Drug safety advances from ICPE 2010

The 26th International Conference on Pharmacoepidemiology and Therapeutic Risk Management (ICPE) was held recently in Brighton, UK. This report focuses on drug safety presentations made at the conference. Table 1 summarizes other relevant presentations not included in the following narrative.

Table 1.

Studies in brief. Abstracts are published in (2010) Pharmacoepidemiology and Drug Safety 19(Suppl 1): S1–S347 and give fuller details of methodology, control groups, limitations and confounders.

Drug/drug group	Study design; number of patients (study group +  control); aim	Main results	Conclusion/ comments	Author (abstract number)
Antibiotics	R, C; 10,154; risk of CDI during hospitalization	CDI-related hospitalization associated with high doses of antibiotics, long duration of therapy and greater number of antibiotics	Increasing number, dose and days of antibiotic exposure increases risk of CDI. Metronidazole had a protective effect	Stevens et al. (7)
Antibiotics	CCS; 614 + 1228; risk of drug-induced liver injury during hospitalization	Risk not related to duration but was related to concomitant use of multiple antibiotics; highest risk with sulphamethoxazole, trimethoprim, ciprofloxacin, imipenem	Antibiotic-induced DILI confirmed; risk increases if more than one antibiotic given at the same time	De Bruin et al. (9)
Antibiotics	P, C; 749,122; risk of oral facial cleft (OFC) with antibiotics	284,102 antibiotic-exposed pregnancies with 1290 OFC cases; cleft lip ± cleft palate was significantly associated with systemic doxycycline and erythromycin, and topical mupirocin	First trimester use of doxycycline, erythromycin and mupirocin is associated with an increased risk of cleft lip ± cleft palate	Moelgaard-Nielsen and Hviid (519)
Aspirin	NCC; 39,680; to evaluate the risk/benefit of aspirin in elderly patients (≥ 66y) with newly treated T2DM	1871 cases of MI: aspirin did not decrease risk of MI in current or past users and increased it in those with CVD; 614 cases of GI bleeding: aspirin increased risk of bleeding especially in current users	Risk/benefit ratio does not favour aspirin use in elderly patients with newly treated T2DM	Sirois et al. (77, poster)
Bisphosphonates	R, C; 28,977 + 2.9 million; estimate incidence rates of oesophageal cancer in osteoporosis and general population cohorts	80% increased risk of oesophageal cancer in patients with osteoporosis especially in females over 50 years old	This provides useful initial data to better evaluate any observed risk of oesophageal cancer with bisphosphonates	Thakrar et al. (61)
Calcineurin inhibitors and corticosteroids (topical)	NCC; 625,915; assess risk of lymphoma in patients (<20y) with atopic dermatitis	Adjusted ORs showed no risk of lymphoma with topical pimecrolimus, tacrolimus or corticosteroids; T-cell lymphoma OR increased with tacrolimus	No increased risk of lymphoma; increased T-cell lymphoma risk with tacrolimus	Arana et al. (28)
Contraceptives, combined oral (COC)	P, C; 59,510; to assess time patterns of VTE risk after COC use	293,227 women-years of observation to date; first-time COC use and restarting after 1 cycle break increased initial VTE risk; switching to another COC with no cycle break did not increase risk	First-time COC use and recurrent COC use after a cycle break increases VTE risk for initial 3–6 months	Dinger et al. (507)
Contraceptives, drospirenone	P, C; 30,597; comparison of SAEs with other occHRTs	Incidence rates for drospirenone/estradiol similar to other occHRTs for VTE but less for arterial thromboembolism, acute MI and stroke	Drospirenone/estradiol has a good cardiovascular safety profile compared with other occHRTs with relatively low incidence of arterial thromboembolic events	Assmann et al. (73, poster)
Contraceptives, drospirenone	P, C; 59,510; risk of long-term progestin/ethinyletradiol (EE) use	259,696 women-years of observation; adverse cardiovascular events similar for all progestin/EE combinations; significantly less drospirenone/EE users went on to use antihypertensive medications	Risk of cardiovascular events with drospirenone/EE similar to other progestin combinations; less drosperinone users start antihypertensive treatment	Dinger et al. (548)
