Adms 1010 Assignment 10-1

1. Modelling of Endpoint Postponement for All-Cause Mortality in Statin Trials

Morten Rix Hansen,1,2 Anton Pottegård,1 Asbjørn Hróbjartsson,3 Per Damkier,1,2 René D Christensen,3 Kasper Søltoft Larsen,1 Malene EL Kristensen,1 Palle M Christensen,1 Jesper Hallas.1,21Clinical Pharmacology, University of Southern Denmark, Odense, Denmark;2Department of Clinical Chemistry and Pharmacology, Odense University Hospital, Odense, Denmark;3The Nordic Cochrane Centre, Rigshospitalet, Copenhagen, Denmark;4Research Unit for General Practice, University of Southern Denmark, Odense, Denmark.

Background: The average postponement of the outcome event has been proposed as a novel method to present the magnitude of effect for preventive medications. This measure has been shown to have better agreement with patient preferences than conventional outcome measures, including the “number needed to treat” (NNT), possibly because it is more intuitively understood. For some interventions, it may also provide a better theoretical frame for how benefit is distributed among participants than the NNT measure. The aim of this study was to present a novel method for modelling endpoint postponement (EP) from trial data and compare it with the usual approach of measuring the area between survival curves. We also present a formalized meta-analysis of modelled EP for all-cause mortality in statin trials.

Methods: We identified 17 placebo-controlled statin trials that fulfilled our inclusion criteria. Eleven of these presented Kaplan–Meier curves for all-cause mortality. Average EP was calculated as the area between Kaplan–Meier curves by counting pixels on magnified prints for these 11 trials. The modelled EP was computed for all trials on the basis of (1) hazard ratio, relative risk or odds ratio; (2) the cumulative event rate in the untreated group; and (3) the trial's running time. The underlying assumption was that the mortality was reasonably stable within the trials' running time. The modelled EP was subjected to a meta-analysis, using inverse variance weighting in a random effect model.

Results: EPs were generally small for estimates based on pixel-counting, −10 and 27 days for trials both primary and secondary intervention that typically ran over 1.9–6.1 years. The modelled EPs varied between −2 and 34 days. The difference between modeled EP and EP based on pixel-counting was between −8 and 12 days.

The results of the meta-analyses will be presented at the meeting.

Conclusions: Based on these trial data, statin treatment results in a surprisingly small gain in average survival. Our modelled EP estimates agreed reasonably with EPs based on pixel-counting. The modeled EP is amenable to meta-analyses and may be a useful approach to presenting the benefit of preventive treatment.

2. Permanent User Bias in Case–Crossover Studies in Pharmacoepidemiology

Jesper Hallas,1,2 Shirley V. Wang,2 Joshua J. Gagne,2 Sebastian Schneeweiss,2 Anton Pottegård.11Clinical Pharmacology, University of Southern Denmark, Odense C, Denmark;2Division of Pharmacoepidemiology and Pharmacoeconomics, Brigham and Women's Hospital, Harvard Medical School, Boston, MA, USA.

Background: In pharmacoepidemiology, the case–crossover design is based on cases that have contrasting drug exposure at the time of an event and at a reference time in the past. If the drug in question should be taken permanently, only certain exposure patterns will occur. These patients cannot be unexposed at the event time and exposed at the reference time, while the opposite pattern can occur if the drug was initiated recently. The resulting odds ratio (OR) would thus be biased upward. As many drugs have a subpopulation of permanent users, this bias might pervade many case–crossover analyses of drug effects.

Objectives: The aims of this study were to demonstrate this “permanent user bias” and to evaluate whether it can be remedied by including a control group (case–time–control design).

Methods: Using nationwide Danish data resources, we conducted case–crossover and case–time–control analyses for combinations of three exposures that are often intended to be used permanently (statins, insulin, and thyroxin) and three outcomes (retinal detachment, wrist fracture, and ischemic stroke), where the true causal relations were expected to be null. Controls were matched on age, gender, and index date, and exposure was ascertained at 2-month intervals over the preceding 12 months.

Results: For retinal detachment, the case–crossover OR was 1.60 (95% confidence interval (CI): 1.42–1.80) for statins, 1.40 (CI: 1.02–1.92) for thyroxin, and 1.53 (CI: 1.04–2.24) for insulin. Estimates for the control population were nearly identical, leading to near-null case–time–control estimates for the three drug classes. For the wrist fracture and stroke outcomes, case–time–control ORs were consistently above unity (1.09, 1.51, and 1.15 for wrist fracture, and 2.27, 1.87, and 1.67 for stroke), suggesting significant residual bias.

Conclusions: In case–crossover studies of drugs, permanent users confer a moderate bias upward, which is partly remedied by using a control group. Additional research is needed to identify the optimal strategy for selecting this control group.

3. Controlling for Frailty in Cancer Comparative Effectiveness Studies of Older Adults

Jennifer L. Lund, Hanna K. Sanoff, Til Sturmer. University of North Carolina at Chapel Hill, Chapel Hill, NC, USA.

Background: Older cancer patients often have multiple comorbidities and functional deficits, which likely impact treatment decisions and outcomes. Using databases that lack functional information may lead to biased estimates of real world comparative effectiveness (CE).

Objectives: The aim of this study was to evaluate the impact of controlling for markers of frailty in a CE study of adjuvant chemotherapy for non-metastatic rectal cancer.

Methods: We identified a cohort of 1404 older (65+ years) non-metastatic rectal cancer patients from 2004 to 2009 using the Surveillance, Epidemiology and End Results-Medicare data, who underwent neoadjuvant therapy and surgery and survived 120 days. Using propensity score methods, we evaluated the CE of adjuvant chemotherapy versus observation on mortality, incrementally adding (i) basic confounders (demographics, cancer features, neoadjuvant treatment, comorbidities); (ii) 17 claims-based frailty indicators (e.g., oxygen use, sepsis); and (iii) 30-day post-surgical hospitalization. Among those receiving adjuvant chemotherapy, we evaluated the CE of adjuvant oxaliplatin versus 5-flurouracil (5-FU) on mortality using the same confounder sets. Standardized mortality ratio weighted Cox proportional hazards models were used to estimate adjusted hazard ratios (aHRs) and 95% confidence intervals.

Results: In total, 738 patients (52%) received adjuvant chemotherapy; 52% received oxaliplatin. Overall mortality was 9.8 per 100 person-years (mean follow-up: 3 years). The crude HR for adjuvant chemotherapy versus observation and mortality was 0.68 (0.56, 0.83); after basic confounder adjustment, the estimate was stable (aHR = 0.68 (0.54, 0.85)). Adjustment for frailty markers attenuated the aHR (0.71 (0.56, 0.90)), and inclusion of post-surgical hospitalization led to further attenuation (aHR = 0.75 (0.59, 0.95)). Among patients receiving adjuvant chemotherapy, the crude HR comparing oxaliplatin versus 5-FU on mortality was 1.0 (0.72, 1.39); adjustment for basic confounders and additional frailty markers produced similar results.

Conclusions: Our results suggest that adjustment for markers of frailty and post-surgical hospitalization may improve the validity of cancer CE studies using non-active comparators.

4. Impact of Violations of the Assumptions of the Self-controlled Case Series Design in Pharmacoepidemiological Studies: An Example of Antidepressants Use and the Risk of Hip Fracture

Md Jamal Uddin,1 Rolf H. H. Groenwold,2 Mohammed Sanni Ali,2 Gianmario Candore,3 Mark C. H. de Groot,4 Patrick C. Souverein,4 Yolanda Alvarez,3 Svetlana V. Belitser,4 Arno W. Hoes,2 Anthonius de Boer,4 Kit C. B. Roes,2 Olaf H. Klungel,4 Helga Gardarsdottir.41Department of Statistics (Epidemiology and Biostatistics), Shahjalal University of Science and Technology, Sylhet, Bangladesh;2Julius Center for Health Sciences and Primary Care, University Medical Center Utrecht, Utrecht, The Netherlands;3European Medicines Agency, EMA, London, UK;4Division of Pharmacoepidemiology and Clinical Pharmacology, University of Utrecht, Utrecht, The Netherlands.

Background: The self-controlled case-series (SCCS) design has been applied to control for time-fixed (un)measured confounding in pharmacoepidemiological studies. Although previous studies acknowledged that violations of the key SCCS assumptions lead to biased exposure effects, little is known about the impact of the violations in empirical studies.

Objectives: The aim of this study was to evaluate the impact of various levels of violation of assumptions of the SCCS design and different definitions of observation/risk periods in a study of antidepressants use and the risk of hip/femur fracture (HF).

Methods: Information on adults with an HF who used antidepressants at any time during the observation period 2001–2009 was extracted from the UK THIN (6632 cases) and the Dutch Mondriaan (136 cases) databases. The incidence rate ratio (IRR) using this design was defined as the rate of events during exposed periods and during all other observed periods. The IRR of HF was estimated using conditional Poisson regression.

