ISPE August 2016, Dublin

32nd International Conference on Pharmacoepidemiology & Therapeutic Risk Management (ICPE), Dublin, August 2016


Abstract 263: Classification of Drugs Implicated in Cases of Proarrhythmia: Results from the Drug-Induced Arrhythmia Risk Evaluation (DARE) Study in England

Abigail Coughtrie, Elijah Behr, Deborah Layton and Saad Shakir

Background: Several drugs have been withdrawn due to QT prolongation (QTp) or Torsade de Pointes (TdP). The DARE study aimed to improve knowledge of the epidemiology of proarrhythmia [documented TdP, ventricular fibrillation (VF) or polymorphic/non-polymorphic VT (non-QTp); exacerbation of preexisting proarrhythmia, newly documented proarrhythmia or conversion of unsustained proarrhythmia to sustained; severe (>500ms) or moderate (450ms male/470ms female) QTp with past medical history (PMH) of presyncope or syncope] by identifying a cohort of cases.

Objectives: Characterise cases and estimate contribution of individual drugs to risk of proarrhythmia.

Methods: Self-reported data on demography, symptoms, PMH and drug history (DH) were gained for proarrythmia cases referred by cardiologists in England 2003-2011. PMH/DH were verified from hospital notes. Implicated drugs were adjudicated by 2 cardiologists and classified (ATC system). Proarrythmia risk and cytochrome (CYP) P450 activity were categorised using CredibleMeds; potential drug-drug interactions were identified using Medscape Drug Interaction Checker. Analysis comprised descriptive statistics and radar plots.

Results: Of 124 cases, 95 (77%) were QTp-related and 29 (23%) non QTp-related; mean age 62 yrs [SD 15]; 63% female; PMH: 66 (53%) high blood pressure, 90 (73%) heart rhythm problems, 34 (27%) heart valve problems; 33 (27%) hypokalaemia at presentation. Of 166 implicated drugs, 70 (42%) were antiarrhythmics (C01B) [40 amiodarone; 23 flecainide]. A single drug was implicated in 90 (73%) pts and multiple drug combinations in 32 (26%) pts [27 >1 CYP inhibitor]. Potential drug-drug interactions were QTp (19), cardiotoxic (5), conditional (6), other (1). Drugs with known/possible/conditional QTp risk were implicated individually in 88 (71%) pts or in combination in 24 (19%) pts.

Conclusions: Antiarrhythmics, non-cardiac drugs and drug combinations were implicated in the 124 clinically-validated proarrhythmia cases. Underlying cardiovascular disease was present in most cases. Selection bias is possible due to referral and adjudication by cardiologists only.


Abstract 290: Utilisation of Once Weekly Exenatide (Bydureon[reg]) for Type 2 Diabetes Mellitus (T2DM): Interim Results from an Observational Cohort Study in England 

Abigail Coughtrie, Deborah Layton, Qing Qiao and Saad Shakir

Background: Bydureon®, licensed in 2011, is indicated for treatment of T2DM in combination with metformin (MET), sulphonylurea (SU), thiazolidinedione (TZD), MET+SU or MET+TZD for patients (pts) with inadequate glycaemic control on maximally tolerated doses of these therapies alone. A postmarketing safety study is underway as part of the EU Risk Management Plan.

Objectives: To describe pts characteristics and utilisation of Bydureon® at interim (once 2500 pts accrued).

Methods: An observational, population-based cohort study in primary care. Pts were identified from dispensed prescriptions (Rx) issued by GPs Sep 2011-Sep 2015 (interim datalock). Questionnaires were sent to GPs 12 months after each pts’ 1st Rx for additional exposure, outcome and risk factor information for selected outcomes of interest recorded in medical charts. Summary descriptive statistics were calculated. A final cohort of 5000 pts is desired.

Results: Interim evaluable cohort = 2538 pts; 1410 (55.6%) males; median age 57 years (IQR: 50, 65). At baseline, 91.5% (n=1984) of pts were obese (BMI>30.0kg/m2) and 50.0% (n=810) had HbA1c>9%, representing very poor diabetes control. A number of pts were diagnosed with T2DM >10 years (42.6%, n=1054) prior to treatment. Prior exposure to exenatide (Byetta®) was reported in 33.2% (n=826) pts. Bydureon® was prescribed in primary (50.2%, n=1245), secondary (46.3%, n=1146) and intermediate (2.9%, n=73) care settings. Bydureon® was used predominantly as 2nd (30.9%, n=744) or 3rd line cotherapy (66.6%, n=1601). MET (85.5%, n=2170) and SU (49.7%, n=1260) were frequently coprescribed antidiabetics at index. A total of 1752 pts (69.0%) continued treatment to the end of the 12-month observation period.