Cotrimoxazole	R, C; 4582 + 4582; measure incidence and risk of cardiac dysrhythmia in cohorts of cotrimoxazole and other sulfonamide antibiotic users	Incidence and risk of cardiac dysrhythmia was similar in both cohorts	Cotrimoxazole was not associated with increased risk for cardiac dysrhythmia	Blumentals et al. (79, poster)
Coxibs	CCS; NR; assess effect of PPI adherence on Coxib-induced upper GI complications	Adjusted ORs showed higher risk of upper GI complications with low adherence to PPIs	A 10% decrease in PPI adherence led to a 9% increase in upper GI complications in Coxib users	Valkhoff et al. 256)
Cyclosporine (CYC)	R, C; 720 + 58,232; risk of squamous cell cancer (SCC) with CYC	Higher risk of SCC with CYC in patients with skin diseases but not with other chronic diseases	CYC does not appear to increase risk of SCC in chronic diseases expect for chronic skin conditions	Lavole et al. (29)
Diuretics	R, C; NR; to find out if diuretics increase risk of BCC	Increased risk of BCC occurred with high-ceiling diuretics over time. Higher risk in patients with sunburns	High-ceiling diuretics associated with increased risk of BCC with continuing use. No other diuretics implicated	Ruiter et al. (345)
LABAs	R; C; 1,121,218; to look at SAEs with LABAs in newly diagnosed and pre-existing patients with asthma	Low/no risk of SAEs with LABAs in newly diagnosed and pre-existing patients with asthma; less risk than SABAs; higher risk in African Americans, pregnant women and severe asthma	A substantial addition to the evidence; LABA use in general does not increase risk of SAEs; African Americans and certain comorbidities need consideration	Guo et al. (150, poster)
Lipid-lowering drugs (LLRs)	R, C; 1,997,090; to measure the risk of nontraumatic tendon rupture (NTTR) with LLRs	8296 NNTR cases: 3705 not exposed to LLRs, 2691 took statin (only or with an other LLR); LLRs increased risk of NNTR; increasing risk with increasing statin dose	LLDs increase risk of NTTR with notable statin dose–response effect	Lawler et al. (549)
NSAIDs	Syst; 1,341,673, to compare risk of stroke in current users of NSAIDs vs. nonusers	Pooled RR of incident overall stroke moderately increased with rofecoxib and diclofenac but not with naproxen, ibuprofen or celecoxib	Rofecoxib and diclofenac associated with increased risk of stroke	Varas-Lorenzo et al. (332, poster)
NSAIDs	Syst; 4,741,623, to compare risk of acute MI in current users of NSAIDs vs. nonusers	Pooled RRs ranged between 1.08 and 1.97: naproxen <celecoxib <ibuprofen <meloxicam <rofecoxib <indomethacin <diclofenac <etodolac <etoricoxib	NSAIDs associated with a mild-to-moderate increased risk of acute MI. Higher risk associated with high NSAID dose apart from naproxen and ibuprofen	Varas-Lorenzo et al. (333, poster)
NSAIDsns and coxibs	CCS; 32,602 + 325,919; risk of atrial fibrillation/flutter (AF)	Adjusted ORs showed >20% increased risk of AF with NSAIDsns or Coxibs; in new users this increased to >77%	NSAIDsns and coxibs increase risk of AF especially among new users	Schmidt et al. (259)
Opioids, benzodiazepines	CCS; 1039 + 2185; to find out if opioid or benzodiazepine use in patients (65–94y) increases pneumonia risk	Of 1039 patients with community-acquired pneumonia, 14% were current users of opioids compared with 8% of controls. Increased risk was especially high with long-acting opioids. No increased risk with benzodiazepines	Opioids but not benzodiazepines increase pneumonia risk in older patients	Dublin et al. (306)
Phenprocoumon	NCC; 246,220; to estimate risk of serious bleeding in phenprocoumon users	2553 cases hospitalized for first major bleeding; combined use with antithrombotic drugs (e.g. clopidogrel), antibiotics (e.g. quinolones, amoxicillin/clavulanate, cotrimoxazole, metronidazole) or NSAIDs increased risk	Antithrombotic drugs, some antibiotics and NSAIDs increased risk of phenprocoumon first hospitalization for serious bleeding	Jobski et al. (509)
