Epidemiology and Biostatistics (eBook)

Dr. Kestenbaum is an Assistant Professor, Department of Medicine, Division of Nephrology, at the University of Washington.

Preface 5
Contents 6
Epidemiology 13
Measures of Disease Frequency 14
1.1 Importance of Measures of Disease Frequency 16
1.2 Prevalence 16
1.3 Incidence 17
1.4 Relationship Between Prevalence and Incidence 20
1.5 Stratification of Disease Frequency by Person, Place, and Time 20
1.5.1 Disease Frequency Measurements Stratified by Characteristics of Person 21
1.5.2 Disease Frequency Measurements Stratified by Characteristics of Place 21
1.5.3 Disease Frequency Measurements Stratified by Characteristics of Time 22
1.5.4 Disease Frequency Measurements To Complement Experimental Data 22
General Considerations in Clinical Research Design 24
2.1 Study Population 25
2.1.1 Definition of the Study Population 25
2.1.2 Choice of Study Population and Generalizability of Study Findings 26
2.1.3 Where to Find Information About the Study Population in a Clinical Research Article 27
2.2 Exposure and Outcome 28
2.2.1 Definition 28
2.2.2 Specifying and Measuring the Exposure and Outcome 29
2.2.3 Where to Find Exposure and Outcome Data in a Clinical Research Article 29
2.3 Interventional Versus Observational Study Designs 30
2.4 Inferring Causation from Association Studies 32
2.4.1 Importance of Distinguishing Causation from Association 32
2.4.2 Factors Favoring an Inference of Causation 33
Case Reports and Case Series 36
Cross-Sectional Studies 39
Cohort Studies 42
5.1 Overview of Cohort Study Design 42
5.2 Ascertainment of Study Data 44
5.2.1 Validity of Measurements 44
5.2.2 Timing of Measurements 45
5.2.3 Uniform Measurements 46
5.2.4 Retrospective Versus Prospective Data Collection 46
5.3 Advantages of Cohort Studies 47
5.3.1 Study of Multiple Outcomes 47
5.3.2 Ability to Discern Temporal Relationship Between Exposure and Outcome 47
5.4 Disadvantages of Cohort Studies 48
5.4.1 Confounding 48
5.4.2 Inability to Examine Diseases That Are Rare or Have a Long Latency 48
5.5 Cohort Studies for Evaluating Medication Use 49
5.6 Analysis of Data From Cohort Studies 50
5.6.1 Incidence Proportion Versus Incidence Rate 50
5.6.2 Relative Risk 51
5.6.3 Attributable Risk (also Called “Risk Difference” or “ Excess Risk”) 53
Case-Control Studies 54
6.1 Case-Control Study Design 56
6.1.1 Overview 56
6.1.2 Selection of Cases 57
6.1.3 Selection of Controls 58
6.2 Advantages of Case-Control Studies 60
6.2.1 Case Control Studies Can Be Ideal for the Study of Rare Diseases or Those with a Long Latency 60
6.2.2 Case-Control Studies Allow for the Study of Multiple Exposures 60
6.3 Disadvantages of Case-Control Studies 61
6.3.1 Observational Study Design 61
6.3.2 Recall Bias 61
6.3.3 Case Control Studies only Provide Information Regarding the Relative Risk ( Odds) of Disease 62
6.4 Analysis of Case-Control Data 62
6.4.1 Theory of the Odds Ratio 62
6.4.2 Practical Calculation of the Odds Ratio 64
6.4.3 Odds Ratios and Relative Risk 64
6.4.4 Case-Control Studies Cannot Estimate the Actual Incidence of a Disease or Outcome 65
Randomized Trials 67
7.1 Rationale for Randomized Trials 67
7.1.1 Kidney Transplant and Mortality 68
7.1.2 Angioplasty versus Fibrinolysis for Patients with Acute Myocardial Infarction 68
7.1.3 Equipoise 69
7.2 Phases of Drug Development 69
7.2.1 Phase I Studies 70
7.2.2 Phase II Studies 70
7.2.3 Phase III/IV Studies 70
7.3 Conduct of Randomized Trials 70
7.3.1 Comparison Group 70
7.3.2 Placebo 71
