Latest drug safety alerts from the 27th International Conference on Pharmacoepidemiology and Therapeutic Risk Management

The 27th International Conference on Pharmacoepidemiology and Therapeutic Risk Management was held in Chicago, IL, USA on 14–17 August 2011. This report focuses on presentations where clinically relevant drug safety alerts were announced or confirmed. The abstract number in each summary refers to the full abstract, which is published in Pharmacoepidemiology and Drug Safety (2011) 20(Suppl 1): S1–S364.

Analgesic drugs

Danish databases were analysed for prescription of chronic paracetamol use during pregnancy (1996–2008) to see if it had led to an increased risk of asthma in offspring (Ane Brigitte Andersen et al. Aarhus University, Denmark; Abstract 166). A total of 976 out of 197,071 (0.5%) children born during this time period had been prenatally exposed to maternal paracetamol use. Increased absolute risk of asthma in exposed versus nonexposed children was most evident in the first trimester: 13% exposed versus 7.5% nonexposed after 2 years, and 22% exposed versus 14% nonexposed after 10 years. The adjusted incidence rate ratio (aIRR) was1.42 (confidence interval [CI] 1.17–1.73).

Whilst opioid addiction in pregnant women is associated with a high risk of preterm birth, the risk with legally prescribed opioids in pregnancy is unknown. A population-based cohort study using birth registry and prescription data in Norway (2004–2009) investigated this risk among pregnant women prescribed opioids for noncancer pain (Marte Handal et al. Norwegian Institute of Public Health, Oslo, Norway; Abstract 173). There were 3710 out of 156,000 pregnancies where opioids were prescribed with an overall prevalence of preterm birth of 4.2%. The odds ratio (OR) for preterm birth increased in a dose-dependent fashion to 3.1 (CI 1.9–5.1) in women who used opioids during all trimesters.

Patricia McGettigan and David Henry (Hull and York Medical School, UK and Institute for Clinical Evaluative Sciences, Toronto, Canada) conducted an updated systematic review and meta-analysis of community-based observational studies (published during the period 2000–2010) to provide estimates of the risk of cardiovascular (CV) events with nonsteroidal anti-inflammatory drugs (NSAIDs) (Abstract 748). Their analysis included 28 reports of case-control studies (184,946 CV events) and 23 cohort studies (outcomes for over 2.3 million individuals exposed to NSAIDs). Using pooled adjusted relative risk (aRR) values, they found the highest RR was seen with rofecoxib (1.5, CI 1.3–1.6) then diclofenac (1.4, 1.3–1.5). Etoricoxib and etodolac also showed increased CV risk though data were sparse. Lowest RR was observed with naproxen (1.1, 1.0–1.2) and celecoxib (1.2, 1.1–1.3). Pair-wise analyses showed etoricoxib had a significantly higher RR than ibuprofen (1.68, 1.1–2.5) and naproxen (1.75, 1.2–2.6), but not significantly higher than rofecoxib and diclofenac. Indomethacin had a higher RR than naproxen (1.2, 1.1–1.4). The RR for naproxen was significantly lower than ibuprofen (0.92, 0.87–0.99), but not celecoxib. Higher doses of ibuprofen and celecoxib were associated with higher RR but this was not observed with naproxen. The authors discouraged the use of indomethacin and stated that the CV risk with diclofenac is a concern especially because it is available without prescription in many countries. Naproxen may have an advantage over ibuprofen and celecoxib because CV risk does not appear to be elevated at higher doses.

In another study of NSAID CV safety, this time focused on lumaricoxib (LUX), Isla Mackenzie and colleagues (University of Dundee, UK) performed meta-analyses of six LUX versus placebo trials and 12 LUX versus other NSAIDs trials (Abstract 821). Pooled ORs for myocardial infarction (MI), stroke CV death and chest pain revealed no detectable difference in CV events between LUX and placebo or other NSAIDs.

Antibiotic drugs

Jon Anderson and colleagues (University Hospital Copenhagen, Denmark) used the Danish Fertility Database (1997–2005) for a nationwide cohort study to test the hypothesis that use of trimethoprim in the first trimester causes a higher frequency of miscarriage (Abstract 19). They found a hazard ratio (HR) of 2.0 (CI 1.4–3.0) with trimethoprim compared with nondrug exposure and suggest further studies are required to confirm these findings.

