Author (year), study design |
Country (time period) |
Population; age in years; % Female |
Subtype of Proton Pump Inhibitor |
Ascertainment of Exposure |
Major adverse cardiovascular event |
Myocardial Infarction |
Follow Up Period |
Ascertainment of Outcome |
Adjustment for potential confounders |
Summary of Key Findings |
Ali36(2015)*, pharmacovigilance analysis. Abstract only. |
United States (January 2004–June 2014) |
Not published;not published; not published. |
Not specified |
Adverse event reports submitted to US Food and Drug Administration (FDA) Adverse Event Reporting System |
IHD: EBGM 0.74 (EB05-EB95 0.72–0.77). |
EBGM 0.69 (EB05-EB95 0.66–0.72) |
N/A |
Outcome defined as structured Medical Dictionary for Regulatory Activities(MedDRA) queries of Preferred Terms denoting to IHD. Safety signals defined as drug-event associations with EB05≥2.0. |
N/A |
Abstract only. PPI use is not associated with IHD. Identified that protopathic bias may contribute to association in previous studies. |
Bigardet al.36 (2005), double blind randomised controlled trial. |
France (May 2002–June 2003) |
N = 181; 53.3; 54.7. |
Lansoprazole. |
Randomised allocation |
Events/PPI group: n = 0/84 Events/control group: n = 1/97 |
Data not available |
Six months |
Not published. |
Exclusion criteria: H2RA use within one week of inclusion or requiring NSAIDs or aspirin during the study period, gastroduodenal ulcer, history of gastric or duodenal surgery. |
N/A – measure of adverse events in randomised controlled trial. |
Charlot et al. 37(2010), retrospective cohort study. |
Denmark (2000–2006) |
n = 56406; 65.0; 31.4. |
Esomeprazole, lansoprazole, omeprazole, pantoprazole. |
National pharmacy prescription registry |
Rehospitalisation for MI, stroke or cardiovascular death: HR 1.43 (95% CI 1.34–1.53), p<0.001. Events/PPI group: n = 2341/8889 Events/control group: n = 4232/22815 |
HR 1.13 (95% CI 1.01–1.25) p=0.033. Events/PPI group: n = 1029/8889 Events/control group: n = 1791/22815 |
One year |
ICD-9 codes from the validated National Patient Registry. |
Age, sex, income, PCI, concomitant treatment, comorbid medical conditions. |
Time-dependant, propensity score-matched Cox proportional hazards model. Found association of cardiovascular risk with PPI use, independent of clopidogrel use. |
Dunn et al. 38 (2013)*, post-hoc analysis of randomized controlled trial. |
United States (1999–2001) |
n = 2116; 61.6; 28.6. |
Lansoprazole, omeprazole, pantoprazole, rabeprazole. |
PPI use was determined by patient's provider and was identified at study baseline and follow-up |
Death, MI, stroke: HR 1.56 (95% CI 1.06–2.30), p=0.025. Events/PPI group: n = 48/325 Events/control group: n = 74/738 |
Data not available |
One year |
Adjudicated by blinded, independent committee. |
Race, diabetes, hyperlipidaemia, congestive heart failure, atrial fibrillation, stable angina, unstable angina, previous MI, previous stroke, PVD, PCI, coronary angiography and CABG. |
Adjusted estimated HR found association between PPI use and major adverse cardiovascular outcomes in placebo wing of The Clopidogrel for Reduction of Events During Observation (CREDO) trial, independent of clopidogrel use. |
Fasset al.39 (2009), double blind randomised controlled trial. |
United States (June 2006 – December 2006) |
N = 947; 47.5; 71.1. |
Dexlansoprazole. |
Randomised allocation |
Events/PPI group: n = 3/630 Events/control group: n = 1/317 |
Events/PPI group: n = 2/630 Events/control group: n = 0/317 |
Four weeks |
Observed by investigators, spontaneous reporting, laboratory parameters and evaluations |
Exclusion criteria: Pregnancy or lactation, Barrett's oesophagus, gastric or duodenal ulcers or surgery, oesophageal strictures, use of PPI or H2RA, use of NSAIDs or high dose aspirin, anticholinergic use, sucralfate or prokinetic agents, uncontrolled systemic disease |
N/A – measure of adverse events in randomised controlled trial. |
Flooket al.40 (2012), double blind randomised controlled trial. |
Canada, Denmark, Norway, The Netherlands (May 2004–July 2005) |
n = 599; 47.0; 47.0 |
Esomeprazole. |
Randomised allocation |
Events/PPI group: n = 0/297 Events/control group: n = 3/302 |
Events/PPI group: n = 0/297 Events/control group: n = 1/302 |
Four weeks |
Not published. |
Exclusion criteria: exercise tolerance test positive for myocardial ischemia prior to inclusion, known cardiac disease, abnormal ECG, pathological troponin, chest pain due to GORD or MSK disease. |
N/A – measure of adverse events in randomised controlled trial. |
