Cheryl A. Flynn, MD, MS

Gastroesophageal reflux disease (GERD) is defined as symptoms or tissue damage that results from the abnormal reflux of gastric contents into the esophagus. A systematic review of population-based studies estimates that heartburn or regurgitation symptoms occur in 21% to 59% of the population during a given year.¹ The frequency of GERD in specific populations is provided in Table 1. Although only 1 in 5 patients with upper intestinal symptoms that occur at least weekly seeks medical attention,² nearly 1% of all visits to a family physician’s office are for GERD or related conditions.³

GERD significantly affects the quality of patients’ lives. In a survey of patients presenting for upper endoscopy with symptoms of at least 3 months’ duration, those with a diagnosis of GERD reported low scores at baseline for general wellbeing. Fortunately, follow-up data reported 4 weeks after treatment note improvement in gastrointestinal symptoms, general well-being, general health, vitality, and depression.⁴

Pathophysiology

The pathogenesis of GERD is multifactorial and is thought to involve lower than normal esophageal sphincter pressures. This allows gastric acidic content to reflux into the distal esophagus, which lacks a protective barrier, causing esophagitis. Inflamed tissue impairs the normal clearance of acid, worsening the esophagitis, which inhibits normal motility.

Although Helicobacter pylori is clearly associated with peptic ulcer disease, its association with GERD is still debated. Data from case control studies actually suggest an inverse association; that is, that the presence of H pylori may be protective against the development of GERD.⁵

Because of the anatomical location of the esophagus, GERD should be considered in the differential diagnosis for presenting complaints other than regurgitation or dyspepsia. For example, approximately 50% of patients with chest pain in whom cardiac etiology is ruled out will ultimately be given a diagnosis of GERD.⁶ Similarly, 10% of patients with chronic cough⁷ and 78% of patients with laryngitis have GERD.⁸ Clear associations between GERD and asthma have been demonstrated, but data from meta-analyses fail to show improvement of asthma symptoms when GERD is appropriately treated.⁹

Diagnosis

The diagnosis of GERD can usually be made without the use of invasive tests. The accuracy of key tests, including clinical history, is outlined in Table 2. One study of patients presenting with dyspepsia (signs and symptoms referable to the upper gastrointestinal tract) found that 56% also have GERD.¹⁰ Another showed that 60% of patients referred for pH monitoring had GERD.¹¹ Since reasonable prevalence estimates for GERD in the family practice setting may be slightly lower, calculations in Table 2 assume that between 40% and 60% of patients with suspected GERD actually have the condition.

Individual symptoms of GERD such as heartburn, regurgitation, belching, or dyspepsia are of limited usefulness in diagnosis. In a survey of patients referred for pH monitoring likelihood ratios hovered near 1, meaning that the presence or absence of the symptom had little impact on the diagnosis.¹¹ The clinician’s overall impression that a patient has GERD, however, is much more useful to rule in disease than any individual symptoms.¹¹ Assuming that half of patients with suspected GERD have the disease, if a clinician suspects GERD, that probability increases to 77%. Most clinicians would find a trial of empiric therapy appropriate at that probability.

The omeprazole test is also helpful in confirming a diagnosis of GERD. It consists of the patient taking 40 mg omeprazole in the morning and 20 mg at night for 1 week. If the symptoms resolve, the test is considered diagnostic of GERD.¹² Some consider this approach to be therapeutic, as well as diagnostic. Beginning with an omeprazole test and reserving invasive testing for those not responding to the medication was cost-effective for patients with noncardiac chest pain.¹³ In a decision analysis, empiric treatment with omeprazole was a cost-effective approach to the management of GERD.¹⁴ Of course, initial endoscopy is indicated for patients with “red flags”: signs and symptoms consistent with obstruction, bleeding, or perforation, and those older than 50 years who are at a higher risk of malignancy.

Upper endoscopy, however, is not very accurate in diagnosing GERD. Among patients with GERD, only 22% have esophageal erythema, and only 48% have erosions or ulcerations. Therefore, because of costs and limited resources, the American Society for Gastrointestinal Endoscopy recommends that endoscopy be reserved for patients presenting with possible GERD who also have symptoms of more serious disease (dysphagia, weight loss, gastrointestinal bleeding) and for those not responding to a reasonable trial of therapy.¹⁵ The goal is to rule out more serious conditions.

Twenty-four–hour pH monitoring, while more accurate than endoscopy, is also reserved as a second-line test. According to the American Gastro-enterological Association guidelines, pH recording is indicated when endoscopy is normal and reflux symptoms persist despite acid suppression therapy or to evaluate extra-esophageal symptoms that may be GERD (ie, atypical chest pain or chronic cough).¹⁶ The goal here is to rule in GERD as the etiology of the patient’s symptoms.

Numerous other tests have been proposed to evaluate the patient with suspected GERD: manometry, scintigraphy, esophograms (as part of upper gastrointestinal series), and the Bernstein test (a provocative challenge with hydrochloric acid infusion). Data on the accuracy of these tests are limited by the lack of an accepted reference standard, and most have fallen out of favor because of their inconvenience or limited accuracy. None are more helpful than pH monitoring or endoscopy, and they are therefore not recommended.

Treatment

Pharmacologic

Treatments for GERD address the pathophysiology of the disease; they tend to target the removal of precipitating factors, using gravity or medications to improve acid clearance, lower the acidity of gastric contents, or improve the esophageal sphincter tone. Short-term treatment goals are to relieve symptoms and heal esophagitis; long-term goals are to prevent relapses and sequelae, such as esophageal stricture formation or adenocarcinoma.

Recommendations for lifestyle modifications to lessen GERD symptoms are extensive and mostly based on pathophysiologic data. Possible triggers for GERD include obesity, tight clothing, fatty foods, alcohol, tobacco, caffeine, onions, peppermint, and chocolate.^17-21 Because symptoms are typically worse at night, elevating the head of the bed or avoiding postprandial recumbency is recommended.^22,23 Eliminating these factors has been shown to decrease the amount of acid in the distal esophagus or improve the pressure readings on manometry. This is disease-oriented evidence; it does not necessarily translate into reduced symptoms for patients. The only patient-oriented evidence that supports lifestyle modification is a small crossover trial in which subjects had less heartburn and belching following a hamburger meal compared with the same meal with onions on the burger.¹⁸ Despite the lack of high-quality evidence of effectiveness, guidelines recommend lifestyle modification as the first-line therapy or adjunct to additional medication.^24,25 Because these lifestyle changes are not thought to be harmful and may have other possible health benefits it is reasonable to recommend them to patients until better studies are published.

Promotility agents are a possible treatment of GERD because they improve acid clearance from the esophagus. Older agents such as metoclopramide and bethanecol have little data to support their use^26,27 and are further limited by their central nervous system side effects. A meta-analysis of trial data found that cisapride heals esophagitis and decreases symptoms of GERD²⁸ but was taken off the market because of problems with cardiac arrhythmias.

Acid suppression strategies available over-the-counter include antacids, alginates, and H2-blockers (H2Bs). Small, short-term trials have demonstrated mixed results with regard to symptom relief and lessened acid exposure from both antacids and alginates.^29,30 Longer follow-up in cohort studies suggests a modest benefit from antacids.³¹ More than 24 randomized controlled trials (RCTs) have demonstrated the benefits of H2Bs in the treatment of GERD for both healing esophagitis and symptom relief. The number needed to treat (NNT) is 5; that is, for every 5 GERD patients treated with H2Bs instead of placebo, 1 patient benefits.³² H2Bs also prevent the long-term recurrence of symptoms (NNT =15).³³ There is no clear benefit of one H2B agent over another. Proton pump inhibitors (PPIs) are also well supported by meta-analyses of RCTs in the short term (NNT = 2) and long term (NNT = 3) treatment of GERD.³³ As with H2Bs, no clear advantage is found between different PPI medications.

STEPS. Using the STEPS approach (Safety, Tolerability, Efficacy, Price, Simplicity) to compare H2Bs and PPIs is one way to help guide management of GERD. H2Bs are not associated with any serious long-term complications. PPIs are theoretically linked to malignancy through 2 mechanisms: proliferation of endocrine cells leading to endocrine neoplasia and atrophic gastritis caused by chronic acid suppression as a precursor to gastric adenocarcinoma. However, studies have not borne out these concerns.³⁴ Thus, both H2Bs and PPIs seem equally safe. The best measure of tolerability is the pooled dropout rate, which is the number of patients dropping out of a study for any reason. Dropout rates are about the same for H2Bs and PPIs.³⁵ Meta-analyses of trials comparing the efficacy of H2Bs to PPIs in the short- and long-term treatment of GERD consistently favor PPIs.^33,35 Of 100 patients with GERD, approximately 25 will benefit from treatment with an H2B, and 75 to 80 will benefit from treatment with a PPI.³³ An interesting comparison of trials suggests that the therapeutic gain of PPIs is greatest in those patients with more severe GERD.³⁶ Little data exist about the prevention of complications of GERD with acid suppression. However, 1 small trial found a decreased rate of esophageal stricture recurrence at 1 year in patients with reflux esophagitis (NNT =7) treated with 30-mg lansoprazole compared with those treated with 300-mg ranitidine.³⁷ These data seem to suggest that high-risk patients are more apt to benefit from PPIs than H2Bs.

Because H2Bs are now available in generic forms, price tips the scales greatly in their favor. The average cost for PPIs is nearly 10 times that of generic H2Bs (3020-mg omeprazole tablets cost $105.55 and 60 150-mg ranitidine tablets cost $10.98 as of 11/0/00). Simplicity is a toss-up: most PPIs are dosed once daily, while H2Bs are given once or twice daily. A general approach is to start patients with GERD on a reasonable dose of H2Bs, and move on to PPIs if symptoms do not resolve in 2 to 4 weeks.^25,54 Some insurance companies require documentation of H2B failure before covering the increased costs of PPIs, despite their efficacy advantage.

Surgery

For patients with chronic or recurrent GERD, surgery offers an additional treatment option. Prospective cohort studies note that open Nissen fundoplication produces an 80% to 93% success rate at 10-year follow-up. Laproscopic procedures are producing similar short-term results, but long-term data are pending.³⁸ One randomized trial demonstrated equivalent 3-year outcomes for those undergoing Nissen fundoplication and those taking 40 mg omeprazole daily.³⁹ Decision analysis data favor the cost effectiveness of open Nissen⁴⁰ if medical treatment will be required for more than 4 years, and laproscopic procedure if medical treatment will be needed for more than 10 years.⁴¹

Prognosis

GERD may be a short-term intermittent problem or may be severe and chronic in nature. Untreated, approximately 15% of patients will have symptom relief.³² Antacids can raise that rate to an estimated 20%,^31,25 while H2Bs are associated with symptom-free rates of approximately 25%.³⁰ The best outcomes are found with PPIs, where recurrence of GERD symptoms is suppressed in 75% to 82% of patients at 1-year follow-up.^33,35

Although many patients will experience recurrences, chronic medications may not be necessary. In 1 study, 677 patients with heartburn and mild esophagitis were randomized to treatment with omeprazole or ranitidine for 2 weeks. If symptoms persisted, the dose of medication was doubled. If symptoms resolved, medication was stopped. Recurrences were treated for 2 to 4 weeks at the previously effective dose. Nearly half the patients were successfully treated with intermittent medication, and nearly 40% of initial responders required no further treatment.⁴²

GERD is associated with esophageal strictures, Barrett esophagus (metaplasia of the distal esophageal columnar cells, thought to be a precursor of dysplasia and cancer) and adenocarcinoma. Limited data are available for the actual incidence of stricture in GERD patients. One retrospective cohort study of patients discharged from veterans’ administration hospitals found that 8.4% of patients with GERD had strictures, and the association between esophageal ulcers and stricture was significant.⁴³ This study likely suffered from significant selection bias, however, and the rate in primary care practice is almost certainly much lower. Barrett was noted in 11.6% of 662 patients with GERD referred from general practice settings for endoscopy.⁴⁴ Again, though, patients with GERD referred for endoscopy are likely to have more severe disease, and a recent meta-analysis suggests that the actual risk of Barrett in unselected patients is closer to 3% to 4%.⁴⁵ A longer duration of symptoms was associated with an increase risk. Patients with Barrett esophagus; negative, low-grade, or indefinite dysplasia; and neither aneuploidy or increased 4N on flow cytometry are at very low risk of esophageal cancer (<2% over 5 years). Only approximately 4% of patients with Barrett go on to develop esophageal cancer.⁴⁶

In a well-conducted case control study in Sweden, reflux symptoms were associated with a 7- to 10-fold increase in the risk of esophageal adenocarcinoma. A dose-response risk was noted for symptom frequency, severity, and duration.⁴⁷ However, because adenocarcinoma of the esophagus is so rare, the authors note that a family physician would need to perform endoscopy on more than 1400 men older than 40 years who have severe GERD symptoms to identify 1 case of cancer. Further, there are no data to suggest that treating GERD will reduce the likelihood of these more serious sequelae.