Case reports suggest that fluoroquinolones (FQs), as a class, are associated with tendon rupture. David Money and colleagues (Food and Drug Administration, Silver Spring, MD, USA) conducted a study using the Military Health System to see whether there are differences among FQs (Abstract 434). Nontraumatic tendon rupture rate was measured in four new-user groups: FQs (all cases), levofloxacin (LEV), ciprofloxacin (CIP), and other antibiotics (ABS). FQs users had an increased rupture risk compared with ABS users (aRR 1.48, CI 1.16–1.86). Rupture rate was higher in LEV users (aRR 2.13, CI 1.60–2.77) than for CIP users (aRR 0.98, CI 0.65–1.41) when compared with ABS users. There was also a significant difference between LEV and CIP users (aRR 2.03, CI 1.64–2.49). Prior steroid use among LEV and CIP users increased rupture rate four fold in LEV users compared with CIP users.

The risk of severe dysglycaemia in outpatients receiving oral antibiotics (measured as hospitalization for dysglycaemia within 30 days of antibiotic therapy) was studied in a population-based cohort study using the Taiwan National Health Insurance Research Database (NHIRD) (Hsu-Wen Chou et al. National Taiwan University, Taiwan; Abstract 653). A total of 1,662,499 patients received antibiotics during the study period (2006–2007). The adjusted ORs (aORs) for the FQs (LEV, CIP, moxifloxacin) compared with macrolides ranged from 1.95 to 2.92 for hyperglycaemia and 3.03 to 3.39 for hypoglycaemia. Cephalosporins showed no detectable dsyglycaemic risk. Thus, oral FQs appear to be associated with an increased risk of severe dysglycaemia.

A retrospective claims-based cohort study investigated the risk of acute myocardial infarction (AMI) with protease inhibitors (PIs) and specific nucleos(t)ide reverse transcriptase inhibitors (NRTIs) in patients with HIV infection (Carman et al. Innovus, Ann Arbor, MI, USA; Abstract 323). Of the 66,286 patients, 474 had AMI. The aIRRs for abacavir and PI use were 1.30 (CI 1.03–1.64) and 1.44 (CI 1.13–1.83), respectively. Further analyses showed increased risks with non-NRTIs, and with recent exposure. The authors suggest the risk needs to be balanced against the effectiveness of the therapies and at managing CV risk factors.

Antidiabetic drugs

Thiazolidinediones (TZDs) are associated with increased fracture risk; however, type 2 diabetes mellitus (T2DM) is a risk factor for osteoporotic fracture. Marloes Bazelier and colleagues (Utrecht University, The Netherlands) conducted a population-based cohort study (Danish National Databases 1996–2007) looking at use of antidiabetic drugs according to severity of disease (Abstract 276). They found that use of TZDs (third stage of disease) and insulin (fourth stage) both increased the fracture risk 1.3 fold, whereas drugs used in the earlier stages of the disease did not. The authors concluded that whilst TZDs appear to increase the risk of osteoporotic fracture, this is partially confounded by T2DM severity.

Anti-epileptic drugs

Older anti-epileptic drugs (AEDs) are considered to be teratogenic; however, there is scant information about newer AEDs. The North American AED Pregnancy Register (1997–2010) was analysed by Sonia Hernandez-Diaz and colleagues (Harvard School of Public Health, Boston, MA, USA) to estimate risk of malformations with AEDs during the first trimester (Abstract 24). Major malformations were higher for older AEDs (e.g. valproate, phenobarbital) than for newer AEDs such as lamotrigine. For example, RR was 5.3 (CI 3.1–9.0) for valproate when compared with lamotrigine. However, the authors noted that seizure control was better with the older AEDs.

Older versus newer AEDs were studied in a cohort study of 40–64-year-old patients (HealthCore Integrated Research Database) and Medicare beneficiaries over 65 years of age who initiated AED therapy between 2001 and 2006 (Patorno et al. Brigham and Women’s Hospital, Boston, MA, USA; Abstract 558). The aim was to evaluate coronary and stroke risk with these drugs. In the younger aged group there were 460 coronary events and 263 stroke events among 159,825 new treatment episodes. In the older group, the corresponding figures were 7043 coronary and 281 stroke events in 43,016 new treatment episodes. Stroke risk was increased in the younger group with older versus newer AEDs (RR 1.9, CI 1.2–3.2), but there was no difference in coronary events. In the older group, coronary risk was increased with older versus newer AEDs (RR 1.2, CI 1.1–1.3) with no meaningful difference for stroke events. Thus, older AEDs increase risk of coronary events in older patients and increase stroke risk in a younger population.