Hall et al. 41 (2009), retrospective cohort study. Abstract only. |
United States (not specified) |
n = 10703; 60.0; 37.5. |
Lansoprazole, omeprazole, pantoprazole. |
Unclear method from healthcare database Intermountain Heart Collaborative Study Registry |
Cardiovascular death, MI: HR 1.38 (95% CI 1.25–1.52), p<0.01. |
Data not available |
One year |
Unclear method from healthcare database Intermountain Heart Collaborative Study Registry. |
Multiple potential confounders, specifics not published. |
Abstract only. Adjusted hazard ratio found association between PPI use and MI irrespective of clopidogrel use. |
Ho et al. 24 (2009), retrospective cohort study with nested case-control analysis. |
United States (October 2003–January 2006) |
n = 8205; 67.0; 1.5. |
Lansoprazole, omeprazole, pantoprazole, rabeprazole. |
Pharmacy dispensing records |
Death or rehospitalisation for ACS: OR 0.98 (95% CI 0.85–1.13), p-value not published. Events/PPI group: n not published Events/control group: n not published |
Data not available |
Between 1 October 2003–31 January 2006 (median time 521 days) |
Vital status file, chart review consistent with electrocardiographic and biochemical criteria, ICD-9 codes using Veteran Affairs (VA) database. |
Age, diabetes, previous MI, PCI, CABG, heart failure, cerebrovascular disease, PVD, cancer, COPD, renal disease, dementia, liver disease, various medications. |
After discharge, a prescription for PPI was not associated with death or rehospitalisation for ACS. |
Howden et al.42 (2009), double blind randomised controlled trial. |
United States (2009) |
N = 451; 48.9; 47.9 |
Dexlansoprazole |
Randomised allocation |
Events/PPI group: n = 1/311 Events/control group: n = 0/140 |
Events/PPI group: n = 1/311 Events/control group: n = 0/140 |
Six months |
Investigator evaluation of reported/monitored adverse events. |
Exclusion criteria: Helicobacter pylori positive, uncontrolled systemic disease, continuous anticoagulant therapy, H2RA use, misoprostol use, prokinetic use, antacid use, chronic use of NSAIDs. |
N/A – measure of adverse events in randomised controlled trial. |
Juurlinket al. 43 (2013)*, self-matched case series. |
Canada (January 1996–December 2008) |
n = 5550; 77.0; 49.0. |
Not specified |
Pharmacy prescription records from the Ontario Drug Benefit Claims Database |
Data not available |
Data not available |
12 weeks |
ICD-9 codes from the Canadian Institute for Health Information's Discharge Abstract Database. |
None, this was a self-matched case series. |
Risk of MI in first 4 weeks (risk interval) increased as compared to weeks 9 to 12 (control interval) OR 1.8 (95% CI 1.7–1.9), p-value not published. Similar results seen in benzodiazepines and H2 receptor antagonists, suggestive of confounding. |
Kiljanderet al.44 (2010), double blind randomised controlled trial. |
Argentina, Bulgaria, Canada, Czech Republic, France, Germany, Hungary, Italy, Mexico, Poland, Portugal, Slovakia, United States (2010) |
n = 960; 44.7; 77.0 |
Esomeprazole. |
Randomised allocation |
Events/PPI group: n = 2/632 Events/control group: n = 0/328 |
Data not available |
Six months |
Not published. |
Exclusion criteria: Significant disease or disorder that may have influenced results or participation, malignant disease, extensive smoking, corticosteroid use, oesophageal or gastric surgery, pregnancy or breast feeding. |
N/A – measure of adverse events in randomised controlled trial. |
Kinoshita et al.45 (2011), double blind randomised controlled trial. |
Japan (September 2004–October 2005) |
n = 285; 47.6; 55.2 |
Rabeprazole. |
Randomised allocation |
Events/PPI group: n = 1/194 Events/control group: n = 0/91 |
Data not available |
Four weeks |
Not published. |
Exclusion criteria: psychiatric/psychosomatic disease, Helicobacter pylori within 6 months, active gastric or duodenal ulcer, angina pectoris, NSAIDs, aspirin treatment, cardiovascular disease, haematological disease, renal disease, hepatic disease, malignancy, pregnancy, lactating. |
N/A – measure of adverse events in randomised controlled trial. |
Kreutz et al. 46 (2010), retrospective cohort study. |
United States (October 2005–September 2006) |
n = 16690; 66.1; 31.0. |
Esomeprazole, lansoprazole, omeprazole, pantoprazole. |
Pharmacy prescription records from medical and pharmacy claims database Medco Health Solutions Inc |
Cardiovascular death, non-fatal MI, non-fatal stroke: HR 1.19 (95% CI 0.84–1.70), p-value not published. Events/PPI group: 24.8% (n not published) Events/control group: 20.8% (n not published) |
Data not available |