Conclusions: These interim results suggest Bydureon® is largely used in obese pts, suggesting possible channelling by prescribers to pts who are obese and/or have poor control of their diabetes. Bydureon® is
largely prescribed as 2nd line co-therapy with MET or SU or 3rd line co-therapy with MET+SU, as per NICE guidelines. This interim analysis will be superseded once cohort accrual and final analysis are complete.


Abstract 335: The Application of the HAS-BLED Criteria within Post Authorisation Safety Studies to Characterise Anticoagulant New User Patients with Non-Valvular Atrial Fibrillation (AF): Interim Results from a Specialist Cohort Event Monitoring (SCEM) Study 

Deborah Layton, Alison Evans, Miranda Davies, Vicki Osborne and Saad AW Shakir

Background: The Rivaroxaban Observational Safety Evaluation (ROSE) SCEM study is being conducted as part of a risk management plan to monitor short-term (first 3 months) safety and utilisation of rivaroxaban (Riv). A contextual comparator of new user patients (pts) prescribed best practice standard care (warfarin-War) is also being identified. Study objectives include advancing the understanding of the pt population prescribed Riv in the secondary care setting.

Objectives: An interim analysis to evaluate use of HAS-BLED to characterise baseline bleeding risk of pts with AF treated for prevention of stroke/systemic embolism with Riv or War.

Methods: An observational, population-based cohort study. The interim cohort was identified through a specialist network from Sep13 to Mar15 (datalock), supported by UK Clinical Research Networks. Data collected via a questionnaire from consenting pt medical charts by specialists incl. baseline pt characteristics within HAS-BLED. Descriptive statistics & univariate analyses [OR(95%CI)] were calculated (% denominator assumes no missing data).

Results: Interim AF cohort; Riv=641, 53% male; War=477, 56% male. Riv pts were more likely than War pts to have a stroke history [39% vs 26%;OR 1.8(95%1.4,2.3)], but less likely to have renal disease [1% vs 4%;OR 0.2 (95%CI0.1,0.6)] or use drugs predisposing to bleeds [2% vs 5%;OR 0.3 (95% CI0.2,0.7)]. Non-significance differences in baseline prevalence were observed of uncontrolled hypertension (3% vs 4%), abnormal liver function (0.5% vs 1%), predisposition to bleeds (4% vs 4%), excess alcohol use (2% vs 2%) and age 65+ yrs (84% vs 83%), respectively.

Conclusions: This SCEM study shows that HAS-BLED criteria can provide a framework for systematic collection of baseline bleeding risk data. In this interim analysis, some differences were observed between the Riv and War AF cohorts in the reported prevalence of baseline bleeding risk factors. Findings from this interim analysis will become obsolete when all available data are analysed for the final report.


Abstract 337: Factors Associated with Rivaroxaban Prescribing in Specialist Care Setting: Interim Results from the Rivaroxaban Observational Safety Evaluation (ROSE) Study

Deborah Layton1,2, Alison Evans1,2, Miranda Davies1,2, Vicki Osborne1,2 and Saad AW Shakir

Background: Increasingly choice of medicines is guided by published guidelines. ROSE is being conducted as part of a risk management plan to monitor short-term (first 3 months) safety and utilisation of rivaroxaban (Riv). A contextual comparator cohort of new user patients (pts) treated via standard care (warfarin-War) is also being identified. Study objectives include advancing the understanding of use of Riv in the secondary care setting.

Objectives: An interim analysis of treatment setting and factors influencing anticoagulant prescribing decisions.

Methods: An observational, population-based cohort study. The interim cohort was identified through a specialist network from Sep13 to Mar15 (data lock date), supported by UK Clinical Research Networks. Data collected via a questionnaire from consenting pt medical charts by specialists included reasons for prescribing such as guidelines. Descriptive statistics were calculated (% denominator assumed no missing data; sub-cohorts were pooled by indication: treatment/prevention of DVT/PE; prevention of stroke/systemic embolism in pts with non-valvular AF).