Results: The IRRs appeared extremely biased when all subjects were censored at their first/last HF or when the analysis was restricted to subjects experiencing hip fracture after initiating antidepressant use. For example, in THIN, IRRs for >365 days of exposure were 1.26 [1.13–1.42] when complete follow-up was considered and 40.1 [32.2–49.9] when censoring was at the first event. However, modest censoring at the first or last event (up to 20%) had a minor impact on the IRRs. Additionally, results were consistent when including subjects who were exposed at the start of follow-up and for different risk period definitions.

Conclusions: The SCCS design is sensitive to violations of the assumptions and yields apparently biased estimates when a significant number of subjects are censored at the event or when the analysis is restricted subjects who experienced hip fracture after initiating antidepressants. The performance of this design may differ across studies and across databases. Therefore, in each SCCS study, correct specification of the SCCS design should be carefully assessed and reported.

5. Probabilistic Multiple-Bias Analyses of Observational Studies on Narcolepsy Following Vaccination with GlaxoSmithKline's Inactivated Adjuvanted (AS03) A/H1N1pdm09 Pandemic Influenza Vaccine

Kaatje Bollaerts,1 Vivek Shinde,2 Gaël Dos Santos,3 Germano Ferreira,4 Vincent Bauchau,4 Catherine Cohet,4 Thomas Verstraeten.11P95 Pharmacovigilance and Epidemiology Services, Leuven, Belgium;2GlaxoSmithKline Vaccines, Philadelphia, PA, USA;3Business & Decision Life Sciences, Brussels, Belgium;4GlaxoSmithKline Vaccines, Wavre, Belgium.

Background: An increase in the incidence of narcolepsy was first observed in Finland and Sweden towards the end of the 2009 H1N1 influenza pandemic. Preliminary epidemiological studies suggested a temporal association with GlaxoSmithKline's (GSK) Dresden-manufactured A/H1N1pdm09 vaccine, leading to a number of additional studies across Europe. Given the public health urgency to investigate the signal, these studies used readily available retrospective data from various sources. The potential for bias in such settings was generally acknowledged. Although several health authorities advocate quantifying the potential impact of biases, this was not systematically carried out in any of the narcolepsy studies.

Objectives: The aim of this study was to quantify the impact of a cascade of potential bias and confounding on the association between GSK's A/H1N1pdm09 vaccine and narcolepsy.

Methods: We apply bias-level multiple-bias analyses to two published studies on the association of the vaccine with narcolepsy: a paediatric cohort study from Finland and a case–control study from France. In particular, we developed Monte Carlo simulation models based on formal models of bias and confounding to evaluate a potential cascade of biases, including confounding by indication and natural H1N1 influenza infection, selection bias, and disease and exposure misclassification. All bias parameters were evidence based to the extent possible.

Results: Given the assumptions made and when accounting for all potential sources of bias, the rate ratio of 13.78 (95%CI: 5.72, 28.11) in the Finnish study was reduced to 4.88 (2.5th to 97.5th percentile: 1.91, 10.84) and the odds ratio of 5.43 (95%CI: 2.6, 10.08) in the French study to 1.93 (2.5th to 97.5th percentile: 0.78, 4.04).

Conclusions: The observed association between GSK's A/H1N1pdm09 vaccine and narcolepsy persists in a multiple-bias sensitivity analysis in the Finnish study but not in the French study. We advocate the use of multiple-bias analyses to better understand the robustness of study findings, and to increase accuracy of data used to inform subsequent benefit-risk decision.

6. Confounding by Drug Formulary Restrictions in Pharmacoepidemiologic Research

Kristian B. Filion, Maria Eberg, Pierre Ernst. Center for Clinical Epidemiology, Lady Davis Institute, Jewish General Hospital, McGill University, Montreal, QC, Canada.

Background: The potential consequences of confounding owing to drug formulary restrictions in pharmacoepidemiologic research remain incompletely understood.

Objectives: The aim of this study was to illustrate this potential bias using the example of fluticasone/salmeterol combination therapy (Advair©), an oral inhaler used for the treatment of asthma and chronic obstructive pulmonary disease, whose use is restricted in the province of Quebec, Canada.

Methods: We identified all new users of fluticasone/salmeterol in Quebec's administrative health databases and classified those who received their initial dispensation of fluticasone/salmeterol between 1 September 1999 and 30 September 2003 as users from the liberal use period and those who received their initial dispensation between 1 January 2004 and 31 October 2006 as users from the restricted period. The primary outcome was time to first hospitalization for respiratory causes within 12 months of cohort entry.

Results: Our cohort included 77 212 new users of fluticasone/salmeterol, 72 154 from the liberal period and 5058 from the restricted period. Compared with the liberal period (crude rate per 100 person-years = 18.7, 95% confidence interval [CI] = 18.3, 18.9), the restricted period (crude rate = 26.2 events per 100 person-years, 95%CI = 24.7, 27.9) was associated with an increased rate of hospitalization for respiratory causes (crude hazards ratio [HR] = 1.41, 95% = 1.32, 1.51). Subsequent adjustment for age, sex, and hospitalization for respiratory causes in the previous year attenuated the association (HR = 1.05, 95%CI = 0.98, 1.12). Further adjustment for comorbidities, respiratory and non-respiratory medications, prescribing physician specialty, and season resulted in a significantly lower rate during the restricted period (HR = 0.78, 95%CI = 0.73, 0.83).

Conclusions: Drug formulary restrictions can result in substantial and unexpected confounding and should be considered during the design and analysis of pharmacoepidemiologic studies.

7. Perioperative Beta-Blocker (BB) Therapy in Non-cardiac Surgery and the Risk of In-hospital Myocardial Infarction (MI), Stroke, and Mortality

Elisabetta Patorno,1 Robert Glynn,1 Sebastian Schneeweiss,1 Jun Liu1, and Brian Bateman.1,21Division of Pharmacoepidemiology and Pharmacoeconomics, Department of Medicine, Brigham and Women's Hospital and Harvard Medical School, Boston, MA, USA;2Department of Anesthesiology, Critical Care, and Pain Medicine, Massachusetts General Hospital, Boston, MA, USA.

Background: Based on two small randomized controlled trials (RCTs) from the 1990s, beta-blockers (BBs) were promoted to prevent perioperative cardiac events in non-cardiac surgery. In 2008, a large RCT (POISE trial) showed decreased myocardial infarction (MI) risk associated with perioperative BB initiation, but an increased risk of stroke and mortality. Subsequent observational studies have produced mixed results.

Objectives: The aim of this study was to evaluate the risk of in-hospital MI, stroke, and mortality associated with perioperative BB initiation in patients undergoing elective non-cardiac surgery.

Methods: From a large US commercial healthcare database, we identified patients ≥18 years who underwent moderate- to high-risk elective non-cardiac surgery between 2004 and 2013 (index surgery), had a surgical visit within a 60-day window prior to surgery (to ensure potential access to treatment), and had no BB use in the 180 days prior to that window. Among those, we identified BB initiators and non-initiators within 60 days of surgery. A 1:1 propensity score (PS) matching was used to balance over 90 baseline characteristics. Relative risks (RR) and 95% confidence intervals (CIs) for MI, stroke (either ischemic or hemorrhagic), and mortality events occurring during the index surgery hospitalization were calculated using logistic regression models.

Results: Of 245 257 patients undergoing surgery, 7548 initiated a BB (3.1%). The risk of in-hospital MI, stroke, and mortality was 0.30%, 0.46%, and 0.23% among BB initiators, and 0.11%, 0.11%, and 0.12% in non-initiators. Among 14 884 1:1 PS-matched patients, there were 21 (0.28%) MIs among BB initiators versus 25 (0.34%) among non-initiators (RR = 0.84, 95%CI = 0.47–1.50), along with 32 (0.43%) versus 18 (0.24%) strokes (RR = 1.78, 1.00–3.18), and 17 (0.23%) versus 12 (0.16%) deaths (RR = 1.42, 0.68–2.97). Sensitivity analyses based on 1:3 fixed ratio PS-matching yielded consistent results.

Conclusions: These preliminary results suggest perioperative BB initiation may be associated with an increased risk of stroke in patients undergoing non-cardiac surgery in routine care, mirroring what was observed in the POISE trial. Further research on how such an effect may vary across individual BB agents and dose is warranted.

8. Evidence of Potential Unmeasured Confounding in a Comparison of Cardioselective β-Blockers and Non-dihydropyridine Calcium Channel Blockers in Patients with Acute Coronary Syndromes and Chronic Obstructive Pulmonary Disease: Results of a Multi-national Study

Yaa-Hui Dong,1 Matthew Alcusky,2 Vittorio Maio,2 Jun Liu,1 Mengdan Liu,2 Li-Chiu Wu,3 Chia-Hsuin Chang,3,4 Mei-Shu Lai,3,4 Joshua J. Gagne.11Division of Pharmacoepidemiology and Pharmacoeconomics, Department of Medicine, Brigham and Women's Hospital and Harvard Medical School, Boston, MA, USA;2Jefferson School of Population Health, Thomas Jefferson University, Philadelphia, PA, USA;3Department of Internal Medicine, National Taiwan University Hospital, Taipei, Taiwan;4Graduate Institute of Epidemiology and Preventive Medicine, College of Public Health, National Taiwan University, Taipei, Taiwan.