One year |
ICD-9-CM codes and Current Procedural Terminology, Fourth Edition (CPT-4) codes from medical and pharmacy claims database Medco Health Solutions Inc. |
Age, sex, heart failure, diabetes, hypertension, renal disease, upper gastrointestinal disease, prior cardiovascular events, past PPI exposure. |
PPI exposure in patients not receiving clopidogrel therapy after PCI suggested that there was no association with an increased risk of cardiovascular events. |
Richter et al.47 (2000), double blind randomised controlled trial. |
United States (2000) |
n = 359; 49.7; 46.2 |
Omeprazole. |
Randomised allocation |
Events/PPI group: n = 2/236 Events/control group: n = 2/123 |
Events/PPI group: n = 1/236 Events/control group: n = 0/123 |
Four weeks |
Patient reporting |
Exclusion criteria: Evidence of oesophageal, gastric or duodenal ulcers, evidence of gastrointestinal bleeding, Zollinger-Ellison syndrome, pancreatitis, malabsorption, inflammatory bowel disease, pulmonary or liver disease, malignancy, uncontrolled diabetes mellitus, cerebral vascular disease, bleeding disorders, PPI or H2RA within 28 days, concomitant anticholinergic, promotility agents, prostaglandin analogues, sucralfate, NSAIDs, high dose aspirin |
N/A – measure of adverse events in randomised controlled trial. |
Schmidt et al. 48(2012), retrospective cohort study. |
Denmark (January 2002–June 2005) |
n = 13001; 64.0; 28.0. |
Esomeprazole, lansoprazole, omeprazole, pantoprazole, rabeprazole. |
Redeemed prescription data from the Danish Nationwide Prescription Database |
First incidence MI, ischaemic stroke, stent thrombosis, revascularization, cardiac death: HR 1.16 (95% CI 0.95–1.43), p=0.19. Events/PPI group: 267 events per 1000 person years (n not published) Events/control group: 263 events per 1000 person years (n not published) |
Data not available |
One year |
ICD codes from the Danial National Registry of Patients. Occurrence of stent thrombosis and cardiac death underwent blinded file review by a committee of cardiac specialists. |
Age, sex, hypertension, obesity, diabetes, various medications. |
In patients following PCI, PPI use and associated cardiovascular risk appeared independent of clopidogrel use. |
Shah et al. 25 (2015), data mining analysis. |
United States (Stanford Translational Research Integrated Database Environment (STRIDE): 1994 – 2011 Practice Fusion, Inc. (PF): 2007 – 2012 Genetic Determinants of Peripheral Arterial Disease (GenePAD): 2004 - 2015 |
STRIDE: n = 70477; 54.1; 55.0. PF: n = 227438; >18 (not clearly specified); not published. GenePAD: n = 1503; not published; not published. |
STRIDE, PF: dexlansoprazole, esomeprazole, lansoprazole, omeprazole, pantoprazole, rabeprazole. GenePAD:Not specified |
STRIDE, PF: prescription records and drug terms from respective databases. GenePAD: Not specified. |
GenePAD: cardiovascular mortality (MI, cardiac arrest, stroke, heart failure or aneurysm rupture): HR 2.00 (95% CI 1.07–3.78), p=0.031. |
STRIDE: OR 1.14 (95% CI 1.06–1.24), p-value not published. PF: OR 1.19 (95% CI 1.09–1.30), p-value not published. Events/PPI group: n not published Events/control group: n not published |
STRIDE: between 1994–2011. PF: 2007–2012. GenePAD: 2004–2015. |
STRIDE, PF: ICD-9 codes from the respective databases. GenePAD: medical record review with confirmation by contact of patient or next of kin. |
STRIDE, PF: age, sex, race, length of observation, number of unique drugs in the record, number of unique disease concepts, clopidogrel use. GenePAD: age, sex, race, total cholesterol, systolic blood pressure, anti-hypertensive medications, lifetime pack-years. |
Patients that take PPIs have an increased risk of MI and cardiovascular death in cohorts in which clopidogrel users were excluded. H2 receptor antagonists were not associated with an increased risk. |
Shih et al. 26 (2014), propensity score-matched cohort study and case-control cross over study. |
Taiwan (2000–2009) |
Propensity score-matched study: n = 252734; 49.3; 48.8. Case-control cross over study: n = 10860; 67.4; 31.1. |
Not specified |
Prescription records from the Longitudinal Health Insurance Database |
Data not available |
Propensity score-matched study: all patients HR 1.58 (95% CI 1.11–2.25), p=0.011. Clopidogrel non-users HR 1.53 (95% CI 1.04–2.23), p-value not published. Events/PPI group: n = 79/126367 Events/control group: n = 50/126367 Case-control cross over study: 14-day window OR 3.67 (95% CI 1.76–6.83), p<0.001. |
Propensity score-matched study: 120 days. Case-control cross over study: 14 days. |
