Biostatistics Manual for Health Research: A Practical Guide to Data Analysis

Front Cover
BIOSTATISTICS MANUAL FOR HEALTH RESEARCH
BIOSTATISTICS MANUAL FOR HEALTH RESEARCH
Copyright
Contents
About the authors
Preface
List of abbreviations
1 - Introduction to biostatistics
1. Background
2. What is biostatistics?
2.1 Why biostatistics rather than statistics?
2.2 Role of biostatistics
3. Statistical inference
4. Aim of the book
5. Two foundational concepts
5.1 Law of large numbers
5.2 Central limit theorem
6. Data and variables
6.1 Data
6.2 Qualitative data
6.2.1 Nominal data
6.2.2 Ordinal data
6.3 Quantitative data
6.3.1 Scale data: interval and ratio
6.3.2 Discrete and continuous
6.4 Cardinal and ordinal data
6.5 Variable
7. Measures of central tendency and dispersion
7.1 Introduction
7.2 Central tendency
7.3 Dispersion
References
2 - Data management and SPSS environment∗
1. Data management
1.1 Introduction
1.2 Best practices for data management
1.3 Data management plan
2. Data documentation sheet
2.1 Template for DDS
2.2 Coding the dataset
3. Data capture and cleaning
3.1 Data capture
3.2 Data checking and cleaning
4. SPSS environment
4.1 Introduction to the software
4.2 The first thing you will see
4.2.1 Data View
4.2.2 Toolbar icons from left to right (Fig. 2.7)
4.2.3 Variable View
5. Data entry and importing in SPSS
5.1 Data entry in SPSS
5.2 Data importing SPSS
6. Data transformation in SPSS
6.1 Compute variables
6.2 Recode into same variables and different variables
6.3 Find and replace
References
3 - Statistical tests of significance∗
1. Hypothesis testing
1.1 Null and alternate hypothesis
1.2 Directional versus nondirectional hypothesis
1.3 Steps in hypothesis testing
2. Statistical tests of significance
2.1 Parametric and nonparametric tests
2.2 Checking for assumptions of parametric tests
2.2.1 Normally distributed data
2.2.2 Homogeneity of variance/homoscedasticity in the sample
2.2.3 Independence of the variable
3. Choosing a statistical test
4. Levels of significance and P-values
5. Errors in health research
5.1 Sampling errors
5.2 Power and confidence interval
5.3 Reducing sampling errors
5.3.1 Which sampling error is worse?
5.4 Nonsampling errors
6. P-values and effect sizes
6.1 Basis of P-values and confidence intervals
6.2 ASA statement on P-values
6.3 Final word: best practices with P-values
References
4 - Parametric tests∗
1. Continuous outcomes
2. Parametric tests
2.1 Introduction
2.2 Assumptions of a parametric test
3. t-Tests: independent and paired
4. Independent t-test
4.1 Applying t-test
4.2 Interpretation of results of t-test
5. Paired t-test
5.1 Applying paired t-test
5.2 Interpretation of results of paired t-test
6. Parametric tests comparison with ﹥2 groups: analysis of variance
6.1 Applying one-way ANOVA
6.2 Interpretation of results of ANOVA
6.3 Post hoc ANOVA test
6.4 Selecting appropriate post hoc ANOVA test
6.5 Applying post hoc ANOVA test
6.6 Interpretation of results of post hoc ANOVA test
7. Repeated-measures ANOVA
7.1 Applying RMANOVA
7.2 Post hoc RMANOVA
7.3 Applying post hoc RMANOVA
7.4 Interpretation of results of post hoc RMANOVA test
8. ANOVA, ANCOVA, MANOVA, and MANCOVA
References
5 - Nonparametric tests∗
1. Nonparametric methods
2. Mann–Whitney U test
2.1 The assumptions for Mann–Whitney U test
2.2 Applying Mann–Whitney U test
2.3 Interpretation of results of Mann–Whitney U test
3. Wilcoxon signed-rank test
3.1 Assumptions of Wilcoxon signed-rank test
3.2 Applying Wilcoxon signed rank-sum test
3.3 Interpretation of results of Wilcoxon signed rank-sum test
4. Nonparametric tests comparison with ﹥2 groups: Kruskal–Wallis test
4.1 The assumptions for Kruskal–Wallis test
4.2 Applying Kruskal–Wallis test
4.3 Interpretation of results of Kruskal–Wallis test
5. Nonparametric tests comparison with ﹥2 related or repeated measures: Friedman test
5.1 Assumptions of Friedman test
5.2 Applying Friedman test
5.3 Interpretation of results of Friedman test
References
6 - Correlation∗
1. Continuous outcome and exposure
2. Correlation versus association
