The following table summarizes the main comparative findings for the three treatment modalities. Discussion regarding the findings of our report follows.

The seemingly straightforward definition of GERD proposed in a guideline from the American College of Gastroenterology (ie, “symptoms or mucosal damage caused by the abnormal reflux of gastric contents into the esophagus.¹) belies the complex issues related to its diagnosis and management. Patients with GERD have a wide spectrum of disease severity while caregivers have to choose from many options for evaluating such patients, embarking upon specific therapy and assessing the response, and considering the options for long-term treatment. This report cannot address all of these issues but does shed light upon the comparative efficacy and safety of the available treatment options.

GERD is generally considered a chronic condition that requires long-term treatment. Thus, we emphasized comparisons of the long-term efficacy and safety of medical, surgical, and endoscopic approaches. As a general rule, the surgical and endoscopic approaches are applicable only to patients who are seeking a long-term strategy for avoidance of symptoms and medication dependency.

It is important to stress that the populations evaluated in all of the studies included in this report focused on patients who had GERD that was well established based upon a formal evaluation that included objective testing (albeit the rigor with which the diagnosis was established varied across studies). By contrast, the majority of patients with symptoms suggestive of GERD are treated clinically without such an evaluation. Many such patients may have symptoms that are caused by other upper digestive disorders such as nonulcer dyspepsia. Thus, the results are directly applicable only to patients in whom there is a relatively high-degree of certainty regarding the underlying diagnosis.

Furthermore, the studies of medical therapy used in this report were based mainly upon previously published meta-analyses. The primary controlled trials included in the meta-analyses enrolled almost exclusively patients who had some degree of esophagitis, thus excluding a population of patients with symptoms compatible with GERD but without visible esophagitis (a population sometimes referred to as having “non-erosive reflux disease”). The pathogenesis, natural history and response to therapy in such patients may be different than patients with GERD and esophagitis.^{155, 156} Similarly, the studies of surgical and endoscopic therapy typically enrolled patients with esophagitis or those who were already on antisecretory medications thus making it unclear whether the patients had non-erosive reflux disease at baseline. Thus, the summary of medical therapy described in this report is not directly applicable to patients with non-erosive reflux disease while the surgical and endoscopic literature has not addressed this population explicitly.

There were differences in the spectrum of disease severity in patients included in the various studies. As a general rule, studies of fundoplication included patients with the widest spectrum of disease severity, at least as assessed by the degree of esophagitis, since they included patients with esophagitis ranging in severity from none to Barrett's esophagus and esophageal strictures. Differences in baseline characteristics of patients may influence the results of some of the comparative trials, particularly those comparing surgery to medical therapy since patients undergoing surgery had relatively more severe disease at baseline in some of the comparative trials. Further complicating comparisons were the different classification systems used across studies for describing symptoms, the degree of esophagitis, and variation in how esophageal pH testing was performed and interpreted.

In addition, there are multiple subtle variations in surgical techniques that may influence the outcomes of surgery. The surgical reports did not provide sufficient detail on these technical nuances to understand their implications on clinical endpoints. We considered all of these factors in deriving our conclusions; nevertheless, accounting for all of these differences is sometimes not attainable.

Treatment outcomes were not defined uniformly across studies. Studies described a variety of subjective outcomes (such as changes in symptoms and quality of life) and more objective outcomes (such as healing of esophagitis, use of medications, and changes in esophageal pH exposure) and did not always use the same definitions to describe these endpoints. There was variability in the rigor with which some of these endpoints were defined, particularly those that were relatively subjective. For example, while some studies used a validated measure of quality of life, others used symptoms scales whose measurement properties have not been well studied. Similar differences were notable for more objective endpoints such as esophageal pH exposure and grading of esophagitis. Some studies used 12-hour while others 24-hour pH measurements and defined normalization differently. Some studies defined “normal” as a DeMeester score of less than <14.7 while others used a criterion of esophageal pH less than 4, less than 4.7 percent of the time. Still others did not define normal and simply used pre-treatment and post-treatment difference to report improvement or worsening. Studies reporting the grade of esophagitis also used varied definitions; some used the Savary-Miller grade while others the Los Angeles Classification or provided only macroscopic descriptions without using any classification system.

