Proton pump inhibitor side effects and drug interactions: Much ado about nothing?

The development and introduction of the first proton pump inhibitor (PPI), omeprazole (Prilosec), for the management of acid-peptic disorders marks one of the great success stories in gastroenterology. Until the latter part of the 20th century, complications of acid-peptic disease were among the most common problems faced in gastroenterology. Severe peptic strictures were once a highly prevalent cause of dysphagia, and operations for peptic ulcer disease were routinely learned by surgical trainees.

PPIs (Table 1) are substituted benzimidazole derivatives that have potent antisecretory effects on gastric acid. They block the terminal step in acid production by irreversibly inhibiting the function of the hydrogen-potassium adenosine triphosphatase present on the luminal aspect of parietal cell membranes in the stomach.

The success of these drugs, with sales total-ling $13.6 billion worldwide in 2009,¹ is not just a result of their potency and effectiveness in improving symptoms and complications of acid-peptic disease. Their safety among pharmacologic agents has been unparalleled. When the drugs were first introduced, their use was limited to short courses out of concern that gastric carcinoids could develop, but decades of use have not shown this issue to be of clinical relevance. Serious, acute adverse effects are also exceedingly uncommon.

However, recent reports have questioned the long-term safety of PPIs. Furthermore, these drugs are too often used in patients who have no valid indication for them,^2,3 exposing these patients to unnecessary risks.

The goals of this review are to analyze the recent literature about the risks of PPIs and to provide a rational approach for managing patients on PPI therapy in light of these concerns.

DO PPIs REDUCE THE EFFECT OF CLOPIDOGREL?

Clopidogrel (Plavix) is a potent antiplatelet agent commonly used in patients with atherosclerotic cardiac or cerebrovascular disease, sometimes in combination with aspirin. Because of the risk of significant gastrointestinal bleeding, a 2008 multisociety task force recommended prescribing a PPI when both clopidogrel and aspirin are used as dual antiplatelet therapy.⁴

However, recent studies have advanced the concern that cotherapy with a PPI reduces the effectiveness of clopidogrel. The pharmacologic basis for this interaction is that the two drugs share a common metabolic pathway (Figure 1). Clopidogrel is an inactive prodrug that requires cytochrome P450 (CYP) enzymes to become active. PPIs, on the other hand, are metabolized from their active to their inactive state by the CYP enzymes. PPIs competitively inhibit one of the principal enzymes, CYP2C19, important in the activation of clopidogrel.⁵ Pharmacodynamic studies have shown that adding a PPI to clopidogrel leads to reduced platelet inhibition.^6–9

Studies of clopidogrel plus PPIs: Discrepant results

This concern prompted epidemiologic studies to assess the clinical importance of this pharmacologic interaction (Table 2). Three large observational studies found a small but statistically significant increase in adverse cardiovascular outcomes in patients who were taking clopidogrel and a PPI compared with those who were taking clopidogrel without a PPI.^10–12 On the other hand, five studies of similar design found no significant increase in cardiovascular events.^9,13–16

To date, only one prospective randomized controlled trial has specifically investigated the effect of PPIs on cardiovascular outcomes in patients using clopidogrel. In this trial, patients on dual antiplatelet therapy with clopidogrel and aspirin were randomized to receive either omeprazole 20 mg or placebo. Analysis of the data revealed no significant increase in the composite end point of cardiovascular events (hazard ratio [HR] 0.99, 95% confidence interval [CI] 0.68–1.44, P = .96), but a statistically significant decrease in composite gastrointestinal events (HR 0.34, 95% CI 0.18–0.63, P < .001).¹⁷

Unfortunately, this trial had to be terminated before the prespecified sample size and duration of follow-up were reached because the study sponsor declared bankruptcy.

One additional recent retrospective cohort study¹⁶ likewise found no significant risk of serious cardiovascular disease related to PPI use in clopidogrel users. It also found that the adjusted incidence of hospitalization for upper gastrointestinal bleeding was 50% lower in patients who used PPIs than in those who did not (HR 0.50, 95% CI 0.39–0.65).

 

 

Do factors other than PPIs account for the higher risk in some of the studies?

The discrepant results of these studies suggest that the higher risk of cardiovascular events may be due, either completely or in part, to a factor other than the pharmacologic interaction of PPIs and clopidogrel. It is difficult to infer causality from the available data. In situations in which no randomized controlled trials exist, one looks to observational (case-control or cohort) studies to try to obtain the best estimate of the actual risk. With PPIs and clopidogrel, a randomized controlled trial was performed but terminated before patient enrollment was complete.

The increased risk found in some of these studies may be real, may be due to chance, or may even represent an increased risk from PPIs alone (although data do not support this possibility).¹⁸ However, the major concern in observational studies is the inability to account for unmeasured confounders, a problem virtually eliminated by randomization strategies in prospective studies.

In the studies that found a higher risk with the combination of omeprazole plus clopidogrel, the principal concern is confounding by indication, in which distortions of the risk estimates arise from an imbalance in prognostic factors between compared treatment groups that remains unmeasured.¹⁹ Stated another way, physicians who believed some patients to be “sicker” or to have a higher risk of serious events may have treated them with a PPI on the basis of factors that remained unaccounted for in the epidemiologic investigation.

This possibility has been reinforced by findings from a nonrandomized subgroup analysis of a randomized controlled trial in which patients who had been receiving a PPI had a higher rate of cardiovascular events whether they received clopidogrel or placebo.²⁰

FDA alert: Avoid using omeprazole or esomeprazole with clopidogrel

Nonetheless, on November 17, 2009, the US Food and Drug Administration (FDA) issued an alert to health care professionals and the public about the potential interaction between clopidogrel and omeprazole.²¹ In this alert, the FDA stated that the use of omeprazole or esomeprazole (Nexium) with clopidogrel should be avoided.

An algorithm to use when considering clopidogrel plus a PPI

Physicians are now left in a bind between the minimal, if any, pooled risk seen in the available data and the FDA recommendation. What is the best action to take?

To decide on a course of action when a patient is using or is being considered for therapy with clopidogrel plus a PPI, an algorithm that takes into account a few straightforward concerns can be suggested (Figure 2).

First, assess the need for dual antiplatelet therapy. If dual antiplatelet therapy (clopidogrel plus aspirin) is required, then a PPI is warranted for gastric protection because the risk of life-threatening bleeding outweighs any increased risk of cardiovascular events.⁴

If antiplatelet monotherapy (clopidogrel alone) is required, then assess the reason for antisecretory therapy.

For complicated disease, such as gastroesophageal reflux disease with Barrett esophagus or peptic strictures, PPI therapy is warranted to prevent progression or recurrence of complications. If the antisecretory therapy is being provided for noncomplicated symptomatic disorders such as nonerosive gastroesophageal reflux disease or dyspepsia, then one should try to “step down” the therapy by lowering the PPI dose as much as possible while still controlling symptoms to the patient’s tolerance, then possibly stepping further by substituting a histamine-2-receptor antagonist, an antacid, or “on-demand” use of PPIs.^22,23

However, if the rationale for antisecretory therapy is simply for gastrointestinal protection, then further risk stratification for gastro intestinal bleeding should be undertaken.⁴ For patients with a high risk of future gastrointestinal bleeding, such as those with prior episodes of bleeding or concurrent use of nonsteroidal anti-inflammatory drugs, antisecretory therapy is still recommended. Therefore, if a patient is on monotherapy with clopidogrel, has no complicated or symptomatic gastrointestinal disorder, and does not have a high risk of gastrointestinal bleeding, then therapy with a PPI should be reconsidered.

