Survey
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
Clinical Epidemiology & Analytics – filling the evidence gap Woodie M. Zachry, III, PhD Global Lead Clinical Epidemiology and Analytics The Present – Overview of CE&A activities Establishing the disease profile – – – – Natural history of the disease Issues in special populations Incidence/prevalence of the disease Risk factors of disease Identifying drug safety issues in collaboration with Pharmacovigilance – Safety issues of Abbott products and other current therapies – Subpopulations at higher risk? – Drug-drug interactions? Providing clinical trial support and instrumentation – Identifying biomarkers/surrogate endpoints and its relationship to outcomes Company Confidential © 2009 Abbott 2 Study Types & Data Sources Study Type Potential Sources of Information Preclinical AEGIS AERS WHO Registry Claims Clinical Trials Database Database Data Systematic Review with Meta Analysis Randomized Controlled Trial Experimental Designs Cohort Case Control Case Report X X X Company Confidential © 2009 Abbott X Literature Cochrane X X X X X X X X X X X X X X X X 3 GRADE – The Grading of Recommendations Assessment, Development and Evaluation (GRADE ) – Provides a system for rating quality of evidence and strength of recommendations that is explicit, comprehensive, transparent, and pragmatic and is increasingly being adopted by organizations worldwide • High quality— Further research is very unlikely to change the estimate of effect • Moderate quality— Further research is likely to have an important impact on the estimate of effect and may change the estimate • Low quality— Further research is very likely to have an important impact on the estimate of effect and is likely to change the estimate • Very low quality— Any estimate of effect is very uncertain Company Confidential © 2009 Abbott 4 Hierarchy of Evidence Meta-analysis RCT Prospective Less Bias Observational Studies Retrospective Less Bias Case-Control Comparison with bias Case Series Uncontrolled Nonsystematic Clinical Experience Company Confidential © 2009 Abbott 5 Multiple EBM Stakeholders Levels of Evidence Chest CONSORT Statement RCTs Clinical Practice Guidelines FDA AHRQ NIH Users’ Guides JAMA HTAs ACP Journal Club Clinical Evidence Company Confidential © 2009 Abbott NICE 6 EMEA QUORUM Statement Systematic Review Meta-Analysis Cochrane Collaborative Where we want to be Evidence Summaries across All Phases of Development and Study Designs Company Confidential © 2009 Abbott Evidence Based Approach 7 Identify Evidence Gaps and Propose Ways to Fill Gaps Case-Control analysis of ambulance, emergency room, or inpatient hospital events for epilepsy and antiepileptic drug formulation changes Woodie M Zachry, III PhD Quynhchau D Doan PhD Jerry D Clewell, PharmD Brien J Smith MD Background Epilepsy Treatment Disease & Treatment – Incidence: 200,000 cases annually in US, Prevalence: 1% from birth to age 20, then 3% by age 75.6 – Treatment choice dependent upon Partial vs. Generalized presentation, history & secondary causes. – “A-rated” compounds are considered to be therapeutically bioequivalent to the reference listed drug (United States Food & Drug Administration Center for Drug Evaluation & Research) • Generic substitution, observational experience – 65% of US physicians surveyed reported caring for a patient who had a breakthrough seizure after a brand to generic switch.1 – 49.2% of foreign physicians surveyed reported problems when switching from brand alternatives to generics.2 – 67.8% of surveyed neurologists reported breakthrough seizures after a switch.3 – 12.9% of Lamotrigine switches had to be switched back due to medical necessity (v.s 1.5-2.9 for Non-AED).4 – 10.8% of patients switching supplier for CBZ, PHT, & VAL had perceived problems validated by GP.5 1. 2. 3. 4. 5. 6. Berg MJ, Gross RA. Physicians and patients perceive that generic drug substitution of anti-epileptic drugs can cause breakthrough seizures - results from a U.S. survey. 60th Annual Meeting of the American Epilepsy Society; Dec 1-5, 2006; San Diego, California. Kramer G. et al. Experience with generic drugs in epilepsy patients: an electronic survey of members of the German, Austrian and Swiss branches of the ILAE. Epilepsia 2007;48, 609-11. Wilner AN. Therapeutic equivalency of generic antiepileptic drugs: results of a survey. Epilepsy Behavior 2004;5(6):995-8. Andermann F, et al. Compulsory generic switching of antiepileptic drugs: high switchback rates ro branded compounds compared with other drug classes. Epilepsia 2007;48(3):464-9. Crawford P. et al. Generic Prescribing for epilepsy. Is it safe? Siezure 1996;5:1-5. Centers for Disease Control and Prevention 2007. http://www.cdc.gov/epilepsy/ Accessed October 10, 2007. Company Confidential © 2009 Abbott 9 Confidence in Treatment-Effect Relationship Low High Case Reports Case-Control Epidemiological Cohort Epidemiological RCCT Hypothesis generation Hypothesis test (without temporal relationship) Hypothesis test (with temporal relationship assessment) Hypothesis test (Cause – Effect relationship inferred) Spontaneous reports to authorities with variable completeness and data quality Subjects selected based on current disease status (yes / no). Exposed Vs. non-exposed subjects assembled before development of disease. Treatment and Control groups studied in randomized, blinded trial Retrospectively evaluate exposure to agent(s) & confounders Baseline confounding variables assessed before disease development. Detection bias Usually not possible to calculate rate of development of disease given the presence or absence of exposure.1,2 Treatment-emergent, temporal relationship to exposure, and incidence of disease can be measured. Causality can be inferred Cannot establish causality Most closely resembles RCT design.1,2 Generalizability limited by inability to detect events in the greater population, and sub-populations. Selection bias Effects of risk factors are most difficult to evaluate Confounding patient factors often not considered Cannot establish causality Limited ability to detect rare events. Cannot establish causality 1Mednick D, Day D. JMCP 1997;3(1):66-75. 2Hennekens, C. Epidemiology in Medicine. 3Harris S. J Cont. Ed. In Health Prof 2000;20:133-45. Company Confidential © 2009 Abbott 10 Methods • Objective: To determine if patients who received epilepsy care in an inpatient setting, emergency room, or ambulance have greater odds of having had a change between A rated AED medication alternatives in the past 6 months when compared to epileptic patients with no evidence of receiving epileptic care in similar settings. Company Confidential © 2009 Abbott 11 Methods • Retrospective claims database analysis utilizing the Ingenix LabRx database • Case-control study – Unmatched & Matched 1:3 for age within 5 years and epilepsy diagnosis type – Index date for case patients: 1st seizure event requiring inpatient admission, emergency room visit, or ambulance during 3Q2006 – 4Q2006 – Index date for control patients: 1st office visit during 3Q2006 – 4Q2006 • Index primary ICD-9 diagnosis of 345.xx excluding 345.6 • 12 and 64 years of age • No inpatient admission, emergency room visit, or ambulance in 6 months prior to index date • Possess at least 145 day supply of AED medication for 6 months prior to index event • Continuous eligibility for 6 months prior to index. Company Confidential © 2009 Abbott 12 Diagnosis Categories • Siezure type • Modifier – Generalized – XXX.X0 – without mention of intractable epilepsy • Convulsive 345.0 • Non-convulsive 345.1 – XXX.X1 – with mention of intractable epilepsy • Petite mal status 345.2 • Grand mal status 345.3 – Partial • Complex partial 345.4 • Simple partial 345.5 • Epilepsia partialis continua 345.7 – Other • Other forms 345.8 • Epilepsy unspecified 345.9 Company Confidential © 2009 Abbott 13 All Patients (Non-Matched) Variable % Male Age (SD) Case Patients Control Patients P value (n=417) (n=5562) (a-priori=0.05) 44.8% 45.1% NS 37.4yrs (14.8) 37.2yrs (14.6) NS Insurance <0.001 Commercial 95.4% 98.1% Medicaid 4.6% 1.9% US Region NS West 12.7% 14.5% Midwest 33.1% 33.8% South 42.0% 40.0% Northeast 12.2% 11.6% Company Confidential © 2009 Abbott 14 Matched Case-Control Patients Variable % Male Age (SD) Case Patients Control Patients P value (n=416) (n=1248) (a-priori=0.05) 45.0% 44.2% NS 37.4yrs (14.8) 37.5yrs (14.7) NS Insurance 0.004 Commercial 95.4% 98.2% Medicaid 4.6% 1.8% US Region NS West 12.7 % 14.3 % Midwest 33.2 % 33.6 % South 41.8 % 39.2 % Northeast 12.3 % 13.0 % Company Confidential © 2009 Abbott 15 All Patients (Non-Matched) Seizure Type Generalized nonintractable Case Patients (n=417) 30.5% Control Patients (n=5562) 35.5% Generalized intractable 9.1% 6.7% Partial nonintractable 19.2% 36.0% Partial intractable 26.4% 16.6% Other, nonintractable 3.1% 1.1% Other, intractable 11.8% 4.1% 2 <0.001 Company Confidential © 2009 Abbott 16 Matched Case-Control Patients Seizure Type Case Patients (n=416) 30.5% Control Patients (n=1248) 30.5% Generalized intractable 9.1% 9.1% Partial nonintractable 19.2% 19.2% Partial intractable 26.4% 26.4% Other, nonintractable 2.9% 2.9% Other, intractable 11.8% 11.8% Generalized nonintractable 2 = NS Company Confidential © 2009 Abbott 17 All Patients (Non-Matched) • Odds of a change between A rated alternatives Patient switched medications Patients did NOT switch medications Case 47 370 Control 346 5216 Odds ratio = 1.915 (95% CI, 1.387 - 2.644) Company Confidential © 2009 Abbott 18 How to calculate an unmatched odds ratio Unmatched analysis