Gastroesophageal reflux disease (GERD) is defined as symptoms or tissue damage that results from the abnormal reflux of gastric contents into the esophagus. A systematic review of population-based studies estimates that heartburn or regurgitation symptoms occur in 21% to 59% of the population during a given year.¹ The frequency of GERD in specific populations is provided in Table 1. Although only 1 in 5 patients with upper intestinal symptoms that occur at least weekly seeks medical attention,² nearly 1% of all visits to a family physician’s office are for GERD or related conditions.³

GERD significantly affects the quality of patients’ lives. In a survey of patients presenting for upper endoscopy with symptoms of at least 3 months’ duration, those with a diagnosis of GERD reported low scores at baseline for general wellbeing. Fortunately, follow-up data reported 4 weeks after treatment note improvement in gastrointestinal symptoms, general well-being, general health, vitality, and depression.⁴

Pathophysiology

The pathogenesis of GERD is multifactorial and is thought to involve lower than normal esophageal sphincter pressures. This allows gastric acidic content to reflux into the distal esophagus, which lacks a protective barrier, causing esophagitis. Inflamed tissue impairs the normal clearance of acid, worsening the esophagitis, which inhibits normal motility.

Although Helicobacter pylori is clearly associated with peptic ulcer disease, its association with GERD is still debated. Data from case control studies actually suggest an inverse association; that is, that the presence of H pylori may be protective against the development of GERD.⁵

Because of the anatomical location of the esophagus, GERD should be considered in the differential diagnosis for presenting complaints other than regurgitation or dyspepsia. For example, approximately 50% of patients with chest pain in whom cardiac etiology is ruled out will ultimately be given a diagnosis of GERD.⁶ Similarly, 10% of patients with chronic cough⁷ and 78% of patients with laryngitis have GERD.⁸ Clear associations between GERD and asthma have been demonstrated, but data from meta-analyses fail to show improvement of asthma symptoms when GERD is appropriately treated.⁹

Diagnosis

The diagnosis of GERD can usually be made without the use of invasive tests. The accuracy of key tests, including clinical history, is outlined in Table 2. One study of patients presenting with dyspepsia (signs and symptoms referable to the upper gastrointestinal tract) found that 56% also have GERD.¹⁰ Another showed that 60% of patients referred for pH monitoring had GERD.¹¹ Since reasonable prevalence estimates for GERD in the family practice setting may be slightly lower, calculations in Table 2 assume that between 40% and 60% of patients with suspected GERD actually have the condition.

Individual symptoms of GERD such as heartburn, regurgitation, belching, or dyspepsia are of limited usefulness in diagnosis. In a survey of patients referred for pH monitoring likelihood ratios hovered near 1, meaning that the presence or absence of the symptom had little impact on the diagnosis.¹¹ The clinician’s overall impression that a patient has GERD, however, is much more useful to rule in disease than any individual symptoms.¹¹ Assuming that half of patients with suspected GERD have the disease, if a clinician suspects GERD, that probability increases to 77%. Most clinicians would find a trial of empiric therapy appropriate at that probability.

The omeprazole test is also helpful in confirming a diagnosis of GERD. It consists of the patient taking 40 mg omeprazole in the morning and 20 mg at night for 1 week. If the symptoms resolve, the test is considered diagnostic of GERD.¹² Some consider this approach to be therapeutic, as well as diagnostic. Beginning with an omeprazole test and reserving invasive testing for those not responding to the medication was cost-effective for patients with noncardiac chest pain.¹³ In a decision analysis, empiric treatment with omeprazole was a cost-effective approach to the management of GERD.¹⁴ Of course, initial endoscopy is indicated for patients with “red flags”: signs and symptoms consistent with obstruction, bleeding, or perforation, and those older than 50 years who are at a higher risk of malignancy.

Upper endoscopy, however, is not very accurate in diagnosing GERD. Among patients with GERD, only 22% have esophageal erythema, and only 48% have erosions or ulcerations. Therefore, because of costs and limited resources, the American Society for Gastrointestinal Endoscopy recommends that endoscopy be reserved for patients presenting with possible GERD who also have symptoms of more serious disease (dysphagia, weight loss, gastrointestinal bleeding) and for those not responding to a reasonable trial of therapy.¹⁵ The goal is to rule out more serious conditions.

Twenty-four–hour pH monitoring, while more accurate than endoscopy, is also reserved as a second-line test. According to the American Gastro-enterological Association guidelines, pH recording is indicated when endoscopy is normal and reflux symptoms persist despite acid suppression therapy or to evaluate extra-esophageal symptoms that may be GERD (ie, atypical chest pain or chronic cough).¹⁶ The goal here is to rule in GERD as the etiology of the patient’s symptoms.

Numerous other tests have been proposed to evaluate the patient with suspected GERD: manometry, scintigraphy, esophograms (as part of upper gastrointestinal series), and the Bernstein test (a provocative challenge with hydrochloric acid infusion). Data on the accuracy of these tests are limited by the lack of an accepted reference standard, and most have fallen out of favor because of their inconvenience or limited accuracy. None are more helpful than pH monitoring or endoscopy, and they are therefore not recommended.

Treatment

Pharmacologic

Treatments for GERD address the pathophysiology of the disease; they tend to target the removal of precipitating factors, using gravity or medications to improve acid clearance, lower the acidity of gastric contents, or improve the esophageal sphincter tone. Short-term treatment goals are to relieve symptoms and heal esophagitis; long-term goals are to prevent relapses and sequelae, such as esophageal stricture formation or adenocarcinoma.

Recommendations for lifestyle modifications to lessen GERD symptoms are extensive and mostly based on pathophysiologic data. Possible triggers for GERD include obesity, tight clothing, fatty foods, alcohol, tobacco, caffeine, onions, peppermint, and chocolate.^17-21 Because symptoms are typically worse at night, elevating the head of the bed or avoiding postprandial recumbency is recommended.^22,23 Eliminating these factors has been shown to decrease the amount of acid in the distal esophagus or improve the pressure readings on manometry. This is disease-oriented evidence; it does not necessarily translate into reduced symptoms for patients. The only patient-oriented evidence that supports lifestyle modification is a small crossover trial in which subjects had less heartburn and belching following a hamburger meal compared with the same meal with onions on the burger.¹⁸ Despite the lack of high-quality evidence of effectiveness, guidelines recommend lifestyle modification as the first-line therapy or adjunct to additional medication.^24,25 Because these lifestyle changes are not thought to be harmful and may have other possible health benefits it is reasonable to recommend them to patients until better studies are published.

Promotility agents are a possible treatment of GERD because they improve acid clearance from the esophagus. Older agents such as metoclopramide and bethanecol have little data to support their use^26,27 and are further limited by their central nervous system side effects. A meta-analysis of trial data found that cisapride heals esophagitis and decreases symptoms of GERD²⁸ but was taken off the market because of problems with cardiac arrhythmias.

Acid suppression strategies available over-the-counter include antacids, alginates, and H2-blockers (H2Bs). Small, short-term trials have demonstrated mixed results with regard to symptom relief and lessened acid exposure from both antacids and alginates.^29,30 Longer follow-up in cohort studies suggests a modest benefit from antacids.³¹ More than 24 randomized controlled trials (RCTs) have demonstrated the benefits of H2Bs in the treatment of GERD for both healing esophagitis and symptom relief. The number needed to treat (NNT) is 5; that is, for every 5 GERD patients treated with H2Bs instead of placebo, 1 patient benefits.³² H2Bs also prevent the long-term recurrence of symptoms (NNT =15).³³ There is no clear benefit of one H2B agent over another. Proton pump inhibitors (PPIs) are also well supported by meta-analyses of RCTs in the short term (NNT = 2) and long term (NNT = 3) treatment of GERD.³³ As with H2Bs, no clear advantage is found between different PPI medications.

STEPS. Using the STEPS approach (Safety, Tolerability, Efficacy, Price, Simplicity) to compare H2Bs and PPIs is one way to help guide management of GERD. H2Bs are not associated with any serious long-term complications. PPIs are theoretically linked to malignancy through 2 mechanisms: proliferation of endocrine cells leading to endocrine neoplasia and atrophic gastritis caused by chronic acid suppression as a precursor to gastric adenocarcinoma. However, studies have not borne out these concerns.³⁴ Thus, both H2Bs and PPIs seem equally safe. The best measure of tolerability is the pooled dropout rate, which is the number of patients dropping out of a study for any reason. Dropout rates are about the same for H2Bs and PPIs.³⁵ Meta-analyses of trials comparing the efficacy of H2Bs to PPIs in the short- and long-term treatment of GERD consistently favor PPIs.^33,35 Of 100 patients with GERD, approximately 25 will benefit from treatment with an H2B, and 75 to 80 will benefit from treatment with a PPI.³³ An interesting comparison of trials suggests that the therapeutic gain of PPIs is greatest in those patients with more severe GERD.³⁶ Little data exist about the prevention of complications of GERD with acid suppression. However, 1 small trial found a decreased rate of esophageal stricture recurrence at 1 year in patients with reflux esophagitis (NNT =7) treated with 30-mg lansoprazole compared with those treated with 300-mg ranitidine.³⁷ These data seem to suggest that high-risk patients are more apt to benefit from PPIs than H2Bs.

Because H2Bs are now available in generic forms, price tips the scales greatly in their favor. The average cost for PPIs is nearly 10 times that of generic H2Bs (3020-mg omeprazole tablets cost $105.55 and 60 150-mg ranitidine tablets cost $10.98 as of 11/0/00). Simplicity is a toss-up: most PPIs are dosed once daily, while H2Bs are given once or twice daily. A general approach is to start patients with GERD on a reasonable dose of H2Bs, and move on to PPIs if symptoms do not resolve in 2 to 4 weeks.^25,54 Some insurance companies require documentation of H2B failure before covering the increased costs of PPIs, despite their efficacy advantage.