There are no strong data to indicate that one particular PPI should be used or avoided if one of the above criteria indicates the concurrent need for clopidogrel and a PPI. In their health alert about the potential interaction, the FDA did not issue the same warning for PPIs other than omeprazole and esomeprazole, but fell short of recommending a change to another PPI because of a lack of data to support or refute a similar interaction.

Because the half-lives of clopidogrel and PPIs are short, separating their administration could in theory decrease or eliminate the risk of competitive inhibition. The PPI could be given in the morning before breakfast and the clopidogrel could be given at night, or the clopidogrel could be given at lunchtime and the PPI before dinner. Although the FDA does not believe this strategy will reduce this interaction,²¹ one expert in the field has suggested it.¹⁸

 

DO PPIs CAUSE OSTEOPOROSIS, FRACTURES?

In the last several years, concern has been raised that PPIs could induce calcium malabsorption²⁴ and lead to osteoporosis, with a subsequent risk of fracture, especially hip fracture,^25–29 which can have devastating effects, including death (Table 3).^24–31

In a widely publicized paper published in 2006, Yang and colleagues²⁵ reported the results of a large nested case-control study in the United Kingdom. The risk of hip fracture was significantly greater in patients who had been using PPIs for at least 1 year than in those who had not. The risk appeared to increase with longer use and higher doses of PPIs.

A similar risk of hip fracture was seen in a larger Danish case-control study published the same year.²⁶ This study also found an increased odds ratio for PPI use in patients with spine fractures as well as in patients with any type of fracture. Interestingly, this study found a lower risk of fracture in patients using a histamine-2-receptor antagonist instead of a PPI.

Targownik et al²⁷ found that the risk of hip fracture was not significantly higher until after 5 years of PPI exposure, with an even stronger risk after 7 years.

However, the data on both association and causal relationship are not uniform.

The Women’s Health Initiative,³⁰ with more than 1 million person-years of followup, found no association between PPI use and hip fracture, but a modest association between PPI use and spine, arm, and wrist fractures, as well as total fractures.

A study in the United Kingdom found that patients without any major risk factors for hip fracture (defined by a risk ratio > 2) accounted for only 25% of cases but 53% of controls. When only these two average-risk groups were compared, the risk of hip fracture was similar in cases and controls.³¹

Corley et al³² also found that the risk of fracture associated with PPI use was only significant in the presence of another risk factor. These findings suggest that residual confounding may be to blame, at least in part, for the estimates of increased risk in the prior studies.

Another way to interpret these data is that PPIs increase the risk in patients at high risk to begin with, but not in those at average risk. This is an example of interaction (or effect modification) in which the risk is unequally distributed across groups with different characteristics.

A recently published study refutes the theory that impaired calcium absorption is responsible for the increase in fractures.³³ In this study, investigators queried the Manitoba Bone Mineral Density Database to determine the relationship between antisecretory therapy with PPIs and osteoporosis or loss of bone mineral density—and they found none. This study may support the theory that residual confounding is the reason for the finding of an increased risk, but it also leaves open the possibility that PPIs induce other changes in bone microstructure that could increase the risk of fracture.

FDA labeling: Possible risk of fracture with PPIs

Based on the data so far, it appears possible that there is a small, albeit statistically significant, association between PPI use and fracture risk. The association is indeed biologically plausible, but it remains to be seen if this association is clinically significant, as the risk is relatively low. Even though the studies had methodologic limitations, on May 25, 2010, the FDA announced a change in the required labeling information for PPIs to indicate a possible risk of fracture with these drugs.³⁴

Reassess the need for chronic PPI therapy

Although patients may worry that they will develop osteoporosis and fractures if they take PPIs, the data do not support a strong risk. Nevertheless, when faced with a patient on chronic PPI therapy, especially with a high dose, providers should use the opportunity to reassess the indication for the PPI to decide if chronic therapy is required, in a matter similar to the algorithm provided for PPI-clopidogrel cotherapy (FIGURE 2). Providers should educate patients about the data, and limit new and recurring PPI prescriptions to patients who require a PPI for appropriate indications, at the lowest dose, and for the shortest time possible.

DO PPIs INCREASE THE RISK OF PNEUMONIA?

Several recent studies have also raised concern about an association between PPI use and pneumonia.

Normally, the stomach remains free of bacteria (except for Helicobacter pylori) because its acidic milieu destroys nearly all bacteria swallowed. If the stomach becomes less acidic, it loses this protective mechanism, and ingested organisms can survive and proliferate.³⁵ In theory, when gastroesophageal reflux occurs, these bacteria could be carried up to the hypopharynx where microaspiration into the lower airways could lead to pneumonia, especially in patients with compromised oropharyngeal protective reflexes (eg, patients on mechanical ventilation).

This possible association came to the attention of the general medical community when a Dutch study,³⁶ in which 5,551 cases of community-acquired pneumonia developed in 364,683 people, found that the incidence of pneumonia was about 4.5 times higher in people exposed to acid-suppressive drugs (both PPIs and histamine-2-receptor antagonists) than in unexposed individuals. Patients who developed pneumonia also had higher odds of significant comorbid conditions, including heart failure and chronic obstructive pulmonary disease. The authors calculated that about one case of pneumonia per 226 patients treated with a PPI would be attributable to the PPI. A major limitation of this study, however, was that only 18% of the patients diagnosed with pneumonia actually had radiologic or microbiologic confirmation of pneumonia.

Other studies later examined the relationship between PPIs and community-acquired pneumonia,^37–41 and most have revealed a modestly higher risk of community-acquired pneumonia in patients exposed to PPIs.

This risk was confirmed in a recent metaanalysis, which found a higher risk of community-acquired pneumonia with PPI use (odds ratio 1.36, 95% CI 1.12–1.65).⁴² However, the authors refrained from drawing definitive conclusions from these data because of significant heterogeneity between the studies. One study³⁷ found that recent onset of use (within 7 days) had a much stronger association with community-acquired pneumonia than longer-term use, which is contradictory to a causal association, since longer-term use should lead to more cases of pneumonia.

Another study investigated the association between acid-suppressive drugs and hospital-acquired pneumonia in nonventilated patients.⁴³ In a 4-year period, there were 63,878 admissions in 42,093 unique patients. Acid-suppressive drugs were prescribed in 32,922 admissions (52%); the drugs included PPIs in 83% of these. Hospital-acquired pneumonia occurred in 2,219 admissions (3.5%), with a higher incidence in patients exposed to acid-suppressive drugs than in the unexposed group (4.6% vs 2.0%). The adjusted odds ratio for pneumonia was 1.3 (95% CI 1.1–1.4) in the exposed group. Subgroup analysis revealed that the association remained significant for PPIs but not for histamine-2-receptor antagonists.

Adequate studies of mechanically ventilated patients in the current era of intravenous PPI use are lacking. Older studies in this group of patients may not be generalizable to current practice because of the reduction in gastric volume with intravenous PPIs that may offset the theoretical risk of aspiration.³⁵

Although the data supporting the association are not exceedingly strong, the relationship is biologically plausible. If there is a risk, it seems to be greatest in the sickest patients, who can least afford to develop pneumonia. Therefore, prudent prescribing should be the rule for both inpatients and outpatients, especially in patients with comorbidities, in whom pneumonia could have serious consequences.

