Statistical Methods in Healthcare

Frederick Faltin, Founder and Managing Director of The Faltin Group, USA. Ron Kenett, Chairman and CEO of the KPA Group, Israel. Fabrizio Ruggeri, CNR IMATI, Italy.

Foreword xix Preface xxi Editors xxiii Contributors xxv Part One STATISTICS IN THE DEVELOPMENT OF PHARMACEUTICAL PRODUCTS 1 Statistical Aspects in ICH, FDA and EMA Guidelines 3 Allan Sampson and Ron S. Kenett Synopsis 3 1.1 Introduction 3 1.2 ICH Guidelines Overview 5 1.3 ICH Guidelines for Determining Efficacy 7 1.4 ICH Quality Guidelines 11 1.5 Other Guidelines 14 1.6 Statistical Challenges in Drug Products Development and Manufacturing 17 1.7 Summary 18 References 19 2 Statistical Methods in Clinical Trials 22 Telba Irony, Caiyan Li and Phyllis Silverman Synopsis 22 2.1 Introduction 22 2.1.1 Claims 23 2.1.2 Endpoints 23 2.1.3 Types of Study Designs and Controls 24 2.2 Hypothesis Testing, Significance Levels, p-values, Power and Sample Size 25 2.2.1 Hypothesis Testing 25 2.2.2 Statistical Errors, Significance Levels and p-values 25 2.2.3 Confidence Intervals 26 2.2.4 Statistical Power and Sample Size 27 2.3 Bias, Randomization and Blinding/Masking 29 2.3.1 Bias 29 2.3.2 Randomization 30 2.3.3 Blinding or Masking 31 2.4 Covariate Adjustment and Simpson s Paradox 32 2.4.1 Simpson s Paradox 32 2.4.2 Statistical Methods for Covariate Adjustment 34 2.5 Meta-analysis, Pooling and Interaction 35 2.5.1 Meta-analysis 35 2.5.2 Pooling and Interaction 37 2.6 Missing Data, Intent-to-treat and Other Analyses Cohorts 38 2.6.1 Missing Data 38 2.6.2 Intent-to-treat (ITT) and Other Analysis Cohorts 39 2.7 Multiplicity, Subgroup and Interim Analyses 40 2.7.1 Multiplicity 40 2.7.2 Subgroup Analyses 41 2.7.3 Interim Analyses 42 2.8 Survival Analyses 43 2.8.1 Estimating Survival Functions 44 2.8.2 Comparison of Survival Functions 45 2.9 Propensity Score 46 2.10 Bayesian Versus Frequentist Approaches to Clinical Trials 48 2.11 Adaptive Designs 50 2.11.1 Sequential Designs 51 2.12 Drugs Versus Devices 53 References 54 Further Reading 54 3 Pharmacometrics in Drug Development 56 Serge Guzy and Robert Bauer Synopsis 56 3.1 Introduction 56 3.1.1 Pharmacometrics Definition 56 3.1.2 Dose-response Relationship 57 3.1.3 FDA Perspective of Pharmacometrics 57 3.1.4 When Should We Perform Pharmacometric Analysis? 58 3.1.5 Pharmacometric Software Tools 58 3.1.6 Organization of the Chapter 58 3.2 Pharmacometric Components 59 3.2.1 Pharmacokinetics (PK) 59 3.2.2 Pharmacodynamics (PD) 59 3.2.3 Disease Progression 59 3.2.4 Simulation of Clinical Trials 59 3.3 Pharmacokinetic/Pharmacodynamic Analysis 60 3.3.1 Compartmental Methods 60 3.4 Translating Dynamic Processes into a Mathematical Framework 61 3.5 Nonlinear Mixed-effect Modeling 63 3.6 Model Formulation and Derivation of the Log-likelihood 63 3.7 Review of the Most Important Pharmacometric Software Characteristics 65 3.7.1 NONMEM 65 3.7.2 PDx-MC-PEM 65 3.7.3 MONOLIX 66 3.7.4 WinBUGS 66 3.7.5 S-ADAPT 66 3.8 Maximum Likelihood Method of Population Analysis 67 3.9 Case Study: Population PK/PD Analysis in Multiple Sclerosis Patients 68 3.9.1 Study Design 68 3.9.2 Model Building 69 3.9.3 The PK Model 69 3.9.4 Platelet Modeling 69 3.9.5 T1 Lesions Model 69 3.10 Mathematical Description of the Dynamic Processes Characterizing the PK/Safety/Efficacy System 70 3.10.1 Optimization Procedure and Phase 2b Simulation Procedures 72 3.10.2 Clinical Simulation Results and Discussion 72 3.10.3 Calculation of the Cumulative Number of T1 Lesions and the Percentage MRI Improvement 73 3.10.4 Estimation of the Percentage of Patients to Reach Platelet Counts Below a Certain Threshold Value 73 3.10.5 Tentative Proposal for the Phase 2b Trial Design 74 3.11 Summary 75 3.11 References 76 4 