⁴ No significant associations were found for other variables examined including sex, BMI, baseline hiatal hernia, baseline esophagitis, symptoms while off PPIs, baseline esophageal acid exposure, baseline PPI dose, duration of prior therapy, study center, investigator prior experience, implant volume, and residual implant.

The third study on Enteryx™ had just 15 patients and found no relationship with the primary outcome (PPI dose reduction ≥50% at 3 month) and baseline PPI, duration of prior PPI therapy, baseline GERD-HRQL score, baseline esophageal acid exposure, Savary-Miller grade of esophagitis, or presence of hiatal hernia.⁷⁶

One study on Stretta™ found that age less than 50 was significantly associated with improved esophageal acid exposure, but there was no difference in medication use or GERD symptoms after treatment.⁵⁷ No significant associations were found for sex, GERD symptoms, years of GERD, medication use, baseline esophageal acid exposure, baseline lower esophageal sphincter pressure, esophageal peristaltic amplitude, baseline acid exposure, or baseline esophagitis.

Miscellaneous factors

Four studies examined miscellaneous factors influencing outcomes of laparoscopic surgery.^{98, 123}–¹²⁵ At 1-year follow-up, Kamolz et al. reported that patients with aerophagia at baseline had less improvement in quality of life compared with patients who did not have aerophagia at baseline.¹²⁴ The retrospective review by Haithcock et al. reported higher rate of conversion from a laparoscopic to an open approach in African-American women compared with Caucasian women.¹²³ One study of laparoscopic antireflux surgery performed on 26 patients in a rural county hospital setting reported that 95% of the patients had symptomatic relief from heartburn and required no medication at follow-up ranged from 2 to 21 months (mean, 10 months). Complication and conversion rates were both 7%. The authors suggested that their results were not different from those at large urban hospitals, at least for the length of the follow-up period.¹²⁵ Yau et al. reported that the overall rate of reoperation (for paraesophageal hiatus hernia, tight esophageal hiatus, recurrent reflux, tight or slipped Nissen fundoplication) after laparoscopic fundoplication was not associated with esophageal length. There was a non-significant increased rate of reoperation for a postoperative paraesophageal hernia in patients with the shortest esophageal length index (esophageal length divided by the patient height multiplied by 100) compared with patients who had the longest esophageal length index (8% vs. 2%, P=0.36).⁹⁸ The authors concluded that the increased risk of this problem was small and therefore, a case could not be made for patients with a manometrically short esophagus to routinely undergo an esophageal lengthening procedure. Because of the paucity of studies, no meaningful conclusions can be derived for each of these factors.

Medical treatments (Table 18)

The Cochrane Systematic Review by Donnellan et al. identified 51 RCTs (6,242 participants) that reported adverse events associated with medical treatments for GERD.³⁵ The drugs used were PPIs and H2RAs. PPIs included omeprazole, lansoprazole, pantoprazole, rabeprazole, and esomeprazole. The duration of the trials ranged from 12 to 52 weeks. Table 18 summarizes the adverse events of headache, diarrhea, and abdominal pain reported in this systematic review. This review found that PPIs had more overall adverse events in comparison with H2RAs (cimetidine, famotidine, nizatidine, ranitidine) and with placebo. Only one trial in this review showed a significant difference in headache occurrence between PPIs and H2RAs (Relative Risk 2.27, 95% CI 1.04, 4.97).¹²⁹

Table 18

Adverse events in RCTs comparing PPI to H2RA or placebo, and PPI healing vs. maintenance dose.

The OHSU EPC report identified 26 head-to-head comparisons of PPIs for acute treatment of GERD that reported adverse events.³³ The number of participants was 23,466, and the medications tested were omeprazole, esomeprazole, rabeprazole, and lansoprazole. The rate of adverse events ranged from 1.5 to 55.7%, and the proportion of patient withdrawal secondary to adverse events ranged from 0.9 to 13%. The most common adverse events were headache, diarrhea, nausea, and abdominal pain. These events were reported similarly between drugs. Other events reported were respiratory infection, eructation, flatulence, dizziness, paresthesia, somnolence, and gastritis. Serum gastrin levels were found to be elevated compared to baseline although the magnitude of increase was small and not considered clinically significant.