Background: Despite reluctance to prescribe β-blockers (BBs) to chronic obstructive pulmonary disease (COPD) patients owing to concerns about acute bronchospasm, some observational studies suggest that BBs reduce mortality in COPD patients. However, these studies have important limitations, such as focusing on prevalent users and a lack of active comparator groups.

Objectives: The aims of this study were to examine the effectiveness of BBs in patients with acute coronary syndromes (ACS) and COPD with a propensity score (PS)-matched, active comparator, new-user cohort design and to assess potential unmeasured confounding by examining a short-term COPD hospitalization outcome.

Methods: We identified patients with ACS and COPD starting a cardioselective BB or a non-dihydropyridine calcium channel blocker in three US claims databases, the Italy Regione Emilia-Romagna Database, and the Taiwan National Health Insurance Database. Outcomes were mortality, CV hospitalizations, and COPD hospitalizations. We used stratified Cox regression models to estimate HRs and 95%CIs for each outcome and in each database after variable-ratio matching on site-specific PSs. We used random-effects meta-analyses to combine HRs across databases.

Results: A total of 16 995 patients were eligible, of whom 4591 died, 3045 had CV hospitalizations, and 868 had COPD hospitalizations. Cardioselective BBs were not associated with reduced risk of mortality (HR 0.90, 95%CI 0.75–1.09) or CV hospitalizations (HR 1.01, 95%CI 0.85–1.21) but were inversely associated with COPD hospitalizations (HR 0.52, 95%CI 0.45–0.61). The HR for COPD hospitalizations restricted to the first 30 days of follow-up was 0.51 (95%CI 0.32–0.82).

Conclusions: This multi-national study found a strong inverse association between cardioselective BBs use and COPD hospitalizations, even in the first 30 days of follow-up, suggestive of severe confounding. The persistence of this bias and the potential for it to affect the mortality and CV hospitalization outcomes call into question the results of prior studies.

9. Outcomes of Carotid Artery Stenting (CAS) in Older Patients with Carotid Stenosis

Jessica J. Jalbert,1,2 Louis L. Nguyen,3 Marie D. Gerhard-Herman,4 Michael R. Jaff,5 Christopher J. White,6 Andrew T. Rothman,1 John D. Seeger,1 Hiraku Kumamaru,1 Lauren A. Williams,1 Chih-Ying Chen,1 Jun Liu,1 Herbert D. Aronow,7 Joseph A. Johnston,8 Thomas G. Brott,9 and Soko Setoguchi.101Division of Pharmacoepidemiology and Pharmacoeconomics, Brigham and Women's Hospital/Harvard Medical School, Boston, MA, USA;2LA-SER Analytica, New York, NY, USA;3Division of Vascular and Endovascular Surgery, Brigham and Women's Hospital, Boston, MA, USA;4Division of Cardiovascular Medicine, Brigham and Women's Hospital, Boston, MA, USA;5Division of Cardiovascular Medicine, Massachusetts General Hospital, Boston, MA, USA;6Ochsner Clinic Foundation, New Orleans, LA, USA;7Cardiac Catheterization Laboratory, St. Joseph Mercy Hospital, Ypsilanti, MI, USA;8Eli Lilly and Company, Indianapolis, IN, USA;9Department of Neurology, Mayo Clinic, Jacksonville, FL, USA;10Duke Clinical Research Institute, Durham, NC, USA.

Background: Outcomes during and after the peri-procedural period (PPP) among Medicare beneficiaries undergoing carotid artery stenting (CAS) have not been described, despite increasing CAS dissemination.

Objectives: The aims of this study were to describe outcomes during and after the PPP among Medicare patients undergoing CAS and to evaluate how patient and provider characteristics affect outcomes.

Methods: We linked 2000–2009 Medicare data to the Carotid Artery Stenting Database (2005–2009), which contain basic clinical information on patients undergoing CAS in Medicaid- and Medicare-certified facilities. Medicare patients aged ≥66 years were followed up from the CAS date for death and stroke/transient ischemic attack (TIA) during and after the PPP (30 days post-procedure). Using Kaplan–Meier estimators, we derived risks overall and among patient subgroups defined by demographics, clinical, and center-level factors. To evaluate how outcome risks differed across subgroups, we used Cox regression while accounting for clustering.

Results: Among 22 516 CAS Medicare patients, mean age was 76.3 years, 60.5% were male, 93.8% were White, 91.2% were high-surgical risk, and 47.4% were symptomatic. Crude 30-day mortality and stroke/TIA risks were 1.7% (95% confidence interval (CI): 1.5–1.8%) and 3.3% (95%CI: 3.0–3.5%), respectively. Age ≥80 years, symptomatic carotid stenosis, and non-elective hospitalizations were associated with increased adjusted hazards of mortality and stroke/TIA during and after the PPP. Presence of a stroke center, government ownership, and ≥500 beds were associated with increased adjusted hazards of peri-procedural mortality and stroke/TIA. Mortality over a mean follow-up time of 2 years was 32.0% (95%CI: 31.0–33.0%) and differed by subgroup, 27.7% (95%CI: 26.4–28.9%) for asymptomatic patients, 37.3% (95%CI: 35.8–38.7) for symptomatic patients, and 41.5% (95%CI: 39.7–43.3) for patients ≥80 years of age.

Conclusions: While peri-procedural risks for CAS approximated those recommended in clinical guidelines, mortality was high, possibly negating expected benefits of CAS. Future research should identify factors predicting which Medicare patients have a hazard of mortality low enough to justify CAS, particularly in patients who are symptomatic and ≥80 years old.

10. Effects of Clopidogrel and Proton Pump Inhibitors on Cardiovascular Events in Patients with Type 2 Diabetes Mellitus After Drug-Eluting Stent Implantation: A Nationwide Cohort Study

Chi-Feng Hsieh,1 Weng-Foung Huang,1 Yi-Ting  Chiang,1 and Chun-Yen Chen.21Institute of Health and Welfare Policy, National Yang-Ming University, Taipei, Taiwan;2Mackay Medical College, New Taipei City, Taiwan.

Objectives: The aim of this study was to investigate whether there is an increased risk of cardiac events in diabetic patients with a combined therapy of clopidogrel (CLO) and proton pump inhibitors (PPIs) after drug-eluting stent (DES) deployment.

Methods: By using National Health Insurance Research Database, all patients undergoing DES (limus-eluting stents [LESs] and paclitaxel-eluting stents [PESs]) deployment who received 90 days of CLO with or without PPI therapy were enrolled. Endpoints were acute coronary syndrome (ACS) and readmission for revascularization (percutaneous coronary intervention or coronary artery bypass graft surgery) after 3, 6, and 12 months.

Results: A total of 6603 diabetic patients received LESs (5933 in the CLO subgroup and 670 in the CLO plus PPIs subgroup), and 3202 patients received PESs (2923 in the CLO subgroup and 279 in the CLO plus PPIs subgroup). The patients who received CLO plus PPIs were at higher risk of myocardial infarction (MI) than those receiving CLO within 1 year after DES deployment (LESs: 6-month hazard ratio [HR] = 1.63, 95% confidence interval [CI], 1.25–2.14 and 1-year HR = 1.37, 95%CI, 1.09–1.71; PESs: 3-month HR = 1.72, 95%CI, 1.02–2.89). Patients with a history of MI who received CLO plus PPIs were at higher risk of ACS after LES implantation (HR = 1.55; 95%CI, 1.11–2.16) than those in the CLO group. Similar results were not found for PESs.

Conclusions: In “real-world” diabetic patients with LES deployment, the combination of PPIs and CLO is associated with higher rates of ACS after 6 months and 1 year. Even after correction for confounding factors, concomitant PPI use remained an independent predictor of cardiac events, emphasizing the clinical importance of this drug–drug interaction.

11. Cardiovascular Safety and Effectiveness of Glucagon-like Peptide-1 Receptor Agonists (GLP-1-RAs) in Routine Care

Elisabetta Patorno,1 Brendan Everett,2 Allison Goldfine,3 Robert Glynn,1 Jun Liu,1 Seoyoung Kim.1,41Division of Pharmacoepidemiology and Pharmacoeconomics, Department of Medicine, Brigham and Women's Hospital and Harvard Medical School, Boston, MA, USA;2Division of Cardiovascular Medicine, Brigham and Women's Hospital, Boston, MA, USA;3Clinical Research, Joslin Diabetes Center, Boston, MA, USA;4Division of Rheumatology, Allergy and Immunology, Brigham and Women's Hospital, Boston, MA, USA.

Background: Evidence from small, randomized trials suggests that glucagon-like peptide-1 receptor agonists (GLP-1-RAs) may reduce cardiovascular risk in patients with type 2 diabetes mellitus (T2DM). The few observational studies available have produced mixed results.

Objectives: The aim of this study was to evaluate whether GLP-1-RAs, when added to metformin, reduce the risk of major adverse cardiovascular events (MACE) compared with the addition of other antidiabetic agents to metformin in routine care.