Interactive Clinical Trial Design 78 Zvia Agur Synopsis 78 4.1 Introduction 79 4.2 Development of the Virtual Patient Concept 80 4.2.1 The Basic Virtual Patient Model 80 4.3 Use of the Virtual Patient Concept to Predict Improved Drug Schedules 86 4.3.1 Modeling Vascular Tumor Growth 86 4.3.2 Synthetic Human Population (SHP) 91 4.4 The Interactive Clinical Trial Design (ICTD) Algorithm 94 4.4.1 Preclinical Phase: Constructing the PK/PD Module 94 4.4.2 Phase I: Finalizing and Validating the PK/PD Module 95 4.4.3 Interim Stage Between Phase I and Phase II: Intensive Simulations of Short-term Treatments 96 4.4.4 Phase II and Phase III: Focusing the Clinical Trials 96 4.4.5 Interactive Clinical Trial Design Method as Compared to Adaptive Clinical Trial Design Methods 99 4.5 Summary 100 Acknowledgements 100 References 100 5 Stage-wise Clinical Trial Experiments in Phases I, II and III 103 Shelemyahu Zacks Synopsis 103 5.1 Introduction 103 5.2 Phase I Clinical Trials 104 5.2.1 Up-and-down Adaptive Designs in Search of the MTD 105 5.2.2 The Continuous Reassessment Method 107 5.2.3 Efficient Dose Escalation Scheme With Overdose Control (EWOC) 109 5.3 Adaptive Methods for Phase II Trials 110 5.3.1 Individual Dosing 110 5.3.2 Termination of Phase II 111 5.4 Adaptive Methods for Phase III 112 5.4.1 Randomization in Clinical Trials 112 5.4.2 Adaptive Randomization Procedures 113 5.4.3 Group Sequential Methods: Testing Hypotheses 119 5.5 Summary 119 References 120 6 Risk Management in Drug Manufacturing and Healthcare 122 Ron S. Kenett Synopsis 122 6.1 Introduction to Risks in Healthcare and Trends in Reporting Systems 122 6.2 Reporting Adverse Events 124 6.3 Risk Management and Optimizing Decisions With Data 126 6.3.1 Introduction to Risk Management 126 6.3.2 Bayesian Methods in Risk Management 128 6.3.3 Basics of Financial Engineering and Risk Management 129 6.3.4 Black Swans and the Taleb Quadrants 130 6.4 Decision Support Systems for Managing Patient Healthcare Risks 131 6.5 The Hemodialysis Case Study 137 6.6 Risk-based Quality Audits of Drug Manufacturing Facilities 142 6.6.1 Background on Facility Quality Audits 142 6.6.2 Risk Dimensions of Facilities Manufacturing Drug Products 143 6.6.3 The Site Risk Assessment Structure 144 6.7 Summary 152 References 152 7 The Twenty-first Century Challenges in Drug Development 155 Yafit Stark Synopsis 155 7.1 The FDA s Critical Path Initiative 155 7.2 Lessons From 60 Years of Pharmaceutical Innovation 156 7.2.1 New-drug Performance Statistics 156 7.2.2 Currently There are Many Players, but Few Winners 156 7.2.3 Time to Approval Standard New Molecular Entities 157 7.3 The Challenges of Drug Development 158 7.3.1 Clinical Trials 158 7.3.2 The Critical-path Goals 159 7.3.3 Three Dimensions of the Critical Path 159 7.3.4 A New-product Development Toolkit 160 7.3.5 Towards a Better Safety Toolkit 160 7.3.6 Tools for Demonstrating Medical Utility 160 7.4 A New Era in Clinical Development 160 7.4.1 Advancing New Technologies in Clinical Development 161 7.4.2 Advancing New Clinical Trial Designs 161 7.4.3 Advancing Innovative Trial Designs 162 7.4.4 Implementing Pharmacogenomics (PGx) During All Stages of Clinical Development 162 7.5 The QbD and Clinical Aspects 163 7.5.1 Possible QbD Clinical Approach 164 7.5.2 Defining Clinical Design Space 164 7.5.3 Clinical Deliverables to QbD 164 7.5.4 Quality by Design in Clinical Development 165 References 166 Part Two STATISTICS IN OUTCOMES ANALYSIS 8 The Issue of Bias in Combined Modelling and Monitoring of Health Outcomes 169 Olivia A. J. Grigg Synopsis 169 8.1 Introduction 170 8.1.1 From the Industrial Setting to the Health Setting: Forms of Bias and the Flexibility of Control Charts 170 8.1.2 Specific Types of Control