35th International Conference on Pharmacoepidemiology & Therapeutic Risk Management (ICPE), Philadelphia, August 2019

Title

Characterisation of Drug Interaction Related Signals Leading to EU Regulatory Action and a Methodological Review of Novel Drug Interaction Signal Detection Methods in the Post-Marketing Setting

Debabrata Roy1, 2, Lorna Hazell1, 2, Vicki Osborne1, 2 and Saad Shakir1, 2.

1Drug Safety Research Unit, Southampton, United Kingdom

2University of Portsmouth, Portsmouth, United Kingdom

Background

A growing rise in polypharmacy is often attributed to an aging population and associated co-morbidities. In patients with complex therapeutic regimens, this increases the potential of harmful drug interactions (DI’s) which may result in unexpected adverse drug reactions. Current signal detection methods used to identify DI-related signals may not be sufficient. Novel methods utilizing big data sets, machine learning or algorithmic techniques may have a crucial role in DI-related signal detection.

Objectives

The aim of this study is to assess current and novel methods used to identify DI-related safety signals in the post-marketing setting.

Methods

Current methods used to detect DI-related signals will be reviewed using publicly available information from the European Medicines Agency (EMA) on signals leading to post-marketing regulatory action between July 2012 and December 2018. A review of the published literature will also be conducted to identify novel methods used to detect DI-related signals which may have utility in the post-marketing setting. A systematic approach will be used to assess the performance of novel DI signal detection methods based on predefined criteria including, for example, sensitivity, precision and applicability.

Results

Data collection, identification of valid data sources and methodological review are currently in progress. So far, one DI-related signal which led to post-marketing regulatory action has been identified from the search whereby dehydration was associated with an interaction between tolvaptan (vasopressin antagonist used to treat inappropriate antidiuretic hormone secretion) and diuretic use. This DI-related signal was identified in 2012 based on 16 spontaneous reports. The identification of this signal resulted in an update to product information, regarding a possible interaction between tolvaptan and diuretic use and the risk of renal dysfunction.

Conclusions

The final results of this study will identify whether the majority of DI-related signals in the post-marketing setting are identified from individual case safety and spontaneous reports. Novel methods, utilizing big data sets and advancements in machine learning and computational power, have the potential to identify previously unknown and unexpected DI-related signals.

Title

Impact Analyses of European Pharmacovigilance Interventions on Public Health Burden

S Lane, E Lynn, J Slattery, S Shakir.

Background

Since implementation of the 2012 pharmacovigilance (PV) legislation in the EU, monitoring the success of PV has become commonplace; however impact of PV regulatory interventions on public health remains mostly unquantified.(3) There is a need for research to identify the best technique for measuring impact of these actions, and guidelines on reporting impact analyses.(3) The overall aim of this study is to quantify the impact of EU marketing authorisation withdrawal, revocation or suspension (“PV regulatory interventions”) in terms of morbidity changes.

Objectives

Interim analysis to assess the feasibility of predictive modelling techniques for estimation of public health impact of PV regulatory interventions.

Methods

Prescription products used in primary care whose marketing authorisations were withdrawn, revoked or suspended in France, Germany and the UK between 2012 and 2016 were previously identified.(4) Annual product utilisation figures for the year prior to PV action were estimated using IMS France, IMS Germany and the British Health Improvement Network (THIN) electronic health record databases. Systematic searches of PubMed/MEDLINE, and European Medicines Agency and national competent authority websites and documents were conducted to identify quantitative studies for the product and adverse drug reaction (ADR) of interest, allowing risks to be calculated. The public health impact of the intervention was estimated in terms of morbidity reduction due to product withdrawal, revocation or suspension by modelling of usage figures and risk of each ADR. Work continues to obtain background risk data to establish the actual reduction in morbidity attributable to removal of the product from market in this territory.

Results

18 products were considered for impact analysis; 9 were excluded as no quantitative studies were identified for the respective ADRs. This interim analysis focused on ketoconazole, metoclopramide and domperidone, and provided a prediction of the number of ADRs avoided per year as a result of marketing authorisation withdrawal, revocation or suspension, and an estimation of the public health impact of each PV action using changes in morbidity as an indicator.