Surgery

For patients with chronic or recurrent GERD, surgery offers an additional treatment option. Prospective cohort studies note that open Nissen fundoplication produces an 80% to 93% success rate at 10-year follow-up. Laproscopic procedures are producing similar short-term results, but long-term data are pending.³⁸ One randomized trial demonstrated equivalent 3-year outcomes for those undergoing Nissen fundoplication and those taking 40 mg omeprazole daily.³⁹ Decision analysis data favor the cost effectiveness of open Nissen⁴⁰ if medical treatment will be required for more than 4 years, and laproscopic procedure if medical treatment will be needed for more than 10 years.⁴¹

Prognosis

GERD may be a short-term intermittent problem or may be severe and chronic in nature. Untreated, approximately 15% of patients will have symptom relief.³² Antacids can raise that rate to an estimated 20%,^31,25 while H2Bs are associated with symptom-free rates of approximately 25%.³⁰ The best outcomes are found with PPIs, where recurrence of GERD symptoms is suppressed in 75% to 82% of patients at 1-year follow-up.^33,35

Although many patients will experience recurrences, chronic medications may not be necessary. In 1 study, 677 patients with heartburn and mild esophagitis were randomized to treatment with omeprazole or ranitidine for 2 weeks. If symptoms persisted, the dose of medication was doubled. If symptoms resolved, medication was stopped. Recurrences were treated for 2 to 4 weeks at the previously effective dose. Nearly half the patients were successfully treated with intermittent medication, and nearly 40% of initial responders required no further treatment.⁴²

GERD is associated with esophageal strictures, Barrett esophagus (metaplasia of the distal esophageal columnar cells, thought to be a precursor of dysplasia and cancer) and adenocarcinoma. Limited data are available for the actual incidence of stricture in GERD patients. One retrospective cohort study of patients discharged from veterans’ administration hospitals found that 8.4% of patients with GERD had strictures, and the association between esophageal ulcers and stricture was significant.⁴³ This study likely suffered from significant selection bias, however, and the rate in primary care practice is almost certainly much lower. Barrett was noted in 11.6% of 662 patients with GERD referred from general practice settings for endoscopy.⁴⁴ Again, though, patients with GERD referred for endoscopy are likely to have more severe disease, and a recent meta-analysis suggests that the actual risk of Barrett in unselected patients is closer to 3% to 4%.⁴⁵ A longer duration of symptoms was associated with an increase risk. Patients with Barrett esophagus; negative, low-grade, or indefinite dysplasia; and neither aneuploidy or increased 4N on flow cytometry are at very low risk of esophageal cancer (<2% over 5 years). Only approximately 4% of patients with Barrett go on to develop esophageal cancer.⁴⁶

In a well-conducted case control study in Sweden, reflux symptoms were associated with a 7- to 10-fold increase in the risk of esophageal adenocarcinoma. A dose-response risk was noted for symptom frequency, severity, and duration.⁴⁷ However, because adenocarcinoma of the esophagus is so rare, the authors note that a family physician would need to perform endoscopy on more than 1400 men older than 40 years who have severe GERD symptoms to identify 1 case of cancer. Further, there are no data to suggest that treating GERD will reduce the likelihood of these more serious sequelae.

Gastroesophageal reflux disease (GERD) is defined as symptoms or tissue damage that results from the abnormal reflux of gastric contents into the esophagus. A systematic review of population-based studies estimates that heartburn or regurgitation symptoms occur in 21% to 59% of the population during a given year.¹ The frequency of GERD in specific populations is provided in Table 1. Although only 1 in 5 patients with upper intestinal symptoms that occur at least weekly seeks medical attention,² nearly 1% of all visits to a family physician’s office are for GERD or related conditions.³

GERD significantly affects the quality of patients’ lives. In a survey of patients presenting for upper endoscopy with symptoms of at least 3 months’ duration, those with a diagnosis of GERD reported low scores at baseline for general wellbeing. Fortunately, follow-up data reported 4 weeks after treatment note improvement in gastrointestinal symptoms, general well-being, general health, vitality, and depression.⁴

Pathophysiology

The pathogenesis of GERD is multifactorial and is thought to involve lower than normal esophageal sphincter pressures. This allows gastric acidic content to reflux into the distal esophagus, which lacks a protective barrier, causing esophagitis. Inflamed tissue impairs the normal clearance of acid, worsening the esophagitis, which inhibits normal motility.

Although Helicobacter pylori is clearly associated with peptic ulcer disease, its association with GERD is still debated. Data from case control studies actually suggest an inverse association; that is, that the presence of H pylori may be protective against the development of GERD.⁵

Because of the anatomical location of the esophagus, GERD should be considered in the differential diagnosis for presenting complaints other than regurgitation or dyspepsia. For example, approximately 50% of patients with chest pain in whom cardiac etiology is ruled out will ultimately be given a diagnosis of GERD.⁶ Similarly, 10% of patients with chronic cough⁷ and 78% of patients with laryngitis have GERD.⁸ Clear associations between GERD and asthma have been demonstrated, but data from meta-analyses fail to show improvement of asthma symptoms when GERD is appropriately treated.⁹

Diagnosis

The diagnosis of GERD can usually be made without the use of invasive tests. The accuracy of key tests, including clinical history, is outlined in Table 2. One study of patients presenting with dyspepsia (signs and symptoms referable to the upper gastrointestinal tract) found that 56% also have GERD.¹⁰ Another showed that 60% of patients referred for pH monitoring had GERD.¹¹ Since reasonable prevalence estimates for GERD in the family practice setting may be slightly lower, calculations in Table 2 assume that between 40% and 60% of patients with suspected GERD actually have the condition.

Individual symptoms of GERD such as heartburn, regurgitation, belching, or dyspepsia are of limited usefulness in diagnosis. In a survey of patients referred for pH monitoring likelihood ratios hovered near 1, meaning that the presence or absence of the symptom had little impact on the diagnosis.¹¹ The clinician’s overall impression that a patient has GERD, however, is much more useful to rule in disease than any individual symptoms.¹¹ Assuming that half of patients with suspected GERD have the disease, if a clinician suspects GERD, that probability increases to 77%. Most clinicians would find a trial of empiric therapy appropriate at that probability.

The omeprazole test is also helpful in confirming a diagnosis of GERD. It consists of the patient taking 40 mg omeprazole in the morning and 20 mg at night for 1 week. If the symptoms resolve, the test is considered diagnostic of GERD.¹² Some consider this approach to be therapeutic, as well as diagnostic. Beginning with an omeprazole test and reserving invasive testing for those not responding to the medication was cost-effective for patients with noncardiac chest pain.¹³ In a decision analysis, empiric treatment with omeprazole was a cost-effective approach to the management of GERD.¹⁴ Of course, initial endoscopy is indicated for patients with “red flags”: signs and symptoms consistent with obstruction, bleeding, or perforation, and those older than 50 years who are at a higher risk of malignancy.

Upper endoscopy, however, is not very accurate in diagnosing GERD. Among patients with GERD, only 22% have esophageal erythema, and only 48% have erosions or ulcerations. Therefore, because of costs and limited resources, the American Society for Gastrointestinal Endoscopy recommends that endoscopy be reserved for patients presenting with possible GERD who also have symptoms of more serious disease (dysphagia, weight loss, gastrointestinal bleeding) and for those not responding to a reasonable trial of therapy.¹⁵ The goal is to rule out more serious conditions.

Twenty-four–hour pH monitoring, while more accurate than endoscopy, is also reserved as a second-line test. According to the American Gastro-enterological Association guidelines, pH recording is indicated when endoscopy is normal and reflux symptoms persist despite acid suppression therapy or to evaluate extra-esophageal symptoms that may be GERD (ie, atypical chest pain or chronic cough).¹⁶ The goal here is to rule in GERD as the etiology of the patient’s symptoms.

Numerous other tests have been proposed to evaluate the patient with suspected GERD: manometry, scintigraphy, esophograms (as part of upper gastrointestinal series), and the Bernstein test (a provocative challenge with hydrochloric acid infusion). Data on the accuracy of these tests are limited by the lack of an accepted reference standard, and most have fallen out of favor because of their inconvenience or limited accuracy. None are more helpful than pH monitoring or endoscopy, and they are therefore not recommended.

Treatment

Pharmacologic

Treatments for GERD address the pathophysiology of the disease; they tend to target the removal of precipitating factors, using gravity or medications to improve acid clearance, lower the acidity of gastric contents, or improve the esophageal sphincter tone. Short-term treatment goals are to relieve symptoms and heal esophagitis; long-term goals are to prevent relapses and sequelae, such as esophageal stricture formation or adenocarcinoma.

Recommendations for lifestyle modifications to lessen GERD symptoms are extensive and mostly based on pathophysiologic data. Possible triggers for GERD include obesity, tight clothing, fatty foods, alcohol, tobacco, caffeine, onions, peppermint, and chocolate.^17-21 Because symptoms are typically worse at night, elevating the head of the bed or avoiding postprandial recumbency is recommended.^22,23 Eliminating these factors has been shown to decrease the amount of acid in the distal esophagus or improve the pressure readings on manometry. This is disease-oriented evidence; it does not necessarily translate into reduced symptoms for patients. The only patient-oriented evidence that supports lifestyle modification is a small crossover trial in which subjects had less heartburn and belching following a hamburger meal compared with the same meal with onions on the burger.¹⁸ Despite the lack of high-quality evidence of effectiveness, guidelines recommend lifestyle modification as the first-line therapy or adjunct to additional medication.^24,25 Because these lifestyle changes are not thought to be harmful and may have other possible health benefits it is reasonable to recommend them to patients until better studies are published.

Promotility agents are a possible treatment of GERD because they improve acid clearance from the esophagus. Older agents such as metoclopramide and bethanecol have little data to support their use^26,27 and are further limited by their central nervous system side effects. A meta-analysis of trial data found that cisapride heals esophagitis and decreases symptoms of GERD²⁸ but was taken off the market because of problems with cardiac arrhythmias.

Acid suppression strategies available over-the-counter include antacids, alginates, and H2-blockers (H2Bs). Small, short-term trials have demonstrated mixed results with regard to symptom relief and lessened acid exposure from both antacids and alginates.^29,30 Longer follow-up in cohort studies suggests a modest benefit from antacids.³¹ More than 24 randomized controlled trials (RCTs) have demonstrated the benefits of H2Bs in the treatment of GERD for both healing esophagitis and symptom relief. The number needed to treat (NNT) is 5; that is, for every 5 GERD patients treated with H2Bs instead of placebo, 1 patient benefits.³² H2Bs also prevent the long-term recurrence of symptoms (NNT =15).³³ There is no clear benefit of one H2B agent over another. Proton pump inhibitors (PPIs) are also well supported by meta-analyses of RCTs in the short term (NNT = 2) and long term (NNT = 3) treatment of GERD.³³ As with H2Bs, no clear advantage is found between different PPI medications.

STEPS. Using the STEPS approach (Safety, Tolerability, Efficacy, Price, Simplicity) to compare H2Bs and PPIs is one way to help guide management of GERD. H2Bs are not associated with any serious long-term complications. PPIs are theoretically linked to malignancy through 2 mechanisms: proliferation of endocrine cells leading to endocrine neoplasia and atrophic gastritis caused by chronic acid suppression as a precursor to gastric adenocarcinoma. However, studies have not borne out these concerns.³⁴ Thus, both H2Bs and PPIs seem equally safe. The best measure of tolerability is the pooled dropout rate, which is the number of patients dropping out of a study for any reason. Dropout rates are about the same for H2Bs and PPIs.³⁵ Meta-analyses of trials comparing the efficacy of H2Bs to PPIs in the short- and long-term treatment of GERD consistently favor PPIs.^33,35 Of 100 patients with GERD, approximately 25 will benefit from treatment with an H2B, and 75 to 80 will benefit from treatment with a PPI.³³ An interesting comparison of trials suggests that the therapeutic gain of PPIs is greatest in those patients with more severe GERD.³⁶ Little data exist about the prevention of complications of GERD with acid suppression. However, 1 small trial found a decreased rate of esophageal stricture recurrence at 1 year in patients with reflux esophagitis (NNT =7) treated with 30-mg lansoprazole compared with those treated with 300-mg ranitidine.³⁷ These data seem to suggest that high-risk patients are more apt to benefit from PPIs than H2Bs.

Because H2Bs are now available in generic forms, price tips the scales greatly in their favor. The average cost for PPIs is nearly 10 times that of generic H2Bs (3020-mg omeprazole tablets cost $105.55 and 60 150-mg ranitidine tablets cost $10.98 as of 11/0/00). Simplicity is a toss-up: most PPIs are dosed once daily, while H2Bs are given once or twice daily. A general approach is to start patients with GERD on a reasonable dose of H2Bs, and move on to PPIs if symptoms do not resolve in 2 to 4 weeks.^25,54 Some insurance companies require documentation of H2B failure before covering the increased costs of PPIs, despite their efficacy advantage.