Methods: Within a large commercial US health insurance database, we identified T2DM patients who intensified metformin with new use of GLP-1-RAs, second-generation sulfonylureas (SUs), DPP-4 inhibitors (DPP-4is), or insulin between 2005 and 2013. Propensity score (PS) matched analyses (1:1 ratio) were used to balance more than 100 baseline characteristics and evaluate the risk of MACE (i.e., a hospitalization for acute coronary syndrome, stroke, or cardiac revascularization) in three separate PS-matched cohorts (GLP-1-RAs vs. SUs, vs. DPP-4is, and vs. insulin). Follow-up started on the day following intensification and ended at a MACE event, insurance disenrollment, or the end of a 365-day period.

Results: After PS matching, we identified 28 182 new users for GLP-1-RAs versus SUs with 138 and 151 MACE events, respectively (incidence rates = 12.2/1000 person years (py) and 13.6/1000 py); 35 526 new users for GLP-1-RAs versus DPP-4is with 209 and 194 events (14.6/1000 py and 13.5/1000 py); and 47 458 new users for GLP-1-RAs versus insulin with 306 and 409 events (16.0/1000 py and 22.2/1000 py). The hazard ratio (95% confidence interval) for MACE was 0.90 (0.71–1.13) for initiators of GLP-1-RAs versus SUs, 1.08 (0.89–1.31) versus DPP-4is, and 0.72 (0.62–0.84) versus insulin. The decreased risk versus insulin disappeared in a sensitivity analysis restricted to patients with baseline HbA1c level available and including this value in the PS model (HR = 0.95, 0.69–1.32).

Conclusions: In this large nationwide investigation, the addition of GLP-1-RAs to metformin was not associated with a reduced risk of MACE when compared with the addition of other antidiabetic agents. More research is needed to define whether specific subgroups of patients with T2DM may benefit from the use of GLP-1-RAs.

12. The Risk of Developing Ménière's Disease in Association with Use of Antiarrhythmic Drugs

Nadja A. Stohler,1,2 Saskia G. Bruderer,1,2 Daniel Bodmer,3 Susan Jick,4 and Christoph R. Meier.1,2,41Department of Pharmaceutical Sciences, Basel Pharmacoepidemiology Unit, Division of clinical Pharmacy and Epidemiology, University of Basel, Basel, Switzerland;2Hospital Pharmacy, University Hospital Basel, Basel, Switzerland;3Department of Biomedicine and Clinic for Otorhinolaryngology, University Hospital Basel, Basel, Switzerland;4Boston Collaborative Drug Surveillance Program, Boston University, School of Medicine, Lexington, MA, USA.

Background: Ménière's disease (MD) is a disease of the inner ear with a poorly understood pathophysiology. The inner ear is compartmentalized and filled with two different fluids, which are separated by membrane barriers: potassium-rich and sodium-poor endolymph on one side, and sodium-poor and potassium-rich perilymph on the other side. Amongst others, K+ and Na+ channels help maintain the ion gradient between the two liquids. Class I and III antiarrhythmic drugs act via Na+ or K+ channels, respectively, and may therefore influence these gradients.

Objectives: The aim of this study was to analyse the association between antiarrhythmics (class I and III) interacting with Na+ and K+ channels and the risk of developing MD.

Methods: We performed a case–control analysis using data from the UK-based Clinical Practice Research Datalink. Between 1993 and 2013, we identified case patients between 18 and 79 years with an incident MD diagnosis and matched four MD-free control subjects to each case. The controls were selected according to the criteria of identical age, sex, number of years of active history in the database prior the index date, and general practice. We stratified exposure by timing of the last prescription prior to the index date (current <180 days, past ≥180 days), and by duration of use. We calculated odds ratios (OR) with 95% confidence intervals (CIs) using conditional logistic regression and adjusting for potential confounders.

Results: Of 7777 cases with incident MD, 66% were female with an overall mean age of 56.7 years (±SD 13.7) at first MD diagnosis. Among these cases and 31 108 matched controls, current use of Na+ channel blocking antiarrhythmics was associated with an OR of 1.92 (95%CI 1.10–3.36), while no association was found between current use of K+ channel blockers and the risk of developing incident MD (OR 1.12, 95%CI 0.81–1.55). These analyses were adjusted for body mass index, smoking, and alcohol consumption.

Conclusions: Our findings suggest that patients who use class I antiarrhythmics acting on Na+ channels, but not class III antiarrhythmics acting on K+ channels, may be at a slightly increased risk of developing MD.

13. Patterns of Antihyperglycemic Drug Use and Effect on Implementation of a New-User Active Comparator Study

Mugdha N. Gokhale, John B. Buse, M. Alison Marquis, and Til Sturmer. University of North Carolina at Chapel Hill, Chapel Hill, NC, USA.

Background: New-user active comparator studies often use data combined from multiple years, which could be an issue if there are changes in treatment initiation patterns during the study period. In recent years, second-line treatment of type 2 diabetes with dipeptidyl peptidase-4 inhibitors (DPP) increased while thiazolidinediones (TZD) decreased.

Objectives: We examined the patterns of initiation of antihyperglycemic drugs and their effect on the implementation of a new-user study comparing DPP versus TZD using combined data from all years.

Methods: Using Medicare claims from 2008 to 2012, we identified initiators of second-line treatment with DPP, TZD, or sulfonylureas (SU) (100%) after a 6-month washout and classified them based on year of initiation. We examined whether the prevalence of baseline variables within DPP and TZD initiators changed based on year of initiation. Second, we assessed whether the determinants of initiating DPP versus TZD changed over time. This was carried out by comparing the odds ratio (OR) and 95%CI for DPP versus TZD for all baseline variables for each year.

Results: During the study period, the proportion of DPP initiators increased from 11.7% (n = 2637) to 30.1% (n = 5024), while TZD initiators decreased from 19.7% (n = 3989) to 2.6% (n = 429) with a clear crossing of curves around 2010. The proportion of SU initiators remained constant around 70%. There were negligible differences in the baseline characteristics within initiators of DPP and TZD across years (<5%). The odds ratios for DPP versus TZD changed very slightly across years even for key variables like baseline diabetes complications (OR 1.2 (1.0–1.3) in 2008; 1.3 (1.1–1.6) in 2012) and metformin use (OR 1.3 (1.2–1.4) in 2008; 1.5 (1.2–1.8) in 2012).

Conclusions: Over the study period, the initiation of DPP increased while TZD initiation decreased. No changes in the baseline characteristics of DPP and TZD initiators combined evince that the indication did not change over time. The determinants to prescribe DPP versus TZD for each year did not change, implying that the market dynamics have little implications for confounding control while comparing DPP versus TZD with a new-user design using combined data from all years.

14. Metformin Use in Patients with Renal Impairment: A Drug Utilization Study in Denmark and the United Kingdom

Christian F. Christiansen,1 Vera Ehrenstein,1 Stine Skovbo,1 Helene Nørrelund,1 Henrik T. Sørensen,1 Lin Li,2 and Susan Jick.21Department of Clinical Epidemiology, Aarhus University Hospital, Aarhus N, Denmark;2Boston Collaborative Drug Surveillance Program, Boston University School of Public Health, Lexington, MA, USA.

Background: According to current recommendations, metformin is contraindicated in patients with severe renal impairment and should be used with caution in patients with mild to moderate renal impairment.

Objectives: The aims of this study were to estimate prevalence of renal impairment in metformin users and to examine utilization of metformin among diabetes patients with renal impairment.

Methods: We conducted this two-country drug utilization study using the Aarhus University Prescription Database, the Laboratory Information System, and the Danish National Patient Registry including data from Northern Denmark and the Clinical Practice Research Datalink from the UK for years 2000 to 2011. We included patients aged ≥30 years with medically treated diabetes. Using cross-sectional analysis, we described patient demographics, comorbidities, and co-medications according to metformin use and renal function, using estimated glomerular filtration rates (eGFR). We also examined changes in metformin use within 90 days after first decline in eGFR after study start.

Results: We included 172 052 diabetes patients in Denmark and the UK. Users of metformin were overall younger and had a lower prevalence of comorbidities and metformin contraindications including renal impairment than users of other antidiabetic drugs. Prevalence of eGFR <60 ml/min/1.73 m2 among new metformin users was 11.0% in Denmark and 25.2% in the UK. In contrast, eGFR values <45 ml/min/1.73 m2 were less prevalent (2.7% of new metformin users in Denmark and 4.9% in the UK). Most metformin users continued taking the medication after the first decline in eGFR. No more than 45% of patients discontinued metformin even after an eGFR drop below 30 ml/min/1.73 m2. There was no clinically significant dose reduction with decreasing eGFR level discernible from the data.

Conclusions: Mild to moderate renal impairment was common among metformin users, most of whom continued metformin after developing severe renal impairment—against current recommendations.