A retrospective cohort study of 1 million subjects from the Taiwan NHIRD database looked at risk of stroke with older AEDs (Lai et al. National Cheng King University, Tainan, Taiwan; Abstract 563). Adjustment for propensity score matching revealed the risk of stroke was higher for phenytoin (HR 1.78, CI 1.01–3.15) and valproate (HR 1.98, CI 1.08–3.60) compared with phenobarbital. Stroke risk was increased with increasing AED dosage.

Bisphosphonates

Bisphosphonates (BPs) have been linked to osteonecrosis (ON) of the jaw; however, it is not known if they cause ON elsewhere. Steven Vlad and colleagues (Boston University School of Medicine, MD, USA) performed a self-controlled case series using The Health Improvement Network (THIN) database (1989–2009) to investigate this (Abstract 274). They found that BPs increased the risk of ON at any site. RR ranged from 2.4 in the first 3 months of therapy up to 3.3 during chronic use.

The THIN database has also been used to find out if BPs are associated with carpal tunnel syndrome (Carvajal et al. Valladolid University, Spain; Abstract 73). The cohort consisted of women over 51 years of age who had received at least one oral BP prescription who were compared against a control group. There were 983 of 30,727 women in the BP cohort who were diagnosed with carpal tunnel syndrome (1.36%), this compared with 570 out of 62,091 women in the control group (0.93%). The aOR was 1.13 (CI 0.99–1.29; p = 0.071), and this was related to drug exposure (aOR among the cohort receiving 60 or more prescriptions was 2.02, CI 1.27–3.23).

Rune Erichsen and colleagues (Aarhus University Hospital, Denmark) investigated the risk of atrial fibrillation with intravenous BPs in cancer patients (Abstract 280). They conducted a population-based cohort study using Danish nationwide databases (2000–2008). Among 3891 BP-exposed patients and 7906 nonexposed patients there were 3.2% and 2.4% who developed atrial fibrillation (adjusted HR [aHR] 2.0, CI 1.5–2.6). This extends the findings of atrial fibrillation that have been reported with relatively lower oral BP doses in osteoporotic patients.

Cancer drugs

Whilst congestive heart failure (CHF) has been documented in randomized controlled trials (RCTs) of bevacizumab in colorectal cancer (CRC), the extent of this risk in general practice, where patients are generally older and with more comorbidities, is unknown. Huei-Ting Tsai and colleagues (Georgetown University, Washington, DC, USA) used the US SEER-Medicare database (2005–2007) to look at CHF or cardiomyopathy (CM) among CRC patients who did and did not receive bevacizumab (Abstract 281). They found 188 CHF or CM events in 3832 CRC patients (median age 75 years): bevacizumab increased the risk of CHF or CM (HR 2.32, CI 1.70–3.16) and this was even more pronounced in patients aged 75 years or older (HR 3.65, CI 2.55–5.23) and in those with prior cardiac disease (HR 3.00, CI 2.07–4.35). Thus, bevacizumab-induced risk of CHF or CM is higher in the general oncology population than reported in RCTs.

A population-based cohort study (with nested case-control analysis) of prostate cancer patients using the UK General Practice Research Database has been conducted by Laurent Azoulay and colleagues (Jewish General Hospital, Montreal, QC, Canada) (Abstract 301). Their objective was to determine if androgen deprivation therapy (ADT) increases the risk of stroke and transient ischemic attacks (TIAs). Of 22,310 newly diagnosed patients with prostate cancer, 938 experienced stroke/TIA. Risk of stroke was greater with ADT therapies compared with patients who had never received ADT: adjusted rate ratio (CI) for gonadotropin-releasing hormone agonists, oral anti-androgens and bilateral orchiectomy were 1.18 (1.00–1.39), 1.47 (1.08–2.01) and 1.62 (1.06–2.48), respectively.

Robert Hawkins and colleagues (School of Cancer and Imaging Studies, Manchester, UK) evaluated the rates of, and reasons for, treatment modifications and adverse events (AEs) associated with use of sunitinib in patients with advanced renal cell carcinoma (aRCC) (Abstract 162). Data from 132 patients who received sunitinib were collected from three UK oncology centres (2005–2010).Of these patients 85.6% experienced AEs (24.2% were grade 3/4): mucositis/stomatitis (42.4%), diarrhoea (38.6%) and fatigue (28.0%). Treatment modifications occurred in 87.1% of patients, and 62.1% were due to AEs – mainly dose reduction followed by interruption then discontinuation. Validation in a larger sample is sought; however, the current results suggest sunitinib tolerability is a considerable clinical challenge in aRCC.