Pioglitazone	R, C; 193,099; assess risk of bladder cancer	90 bladder cancers among 30,173 treated with drug; no risk from ‘ever use’; however, increased risk suggested with >24 months therapy	Ever use of pioglitazone not associated with increased risk of bladder cancer. Longer-term analysis is required to assess possible risk with prolonged therapy	Lewis et al. (30)
PPIs	R, C; 209175 + 233166; does PPI therapy lead to bisphosphonates prescription?	OR for anti-osteoporotic drug prescription increased with PPI use especially with regard to duration of PPI therapy and dosage	Possible association between PPI usage and subsequent need to bisphosphonate	McGowan et al. (20)
PPIs	NCC; 208,951; does PPI exposure in utero lead to increase in cardiac birth defects?	2445 cases of cardiac malformations among 208,951 pregnancies. Maternal PPI use in pregnancy increases cardiac birth defects twofold	The authors consider the implications of this result for PPI treatment of pregnancy-related gastroesophageal reflux disease	Rhim et al. (566)
Spironolactone	CCS; 4028 + 40,171; is spironolactone associated with increased risk of GI bleeding?	Of 4028 cases with a first-time clinical diagnosis of GI bleeding, the adjusted rate ratio for GI bleeding in spironolactone users was twofold compared with control group	The author cautions the use of spironolactone in patients at high risk of GI bleeding until more evidence is accumulated	Delaney et al. (510)
SSRIs	CCS; 4862 + 19,448; is there an association between SSRI use and uncomplicated peptic ulcer (UPU)?	Adjusted ORs showed increased risk with SSRIs but not TCAs	Authors suggest that SSRI use needs to be carefully considered in groups at high risk of UPU	Dall et al. (398, poster)
SSRIs	R, C; 40,999; do SSRIs increase risk of bleeding in post-MI patients taking antiplatelet medications?	Higher risk of hospitalization for major bleeding when patients taking SSRIs: e.g. 47% increased risk for SSRIs + aspirin versus aspirin alone	Adding an SSRI to aspirin therapy increases risk of major bleeding. Future study may evaluate whether risk of bleeds is balanced against benefit of reduced MI with SSRIs	Labos et al. (283)
TCAs	SGA; 3914; is there an association between baseline depression and incident hypertension?	622 of 3914 patients had baseline depression; 477 of 3914 developed hypertension; baseline depression not associated with hypertension but TCA use increased relative risk by 20%	TCAs may be associated with increased rates of hypertension. Author suggests replicating study in earlier cohort where TCAs were standard of care	Delaney et al. (299)
TCAs, SSRIs	CCS; 16,717 + 61,517; assess risk of fracture with antidepressant drugs	60% of cases had osteoporotic fracture. Both SSRIs and TCAs significantly increased risk; affinity for 5-HT transporter also shown to increase risk	Affinity for 5-HT transporter in bone may explain reason for osteoporotic fracture with TCAs and SSRIs	Verdel et al. (282)
Tiotropium	NCC; 6788; assess risk of drug-related cerebrovascular events in COPD patients	Adjusted ORs showed no increased risk of cerebrovascular events compared with long-acting beta agonists (LABAs)	Tiotropium showed no increased risk of cerebrovascular risk in COPD patients when compared with LABAs	Haag et al. (22)
TKIs	R, C; 537; are LFTs elevated in cancer patients receiving TKIs?	ALT levels tended to be elevated with TKI use	Clinically significant LFT elevation rare; any elevation usually transient	Nordstron et al. (26)
TNFIs	R,C; NR; risk of infection among RA patients	Incident rate of infection: obDMARDs>cytoDMARDs> non-cytoDMARDs =  TNFIs = MTX	Risk of infection with non-cyto DMARDs or TNFIs is comparable to MTX	Chen et al. (10)
Warfarin	R, C; 17,661; to assess the risk of bleeding in older patients (>65y) when warfarin is taken with other medications that may amplify the risk	Significant increases in bleeding-related hospitalizations when warfarin was given with aspirin ± clopidogrel, amiodarone and antibiotics	Prescribers need to take note of the excess risk of adding these medications to warfarinized patients and seek alternatives or monitor closely	Vitry et al. (307)