 

PPIs AND ENTERIC INFECTIONS

Traditionally, gastric acid was not believed to be important in protecting against Clostridium difficile infection because acid-resistant spores were presumed to be the principal vector of transmission.⁴⁴ Recently, this thought has been challenged, as several studies have found a higher risk of C difficile infection in PPI users. In theory, PPIs may increase the risk of C difficile infection by increasing the ability of the spore to convert to the vegetative form and to survive intraluminally.

A recent meta-analysis of 11 papers, including nearly 127,000 patients, found a significant relationship between PPI use and C difficile infection, with an odds ratio of 2.05 (95% CI 1.47–2.85).⁴⁵ Further supporting the hypothesis of a direct causative association, a recent study found a significant dose-response, with more aggressive acid-suppression associated with higher odds ratios.⁴⁶ In view of this association, patients using PPIs who develop diarrhea should be evaluated for C difficile, perhaps even in the absence of other risk factors.

Other enteric infections have been found to be associated with PPIs.^44,45 Small intestinal bacterial overgrowth, a condition that is associated with bloating, diarrhea, and malabsorption, has recently been associated with PPI use, although the significance of the association is uncertain.⁴⁷

Based on a change in the intestinal flora, recent reports have additionally implied that there is a relationship between PPI use and the development of spontaneous bacterial peritonitis in hospitalized cirrhotic patients with ascites. One study found a strong association (odds ratio 4.3, 95% CI 1.3–11.7) between PPIs and spontaneous bacterial pneumonitis,⁴⁸ whereas another study found no significant association (odds ratio 1.0, 95% CI 0.4–2.6).⁴⁹

Both studies were small case-control studies of hospitalized patients. No firm conclusion can be drawn about the relevance of this association from these investigations at this point.

PPIs AND ACUTE INTERSTITIAL NEPHRITIS

Several case reports have implicated PPIs as a cause of acute interstitial nephritis.

A systematic review from 2007 found 64 cases documented in the literature, 12 of which were considered certainly associated, and 9 of which were probably associated.⁵⁰ Initial symptoms were nonspecific and included nausea, malaise, and fever. With such extensive use worldwide as the denominator, the authors concluded that acute interstitial nephritis was a rare, idiosyncratic occurrence related to PPI use, but did not find enough evidence to support a causative relationship. Despite the rarity of the syndrome, they recommended maintaining a high level of clinical suspicion to detect acute interstitial nephritis early in its course, especially soon after the initiation of PPI therapy.

POSSIBLE ASSOCIATIONS WITH IRON AND B₁₂ DEFICIENCIES

Long-term PPI therapy has been thought to be associated with micronutrient deficiencies, especially of iron and vitamin B₁₂. Hydrochloric acid in the stomach assists in the dissociation of iron salts from food and the reduction of ferric iron to the more soluble ferrous iron.⁵¹ Gastric acid also facilitates the release of vitamin B₁₂ bound to proteins within ingested foodstuffs to permit binding to R-proteins for eventual absorption in the terminal ileum.^51,52

Despite the biologic plausibility of these deficiencies, there is currently little evidence to support a clinically relevant association to recommend a change in current practice.

NO THERAPY IS COMPLETELY WITHOUT RISK

Although concerns have been raised about the long-term safety of PPIs, the preponderance of the evidence does not strongly support the apprehensions publicized over the last few years. When translating these studies into the routine management of patients, it is important to recall some very basic tenets of good patient care.

No therapy is completely without risk—whether pharmacologic, surgical, or psychological, and no matter how benign or straightforward. Consequently, no drug, procedure, or treatment plan should be ordered without a valid indication. Even with an indication, the risk-benefit ratio of the therapy prescribed should always be considered. If the indication for the PPI is weak or uncertain, then even a slight risk tips the balance away from the drug, and the drug should be discontinued.

When seeing patients in long-term care, the indication and necessity for all drugs, including PPIs, should be reviewed. The algorithm proposed in Figure 2 can be adapted for virtually any of the possible associations.

Consider the indication for the PPI. Was the PPI started during a hospitalization and then routinely continued after discharge? This is one situation in which the use of a PPI could potentially be discontinued.²

For complicated acid-peptic disease, dose reduction or cessation of PPI therapy may not be possible.

If the PPI was prescribed only for symptom relief, as in cases of dyspepsia or nonerosive gastroesophageal reflux disease, reduce the dose of PPI to as low as possible to maintain symptom control. Should chronic therapy still be required, no specific monitoring is recommended, apart from routine monitoring that takes place in the course of patient care.

Lastly, because of the media attention that several of these concerns have garnered, patients may still harbor significant concerns about PPIs, even their short-term use. In such cases, the prescriber should take the opportunity to communicate the reason for the decision to prescribe the PPI, as well as the best available data about the risks PPIs may pose. None of these outcomes is very common in the absence of PPIs, with the possible exception of recurrent cardiovascular events, and the risks provided in all of these studies are relative to the baseline risk. Even if the risk of a particular outcome doubles with long-term PPI use, twice a small risk remains a small risk.

The development and introduction of the first proton pump inhibitor (PPI), omeprazole (Prilosec), for the management of acid-peptic disorders marks one of the great success stories in gastroenterology. Until the latter part of the 20th century, complications of acid-peptic disease were among the most common problems faced in gastroenterology. Severe peptic strictures were once a highly prevalent cause of dysphagia, and operations for peptic ulcer disease were routinely learned by surgical trainees.

PPIs (Table 1) are substituted benzimidazole derivatives that have potent antisecretory effects on gastric acid. They block the terminal step in acid production by irreversibly inhibiting the function of the hydrogen-potassium adenosine triphosphatase present on the luminal aspect of parietal cell membranes in the stomach.

The success of these drugs, with sales total-ling $13.6 billion worldwide in 2009,¹ is not just a result of their potency and effectiveness in improving symptoms and complications of acid-peptic disease. Their safety among pharmacologic agents has been unparalleled. When the drugs were first introduced, their use was limited to short courses out of concern that gastric carcinoids could develop, but decades of use have not shown this issue to be of clinical relevance. Serious, acute adverse effects are also exceedingly uncommon.

However, recent reports have questioned the long-term safety of PPIs. Furthermore, these drugs are too often used in patients who have no valid indication for them,^2,3 exposing these patients to unnecessary risks.

The goals of this review are to analyze the recent literature about the risks of PPIs and to provide a rational approach for managing patients on PPI therapy in light of these concerns.

DO PPIs REDUCE THE EFFECT OF CLOPIDOGREL?

