Material and Methods: reporting statistical
anlysis
Why is it important to describe statistical methods in detail in
biomedical research?
• To verify the reported results
• Transparency
• Reproducibility of the findings
Where is it placed
• Placement in Methods Section:
Detailed statistical analysis is typically placed at the
end of the Methods section.
Avoid repeating detailed analysis in the Results
section.
What is the principle behind it
• “Describe statistical methods with enough
detail to enable a knowledgeable reader with
access to the original data to judge its
appropriateness for the study and to verify the
reported results” (www.icmje.org)
International Committee of Medical Journal Editors. Preparing a Manuscript for Submissio to a Medical
Journal. 2020/4/10]; Available from: http://www.icmje.org/recommendations/browse/ manuscript-
preparation/preparing-for-submission.html.
Importance of descriptive statistics
• Descriptive statistics provide the foundational data
necessary to understand and replicate the analysis.
• They allow researchers to verify calculations and
support the transparency of reported findings.
• Numerators and Denominators: Essential for
interpreting percentages (e.g., risk ratios, odds ratios,
or hazard ratios).
• Group Comparisons: Clearly state the sample sizes for
each group to ensure reproducibility and
comparability.
Reporting Guidelines
Randomized trial CONSORT
Observational study STROBE
Systematic review PRISMA
Study protocol SPIRIT, PRISMA-P
Prediction model study TRIPOD
Case report CARE
Clinical practice guideline AGREE, RIGHT
Qualitative research SRQR, COREQ
Animal preclinical study ARRIVE
Quality improvement study SQUIRE
Economic evaluation CHEERS
Essential Components of Methods Section
• Study Design:
Objectives, patient population, and inclusion/exclusion
criteria.
For clinical trials, include parameters such as Type I
error, study power, and primary endpoint effect size.
• Statistical Analysis Plan:
Pre-determined before analysis begins.
Data Collection methods
Specify significance level (e.g., p < 0.05) and any
adjustments for multiple comparisons.
Study Area Description
• Description of the hospital or lab (e.g., type, size,
specialty, and services provided).
• Characteristics of the patient population (e.g.,
demographics, common diagnoses, or conditions
relevant to the study).
• Sample source (e.g., blood, urine, imaging data)
and criteria for inclusion/exclusion.
• Study timeframe and location (e.g., data collected
from January to June 2024 at XYZ Diagnostic Lab).
Variables in Statistical Analysis
• Types of Variables:
– Independent: Intervention (e.g., drug dosage levels).
– Dependent: Outcome measured (e.g., systolic blood
pressure reduction).
– Extraneous: Patient's age, gender, or pre-existing
conditions.
• Example:
– "We used logistic regression to assess the relationship
between smoking (independent variable) and chronic
cough (dependent variable), adjusting for age and BMI."
Statistical Tests and Assumptions
• Common Tests:
– Independent t-test: Compare two groups (e.g.,
treated vs. placebo).
– ANOVA: Compare multiple groups (e.g., dosage
levels).
• Assumptions:
– Normality tested with Shapiro-Wilk test.
– Homogeneity of variances checked via Levene’s
test.
Handling Confounders
• Adjustment:
– Adjust for confounders such as BMI and smoking
using multivariable regression.
• Study-Specific Criteria:
– Example: Hypertension defined as systolic BP ≥
140 mmHg.
Handling Missing Data
Methods to Specify:
Imputation techniques.
Exclusion criteria and reasons for exclusions.
• Example:
– "Missing data on smoking (n=86) were excluded,
leaving a final cohort of 702 patients."
• Slide 8: Advanced Analytical Techniques
• Numerical Methods:
– Normalize skewed data using logarithmic
transformation.
• Graphical Methods:
– Box plots for visualizing distribution.
– Scatterplots for examining correlations.
• Post hoc Analysis:
• Tukey’s HSD for pairwise comparisons
Reporting Confidence and Significance Levels
• Confidence Intervals:
– Example: "The drug reduced systolic BP by 8
mmHg (95% CI: 5–11 mmHg, P=0.03)."
• Significance Thresholds:
– Results considered significant if p < 0.05.
• Slide 10: Avoiding Misinterpretations
• P Values:
– Do not state: "The drug was ineffective" (P ≥ 0.05).
– Instead: "No evidence of a drug effect was
observed."
• Trends:
– Avoid calling P > 0.05 a "trend."
• Slide 11: Statistical Software
• Transparency in Tools:
– Report software and version used.
– Example: "Analyses were conducted using SPSS
v28 and R v4.2.1."
• Collaboration:
– Include details of statistical review by
biostatisticians.
• Slide 12: Reporting Cohort Selection
• Details to Include:
– Dates (e.g., diagnosed March 2013 to December
2017).
– Inclusion criteria: Newly diagnosed lung cancer.
– Exclusions: Prior surgery (n=43), missing data
(n=86).
– Final cohort size: 702 patients.
• Slide 13: Reporting Study Questions
• Clarity:
– List each primary question and the associated
statistical approach.
– Justify the choice of methods.
• Example:
– "Sample size of 120 per group calculated to detect
a 20% difference with 80% power."
• Slide 14: Reporting Results
• Numerators and Denominators:
– Example: "20 out of 50 patients (40%) showed
improvement."
