Ceresoli M., Abu-Zidan F.M., Staudenmayer K.L., Catena F., Coccoli F. (eds.) Statistics and Research Methods for Acute Care and General Surgeons

Cham: Springer, 2023. — 177 p.The main aim of this book is to offer an easy tool to read a scientific article with greater awareness, to understand and evaluate it more thoroughly, and to better plan research. Today, in the era of evidence-based medicine, both research and daily patient-focused clinical practice are no longer possible without a thorough knowledge of the literature and its continuous updates.
Written by surgeons for surgeons, this practical book makes the basic concept of statistics and research methodology easy to understand and apply for young surgeons and researchers, students and residents.Preface: Why a Statistics Manual in the Series of “Hot Topics in Acute Care Surgery”?Designing Your Research
Study Typology: An OverviewThe Need for Evidence-Based Medicine
Research Studies
Primary and Secondary Research Studies
Primary Studies
Laboratorial Research
Clinical Studies
Clinical Observational Studies
Case Reports and Case Series
Clinical Experimental Studies
Epidemiological Research
Cross-Sectional Studies
Case–Control Studies
Cohort Studies
Ecological Studies
Secondary Studies
Narrative Reviews
Systematic Reviews and Meta-AnalysisDiagnostic Studies Made EasyNature of a Diagnostic Study
The Need for a Gold Standard
Components of Diagnostic Studies
Predictive Values
Prior Probability of the Disease (Prevalence)
The Likelihood Ratio (LR)
Receiver Operating Characteristics (ROC) Curves
Choosing a Cut-off Point: The Youden Index
Common Errors Encountered in Submitted Diagnostic Studies
Further Reading
Common Pitfalls in Research Design and Its ReportingUnclear Research Question
Lack of Planning (Failing to Plan Is Planning to Fail)
Using the Wrong Research Tool
Selecting the Wrong Population
Addressing the Missing Data
Correlation and Prediction
Statistical and Clinical Significance
Reporting of the DataBasic Statistical Analysis
Introduction to Statistical MethodThe Hypothesis
The Aim
The Errors
Type I Error
Type I Error Rate
Type II Error
Statistical Power
Type II Error Rate
Trade-Off between Type I and Type II Errors
Is a Type I or Type II Error Worse?
Sample Size Calculation
The P Value
Results and Interpretation
Bias
Selection Bias
Classification Bias
Confounding Bias
Other Types of BiasAnalyzing Continuous Variables: Descriptive Statistics, Dispersion and ComparisonQualitative Variables
Quantitative Variables
Discrete Variables
Continuous Variables
Describing Data
Data Distribution
Test for Normality Assessment
Descriptive Measures
Dispersion Measure
Graphical Representations
Data Comparison: It Is All About Probability
Paired Data vs Independent Data
Parametric vs Non-Parametric Statistics
Commonest Tests
Linear Correlation
Pearson Correlation
Pearson Coefficient Interpretation
Spearman’s Rank Correlation Coefficient
Further Reading
Analyzing Categorical Variable: Descriptive Statistics and ComparisonsConfidence Interval of Proportions
Absolute Risk Reduction and Number Needed-to-Treat
Relative Risk and Relative Risk Reduction
Odds Ratio
Chi-Squared Test and Fisher’s Exact Test
Matched Data
Chi-Squared Test for Trend
Standardized DifferencesAdvanced Statistics
Multivariate AnalysisStatistical Models
Different Types of Regressions and Multiple Regression
Example: The Dataset
Linear Regression Models
Building a Linear Regression Model: It All Comes Down to the Straight Line Equation
Interpreting the Linear Regression Model
Interpreting the Parameters of the Model
Interpreting % Confidence Intervals
Assumptions of Linear Regression
Multiple Regression
Choice of Predictors
Example: Inclusion of Predictors for Multivariable Analysis
Adjustment for Confounders
Logistic Regression Models
The Logistic Regression Model
Interpreting the Logistic Regression Model: An Example
Further Reading
Survival Analysis
Generalities About Time-to-Event Data
The Variables We Need to Make Analysis: Event and Time
The Survival Curve and Life Tables: The Kaplan–Meier Method
Comparing Survival Curves: The Log-Rank Test
A Regression Model to Assess the Association of Multiple Predictors with a Survival Outcome: The Cox “Proportional Hazards” Model
Meta-AnalysisThe Question
Systematic Review of the Literature
Meta-analysis Appropriateness: Study Inclusion
Study Quality Assessment and the Risk of Bias
Results: Effect Measure
Binary Outcomes/Dichotomous Data
Continuous Data (Also Scale Data or Counts of Events)
Results: The Forest Plot
Results: Heterogeneity
Interpretation of the Results
Sensitivity Analysis
Common Mistakes Encountered in Submitted Systematic Review Manuscripts
ConclusionsRandomized Trials and Case–Control Matching Techniques
Introduction to Randomized Trials
Ethical Concerns Are Also Related to RCTs
Placebo Effect
Hypothesis Testing and Sample Size Calculation
Reporting the Trials
Randomized Controlled Trials Designs and Techniques
Strengths of Randomized Trials
Limitation of Randomized Trials
Case–Control Studies and Case–Control Matching Techniques
Propensity Score and Inverse ProbabilityDifference-In-Difference Techniques and Causal InferenceMachine Learning Techniques
Machine Learning and Artificial Intelligence
Machine Learning Terminologies and Concepts
Algorithms, Models, Inputs, and Outputs
Dimensions
Overfitting vs Underfitting
Bias vs Variance Tradeoff
Model Flexibility
Feature Selection and Dimension Reduction
Performance Metrics
Training, Validation, and Test Sets
Cross-Validation
Evolution of a Family of Machine Learning Algorithms: From Linear Models to Deep Learning
Supervised Learning
Logistic and Linear Regression
Generalized Additive Models
Deep LearningStatistical Editor’s Practical Advice for Data AnalysisWhat Is the Objective of the Analysis?
What Is the Type of Data?
Are the Data Normally Distributed?
How Many Groups Are Compared?
What Is the Number of Subjects in Each Group?
Are the Compared Data Related or Unrelated?
Which Test to Use?Further Reading