3. Pearson's correlation test
3.1 Applying Pearson's correlation
3.2 Results of Pearson's correlation
4. Spearman's correlation test
4.1 Applying Spearman's correlation to continuous data
4.2 Results of Spearman's correlation
4.3 Applying Spearman's correlation on ordinal data
4.4 Results of Spearman's correlation on ordinal data
5. Correlation versus concordance
6. Agreement: Kendall's τ, Kendall's W, and kappa
6.1 Kendall's tau (τ)
6.2 Kendall's coefficient of concordance W
6.3 Kappa test (κ)
7. Measuring concordance/agreement
7.1 Scenario 1: Nominal data
7.2 Scenario 2: Ordinal data with two raters
7.3 Scenario 3: Ordinal data with more than two raters
7.4 The real difference between the three tests
7.4.1 Case 1: Amar and Akbar (Table 6.3)
7.4.2 Case 2: Amar and Anthony (Table 6.4)
7.4.3 Case 3: Akbar and Anthony (Table 6.5)
7.4.4 Case 4: Amar is a standard program
7.4.5 Case 5. All three are nonstandard raters (Table 6.6)
References
7 - Categorical variables∗
1. Categorical variables
2. Independent exposure variables: chi-square test
2.1 The assumptions for chi-square test
2.2 Applying the chi-square (χ2) test
2.3 Interpretation of results of the chi-square (χ2) test
2.3.1 Guide to interpret effect sizes in chi-square test
2.4 Post hoc comparison of the chi-square test
3. Alternatives to chi-square test
3.1 Fisher exact test
3.1.1 The assumptions for Fisher exact test
3.1.2 Applying the Fisher exact test
3.2 Chi-square with continuity correction
3.3 Likelihood ratio test
3.4 Linear by linear association
3.5 Cochran–Mantel–Haenszel chi-square test
3.5.1 Applying the Cochran–Mantel–Haenszel chi-square test
3.5.2 Interpreting the results of the Cochran–Mantel–Haenszel chi-square test
4. Two related exposure variables: McNemar's test
4.1 The assumptions for McNemar's test
4.2 Applying McNemar's test
4.3 Interpretation of results of McNemar's test
5. More than two related exposure variables: Cochran's Q test
5.1 The assumptions for Cochran's Q test
5.2 Applying Cochran's Q test
5.3 Interpretation of results of the Cochran Q test
6. Analyzing the summary data
References
8 - Validity∗
1. Validity
1.1 Types of validity
2. Diagnostic test evaluation
2.1 Tests for diagnostic test evaluation
2.2 Sensitivity and specificity
2.2.1 Sensitivity
2.2.2 Specificity
2.2.3 Relative importance of sensitivity and specificity
2.3 Positive and negative predictive values
2.3.1 Concept of conditional probability
2.3.2 Probability of the disease and validity of the test
2.4 Positive and negative likelihood ratio
2.5 Diagnostic accuracy
3. Diagnostic test evaluation: calculations
3.1 Applying diagnostic test evaluation
3.2 Interpretation of results of diagnostic test evaluation
4. Combining screening tests
5. Continuous data and ROC curves
5.1 Understanding the ROC curve
5.1.1 Area under curve
5.1.2 Criteria for cut-off value
5.2 Applying the ROC curve
5.3 Interpretation of the results of the ROC curve
5.4 Comparing multiple ROC curves
References
9 - Reliability and agreement∗
1. Reliability and agreement
1.1 Validity and reliability
1.2 Reliability versus agreement
1.3 Types of reliability
2. Reliability methods for categorical variables
3. Cohen's kappa test
3.1 Assumption of Cohen's kappa test
3.2 Applying Cohen's kappa test
3.3 Interpretation of results of Cohen's kappa test
4. Weighted Cohen's kappa test
4.1 Assumption of weighted Cohen’s kappa test
4.2 Applying weighted Cohen’s kappa test
4.2.1 Applying weighted Cohen's kappa test in SPSS version 27 and above
4.2.2 Applying weighted Cohen's kappa test in SPSS version 26 and below
4.2.3 Calculating weights
4.2.4 Creating the weight variable
4.2.5 Applying weights
4.2.6 Applying Cohen's kappa test
4.3 Interpretation of results of weighted Cohen's kappa test
5. Fleiss kappa test
5.1 Assumptions of Fleiss kappa test
5.2 Applying Fleiss kappa test
5.3 Interpretation of results of Fleiss kappa test
6. Agreement and concordance: which test to use?
7. Reliability for continuous variables: intraclass correlation
7.1 Assumptions of ICC
7.1.1 Models
7.1.2 Type of agreement
7.2 Applying ICC