Chart 171 8.2 Example I: Re-estimating an Infection Rate Following a Signal 172 8.2.1 Results From a Shewhart and an EWMA Chart 172 8.2.2 Results From a CUSUM, and General Concerns About Bias 173 8.2.3 More About the EWMA as Both a Chart and an Estimator 174 8.3 Example II: Correcting Estimates of Length-of-stay Measures to Protect against Bias Caused by Data Entry Errors 175 8.3.1 The Multivariate EWMA Chart 175 8.3.2 A Risk Model for Length of Stay Given Patient Age and Weight 176 8.3.3 Risk Adjustment 176 8.3.4 Results From a Risk-adjusted Multivariate EWMA Chart 177 8.3.5 Correcting for Bias in Estimation Through Regression 178 8.4 Discussion 182 References 182 9 Disease Mapping 185 Annibale Biggeri and Dolores Catelan Synopsis 185 9.1 Introduction 186 9.2 Epidemiological Design Issues 186 9.3 Disease Tracking 187 9.4 Spatial Data 188 9.5 Maps 188 9.6 Statistical Models 191 9.7 Hierarchical Models for Disease Mapping 192 9.7.1 How to Choose Priors in Disease Mapping? 194 9.7.2 More on the BYM Model and the Clustering Term 195 9.7.3 Model Checking 200 9.8 Multivariate Disease Mapping 200 9.9 Special Issues 202 9.9.1 Gravitational Models 202 9.9.2 Wombling 202 9.9.3 Some Specific Statistical Modeling Examples 203 9.9.4 Ecological Bias 205 9.9.5 Area Profiling 207 9.10 Summary 210 References 210 10 Process Indicators and Outcome Measures in the Treatment of Acute Myocardial Infarction Patients 219 Alessandra Guglielmi, Francesca Ieva, Anna Maria Paganoni and Fabrizio Ruggeri Synopsis 219 10.1 Introduction 220 10.2 A Semiparametric Bayesian Generalized Linear Mixed Model 222 10.3 Hospitals Clustering 223 10.4 Applications to AMI Patients 224 10.5 Summary 227 References 228 11 Meta-analysis 230 Eva Negri Synopsis 230 11.1 Introduction 231 11.2 Formulation of the Research Question and Definition of Inclusion/Exclusion Criteria 232 11.3 Identification of Relevant Studies 233 11.4 Statistical Analysis 234 11.5 Extraction of Study-specific Information 234 11.6 Outcome Measures 235 11.6.1 Binary Outcome Measures 235 11.6.2 Continuous Outcome Measures 236 11.7 Estimation of the Pooled Effect 237 11.7.1 Fixed-effect Models 237 11.7.2 Random-effects Models 240 11.7.3 Random-effects vs. Fixed-effects Models 241 11.8 Exploring Heterogeneity 242 11.9 Other Statistical Issues 243 11.10 Forest Plots 243 11.11 Publication and Other Biases 245 11.12 Interpretation of Results and Report Writing 246 11.13 Summary 247 References 247 Part Three STATISTICAL PROCESS CONTROL IN HEALTHCARE 12 The Use of Control Charts in Healthcare 253 William H. Woodall, Benjamin M. Adams and James C. Benneyan Synopsis 253 12.1 Introduction 253 12.2 Selection of a Control Chart 255 12.2.1 Basic Shewhart-type Charts 255 12.2.2 Use of CUSUM and EWMA Charts 257 12.2.3 Risk-adjusted Monitoring 259 12.3 Implementation Issues 261 12.3.1 Overall Process Improvement System 261 12.3.2 Sampling Issues 262 12.3.3 Violations of Assumptions 262 12.3.4 Measures of Control Chart Performance 263 12.4 Certification and Governmental Oversight Applications 263 12.5 Comparing the Performance of Healthcare Providers 264 12.6 Summary 265 Acknowledgements 265 References 265 13 Common Challenges and Pitfalls Using SPC in Healthcare 268 Victoria Jordan and James C. Benneyan Synopsis 268 13.1 Introduction 268 13.2 Assuring Control Chart Performance 269 13.3 Cultural Challenges 270 13.3.1 Philosophical and Statistical Literacy 270 13.3.2 Acceptable Quality Levels 271 13.4 Implementation Challenges 272 13.4.1 Data Availability and Accuracy 272 13.4.2 Rational Subgroups 273 13.4.3 Specification Threshold Approaches 273 13.4.4 Establishing Versus Maintaining Stability 275 13.5 Technical Challenges 276 13.5.1 Common Errors 276 13.5.2 Subgroup Size