There were only two RCTs of maintenance treatment of GERD that included participants who completed 5 years of follow-up.^{130, 131} Thjodleiffson et al. reported adverse event data for omeprazole 20 mg, rabeprazole 10 mg and 20 mg.¹³¹ The study recorded 1,086 adverse events. The most commonly reported event was diarrhea: 6.4% in rabeprazole 20 mg, 3.7% in rabeprazole 10 mg, and 4.8% in omeprazole 20 mg. There were 177 treatment-related adverse events. The majority of them were mild in intensity, and only 13% were considered to be severe (details not provided). There was no significant difference in the number of adverse events reported or the number of withdrawals due to adverse events in different PPI groups. Caos et al. compared rabeprazole 10 mg and 20 mg with placebo and reported that 8% of the patients experienced adverse events.¹³⁰ The most common adverse events were rhinitis, diarrhea, flu syndrome, headache, pharyngitis, back pain, and abdominal pain. The proportion of these adverse events was significantly higher in the two treatment arms compared to placebo (P<0.018).

The OHSU EPC report also identified a nested case-control study of 10,008 lansoprazole users who were followed for 4 years. The authors found a trend for diarrhea to be dose related; it was reported in 5%, 3.7%, and 1.5% of patients using 60 mg or more, 30 mg, and 15 mg or less, respectively. The dose of lansoprazole was reduced or discontinued in 42.1% of patients who reported diarrhea.¹³² There were no studies with a follow-up longer than one year for pantoprazole and esomeprazole.

Several studies assessed the association of acid suppression therapies (PPIs and H2RAs) with the development of pneumonia, hypergastrinemia, atrophic gastritis, and vitamin B12 malabsorption. A population-based cohort study comprised 364,683 individuals who subsequently developed 5,551 first occurrences of pneumonia during follow-up.¹¹ Incidence rates of pneumonia in acid-suppressive drug users and non acid-suppressive drug users were 2.45 and 0.6 per 100 person years, respectively. The relative risk of pneumonia among persons currently using PPIs compared with those who stopped was 1.89 (95% CI 1.36–2.62), after adjusting for matching factors, respiratory illness, long-term heart failure, diabetes mellitus, use of antibiotics, and use of immunosuppressants.

Another report included 230 patients whose severe reflux esophagitis had healed after up to 12 weeks of treatment and were followed for up to 11 years.¹³³ Patients older than 65 years of age had twice as many adverse events (including deaths) compared with younger patients. The annual incidence of gastric corpus mucosal atrophy was 4.6% and 0.7% in H. pylori-positive and -negative patients, respectively. It was observed mainly in elderly patients who had moderate/severe gastritis at baseline. Corpus intestinal metaplasia was rare, but no dysplasia or neoplastic changes were observed.

Two studies addressed the effect of acid suppression therapy on gastric atrophy.^{13, 14} Kuipers et al. compared two groups of H. pylori infected patients with chronic GERD.¹³ One group was treated with omeprazole for 5 years, the other underwent antireflux surgery and had follow up for more than 5 years. None of the surgically treated patients developed atrophic gastritis, but 31% of the medically treated patients showed evidence of some degree of atrophic gastritis. However, a subsequent RCT comparing omeprazole to antireflux surgery after 3 years showed no significant differences between the two groups in the development of gastric glandular atrophy or the occurrence of intestinal metaplasia.¹⁴

One case-control study revealed that omeprazole treatment was associated with a 10-fold increased risk of campylobacter infection.¹⁰ Risk for current users (1 month before infection) but not former users suggests that the relation was causal. No relation was seen with H2RA. Another cohort study was identified from a pharmacy database of 1,187 inpatients who received antibiotics for more than 9 months at a Canadian teaching hospital.⁹ C. difficile diarrhea was significantly associated with use of a PPI (adjusted OR 2.1, 95% CI 1.3–3.4). From the same report, but in a case-control design, C. difficile diarrhea also was associated with use of PPI (adjusted OR 2.7, 95% CI 1.4–5.2). A retrospective case-control study showed that PPI use within the preceding 8 weeks was associated with increased risk of C. difficile diarrhea (OR 2.5, 95% CI 1.5–4.2).¹³⁴

Drug Interactions

Table 10 of the OHSU EPC report lists the clinically significant drug interactions with PPIs.³³ All PPIs reduce the absorption of drugs that require an acidic gastric pH for maximal absorption, such as ketoconazole, iron, digoxin, delaviridine, indinivir, and enteric-coated salicylates. Omeprazole interacts with several drugs, specifically carbamazepine, phenytoin, diazepam, methotrexate, and trovafloxacin. So far, no clinically significant drug interactions have been reported, with the exception that there is decreased clearance of theophylline with lansoprazole. Esomeprazole has the potential to interact with cytochrome 2C19 and thereby may cause interactions with diazepam, phenytoin, and warfarin.¹³⁵

Table 10

Medical Treatment for GERD: Relapse of symptoms on treatments.