Results: Interim cohort = 2022 (54% Riv vs 46% War); median age 71yrs (IQR 59, 80); 56% (1131) male; 59% (1193) were outpts; AF cohort =1118: Riv (59%), War (51%) and the DVT/PE cohort =860: Riv (39%), War (46%). Irrespective of indication, the most frequently provided prescribing reason was clinical judgement [Riv (98%), War (82%)]. Other reasons were NICE guidance [Riv (25%), War (40%)], lifestyle impact [Riv (29%), War (3%)] and formulary guidelines [Riv (15%), War (28%)]. Similar patterns were observed when data were stratified by indication group.

Conclusions: This interim analysis shows that in UK secondary care clinical practice, prescriber clinical judgement is the overriding factor affecting treatment choice. Whilst guidelines are important, their influence on prescribing decision differs between the two drugs. Non-clinical lifestyle considerations appear to be more important for patients treated with Riv than War. A formal multi-level analysis of determinants of prescribing is planned for the final analysis.


Abstract 338: Cohort Characteristics and Determinants of Prescribing Rivaroxaban in Primary Care in England: Interim Results from a Post Authorisation Safety Study (PASS)

Sandeep Dhanda, Miranda Davies, Deborah Layton, Vicki Osborne and Saad AW Shakir

Background: A UK PASS is being conducted as part of a risk management plan to monitor the safety and use of rivaroxaban (Riv) as prescribed for medical conditions requiring anticoagulation. Increasingly, patient (pt) prescribing is influenced by published guidelines. One study objective was to evaluate Riv prescribing in real-life clinical practice in primary care in England.

Objectives: A planned interim analysis to describe cohort characteristics and determinants of prescribing.

Methods: An observational cohort study. Interim cohort was identified from dispensed prescriptions of Riv in England from Dec 2011 to Jul 2015 (datalock). Questionnaires requesting information on drug utilisation were sent to prescribing general practitioners (GPs) at ≥3, and ≥12 months after 1st Riv prescription issued for each pt. Summary descriptive statistics were calculated; % denominator where response given; pts with single indications reported were analysed within mutually exclusive groups.

Results: Interim evaluable cohort = 8372 pts (median age 75 years (IQR 64-83)); 4223 (50.5%) female; indication of non-valvular atrial fibrillation (AF) reported for 4764 pts (56.9% cohort) and venous thromboembolism (VTE) for 2286 pts (27.3% cohort). More than half of pts (4862, 59.4%) were initiated on 20mg od. Riv was initiated more frequently in secondary vs. primary care setting (4467, 55.3% vs. 3530, 43.7%). Recommendation from a specialist was the most frequent reason for prescribing (4509, 56.3%), followed by clinical judgement (2029, 25.3%), lifestyle/anticoagulation monitoring needs (1695, 21.1%) and recommendation by guidelines (1073, 13.4%).

Conclusions: The interim analysis reveals that Riv is largely prescribed within the licensed indications of AF and VTE. In terms of GP prescribing determinants, recommendation from a specialist would appear to be most influential, however approximately one fifth of GPs cite pt lifestyle factors (anticoagulation monitoring needs) as critical to the decision to prescribe. These results will become obsolete when further evaluation and validation are complete for the final report.


Abstract 339: Characterisation of Baseline Risk Factors for Bleeding Outcomes in Patients with Non-Valvular Atrial Fibrillation (AF) Prescribed Rivaroxaban in Primary Care in England: Interim Results from a Post Authorisation Safety Study (PASS) 

Sandeep Dhanda, Miranda Davies, Deborah Layton, Vicki Osborne and Saad AW Shakir

Background: Rivaroxaban, first approved in the EU in 2008, is one of the novel oral anticoagulants. To advance the understanding of the safety and use of rivaroxaban, a UK PASS is being conducted in primary
care as part of a risk management plan. Study objectives include characterisation of baseline risk of bleeding in patients (pts) with AF.

Objectives: To characterise baseline risk factors for bleeding in pts prescribed rivaroxaban in primary care in England.