Selection 278 13.5.3 Over-use of Supplementary Rules 279 13.5.4 g Charts 280 13.5.5 Misuse of Individuals Charts 281 13.5.6 Distributional Assumptions 282 13.6 Summary 284 References 285 14 Six Sigma in Healthcare 286 Shirley Y. Coleman Synopsis 286 14.1 Introduction 287 14.2 Six Sigma Background 288 14.3 Development of Six Sigma in Healthcare 289 14.4 The Phases and Tools of Six Sigma 292 14.5 DMAIC Overview 292 14.5.1 Define 292 14.5.2 Measure 293 14.5.3 Analyse 295 14.5.4 Improve 296 14.5.5 Control 297 14.5.6 Transfer 298 14.6 Operational Issues of Six Sigma 298 14.6.1 Personnel 298 14.6.2 Project Selection 300 14.6.3 Training 301 14.6.4 Kaizen Workshops 301 14.6.5 Organisation of Training 302 14.7 The Way Forward for Six Sigma in Healthcare 303 14.7.1 Variations 303 14.7.2 Six Sigma and the Complementary Methodology of Lean Six Sigma 304 14.7.3 Implementation Issues 305 14.7.4 Implications of Six Sigma for Statisticians 306 14.8 Summary 307 References 307 15 Statistical Process Control in Clinical Medicine 309 Per Winkel and Nien Fan Zhang Synopsis 309 15.1 Introduction 310 15.2 Methods 310 15.2.1 Control Charts 310 15.2.2 Measuring the Quality of a Process 311 15.2.3 Logistic Regression 311 15.2.4 Autocorrelation of Process Measurements 312 15.2.5 Simulation 312 15.3 Clinical Applications 313 15.3.1 Measures and Indicators of Quality of Healthcare 313 15.3.2 Applications of Control Charts 314 15.4 A Cautionary Note on the Risk-adjustment of Observational Data 324 15.5 Summary 328 Appendix A 328 15.A.1 The EWMA Chart 328 15.A.2 Logistic Regression 329 15.A.3 Autocovariance and Autocorrelation 330 Acknowledgements 330 References 330 Part Four APPLICATIONS TO HEALTHCARE POLICY AND IMPLEMENTATION 16 Modeling Kidney Allocation: A Data-driven Optimization Approach 335 Inbal Yahav Synopsis 335 16.1 Introduction 335 16.1.1 Literature Review 338 16.2 Problem Description 340 16.2.1 Notation 340 16.2.2 Choosing Objectives 341 16.3 Proposed Real-time Dynamic Allocation Policy 342 16.3.1 Stochastic Optimization Formulation 342 16.3.2 Knowledge-based Real-time Allocation Policy 343 16.4 Analytical Framework 344 16.4.1 Data 344 16.4.2 Model Estimation 344 16.5 Model Deployment 345 16.5.1 Stochastic Optimization Analysis 346 16.5.2 Knowledge-based Real-time Policy 347 16.6 Summary 350 Acknowledgement 352 References 352 17 Statistical Issues in Vaccine Safety Evaluation 353 Patrick Musonda Synopsis 353 17.1 Background 353 17.2 Motivation 354 17.3 The Self-controlled Case Series Model 354 17.4 Advantages and Limitations 357 17.5 Why Use the Self-controlled Case Series Method 358 17.6 Other Case-only Methods 358 17.7 Where the Self-controlled Case Series Method Has Been Used 359 17.8 Other Issues That were Explored in Improving the SCCM 360 17.9 Summary of the Chapter 362 References 362 18 Statistical Methods for Healthcare Economic Evaluation 365 Caterina Conigliani, Andrea Manca and Andrea Tancredi Synopsis 365 18.1 Introduction 365 18.2 Statistical Analysis of Cost-effectiveness 366 18.2.1 Incremental Cost-effectiveness Plane, Incremental Cost-effectiveness Ratio and Incremental Net Benefit 366 18.2.2 The Cost-effectiveness Acceptability Curve 368 18.3 Inference for Cost-effectiveness Data From Clinical Trials 369 18.3.1 Bayesian Parametric Modelling 370 18.3.2 Semiparametric Modelling and Nonparametric Statistical Methods 373 18.3.3 Transformation of the Data 374 18.4 Complex Decision Analysis Models 375 18.4.1 Markov Models 376 18.5 Further Extensions 378 18.5.1 Probabilistic Sensitivity Analysis and Value of Information Analysis 379 18.5.2 The Role of Bayesian Evidence Synthesis 380 18.6 Summary 383 References 383 19 Costing and Performance in Healthcare Management 386 Rosanna Tarricone and