Conclusion

This interim analysis tested a method for predicting public health impact of PV interventions on a subset of products based on drug utilisation data and expected changes in morbidity. Results suggest the method could be useful in determining public health impact of future PV actions. The predictive modelling method will be further evaluated prior to completion of this study.

References

• Mammi M, Citraro R, Torcasio G, Cusato G, Palleria C, di Paola ED. Pharmacovigilance in pharmaceutical companies: An overview. J Pharmacol Pharmacother. 2013;4(Suppl 1):S33-7.
• Mazzitello C, Esposito S, De Francesco AE, Capuano A, Russo E, De Sarro G. Pharmacovigilance in Italy: An overview. J Pharmacol Pharmacother. 2013;4(Suppl 1):S20-8.
• Goedecke T, Morales D, Pacurariu A, Kurz X. Measuring the impact of medicines regulatory interventions – systematic review and methodological considerations. Br J Clin Pharmacol. 2017.
• Lane S, Lynn E, Shakir S. Investigation assessing the publicly available evidence supporting postmarketing withdrawals, revocations and suspensions of marketing authorisations in the EU since 2012. BMJ Open. 2018;8(1).

Title

Monitoring the Safety of Exenatide Once-Weekly (Bydureon®) in Primary Care in England: Results from a PASS

Authors: Sandeep Dhanda 1, 2, Vicki Osborne 1, 2, Debabrata Roy 1, 2, Saad Shakir 1, 2

Affiliations: 1 Drug Safety Research Unit, 2 University of Portsmouth

Background

Clinical trials have reported specific safety concerns such as acute pancreatitis (AP) in patients (pts) taking Bydureon® (exenatide once-weekly) for type 2 diabetes mellitus (T2DM). A Post-Authorisation Safety Study (PASS) was conducted to monitor the real-world safety of Bydureon® in primary (1o) care in England.

Objectives

To estimate the incidence of targeted safety outcomes in pts prescribed Bydureon® in 1o care.

Methods

Pts identified from dispensed Bydureon® prescriptions in England (2012-2016). Questionnaires sent to prescribing general practitioners (GPs) at ≥12-months observation collected event information. Events of interest included AP, pancreatic cancer (PC), thyroid neoplasm (TN), gallstones, biliary colic or cholecystitis (GBC), acute renal failure (ARF), type 1 hypersensitivity (T1H), and cardiac events (CE). 12-month incidence estimates (on Bydureon® or ≤10 weeks after stopping) were calculated; results were stratified according to prior exenatide use (i.e. Byetta®).

Results

Questionnaire response rate= 37.2% (7752/20860). Total cohort = 6294 pts prescribed Bydureon® for T2DM (median age 57 years [IQR 50, 65]; 55.2% male). Exenatide naïve n=4556 (72.4%), previous Byetta® users n=1629 (25.9%), previous Byetta® use unknown n=109 (1.7%).

Risk of AP in total cohort 0.2% (95% CI [0.1, 0.4]; n=14). 2 pts had a prior history. Risk of AP was similar for exenatide naïve (0.2% (95% CI [0.1, 0.4]; n=10) and previous Byetta® users (0.2% (95% CI [0.0, 0.5]; n=3). Rate of AP in total cohort was 2.5/1000 person-years (95% CI [1.5, 4.3]) and no statistically significant difference was observed between the 2 user groups. Time-to-event analyses suggested no clear pattern in the hazard function of AP over time. For 12 of the 14 pts, Bydureon® was stopped due to AP. AP complications; necrosis/pseudocyst n=1, fatal outcome n=1.

PC 0.1% (n=4); a fatal outcome was reported in 3 of these pts. TN 0.0% (n=0). GBC 0.6% (95% CI [0.4, 0.8]; n=38); 15 of which had a prior history of GBC. ARF 0.5% (95% CI [0.3, 0.7]; n=29). T1H 0.7% (95% CI [0.5, 0.9]; n=44). CE 3.6% (95% CI [3.2, 4.1]; n=227). No statistically significant differences in risk between the 2 user groups were observed.

Conclusions

Incidence of AP was low in Bydureon® users and consistent with prior clinical trial/observational data. No unexpected findings were identified. This study has unique strengths, including collection of timely, granular real-world data from GPs, facilitating accurate estimates. The study is part of a broader literature on the safety of Bydureon® and conclusions should be made in context with other post-marketing findings.