Drug interactions with cimetidine were not reported in the aforementioned meta-analyses. However, a literature search showed that cimetidine interacts with many drugs: when added to warfarin it increases INR and may be dose dependent.¹³⁶ Cimetidine inhibits cytochrome P-450 and other isoenzymes involved in the metabolism of theophylline, phenytoin, diazepam, caffeine, and other drugs. Interaction with beta-blockers results in significant sinus bradycardia and hypotension.

Surgical treatments (Tables 19, 20, 21)

We identified two recently published systematic reviews that reported on perioperative mortality and morbidity and adverse events associated with specific surgical treatments for GERD.^{137, 138} We also identified one database analysis of major adverse outcomes in antireflux surgery from the Nationwide Inpatient Sample of the Health Care Utilization Project.¹³⁹ In addition, from 106 articles identified through MEDLINE search that met our study inclusion criteria, we identified three RCTs, five prospective cohorts, nine retrospective cohorts and two case-control surgical studies of various fundoplication techniques for GERD in 24 publications that reported adverse event data which were not included in the two systematic reviews (except for one study²⁴ that provided subsequent long-term⁸ follow-up data).^{8, 23}–^{25, 45, 46, 65, 69, 82, 93, 95, 103, 109, 140}–¹⁵⁰ The first systematic review examined outcomes for 41 papers published between 1974 and 2002;¹³⁸ the second review examined outcomes for another 41 papers published between 1993 and 2003.¹³⁷ Because of the overlap between the two reviews, primary studies common to both reviews were labeled clearly in the evidence tables.

Table 22

Summary of medical, surgical and endoscopic treatments.

Table 23

Summary of Comparative Data in Treatments of GERD.

Table 19

Intraoperative complications (and those occurring within 30 days) for surgical and endoscopic procedures.

Table 20

Complications occurring more than 30 days after surgical or endoscopic procedures.

Table 21

Complications occurring after surgical and endoscopic procedures (time period, uncertain).

One RCT²⁴ and one subsequent long-term follow-up report⁸ compared medical with surgical therapy for GERD and provided data for operative and postoperative complications (Spechler 1992²⁴ [included in Carlson 2001¹³⁷]). The RCT from the VA Cooperative Study²⁴ compared Nissen fundoplication to continuous medical therapy or to stepwise medications such as Maalox®, ranitidine, metoclopramide, and sucralfate in 247 patients for up to 2 years. Ten of 82 surgical patients had one or more operative complications including six with splenic injury and four with a gastrointestinal perforation; 12 of 82 had one or more postoperative complications including wound infection, pulmonary embolism, two with ileus lasting more than 5 days, two with bleeding requiring transfusion, and two with abscesses. No complications were fatal, and all resolved without apparent sequelae. “Side effects” occurred in 84% of surgical patients, 88% in those receiving continuous medical therapy, and in 88% of those given stepwise medication for symptoms. Early satiety, inability to belch, and inability to vomit were increased for surgical therapy compared to either medical therapy (P<0.05). A follow-up to this study was published in 2001 - mean duration of follow-up for surgical patients was 9.1 years and, 10.6 years for medically treated patients.⁸ There were no significant differences between groups in the frequency of any symptoms described in the earlier paper: increased abdominal girth (34% surgical versus 36% medical), inability to belch (29% surgical versus 20% medical), and inability to vomit (32% surgical versus 20% medical).

Another RCT compared laparoscopic total fundoplication with anterior partial fundoplication.³⁷ There were 107 patients with available follow-up for 5 years. The two groups did not differ significantly in the rate of adverse events such as dysphagia (27% for total and 18% for partial fundoplication) or diarrhea (27% for total and 24% for partial fundoplication). However, abdominal bloating at 5 years was significantly higher among the total fundoplication patients when compared to partial fundoplication group (75% versus 44%, respectively, P=0.002). Inability to belch at 5 years was also significantly higher among the total fundoplication patients when compared with the partial fundoplication group (43% versus 20%, respectively, P=0.018).