7.3.3 Block Randomization 72
7.3.4 Biological Versus Clinical Endpoints 73
7.4 Limitations of Randomized Controlled Trials 73
7.4.1 Generalizability of the Study Population 73
7.4.2 Generalizability of the Study Environment 74
7.4.3 Limited Question 75
7.4.4 Limited Clinical Applicability 75
7.4.5 Randomized Design Accounts only for Confounding 76
7.5 Analysis of Randomized Controlled Trial Data 76
7.5.1 Measures of Effect 76
7.5.2 Numbers Needed to Treat/Harm 77
7.5.3 Measures of Effect in Journal Articles 77
7.5.4 Intention-to Treat-Analysis 78
7.5.5 Subgroup Analyses 79
Misclassification 82
8.1 Definition of Misclassification 82
8.2 Nondifferential Misclassification 83
8.2.1 Example of Nondifferential Misclassification of the Exposure 83
8.2.2 Definition and Impact of Nondifferential Misclassification of the Exposure 85
8.2.3 Nondifferential Misclassification of the Outcome 88
8.2.4 Definition and Impact of Nondifferential Misclassification of the Outcome 88
8.3 Differential Misclassification 91
8.4 Assessment of Misclassification in Clinical Research Articles 96
Introduction to Confounding 97
9.1 Confounding and the Interpretation of Clinical Data 97
9.2 Formal Evaluation of a Potential Confounding Factor 100
9.2.1 Evaluation of a Confounder: Association with Exposure 101
9.2.2 Evaluation of a Confounder: Association with Outcome 101
9.2.3 Evaluation of a Confounder: Not in the Causal Pathway of Association 102
9.2.4 Other Examples of Factors That Reside on the Causal Pathway of Association 104
9.3 Scientifically Meaningful Versus Statistical Associations 104
9.4 Evaluation of a Confounder in Clinical Research Articles 105
9.5 Confounding-by-Indication 106
Methods to Control for Confounding 107
10.1 Restriction 108
10.1.1 Method of Restriction 108
10.1.2 Pros and Cons of Restriction as a Means to Control for Confounding 108
10.1.3 Restriction to Control for Confounding-by-Indication 109
10.2 Stratification 109
10.2.1 Method of Stratification 109
10.2.2 Pros and Cons of Stratification as a Means to Control for Confounding 111
10.2.3 Stratum-Specific Associations 111
10.3 Matching 112
10.3.1 Method of Matching 112
10.3.2 Pros and Cons of Matching as a Means to Control Confounding 113
10.4 Regression 114
10.5 Randomization 114
10.6 Interpreting Study Results After Adjustment for Confounding 115
10.7 Unadjusted Versus Adjusted Associations: Confounding 115
10.8 Confounding: An Advanced Example 116
Effect Modification 118
11.1 Concept of Effect Modification 118
11.2 Synergy Between Exposure Variables 119
11.3 Effect Modification Versus Confounding 120
11.4 Evaluation of Effect Modification 121
11.4.1 Epidemiologic Evaluation of Effect Modification 121
11.4.2 Statistical Evaluation of Effect Modification 121
11.5 Effect Modification in Clinical Research Articles 122
11.6 Effect Modification on the Relative and Absolute Scales 123
Screening 126
12.1 General Principles of Screening 127
12.2 Qualities of Diseases Appropriate for Screening 127
12.2.1 The Disease should be Important in the Screened Population 127
12.2.2 Early Recognition and Treatment of the Disease Should Prevent Clinical Outcomes 128
12.2.3 The Disease Should have a Preclinical Phase 128
12.3 Qualities of Screening Tests 128
12.3.1 General Qualities 128
12.3.2 Reliability and Validity 128
12.4 Validity of Screening Tests 129
12.4.1 Sensitivity and Specificity 129
12.4.2 Positive and Negative Predictive Value 130
12.4.3 Screening Tests with Continuous Values 134
12.5 Reliability of Screening Tests 137