ALT, alanine aminotransferase; BCC, basal cell carcinoma; C, cohort; CCS, case control study; CDI, Clostridium difficile infection; COPD, chronic obstructive pulmonary disease; Coxibs, cyclooxygenase-2 specific inhibitors; DILI, drug-induced liver injury; DMARDs, disease-modifying antirheumatic drugs (cyto, cytotoxic; non-cyto, non-cytotoxic; ob, other biologic); GI, gastrointestinal; ICS, inhaled corticosteroid; LABAs, long-acting beta agonists; LFTs, liver function tests; MTX, methotrexate; NCC, nested case control; NR, not reported; NSAIDsns, nonsteroidal anti-inflammatory drugs—nonselective; occHRTs, oral continuous combined hormone replacement therapies; OR, odds ratio; P, prospective; PPI, proton pump inhibitor; R, retrospective; RA, rheumatoid arthritis; RR, relative risk; SAEs, serious adverse events; SGA, subgroup analysis; SSRIs, selective serotonin reuptake inhibitors; Syst, systematic review; TCAs, tricyclic antidepressants; TKIs, tyrosine kinase inhibitors; TNFIs, tumour necrosis factor inhibitors; VTE, venous thromboembolism; y, years.

Antidiabetic drugs in the spotlight

Cancer risk

A meta-analysis of four randomized controlled trials (RCTs) that looked at improving glycaemic control in type 2 diabetes mellitus (T2DM) was presented by Jeffrey Johnson (University of Alberta, Edmonton, AB, Canada). His group was aiming to find out whether the increased cancer risk seen in T2DM might be due to poor metabolic control. In four trials that reported cancer mortality (UKPDS 33, UKPDS 34, ACCORD and VADT) the risk estimate was not significantly different compared with controls. Three trials reporting cancer incidence (ADVANCE, PROactive and RECORD) showed similar results. While this study has limitations (i.e. it is a secondary analysis and was not a full systematic review), the author suggests that improving glycaemic control in T2DM patients does not reduce cancer risk.

In another presentation, Hui Zhang (Harvard School of Public Health, Boston, MA, USA) showed that a retrospective cohort study of electronic medical records from the UK General Practice Research Database (GPRD) did not find evidence that hypoglycaemic drugs are associated with an increased risk of cancer. Their study comprised 63,838 incident diabetic patients and 4632 cancer cases. Hazard ratios (HRs) for each additional year of use of rosiglitazone, pioglitazone, metformin, sulphonylurea and insulin were not significant.

These results are at odds with the putative protective effects of metformin demonstrated both in vitro [Memmott and Dennis, 2010] and in vivo [Jalving et al. 2010].

Cardiovascular risk

Linda Levesque (McGill University, Montreal, QC, Canada) looked specifically at cardiovascular risk associated with metformin usage. Her group performed a nested case–control (NCC) analysis on a cohort of 42,775 adults who were treated with a hypoglycaemic drug from 1978 to 1999 (i.e. specifically to exclude thiazolidinedione usage). The study endpoints were first hospitalization for acute myocardial infarction (MI), death due to ischaemic heart disease (IHD) or first hospitalization for congestive heart failure (CHF). There were 4806 hospitalizations for acute MI and 5439 cardiac deaths. There was a significantly lower risk of MI with metformin compared with sulphonylurea usage (adjusted rate ratio [RR] 0.81; 95% confidence interval [CI] 0.69–0.95), especially as duration of therapy progressed. Dr Levesque commented that this reduction is on a par with that seen with statin drugs. Metformin had no effect on cardiovascular death, however. This study adds to the data supporting metformin for first-line use in T2DM.

MI, fractures and macular oedema have been documented with thiazolidinedione use in T2DM patients. Their safety profile has been examined by Elisabetta Poluzzi and colleagues (University of Bologna, Bologna, Italy) using the FDA Adverse Event Reporting Database (AERS). A case/noncase method was used to identify adverse drug reactions with thiazolidinediones A reporting odds ratio (OR) was used to ascertain risk compared to other antidiabetic drugs. The frequency of cardiovascular reactions, macular oedema and fractures were significantly higher with thiazolidinediones. In particular, the reporting OR for congestive cardiac failure was 6.11 (CI 5.68-6.56) versus 0.66 (CI 0.60-0.73).

Acute pancreatitis

Raymond Schlienger (Novartis, Basel, Switzerland) presented results of a cohort study (NCC) which looked at the risk of acute pancreatitis (AP) in T2DM associated with antidiabetic treatment. There were 419 AP patients selected from a starting group of 85,525 T2DM patients and 200,000 from the general population taken from The Health Improvement Network (THIN) database. The results showed that T2DM causes a moderate increase in AP risk (relative AP risk OR 1.35; CI 0.98–1.86). Analysis of drug use showed that insulin treatment (adjusted OR 0.35; CI 0.20–0.61) and long-term metformin (adjusted OR 0.49; CI 0.27–0.89) use seems to reduce, but long-term sulphonylurea (adjusted OR 1.69; CI 1.03–2.78) use increased, AP risk.