The maximal duration of follow-up also differed across the three interventions. Most of the primary studies included in the meta-analyses of medical therapy had follow-up duration of approximately one year. By contrast, for surgical therapy, we focused on cohort studies with at least five-years of follow-up while the follow-up on the studies comparing surgical and medical approaches ranged from 1 to 10 years. The endoscopic studies generally had follow-up of 1 year or less with the longest reported follow-up of 27 months.

All studies that were included had important limitations. Many of the surgical studies reported more than 50% of the patients were lost to follow-up at 5 or more years. Some of the uncontrolled studies and non-randomized comparative studies presented data only on evaluable patients (without an intention-to-treat analysis). Because many of the studies were non-randomized or lacked a suitable control group and focused on endpoints that can be subjective or vulnerable to a placebo effect, it was particularly difficult to confidently attribute benefits to the intervention. This was illustrated in a prospective cohort study by Cohen et al.,⁵⁴ which evaluated one of the endoscopic approaches. The authors reported that 67% of patients were off PPIs at 24 months. The clinical impact of this important endpoint is diminished when considering that 41% of the patients in the sham group in the controlled trial by Deviere et al.⁶ were also off PPIs at the end of the study.

We summarized the various objective and subjective outcomes that are considered to be important in the care of patients with GERD, without attempting to define treatment success or failure. We relied on the definitions of endpoints used in the individual studies and did not attempt to adjust results in any way to increase comparability. Thus, whether variations in definitions of endpoints could account for differences in results across studies is unclear.

Two long-term RCTs that directly compared medical with surgical treatments qualified for inclusion in this report. One reported that there was no difference in heartburn symptoms between the medical and surgical patients at pre-defined time points through 5 years. The other reported significantly improved GERD symptoms in the surgical arm compared to the medical arm, after both groups discontinued all antireflux medications during the week of assessment. However, the difference was no longer significant when the assessment was made while both groups were kept on their usual antireflux medications.

Neither of the two long-term RCTs on medical versus surgical treatments demonstrated superiority of one treatment modality over the other in healing of esophagitis. Whether any of the endoscopic procedures has a favorable impact on patients who have esophagitis is also unclear. Two uncontrolled studies of Enteryx™ showed a worsening of esophagitis in approximately one-third of patients. Whether this represents chance, ineffectiveness of the procedure, or an adverse effect of treatment is uncertain.

Esophageal acid exposure improved to a greater extent in surgical arms compared to medical arms in two RCTs, at least for up to 1-year of follow-up. Uncontrolled studies of all the endoscopic procedures suggested improvement or normalization in pH in some patients, but there are insufficient data to determine the magnitude of improvement relative to one another, or correlation of pH changes with other outcome measures. Normalization of esophageal pH exposure was observed in 30 to 50% of patients in uncontrolled endoscopic studies, which is similar to what has been reported with medical therapy and inferior to what has been described with fundoplication (at least with short-term follow-up).

In most surgical cohort studies, 90% of patients were off antireflux medications at 5-year follow-up. However, in one of the medical versus surgical RCTs, approximately two-thirds of the surgical patients were still dependent on some form of antireflux medications at 10 years follow-up.⁸ The reasons for these different outcomes are unclear. But, it is important to note that some patients who continued to take antisecretory medications following surgery do not have objective evidence of GERD.¹⁵⁷–¹⁵⁹ Another possible contributing factor may have been that the patients in the study by Spechler et al.⁸ had relatively more severe disease at baseline. However, the Spechler study had the longest follow-up, and the high dropout rates in many of the surgical cohort studies did not permit a confident appraisal of the actual proportion of patients who were off all antisecretory medications on an intention-to-treat basis. The proportion of patients who did not require any antisecretory medications was reported infrequently in endoscopic studies but was in the range of only 25 to 40% in studies in which it was described, a proportion similar to the control rate observed in one of the sham-controlled studies.

Improvement in outcomes relevant to patients (such as symptoms, quality of life, and need for medications) would ideally correlate well with objective measures such as normalization of esophageal pH exposure and healing of esophagitis. The various studies included in this report underscore that these objectives do not always occur concordantly. Furthermore, the degree to which certain objectives should be achieved (such as normalization of pH or complete healing of esophagitis) has not been established.