15. Channeling of Linagliptin to Patients with Renal or Hepatic Impairment

Chandrasekar Gopalakrishnan,1 Elisabetta Patorno,1 Olesya I. Zorina,2 Sebastian Schneeweiss,1 Dorothee B. Bartels,2 Jun Liu,1 and John D. Seeger.11Division of Pharmacoepidemiology and Pharmacoeconomics, Department of Medicine, Brigham and Women's Hospital, Harvard Medical School, Boston, MA, USA;2Corporate Department of Global Epidemiology, Boehringer Ingelheim GmbH, Ingelheim, Germany.

Background: Prescribing of medications may be selective with respect to factors that could influence effectiveness or tolerability. Linagliptin (LINA) is a dipeptidyl peptidase-4 inhibitor (DPP-4i) recently approved in the USA for treatment of type 2 diabetes mellitus (T2DM). LINA is the only diabetes medication whose labeling does not require dose adjustment in patients with renal or hepatic impairment, and this could translate into preferential prescribing of LINA to such patients.

Objectives: We sought to assess the extent of channeling of LINA and whether this channeling matches expectations arising from the labeling.

Methods: Within a large, nationwide US health insurance database (Optum Clinformatics), T2DM patients who initiated LINA or other non-insulin glucose lowering agents between May 2011 and June 2012 were described in terms of medical characteristics at time of initiation.

Results: Of 155 345 T2DM patients who initiated a non-insulin diabetes agent, 2820 (1.8%) did so with LINA. The prevalence of baseline kidney disease among patients initiating LINA (11.9%) was higher than initiators of other DPP-4i (6.5%), sulfonylureas (6.9%), glitazones (6.8%) or metformin (3.9%), with only meglitinide initiators (14.2%) having a higher prevalence. Although rare, patients initiating LINA tended to have a slightly higher prevalence of baseline chronic liver disease (3.0%) than other DPP-4i (2.6%), metformin (2.5%), sulfonylurea (2.9%) and glitazones (2.6%). These conditions and further comorbidities translated into higher average Charlson comorbidity scores among LINA (1.68) initiators compared with other DPP-4i (1.43), metformin (1.27), sulfonylureas (1.44) and glitazones (1.37), with only meglitinide initiators having a higher score (1.79) than LINA. These patterns were also observed in another US insurance claims database (MarketScan).

Conclusions: Patients initiating LINA exhibit a higher burden of comorbidities, including renal and hepatic disease, which reflect labeling and need to be accounted for in comparative effectiveness or safety studies of LINA.

16. Effectiveness of Intensification Therapy in Patients with Type 2 Diabetes Who Used Basal Insulin Only

Reimar W. Thomsen,1 Lisbeth M. Baggesen,1 Mette Søgaard,1 Lars Pedersen,1 Helene Nørrelund,1 Esben S. Buhl,2 Christiane L. Haase,2 and Søren P. Johnsen.11Department of Clinical Epidemiology, Aarhus University Hospital, Aarhus, Denmark;2Novo Nordisk Scandinavia AB, Copenhagen, Denmark.

Background: Limited data exist on the glycemic benefit of different intensification therapies in type 2 diabetes (T2D) patients who use basal insulin only in a routine clinical setting.

Objectives: The aim of this study was to examine the effect of different intensification therapies on glycemic control among T2D patients who received basal insulin only therapy.

Methods: We identified a cohort of all individuals with a first basal insulin only prescription in Northern Denmark, 2000–2012, and identified all add-on intensification therapy with bolus insulin, premixed insulin, or GLP-1 receptor agonists. We used Poisson regression to compute adjusted relative risks (aRRs) of reaching an HbA1c target value of <7.0% adjusted for age, gender, comorbidities, and baseline HbA1c.

Results: We included 6114 patients with a first basal insulin prescription and HbA1c measurements. Of these 2156 (35.3%) who received intensification therapy after a median of 11 months, 59.2% (n = 1276) were intensified with premixed insulin, 27.5% (n = 592) with bolus insulin, and 11.8% (n = 255) with GLP-1 agonists, and only 1.5% (n = 33) received more than one add-on regimen simultaneously. Overall, 25.9% attained an HbA1c target of <7% within 3–6 months after intensification. Reductions in median HbA1c were 0.9 percentage points (pp) for premixed insulin, 0.4 pp for bolus insulin, 0.9 pp for GLP-1 agonists, and 1.1 pp for patients with >1 intensification drug. Compared with the large group of premixed insulin intensification as reference, aRRs of attaining an HbA1c <7% were 1.01 (95%CI 0.84–1.21) for bolus insulin and 1.28 (95%CI 1.02–1.60) for GLP-1 agonists.

Conclusions: In this population-based study, one-fourth of T2D patients reached a target HbA1c <7% within 3–6 months after intensification of their basal insulin only therapy. GLP-1 agonists were associated with higher target attainment than premixed insulin intensification.

17. Statin Initiation in the Context of Diabetes Risk

Robert J. Glynn, and Paul M. Ridker. Department of Medicine, Brigham and Women's Hospital, Boston, MA, USA.

Background: Statins reduce cardiovascular risk in many types of patients but can also increase risk of being diagnosed with diabetes. Simple prediction models for incident type 2 diabetes have excellent performance characteristics, but it is unclear whether they can identify patients at particular risk at the time of statin initiation.

Objectives: We evaluated the predictive performance of a Framingham risk score for diabetes (PW Wilson et al.; Arch Intern Med 2007) in a multi-cultural statin trial population and tested whether risk of diabetes on statin therapy was modified across levels of this score.

Methods: The double-blind JUPITER trial randomized 17 603 men and women without prior cardiovascular disease or diabetes to daily rosuvastatin 20 mg or placebo and followed them up for up to 5 years for the primary endpoint of a first confirmed cardiovascular event and the secondary endpoint of physician-reported diabetes.

Results: The externally derived diabetes risk score performed well: separately in those randomized to rosuvastatin and placebo, the score had excellent discrimination (C-statistic ≥ 0.83) and acceptable calibration (Hosmer–Lemeshow p > 0.05) with rates of newly diagnosed diabetes ranging from below 0.2/100 person-years (p-yrs) among those in the bottom score quartile to above 3.0/100 p-yrs in the top quartile. Based on 406 cases of newly diagnosed diabetes during follow-up, those assigned to active rosuvastatin had a 29% increased rate relative to placebo, adjusted for predicted risk (relative risk 1.29; 95% confidence interval (CI): 1.06–1.57) with no evidence for heterogeneity across risk categories on this scale (homogeneity p = 0.67). However, on the rate difference scale, heterogeneity was apparent (homogeneity p = 0.033) with no meaningfully higher rate of newly diagnosed diabetes in those assigned to active rosuvastatin if they had a below median baseline risk score (risk difference (RD) 0.04; 95%CI: −0.09 to 0.17 cases/100 p-yrs) but a significantly increased rate with rosuvastatin in those with above-median baseline risk (RD 0.52; 95%CI: 0.10 to 0.94 cases/100 p-yrs).

Conclusions: A clinical diabetes risk score can partition patients at the time of statin initiation into some with low risk and others with higher risk of newly diagnosed diabetes during treatment.

18. Impact of Adherence to Oral Antidiabetics on All-Cause Mortality: A Population-Based Study

Patrice Simard,1 Nancy Presse,1 Louise Roy,2 Marc Dorais,3 Brian White-Guay,1 Agnès Rakel,2 and Sylvie Perreault.11Faculté de Pharmacie, Université de Montréal, Montreal, QC, Canada;2Centre Hospitalier de l'Université de Montréal, Hôpital Saint-Luc, Montreal, QC, Canada;3StatSciences Inc, Notre-Dame de l'Ile-Perrot, QC, Canada.

Background: Oral antidiabetics (OAD) have been shown to reduce the risk of mortality, particularly among metformin users, which had risk reductions of 36% for cardiovascular-related death and 33–40% for all-cause mortality. However, very few studies have assessed adherence to OAD and all-cause mortality.

Objectives: A population-based nested case–control study design was used to investigate the relationship between adherence to OAD and all-cause mortality among incident users of OAD.

Methods: Incident OAD users were identified using healthcare databases of residents covered by the public drug insurance plan of the Province of Quebec, Canada. Patients initiated OAD therapy between 2000 and 2009 and were aged 45–85 years at cohort entry. A nested case–control design was conducted to study mortality occurrence. Each case was matched to 10 controls by gender, age and duration of follow-up. The adherence to OAD was measured by calculating the medication possession ratio. Conditional logistic regression models were used to estimate the association between adherence to OAD and all-cause mortality adjusting for various potential confounders.

Results: The cohort included 63 859 incident OAD users at entry: mean age was 68 years, 45% were male, 37% had coronary artery disease, 82% had hypertension, and 62% had dyslipidemia. Most patients initiated their OAD treatment with biguanides (78%) and sulfonylureas (12%). The average follow-up time was 48 months. Among those deemed adherent, the risk of mortality was decreased compared with non-adherent (rate ratio: 0.67 [95%CI 0.64–0.70]). The likelihood for mortality was higher for patients with heart failure (1.56 [1.49–1.65]), ≥2 cardiovascular diseases (1.45 [1.39–1.52]), amputations (2.03 [1.42–2.91]), chronic viral infections (1.73 [1.44–2.07]), corticosteroid use (1.69 [1.56–1.89]), and ≥1 hospital admissions (1.73 [1.65–1.80]). Conversely, mortality was least likely for patients with dyslipidemia (0.76 [0.73–0.79]) and hypertension (0.88 [0.83–0.94]).