Clopidogrel (Plavix) is a potent antiplatelet agent commonly used in patients with atherosclerotic cardiac or cerebrovascular disease, sometimes in combination with aspirin. Because of the risk of significant gastrointestinal bleeding, a 2008 multisociety task force recommended prescribing a PPI when both clopidogrel and aspirin are used as dual antiplatelet therapy.⁴

However, recent studies have advanced the concern that cotherapy with a PPI reduces the effectiveness of clopidogrel. The pharmacologic basis for this interaction is that the two drugs share a common metabolic pathway (Figure 1). Clopidogrel is an inactive prodrug that requires cytochrome P450 (CYP) enzymes to become active. PPIs, on the other hand, are metabolized from their active to their inactive state by the CYP enzymes. PPIs competitively inhibit one of the principal enzymes, CYP2C19, important in the activation of clopidogrel.⁵ Pharmacodynamic studies have shown that adding a PPI to clopidogrel leads to reduced platelet inhibition.^6–9

Studies of clopidogrel plus PPIs: Discrepant results

This concern prompted epidemiologic studies to assess the clinical importance of this pharmacologic interaction (Table 2). Three large observational studies found a small but statistically significant increase in adverse cardiovascular outcomes in patients who were taking clopidogrel and a PPI compared with those who were taking clopidogrel without a PPI.^10–12 On the other hand, five studies of similar design found no significant increase in cardiovascular events.^9,13–16

To date, only one prospective randomized controlled trial has specifically investigated the effect of PPIs on cardiovascular outcomes in patients using clopidogrel. In this trial, patients on dual antiplatelet therapy with clopidogrel and aspirin were randomized to receive either omeprazole 20 mg or placebo. Analysis of the data revealed no significant increase in the composite end point of cardiovascular events (hazard ratio [HR] 0.99, 95% confidence interval [CI] 0.68–1.44, P = .96), but a statistically significant decrease in composite gastrointestinal events (HR 0.34, 95% CI 0.18–0.63, P < .001).¹⁷

Unfortunately, this trial had to be terminated before the prespecified sample size and duration of follow-up were reached because the study sponsor declared bankruptcy.

One additional recent retrospective cohort study¹⁶ likewise found no significant risk of serious cardiovascular disease related to PPI use in clopidogrel users. It also found that the adjusted incidence of hospitalization for upper gastrointestinal bleeding was 50% lower in patients who used PPIs than in those who did not (HR 0.50, 95% CI 0.39–0.65).

 

 

Do factors other than PPIs account for the higher risk in some of the studies?

The discrepant results of these studies suggest that the higher risk of cardiovascular events may be due, either completely or in part, to a factor other than the pharmacologic interaction of PPIs and clopidogrel. It is difficult to infer causality from the available data. In situations in which no randomized controlled trials exist, one looks to observational (case-control or cohort) studies to try to obtain the best estimate of the actual risk. With PPIs and clopidogrel, a randomized controlled trial was performed but terminated before patient enrollment was complete.

The increased risk found in some of these studies may be real, may be due to chance, or may even represent an increased risk from PPIs alone (although data do not support this possibility).¹⁸ However, the major concern in observational studies is the inability to account for unmeasured confounders, a problem virtually eliminated by randomization strategies in prospective studies.

In the studies that found a higher risk with the combination of omeprazole plus clopidogrel, the principal concern is confounding by indication, in which distortions of the risk estimates arise from an imbalance in prognostic factors between compared treatment groups that remains unmeasured.¹⁹ Stated another way, physicians who believed some patients to be “sicker” or to have a higher risk of serious events may have treated them with a PPI on the basis of factors that remained unaccounted for in the epidemiologic investigation.

This possibility has been reinforced by findings from a nonrandomized subgroup analysis of a randomized controlled trial in which patients who had been receiving a PPI had a higher rate of cardiovascular events whether they received clopidogrel or placebo.²⁰

FDA alert: Avoid using omeprazole or esomeprazole with clopidogrel

Nonetheless, on November 17, 2009, the US Food and Drug Administration (FDA) issued an alert to health care professionals and the public about the potential interaction between clopidogrel and omeprazole.²¹ In this alert, the FDA stated that the use of omeprazole or esomeprazole (Nexium) with clopidogrel should be avoided.

An algorithm to use when considering clopidogrel plus a PPI

Physicians are now left in a bind between the minimal, if any, pooled risk seen in the available data and the FDA recommendation. What is the best action to take?

To decide on a course of action when a patient is using or is being considered for therapy with clopidogrel plus a PPI, an algorithm that takes into account a few straightforward concerns can be suggested (Figure 2).

First, assess the need for dual antiplatelet therapy. If dual antiplatelet therapy (clopidogrel plus aspirin) is required, then a PPI is warranted for gastric protection because the risk of life-threatening bleeding outweighs any increased risk of cardiovascular events.⁴

If antiplatelet monotherapy (clopidogrel alone) is required, then assess the reason for antisecretory therapy.

For complicated disease, such as gastroesophageal reflux disease with Barrett esophagus or peptic strictures, PPI therapy is warranted to prevent progression or recurrence of complications. If the antisecretory therapy is being provided for noncomplicated symptomatic disorders such as nonerosive gastroesophageal reflux disease or dyspepsia, then one should try to “step down” the therapy by lowering the PPI dose as much as possible while still controlling symptoms to the patient’s tolerance, then possibly stepping further by substituting a histamine-2-receptor antagonist, an antacid, or “on-demand” use of PPIs.^22,23

However, if the rationale for antisecretory therapy is simply for gastrointestinal protection, then further risk stratification for gastro intestinal bleeding should be undertaken.⁴ For patients with a high risk of future gastrointestinal bleeding, such as those with prior episodes of bleeding or concurrent use of nonsteroidal anti-inflammatory drugs, antisecretory therapy is still recommended. Therefore, if a patient is on monotherapy with clopidogrel, has no complicated or symptomatic gastrointestinal disorder, and does not have a high risk of gastrointestinal bleeding, then therapy with a PPI should be reconsidered.

There are no strong data to indicate that one particular PPI should be used or avoided if one of the above criteria indicates the concurrent need for clopidogrel and a PPI. In their health alert about the potential interaction, the FDA did not issue the same warning for PPIs other than omeprazole and esomeprazole, but fell short of recommending a change to another PPI because of a lack of data to support or refute a similar interaction.

Because the half-lives of clopidogrel and PPIs are short, separating their administration could in theory decrease or eliminate the risk of competitive inhibition. The PPI could be given in the morning before breakfast and the clopidogrel could be given at night, or the clopidogrel could be given at lunchtime and the PPI before dinner. Although the FDA does not believe this strategy will reduce this interaction,²¹ one expert in the field has suggested it.¹⁸

 

DO PPIs CAUSE OSTEOPOROSIS, FRACTURES?

In the last several years, concern has been raised that PPIs could induce calcium malabsorption²⁴ and lead to osteoporosis, with a subsequent risk of fracture, especially hip fracture,^25–29 which can have devastating effects, including death (Table 3).^24–31

In a widely publicized paper published in 2006, Yang and colleagues²⁵ reported the results of a large nested case-control study in the United Kingdom. The risk of hip fracture was significantly greater in patients who had been using PPIs for at least 1 year than in those who had not. The risk appeared to increase with longer use and higher doses of PPIs.

A similar risk of hip fracture was seen in a larger Danish case-control study published the same year.²⁶ This study also found an increased odds ratio for PPI use in patients with spine fractures as well as in patients with any type of fracture. Interestingly, this study found a lower risk of fracture in patients using a histamine-2-receptor antagonist instead of a PPI.

Targownik et al²⁷ found that the risk of hip fracture was not significantly higher until after 5 years of PPI exposure, with an even stronger risk after 7 years.

However, the data on both association and causal relationship are not uniform.

The Women’s Health Initiative,³⁰ with more than 1 million person-years of followup, found no association between PPI use and hip fracture, but a modest association between PPI use and spine, arm, and wrist fractures, as well as total fractures.

A study in the United Kingdom found that patients without any major risk factors for hip fracture (defined by a risk ratio > 2) accounted for only 25% of cases but 53% of controls. When only these two average-risk groups were compared, the risk of hip fracture was similar in cases and controls.³¹

Corley et al³² also found that the risk of fracture associated with PPI use was only significant in the presence of another risk factor. These findings suggest that residual confounding may be to blame, at least in part, for the estimates of increased risk in the prior studies.