Disclosure

The DSRU is an independent charity (No 327206), which works in association with the University of Portsmouth. The DSRU has received funding from AstraZeneca AB, the manufacturer of Bydureon®. The DSRU makes the final decision on the publication of external communications.

Title

Utilisation of Exenatide Once-Weekly (Bydureon®) in Primary Care in England: Results From a PASS

Authors: Sandeep Dhanda 1, 2, Vicki Osborne 1, 2, Debabrata Roy 1, 2, Saad Shakir 1, 2

Affiliations: 1 Drug Safety Research Unit, 2 University of Portsmouth

Background

Exenatide extended-release (Bydureon®) is a glucagon-like peptide-1 receptor agonist administered once-weekly by subcutaneous injection. Approved in 2011, Bydureon® is licensed to improve glycaemic control in combination with other antidiabetes medications (ADMs) in patients (pts) with type 2 diabetes mellitus (T2DM) when current therapy, with diet and exercise, is inadequate. A Post Authorisation Safety Study (PASS) was conducted to monitor the use and safety of Bydureon® in primary (1o) care in England.

Objectives

To describe the utilisation of Bydureon® in 1o care.

Methods

A cohort study identified pts from dispensed Bydureon® prescriptions in England (2012-2016). Pt characteristics and drug utilisation data was collected from prescribing general practitioners (GPs) via questionnaires sent ≥12 months after the 1st Bydureon® prescription issued for each pt. Summary descriptive statistics were calculated.

Results

Questionnaire response rate = 37.2% (7752/20860). Evaluable cohort = 6294 pts prescribed Bydureon® for T2DM (median age 57 years [IQR 50, 65]; 55.2% male). 2 pts were <18 years. 16 pts had an off-label indication of type 1 diabetes mellitus and were excluded from analyses. At baseline, where specified, 91.7% of pts (n=2390) were obese (BMI ≥30 kg/m2) and/or had an HbA1c ≥7.5% (n=1986, 90.0%); of those pts with a raised HbA1c, 66.0% (n=1311) had measurements of ≥9.0% indicating very poor diabetic control. Bydureon® was mostly prescribed in 1o care (n=3269, 51.9%); ‘specialist decision’ was the overwhelming reason for prescribing (n=3113, 49.5%). The majority of pts were exenatide naïve (n=4556, 72.4%), whilst 25.9% (n=1629) were previous Byetta® (exenatide twice-daily) users. For 109 pts (1.7%) prior exposure to Byetta® was unknown. Nearly all pts (n=5948, 94.5%) were taking Bydureon® 2mg once/week, in accordance with the product label. Pts predominantly started Bydureon® as triple therapy (n=3876, 61.6%) or dual therapy (n=1936, 30.8%); monotherapy use was lower (n=161, 2.6%). Metformin (n=5753, 81.5%) and sulphonylureas (n=2382, 45.0%) were the most frequent concomitant ADMs at index.

Conclusions

Bydureon® was largely used in accordance with prescribing recommendations in 1o care in England, with most frequent use in pts with poor diabetic control where treatment with alternative ADMs may have been unsuccessful. In addition, characteristics of pts prescribed Bydureon® are in keeping with the T2DM pt profile. This study design allowed for the timely collection of drug utilisation data directly from GPs and suggests shared Bydureon® T2DM treatment arrangements between 1o and 2o care.

Disclosure

The DSRU is an independent charity (No 327206), which works in association with the University of Portsmouth. The DSRU has received funding from AstraZeneca AB, the manufacturer of Bydureon®. The DSRU makes the final decision on the publication of external communications.

Title

Utilisation Of Dulaglutide In Primary Care In England: Interim Results From A Post Authorisation Safety Study

Authors: Miranda Davies 1, 2, Lorna Hazell 1, 2, Sandeep Dhanda 1, 2, Vicki Osborne 1,2, Ayad K Ali 3, Saad Shakir 1, 2

Affiliations: 1 Drug Safety Research Unit, 2 University of Portsmouth, 3 Eli Lilly and Company

Background

Dulaglutide is a once-weekly injectable long acting glucagon-like peptide-1 receptor agonist (GLP-1 RA) authorised in England for the treatment of type 2 diabetes mellitus (T2DM).