Methods: An observational cohort study. Interim cohort identified from dispensed prescriptions of rivaroxaban in England from Dec 2011- Jul 2015. Risk factors for bleeding were collected from prescribing general practitioners (GPs) via questionnaires sent ≥3, and ≥12 months after 1st prescription issued for each pt. Summary descriptive statistics were calculated (% denominator is of interim AF cohort assuming no missing data; pts with other or multiple indications excluded).

Results: The interim evaluable 3-month AF cohort consisted of 4764 pts [median age 77 years (IQR 69-84)]; 2316 (48.7%) female). Relevant risk factors reported at baseline included: age>65 yrs (4193, 88.0%), stroke history (662, 13.9%), history of bleeding (114, 2.4%), excessive alcohol intake (85, 1.8%), uncontrolled hypertension (40, 0.8%), renal disease (38, 0.8%), medication use predisposing to bleeds (26, 0.6) and abnormal liver function (20, 0.4%). 4198 (88.2%) pts had a HAS-BLED score of 1 (3490, 73.3%) or 2 (708, 14.9%). 88 pts (1.9%) had a score of ≥3 indicating high risk of bleeding.

Conclusions: The results from this interim analysis show that the prevalence of stroke history is very common in patients with AF starting rivaroxaban, followed by a history of bleeding and excessive alcohol intake. A HAS-BLED score necessitating caution or regular review (i.e ≥3) was reported in 1.9% of the cohort. These results will be superseded after validation and follow-up are complete for the final analysis.


Abstract 504: Use of New Oral Anticoagulants in the Secondary Care Setting vs Primary Care in the UK: Interim Results from Two Post-Authorisation Studies

Deborah Layton, Alison Evans, Miranda Davies, Sandeep Dhanda and Saad AW Shakir

Background: The Risk Management Plan for rivaroxaban (Riv) included studies on the utilisation and safety monitoring of Riv prescribed in primary care (a Modified Prescription-Event Monitoring (M-PEM) study; all indications) and secondary care (a Specialist Cohort-Event Monitoring (SCEM) study; selected indications incl. prevention of stroke and systemic embolism in non-valvular atrial fibrillation (AF)).

Objectives: An ad hoc interim analysis to describe the baseline bleeding risk characteristics of two interim study Riv cohorts with AF.

Methods: Both studies used an observational cohort design. Data (incl.selected pt baseline characteristics as per HAS-BLED) were collected from forms sent to specialists in secondary care Sep 2013 to Mar 2015 (datalock) and General Practitioners (GPs) in primary care Dec 2011 to Jul 2015 (datalock). Descriptive statistics & univariate analyses [OR (95%CI)] were calculated (% denominator assumes no missing data). AF cohorts exclude pts with >1 indication.

Results: Interim AF cohort; SCEM=641, 53% male; M-PEM=4764, 52% male. SCEM pts were more likely than M-PEM pts to have a history of stroke [39% vs 14%; OR 3.9 (2.3,4.7)], uncontrolled hypertension [3% vs 1%; OR 3.8 (2.2,6.6)], clinical predisposition to bleeds [4% vs 2%; OR 1.7 (1.1, 2.7); or use drugs predisposing to bleeds [2% vs 1%; OR 3.5 (1.7,6.9)]. Baseline prevalence of abnormal liver function, renal disease, excess alcohol use & age 65+ yrs in the 2 cohorts were similar. The HAS-BLED score distribution differed (ranksum p<0.001); HAS-BLED≥3 [6%vs 2%; OR 3.3 (2.3, 5.0)].

Conclusions: In this ad-hoc analysis, SCEM AF pts appeared to have a higher burden of baseline bleeding risk factors than M-PEM AF pts. This appears to be as expected within healthcare systems, where complex pts with multiple morbidities may be managed by specialists in secondary care. Considerations include differences in the recording of data in medical records held by specialists compared to GPs and interim cohort sample size. Nevertheless, these findings support the need for systematic surveillance across healthcare settings to evaluate full spectrum of pts at risk.


Abstract 700: Risk Management and Specialist Cohort-Event Monitoring [ndash] Responding to Change

Deborah Layton and Saad AW Shakir

Background: It is recognised that there is a paucity of data sources to conduct studies in the EU on hospitalised patients (pts) and those seeing specialists.SCEM is method which has been used to meet the requirements of Post-Authorisation Safety Studies (PASS) to address this need for systematic large scale safety surveillance of new medicines mostly initiated by hospital specialists in a hospital or other secondary care settings (e.g. outpatient clinics). To date, four SCEM studies have been included with Risk Management Plans (RMP) of recently marketed drugs.