Another way to interpret these data is that PPIs increase the risk in patients at high risk to begin with, but not in those at average risk. This is an example of interaction (or effect modification) in which the risk is unequally distributed across groups with different characteristics.

A recently published study refutes the theory that impaired calcium absorption is responsible for the increase in fractures.³³ In this study, investigators queried the Manitoba Bone Mineral Density Database to determine the relationship between antisecretory therapy with PPIs and osteoporosis or loss of bone mineral density—and they found none. This study may support the theory that residual confounding is the reason for the finding of an increased risk, but it also leaves open the possibility that PPIs induce other changes in bone microstructure that could increase the risk of fracture.

FDA labeling: Possible risk of fracture with PPIs

Based on the data so far, it appears possible that there is a small, albeit statistically significant, association between PPI use and fracture risk. The association is indeed biologically plausible, but it remains to be seen if this association is clinically significant, as the risk is relatively low. Even though the studies had methodologic limitations, on May 25, 2010, the FDA announced a change in the required labeling information for PPIs to indicate a possible risk of fracture with these drugs.³⁴

Reassess the need for chronic PPI therapy

Although patients may worry that they will develop osteoporosis and fractures if they take PPIs, the data do not support a strong risk. Nevertheless, when faced with a patient on chronic PPI therapy, especially with a high dose, providers should use the opportunity to reassess the indication for the PPI to decide if chronic therapy is required, in a matter similar to the algorithm provided for PPI-clopidogrel cotherapy (FIGURE 2). Providers should educate patients about the data, and limit new and recurring PPI prescriptions to patients who require a PPI for appropriate indications, at the lowest dose, and for the shortest time possible.

DO PPIs INCREASE THE RISK OF PNEUMONIA?

Several recent studies have also raised concern about an association between PPI use and pneumonia.

Normally, the stomach remains free of bacteria (except for Helicobacter pylori) because its acidic milieu destroys nearly all bacteria swallowed. If the stomach becomes less acidic, it loses this protective mechanism, and ingested organisms can survive and proliferate.³⁵ In theory, when gastroesophageal reflux occurs, these bacteria could be carried up to the hypopharynx where microaspiration into the lower airways could lead to pneumonia, especially in patients with compromised oropharyngeal protective reflexes (eg, patients on mechanical ventilation).

This possible association came to the attention of the general medical community when a Dutch study,³⁶ in which 5,551 cases of community-acquired pneumonia developed in 364,683 people, found that the incidence of pneumonia was about 4.5 times higher in people exposed to acid-suppressive drugs (both PPIs and histamine-2-receptor antagonists) than in unexposed individuals. Patients who developed pneumonia also had higher odds of significant comorbid conditions, including heart failure and chronic obstructive pulmonary disease. The authors calculated that about one case of pneumonia per 226 patients treated with a PPI would be attributable to the PPI. A major limitation of this study, however, was that only 18% of the patients diagnosed with pneumonia actually had radiologic or microbiologic confirmation of pneumonia.

Other studies later examined the relationship between PPIs and community-acquired pneumonia,^37–41 and most have revealed a modestly higher risk of community-acquired pneumonia in patients exposed to PPIs.

This risk was confirmed in a recent metaanalysis, which found a higher risk of community-acquired pneumonia with PPI use (odds ratio 1.36, 95% CI 1.12–1.65).⁴² However, the authors refrained from drawing definitive conclusions from these data because of significant heterogeneity between the studies. One study³⁷ found that recent onset of use (within 7 days) had a much stronger association with community-acquired pneumonia than longer-term use, which is contradictory to a causal association, since longer-term use should lead to more cases of pneumonia.

Another study investigated the association between acid-suppressive drugs and hospital-acquired pneumonia in nonventilated patients.⁴³ In a 4-year period, there were 63,878 admissions in 42,093 unique patients. Acid-suppressive drugs were prescribed in 32,922 admissions (52%); the drugs included PPIs in 83% of these. Hospital-acquired pneumonia occurred in 2,219 admissions (3.5%), with a higher incidence in patients exposed to acid-suppressive drugs than in the unexposed group (4.6% vs 2.0%). The adjusted odds ratio for pneumonia was 1.3 (95% CI 1.1–1.4) in the exposed group. Subgroup analysis revealed that the association remained significant for PPIs but not for histamine-2-receptor antagonists.

Adequate studies of mechanically ventilated patients in the current era of intravenous PPI use are lacking. Older studies in this group of patients may not be generalizable to current practice because of the reduction in gastric volume with intravenous PPIs that may offset the theoretical risk of aspiration.³⁵

Although the data supporting the association are not exceedingly strong, the relationship is biologically plausible. If there is a risk, it seems to be greatest in the sickest patients, who can least afford to develop pneumonia. Therefore, prudent prescribing should be the rule for both inpatients and outpatients, especially in patients with comorbidities, in whom pneumonia could have serious consequences.

 

PPIs AND ENTERIC INFECTIONS

Traditionally, gastric acid was not believed to be important in protecting against Clostridium difficile infection because acid-resistant spores were presumed to be the principal vector of transmission.⁴⁴ Recently, this thought has been challenged, as several studies have found a higher risk of C difficile infection in PPI users. In theory, PPIs may increase the risk of C difficile infection by increasing the ability of the spore to convert to the vegetative form and to survive intraluminally.

A recent meta-analysis of 11 papers, including nearly 127,000 patients, found a significant relationship between PPI use and C difficile infection, with an odds ratio of 2.05 (95% CI 1.47–2.85).⁴⁵ Further supporting the hypothesis of a direct causative association, a recent study found a significant dose-response, with more aggressive acid-suppression associated with higher odds ratios.⁴⁶ In view of this association, patients using PPIs who develop diarrhea should be evaluated for C difficile, perhaps even in the absence of other risk factors.

Other enteric infections have been found to be associated with PPIs.^44,45 Small intestinal bacterial overgrowth, a condition that is associated with bloating, diarrhea, and malabsorption, has recently been associated with PPI use, although the significance of the association is uncertain.⁴⁷

Based on a change in the intestinal flora, recent reports have additionally implied that there is a relationship between PPI use and the development of spontaneous bacterial peritonitis in hospitalized cirrhotic patients with ascites. One study found a strong association (odds ratio 4.3, 95% CI 1.3–11.7) between PPIs and spontaneous bacterial pneumonitis,⁴⁸ whereas another study found no significant association (odds ratio 1.0, 95% CI 0.4–2.6).⁴⁹

Both studies were small case-control studies of hospitalized patients. No firm conclusion can be drawn about the relevance of this association from these investigations at this point.

PPIs AND ACUTE INTERSTITIAL NEPHRITIS

Several case reports have implicated PPIs as a cause of acute interstitial nephritis.

A systematic review from 2007 found 64 cases documented in the literature, 12 of which were considered certainly associated, and 9 of which were probably associated.⁵⁰ Initial symptoms were nonspecific and included nausea, malaise, and fever. With such extensive use worldwide as the denominator, the authors concluded that acute interstitial nephritis was a rare, idiosyncratic occurrence related to PPI use, but did not find enough evidence to support a causative relationship. Despite the rarity of the syndrome, they recommended maintaining a high level of clinical suspicion to detect acute interstitial nephritis early in its course, especially soon after the initiation of PPI therapy.