The Thomson Reuters MarketScan database (2002–2009) was used to conduct a retrospective cohort study of the comparative risk of MI, stroke and revascularization among cancer patients who had, or had not, received any of the vascular endothelial growth factor (VEGF) inhibitor drugs, bevacizumab, sunitinib or sorafenib (David Nimke et al. Abbott Laboratories, Abbott Park, IL, USA; Abstract 618). Exposure to VEGF inhibitors did not increase the risk of MI or revascularization compared with VEGF-naïve patients; however, it did significantly increase stroke risk (aHR 2.27, CI 1.71–2.98). Thus, VEGF-inhibitor therapy is an independent risk factor for stroke.

CV drugs

Statins are purported to have anti-inflammatory and cardioprotective benefits. Mary De Vera and colleagues (Arthritis Research Center of Canada, Vancouver, BC, Canada) conducted a longitudinal study to find out if starting statins in patients with rheumatoid arthritis (RA) lowers the risk of AMI (Abstract 27). Analysis of a population-based cohort (1996–2006) showed that the HR for AMI in patients receiving statins versus those who did not was 0.69 (CI 0.54–0.90). Thus, statins given to RA patients did appear to lower the risk of AMI.

Hilda de Jong and colleagues (Maastricht University Medical Center, The Netherlands) assessed the association between statin use and occurrence of polymyalgia rheumatica and temporal arteritis (PR/TA), both of which have been case reported with statin use in the literature. Individual case safety reports (ICSR) in the World Health Organization global ICSR database (Vigibase) led to the identification of 327 reports of PR/TA, which were compared with 1635 other adverse drug reactions (ADRs) as non-cases (Abstract 155). Statins were reported as the suspected agent in 29.4% of PR/TA cases versus 2.9% in non-cases: adjusting for covariates showed a significant association between statin use and occurrence of PR/TA (reporting OR 14.21, CI 9.89–20.85).

Drugs for asthma and chronic obstructive pulmonary disease

Sonal Singh and colleagues (Johns Hopkins University, Baltimore, MD, USA) conducted a systematic review of all observational studies and RCTs in order to evaluate the association between chronic use of inhaled corticosteroids (ICS) and fractures among patients with chronic obstructive pulmonary disease (COPD) (Abstract 278). An increased odds of fracture was seen in both meta-analyses of RCTs (Peto OR 1.27, CI 1.01–1.58) and observational studies (OR 1.21, CI 1.12–1.32).

Corticosteroids may affect blood glucose control; thus, researchers in Adelaide, Australia have conducted a retrospective study to find out if corticosteroid use (inhaled and systemic) for COPD treatment in patients who also have T2DM increased the risk of diabetes-related hospitalization (Gillian Caughey et al. University of South Australia; Abstract 650). Patients who were new users of metformin or sulphonylurea drugs were identified using claims data from the Department of Veterans’ Affairs (2000–2008). There were 24,330 new users during this time period of which 1504 had COPD. A total of 954 COPD/T2DM patients received corticosteroids and 20% of them had a diabetes-related hospitalization during 5 years of follow up. Dose stratification revealed a 50% increased risk of diabetes-related hospitalization in those receiving a defined daily dose of corticosteroid ≥0.3 (units not given; subhazard ratio 1.50, CI 1.01–2.22, p < 0.05, compared with no corticosteroid use). The authors suggest close monitoring of blood glucose levels in such patients.

The US Adverse Events Reporting System (2004–2010) was used in a retrospective analysis of adverse reports for behavioural or psychiatric problems (Medical Dictionary for Regulatory Activities preferred terms) with use of ICS, intranasal corticosteroids (INCS) or systemic corticosteroids (SCS) (Ayad Ali and Abraham Hartzema, University of Florida, Gainesville, FL, USA; Abstract 142). There were 68,101 reports related to corticosteroid use over the study period (78% for SCS; 11.3% ICS; 10.7% INCS). Proportional reporting ratios (PPRs) for behavioural or psychiatric problems were higher for ICS (1.40, CI 1.20–1.63) and INCS (1.71, CI 1.48–1.97) than for SCS (0.48, CI 0.43–0.54). The PPR safety signal was highest for inhaled mometasone (2.50, CI 0.38–16.8).