Objectives: To discuss design considerations of SCEM.

Methods: A description of design of SCEM, incl. pt identification, specialist research frameworks, study size, data collection and study duration.

Results: SCEM assembles new user cohorts on the basis of a common exposure (study drug). The sampling frame comprises all secondary care settings likely to use the study drug in clinical practice. Specialists register within a research network established with collaborative support from the UK NIHR Clinical Research Network.Pt inclusion criteria are minimal and eligible pts for are those whom the clinical decision to treat has been made by a specialist, prior to pt consent. NHS ethics approval is required. The desired pt sample size is powered to examine primary safety issues identified in the RMP, and study duration planned to maximise cohort accrual. Longitudinal data collection (min 3 mths per pt) is supported via specialist completed questionnaires (secondary use pt medical charts). Inclusion of a comparator cohort is possible, dependent on the availability of an ideal counterfactual treatment.

Conclusions: Since the adoption of a new medicine into clinical practice in the UK is often initially facilitated by specialists, there is a need for data capture across all clinical settings to ensure that exposed populations are characterised and monitored. SCEM studies attempt to overcome some limitations of PASS ( such as potential selection bias and under-reporting) conducted exclusively in the primary care setting, or using primary care based data sources with partial information on pts health experience since specialist initiation.


Abstract 873: Methodological Approaches to Enhanced Safety Surveillance for Seasonal Flu Vaccines: Early Experience in the UK

Lorna Hazell, Saad AW Shakir, Andrew Finlay, Hannah Coulter, Robert S Brody and Robert P Wise

Background: In 2014, the European Medicines Agency issued draft guidance for enhanced safety surveillance (ESS) on seasonal influenza (flu) vaccines. In the UK, conventional vaccine surveillance is through spontaneous reporting which is limited by under-reporting and lack of denominator data. We have therefore developed and implemented new methods for ESS.

Objectives: To describe pilot experience of ESS in the UK

Methods: In 2014/5, an ‘active’ surveillance study was conducted on an intranasal flu vaccine used in children (Q-LAIV: quadrivalent live attenuated influenza vaccine). Vaccinees were invited to participate after routine vaccination at participating GP practices and schools. Postal or electronic questionnaires solicited data on adverse events (AE) of interest 14 days after vaccination. Incidence rates were calculated for reported AEs.

In contrast, for 2015/6, a ‘passive’ ESS programme was conducted on the same vaccine. Vaccinees were handed a Safety Report Card (SRC) after vaccination and asked to report any suspected adverse drug reactions (sADRs) directly to the vaccine manufacturer. Reporting rates for sADRs were calculated using concurrent denominator data provided by immunisation sites.

Both pilots were collaborations between DSRU and AstraZeneca, were supported by the UK Clinical Research Network and received Ethics Committee approvals.

Results: In the ‘active’ design, 385 (64%) of 600 recruited vaccinees responded (46 immunisation sites), of which 237 experienced AEs. In the ‘passive’ ESS, 165 vaccinees reported sADRs (1.9% of 8,753 SRCs issued; 67 sites). In both years, the most frequently reported symptoms were pyrexia, malaise, rhinorrhoea, cough and headache. One serious event was reported in the passive phase (flu-like symptoms requiring hospitalisation).

Conclusions: Both active and passive ESS are feasible methods to collect patient-reported safety outcomes that may otherwise go unrecorded. Whilst both methods can capture denominator data, larger samples are needed to detect rare events. Reporting rates in the passive model were low. Promotion of safety surveillance, however, requires caution so as not to affect vaccine uptake.

DSRU, a not-for-profit organisation, and AstraZeneca who provided funding, collaborated on both pilot studies.


Abstract 874: Passive Enhanced Safety Surveillance In Children Receiving Fluenz® Tetra Vaccination In England During The Early 2015/2016 Influenza Season

Lorna Hazell, Saad AW Shakir, Andrew Finlay, Hannah Coulter, Robert S Brody and Robert P Wise

Background: Fluenz® Tetra is a quadrivalent, live attenuated, intranasal, influenza vaccine recommended for use in children vaccinated as part of the seasonal influenza immunisation programme in the UK. We report results from a pilot safety surveillance programme in England consistent with regulatory guidance for all influenza vaccines.