Surgery

For patients with chronic or recurrent GERD, surgery offers an additional treatment option. Prospective cohort studies note that open Nissen fundoplication produces an 80% to 93% success rate at 10-year follow-up. Laproscopic procedures are producing similar short-term results, but long-term data are pending.³⁸ One randomized trial demonstrated equivalent 3-year outcomes for those undergoing Nissen fundoplication and those taking 40 mg omeprazole daily.³⁹ Decision analysis data favor the cost effectiveness of open Nissen⁴⁰ if medical treatment will be required for more than 4 years, and laproscopic procedure if medical treatment will be needed for more than 10 years.⁴¹

Prognosis

GERD may be a short-term intermittent problem or may be severe and chronic in nature. Untreated, approximately 15% of patients will have symptom relief.³² Antacids can raise that rate to an estimated 20%,^31,25 while H2Bs are associated with symptom-free rates of approximately 25%.³⁰ The best outcomes are found with PPIs, where recurrence of GERD symptoms is suppressed in 75% to 82% of patients at 1-year follow-up.^33,35

Although many patients will experience recurrences, chronic medications may not be necessary. In 1 study, 677 patients with heartburn and mild esophagitis were randomized to treatment with omeprazole or ranitidine for 2 weeks. If symptoms persisted, the dose of medication was doubled. If symptoms resolved, medication was stopped. Recurrences were treated for 2 to 4 weeks at the previously effective dose. Nearly half the patients were successfully treated with intermittent medication, and nearly 40% of initial responders required no further treatment.⁴²

GERD is associated with esophageal strictures, Barrett esophagus (metaplasia of the distal esophageal columnar cells, thought to be a precursor of dysplasia and cancer) and adenocarcinoma. Limited data are available for the actual incidence of stricture in GERD patients. One retrospective cohort study of patients discharged from veterans’ administration hospitals found that 8.4% of patients with GERD had strictures, and the association between esophageal ulcers and stricture was significant.⁴³ This study likely suffered from significant selection bias, however, and the rate in primary care practice is almost certainly much lower. Barrett was noted in 11.6% of 662 patients with GERD referred from general practice settings for endoscopy.⁴⁴ Again, though, patients with GERD referred for endoscopy are likely to have more severe disease, and a recent meta-analysis suggests that the actual risk of Barrett in unselected patients is closer to 3% to 4%.⁴⁵ A longer duration of symptoms was associated with an increase risk. Patients with Barrett esophagus; negative, low-grade, or indefinite dysplasia; and neither aneuploidy or increased 4N on flow cytometry are at very low risk of esophageal cancer (<2% over 5 years). Only approximately 4% of patients with Barrett go on to develop esophageal cancer.⁴⁶

In a well-conducted case control study in Sweden, reflux symptoms were associated with a 7- to 10-fold increase in the risk of esophageal adenocarcinoma. A dose-response risk was noted for symptom frequency, severity, and duration.⁴⁷ However, because adenocarcinoma of the esophagus is so rare, the authors note that a family physician would need to perform endoscopy on more than 1400 men older than 40 years who have severe GERD symptoms to identify 1 case of cancer. Further, there are no data to suggest that treating GERD will reduce the likelihood of these more serious sequelae.

BACKGROUND: Current recommendations for reducing the neonatal morbidity and mortality caused by group B streptococcus (GBS) infections support either screening pregnant women at 35 to 37 weeks’ gestation and treating those with positive cultures or treating GBS during labor on the basis of risk factors such as fever or prolonged rupture of membranes. Both of these strategies result in overtreatment (GBS-negative mothers receiving antibiotics) and undertreatment (GBS-positive mothers not receiving antibiotics). A rapid test performed and available at the time of labor may be a useful guide to appropriately treating at-risk mothers, if such a test is sufficiently accurate.

POPULATION STUDIED: A total of 112 pregnant women hospitalized for delivery in Quebec City, Canada, were enrolled. Of these, 57 had intact membranes, and 55 had membranes already ruptured. The authors did not comment on how the women were recruited or enrolled in the study. The prevalence of GBS colonization in this study population was 29.5%, which is similar to the prevalence reported in other community-based studies. The results are therefore likely to be representative of those patients seen by family physicians practicing obstetrics.

STUDY DESIGN AND VALIDITY: Investigators obtained rectal, vaginal, and rectovaginal swabs from the women at the time of admission and again after the rupture of membranes (for those whose membranes were intact on admission). Three tests were performed on each swab at the same laboratory: a standard GBS culture, a conventional polymerase chain reaction (PCR) assay, and a new rapid PCR assay. It was not stated whether those performing the PCR assays were blinded to the culture results.

OUTCOMES MEASURED: PCR assay accuracy was reported in terms of sensitivity, specificity, and predictive values, using the GBS culture as the gold standard.

RESULTS: Using the combined vaginal and anal swab before the rupture of membranes (N=57), each PCR test had a sensitivity and specificity of 100%. For samples obtained after the membranes had ruptured, the sensitivity dropped to only 93.8% for each test (only one false-negative result), while the specificity remained 100%. This corresponds to a positive predictive value of 100% and a negative predictive value of 98.8%. The positive likelihood ratio calculated from this data is 194; the negative likelihood ratio is 0.03. Results were available for the new PCR assay in 45 minutes, for the conventional PCR assay in 100 minutes, and for the GBS culture in 36 hours.

BACKGROUND: Current recommendations for reducing the neonatal morbidity and mortality caused by group B streptococcus (GBS) infections support either screening pregnant women at 35 to 37 weeks’ gestation and treating those with positive cultures or treating GBS during labor on the basis of risk factors such as fever or prolonged rupture of membranes. Both of these strategies result in overtreatment (GBS-negative mothers receiving antibiotics) and undertreatment (GBS-positive mothers not receiving antibiotics). A rapid test performed and available at the time of labor may be a useful guide to appropriately treating at-risk mothers, if such a test is sufficiently accurate.

POPULATION STUDIED: A total of 112 pregnant women hospitalized for delivery in Quebec City, Canada, were enrolled. Of these, 57 had intact membranes, and 55 had membranes already ruptured. The authors did not comment on how the women were recruited or enrolled in the study. The prevalence of GBS colonization in this study population was 29.5%, which is similar to the prevalence reported in other community-based studies. The results are therefore likely to be representative of those patients seen by family physicians practicing obstetrics.

STUDY DESIGN AND VALIDITY: Investigators obtained rectal, vaginal, and rectovaginal swabs from the women at the time of admission and again after the rupture of membranes (for those whose membranes were intact on admission). Three tests were performed on each swab at the same laboratory: a standard GBS culture, a conventional polymerase chain reaction (PCR) assay, and a new rapid PCR assay. It was not stated whether those performing the PCR assays were blinded to the culture results.

OUTCOMES MEASURED: PCR assay accuracy was reported in terms of sensitivity, specificity, and predictive values, using the GBS culture as the gold standard.

RESULTS: Using the combined vaginal and anal swab before the rupture of membranes (N=57), each PCR test had a sensitivity and specificity of 100%. For samples obtained after the membranes had ruptured, the sensitivity dropped to only 93.8% for each test (only one false-negative result), while the specificity remained 100%. This corresponds to a positive predictive value of 100% and a negative predictive value of 98.8%. The positive likelihood ratio calculated from this data is 194; the negative likelihood ratio is 0.03. Results were available for the new PCR assay in 45 minutes, for the conventional PCR assay in 100 minutes, and for the GBS culture in 36 hours.

BACKGROUND: Current recommendations for reducing the neonatal morbidity and mortality caused by group B streptococcus (GBS) infections support either screening pregnant women at 35 to 37 weeks’ gestation and treating those with positive cultures or treating GBS during labor on the basis of risk factors such as fever or prolonged rupture of membranes. Both of these strategies result in overtreatment (GBS-negative mothers receiving antibiotics) and undertreatment (GBS-positive mothers not receiving antibiotics). A rapid test performed and available at the time of labor may be a useful guide to appropriately treating at-risk mothers, if such a test is sufficiently accurate.

POPULATION STUDIED: A total of 112 pregnant women hospitalized for delivery in Quebec City, Canada, were enrolled. Of these, 57 had intact membranes, and 55 had membranes already ruptured. The authors did not comment on how the women were recruited or enrolled in the study. The prevalence of GBS colonization in this study population was 29.5%, which is similar to the prevalence reported in other community-based studies. The results are therefore likely to be representative of those patients seen by family physicians practicing obstetrics.

STUDY DESIGN AND VALIDITY: Investigators obtained rectal, vaginal, and rectovaginal swabs from the women at the time of admission and again after the rupture of membranes (for those whose membranes were intact on admission). Three tests were performed on each swab at the same laboratory: a standard GBS culture, a conventional polymerase chain reaction (PCR) assay, and a new rapid PCR assay. It was not stated whether those performing the PCR assays were blinded to the culture results.

OUTCOMES MEASURED: PCR assay accuracy was reported in terms of sensitivity, specificity, and predictive values, using the GBS culture as the gold standard.

RESULTS: Using the combined vaginal and anal swab before the rupture of membranes (N=57), each PCR test had a sensitivity and specificity of 100%. For samples obtained after the membranes had ruptured, the sensitivity dropped to only 93.8% for each test (only one false-negative result), while the specificity remained 100%. This corresponds to a positive predictive value of 100% and a negative predictive value of 98.8%. The positive likelihood ratio calculated from this data is 194; the negative likelihood ratio is 0.03. Results were available for the new PCR assay in 45 minutes, for the conventional PCR assay in 100 minutes, and for the GBS culture in 36 hours.

CLINICAL QUESTION
What is the association between bacterial vaginosis and preterm delivery?

Prematurity, whether defined by gestational age or birth weight, increases the risk of neonatal morbidity and mortality, as well as early childhood morbidity. Preterm birth, defined as delivery before 37 weeks’ gestation, accounts for 8% to 10% of all births¹ and leads to nearly 75% of all neonatal mortality and 50% of all long-term neurologic damage in children.² On average, first-year medical costs for infants born weighing less than 2500 grams exceed that of a full-term infant by $15,000.³

Between 25% and 60% of preterm births are thought to be attributable to maternal infections,^4,5 and are thus considered preventable. Bacterial vaginosis (BV) has been suggested as one potentially treatable risk factor for preterm delivery. BV is fairly common, with a prevalence ranging from 10% to 30% in an typical obstetrical population⁶ to more than 50% in some high-risk groups.⁷

Although otherwise thought to be a fairly benign condition, in pregnancy BV is estimated to confer a two- to threefold increased risk of prematurity.^4,8 Yet the relative risks from the literature range from^09,10> to 6.9.¹¹ These variations may be attributable to differences in study design, sample size, or confounders. The purpose of our meta-analysis was to estimate the magnitude of risk that BV poses on prematurity and pregnancy complications that may lead to prematurity.

Methods

Data Sources

To identify potential studies for inclusion, 2 independent investigators (CF and AH) conducted a MEDLINE search (1966-1996), using the terms “bacterial vaginosis,” “gardnerella” and “prematurity,” “labor-premature onset, ” “rupture of membranes-premature,” “preterm delivery,” or “preterm infant” as both medical subject headings and text words. The bibliographies of obstetric texts, all included studies, relevant reviews, and the Cochrane Library were also reviewed. Finally, we contacted several authors who had published articles on the subject in an attempt to identify any unpublished data.

Study Selection

Studies were included if they met the following criteria: (1) the population studied was pregnant women; (2) the risk factor considered was the presence of BV; (3) the outcomes measured included either gestational age or birth weight; secondary outcomes considered were preterm premature rupture of membranes and preterm onset of labor; and (4) study design was either case control or cohort trial evaluating the benefit of treating BV in pregnancy. Trials were included if sufficient data were available to compare the outcomes of those women who had BV with those who did not in the control cohort. Inclusion criteria were applied independently by the 2 investigators; differences were settled by consensus.

Non-English language papers and those containing duplication of previous data were excluded. For articles in which the data presentation prohibited the linking of BV to the outcomes of interest, we contacted the authors in writing for the original data. If the original data were unavailable, we excluded the study from final analysis.