The numbers and rates of adverse events or complications extracted from the reviews and the primary papers are tabulated in the summary tables below.

The meta-analysis of nine RCTs comparing partial versus total wrap and six RCTs comparing open versus laparoscopic fundoplication reported overall morbidity in the trials from 9.4 to 13.1%. The need for conversion arose in 17 cases (7.3%).¹³⁸ The mortality rate for 10,489 laparoscopic procedures (ie, Nissen, Toupet, Nissen-Rossetti) was 0.08%. The causes of death were myocardial infarction, duodenal or gastric perforation, and unspecified. The rate of open conversion as reported in 34 of 41 papers (8,620 cases) was 3.1%.¹³⁷ Four studies of laparoscopic Nissen fundoplication reported a mean conversion rate of 1.6%.^{93, 137, 138, 146}

The most commonly reported complications occurring intraoperatively or within 30 days after open fundoplication were splenectomy, dysphagia, inability to belch, and inability to vomit; while the most commonly reported complications for laparoscopic procedures were gastric or esophageal injury or perforation, splenic injury or splenectomy, pneumothorax, bleeding, pneumonia, fever, wound infections, bloating, and dysphagia. Gastrointestinal injury was reported in six surgical studies and one review, the rate of event ranged from <1% to 3.4%. Pneumothorax was reported in four studies and one review. Bleeding was reported in four studies and one review. Splenic injury was reported in six studies and one review.^{23, 25, 65, 137, 139, 146, 150}

Dysphagia 30 days after the procedure ranging in frequency from 2.5 to 27% was reported in many surgical studies.^{23, 37, 65, 137, 138, 148} Nine other surgical studies also reported high postsurgical dysphagia rates, but the data were not reported with sufficient detail to determine when these adverse events occurred relative to the procedure.^{25, 46, 69, 93, 95, 109, 144, 148, 150} One study reported that 45 out of 142 patients (32%) with either mild or absent bloating preoperatively had an increase in bloating 6 months after surgery (However, patients in whom bloating was moderate preoperatively showed little change and patients with severe bloating preoperatively showed the greatest improvement in bloating at 6 months follow-up.).¹⁴³ One RCT reported 27% dysphagia and 75% bloating 5 years after the procedure.³⁷

Length of stay

Length of stay (LOS) after a primary minimally invasive antireflux procedure often offers a convenient and objective measure of overall morbidity associated with operating technique. Eight papers reported this information and LOS varied greatly from paper to paper and from procedure to procedure.^{45, 93, 103, 137}–^{139, 144, 146}. The meta-analysis by Carlson et al.¹³⁷ reviewed 28 papers from 1993 to 2000 and reported average LOS of 2.8 days and stated that there was no real trend to a decreased length of stay with later publication, as might be expected. The meta-analysis by Catarci et al.¹³⁸ combined the results of six RCTs from 1997 to 2002 and reported a LOS of 3.1 days for laparoscopic Nissen surgery and 5.2 days for open Nissen. Leggett et al.¹⁴⁶ reported 2.7 days for Nissen and 2.3 days for Nissen-Rossetti. In 2003, Anvari et al.⁴⁵ reported 2.8 days for Nissen, but Klaus et al.¹⁰³ in the same year reported 1.5 days for laparoscopic Nissen or Toupet procedures combined.

Learning curve

Seven papers (two prospective and five retrospective studies) explicitly discussed the implications of the learning curve in antireflux procedures.^{46, 104}–^{108, 139} A prospective study from Australia of 280 cases undertaken by 11 surgeons over 46 months demonstrated that the complication, reoperation, and conversion rates (laparoscopic to open surgery) were all higher in the first 50 cases performed by the group of surgeons, and in the first 20 cases by each individual surgeon.¹⁰⁸ Rates were higher in the initial 20 cases and in the initial five individual cases. The early experience was associated with prolonged operating times and technical difficulties because of the need to adjust to new instruments and an altered method of vision. A prospective study from Canada of 100 consecutive cases undertaken by four general surgeons reported a laparoscopic failure rate of 26% (18/68) during a surgeon's first 20 operations and 11% (3/28) thereafter (P<0.09); corresponding conversion rates were 22% and 4% (P<0.05).¹⁰⁶ During a surgeon's first 20 operations, the failure rate rose from 21% (12/67) to 55% 6/11 (P<0.04) if a second surgeon did not assist. The authors concluded that the individual learning curve requires about 20 operations to surmount.