POSSIBLE ASSOCIATIONS WITH IRON AND B₁₂ DEFICIENCIES

Long-term PPI therapy has been thought to be associated with micronutrient deficiencies, especially of iron and vitamin B₁₂. Hydrochloric acid in the stomach assists in the dissociation of iron salts from food and the reduction of ferric iron to the more soluble ferrous iron.⁵¹ Gastric acid also facilitates the release of vitamin B₁₂ bound to proteins within ingested foodstuffs to permit binding to R-proteins for eventual absorption in the terminal ileum.^51,52

Despite the biologic plausibility of these deficiencies, there is currently little evidence to support a clinically relevant association to recommend a change in current practice.

NO THERAPY IS COMPLETELY WITHOUT RISK

Although concerns have been raised about the long-term safety of PPIs, the preponderance of the evidence does not strongly support the apprehensions publicized over the last few years. When translating these studies into the routine management of patients, it is important to recall some very basic tenets of good patient care.

No therapy is completely without risk—whether pharmacologic, surgical, or psychological, and no matter how benign or straightforward. Consequently, no drug, procedure, or treatment plan should be ordered without a valid indication. Even with an indication, the risk-benefit ratio of the therapy prescribed should always be considered. If the indication for the PPI is weak or uncertain, then even a slight risk tips the balance away from the drug, and the drug should be discontinued.

When seeing patients in long-term care, the indication and necessity for all drugs, including PPIs, should be reviewed. The algorithm proposed in Figure 2 can be adapted for virtually any of the possible associations.

Consider the indication for the PPI. Was the PPI started during a hospitalization and then routinely continued after discharge? This is one situation in which the use of a PPI could potentially be discontinued.²

For complicated acid-peptic disease, dose reduction or cessation of PPI therapy may not be possible.

If the PPI was prescribed only for symptom relief, as in cases of dyspepsia or nonerosive gastroesophageal reflux disease, reduce the dose of PPI to as low as possible to maintain symptom control. Should chronic therapy still be required, no specific monitoring is recommended, apart from routine monitoring that takes place in the course of patient care.

Lastly, because of the media attention that several of these concerns have garnered, patients may still harbor significant concerns about PPIs, even their short-term use. In such cases, the prescriber should take the opportunity to communicate the reason for the decision to prescribe the PPI, as well as the best available data about the risks PPIs may pose. None of these outcomes is very common in the absence of PPIs, with the possible exception of recurrent cardiovascular events, and the risks provided in all of these studies are relative to the baseline risk. Even if the risk of a particular outcome doubles with long-term PPI use, twice a small risk remains a small risk.

The development and introduction of the first proton pump inhibitor (PPI), omeprazole (Prilosec), for the management of acid-peptic disorders marks one of the great success stories in gastroenterology. Until the latter part of the 20th century, complications of acid-peptic disease were among the most common problems faced in gastroenterology. Severe peptic strictures were once a highly prevalent cause of dysphagia, and operations for peptic ulcer disease were routinely learned by surgical trainees.

PPIs (Table 1) are substituted benzimidazole derivatives that have potent antisecretory effects on gastric acid. They block the terminal step in acid production by irreversibly inhibiting the function of the hydrogen-potassium adenosine triphosphatase present on the luminal aspect of parietal cell membranes in the stomach.

The success of these drugs, with sales total-ling $13.6 billion worldwide in 2009,¹ is not just a result of their potency and effectiveness in improving symptoms and complications of acid-peptic disease. Their safety among pharmacologic agents has been unparalleled. When the drugs were first introduced, their use was limited to short courses out of concern that gastric carcinoids could develop, but decades of use have not shown this issue to be of clinical relevance. Serious, acute adverse effects are also exceedingly uncommon.

However, recent reports have questioned the long-term safety of PPIs. Furthermore, these drugs are too often used in patients who have no valid indication for them,^2,3 exposing these patients to unnecessary risks.

The goals of this review are to analyze the recent literature about the risks of PPIs and to provide a rational approach for managing patients on PPI therapy in light of these concerns.

DO PPIs REDUCE THE EFFECT OF CLOPIDOGREL?

Clopidogrel (Plavix) is a potent antiplatelet agent commonly used in patients with atherosclerotic cardiac or cerebrovascular disease, sometimes in combination with aspirin. Because of the risk of significant gastrointestinal bleeding, a 2008 multisociety task force recommended prescribing a PPI when both clopidogrel and aspirin are used as dual antiplatelet therapy.⁴

However, recent studies have advanced the concern that cotherapy with a PPI reduces the effectiveness of clopidogrel. The pharmacologic basis for this interaction is that the two drugs share a common metabolic pathway (Figure 1). Clopidogrel is an inactive prodrug that requires cytochrome P450 (CYP) enzymes to become active. PPIs, on the other hand, are metabolized from their active to their inactive state by the CYP enzymes. PPIs competitively inhibit one of the principal enzymes, CYP2C19, important in the activation of clopidogrel.⁵ Pharmacodynamic studies have shown that adding a PPI to clopidogrel leads to reduced platelet inhibition.^6–9

Studies of clopidogrel plus PPIs: Discrepant results

This concern prompted epidemiologic studies to assess the clinical importance of this pharmacologic interaction (Table 2). Three large observational studies found a small but statistically significant increase in adverse cardiovascular outcomes in patients who were taking clopidogrel and a PPI compared with those who were taking clopidogrel without a PPI.^10–12 On the other hand, five studies of similar design found no significant increase in cardiovascular events.^9,13–16

To date, only one prospective randomized controlled trial has specifically investigated the effect of PPIs on cardiovascular outcomes in patients using clopidogrel. In this trial, patients on dual antiplatelet therapy with clopidogrel and aspirin were randomized to receive either omeprazole 20 mg or placebo. Analysis of the data revealed no significant increase in the composite end point of cardiovascular events (hazard ratio [HR] 0.99, 95% confidence interval [CI] 0.68–1.44, P = .96), but a statistically significant decrease in composite gastrointestinal events (HR 0.34, 95% CI 0.18–0.63, P < .001).¹⁷

Unfortunately, this trial had to be terminated before the prespecified sample size and duration of follow-up were reached because the study sponsor declared bankruptcy.

One additional recent retrospective cohort study¹⁶ likewise found no significant risk of serious cardiovascular disease related to PPI use in clopidogrel users. It also found that the adjusted incidence of hospitalization for upper gastrointestinal bleeding was 50% lower in patients who used PPIs than in those who did not (HR 0.50, 95% CI 0.39–0.65).

 

 

Do factors other than PPIs account for the higher risk in some of the studies?

The discrepant results of these studies suggest that the higher risk of cardiovascular events may be due, either completely or in part, to a factor other than the pharmacologic interaction of PPIs and clopidogrel. It is difficult to infer causality from the available data. In situations in which no randomized controlled trials exist, one looks to observational (case-control or cohort) studies to try to obtain the best estimate of the actual risk. With PPIs and clopidogrel, a randomized controlled trial was performed but terminated before patient enrollment was complete.

The increased risk found in some of these studies may be real, may be due to chance, or may even represent an increased risk from PPIs alone (although data do not support this possibility).¹⁸ However, the major concern in observational studies is the inability to account for unmeasured confounders, a problem virtually eliminated by randomization strategies in prospective studies.

In the studies that found a higher risk with the combination of omeprazole plus clopidogrel, the principal concern is confounding by indication, in which distortions of the risk estimates arise from an imbalance in prognostic factors between compared treatment groups that remains unmeasured.¹⁹ Stated another way, physicians who believed some patients to be “sicker” or to have a higher risk of serious events may have treated them with a PPI on the basis of factors that remained unaccounted for in the epidemiologic investigation.