There have been spontaneous case reports of AP with the new incretin mimetic, exenatide, but a retrospective cohort study found no association. Made Wenten (Amylin Pharmaceuticals, San Diego, CA, USA) gave results from a follow-up study using a large US health insurance claims database. Data on 24,000 patients taking exenatide and 450,000 taking other antidiabetic drugs were analysed using a discrete-time survival model with a study endpoint of first occurrence of AP. No excess risk of AP was found with exenatide (HR 0.95; CI 0.65–1.38).

Neuropsychiatric drugs

Cardiovascular risk

The results of a NCC study, which sought to determine the risk of heart failure (HF) in a cohort of patients with Parkinson’s disease, were presented by Dr G Trifiro (Erasmus Medical Centre, Rotterdam, The Netherlands). The cohort was taken from four different European databases and consisted of 10,790 new users of dopamine agonists (DAs) and 14,669 new users of levodopa where follow up revealed 518 incident cases of HF. There was no difference in HF risk with ergot DAs and levodopa; however, pramipexole was associated with increased risk (OR 1.49; CI 1.08–2.05). Although there may be a biological cause for a differential effect among various DAs, Dr Trifiro added that detection bias and residual confounding could not be ruled out.

In another study, health databases in Quebec, Canada, were searched to identify a population cohort of over 71,000 older adults who had initiated antidepressant therapy between 1997 and 2004. NCC analysis examined the primary endpoint of first hospitalization for acute MI or cardiac death due to IHD. Linda Levesque (Queen’s University, Kingston, Canada) stated that their analysis showed venlafaxine users were at a significantly lower risk of MI or death compared with those who had taken other antidepressants (adjusted RR 0.80; CI 0.66–0.97) and this was even more pronounced among patients with established cardiovascular disease. No benefit with selective serotonin reuptake inhibitors (SSRIs) was shown, although patients with a history of previous MI may show benefit. Dr Levesque postulated that venlafaxine’s beneficial effect may be due to prevention of platelet aggregation, and that RCTs are warranted to confirm these findings.

Annalisa Rubino (Florence University, Florence, Italy) reported results of her group’s NCC study of the THIN database, where they analysed the incident rate of cardiac valvulopathy among patients who had received serotoninergic drugs between 1990 and 2008. The calculation of ORs showed there was no evidence that his group of drugs caused cardiac valvulopathy.

Several antipsychotic drugs have been associated with sudden cardiac death (CD) and ventricular arrhythmia (VA). Medicaid data from five large US states was analysed by Charles Leonard and coworkers (University of Pennsylvania School of Medicine, Philadelphia, PA, USA) for new antipsychotic users (aged 30–75 years from 1999 to 2003) and outcome was determined by emergency department visit for CD/VA with olanzapine used as the reference group. The adjusted HRs compared with olanzapine were 2.06 for chlorpromazine, 1.72 for haloperidol, and 0.73 for quetiapine–risperidone and perphenazine were close to unity. A subanalysis of first-prescription exposures increased the HRs for chlorpromazine and haloperidol. Dr Leonard commented that these results are consistent with the Arizona Center for Education and Research on Therapeutics which lists chlorpromazine and haloperidol (as well as mesoridazine, mimozide and thioridazine) as ‘high risk’ for torsades de pointes.

Krista Huybrechts (Boston University School of Public Health, Boston, MA, USA) and colleagues conducted a cohort study of almost 11,000 patients aged 65 years or older who were given psychotropic medications shortly after being admitted to nursing homes in British Columbia (between 1996 and 2006). They looked at four measures: risk of death and hospitalization for three major medical events (femur fracture, pneumonia and HF). Proportional hazard models looking at risk within 180 days of treatment initiation showed that patients on conventional antipsychotic medicines or antidepressants had an increased risk of death (HR 1.47, CI 1.14–1.91 and HR 1.20, CI 0.96–1.50, respectively). A similar increased risk of femur fracture was also observed in both drug groups. Patients receiving benzodiazepines had an increased risk of death (HR 1.28, CI 1.04–1.58) and HF (HR 1.54, CI 0.89–2.67). Dr Huybrechts concluded that conventional antipsychotic medications, antidepressants and benzodiazepines appear to have at least the same magnitude of risk for adverse health outcomes as typical antipsychotic medications.