Objectives

To assess the utilisation of dulaglutide in real-world primary care practice in England.

Methods

A descriptive cohort study (Modified Prescription-Event Monitoring) identified dulaglutide users from dispensed primary care prescriptions of dulaglutide between 2015-2017 in England. Patient characteristics and drug utilisation data were collected from prescribing general practitioners (GPs) via questionnaires sent ≥12 months after the 1st prescription issued for each patient. Summary descriptive statistics were calculated.

Results

1467 evaluable dulaglutide users with T2DM diagnosis confirmed. Median age: 58 yrs (IQR 51, 66); 775 males (52.8%). Baseline obesity (Body Mass Index (BMI) ≥30 kg/m2) was present in 45.9% of the cohort (89.9% where BMI specified), while poor T2DM control, indicated by Haemoglobin A1c ≥58 mmol/mol (7.5%) was observed in 60.1% of patients (93.1% where specified). Median duration of T2DM was 10.0 yrs (IQR 6.0, 14.1). The decision to initiate dulaglutide was most frequently made by a specialist nurse (n=766, 52.2%) followed by the GP (n=352, 24.0%), and hospital doctor (n=299, 20.4%). Most patients were prescribed 0.75mg once- weekly (n=233, 15.9%) or 1.5mg once-weekly (n=762, 51.9%). There were a small number of reports (n=14, 0.9%) of >once-weekly usage. In the majority of patients, dulaglutide was prescribed as either add-on dual therapy (n=469, 32.0%) or add-on triple therapy (n=908, 61.9%); the most frequently reported concomitant anti-diabetes medication (ADM) was metformin (n=624, 42.5%). In sub-populations not previously studied in clinical trials, the most frequently reported usage was in patients ≥75 yrs (n=80, 5.5%).

Conclusions

This interim analysis suggests that dulaglutide is largely being prescribed in accordance with prescribing recommendations and national T2DM management guidelines. Results suggest most frequent use in patients with poor diabetic control where existing treatment with alternative ADM may not have achieved optimum control. This preliminary analysis of baseline data will be updated in the final report.

Disclosure Statement: The DSRU is an independent charity (No 327206) which works in association with the University of Portsmouth. The DSRU has received funding from Eli Lilly and Co., the manufacturer of Trulicity®.

Title

The Association between Direct Oral Anticoagulants and Gynaecological Bleeding in Women v. Standard Therapy: A Systematic Review

Authors: Sandeep Dhanda 1, 2 Vicki Osborne 1, 2, Saad Shakir 1, 2

Affiliations: 1 Drug Safety Research Unit, 2 University of Portsmouth

Background

Women receiving anticoagulation are susceptible to the added risk of gynaecological bleeding (GB). This may occur in premenopausal women who experience physiological bleeding related to the ovulatory cycle and for whom the risk is greater, or postmenopausal women. Thus, the use of anticoagulation in women requires special consideration in terms of GB risk. It may be that DOACs overall or a particular DOAC are the wrong choice of treatment for women at risk.

Objectives

To examine the association between DOACs and GB in women, by comparing risk in users of DOACs to users of standard anticoagulant (AC) therapy (e.g. vitamin K antagonists (VKAs)).

Methods

4 databases were searched (July 2018) for clinical trials (CTs) and observational studies (OS) reporting GB in women taking DOACs or a comparator of standard AC therapy for any indication. Abstracts were screened followed by full review of a subset for eligibility. Studies reporting GB with DOACs but with no standard AC comparator were excluded. Due to a high degree of heterogeneity between studies in terms of study design, indications, outcome measures and use of different denominators, a meta-analysis was not performed.

Results

Records identified n=594; 16 eligible for inclusion (CTs n=7, OS n=9). Of the 6 studies specifically designed to evaluate GB in women with DOACs v. standard AC, 3 reported a significantly increased risk of this outcome with DOACs v. standard AC (rivaroxaban (riva) n=2 (HR 2.13; 95% CI [1.57, 2.89], risk ratio 2.3; p=0.009), and edoxaban (edox) n=1 (HR 1.7; 95% CI [1.1, 2.5])). Conversely, no significant difference for 2 studies (riva n=1, apixaban (apix) n=1) and a lower risk for 1 (dabigatran (dabi)) was observed (OR 0.59; 95% CI [0.30, 0.90]). The remaining 10 studies were not designed to evaluate GB with DOACs but reported this within study results; 1 reported a higher risk of GB with dabi v. standard AC (HR 2.27; 95% CI [1.32, 3.90]), and for the remaining 9, results were inconclusive due to non-significant/lack of hypothesis test results.