This possibility has been reinforced by findings from a nonrandomized subgroup analysis of a randomized controlled trial in which patients who had been receiving a PPI had a higher rate of cardiovascular events whether they received clopidogrel or placebo.²⁰

FDA alert: Avoid using omeprazole or esomeprazole with clopidogrel

Nonetheless, on November 17, 2009, the US Food and Drug Administration (FDA) issued an alert to health care professionals and the public about the potential interaction between clopidogrel and omeprazole.²¹ In this alert, the FDA stated that the use of omeprazole or esomeprazole (Nexium) with clopidogrel should be avoided.

An algorithm to use when considering clopidogrel plus a PPI

Physicians are now left in a bind between the minimal, if any, pooled risk seen in the available data and the FDA recommendation. What is the best action to take?

To decide on a course of action when a patient is using or is being considered for therapy with clopidogrel plus a PPI, an algorithm that takes into account a few straightforward concerns can be suggested (Figure 2).

First, assess the need for dual antiplatelet therapy. If dual antiplatelet therapy (clopidogrel plus aspirin) is required, then a PPI is warranted for gastric protection because the risk of life-threatening bleeding outweighs any increased risk of cardiovascular events.⁴

If antiplatelet monotherapy (clopidogrel alone) is required, then assess the reason for antisecretory therapy.

For complicated disease, such as gastroesophageal reflux disease with Barrett esophagus or peptic strictures, PPI therapy is warranted to prevent progression or recurrence of complications. If the antisecretory therapy is being provided for noncomplicated symptomatic disorders such as nonerosive gastroesophageal reflux disease or dyspepsia, then one should try to “step down” the therapy by lowering the PPI dose as much as possible while still controlling symptoms to the patient’s tolerance, then possibly stepping further by substituting a histamine-2-receptor antagonist, an antacid, or “on-demand” use of PPIs.^22,23

However, if the rationale for antisecretory therapy is simply for gastrointestinal protection, then further risk stratification for gastro intestinal bleeding should be undertaken.⁴ For patients with a high risk of future gastrointestinal bleeding, such as those with prior episodes of bleeding or concurrent use of nonsteroidal anti-inflammatory drugs, antisecretory therapy is still recommended. Therefore, if a patient is on monotherapy with clopidogrel, has no complicated or symptomatic gastrointestinal disorder, and does not have a high risk of gastrointestinal bleeding, then therapy with a PPI should be reconsidered.

There are no strong data to indicate that one particular PPI should be used or avoided if one of the above criteria indicates the concurrent need for clopidogrel and a PPI. In their health alert about the potential interaction, the FDA did not issue the same warning for PPIs other than omeprazole and esomeprazole, but fell short of recommending a change to another PPI because of a lack of data to support or refute a similar interaction.

Because the half-lives of clopidogrel and PPIs are short, separating their administration could in theory decrease or eliminate the risk of competitive inhibition. The PPI could be given in the morning before breakfast and the clopidogrel could be given at night, or the clopidogrel could be given at lunchtime and the PPI before dinner. Although the FDA does not believe this strategy will reduce this interaction,²¹ one expert in the field has suggested it.¹⁸

 

DO PPIs CAUSE OSTEOPOROSIS, FRACTURES?

In the last several years, concern has been raised that PPIs could induce calcium malabsorption²⁴ and lead to osteoporosis, with a subsequent risk of fracture, especially hip fracture,^25–29 which can have devastating effects, including death (Table 3).^24–31

In a widely publicized paper published in 2006, Yang and colleagues²⁵ reported the results of a large nested case-control study in the United Kingdom. The risk of hip fracture was significantly greater in patients who had been using PPIs for at least 1 year than in those who had not. The risk appeared to increase with longer use and higher doses of PPIs.

A similar risk of hip fracture was seen in a larger Danish case-control study published the same year.²⁶ This study also found an increased odds ratio for PPI use in patients with spine fractures as well as in patients with any type of fracture. Interestingly, this study found a lower risk of fracture in patients using a histamine-2-receptor antagonist instead of a PPI.

Targownik et al²⁷ found that the risk of hip fracture was not significantly higher until after 5 years of PPI exposure, with an even stronger risk after 7 years.

However, the data on both association and causal relationship are not uniform.

The Women’s Health Initiative,³⁰ with more than 1 million person-years of followup, found no association between PPI use and hip fracture, but a modest association between PPI use and spine, arm, and wrist fractures, as well as total fractures.

A study in the United Kingdom found that patients without any major risk factors for hip fracture (defined by a risk ratio > 2) accounted for only 25% of cases but 53% of controls. When only these two average-risk groups were compared, the risk of hip fracture was similar in cases and controls.³¹

Corley et al³² also found that the risk of fracture associated with PPI use was only significant in the presence of another risk factor. These findings suggest that residual confounding may be to blame, at least in part, for the estimates of increased risk in the prior studies.

Another way to interpret these data is that PPIs increase the risk in patients at high risk to begin with, but not in those at average risk. This is an example of interaction (or effect modification) in which the risk is unequally distributed across groups with different characteristics.

A recently published study refutes the theory that impaired calcium absorption is responsible for the increase in fractures.³³ In this study, investigators queried the Manitoba Bone Mineral Density Database to determine the relationship between antisecretory therapy with PPIs and osteoporosis or loss of bone mineral density—and they found none. This study may support the theory that residual confounding is the reason for the finding of an increased risk, but it also leaves open the possibility that PPIs induce other changes in bone microstructure that could increase the risk of fracture.

FDA labeling: Possible risk of fracture with PPIs

Based on the data so far, it appears possible that there is a small, albeit statistically significant, association between PPI use and fracture risk. The association is indeed biologically plausible, but it remains to be seen if this association is clinically significant, as the risk is relatively low. Even though the studies had methodologic limitations, on May 25, 2010, the FDA announced a change in the required labeling information for PPIs to indicate a possible risk of fracture with these drugs.³⁴

Reassess the need for chronic PPI therapy

Although patients may worry that they will develop osteoporosis and fractures if they take PPIs, the data do not support a strong risk. Nevertheless, when faced with a patient on chronic PPI therapy, especially with a high dose, providers should use the opportunity to reassess the indication for the PPI to decide if chronic therapy is required, in a matter similar to the algorithm provided for PPI-clopidogrel cotherapy (FIGURE 2). Providers should educate patients about the data, and limit new and recurring PPI prescriptions to patients who require a PPI for appropriate indications, at the lowest dose, and for the shortest time possible.

DO PPIs INCREASE THE RISK OF PNEUMONIA?

Several recent studies have also raised concern about an association between PPI use and pneumonia.

Normally, the stomach remains free of bacteria (except for Helicobacter pylori) because its acidic milieu destroys nearly all bacteria swallowed. If the stomach becomes less acidic, it loses this protective mechanism, and ingested organisms can survive and proliferate.³⁵ In theory, when gastroesophageal reflux occurs, these bacteria could be carried up to the hypopharynx where microaspiration into the lower airways could lead to pneumonia, especially in patients with compromised oropharyngeal protective reflexes (eg, patients on mechanical ventilation).