Other risks

Joshua Gagne (Harvard Medical School, Boston, MA, USA) presented results of a case-crossover study, which looked at the risk of seizure-related events (SREs) following refilling of anti-epileptic drugs (AEDs) and whether switching from branded to generic products carried an increased risk. They found that refilling an AED prescription with the same brand or generic produced a 2.3-fold increased risk of SREs. Switching to a different brand/generic produced only a small increase in risk (adjusted OR 1.19, CI 0.35–3.99), which Dr Gagne said was a much smaller effect than has been reported in previous studies.

The increased risk of suicide and mortality associated with anticonvulsant drugs was investigated by Elisabeth Patorno and colleagues (Harvard Medical School, Boston, MA, USA) in a cohort study that included 297,620 new episodes of treatment with an anticonvulsant. Outcomes were suicide attempts, completed suicides and violent deaths. Compared with topiramate (reference drug), attempted/completed suicides were increased with gabapentin (HR 1.42; CI 1.11–1.80), lamotrigine (HR 1.84; CI 1.43–2.37), oxcarbazepine (HR 2.07; CI 1.52–2.80), tiagabine (HR 2.41; CI 1.65–3.52) and valproate (HR 1.65; CI 1.25–2.19) and the risk was especially high within the first 30 days of treatment. In a secondary analysis, gabapentin showed increased risk compared with carbamazepine (reference drug) in both young and old patient groups, patients with mood disorders and patients with epilepsy/seizure disorders.

The risk of fractures in patients with multiple sclerosis (MS) has been assessed by Frank de Vries and coworkers (Utrecht University, Utrecht, The Netherlands). Dr de Vries presented results of a population-based cohort study from the UK GPRD where 5565 MS patients were matched with 33,864 non-MS controls, although the adjusted HR for any fracture was 1.2 (CI 1.1–1.4) there was a threefold increased risk of hip fracture in MS patients. Fracture risk was significantly higher in patients who had taken oral glucocorticosteroids or antidepressants. The glucocorticosteroid risk increased with dose, i.e. ≥7.5 mg/day prednisolone equivalents produced a 2.3-fold increased risk compared with controls. Thus, fracture risk assessment is warranted in MS patients receiving oral steroids or antidepressants.

Cardiovascular risks with aspirin and NSAIDs

Annica Bergendal (Karolinska Institutet, Stockholm, Sweden) presented results from the TEHS population-based case–control study looking at risk of venous thromboembolism (VTE) with nonsteroidal anti-inflammatory drugs (NSAIDs), aspirin, paracetamol or statins. There were 1415 women (18–64 years old) with a first-time VTE in the case group and these were matched by 1350 women from the Swedish population register. Use of NSAIDs, aspirin, paracetamol or statins during the last 3 months prior to a VTE event (or interview date in the control group) was recorded. Dr Bergendal presented preliminary results showing no VTE risk with any of the drugs. In fact, there is some suggestion of a mild protective effect with conventional dose aspirin and paracetamol, whilst frequent use low-dose aspirin shows significant protection (OR 0.50, CI 0.29–0.95). This may be confounded by the fact that frequent users of low-dose aspirin tend to be ‘healthy users’ which might bias results.

Many patients with intracerebral haemorrhage (ICH) may require antiplatelet medications, e.g. to prevent ICH ischaemic events; however, there is very little evidence or guidance about this as well as a perception that they are contraindicated. Robert Flynn and colleagues (University of Hertfordshire, Hatfield, UK) followed up a cohort of stroke patients with confirmed ICH from Tayside, Scotland. They looked at use of aspirin, dipyridamole or clopidogrel with the primary outcome being recurrent ICH. In his presentation, Dr Flynn stated that of 942 initial ICH events there were 417 patients who survived to follow up (total follow up of 1510 patient years). Of these, 120 received antiplatelet medications. Median time from discharge to first use was 14.8 months. There were 14 recurrent ICH. Interestingly, there were 29 cases of subsequent ischaemic stroke and 44 cases of subsequent ischaemic stroke or MI. HRs were not significant on any of these outcomes. Thus, antiplatelet medications had no significant impact on outcomes, which calls into question why they should be contraindicated in recurrent ICH. Dr Flynn concluded that their study rules out neither moderate harm or benefit associated with antiplatelet medicine prescribing following ICH. It should be borne in mind that these are very small numbers. Further studies are warranted to confirm these preliminary findings.