A retrospective chart review of the first 100 laparoscopic fundoplications in Canada was analyzed.¹⁰⁴ Two surgeons performed the majority of the procedures and routinely assisted each other. Patients were grouped chronologically with the first 50 cases defined as early institutional experience and a surgeon's first 20 cases defined as early personal experience. Operative time was longer in both the early institutional and personal experience. The rate of dysphagia requiring intervention was significantly higher during the early institutional experience (22% versus 4%, P=0.017). The rate of dysphagia requiring intervention was also higher with early personal experience than with late personal experience, but the difference did not reach statistical significance (19% versus 8%, P=NS). The conversion rate was 0%, reoperation rate was 1% and the length of stay was 2.5 days. These outcomes were unaffected by a learning curve.

A retrospective study by case-note analysis of 61 patients undergoing laparoscopic fundoplication in England was analyzed.¹⁰⁵ The authors reported a significant decline in conversion rate with time (P=0.002). In the first set of 12 patients, there were seven conversions; in the last subset of 13, there was only one. A review of 241 consecutive patients undergoing laparoscopic fundoplication in a US hospital over 6 years was evaluated.¹⁰⁷ Comparing the first 25 attempts with the second 25 attempts, there were 14 conversions (56%) versus four (16%) (P<0.01), respectively. Operative times decreased significantly and intraoperative complication rates were five (20%) and one (4%), respectively. The authors described their learning curve as very steep.

Endoscopic treatments (Tables 19, 20, 21,)

For endoscopic studies, we identified only three comparative studies: one non-randomized parallel group comparison of laparoscopic fundoplication and the EndoCinch™ device,⁶⁹ one RCT that compared two different configurations of stitching using Endocinch™, ⁶⁸ and another RCT that compared Enteryx™ injection versus sham.⁶ There were a total of 20 studies in 25 publications of various techniques and devices for endoscopic surgery.^{6, 7, 54}–^{57, 59, 60, 62}–^{69, 72}–^{76, 79}–^{81, 151} Finally there were 47 individual adverse events reports submitted by the manufacturers since 2000 to the FDA.

The frequency and rate of adverse events varied widely. The major intraoperative complications reported in the endoscopic studies were pain/discomfort (eg, chest/retrosternal, pharyngeal, gastric, and abdominal) and major bleeding episodes. Ten endoscopic studies in 12 publications reported on postsurgical pain.^{7, 56, 57, 63, 67}–^{69, 74}–^{76, 80, 81} Bleeding was reported in eight studies.^{7, 59, 68, 69, 72, 75, 80, 81} One study of Enteryx™ reported 85% of patients developed retrosternal pain.⁵⁴ A second study of Enteryx™ reported 77%⁷³ of patients developed chest pain and a third study reported 69%⁶ of patients developed retrosternal and epigastric pain. One study on Stretta™ reported chest pain in 1.7% of patients.⁵⁷In addition, short-term dysphagia occurred in eight studies reported in 10 publications.^{54, 56, 57, 63, 67, 72, 73, 76, 79, 80} Five studies in seven publications reported dysphagia persisting more than 30 days after endoscopic procedure (ranged from <1% to 13%).^{54, 55, 57, 62, 66, 67, 73}

A case report published in 2005 described a pneumomediastinum following Enteryx™ injection procedure.¹⁵² Other complications involving the mediastinum after the Enteryx™ procedure had also been described.¹⁵³ A 2004 report described development of an abscess at the gastroesophageal junction in continuity with the esophagus following Enteryx™ injection; the patient in this case died but the actual cause of death could not be determined because this patient had multiple co-morbidities and the family refused an autopsy.¹⁵⁴ A search of the FDA/MAUDE database (Manufacturer and User Facility Device Experience)¹⁶ on May 31, 2005, revealed the following: the manufacturers of Enteryx™ reported 26 adverse events from 2002-2005 including five deaths, six incidents of chest pain, five polymer exudation, three pneumonia, and three dysphagia. The manufacturers of Stretta™ reported 26 adverse events from 2000-2005; 18 occurred in 2000-2001. These included three deaths, four chest pain, three nausea/vomiting, two bouts of abdominal pain, gas, or bloating, two burns, and two pleural effusion.

Summary of results

The following table included information on the data source, population studied, limitations of the included studies, as well as a result summary on symptoms and quality of life, esophagitis healing, esophageal acid exposure, medication use, variables that either do or do not affect outcomes, and adverse event data.