Conclusions: Adherence to OAD seems to be associated with a risk reduction of mortality. Further research is needed to confirm this risk. Residual confounding may remain a potential issue.

19. Impact of Assisted Reproductive Therapy (ART) on Infant Health and Mortality

Katherine Jane Sullivan, Allison Schroeder, David Tabano, and Neel Vaidya. Center for Pharmaceutical Outcomes Research, University of Colorado Anschutz Medical Campus, Skaggs School of Pharmacy, Aurora, CO, USA.

Background: Use of assisted reproductive therapy (ART) for conception nearly doubled from 1999 to 2008 and continues to increase. Literature evaluating outcomes from ART including birth weight and overall newborn survival is limited; although in-depth information on ART as a whole is available, there is no Colorado-specific data on neonatal intensive care unit (NICU) admission rates or mortality following ART.

Objectives: The aims of this study were (i) to quantify NICU admission risk to infants conceived via ART compared with non-ART and (ii) to compare mortality rates within the first year of life between ART NICU infants and non-ART NICU infants.

Methods: Using data from the Colorado Department of Public Health and Environment, Colorado Birth Certificate Database from 2007 to 2012, a retrospective cohort study using multivariable logistic regression was performed. Rates of NICU admission in ART and non-ART infants were determined, then 1-year mortality rates in ART NICU infants and non-ART NICU infants were compared. Separate analyses were performed on singleton-only births. Non-hospital delivery, delivery prior to 22 weeks' gestation, discharge and readmission to NICU, and missing data were excluded. All mothers were between the ages of 25 and 45 years, were non-smokers, and had a live birth.

Results: A total of 190 795 live births in 2007–2012 were included into the birth cohort for analysis (12 666 ART births; 178 129 non-ART births). ART births had a 52% increased risk of being admitted to the NICU compared with non-ART births (OR 1.52 [95%CI 1.38, 1.69]); singleton-only ART births had a 39% greater risk of being admitted to the NICU compared with singleton non-ART births (OR 1.39 [95%CI 1.18, 1.65]). The risk of mortality within 1 year after birth among the ART NICU admissions is 46.7% lower compared with non-ART NICU admissions; however, it was not a statistically significant difference (OR 0.533 [95%CI 0.283, 1.002]).

Conclusions: ART births in Colorado have a higher risk of NICU admission compared with non-ART births. Although ART increases NICU admission risk, mortality risk within 1 year of birth among ART births admitted to the NICU was not statistically different from non-ART births.

20. Safety of Seasonal Influenza Vaccines in Pregnancy: VAMPSS Update

Christina D. Chambers,1,2 Carol Louik,3 Kenneth L. Jones,1 Allen A. Mitchell,3 and Michael Schatz.4,51Department of Pediatrics, University of California San Diego, La Jolla, CA, USA;2Department of Family Medicine and Public Health, University of California San Diego, La Jolla, CA, USA;3Slone Epidemiology Center, Boston University, Boston, MA, USA;4Department of Allergy and Immunology, Kaiser Permanente, San Diego, CA, USA;5American Academy of Allergy, Asthma and Immunology, Milwaukee, WI, USA.

Background: Annual variations in seasonal influenza vaccines require that their pregnancy safety be monitored.

Objectives: The aim of this study was to evaluate the fetal safety of influenza vaccines available in the USA in years following the pandemic H1N1 season.

Methods: The Vaccines and Medications in Pregnancy Surveillance System (VAMPSS) continued to monitor risks for birth defects and other adverse pregnancy outcomes in women who received seasonal influenza vaccine in pregnancy. The cohort arm of VAMPSS conducted by the OTIS Collaborative Research Group prospectively followed pregnant women in the USA or Canada who did or did not receive seasonal influenza vaccine in any trimester. In parallel, the case–control arm of VAMPSS conducted by the Slone Epidemiology Center Birth Defects Study interviewed mothers of malformed cases and controls from five US regions about influenza vaccination during pregnancy.

Results: The VAMPSS cohort arm completed follow-up of 1205 vaccinated and 448 unvaccinated women between 2010 and 2014. The relative risk (RR) for major birth defects overall in first-trimester vaccine-exposed compared with unexposed approximated 1.0 (RR 1.23, 95% confidence interval (CI) 0.60, 2.50); no increases were noted within seasons. Rates of spontaneous abortion, preterm delivery, and small for gestational age infants were also similar between groups. The VAMPSS case–control arm completed maternal interviews for 4118 cases with specific malformations and 2084 non-malformed controls in the 2011/2012 or 2012/2013 seasons. For 41 specific defects evaluated in either season or both seasons combined, adjusted odds ratios approximated 1.0; only one lower bound of the 95%CI exceeded 1.0. Twenty eight of the 41 specific defects met the VAMPSS criteria of no evidence of risk (upper bound 95%CI < 4.0).

Conclusions: Continued surveillance of each seasonal formulation of influenza vaccine received by pregnant women (regardless of trimester) provides reassurance that these vaccines were not associated with increased risks for adverse pregnancy outcomes.

21. Otic Quinolones After Tympanostomy Tubes Associated with Persistent Tympanic Membrane Perforations

Adel A. Alrwisan,1 Almut G. Winterstein,2 and Patrick J. Antonelli.31Pharmaceutical Outcomes and Policy, University of Florida, Gainesville, FL, USA;2Pharmaceutical Outcomes and Policy, University of Florida, Gainesville, FL, USA;3Department of Otolaryngology, University of Florida, Gainesville, FL, USA.

Background: Systemic quinolones have been linked to collagen disorders (e.g., tendon rupture). Otic quinolone administration exposes the tympanic membrane to higher concentrations, but no study has examined the effect on tympanic membrane perforation (TMPs).

Objectives: We thought to investigate whether otic quinolones with or without corticosteroids carry an additional risk of TMP compared with otic neomycin plus hydrocortisone (HC) preparations.

Methods: This retrospective cohort study utilized administrative claims data of children eligible for Medicaid services in 29 states between 1999 and 2006. Children entered the cohort after placement of tympanostomy tubes (TTs) and dispensing of quinolone or neomycin + HC eardrops. Children were followed up 24 months after their first eardrop prescription. We defined persistent TMP as the need for tympanoplasty. A Cox regression model adjusted for age, sex, race, adenoidectomy, and TT reinsertion was used to compare the rate of TMP between quinolone and neomycin + HC exposed children.

Results: A total of 96 595 children had a risk of 19 and 14 TMPs/10 000 patient-years for quinolones and neomycin + HC, respectively. Patients exposed to quinolones had a higher risk of TMP with an adjusted hazard ratio of 1.92 (95% confidence interval (CI) 1.31–2.8). Stratified by quinolone, the hazard ratios were 1.79 (95%CI 1.21–2.65) for ofloxacin, 2.17 (95%CI 1.39–3.37) for ciprofloxacin + HC, and 2.58 (95%CI 1.51–4.42) for ciprofloxacin + dexamethasone.

Conclusions: In children with TTs, exposure to otic quinolones is associated with increased risk of persistent TMPs. Eardrops with quinolones and corticosteroids may pose a greater risk of persistent TMPs. This risk must be balanced against the risk of hearing loss with otic neomycin use.

22. Immunization Status at Discharge from the Neonatal Intensive Care Unit (NICU)

Sudha R. Raman,1 Stephen D. DeMeo,2 Christoph P. Hornik,1 Reese Clark,3 and P. Brian Smith.11Duke Clinical Research Institute, Duke University, Durham, NC, USA;2Department of Pediatrics, Duke University Medical Center, Durham, NC, USA;3Pediatrix-Obstetrix Center for Research and Education, Sunrise, FL, USA.

Background: Underimmunization at discharge from the neonatal intensive care unit (NICU) can persist into early childhood, putting vulnerable infants at risk of vaccine preventable diseases. Few studies have assessed immunization status of infants at NICU discharge.

Objectives: The aim of this study was to assess the immunization status of infants >60 days of age at time of discharge from the NICU and assess for trends in vaccination status over time.

Methods: We conducted a multicenter retrospective cohort study using electronic medical record data from 322 US NICUs managed by the Pediatrix Medical Group. The cohort included infants discharged home between 2007 and 2012, who had been in the NICU continuously from day of life 0 to 60, excluding those diagnosed with any immunodeficiency. The infants were grouped by age at discharge (2 months: 60–119 days; 4 months: 120–179 days; 6 months: 180–365 days). We assessed whether infants were immunized with five vaccine types [diphtheria/tetanus/acellular pertussis (DTaP), inactivated polio virus (IPV), Haemophilus influenza B (HiB), pneumococcal conjugate vaccine 7-valent/13-valent, hepatitis B (HepB)]. The outcome was the proportion of infants who were up to date (UTD) at discharge, defined as at least one dose of each vaccine type for the 2-month group, two doses for the 4-month group, and three doses for the 6-month group [except HepB (two doses) and HiB (two doses if started with DTaP/IPV/HepB combination vaccine)].