Data Extraction

The 2 investigators also independently performed data extraction. Disagreements were settled by discussion and consensus.

The population data we collected included country, medical setting, and baseline risk of prematurity. We recorded inclusion and exclusion criteria, but those factors were insufficient to categorize the study population as either high, normal, or low risk. Therefore, baseline risk was determined by calculating the incidence of preterm delivery in the control group for each study. Since the standard incidence of preterm delivery is 8% to 10%, we considered as high risk those studies with rates greater than 10% in the non-BV group; those below were categorized as low or normal risk.

We recorded the method and timing of BV diagnosis. In studies with multiple methods of determining BV, we collected the results for each method. The gram stain/wet mount result was preferentially used in the final analysis, as this is the most relevant technique for clinical assessment of BV. Vaginal culture data could only be correlated to outcome by individual microbes, so we limited data recording to Gardnerella vaginalis only. We recorded the timing of the diagnosis of BV as presented (ie, weeks of gestational age, trimester, during labor), then grouped the data by trimester. If more than one sample was collected in an individual study, we recorded all results but used the earliest in the analysis.

We recorded outcome data dichotomously in 2 x 2 tables for all 4 study outcomes. Extracted information on the handling of confounding included the method used, the adjusted odds ratio with confidence interval, and the variables included in the final model.

In studies that reported preterm delivery results at multiple gestational age cutoffs, we used only the 37-week cutoff. If studies used a definition other than 37 weeks, their data were noted for subanalysis. Some case control studies defined cases as women in preterm labor, but reported data separately from those cases that delivered prematurely from those that delivered after 37 weeks. For the preterm delivery outcome, the preterm labor cases who delivered at term were included in the control group.

Validity Assessment

We used validity assessment worksheets that were developed specifically for this project using a summary of previously published criteria.^12-14 Each included study was critically appraised independently by the 2 investigators, and their assessments were compared. A third investigator (LM) settled disagreements. We did not use validity criteria to exclude any study from analysis.

Data Synthesis

We calculated summary estimates of risk as odds ratios using both the fixed and random effects models.^15-17 Additionally, we combined cohort studies to generate summary relative risk estimates using both models. Precision is reported as 95% confidence intervals for each statistic. These calculations were generated using Review Manager 3.0 software.¹⁸ We evaluated homogeneity using the chi-square statistic:¹⁵ the greater the P value, the more homogeneous the studies.

We conducted subanalyses by study design, baseline population risk of prematurity, method and timing of BV diagnosis, and country of study population. When the subgroupings resulted in any category having fewer than 3 studies, we did not calculate a summary statistic for that group. We excluded those studies and generated a new pooled risk assessment for the alternate groups only.

To address the issue of confounders, we used the general variance-based model^15,19,20 to combine the adjusted odds ratios of individual studies into a summary odds ratio with its 95% confidence interval.

Results

Data Sources and Study Selection

Our literature review identified 233 studies; no unpublished data were discovered. Reviewing the abstracts identified 39 studies for possible inclusion (27 observational studies and 12 trials). We excluded 11 observational studies because they had an inadequate or no control group (2),^21,22 no vaginal assessment of BV (2),^23,24 repeated data (1),²⁵ or the inability to link BV with pregnancy outcomes (6).^9,26-30 Nine of the 12 trials did not present their control group cohort data in a way that distinguished the outcomes by the presence or absence of BV and were therefore excluded.^31-39 We included 19 studies^10,11,40-56 in the final analysis: 8 case control trials and 11 cohort studies. Three from this latter group consisted of the placebo group of randomized controlled trials.

We estimated the likelihood of publication bias by generating a funnel plot.¹⁵ The graph of study size versus the logarithm of the ratio results, although funnel-shaped, is not completely symmetric.* Data from small studies demonstrating a protective effect of BV in pregnancy are missing.

Data Extraction

Twelve of the 19 included studies drew patients from a university or tertiary care hospital setting; the 7 others were clinic based Table 1. Two studies were conducted in nonindustrialized countries (Nairobi and Indonesia). The remaining studies were performed in the United States (11), Australia (3), the United Kingdom (1), Sweden (1), and Finland (1). The baseline prevalence of preterm delivery ranged from 1.1% to 64.9%, with a mean of 20.0% and a median of 13.8%.

To diagnose bacterial vaginosis, 13 studies used gram stain alone or in combination with a wet mount, 2 used gas-liquid chromatography, and 4 used vaginal swab cultures. The timing of risk factor detection varied among the studies, ranging from the first prenatal visit to the time of labor. We grouped the studies by trimester as precisely as possible, resulting in 10 studies diagnosing BV in the first or second trimester, 4 with second or third trimester assessment, and 4 studies discovering BV in the third trimester only (which in most cases was at the time of labor).

In most studies, gestational age was determined by the best obstetric clinical estimate — using the date of the mother’s last menstrual period, detection of fetal heart tones, fundal heights, and obstetrical ultrasound. In one study, gestational age was determined only by pediatric assessment. Preterm delivery was evaluated as a dichotomous outcome in 18 of the 19 included studies. Two studies used 35 weeks’ gestation or less to define a preterm infant; one study used 36 weeks. All others adhered to the standard definition of preterm as any gestation with a duration of less than 37 weeks. Low birth weight was defined as an infant weighing less than 2500 grams [at birth] in all 6 studies that reported this outcome.

Of the 7 studies evaluating preterm premature rupture of membranes, 2 used 36 weeks as the cutoff for preterm; the other 5 used 37 weeks. Only 2 studies reported the method to determine membrane rupture (both used pH and ferning criteria). The time from rupture of membranes to labor onset varied from 1 to 6 hours in the 4 studies defining this period.

For the outcome of preterm onset of labor, all 9 studies defined preterm as gestational age less than 37 weeks. Most studies defined labor as regular painful uterine contractions; only 2 required cervical change. Two others considered treatment for preterm labor as the definition of preterm onset of labor.

Validity Assessment

Details about validity assessment and the effect of biases on the summary estimates of individual studies are available elsewhere.* Two biases were common. First was the misclassification of either the predictor or the outcomes, which tended to underestimate risk. Second was the issue of confounding variables, which tended to overestimate the odds ratio.

Data Synthesis

The Figure shows that women with BV were more likely to deliver a preterm infant (odds ratio, fixed effects model [OR<->FIXED<->] 1.85; 95% CI, 1.62-2.11) or an infant weighing less than 2500 grams (OR<->FIXED<-> 1.57; 95% CI, 1.32-1.87).

For the secondary outcomes of preterm premature rupture of membranes and preterm onset of labor, the resultant ORs<->FIXED <->were 1.83 (95% CI, 1.39-2.44) and 2.19 (95% CI, 1.73-2.76), respectively. The studies combined for preterm onset of labor met statistical requirements of homogeneity (P <\>>.25); those for preterm delivery, low birth weight, and preterm premature rupture of membranes did not. Recalculation of the odds ratio using the random effects model did not result in the loss of statistical significance for any of the main outcomes Table 2.

Pooling only cohort studies to generate a summary relative risk also resulted in a persistently elevated risk of prematurity for those mothers with BV, ranging from a 1.44- to a 2.86-fold increase Table 2, and homogeneity criteria were met for low birth weight, preterm premature rupture of membranes, and preterm onset of labor, (P <\>>.33) but not for preterm delivery. BV was significantly associated with preterm delivery in nearly all the subanalyses conducted Table 3.

Seven of the 18 studies evaluating preterm delivery did not perform regression analysis to evaluate for confounding; 4 others did this analysis but did not report an adjusted risk for BV. Thus, only 7 studies had controlled data available for a summary estimate. These studies, with their respective adjusted odds ratios and confounders considered, are listed in Table 5. As expected, the resultant summary estimate of the adjusted odds ratio was lower than that obtained from unadjusted data but remained significant clinically (OR = 1.60), as well as statistically (95% CI, 1.44-1.74).

Discussion

Our study pooled data representing more than 17,000 patients and the results show BV to be a significant risk factor for preterm and low birth weight deliveries. Additionally, BV is significantly associated with preterm onset of labor and preterm premature rupture of membranes. Summary relative risks calculated using cohort data only, although lower than the odds ratios, also showed a significant association between BV and all prematurity outcomes. Although often used interchangeably with relative risks, odds ratios tend to overestimate risk in cases of a positive association and a nonrare outcome. In our study, odds ratios exceeded relative risks for 3 of the 4 outcomes. This may reflect violation of the rare disease assumption or may simply be due to pooling a different subset of studies.

We believe the nearly twofold increase in prematurity with BV is especially robust for several reasons. First, the results are statistically significant regardless of the statistics used to generate them: odds ratio or relative risk, fixed or random effects models.

Second, decisions regarding data handling in our study were made to produce the most conservative estimate of association. For example, in case control studies that defined cases by the presence of preterm labor, those women with preterm onset of labor who gave birth at term were analyzed with the control group in the preterm delivery analysis. These women may be more likely to have BV, and moving them to the control group for the purpose of analysis would tend to underestimate the risk. In addition, for studies using vaginal cultures for diagnosing BV, we extracted data only for G vaginalis culture, which might lead to overdiagnosis, since the presence of G vaginalis can be a normal finding. This nondifferential error, as well as the potential imprecision of clinical estimates of gestational age, would tend to bias the summary estimate toward the null. Despite these potential underestimations, BV remained a significant risk factor for prematurity.

Finally, BV remained a significant risk for preterm delivery regardless of the subanalysis groupings. Pooling data from different populations with variable baseline risk or in different settings may lend confidence to the generalizability of these estimates. Caution is warranted, however, when drawing conclusions from any specific subanalysis.

All studies included in this analysis were observational in design, which raises 2 particular concerns: causality and confounding. Because neither the Mantel-Haenszel or DerSimonian and Laird methods for pooling risk estimates can incorporate confounding, we calculated separately the adjusted odds ratio pooled from risk estimates generated by regression analyses in individual studies. This summary of adjusted odds ratios still demonstrated a statistically significant elevated risk of 1.6 for preterm delivery in women with BV. This value may be overestimated, because 4 studies did not report regression results for the variable of BV. Presumably, these values were near the null but were likely not statistically significant, or they would have been reported. A lower summary adjusted odds ratio may have resulted were inclusion of these results possible.

Although causality cannot be proven by observational studies or by meta-analytic combination of such studies, several of the criteria suggesting causality are met.^57,58 The strength of the association is relatively small, ranging from 1.4 to 2.4. Although it is possible that confounders account for all this association, we think that is unlikely given that the controlled summary odds ratio remained significant. This meta-analysis did not address the dose response question directly, although some individual studies reported a stronger association with prematurity outcomes for those who had heavier colonization or higher BV scores. In all but one study this risk factor preceded the outcome, although in some cases vaginal assessment was done at the time of labor. Results are consistent, as can be seen from the summary figures: No study found a protective effect, and only one had a null value. The presence of BV makes biological sense as a contributor to preterm labor and thus to preterm premature rupture of membranes and preterm delivery. One proposed mechanism for this association is the production of phospholipase by the bacteria associated with BV. These enzymes can initiate prostaglandin synthesis, which is one step in the physiology of normal labor activation.⁵³ Infections in pregnancy are also associated with fetuses that were small for the gestational age,⁵⁹ which is another mechanism for low birth weight. Taken together, these factors lend support to a causal association for BV and prematurity.

Identifying BV in pregnancy as a modifiable risk factor for prematurity raises the obvious question of intervention. Accumulating evidence demonstrates that treating pregnant women who have BV with certain oral antibiotics can decrease the risk of prematurity. Clindamycin taken orally by pregnant women with BV decreased preterm deliveries and low birth weight infants by approximately 50%.^38,55 Hauth⁴² combined oral erythromycin and metronidazole and found a decreased rate in preterm births among a group of high-risk women with BV. Neither oral amoxicillin³¹ or intravaginal clindamycin³³ have been shown to affect pregnancy outcomes.