A retrospective cohort study using the Korean Health Insurance Review and Assessment Service claims database was used by Chul Woo Ree and colleagues (Seoul National University College of Medicine, Republic of Korea) to determine if leucotriene receptor antagonists (LRAs) increase the risk of depression in older patients with asthma compared with ICS (Abstract 576). There were 8545 patients eligible for analysis (no psychiatric disorder or treatment, no previous asthma controller medications). Of these, 119 had new onset depression: the risk for depression with LRAs compared with ICS was increased (HR 1.50, CI 1.02–2.21).

Neuropsychiatric drugs

Helle Keiler and colleagues (Karolinska Institutet, Stockholm, Sweden) assessed whether or not selective serotonin reuptake inhibitors (SSRIs) during late pregnancy are associated with persistent pulmonary hypertension in the newborn (PPHN) (Abstract 22).They conducted a cohort study of over 1.6 million Nordic infants where they obtained information on SSRI use during pregnancy and found that late use of SSRIs in pregnancy increased the risk of PPHN (aOR 2.3, CI 1.3–4.2). PPHN risk was of similar magnitude across all SSRIs, thus suggesting a class effect.

A postmarketing study was conducted by M. Davies and colleagues from the Drug Safety Research Unit (Southampton, UK) to examine the risk of cardiac events during atomoxetine therapy (Abstract 445). They employed an observational cohort technique (modified-prescription event monitoring), where cardiac event risk was monitored among patients before and after starting atomoxetine during the period 2004–2009. The cohort comprised 4711 patients (median age 12 years). Of the five cardiac events that were monitored (cardiac arrest, chest pain, arrhythmia, tachycardia, prolonged QT interval), a significant increased risk was found for chest pain (RR 3.29, CI 1.41–7.66, p < 0.05) and tachycardia (RR 2.36, CI 1.17–4.78, p < 0.05). The authors suggest that atomoxetine should be used with caution in patients with a history of cardiac disease; however, residual confounding of the results is possible and other pharmacoepidemiological data should also be considered.

A retrospective cohort study of elderly noninstitutionalized Medicaid/Medicare patients was conducted to examine all-cause mortality resulting from new prescriptions of antipsychotic medications (APMs) (Tobias Gerhard et al. Rutgers University, New Brunswick, NJ, USA; Abstract 603). There were 151,497 patients (mean age 79 years, 76% female). Comparison with risperidone produced an increased mortality risk with haloperidol (HR 1.57, CI 1.44–1.70) and similar or lower risks with aripiprazole (0.73, CI 0.54–0.99), quetiapine (0.75, CI 0.69–0.82), olanzapine (0.91, CI 0.85–0.98) and ziprasidone (0.81, CI 0.59–1.11). Thus, all-cause mortality risk differs between individual APMs in older patients.

It is well known that all-cause mortality is higher with conventional versus atypical APMs in older patients living in nursing homes. Krista Huybrechts and colleagues (Brigham and Women’s Hospital, Boston, MA, USA; Abstract 604) conducted a cohort study of Medicaid-eligible patients who had started APMs after entering nursing homes between 2001 and 2005 in order to establish the clinical mechanisms behind this increased mortality. There were 83,959 patients in the cohort. Conventional APMs increased the risk of all-cause mortality compared with atypical ones (RR 1.55, CI 1.31–1.82), and death due to stroke (1.90, CI 1.19–3.03), other circulatory system disorders (1.42, CI 1.09–1.84) and respiratory disorders (1.77, CI 1.08–2.89). Thus, the increased risk of death in nursing homes with conventional versus atypical APMs appears to be due to circulatory and respiratory disorders.

The French pharmacovigilance database (containing spontaneous ADR reports 1985–2010) was analysed to find cases of hepatotoxicity related to Alzheimer drug treatment (Geric Maura et al. University Hospital of Reims, France; Abstract 706). There were 45 hepatotoxicity ADRs reported where an Alzheimer drug treatment was reported as the suspected drug – 30 serious adverse events (SAEs) and six fatal. Donepezil was associated with 30 cases of hepatobiliary ADRs (only six, five and four cases with memantine, galantamine and rivastigmine, respectively). Serum alanine aminotransferase levels were elevated in 34 cases, cholestasis occurred in 20 cases, and three cases of fulminant hepatitis were recorded (two with donepezil, one with rivastigmine). The authors conclude that physicians should take special care to look out for hepatotoxicity with Alzheimer drugs, especially donepezil.

Oral contraceptives and hormonal therapies

What is the impact of obesity on safety and efficacy of combined oral contraceptives (COCs)? Juergen Dinger and colleagues (Berlin Center for Epidemiology and Health Research, Germany) addressed this issue by analysing 185,000 patients from four international cohort studies (Abstract 34). Their results showed that obese users of COCs have much higher health risks than non-obese users (e.g. 252 versus 186 SAEs/10,000 women years in obese versus non-obese women.