Results: Of 24,599 infants in the cohort, 80% of the 2-month group (17 251/21 673), 39% (1010/2582) of the 4-month group, and 26% (78/304) of the 6-month group were UTD, with only 4% of all infants being completely unimmunized at discharge (n = 1101). Having UTD status was most common for HepB in the three groups (97%, 51%, and 79% in the 2-, 4-, and 6-month group, respectively). UTD status at discharge was higher in 2012 versus 2007 for all groups (2 months: 85% vs. 75%; 4 months: 45% vs. 32%; 6 months: 29% vs. 12%).

Conclusions: Many infants are underimmunized at NICU discharge, especially if discharged after 4 months. Up-to-date immunization status has improved over time. The influence of infant clinical characteristics on vaccination status should be examined.

23. Evaluation of a Composite Endpoint for Measuring Neonatal Benefit in Clinical Trials of Tocolytics Using Electronic Medical Records (EMRs)

Jeanne M. Pimenta,1 Emma Brinkley,2 Kathleen J. Beach,3 Timothy H. Montague,4 and Christina D. Mack.21Worldwide Epidemiology, GlaxoSmithKline, Uxbridge, Middlesex, UK;2Epidemiology, Real-World and Late Phase Research, Quintiles, Research Triangle Park, NC, USA;3Maternal Neonatal Health Unit, GlaxoSmithKline, Research Triangle Park, NC, USA;4Clinical Statistics, GlaxoSmithKline, Philadelphia, PA, USA.

Background: Neonatal benefit resulting from prolongation of pregnancy is a required endpoint in randomized controlled trials (RCTs) of new tocolytics. We previously defined a composite endpoint (CE) of neonatal mortality and morbidity combining outcomes of death, respiratory distress syndrome (RDS), bronchopulmonary dysplasia, intraventricular hemorrhage grade 3/4, periventricular leukomalacia, necrotizing enterocolitis + surgery, treated retinopathy of prematurity (ROP), sepsis and meningitis. Initially, the CE rate was assessed using data from one 3° referral center (Medical University of South Carolina (MUSC)) and was used to determine RCT sample size.

Objectives: The aim of this study was to understand if the CE rate was representative across the USA. We compared results with a diverse population from four integrated data networks.

Methods: Retrospective analyses were conducted using mother–infant linked pairs in electronic medical records from seven US States in Quintiles' COMparative effectiveness PAtient Safety and Surveillance (COMPASS) Network (2001–2012), funded by GSK. As in MUSC, the frequency of the CE was assessed among singleton infants born ≥24 weeks' gestational age (w GA); those with congenital conditions or born to women with eclampsia/HELLP, placental conditions, or infection were excluded.

Results: Comparing uncomplicated, singleton births in COMPASS (n = 56 572) and MUSC (17 957), preterm birth rates (<37w GA) were 4.7% and 8.3%, respectively; 2.3% and 3.4% had one or more of the morbidities in the CE. In MUSC, the CE was 96–100% among 24–29w GA infants; rates were 20–30% lower in COMPASS, declining steadily from 89 to 78%. CE rates remained lower in week 30 in COMPASS (78% vs. 51%) but merged by week 31 and remained similar for later GA births. Differences in the CE during 24–30 weeks reflect lower continuous positive airway pressure/ventilator use in RDS and laser treatment in ROP cases in COMPASS.

Conclusions: Exact comparisons between populations are challenging owing to differences in distribution of race, geography, hospital type/size, and medical coding and treatment protocols. As a result of lower CE rates among early preterms, an adaptive study design allowing sample size re-estimation was implemented for planned RCTs.

24. Validity of Delivery Outcomes Within the Medicaid Analytic eXtract (MAX) Database

Caitlin A. Knox,1,2 Juan Hincapie Castillo,1 and Almut G. Winterstein.1,31Pharmaceutical Outcomes and Policy, University of Florida, Gainesville, FL, USA;2Safety and Epidemiology, HealthCore, Andover, MA, USA;3Department of Epidemiology, University of Florida, Gainesville, FL, USA.

Background: Previous research has developed and validated algorithms to identify premature birth utilizing both maternal and infant information in administrative claims data. However, research regarding the identification of delivery outcomes utilizing only maternal claims is limited.

Objectives: The aim of this study was to assess the validity of delivery outcomes including the identification of cesarean section delivery, as well as premature deliveries in the Medicaid Analytic eXtract (MAX) database.

Methods: We identified women aged 12 to 55 years with a claim for a live birth using the 1999–2006 Florida and Texas MAX data. ICD9-CM codes were used to ascertain potential cases in the MAX claims, and then we reviewed the corresponding BCR for each subject. We calculated sensitivity, specificity, and positive predictive value (PPV) using the BCR as reference.

Results: We identified 67 937 live births in MAX for women with 12 months' continuous eligibility before the start of pregnancy until delivery. After linkage with the BCR, among the 6760 births identified as premature in MAX, the sensitivity was 46.0%, with specificity of 97.0% and a PPV of 72.8%. Utilizing CPT-4 procedure codes, ICD-9-CM diagnosis codes, we identified almost 13 000 live births that occurred via cesarean section delivery in MAX with a sensitivity of 54.3%, a specificity of 99.5%, and a PPV of 98.1%.

Conclusions: We were able to conduct an objective evaluation of the validity of diagnostic codes used to identify delivery outcomes. Algorithms for both premature and cesarean section delivery using only maternal administrative claims produced moderate and high specificity, respectively. Our results can be used to conduct corrections of measurement error in epidemiologic studies dealing with pregnancy and related outcomes.

25. Using Pharmacoepidemiologic Studies to Inform Drug Policy

Sandra Knowles,1,3 Mina Tadrous,1,2,3 Diana Martins,2 Tara Gomes,1,2,3 David Juurlink,2,3,4 and Muhammad Mamdani.1,2,31Li Ka Shing Knowledge Institute, St. Michael's Hospital, Toronto, ON, Canada;2Institute for Clinical Evaluative Sciences, Toronto, ON, Canada;3University of Toronto, Toronto, ON, Canada;4Department of Medicine, Sunnybrook Health Sciences Centre, Toronto, ON, Canada.

Background: Pharmacoepidemiologic studies generate information regarding drug utilization, benefits and risks, which can be incorporated into drug class reviews to provide important contextualization to help inform public drug policy. The Ontario Drug Policy Research Network framework for drug class reviews combines multiple types of evidence to generate policy recommendations for the Ontario government. Within this model, pharmacoepidemiology studies have been used as a tool to quantify and describe the use, adherence and safety of medications within a drug class.

Objectives: We describe the role of pharmacoepidemiology research in formulary modernization for the Ontario Public Drug Programs using two drug class reviews.

Methods: Using linked databases in Ontario including prescription records and physicians' claims data, studies are conducted for health resource utilization, adherence and patient characteristics. Public drug program prescription records for six other provinces are also analyzed. National and provincial trends on all retail prescriptions are captured through IMS Geographic Prescription Monitor.

Results: In the review of triptans for migraines, a potential safety issue was identified based on Ontario utilization data; approximately 10% of triptan users were considered at risk for medication overuse headache. This led to policy recommendations that included expanded availability of triptans but with monthly quantity limits. A review of testosterone replacement therapy (TRT) showed a 29-fold increase in use of topical TRT over 7 years in men 65 years and older. Approximately one-third of new testosterone users had no lab test for testosterone levels in the year prior to their first prescription for TRT, despite current policy stipulating that baseline testing be carried out. These data, in conjunction with safety concerns and questionable efficacy of TRT in older men, resulted in more restrictive policy recommendations for TRT.

Conclusions: Pharmacoepidemiologic studies provide essential information regarding drug utilization, benefits and risks; these data have been incorporated into our comprehensive drug class reviews to help inform public drug policy.

26. Impact of Texas's 2010 “Pill Mill” Law on Opioid Prescribing and Utilization

Tatyana Lyapustina,1,2 Lainie Rutkow,3 Matthew Daubresse,1,4 and Caleb Alexander.1,4,51Department of Epidemiology, Johns Hopkins Bloomberg School of Public Health, Baltimore, MD, USA;2Johns Hopkins School of Medicine, Baltimore, MD, USA;3Health Policy and Management, Johns Hopkins Bloomberg School of Public Health, Baltimore, MD, USA;4Center for Drug Safety and Effectiveness, Johns Hopkins Bloomberg School of Public Health, Baltimore, MD, USA;5Division of General Internal Medicine, Johns Hopkins Medicine, Baltimore, MD, USA.

Background: One of the ways that states address the abuse and diversion of prescription opioid pain relievers is through strengthening the regulation of pain management clinics; however, the effect of such measures remains unclear.

Objectives: The aim of this study was to quantify the impact of Texas's 2010 “pill mill” law on overall and high risk opioid prescribing and utilization.