Limitations

As with any meta-analysis, one major limitation of this work is the appropriateness of combining results from different studies. Statistical homogeneity was met in 1 of the 4 analyses for summary odds ratios (pooling all studies possible), but in 3 of the 4 analyses when only cohort studies were pooled. This discrepancy suggests that study design was likely a key source of heterogeneity in this review. In addition to study design differences, we expected heterogeneity given the range of risks reported in individual studies, the varying population risks, the disparate methods, the timing of BV diagnosis, and the different definitions for the measured outcomes. One method for addressing heterogeneity is to use the random effects model, which accounts for variability between studies when estimating the precision of the risk. When we analyzed the data using this method, none of our conclusions changed.

A second limitation of this project is the possibility of publication bias. Our funnel plot reflects the absence of studies finding that BV protects pregnant women from delivering preterm infants. By chance alone, some studies may find this result, but it is unlikely that such a study would be published. Using Orwin’s method⁶⁰ to calculate a fail-safe N of 60 and our weighted summary effect size of 0.052 (number needed to harm [NNH] = 19), more than 75 studies showing no effect would be needed to drop the risk difference to 0.01 (NNH = 100). Given our systematic and complete search, we think it unlikely that publication bias accounts for our findings, despite the asymmetry of the funnel plot.

Implications for Future Research

The association of BV with prematurity remains at the disease-oriented level of evidence. Although neonatal and infant morbidity and mortality are increased with preterm delivery, the clinical impact of BV or its treatment on these patient-oriented outcomes remains unclear. Additionally, since approximately 50% of the pregnant women with BV are asymptomatic, some advocate universal screening for BV during pregnancy.⁶¹ A large randomized controlled trial evaluating patient-oriented health benefits as well as costs is warranted before this becomes a part of routine prenatal care.

Recommendations for clinical practice

BV in pregnancy is associated with a significant risk of preterm delivery. Evidence suggests that oral treatment with certain antibiotics can decrease this risk, especially in those with a previous preterm birth.⁶² Thus, if identified during pregnancy, BV should be treated. There is insufficient data to recommend screening for BV during pregnancy.

Acknowledgments

This project was done as part of Dr Flynn’s requirements for her master’s degree. She thanks the members of her thesis committee for their invaluable feedback. The authors also thank Shirley Killian for her assistance with article retrieval, Tangelia Pruitt for her assistance with bibliographic management, and Bill Grant for his statistical guidance.

CLINICAL QUESTION
What is the association between bacterial vaginosis and preterm delivery?

Prematurity, whether defined by gestational age or birth weight, increases the risk of neonatal morbidity and mortality, as well as early childhood morbidity. Preterm birth, defined as delivery before 37 weeks’ gestation, accounts for 8% to 10% of all births¹ and leads to nearly 75% of all neonatal mortality and 50% of all long-term neurologic damage in children.² On average, first-year medical costs for infants born weighing less than 2500 grams exceed that of a full-term infant by $15,000.³

Between 25% and 60% of preterm births are thought to be attributable to maternal infections,^4,5 and are thus considered preventable. Bacterial vaginosis (BV) has been suggested as one potentially treatable risk factor for preterm delivery. BV is fairly common, with a prevalence ranging from 10% to 30% in an typical obstetrical population⁶ to more than 50% in some high-risk groups.⁷

Although otherwise thought to be a fairly benign condition, in pregnancy BV is estimated to confer a two- to threefold increased risk of prematurity.^4,8 Yet the relative risks from the literature range from^09,10> to 6.9.¹¹ These variations may be attributable to differences in study design, sample size, or confounders. The purpose of our meta-analysis was to estimate the magnitude of risk that BV poses on prematurity and pregnancy complications that may lead to prematurity.

Methods

Data Sources

To identify potential studies for inclusion, 2 independent investigators (CF and AH) conducted a MEDLINE search (1966-1996), using the terms “bacterial vaginosis,” “gardnerella” and “prematurity,” “labor-premature onset, ” “rupture of membranes-premature,” “preterm delivery,” or “preterm infant” as both medical subject headings and text words. The bibliographies of obstetric texts, all included studies, relevant reviews, and the Cochrane Library were also reviewed. Finally, we contacted several authors who had published articles on the subject in an attempt to identify any unpublished data.

Study Selection

Studies were included if they met the following criteria: (1) the population studied was pregnant women; (2) the risk factor considered was the presence of BV; (3) the outcomes measured included either gestational age or birth weight; secondary outcomes considered were preterm premature rupture of membranes and preterm onset of labor; and (4) study design was either case control or cohort trial evaluating the benefit of treating BV in pregnancy. Trials were included if sufficient data were available to compare the outcomes of those women who had BV with those who did not in the control cohort. Inclusion criteria were applied independently by the 2 investigators; differences were settled by consensus.

Non-English language papers and those containing duplication of previous data were excluded. For articles in which the data presentation prohibited the linking of BV to the outcomes of interest, we contacted the authors in writing for the original data. If the original data were unavailable, we excluded the study from final analysis.

Data Extraction

The 2 investigators also independently performed data extraction. Disagreements were settled by discussion and consensus.

The population data we collected included country, medical setting, and baseline risk of prematurity. We recorded inclusion and exclusion criteria, but those factors were insufficient to categorize the study population as either high, normal, or low risk. Therefore, baseline risk was determined by calculating the incidence of preterm delivery in the control group for each study. Since the standard incidence of preterm delivery is 8% to 10%, we considered as high risk those studies with rates greater than 10% in the non-BV group; those below were categorized as low or normal risk.

We recorded the method and timing of BV diagnosis. In studies with multiple methods of determining BV, we collected the results for each method. The gram stain/wet mount result was preferentially used in the final analysis, as this is the most relevant technique for clinical assessment of BV. Vaginal culture data could only be correlated to outcome by individual microbes, so we limited data recording to Gardnerella vaginalis only. We recorded the timing of the diagnosis of BV as presented (ie, weeks of gestational age, trimester, during labor), then grouped the data by trimester. If more than one sample was collected in an individual study, we recorded all results but used the earliest in the analysis.

We recorded outcome data dichotomously in 2 x 2 tables for all 4 study outcomes. Extracted information on the handling of confounding included the method used, the adjusted odds ratio with confidence interval, and the variables included in the final model.

In studies that reported preterm delivery results at multiple gestational age cutoffs, we used only the 37-week cutoff. If studies used a definition other than 37 weeks, their data were noted for subanalysis. Some case control studies defined cases as women in preterm labor, but reported data separately from those cases that delivered prematurely from those that delivered after 37 weeks. For the preterm delivery outcome, the preterm labor cases who delivered at term were included in the control group.

Validity Assessment

We used validity assessment worksheets that were developed specifically for this project using a summary of previously published criteria.^12-14 Each included study was critically appraised independently by the 2 investigators, and their assessments were compared. A third investigator (LM) settled disagreements. We did not use validity criteria to exclude any study from analysis.

Data Synthesis

We calculated summary estimates of risk as odds ratios using both the fixed and random effects models.^15-17 Additionally, we combined cohort studies to generate summary relative risk estimates using both models. Precision is reported as 95% confidence intervals for each statistic. These calculations were generated using Review Manager 3.0 software.¹⁸ We evaluated homogeneity using the chi-square statistic:¹⁵ the greater the P value, the more homogeneous the studies.

We conducted subanalyses by study design, baseline population risk of prematurity, method and timing of BV diagnosis, and country of study population. When the subgroupings resulted in any category having fewer than 3 studies, we did not calculate a summary statistic for that group. We excluded those studies and generated a new pooled risk assessment for the alternate groups only.

To address the issue of confounders, we used the general variance-based model^15,19,20 to combine the adjusted odds ratios of individual studies into a summary odds ratio with its 95% confidence interval.

Results

Data Sources and Study Selection

Our literature review identified 233 studies; no unpublished data were discovered. Reviewing the abstracts identified 39 studies for possible inclusion (27 observational studies and 12 trials). We excluded 11 observational studies because they had an inadequate or no control group (2),^21,22 no vaginal assessment of BV (2),^23,24 repeated data (1),²⁵ or the inability to link BV with pregnancy outcomes (6).^9,26-30 Nine of the 12 trials did not present their control group cohort data in a way that distinguished the outcomes by the presence or absence of BV and were therefore excluded.^31-39 We included 19 studies^10,11,40-56 in the final analysis: 8 case control trials and 11 cohort studies. Three from this latter group consisted of the placebo group of randomized controlled trials.

We estimated the likelihood of publication bias by generating a funnel plot.¹⁵ The graph of study size versus the logarithm of the ratio results, although funnel-shaped, is not completely symmetric.* Data from small studies demonstrating a protective effect of BV in pregnancy are missing.

Data Extraction

Twelve of the 19 included studies drew patients from a university or tertiary care hospital setting; the 7 others were clinic based Table 1. Two studies were conducted in nonindustrialized countries (Nairobi and Indonesia). The remaining studies were performed in the United States (11), Australia (3), the United Kingdom (1), Sweden (1), and Finland (1). The baseline prevalence of preterm delivery ranged from 1.1% to 64.9%, with a mean of 20.0% and a median of 13.8%.

To diagnose bacterial vaginosis, 13 studies used gram stain alone or in combination with a wet mount, 2 used gas-liquid chromatography, and 4 used vaginal swab cultures. The timing of risk factor detection varied among the studies, ranging from the first prenatal visit to the time of labor. We grouped the studies by trimester as precisely as possible, resulting in 10 studies diagnosing BV in the first or second trimester, 4 with second or third trimester assessment, and 4 studies discovering BV in the third trimester only (which in most cases was at the time of labor).

In most studies, gestational age was determined by the best obstetric clinical estimate — using the date of the mother’s last menstrual period, detection of fetal heart tones, fundal heights, and obstetrical ultrasound. In one study, gestational age was determined only by pediatric assessment. Preterm delivery was evaluated as a dichotomous outcome in 18 of the 19 included studies. Two studies used 35 weeks’ gestation or less to define a preterm infant; one study used 36 weeks. All others adhered to the standard definition of preterm as any gestation with a duration of less than 37 weeks. Low birth weight was defined as an infant weighing less than 2500 grams [at birth] in all 6 studies that reported this outcome.

Of the 7 studies evaluating preterm premature rupture of membranes, 2 used 36 weeks as the cutoff for preterm; the other 5 used 37 weeks. Only 2 studies reported the method to determine membrane rupture (both used pH and ferning criteria). The time from rupture of membranes to labor onset varied from 1 to 6 hours in the 4 studies defining this period.

For the outcome of preterm onset of labor, all 9 studies defined preterm as gestational age less than 37 weeks. Most studies defined labor as regular painful uterine contractions; only 2 required cervical change. Two others considered treatment for preterm labor as the definition of preterm onset of labor.

Validity Assessment

Details about validity assessment and the effect of biases on the summary estimates of individual studies are available elsewhere.* Two biases were common. First was the misclassification of either the predictor or the outcomes, which tended to underestimate risk. Second was the issue of confounding variables, which tended to overestimate the odds ratio.

Data Synthesis

The Figure shows that women with BV were more likely to deliver a preterm infant (odds ratio, fixed effects model [OR<->FIXED<->] 1.85; 95% CI, 1.62-2.11) or an infant weighing less than 2500 grams (OR<->FIXED<-> 1.57; 95% CI, 1.32-1.87).

For the secondary outcomes of preterm premature rupture of membranes and preterm onset of labor, the resultant ORs<->FIXED <->were 1.83 (95% CI, 1.39-2.44) and 2.19 (95% CI, 1.73-2.76), respectively. The studies combined for preterm onset of labor met statistical requirements of homogeneity (P <\>>.25); those for preterm delivery, low birth weight, and preterm premature rupture of membranes did not. Recalculation of the odds ratio using the random effects model did not result in the loss of statistical significance for any of the main outcomes Table 2.

Pooling only cohort studies to generate a summary relative risk also resulted in a persistently elevated risk of prematurity for those mothers with BV, ranging from a 1.44- to a 2.86-fold increase Table 2, and homogeneity criteria were met for low birth weight, preterm premature rupture of membranes, and preterm onset of labor, (P <\>>.33) but not for preterm delivery. BV was significantly associated with preterm delivery in nearly all the subanalyses conducted Table 3.