Risk factor Cohort Status Case Control Exposed a b a+b Not Exposed c d c+d a+c b+d n Switch No Switch Equations OR = ad/bc SE = SQRT(1/a+1/b+1/c+1/d) CI = EXP(logeOR + 1.96SE) Example Example Calculation Case Control OR estimate 47 346 393 SE 370 5216 5586 1.96*SE 417 5562 lnOR Lower Limit Upper Limit Company Confidential © 2009 Abbott 19 1.91 0.16 0.32 0.65 0.33 0.97 1.39 2.64 Matched Case-Control Patients • Odds of a change between A rated alternatives Patient switched medications Patients did NOT switch medications Case 47 369 Control 81 1167 Odds ratio = 1.811 (95% CI, 1.247 – 2.629) Company Confidential © 2009 Abbott 20 Matched primary analysis Case with exposure Number of controls with exposure 0 1 2 yes 40 7 0 no 298 68 3 total i = # of exposures mi = number of t i where the case is exposed ti = the total number of sets with i exposures i mi 1 2 3 108 10 0 1.810811 0.190201 0.190201 0.372795 0.593775 0.22098 1.247298 3 0 0 iti 108 10 0 118 H i(4-i)ti ti 1 2 3 OR estimate SE 1.96*SE lnOR Lower Limit 40 7 0 47 total i ti total sets 47 369 416 i(4-i)ti 108 20 0 128 324 40 0 364 (4-i)m i i(ti -m i ) 120 68 14 6 0 0 134 74 J (iOR+4-i)2 H/J 324 23.1439 13.99937 40 31.60263 1.265717 0 41.37619 0 15.26509 Cochran-Mantel-Haenszel Statistic MH calc 60 58 3364 MH stat 9.241758 P value 0.0024 21 Matched Case-Control Patients Excluding Medicaid Patients • Odds of a change between A rated alternatives Patient switched medications Patients did NOT switch medications Case 45 352 Control 79 1146 Odds ratio = 1.855 (95% CI, 1.262 – 2.726) Company Confidential © 2009 Abbott 22 Matched Case-Control Patients Excluding Patients Who Changed Dosage Schedule • Odds of a change between A rated alternatives Patient switched medications Patients did NOT switch medications Case 22 205 Control 49 918 Odds ratio = 2.011 (95% CI, 1.189 – 3.4) Company Confidential © 2009 Abbott 23 Discussion • This study tested a hypothesis and found a relationship between emergent and inpatient care visits and previous AED formulation switching. This is concordant with problems identified in the survey and case study literature. – surveyed physicians believe there may be potential safety problems associated with switching between AED formulations for the same medication – There is some evidence of a significant percentage of patients who must switch back to a branded formulation after trying a generic formulation. Company Confidential © 2009 Abbott 24 Discussion • This study assumes that patients experiencing break-through seizures will seek care in emergency and inpatient settings more often than ambulatory settings. • Study subjects seeking care for break through events in an ambulatory setting may have attenuated the true magnitude of the significant relationship found in this study. • Attempts were made to strengthen the assumption that subjects were taking AEDs. However, claims data only records the date a prescription was filled, not when or if the patient took the medication. • Subtle differences in formulations may take time to accumulate and effect outcomes. However, the majority of formulation changes occurred within 2 months of the index event. Company Confidential © 2009 Abbott 25 Discussion • Several factors may play a role in break through seizures that were not controlled for in this analysis (e.g., sleep deprivation, alcohol intake, hormonal influences). These effects may be additive to or even supersede formulation changes in precipitating break-through seizures. • Zonisamide became available as a generic during the study time period. The high percentage of zonisamide formulation changes may have played a role in the significant relationship discovered. • Case-control studies cannot establish a temporal association between AED formulation switches and outcomes. Company Confidential © 2009 Abbott 26 Conclusions • This analysis has found an association between patients who utilized an ER, ambulance or inpatient hospital for epilepsy and the prior occurrence of AED formulation switching involving “A” rated generics. – After matching by age and epilepsy diagnosis, Cases had 81% greater odds of prior “A” rated switches compared to matched controls. – The case population had significantly more Medicaid patients. – Post hoc analyses excluding patients who had a dosage change and Medicaid patients did not change the significance of the original analysis. – Further investigations are warranted to better understand a possible cause-effect relationship. Company Confidential © 2009 Abbott 27 Company Confidential © 2009 Abbott 28 Hierarchy of Evidence Meta-analysis RCT Prospective Less Bias Observational Studies Retrospective Less Bias Case-Control Comparison with bias Case Series Uncontrolled Nonsystematic Clinical Experience Company Confidential © 2009 Abbott 29