Methods: We used the IMS Health LRx LifeLink database to examine anonymized, patient-level pharmacy claims for a closed cohort of individuals filling prescription opioids in Texas between September 2009 and August 2011. Our primary outcomes were derived at a monthly level and included average morphine equivalent dose (MED) per transaction, opioid volume, number of opioid prescriptions, and quantity of opioids dispensed. We compared observed values among our cohort with the counterfactual, which we estimated based on pre-intervention levels and trends. We conducted sensitivity analyses by varying the length of the observation period and examining an open rather than closed cohort.

Results: Our final cohort included 8.3 million patients, 737 123 providers, and 2829 pharmacies. During the 24-month observation period, there were 724 million prescription transactions in Texas, of which 10.5% were for opioids. Overall, Texas's “pill mill” law was associated with declines in average MED (0.57 mg/month, 95% confidence interval (CI) 0.057–1.09), monthly opioid volume (9.99 kg/month, 95%CI 7.11–12.86), monthly number of opioid prescriptions (12 200 prescriptions/month, 95%CI 9150–15 300), and monthly quantity of opioids dispensed (714 000 pills/month, 95%CI 550 000–877 000). These reductions were concentrated among prescribers and patients with the highest opioid prescribing and utilization at baseline. Sensitivity analyses supported the findings of the main results.

Conclusions: Following the implementation of Texas's 2010 “pill mill” law, there were moderate, statistically significant reductions in average MED, opioid volume, number of opioid prescriptions, and quantity of opioids dispensed within the state. These reductions were limited to individuals with higher risk baseline behaviors.

27. Risk of Major Adverse Cardiovascular Events and Transfusion Among US Hemodialysis Patients After CMS and FDA Policy Changes in 2011

Cunlin Wang,1 Robert Kane,1 Mark Levenson,1 Jeffrey Kelman,2 Mickel Wernecke,3 Joo-Yeon Lee,1 Chris Worrall,2 Thomas MaCurdy,3 and David J. Graham.11Center for Drug Evaluation and Research, Food and Drug Administration, Silver Spring, MD, USA;2Center for Medicare and Medicaid Service, Baltimore, MD, USA;3ACUMEN, LLC, Burlingame, CA, USA.

Background: In 2011, the Centers for Medicare and Medicaid Services (CMS) changed its reimbursement policy for erythropoiesis stimulating agents (ESA) used in dialysis. Additionally, the Food and Drug Administration revised the prescribing information for ESAs to reduce their use. Effects of these changes are not yet fully understood.

Objectives: The aim of this study was to compare the risk of major adverse cardiovascular events (MACE) and blood transfusions in dialysis patients before and after the policy changes.

Methods: Patients from the CMS End Stage Renal Disease Program were divided retrospectively into two cohorts based on the date of dialysis initiation and were followed up until the first occurrence of renal transplantation, Medicare disenrollment, switch to peritoneal dialysis, or a study outcome. The window for the pre-policy cohort was January 2008 to December 2009, and for the post-policy cohort, July 2011 to June 2013. Cox regression was used to estimate the relative risk between the two periods for all study outcomes: MACE and its components (acute myocardial infarction (AMI), stroke, and all-cause mortality), hospitalized congestive heart failure (CHF), venous thromboembolism (VTE), and transfusion episodes.

Results: A total of 69 718 incident dialysis patients were included in the analysis. Baseline characteristics at the dialysis initiation were similar between cohorts. Decreases in both the percentage of patients receiving ESAs and the average ESA dose were observed after the policy changes, independent of patients' hemoglobin levels. Post-policy, the risk of stroke decreased (hazard ratio (HR) = 0.77, 95% confidence interval (CI) 0.64–0.93) among patients initiating dialysis. The risk of MACE, death, CHF, and VTE was similar between cohorts, and the risk of AMI was slightly lower in the post-policy cohort: HR = 0.89, 95%CI 0.77–1.02, p < 0.1. Blood transfusions increased post-policy (HR = 1.09, 95%CI 1.07–1.12) but trended toward equilibrium at the end of 2-year follow-up.

Conclusions: Post-policy, there was a significant reduction in the risk of stroke, no change in the risk of MACE, and a slight decrease in the risk of AMI, while transfusion use increased. Additional analysis is needed to confirm whether the risk of transfusion regressed to pre-policy level over time.

28. The Effect of Medicaid Peer Review Prior Authorization Policies on Pediatric Use of Antipsychotic Medications

Mehmet Burcu,1 Julie M. Zito,1 Stephen McKean,2 Rob Warnock,2 Peter Herbert,2 Sumit Verma,2 Chris M. Worrall,3 Steve Chu,3 and Ali Mohamadi.41Pharmaceutical Health Services Research, University of Maryland Baltimore, Baltimore, MD, USA;2Acumen, LLC, Burlingame, CA, USA;3Centers for Medicare and Medicaid Services, Washington, DC, USA;4Center for Drug Evaluation and Research, US Food and Drug Administration, Silver Spring, MD, USA.

Background: Several state Medicaid agencies have recently adopted novel approaches targeting expanded use of atypical antipsychotic (AAP) medications in children, namely peer review prior authorization (PA) policies. Physicians are required to receive pre-approval from these Medicaid agencies through contracted clinicians (peer reviewers) to be able to prescribe AAPs to children under a certain age.

Objectives: The aim of this study was to assess the impact of peer review PA policies on the use of AAPs and other psychotropic medication classes among Medicaid-insured youth (0–17 years).

Methods: We used administrative Medicaid claims data from four geographically diverse states. PA policies were implemented for children ages <5 years in state A, <6 years in states B and C, and <8 years in state D. We used interrupted time-series design to assess monthly use of AAPs and other psychotropic medication classes across 36 months. In multivariable logistic regression models with generalized estimating equations, we added interaction terms for time period and age group to assess whether changes in medication use differed by age group in the post-policy versus pre-policy periods.

Results: Compared with the pre-policy period, AAP use in the post-policy period decreased significantly for children ages <5–8 years in state A from 0.06% to 0.04% (adjusted odds ratio (OR) = 0.65 [0.53–0.79]), in state B from 0.13% to 0.08% (OR = 0.71 [0.64–0.77]), in state C from 0.15% to 0.14% (OR = 0.91 [0.85–0.98]), and in state D from 0.35% to 0.26% (OR = 0.75 [0.72–0.78]). In contrast, AAP use increased significantly for older youth in states B, C, and D, except for those in state A (OR = 0.93 [0.91–0.96]). Restrictions on AAP use did not result in increased use of other psychotropic medication classes, which mostly remained stable or declined.

Conclusions: With the implementation of peer review PA policies, the use of AAP medications declined substantially in children younger than 5–8 years, with no apparent substitution of other psychotropic medication classes. The policy had no discernible effect in older youth whose AAP use continues to be high. Long-term consequences of these policies on clinical outcomes warrant further research.

29. Education and Presence of an Infectious Diseases (ID) Pharmacist in an Outpatient Primary Care Clinic

Diane M. Gomes,1,2 Aisling R. Caffrey,1,2 and Kerry L. LaPlante.1,21Infectious Diseases Research Program, Providence Veteran Affairs Medical Center, Providence, RI, USA;2College of Pharmacy, University of Rhode Island, Kingston, RI, USA.

Background: In the USA, antibiotics are the second most commonly prescribed medications in ambulatory settings. Preventing unnecessary antibiotic use in the primary care setting is essential to overall stewardship efforts.

Objectives: The aim of this study was to assess the change in antibiotic prescribing for outpatient infections after implementation of education and infectious diseases (ID) pharmacist presence in a primary care clinic.

Methods: The Providence VA Medical Center consists of eight primary care teams. In March 2014, two teams with medical residents received the following intervention: presence of an ID pharmacist during clinic hours and live education on prescribing guidelines for outpatient infections (e.g., sinusitis, bronchitis). The other six teams served as the control group. Primary care visits for infections, identified from diagnosis codes, were extracted for the pre-intervention (March to September 2013) and post-intervention (March to September 2014) periods. We then identified antibiotic prescriptions 1 day prior to the visit or within 3 days after the visit. Between-group differences were assessed using X2, Fisher exact, or Cochran–Mantel–Haenszel.

Results: There were 350 primary care visits associated with an infection diagnosis of interest during the pre (n = 137) and post (n

ADMS 1010 Exploring the functions of Business Section: P Muhammad Usman Eytan Lasry 211250685 Building on ideas from the management century, I would propose that we have seen three “ages” of management since the industrial revolution, with each putting the emphasis on a different theme: execution, expertise, and empathy. First era consists of “age of science management” which lasted from 1880 until the end of The Second World War. Second era, consists of application of humanistic psychology, started late 1940s and ended during 1980s. Finally, the third era of globalism started in mid 1980s and is currently present. The first era of scientific management developed theories that emphasized efficiency, lack of variation, consistency of production, and predictability. The goal was to optimize the outputs that could be generated from a specific set of inputs. During this era, the “principals of management” were developed from the findings of Fredrick

0 thoughts on “Adms 1010 Assignment 10-1

Leave a Reply

Your email address will not be published. Required fields are marked *