Seven of the 18 studies evaluating preterm delivery did not perform regression analysis to evaluate for confounding; 4 others did this analysis but did not report an adjusted risk for BV. Thus, only 7 studies had controlled data available for a summary estimate. These studies, with their respective adjusted odds ratios and confounders considered, are listed in Table 5. As expected, the resultant summary estimate of the adjusted odds ratio was lower than that obtained from unadjusted data but remained significant clinically (OR = 1.60), as well as statistically (95% CI, 1.44-1.74).

Discussion

Our study pooled data representing more than 17,000 patients and the results show BV to be a significant risk factor for preterm and low birth weight deliveries. Additionally, BV is significantly associated with preterm onset of labor and preterm premature rupture of membranes. Summary relative risks calculated using cohort data only, although lower than the odds ratios, also showed a significant association between BV and all prematurity outcomes. Although often used interchangeably with relative risks, odds ratios tend to overestimate risk in cases of a positive association and a nonrare outcome. In our study, odds ratios exceeded relative risks for 3 of the 4 outcomes. This may reflect violation of the rare disease assumption or may simply be due to pooling a different subset of studies.

We believe the nearly twofold increase in prematurity with BV is especially robust for several reasons. First, the results are statistically significant regardless of the statistics used to generate them: odds ratio or relative risk, fixed or random effects models.

Second, decisions regarding data handling in our study were made to produce the most conservative estimate of association. For example, in case control studies that defined cases by the presence of preterm labor, those women with preterm onset of labor who gave birth at term were analyzed with the control group in the preterm delivery analysis. These women may be more likely to have BV, and moving them to the control group for the purpose of analysis would tend to underestimate the risk. In addition, for studies using vaginal cultures for diagnosing BV, we extracted data only for G vaginalis culture, which might lead to overdiagnosis, since the presence of G vaginalis can be a normal finding. This nondifferential error, as well as the potential imprecision of clinical estimates of gestational age, would tend to bias the summary estimate toward the null. Despite these potential underestimations, BV remained a significant risk factor for prematurity.

Finally, BV remained a significant risk for preterm delivery regardless of the subanalysis groupings. Pooling data from different populations with variable baseline risk or in different settings may lend confidence to the generalizability of these estimates. Caution is warranted, however, when drawing conclusions from any specific subanalysis.

All studies included in this analysis were observational in design, which raises 2 particular concerns: causality and confounding. Because neither the Mantel-Haenszel or DerSimonian and Laird methods for pooling risk estimates can incorporate confounding, we calculated separately the adjusted odds ratio pooled from risk estimates generated by regression analyses in individual studies. This summary of adjusted odds ratios still demonstrated a statistically significant elevated risk of 1.6 for preterm delivery in women with BV. This value may be overestimated, because 4 studies did not report regression results for the variable of BV. Presumably, these values were near the null but were likely not statistically significant, or they would have been reported. A lower summary adjusted odds ratio may have resulted were inclusion of these results possible.

Although causality cannot be proven by observational studies or by meta-analytic combination of such studies, several of the criteria suggesting causality are met.^57,58 The strength of the association is relatively small, ranging from 1.4 to 2.4. Although it is possible that confounders account for all this association, we think that is unlikely given that the controlled summary odds ratio remained significant. This meta-analysis did not address the dose response question directly, although some individual studies reported a stronger association with prematurity outcomes for those who had heavier colonization or higher BV scores. In all but one study this risk factor preceded the outcome, although in some cases vaginal assessment was done at the time of labor. Results are consistent, as can be seen from the summary figures: No study found a protective effect, and only one had a null value. The presence of BV makes biological sense as a contributor to preterm labor and thus to preterm premature rupture of membranes and preterm delivery. One proposed mechanism for this association is the production of phospholipase by the bacteria associated with BV. These enzymes can initiate prostaglandin synthesis, which is one step in the physiology of normal labor activation.⁵³ Infections in pregnancy are also associated with fetuses that were small for the gestational age,⁵⁹ which is another mechanism for low birth weight. Taken together, these factors lend support to a causal association for BV and prematurity.

Identifying BV in pregnancy as a modifiable risk factor for prematurity raises the obvious question of intervention. Accumulating evidence demonstrates that treating pregnant women who have BV with certain oral antibiotics can decrease the risk of prematurity. Clindamycin taken orally by pregnant women with BV decreased preterm deliveries and low birth weight infants by approximately 50%.^38,55 Hauth⁴² combined oral erythromycin and metronidazole and found a decreased rate in preterm births among a group of high-risk women with BV. Neither oral amoxicillin³¹ or intravaginal clindamycin³³ have been shown to affect pregnancy outcomes.

Limitations

As with any meta-analysis, one major limitation of this work is the appropriateness of combining results from different studies. Statistical homogeneity was met in 1 of the 4 analyses for summary odds ratios (pooling all studies possible), but in 3 of the 4 analyses when only cohort studies were pooled. This discrepancy suggests that study design was likely a key source of heterogeneity in this review. In addition to study design differences, we expected heterogeneity given the range of risks reported in individual studies, the varying population risks, the disparate methods, the timing of BV diagnosis, and the different definitions for the measured outcomes. One method for addressing heterogeneity is to use the random effects model, which accounts for variability between studies when estimating the precision of the risk. When we analyzed the data using this method, none of our conclusions changed.

A second limitation of this project is the possibility of publication bias. Our funnel plot reflects the absence of studies finding that BV protects pregnant women from delivering preterm infants. By chance alone, some studies may find this result, but it is unlikely that such a study would be published. Using Orwin’s method⁶⁰ to calculate a fail-safe N of 60 and our weighted summary effect size of 0.052 (number needed to harm [NNH] = 19), more than 75 studies showing no effect would be needed to drop the risk difference to 0.01 (NNH = 100). Given our systematic and complete search, we think it unlikely that publication bias accounts for our findings, despite the asymmetry of the funnel plot.

Implications for Future Research

The association of BV with prematurity remains at the disease-oriented level of evidence. Although neonatal and infant morbidity and mortality are increased with preterm delivery, the clinical impact of BV or its treatment on these patient-oriented outcomes remains unclear. Additionally, since approximately 50% of the pregnant women with BV are asymptomatic, some advocate universal screening for BV during pregnancy.⁶¹ A large randomized controlled trial evaluating patient-oriented health benefits as well as costs is warranted before this becomes a part of routine prenatal care.

Recommendations for clinical practice

BV in pregnancy is associated with a significant risk of preterm delivery. Evidence suggests that oral treatment with certain antibiotics can decrease this risk, especially in those with a previous preterm birth.⁶² Thus, if identified during pregnancy, BV should be treated. There is insufficient data to recommend screening for BV during pregnancy.

Acknowledgments

This project was done as part of Dr Flynn’s requirements for her master’s degree. She thanks the members of her thesis committee for their invaluable feedback. The authors also thank Shirley Killian for her assistance with article retrieval, Tangelia Pruitt for her assistance with bibliographic management, and Bill Grant for his statistical guidance.

CLINICAL QUESTION
What is the association between bacterial vaginosis and preterm delivery?

Prematurity, whether defined by gestational age or birth weight, increases the risk of neonatal morbidity and mortality, as well as early childhood morbidity. Preterm birth, defined as delivery before 37 weeks’ gestation, accounts for 8% to 10% of all births¹ and leads to nearly 75% of all neonatal mortality and 50% of all long-term neurologic damage in children.² On average, first-year medical costs for infants born weighing less than 2500 grams exceed that of a full-term infant by $15,000.³

Between 25% and 60% of preterm births are thought to be attributable to maternal infections,^4,5 and are thus considered preventable. Bacterial vaginosis (BV) has been suggested as one potentially treatable risk factor for preterm delivery. BV is fairly common, with a prevalence ranging from 10% to 30% in an typical obstetrical population⁶ to more than 50% in some high-risk groups.⁷

Although otherwise thought to be a fairly benign condition, in pregnancy BV is estimated to confer a two- to threefold increased risk of prematurity.^4,8 Yet the relative risks from the literature range from^09,10> to 6.9.¹¹ These variations may be attributable to differences in study design, sample size, or confounders. The purpose of our meta-analysis was to estimate the magnitude of risk that BV poses on prematurity and pregnancy complications that may lead to prematurity.

Methods

Data Sources

To identify potential studies for inclusion, 2 independent investigators (CF and AH) conducted a MEDLINE search (1966-1996), using the terms “bacterial vaginosis,” “gardnerella” and “prematurity,” “labor-premature onset, ” “rupture of membranes-premature,” “preterm delivery,” or “preterm infant” as both medical subject headings and text words. The bibliographies of obstetric texts, all included studies, relevant reviews, and the Cochrane Library were also reviewed. Finally, we contacted several authors who had published articles on the subject in an attempt to identify any unpublished data.

Study Selection

Studies were included if they met the following criteria: (1) the population studied was pregnant women; (2) the risk factor considered was the presence of BV; (3) the outcomes measured included either gestational age or birth weight; secondary outcomes considered were preterm premature rupture of membranes and preterm onset of labor; and (4) study design was either case control or cohort trial evaluating the benefit of treating BV in pregnancy. Trials were included if sufficient data were available to compare the outcomes of those women who had BV with those who did not in the control cohort. Inclusion criteria were applied independently by the 2 investigators; differences were settled by consensus.

Non-English language papers and those containing duplication of previous data were excluded. For articles in which the data presentation prohibited the linking of BV to the outcomes of interest, we contacted the authors in writing for the original data. If the original data were unavailable, we excluded the study from final analysis.

Data Extraction

The 2 investigators also independently performed data extraction. Disagreements were settled by discussion and consensus.

The population data we collected included country, medical setting, and baseline risk of prematurity. We recorded inclusion and exclusion criteria, but those factors were insufficient to categorize the study population as either high, normal, or low risk. Therefore, baseline risk was determined by calculating the incidence of preterm delivery in the control group for each study. Since the standard incidence of preterm delivery is 8% to 10%, we considered as high risk those studies with rates greater than 10% in the non-BV group; those below were categorized as low or normal risk.

We recorded the method and timing of BV diagnosis. In studies with multiple methods of determining BV, we collected the results for each method. The gram stain/wet mount result was preferentially used in the final analysis, as this is the most relevant technique for clinical assessment of BV. Vaginal culture data could only be correlated to outcome by individual microbes, so we limited data recording to Gardnerella vaginalis only. We recorded the timing of the diagnosis of BV as presented (ie, weeks of gestational age, trimester, during labor), then grouped the data by trimester. If more than one sample was collected in an individual study, we recorded all results but used the earliest in the analysis.

We recorded outcome data dichotomously in 2 x 2 tables for all 4 study outcomes. Extracted information on the handling of confounding included the method used, the adjusted odds ratio with confidence interval, and the variables included in the final model.

In studies that reported preterm delivery results at multiple gestational age cutoffs, we used only the 37-week cutoff. If studies used a definition other than 37 weeks, their data were noted for subanalysis. Some case control studies defined cases as women in preterm labor, but reported data separately from those cases that delivered prematurely from those that delivered after 37 weeks. For the preterm delivery outcome, the preterm labor cases who delivered at term were included in the control group.

Validity Assessment

We used validity assessment worksheets that were developed specifically for this project using a summary of previously published criteria.^12-14 Each included study was critically appraised independently by the 2 investigators, and their assessments were compared. A third investigator (LM) settled disagreements. We did not use validity criteria to exclude any study from analysis.

Data Synthesis

We calculated summary estimates of risk as odds ratios using both the fixed and random effects models.^15-17 Additionally, we combined cohort studies to generate summary relative risk estimates using both models. Precision is reported as 95% confidence intervals for each statistic. These calculations were generated using Review Manager 3.0 software.¹⁸ We evaluated homogeneity using the chi-square statistic:¹⁵ the greater the P value, the more homogeneous the studies.