7.3 Interpretation of results of ICC
8. Cronbach's alpha
8.1 Applying Cronbach's alpha
8.2 Interpretation of results of Cronbach's alpha
References
10. Survival analysis∗
1. Time to event as a variable
2. Survival analysis
2.1 Applications in health research
2.2 Terminology
2.3 Types of survival analysis
2.3.1 Kaplan–Meier survival method
2.3.2 Life table analysis
2.3.3 Cox proportional hazards regression
2.3.4 Exponential and Weibull survival analysis
2.3.5 Log-rank test and frailty models
3. Kaplan–Meier survival method
3.1 Assumptions of the Kaplan–Meier survival method
3.2 Applying the Kaplan–Meier survival method
3.2.1 Datasheet for Kaplan–Meier survival
3.3 Interpretation of results of the Kaplan–Meier analysis
3.4 Writing the results of the Kaplan–Meier
4. Cox regression survival method
4.1 Assumptions of Cox regression
4.2 Applying Cox regression
4.2.1 Datasheet for Cox regression
4.3 Interpretation of results of the Cox regression analysis
4.4 Cox regression with a time-dependent covariate
4.4.1 Interpretation of results of the Cox regression with a time-dependent covariate
4.4 Writing the results of Cox regression
References
11 - Regression and multivariable analysis∗
1. Regression and multivariable analysis
1.1 Univariate, univariable and multivariable: terminologies
1.2 Dependence and interdependence methods
1.3 Other applications of regression
1.3.1 Hypothesis testing
1.3.2 Data reduction
1.3.3 Grouping and classification
1.3.4 Investigation of dependents among variables
1.3.5 Prediction and forecasting
2. Regression analysis
2.1 Approaching regression and multivariable analysis
3. Linear regression
3.1 Terminologies in linear regression
3.2 The assumptions for linear regression
4. Simple linear regression analysis
4.1 Applying the simple linear regression
4.2 Interpretation of results of the simple linear regression
4.3 Applying the robust linear regression model
4.4 Interpretation of results of the simple linear regression with robust standard errors
4.5 Bootstrapping in linear regression
4.6 Writing the results of the simple linear regression
5. Multiple linear regression analysis
5.1 Entering predictors into the model
5.1.1 Finalizing the list of predictors/explanatory variables
5.1.2 Method—order of variable entry in the model
5.2 Applying the multiple linear regression
5.3 Interpretation of results of the multiple linear regression
5.4 Writing the results of the multiple linear regression
6. Logistic regression analysis
6.1 The assumptions of logistic regression
7. Multiple logistic regression analysis
7.1 Entering predictors into the model
7.1.1 Finalizing the list of predictors/explanatory variables
7.1.2 Method—order of variable entry in the model
7.2 Applying the multiple logistic regression
7.3 Interpretation of results of the multiple logistic regression
7.4 Writing the results of the multiple logistic regression
8. Multivariable analysis
References
12 - Annexures
Annexure 1: Choice of statistical tests (Fig. 12.1).
Annexure 2: Notes on data used in the book
Annexure 3: Guidelines for statistical reporting in journals: SAMPL guidelines
3.1 Background
3.2 SAMPL guidelines for reporting statistical methods
3.2.1 Preliminary analyses
3.2.2 Primary analyses
3.2.3 Supplementary analyses
3.3 SAMPL guidelines for reporting statistical results
3.3.1 Reporting numbers and descriptive statistics
3.3.2 Reporting risk, rates, and ratios
3.3.3 Reporting hypothesis tests
3.3.4 Reporting association analyses
3.3.5 Reporting correlation analyses
3.3.6 Reporting regression analyses
3.3.7 Reporting analyses of variance or covariance
3.3.8 Reporting survival (time-to-event) analyses
3.3.9 Reporting Bayesian analyses
Annexure 4: Standards for reporting of diagnostic accuracy: STARD guidelines
Annexure 5: Guidelines for reporting reliability and agreement studies: GRRAS guidelines
5.1 Background
5.2 Salient features of the guidelines
Annexure 6: Proposed agenda for biostatistics for a health research workshop
References
Index
A
B
C
D
E
F
G
H
I
J
K
L
M
N
O
P
Q
R
S
T
U
V
W
Y
Z
Back Cover