Proponents of surgery and endoscopic approaches have pointed to a population of patients with “medically refractory” GERD who might benefit from an alternate approach. However, consensus has not been achieved on the definition of “medically refractory” and thus there was corresponding variability in the surgical and endoscopic studies that enrolled patients who “failed” medical treatment. Thus, we could not clearly identify criteria that defined a population of patients with an inadequate response to medical therapy who are likely to respond to surgery or endoscopic therapy. By contrast, we did find evidence that surgical outcomes were better in patients who had responded symptomatically to medical therapy. Thus, there is need for further clarification of the population with medically refractory GERD that might benefit from fundoplication or endoscopic therapy.

Multiple studies have compared one PPI to another. While some differences have been reported, the magnitude of differences has been small and of uncertain clinical importance. Furthermore, most of the comparative trials evaluated doses of the various PPIs that have been approved by the United States Food and Drug Administration (FDA). The doses of various PPIs that would be considered pharmacologically equivalent have not been well established and thus it is possible that all the various PPIs could be equivalent when administered at certain doses. Clinicians often increase the dose of a PPI in patients who do not have an adequate clinical response and do not necessarily rely upon FDA-approved doses and dosing intervals.

We sought and critically evaluated studies that attempted to correlate the outcomes of treatment with baseline patient or treatment-related characteristics. Such associations could be important for guiding patients toward specific options. However, comparison of these studies was difficult because of the heterogeneous study designs and because the reporting of data was often incomplete. Most of the data concerning open and laparoscopic surgery were derived from observational studies of single-center convenience samples. The studies were generally few in number and did not have many subjects. Some studies reported qualitative results without giving actual baseline or post-surgical data. In addition, a diverse range of factors and outcomes were analyzed. Often there was no study or only one study reporting on the association of a specific factor with a specific outcome. Several baseline patient-characteristics were associated with treatment outcomes but the strength of these associations was unclear. Relapse with maintenance medical therapy appears to be more likely in patients with severe esophagitis, younger age, regurgitation, and LES incompetence at baseline.

The quality of reporting of adverse events was incomplete and inconsistent across studies. Some studies did not report specific adverse events (even though they were likely to have occurred) and the definitions of adverse events differed across studies. Furthermore, studies differed in how they classified adverse events as being temporary or persistent or even when they had occurred and none used an acceptable standard or scale for defining their severity. We also restricted our review of adverse events to treatments of GERD and we did not examine adverse event data from studies of PPI use in a non-GERD setting.

Despite a large body of evidence (and an even larger clinical experience) with medical therapy, there are relatively few studies that have systematically sought side effects of long-term medical therapy with PPIs. While this class of drugs has proven to be generally safe in the short-term, concerns have been identified with regard to long-term safety. Also, the potential need for life-long therapy raises the possibility that additional concerns might arise with longer experience with these drugs.

Published experience with all of the approved endoscopic procedures has been outpaced by clinical experience. The manufacturer of the Stretta™ procedure, for example, reports in their marketing media that more than 7000 procedures have been performed worldwide. There is little information regarding patients treated outside of the studies described in this report. An exception is post-marketing adverse events, which are reported voluntarily by manufacturers to the Food and Drug Administration. It should be noted that one of the devices (Enteryx™) was voluntarily removed from the market due to safety concerns that were not fully appreciated during studies that led to its approval.

In conclusion, PPIs and fundoplication are similarly effective in relieving symptoms and improving quality of life in patients with GERD that has been established based on objective testing. Whether medical therapy or surgery is more effective in preventing long-term complications from GERD remains unclear, however. The benefits of surgery must be balanced against its short- and long-term risks, particularly since some of the long-term side effects may be permanent. Experience of the surgeon may also weigh into the decision regarding surgery, although there were limited data from which to explicitly understand the relationship between the surgeon's experience and long-term outcomes or some of the technical nuances of the surgical approach that might have a bearing on surgical outcomes.

The quality, quantity, and consistency of studies on the endoscopic approaches to treatment of GERD are inferior to those of medical or surgical therapy, which can be expected since endoscopic approaches are new developments and data are evolving. At present, their efficacy compared with continued (or intensified) medical therapy is unclear. Sham controlled trials have demonstrated that some of the benefits of these procedures observed in the uncontrolled trials may not be directly attributable to the interventions, thus underscoring the need for additional sham-controlled trials. Although these devices are already commercially available, their long-term efficacy and safety have not yet been established.