We conducted subanalyses by study design, baseline population risk of prematurity, method and timing of BV diagnosis, and country of study population. When the subgroupings resulted in any category having fewer than 3 studies, we did not calculate a summary statistic for that group. We excluded those studies and generated a new pooled risk assessment for the alternate groups only.

To address the issue of confounders, we used the general variance-based model^15,19,20 to combine the adjusted odds ratios of individual studies into a summary odds ratio with its 95% confidence interval.

Results

Data Sources and Study Selection

Our literature review identified 233 studies; no unpublished data were discovered. Reviewing the abstracts identified 39 studies for possible inclusion (27 observational studies and 12 trials). We excluded 11 observational studies because they had an inadequate or no control group (2),^21,22 no vaginal assessment of BV (2),^23,24 repeated data (1),²⁵ or the inability to link BV with pregnancy outcomes (6).^9,26-30 Nine of the 12 trials did not present their control group cohort data in a way that distinguished the outcomes by the presence or absence of BV and were therefore excluded.^31-39 We included 19 studies^10,11,40-56 in the final analysis: 8 case control trials and 11 cohort studies. Three from this latter group consisted of the placebo group of randomized controlled trials.

We estimated the likelihood of publication bias by generating a funnel plot.¹⁵ The graph of study size versus the logarithm of the ratio results, although funnel-shaped, is not completely symmetric.* Data from small studies demonstrating a protective effect of BV in pregnancy are missing.

Data Extraction

Twelve of the 19 included studies drew patients from a university or tertiary care hospital setting; the 7 others were clinic based Table 1. Two studies were conducted in nonindustrialized countries (Nairobi and Indonesia). The remaining studies were performed in the United States (11), Australia (3), the United Kingdom (1), Sweden (1), and Finland (1). The baseline prevalence of preterm delivery ranged from 1.1% to 64.9%, with a mean of 20.0% and a median of 13.8%.

To diagnose bacterial vaginosis, 13 studies used gram stain alone or in combination with a wet mount, 2 used gas-liquid chromatography, and 4 used vaginal swab cultures. The timing of risk factor detection varied among the studies, ranging from the first prenatal visit to the time of labor. We grouped the studies by trimester as precisely as possible, resulting in 10 studies diagnosing BV in the first or second trimester, 4 with second or third trimester assessment, and 4 studies discovering BV in the third trimester only (which in most cases was at the time of labor).

In most studies, gestational age was determined by the best obstetric clinical estimate — using the date of the mother’s last menstrual period, detection of fetal heart tones, fundal heights, and obstetrical ultrasound. In one study, gestational age was determined only by pediatric assessment. Preterm delivery was evaluated as a dichotomous outcome in 18 of the 19 included studies. Two studies used 35 weeks’ gestation or less to define a preterm infant; one study used 36 weeks. All others adhered to the standard definition of preterm as any gestation with a duration of less than 37 weeks. Low birth weight was defined as an infant weighing less than 2500 grams [at birth] in all 6 studies that reported this outcome.

Of the 7 studies evaluating preterm premature rupture of membranes, 2 used 36 weeks as the cutoff for preterm; the other 5 used 37 weeks. Only 2 studies reported the method to determine membrane rupture (both used pH and ferning criteria). The time from rupture of membranes to labor onset varied from 1 to 6 hours in the 4 studies defining this period.

For the outcome of preterm onset of labor, all 9 studies defined preterm as gestational age less than 37 weeks. Most studies defined labor as regular painful uterine contractions; only 2 required cervical change. Two others considered treatment for preterm labor as the definition of preterm onset of labor.

Validity Assessment

Details about validity assessment and the effect of biases on the summary estimates of individual studies are available elsewhere.* Two biases were common. First was the misclassification of either the predictor or the outcomes, which tended to underestimate risk. Second was the issue of confounding variables, which tended to overestimate the odds ratio.

Data Synthesis

The Figure shows that women with BV were more likely to deliver a preterm infant (odds ratio, fixed effects model [OR<->FIXED<->] 1.85; 95% CI, 1.62-2.11) or an infant weighing less than 2500 grams (OR<->FIXED<-> 1.57; 95% CI, 1.32-1.87).

For the secondary outcomes of preterm premature rupture of membranes and preterm onset of labor, the resultant ORs<->FIXED <->were 1.83 (95% CI, 1.39-2.44) and 2.19 (95% CI, 1.73-2.76), respectively. The studies combined for preterm onset of labor met statistical requirements of homogeneity (P <\>>.25); those for preterm delivery, low birth weight, and preterm premature rupture of membranes did not. Recalculation of the odds ratio using the random effects model did not result in the loss of statistical significance for any of the main outcomes Table 2.

Pooling only cohort studies to generate a summary relative risk also resulted in a persistently elevated risk of prematurity for those mothers with BV, ranging from a 1.44- to a 2.86-fold increase Table 2, and homogeneity criteria were met for low birth weight, preterm premature rupture of membranes, and preterm onset of labor, (P <\>>.33) but not for preterm delivery. BV was significantly associated with preterm delivery in nearly all the subanalyses conducted Table 3.

Seven of the 18 studies evaluating preterm delivery did not perform regression analysis to evaluate for confounding; 4 others did this analysis but did not report an adjusted risk for BV. Thus, only 7 studies had controlled data available for a summary estimate. These studies, with their respective adjusted odds ratios and confounders considered, are listed in Table 5. As expected, the resultant summary estimate of the adjusted odds ratio was lower than that obtained from unadjusted data but remained significant clinically (OR = 1.60), as well as statistically (95% CI, 1.44-1.74).

Discussion

Our study pooled data representing more than 17,000 patients and the results show BV to be a significant risk factor for preterm and low birth weight deliveries. Additionally, BV is significantly associated with preterm onset of labor and preterm premature rupture of membranes. Summary relative risks calculated using cohort data only, although lower than the odds ratios, also showed a significant association between BV and all prematurity outcomes. Although often used interchangeably with relative risks, odds ratios tend to overestimate risk in cases of a positive association and a nonrare outcome. In our study, odds ratios exceeded relative risks for 3 of the 4 outcomes. This may reflect violation of the rare disease assumption or may simply be due to pooling a different subset of studies.

We believe the nearly twofold increase in prematurity with BV is especially robust for several reasons. First, the results are statistically significant regardless of the statistics used to generate them: odds ratio or relative risk, fixed or random effects models.

Second, decisions regarding data handling in our study were made to produce the most conservative estimate of association. For example, in case control studies that defined cases by the presence of preterm labor, those women with preterm onset of labor who gave birth at term were analyzed with the control group in the preterm delivery analysis. These women may be more likely to have BV, and moving them to the control group for the purpose of analysis would tend to underestimate the risk. In addition, for studies using vaginal cultures for diagnosing BV, we extracted data only for G vaginalis culture, which might lead to overdiagnosis, since the presence of G vaginalis can be a normal finding. This nondifferential error, as well as the potential imprecision of clinical estimates of gestational age, would tend to bias the summary estimate toward the null. Despite these potential underestimations, BV remained a significant risk factor for prematurity.

Finally, BV remained a significant risk for preterm delivery regardless of the subanalysis groupings. Pooling data from different populations with variable baseline risk or in different settings may lend confidence to the generalizability of these estimates. Caution is warranted, however, when drawing conclusions from any specific subanalysis.

All studies included in this analysis were observational in design, which raises 2 particular concerns: causality and confounding. Because neither the Mantel-Haenszel or DerSimonian and Laird methods for pooling risk estimates can incorporate confounding, we calculated separately the adjusted odds ratio pooled from risk estimates generated by regression analyses in individual studies. This summary of adjusted odds ratios still demonstrated a statistically significant elevated risk of 1.6 for preterm delivery in women with BV. This value may be overestimated, because 4 studies did not report regression results for the variable of BV. Presumably, these values were near the null but were likely not statistically significant, or they would have been reported. A lower summary adjusted odds ratio may have resulted were inclusion of these results possible.

Although causality cannot be proven by observational studies or by meta-analytic combination of such studies, several of the criteria suggesting causality are met.^57,58 The strength of the association is relatively small, ranging from 1.4 to 2.4. Although it is possible that confounders account for all this association, we think that is unlikely given that the controlled summary odds ratio remained significant. This meta-analysis did not address the dose response question directly, although some individual studies reported a stronger association with prematurity outcomes for those who had heavier colonization or higher BV scores. In all but one study this risk factor preceded the outcome, although in some cases vaginal assessment was done at the time of labor. Results are consistent, as can be seen from the summary figures: No study found a protective effect, and only one had a null value. The presence of BV makes biological sense as a contributor to preterm labor and thus to preterm premature rupture of membranes and preterm delivery. One proposed mechanism for this association is the production of phospholipase by the bacteria associated with BV. These enzymes can initiate prostaglandin synthesis, which is one step in the physiology of normal labor activation.⁵³ Infections in pregnancy are also associated with fetuses that were small for the gestational age,⁵⁹ which is another mechanism for low birth weight. Taken together, these factors lend support to a causal association for BV and prematurity.

Identifying BV in pregnancy as a modifiable risk factor for prematurity raises the obvious question of intervention. Accumulating evidence demonstrates that treating pregnant women who have BV with certain oral antibiotics can decrease the risk of prematurity. Clindamycin taken orally by pregnant women with BV decreased preterm deliveries and low birth weight infants by approximately 50%.^38,55 Hauth⁴² combined oral erythromycin and metronidazole and found a decreased rate in preterm births among a group of high-risk women with BV. Neither oral amoxicillin³¹ or intravaginal clindamycin³³ have been shown to affect pregnancy outcomes.

Limitations

As with any meta-analysis, one major limitation of this work is the appropriateness of combining results from different studies. Statistical homogeneity was met in 1 of the 4 analyses for summary odds ratios (pooling all studies possible), but in 3 of the 4 analyses when only cohort studies were pooled. This discrepancy suggests that study design was likely a key source of heterogeneity in this review. In addition to study design differences, we expected heterogeneity given the range of risks reported in individual studies, the varying population risks, the disparate methods, the timing of BV diagnosis, and the different definitions for the measured outcomes. One method for addressing heterogeneity is to use the random effects model, which accounts for variability between studies when estimating the precision of the risk. When we analyzed the data using this method, none of our conclusions changed.

A second limitation of this project is the possibility of publication bias. Our funnel plot reflects the absence of studies finding that BV protects pregnant women from delivering preterm infants. By chance alone, some studies may find this result, but it is unlikely that such a study would be published. Using Orwin’s method⁶⁰ to calculate a fail-safe N of 60 and our weighted summary effect size of 0.052 (number needed to harm [NNH] = 19), more than 75 studies showing no effect would be needed to drop the risk difference to 0.01 (NNH = 100). Given our systematic and complete search, we think it unlikely that publication bias accounts for our findings, despite the asymmetry of the funnel plot.

Implications for Future Research

The association of BV with prematurity remains at the disease-oriented level of evidence. Although neonatal and infant morbidity and mortality are increased with preterm delivery, the clinical impact of BV or its treatment on these patient-oriented outcomes remains unclear. Additionally, since approximately 50% of the pregnant women with BV are asymptomatic, some advocate universal screening for BV during pregnancy.⁶¹ A large randomized controlled trial evaluating patient-oriented health benefits as well as costs is warranted before this becomes a part of routine prenatal care.

Recommendations for clinical practice

BV in pregnancy is associated with a significant risk of preterm delivery. Evidence suggests that oral treatment with certain antibiotics can decrease this risk, especially in those with a previous preterm birth.⁶² Thus, if identified during pregnancy, BV should be treated. There is insufficient data to recommend screening for BV during pregnancy.

Acknowledgments

This project was done as part of Dr Flynn’s requirements for her master’s degree. She thanks the members of her thesis committee for their invaluable feedback. The authors also thank Shirley Killian for her assistance with article retrieval, Tangelia Pruitt for her assistance with bibliographic management, and Bill Grant for his statistical guidance.