ICD-9-CM codes from the Longitudinal Health Insurance Database, sampled from the validated National Health Insurance Research Database. |
Propensity score-matched study: age, sex, traditional risk factors for MI. Case-control cross over study: age, sex, diabetes, antiplatelet agents, NSAIDs. |
Use of PPIs shown to have independent association with increased risk of MI as determined by Cox proportional hazard models and adjustment for multiple known confounders. |
Syed et al. 49 (2016)*, retrospective cohort study. Abstract only. |
United States (January 2005–August 2015) |
Not clearly specified; not published; not published. |
Not specified |
Unclear method conducted using the Explorys Platform |
Acute coronary event, MI, acute restenosis after revascularization: OR 1.60 (95% CI 1.58–1.61), p<0.01. |
Data not available |
Between January 2005–August 2015 |
Unclear method, conducted using the Explorys Platform. |
Not specified. |
Abstract only. Association between PPI use and IHD appears due to concomitant clopidogrel use and not due to PPI independently, as decreased risk of IHD in patients treated with PPI for GORD as compared to patients with GORD not treated with PPI. Increased risk found in patients on concomitant clopidogrel therapy. |
Talley et al.50 (2001), double blind randomised controlled trial. |
Denmark, Finland, Norway, Sweden (2001) |
n = 342; 49.0; 44.0% |
Esomeprazole. |
Randomised allocation |
Events/PPI group: n = 1/170 Events/control group: n = 0/172 |
Data not available |
Six months |
Spontaneous, elicited by questioning or observed by investigators. |
Exclusion criteria: Concomitant therapy with NSAIDs, salicylates, diazepam, quinidine, warfarin, diphenylhydantoins, mephenytoin, anticholinergics or prostaglandin antagonists. |
N/A – measure of adverse events in randomised controlled trial. |
Turkiewiczet al. 51 (2015)*, case-crossover study. |
Sweden (October 2005–December 2006) |
n = 3490; 73.4; 39.0. |
Not specified |
Prescription and dispensation records from the Swedish Drug Register with use of Anatomical Therapeutical Chemical (ATC) classification to define PPIs |
Data not available |
Data not available |
93 days |
ICD-10 codes from the Skåne Healthcare Register |
Time-invariant confounding, specifics not published. |
Risk of MI 0–3 days (hazard period) after starting PPI increased as compared with risk of MI 3–93 days after starting a PPI (control period). Prescription of PPIs in the hazard period versus the control period OR 1.66 (95% CI 1.00–2.76), p-value not published. Dispensation of PPIS in the hazard period versus the control period OR 1.29 (95% CI 0.92–1.83), p-value not published. Higher odds ratio found for prescription of PPIs as compared to dispensation of PPIs, supporting a conclusion that the biological effect of PPI on AMI is unlikely, or association between PPI and AMI is due to failure to diagnose. |
Valkhoffet al. 52 (2011), case-control study. |
The Netherlands (January 1999–December 2008) |
n = 23655; 64.7; 32.8. |
Esomeprazole, lansoprazole, omeprazole, pantoprazole, rabeprazole. |
Pharmacy dispensation records from the PHARMO Record Linkage System |
Data not available |
OR 1.38 (95% CI 1.18–1.61), p-value not published. Events/PPI group: n = 237/23335 Events/control group: n = 766/110002 |
Between 1 January 1999–31 December 2008 (median time 42.6 months) |
ICD-9-CM codes from the PHARMO Record Linkage System. |
Age, sex, Concomitant medical conditions, length of stay, days of follow up, PCI, multiple cardiovascular and sugar lowering medications, total number of prescriptions prior to baseline MI. |
Increased risk for recurrent MI in PPI users in the absence of clopidogrel may illustrate increased comorbidity in PPI users in general. |
Wang et al. 53 (2014)*, cohort study. |
Sweden (January 2006–December 2008) |
n = 2285; >65 (not clearly specified); 44.0. |
Esomeprazole, lansoprazole, omeprazole, pantoprazole, rabeprazole. |
Prescription and dispensation data from the Swedish Prescribed Drug Register |
Data not available |
Data not available |
90 days |
ICD-10 codes from the Swedish Patient Register. |
Age, sex, history of cardiovascular disease, history of bleeding, co-morbidity score. |
Use of PPIs increases the risk of recurrent cardiovascular disease (HR 1.11 [95% CI 0.75–1.65]) when compared with concomitant use of PPIs and clopidogrel (HR 1.00 [reference]). However, as compared to no PPI and no clopidogrel use (HR 1.54 [95% CI 1.05–2.24]), PPI use is associated with a lower risk of cardiovascular disease. |
Comments