Conclusions

Gynaecological bleeding appears to be a common complication of DOACs, however, there is variability between DOACs. More evidence is available for riva, in which the risk of GB appears to be higher than with standard AC. Although limited, evidence suggests a higher risk with edox, lower risk with dabi and no difference with apix v. standard AC. Further research including head-to-head CTs and OS specifically investigating GB is needed to draw firm conclusions on which class of ACs (VKAs v. DOACs) are more favourable for women and the difference in risk between individual DOACs.

Title

Risk Factors of Major Bleeding in Patients Prescribed Rivaroxaban in Primary Care in England

Roy* 1, 2, S. Dhanda1, 2, L. Wise1, S. Shakir1, 2

1Drug Safety Research Unit, Southampton, 2University of Portsmouth, Portsmouth, United Kingdom

Background

The association between oral anticoagulant use and bleeding risk has been widely reported. However, patients may have underlying risk factors which predispose to bleeding. It is valuable to understand the predictors for major bleeding in patients prescribed rivaroxaban.

Objectives

Multivariable logistic regression analyses to explore potential risk factors for major bleeding within gastrointestinal, urogenital and intracranial sites.

Methods

A case/non-case design evaluated the association between clinical risk factors and major bleeding in rivaroxaban patients (N=17546) enrolled to a 12-month cohort study in England (2012-2016). Clinical risk factors for bleeding and bleeding outcomes were collected from general practitioners via questionnaires sent at ≥3, and ≥12 months observation. Multivariable logistic regression analyses examined the association for each site and models were based on two approaches; clinical risk factors selection model and HAS-BLED model. The clinical risk factors selection model included all reported clinical risk factors for bleeding. The HAS-BLED model included HAS-BLED risk score categories (low, moderate, high) and gender. Statistically significant (p<0.05) associations are presented in the results.

Results

Gastrointestinal major bleeds (n=176)

Clinical risk factors selection model

Age 65-74 vs. <65 years OR 2.4 [95% CI 1.3, 4.6]; Age ≥75 vs. <65 years OR 4.2 [95% CI 2.3, 7.5]; Predisposition to/history of bleeding OR 4.8 [95% CI 3.1, 7.5] HAS-BLED model Moderate vs. low OR 4.0 [95% CI 2.1, 7.6]; High vs. low OR 8.9 [95% CI 4.0, 19.9] Urogenital major bleeds (n=36) Clinical risk factors selection model Age 65-74 vs. <65 years OR 0.2 [95% CI 0.1, 0.7]; Female vs. male OR 2.9 [95% CI 1.4, 6.1]; Malignancy OR 2.6 [95% CI 1.1, 6.3] HAS-BLED model Female vs. male OR 2.7 [95% CI 1.3, 5.6] Intracranial major bleeds (n=57) Clinical risk factors selection model Age ≥75 vs. <65 years OR 2.8 [95% CI 1.1, 6.9]; History of cerebrovascular accident (including haemorrhagic)/transient ischaemic attack OR 2.2 [95% CI 1.3, 3.9]; Predisposition to/history of bleeding OR 2.6 [95% CI 1.0, 6.7] HAS-BLED model Moderate vs. low OR 3.3 [95% CI 1.2, 9.1]; High vs. low OR 9.0 [95% CI 2.5, 31.9]

Conclusion

Overall, findings from both models are in keeping with known clinical risk factors for bleeding. For urogenital major bleeds the higher risk in females and younger age group may be related to menstrual bleeding. In clinical practice it is recommended that a bleeding risk assessment, including individual patient characteristics, should be performed prior to anticoagulation.

Disclosure of Interest

The DSRU is an independent charity (No 327206) which works in association with the University of Portsmouth. The DSRU has received funding from Bayer, the manufacturer of Xarelto®. The DSRU makes the final decision on the publication of external communications.