This possible association came to the attention of the general medical community when a Dutch study,³⁶ in which 5,551 cases of community-acquired pneumonia developed in 364,683 people, found that the incidence of pneumonia was about 4.5 times higher in people exposed to acid-suppressive drugs (both PPIs and histamine-2-receptor antagonists) than in unexposed individuals. Patients who developed pneumonia also had higher odds of significant comorbid conditions, including heart failure and chronic obstructive pulmonary disease. The authors calculated that about one case of pneumonia per 226 patients treated with a PPI would be attributable to the PPI. A major limitation of this study, however, was that only 18% of the patients diagnosed with pneumonia actually had radiologic or microbiologic confirmation of pneumonia.

Other studies later examined the relationship between PPIs and community-acquired pneumonia,^37–41 and most have revealed a modestly higher risk of community-acquired pneumonia in patients exposed to PPIs.

This risk was confirmed in a recent metaanalysis, which found a higher risk of community-acquired pneumonia with PPI use (odds ratio 1.36, 95% CI 1.12–1.65).⁴² However, the authors refrained from drawing definitive conclusions from these data because of significant heterogeneity between the studies. One study³⁷ found that recent onset of use (within 7 days) had a much stronger association with community-acquired pneumonia than longer-term use, which is contradictory to a causal association, since longer-term use should lead to more cases of pneumonia.

Another study investigated the association between acid-suppressive drugs and hospital-acquired pneumonia in nonventilated patients.⁴³ In a 4-year period, there were 63,878 admissions in 42,093 unique patients. Acid-suppressive drugs were prescribed in 32,922 admissions (52%); the drugs included PPIs in 83% of these. Hospital-acquired pneumonia occurred in 2,219 admissions (3.5%), with a higher incidence in patients exposed to acid-suppressive drugs than in the unexposed group (4.6% vs 2.0%). The adjusted odds ratio for pneumonia was 1.3 (95% CI 1.1–1.4) in the exposed group. Subgroup analysis revealed that the association remained significant for PPIs but not for histamine-2-receptor antagonists.

Adequate studies of mechanically ventilated patients in the current era of intravenous PPI use are lacking. Older studies in this group of patients may not be generalizable to current practice because of the reduction in gastric volume with intravenous PPIs that may offset the theoretical risk of aspiration.³⁵

Although the data supporting the association are not exceedingly strong, the relationship is biologically plausible. If there is a risk, it seems to be greatest in the sickest patients, who can least afford to develop pneumonia. Therefore, prudent prescribing should be the rule for both inpatients and outpatients, especially in patients with comorbidities, in whom pneumonia could have serious consequences.

 

PPIs AND ENTERIC INFECTIONS

Traditionally, gastric acid was not believed to be important in protecting against Clostridium difficile infection because acid-resistant spores were presumed to be the principal vector of transmission.⁴⁴ Recently, this thought has been challenged, as several studies have found a higher risk of C difficile infection in PPI users. In theory, PPIs may increase the risk of C difficile infection by increasing the ability of the spore to convert to the vegetative form and to survive intraluminally.

A recent meta-analysis of 11 papers, including nearly 127,000 patients, found a significant relationship between PPI use and C difficile infection, with an odds ratio of 2.05 (95% CI 1.47–2.85).⁴⁵ Further supporting the hypothesis of a direct causative association, a recent study found a significant dose-response, with more aggressive acid-suppression associated with higher odds ratios.⁴⁶ In view of this association, patients using PPIs who develop diarrhea should be evaluated for C difficile, perhaps even in the absence of other risk factors.

Other enteric infections have been found to be associated with PPIs.^44,45 Small intestinal bacterial overgrowth, a condition that is associated with bloating, diarrhea, and malabsorption, has recently been associated with PPI use, although the significance of the association is uncertain.⁴⁷

Based on a change in the intestinal flora, recent reports have additionally implied that there is a relationship between PPI use and the development of spontaneous bacterial peritonitis in hospitalized cirrhotic patients with ascites. One study found a strong association (odds ratio 4.3, 95% CI 1.3–11.7) between PPIs and spontaneous bacterial pneumonitis,⁴⁸ whereas another study found no significant association (odds ratio 1.0, 95% CI 0.4–2.6).⁴⁹

Both studies were small case-control studies of hospitalized patients. No firm conclusion can be drawn about the relevance of this association from these investigations at this point.

PPIs AND ACUTE INTERSTITIAL NEPHRITIS

Several case reports have implicated PPIs as a cause of acute interstitial nephritis.

A systematic review from 2007 found 64 cases documented in the literature, 12 of which were considered certainly associated, and 9 of which were probably associated.⁵⁰ Initial symptoms were nonspecific and included nausea, malaise, and fever. With such extensive use worldwide as the denominator, the authors concluded that acute interstitial nephritis was a rare, idiosyncratic occurrence related to PPI use, but did not find enough evidence to support a causative relationship. Despite the rarity of the syndrome, they recommended maintaining a high level of clinical suspicion to detect acute interstitial nephritis early in its course, especially soon after the initiation of PPI therapy.

POSSIBLE ASSOCIATIONS WITH IRON AND B₁₂ DEFICIENCIES

Long-term PPI therapy has been thought to be associated with micronutrient deficiencies, especially of iron and vitamin B₁₂. Hydrochloric acid in the stomach assists in the dissociation of iron salts from food and the reduction of ferric iron to the more soluble ferrous iron.⁵¹ Gastric acid also facilitates the release of vitamin B₁₂ bound to proteins within ingested foodstuffs to permit binding to R-proteins for eventual absorption in the terminal ileum.^51,52

Despite the biologic plausibility of these deficiencies, there is currently little evidence to support a clinically relevant association to recommend a change in current practice.

NO THERAPY IS COMPLETELY WITHOUT RISK

Although concerns have been raised about the long-term safety of PPIs, the preponderance of the evidence does not strongly support the apprehensions publicized over the last few years. When translating these studies into the routine management of patients, it is important to recall some very basic tenets of good patient care.

No therapy is completely without risk—whether pharmacologic, surgical, or psychological, and no matter how benign or straightforward. Consequently, no drug, procedure, or treatment plan should be ordered without a valid indication. Even with an indication, the risk-benefit ratio of the therapy prescribed should always be considered. If the indication for the PPI is weak or uncertain, then even a slight risk tips the balance away from the drug, and the drug should be discontinued.

When seeing patients in long-term care, the indication and necessity for all drugs, including PPIs, should be reviewed. The algorithm proposed in Figure 2 can be adapted for virtually any of the possible associations.

Consider the indication for the PPI. Was the PPI started during a hospitalization and then routinely continued after discharge? This is one situation in which the use of a PPI could potentially be discontinued.²

For complicated acid-peptic disease, dose reduction or cessation of PPI therapy may not be possible.

If the PPI was prescribed only for symptom relief, as in cases of dyspepsia or nonerosive gastroesophageal reflux disease, reduce the dose of PPI to as low as possible to maintain symptom control. Should chronic therapy still be required, no specific monitoring is recommended, apart from routine monitoring that takes place in the course of patient care.

Lastly, because of the media attention that several of these concerns have garnered, patients may still harbor significant concerns about PPIs, even their short-term use. In such cases, the prescriber should take the opportunity to communicate the reason for the decision to prescribe the PPI, as well as the best available data about the risks PPIs may pose. None of these outcomes is very common in the absence of PPIs, with the possible exception of recurrent cardiovascular events, and the risks provided in all of these studies are relative to the baseline risk. Even if the risk of a particular outcome doubles with long-term PPI use, twice a small risk remains a small risk.