Objectives: To measure and assess the frequencies of suspected adverse drug reactions (sADRs) in children receiving Fluenz® Tetra during the early 2015/2016 influenza season in England.

Methods: Passive enhanced safety surveillance was conducted through stimulated spontaneous reporting of sADRs. Vaccinees or parents/guardians received a Safety Report Card (SRC) to return if children experienced sADRs after vaccination with Fluenz® Tetra; no time limit for reporting was specified. At participating sites, 42 general practices and 25 primary schools in England, immunisation teams provided numbers of SRCs distributed. The study was approved by an NHS Research Ethics Committee (North West – Liverpool East).

Results: Between 8th October 2015 and 10th January 2016, 8,753 SRCs were issued for 4,134 children (47.2%) aged 2 to 4 years, 4,078 (46.6%) aged 5 to 10 years and 541 (6.2%) aged 11 to 17 years. Of 323 SRCs returned during this period, 165 reported at least one sADR. The most frequently reported sADRs were rhinorrhoea (n=54), cough (35), and pyrexia (31). One serious and unexpected sADR involved a child with flu-like symptoms requiring a hospital visit 2 days after vaccination. The child recovered from the sADR.

Conclusions: These data are broadly comparable with the frequency of adverse events reported for Fluenz® Tetra in clinical trial and post-marketing data, despite differences in methods. No evidence from the limited data available suggests an increased frequency of minor expected events or other safety signals.

Study co-sponsored by DSRU and AstraZeneca.


Abstract 1090: Changes in Health Parameters Over 12 Months in Users of Once Weekly Exenatide (Bydureon[reg])

Abigail Coughtrie, Deborah Layton, Qing Qiao and Saad Shakir

Background: Bydureon®, licensed in 2011, is indicated for treatment of type 2 diabetes mellitus (T2DM). Favourable changes in health parameters in T2DM patients (pts) can help reduce cardiovascular disease risk and improve glucose homeostasis. A post-marketing safety study is underway as part of the EU Risk Management Plan.

Objectives: To quantify mean changes at group level in health parameters (weight, body mass index (BMI), systolic blood pressure (SBP) and HbA1c) in Bydureon® users at interim.

Methods: An observational, population-based cohort study in primary care in UK. Pts were identified from dispensed prescriptions (Rx) issued by GPs Sep 2011-Sep 2015 (interim datalock). GPs were sent questionnaires 12 months after each pts’ 1st Rx for exposure, outcome and risk factor information from medical charts. For pts continuing treatment to the end of 12 months, changes in each pts’ health parameters were calculated from index (1st Rx) to 12 months post-index (12m) to estimate mean change (Δ) at group level. Denominators are pts with both index and 12m data provided.

Results: Interim evaluable cohort=2538 pts; 55.6% (n=1410) males; median age 57 yrs (IQR 50;65). At end of the 12 months 1752/2538 (69.0%) pts remained on treatment. Of these pts, 1384/1495 (92.6%) were obese and 521/1129 (46.1%) had HbA1c>9% at index. Mean weight Δ=-3.0kg [SD 6.7] (index 109.4kg [22.0]; 12m 106.4kg [21.7]; N=1395). Mean BMI Δ=-0.9kg/m2 [2.5] (index 38.0kg/m2 [6.7]; 12m 37.0kg/m2 [6.6]; N=1292). Mean SBP Δ=-2.1mmHg [16.2] (index 133.8mmHg [14.2]; 12m 131.6mmHg [14.2]; N=1509). Mean HbA1c Δ=-7.6mmol/mol [18.9] (index 75.9mmol/mol [19.3]; 12m 68.3mmol/mol [19.5]; N=1057).

Conclusions: Overall at aggregate level.,Bydureon® use appears to be associated with favourable changes in health parameters over 12 months in this single group study in primary care setting. However the clinical
significance of such changes is unclear. Limitations are lack of adjustment for concurrent medications and potential regression towards the mean as pts prescribed Bydureon® may be extreme relative to the T2DM population (obese and/or poor diabetes control). This analysis will be superseded once final cohort accrual and analysis are complete.