Preface 

xiii  

Biostatistics and Clinical Practice 


1  (10) 

The Changing Medical Environment 


1  (3) 

What Do Statistical Procedures Tell You? 


4  (2) 

Why Not Depend on the Journals? 


6  (3) 

Why Has the Problem Persisted? 


9  (2) 


11  (21) 


13  (1) 


14  (1) 


15  (1) 


16  (5) 

How to Make Estimates from a Limited Sample 


21  (1) 

How Good Are These Estimates? 


22  (7) 


29  (3) 

How to Test for Differences between Groups 


32  (33) 


32  (5) 

Two Different Estimates of the Population Variance 


37  (2) 


39  (8) 


47  (18) 

Relationship of Drug Prescribing to Length of Hospital Stay 


48  (4) 

Halothane versus Morphine for OpenHeart Anesthesia 


52  (5) 

Menstrual Dysfunction in Distance Runners 


57  (8) 

The Special Case of Two Groups: The t Test 


65  (43) 


67  (2) 

The Standard Deviation of a Difference or a Sum 


69  (3) 

Use of t to Test Hypotheses about Two Groups 


72  (7) 

What If the Two Samples Are Not the Same Size? 


79  (2) 


81  (3) 

Relationship of Drug Prescribing to Length of Hospital Stay 


82  (1) 

Halothane versus Morphine for OpenHeart Surgery 


82  (2) 

The t Test Is an Analysis of Variance 


84  (2) 

Common Errors in the Use of the t Test and How to Compensate for Them 


86  (3) 

How to Use t Tests to Isolate Differences between Groups in Analysis of Variance 


89  (3) 

More on Menstruation and Jogging 


91  (1) 

A Better Approach to Multiple Comparison Testing: The StudentNewmanKeuls Test 


92  (6) 

Still More on Menstruation and Jogging 


94  (3) 


97  (1) 

Multiple Comparisons against a Single Control 


98  (5) 


98  (1) 


99  (4) 


103  (5) 

How to Analyze Rates and Proportions 


108  (43) 


109  (4) 

Estimating Proportions from Samples 


113  (6) 

Hypothesis Tests for Proportions 


119  (8) 

The Yates Correction for Continuity 


121  (1) 

Mortality Associated with Anesthesia for OpenHeart Surgery with Halothane or Morphine 


122  (2) 

Prevention of Thrombosis in People Receiving Hemodialysis 


124  (3) 

Another Approach to Testing Nominal Data: Analysis of Contingency Tables 


127  (7) 

The ChiSquare Test Statistic 


130  (4) 

ChiSquare Applications to Experiments with More Than Two Treatments or Outcomes 


134  (6) 

Subdividing Contingency Tables 


137  (3) 


140  (11) 

What Does "Not Significant" Really Mean? 


151  (33) 


152  (4) 


156  (2) 

What Determines a Test's Power? 


158  (13) 

The Size of the Type I Error α 


158  (3) 

The Size of the Treatment Effect 


161  (2) 

The Population Variability 


163  (2) 

Bigger Samples Mean More Powerful Tests 


165  (2) 

What Determines Power? A Summary 


167  (3) 

Another Look at Halothane versus Morphine for OpenHeart Surgery 


170  (1) 

Power and Sample Size Computations for Analysis of Variance 


171  (4) 

Power, Menstruation, and Running 


174  (1) 

Power and Sample Size for Contingency Tables 


175  (2) 

Physicians, Perspiration, and Power 


176  (1) 

Practical Problems in Using Power 


177  (1) 

What Difference Does It Make? 


178  (6) 


184  (29) 

The Size of the Treatment Effect Measured as the Difference of Two Means 


185  (3) 


188  (1) 


189  (3) 

What Does "Confidence" Mean? 


192  (2) 

Confidence Intervals Can Be Used to Test Hypotheses 


194  (3) 

Confidence Interval for the Population Mean 


197  (4) 

The Size of the Treatment Effect Measured as the Difference of Two Rates or Proportions 


198  (1) 

Difference in Mortality Associated with Anesthesia for OpenHeart Surgery 


199  (1) 

Difference in Thrombosis with Aspirin in People Receiving Hemodialysis 


200  (1) 

How Negative Is a "Negative" Clinical Trial? 