What is the impact of obesity on safety and efficacy of combined oral contraceptives (COCs)? Juergen Dinger and colleagues (Berlin Center for Epidemiology and Health Research, Germany) addressed this issue by analysing 185,000 patients from four international cohort studies (Abstract 34). Their results showed that obese users of COCs have much higher health risks than non-obese users (e.g. 252 versus 186 SAEs/10,000 women years in obese versus non-obese women.

The risk of venous thromboembolism (VTE) was compared in a transdermal estradiol system (ETS) and an oral oestrogen-only hormonal therapy (OHT) in a claims analysis from the Thomson Reuters MarketScan database (2002–2009) (Francois Laliberte et al. Analysis Group Inc., Boston, MA, USA; Abstract 417). There were 27,018 matched women in the ETS and OHT groups. There were 115 ETS users and 164 OHT women who developed VTE (unadjusted IRR 0.72, CI 0.57–0.91, p = 0.006). After adjustment, the lower VTE incidence with ETS remained significant (33% reduction, p = 0.013).

Proton pump inhibitors

Sabine Meijvis and colleagues (St Antonius Hospital, Nieuwegein, The Netherlands) conducted a population-based case-control study to assess the possible cause of community-acquired pneumonia (CAP) following initiation of proton pump inhibitor (PPI) therapy (Abstract 415). PPIs within the first 30 days of treatment were associated with an increased risk of CAP (aOR 3.2, CI 1.4–7.2); however, the authors did not find that gastrointestinal bacteria were contributing to oropharyngeal overgrowth (and causing CAP), nor were PPIs being prescribed wrongly for early symptoms of CAP. The authors theorize that other, possibly immunomodulatory, effects of PPIs increase the risk of CAP.

The risk of asthma in offspring resulting from the use of PPIs in pregnancy was investigated by Ane Birgitte Andersen and colleagues (Aarhus University, Aarhus, Denmark) (Abstract 20). Examination of Danish databases over 1996–2008 found an aIRR for asthma of 1.52 (CI 1.32–1.76) for children exposed to PPIs in the first trimester compared with nonexposure. Slightly lower aIRRs were observed for children exposed in the second and third trimesters. Whilst this may suggest that prenatal exposure to PPIs increases asthma risk in offspring, the results are confounded by findings of another analysis that suggested similar asthma propensity with H₂-receptor antagonists (H2RAs).

Dispensing patterns and pregnancy outcomes in women prescribed PPIs during pregnancy were analysed in a cohort of births in Western Australia from 2002 to 2205 (Lyn Colvin et al. University of Western Australia, Perth, WA, Australia; Abstract 104). There were 732 women who received a PPI during pregnancy. They were more likely than the control group to be aged between 35 and 45 years, be smokers, have multiple births, live in a lower socio-economic area, and have essential hypertension or asthma. They were found to be at higher risk than the control group for postpartum haemorrhage, and neonates were more likely to be born preterm and at a lower birth weight. A registered major birth defect was found in 18 (5%) case group and 3953 (4.0%) control group (OR 1.2, CI 0.8–2.0) neonates.

A prospective study using claims data from the Military Health System (2005–2009) evaluated the risk of hip/femur fractures with PPIs versus H2RAs (Rosenie Thelus Jean and colleagues, US Army Pharmacovigilance Department, Silver Spring, MD, USA; Abstract 61). The group consisted of patients who had taken no PPIs or H2RAs in the prior 180 days who were then followed for 365 days for fracture risk. Their preliminary findings showed that PPIs increased hip/femur fracture risk (RR 1.16, CI 1.13–1.19) compared with H2RAs. Pantoprazole showed the highest RR (1.91, CI 1.83–1.99). The authors are now looking at dose, duration, concomitant drugs and comorbidities to validate these results.

In contrast, Jeffrey Munson and colleagues (Dartmouth Medical School, Hanover, NH, USA) conducted a retrospective cohort study to look at PPI use during corticosteroid therapy (HealthCore Integrated Research Database) (Abstract 277). From a total of 13,749 patients followed for an average of 15 months, and controlled for baseline confounders, they found that PPI use in corticosteroid users was associated with hospitalization for hip fracture; however, when controlled for time-varying confounders, this association was not found. The authors suggest there is no clear increased risk of fracture risk by adding PPIs to corticosteroid therapy.