Aleksandra Torbica Synopsis 386 19.1 Introduction 387 19.2 Theoretical Approaches to Costing Healthcare Services: Opportunity Cost and Shadow Price 387 19.3 Costing Healthcare Services 388 19.3.1 Measuring Full Costs of Healthcare Services 389 19.3.2 Definition of the Cost Object (Output) 389 19.3.3 Classification of Cost Components (Direct vs. Non-direct Costs) 390 19.3.4 Selection of Allocation Methods 390 19.3.5 Calculation of Full Costs 392 19.4 Costing for Decision Making: Tariff Setting in Healthcare 392 19.4.1 General Features of Cost-based Pricing and Tariff Setting 393 19.4.2 Cost-based Tariff Setting in Practice: Prospective Payments System for Hospital Services Reimbursement 394 19.5 Costing, Tariffs and Performance Evaluation 395 19.5.1 Definition of Final Cost Object 396 19.5.2 Classification and Evaluation of Cost Components 396 19.5.3 Selection of Allocative Methods and Allocative Basis 397 19.5.4 Calculation of the Full Costs 397 19.5.5 Results 398 19.6 Discussion 400 19.7 Summary 402 References 403 Part Five APPLICATIONS TO HEALTHCARE MANAGEMENT 20 Statistical Issues in Healthcare Facilities Management 407 Daniel P. O Neill and Anja Drescher Synopsis 407 20.1 Introduction 407 20.2 Healthcare Facilities Management 409 20.2.1 Description 409 20.2.2 Relevant Data 410 20.3 Operating Expenses and the Cost Savings Opportunities Dilemma 412 20.4 The Case for Baselining 413 20.5 Facilities Capital ... is it Really Necessary? 414 20.5.1 Facilities Capital Management 414 20.5.2 A Census of Opportunities 415 20.5.3 Prioritization and Efficiency Factors 416 20.5.4 Project Management 417 20.6 Defining Clean, Orderly and in Good Repair 418 20.6.1 Customer Focus 418 20.6.2 Metrics and Methods 419 20.7 A Potential Objective Solution 420 20.8 Summary 424 References 425 21 Simulation for Improving Healthcare Service Management 426 Anne Shade Synopsis 426 21.1 Introduction 426 21.2 Talk-through and Walk-through Simulations 427 21.3 Spreadsheet Modelling 428 21.4 System Dynamics 429 21.5 Discrete Event Simulation 429 21.6 Creating a Discrete Event Simulation 431 21.7 Data Difficulties 432 21.8 Complex or Simple? 434 21.9 Design of Experiments for Validation, and for Testing Robustness 436 21.10 Other Issues 438 21.11 Case Study No. 1: Simulation for Capacity Planning 439 21.12 Case Study No. 2: Screening for Vascular Disease 440 21.13 Case Study No. 3: Meeting Waiting Time Targets in Orthopaedic Care 441 21.14 Case Study No. 4: Bed Capacity Implications Model (BECIM) 442 21.15 Summary 443 References 444 22 Statistical Issues in Insurance/payor Processes 445 Melissa Popkoski Synopsis 445 22.1 Introduction 445 22.2 Prescription Drug Claim Processing and Payment 446 22.2.1 General Process: High-level Outline 446 22.2.2 Prescription Drug Plan Part D Claims Payment Process 447 22.3 Case Study: Maximizing Part D Prescription Drug Claim Reimbursement 450 22.4 Looking Ahead 453 22.5 Summary 454 Reference 455 23 Quality of Electronic Medical Records 456 Dario Gregori and Paola Berchialla Synopsis 456 23.1 Introduction 456 23.2 Quality of Electronic Data Collections 459 23.2.1 Administrative Databases 461 23.2.2 Health Surveys 461 23.2.3 Patient Medical Records 462 23.2.4 Clinical Trials 462 23.2.5 Clinical Epidemiology Studies 462 23.3 Data Quality Issues in Electronic Medical Records 462 23.4 Procedure to Enhance Data Quality 464 23.4.1 Clinical Vocabularies 466 23.4.2 Ontologies 466 23.4.3 Potential Technical Challenges for EMR Data Quality 467 23.4.4 Data Warehousing 469 23.5 Form Design and On-entry Procedures 469 23.5.1 Data Capture 470 23.5.2 Data Input 470 23.5.3 Error Prevention 471 23.5.4 Physician-entered Data 471 23.6 Quality of Data Evaluation 472 23.7 Summary 475 References 475 Index 481