12.5.1 Variation in Measurement Tools and Within an Individual 137
12.5.2 Measures of Reliability 138
12.6 Types of Bias in Screening Studies 139
12.6.1 Referral Bias 139
12.6.2 Lead Time Bias 140
12.6.3 Length Bias Sampling 141
12.6.4 Overdiagnosis Bias 142
12.7 Association versus Prediction 142
Diagnostic Testing 144
13.1 General Considerations in Medical Testing 144
13.2 Likelihood Ratios 148
Biostatistics 156
Summary Measures in Statisitics 157
14.1 Types of Variables 157
14.2 Univariate Statistics 158
14.2.1 Histograms 158
14.2.2 Measures of Location and Spread 160
14.2.3 Quantiles 162
14.2.4 Univariate Statistics for Binary Data 163
14.3 Bivariate Statistics 163
14.3.1 Tabulation Across Categories 163
14.3.2 Correlation 164
14.3.3 Quantile–Continuous Variable Plots 166
Introduction to Statistical Inference 167
15.1 Definition of a Population, Sample, and random Sample 167
15.2 Statistical Inference 168
15.3 Generalizability 169
15.4 Confidence Intervals 169
15.5 P-values 172
15.6 Confidence Intervals and p-values in Clinical Research 173
Hypothesis Testing 175
16.1 Study Hypothesis and Null Hypothesis 176
16.2 Distribution of Sampling Means 177
16.3 Properties of the Distribution of Sampling Means 178
16.3.1 Normal (Bell-Shaped) Distribution for Reasonably Large Sample Sizes 178
16.3.2 Mean Equal to the Population Mean 179
16.3.3 Spread of the Distribution Related to Population Variation and Sample Size 179
16.3.4 Distribution of Sampling Means: Summary 181
16.4 Conducting the Experiment 181
Interpreting Hypothesis Tests 185
17.1 Common Tests of Hypothesis in Clinical Research 185
17.1.1 T-Tests 185
17.1.2 Chi-Square Tests 186
17.1.3 ANOVA Tests 186
17.2 An Imperfect System 187
17.2.1 Type I Errors 187
17.2.2 Type II Errors 188
17.2.3 Power 188
Linear Regression 192
18.1 Describing the Association Between Two Variables 192
18.2 Univariate Linear Regression 195
18.2.1 The Linear Regression Equation 195
18.2.2 Residuals and the Sum of Squares 196
18.2.3 Absolute Versus Relative Fit 197
18.3 Interpreting Results from Univariate Regression Equations 198
18.3.1 Interpreting Continuous Covariates 198
18.3.2 Interpreting Binary Covariates 198
18.4 Special Considerations 200
18.4.1 Influential Points 200
18.4.2 Nonlinear Associations 201
18.4.3 Extrapolating the Regression Equation Beyond the Study Data 203
18.5 Multiple Linear Regression 203
18.5.1 Definition of the Multivariate Model 203
18.5.2 Interpreting Results from the Multiple Regression Model 204
18.6 Confounding and Effect Modification in Regression Models 207
18.6.1 Confounding 207
18.6.2 Effect Modification 208
Non-Linear Regression 211
19.1 Regression for Ratios 211
19.2 Logistic Regression 213
19.3 Application of Logistic Regression Models 215
Survival Analysis 217
20.1 Limitations of Incidence Measures for Evaluating Risk 217
20.1.1 Incidence Measures: Oversimplification of Study Results Over time 218
20.1.2 Incidence Measures: Crude Handling of Participant Dropout 218
20.2 Survival Data 219
20.3 Statistical Testing of Survival Data 221
20.4 Definitions of Events and Censoring 222
20.5 Kaplan–Meier Estimation 223
20.5.1 Kaplan–Meier Estimation of S(t) 223
20.5.2 Kaplan–Meier Estimation of S(t) with Censored Data 224
20.6 Cox’s Proportional Hazards Model 226
20.6.1 Description of the Proportional Hazards Model 226
20.6.2 Interpreting Survival Data and the Proportional Hazards Model 229
20.6.3 Survival Versus Hazard Ratio Data 230
References 231
Author Index 234
Subject Index 237