201  (1) 

Confidence Interval for Rates and Proportions 


201  (6) 

The Fraction of Articles with Statistical Erros 


203  (1) 

Exact Confidence Intervals for Rates and Proportions 


204  (3) 

Confidence Interval for the Entire Population 


207  (6) 


213  (69) 


214  (5) 

The Population Parameters 


217  (2) 

How to Estimate the Trend from a Sample 


219  (19) 

The Best Straight Line through the Data 


219  (7) 

Variability about the Regression Line 


226  (2) 

Standard Errors of the Regression Coefficients 


228  (4) 

How Convincing Is the Trend? 


232  (2) 

Confidence Interval for the Regression Line 


234  (1) 

Confidence Interval for an Observation 


235  (3) 

How to Compare Two Regression Lines 


238  (7) 

Overall Test for Coincidence of Two Regression Lines 


239  (2) 

Relationship between Weakness and Muscle Wasting in Rheumatoid Arthritis 


241  (4) 

Correlation and Correlation Coefficients 


245  (11) 

The Pearson ProductMoment Correlation Coefficient 


247  (2) 

The Relationship between Regression and Correlation 


249  (3) 

How to Test Hypotheses about Correlation Coefficients 


252  (1) 

Dietary Fat and Breast Cancer 


253  (3) 

The Spearman Rank Correlation Coefficient 


256  (6) 

Variation among Interns in Use of Laboratory Tests: Relation to Quality of Care 


259  (3) 

Power and Sample Size in Regression and Correlation 


262  (4) 

Comparing Two Different Measurements of the Same Thing: The BlandAltman Method 


266  (5) 

Assessing Mitral Regurgitation with Echocardiography 


267  (4) 


271  (11) 

Experiments When Each Subject Receives More than One Treatment 


282  (41) 

Experiments When Subjects Are Observed before and after a Single Treatment: The Paired t Test 


283  (8) 

Cigarette Smoking and Platelet Function 


286  (5) 

Another Approach to Analysis of Variance 


291  (11) 


292  (7) 

Accounting for All the Variability in the Observation 


299  (3) 

Experiments When Subjects Are Observed after many Treatments: RepeatedMeasures Analysis of Variance 


302  (12) 

Oral Hydralazine Therapy for Pulmonary Hypertension 


308  (5) 

How to Isolate Differences in RepeatedMeasures Analysis of Variance 


313  (1) 

Power in RepeatedMeasures Analysis of Variance 


314  (1) 

Experiments When Outcomes Are Measured on a Nominal Scale: McNemar's Test 


314  (9) 

Skin Reactivity in People with Cancer 


314  (9) 

Alternatives to Analysis of Variance and the t Test Based on Ranks 


323  (50) 

How to Choose between Parametric and Nonparametric Methods 


324  (3) 

Two Different Samples: The MannWhitney RankSum Test 


327  (11) 

The Leboyer Approach to Childbirth 


333  (5) 

Each Subject Observed before and after One Treatment: The Wilcoxon SignedRank Test 


338  (8) 

Cigarette Smoking and Platelet Function 


345  (1) 

Experiments with Three or More Groups When Each Group Contains Different Individuals: The KruskalWallis Statistic 


346  (9) 

Elimination of Caffeine Impaired by Oral Contraceptives 


349  (1) 

Nonparametric Multiple Comparisons 


350  (3) 


353  (2) 

Experiments in Which Each Subject Receives More than One Treatment: The Friedman Test 


355  (11) 

Oral Hydralazine Therapy for Primary Pulmonary Hypertension 


360  (1) 

Multiple Comparisons after Friedman's Test 


361  (1) 

Effect of Secondhand Smoke on Angina Pectoris 


362  (4) 


366  (7) 

How to Analyze Survival Data 


373  (30) 


374  (3) 

Estimating the Survival Curve 


377  (10) 


384  (1) 

Standard Errors and Confidence Limits for the Survival Curve 


384  (3) 

Comparing Two Survival Curves 


387  (9) 

Bone Marrow Transplantation to Treat Adult Leukemia 


389  (7) 

The Yates Correction for the LogRank Test 


396  (1) 


396  (2) 


398  (1) 


399  (4) 

What Do the Data Really Show? 


403  (22) 


404  (2) 


406  (8) 

Internal Mammary Artery Ligation to Treat Angina Pectoris 


407  (1) 

The Portacaval Shunt to Treat Cirrhosis of the Liver 


408  (3) 

Is Randomization of People Ethical? 


411  (2) 

Is a Randomized Controlled Trial Always Necessary? 


413  (1) 

Does Randomization Ensure Correct Conclusions? 


414  (5) 

Problems with the Population 


419  (2) 

How You Can Improve Things 


421  (4) 
Appendixes 

425  (30) 


425  (6) 


431  (9) 


440  (15) 
Index 

455  