• Effect Sizes:
– "Symptom scores reduced by 15 points (95% CI: 7–
13, P < 0.001)."
• Slide 15: Hypothesis Testing and CIs
• Correct Use:
– Use CIs for estimation, not hypothesis testing.
• Example:
– Event rates: 70% vs. 50%, P = 0.066, but OR 95% CI
excludes 1.
• Slide 17: Transparency Example
• Methods Clarity:
– "We used an independent samples t-test to
compare mean BP between Group A (n=50) and
Group B (n=48), significance set at 0.05."
• CI Reporting:
– "The odds ratio was 2.5 (95% CI: 1.8–3.5)."
• Slide 18: Clinical Relevance vs. Statistical
Significance
• Example:
– "A marker may predict outcomes statistically but
may not warrant clinical decision-making."
• Slide 19: Bayesian Analyses
• Details:
– Provide sufficient descriptions in an appendix.
– Allow interested readers to replicate the methods.
• Slide 20: Variable Selection Techniques
• Steps to Include:
– Initial variables, thresholds, and criteria.
– Example: "Stepwise selection with p < 0.05
criteria."
• Slide 21: Good Practices in Reporting
• Pre-Analysis Planning:
– Document decisions before starting analysis.
• Example:
– "Adjusted for age, gender, and BMI in
multivariable regression."
• Slide 22: Sample Size Calculations
• Clinical Trials:
– "120 participants per group calculated to detect a
20% difference with 80% power."
• Confidence:
– Justify assumptions for sample size estimates.
• Slide 23: Reporting of Interim Analyses
• Details:
– Planned analyses, stopping rules, and
adjustments.
• Example:
– "Interim analysis at 50% accrual for safety
outcomes."
• Slide 24: Reporting Limitations
• Acknowledge:
– Potential biases and limitations in methods.
– Example: "Missing data on 10% of participants
may bias results."
Study Description
• Type of facility: Tertiary care hospital specializing in
cardiology.
• Location: Urban hospital in Springfield with 500-bed capacity.
• Key services: Advanced diagnostic imaging and laboratory
facilities.
Patient Population:
• Demographics: Adults aged 30-65 years undergoing
hypertension screening.
• Inclusion criteria: Patients referred for echocardiography.
• Exclusion criteria: Patients with incomplete clinical data.
Study Description
Sample Collection and Handling:
• Source: Blood samples drawn in fasting state
and processed within 2 hours.
• Imaging data: Collected using standardized
protocols.
Workflow Overview:
• Include a schematic diagram of the patient flow
and sample processing steps to enhance clarity.
Sampling Design
• Structured approach:
• Controls used.
• Variables measured.
• Replicates.
• Data format:
• Quantitative measures (e.g., counts,
percentages).
Data Collection Protocol
• Logical sequence:
• Pre-experiment handling and care.
• Procedures followed.
• Quantitative details:
• Amounts, durations, timings.
Study Design Considerations
• - Hypotheses tested.
• - Controls used.
• - Variables measured.
• - Replication details.
• - Subheadings for multiple studies.
Example - Study Design
• - **Hypothesis:** Impact of nutrient levels on
plant growth.
• - **Variables:** Plant height, leaf size.
• - **Controls:** Consistent water supply.
• - **Replication:** 5 replicates per group.
Data Analysis Overview
• Summarize how data were processed:
• Descriptive statistics (e.g., mean, SD).
• Statistical analyses (e.g., t-tests, ANOVA).
• Use of subheadings to group analyses.
Statistical Software
• - Common tools:
• - SPSS, R, Python, etc.
• - Include specific version used.
• - Example:
• - "Data were analyzed using R version 4.2.2."
Data Summarization
• - Reporting measures:
• - Means ± SD or SEM.
• - Percentages or proportions.
• - Example:
• - "Average height was reported as mean ±
SD."
Data Transformations
• - Examples:
• - Log transformation for normality.
• - Square root transformation for variance
stabilization.
Statistical Tests
• - Examples:
• - Paired t-test: Compare pre/post-treatment.
• - One-way ANOVA: Compare group means.
Significance Threshold
• - Standard significance level:
• - Alpha = 0.05.
• - Example:
• - "Results with p < 0.05 were considered
statistically significant."
Common Challenges
• - Wordiness:
• - Combine related actions into concise
sentences.
• - Ambiguity:
• - Use clear designators (e.g., Group A vs.
Group B).

Methods 17-01-2025.pptx statistical methods

  • 1.
    Material and Methods:reporting statistical anlysis
  • 2.
    Why is itimportant to describe statistical methods in detail in biomedical research? • To verify the reported results • Transparency • Reproducibility of the findings
  • 3.
    Where is itplaced • Placement in Methods Section: Detailed statistical analysis is typically placed at the end of the Methods section. Avoid repeating detailed analysis in the Results section.
  • 4.
    What is theprinciple behind it • “Describe statistical methods with enough detail to enable a knowledgeable reader with access to the original data to judge its appropriateness for the study and to verify the reported results” (www.icmje.org) International Committee of Medical Journal Editors. Preparing a Manuscript for Submissio to a Medical Journal. 2020/4/10]; Available from: http://www.icmje.org/recommendations/browse/ manuscript- preparation/preparing-for-submission.html.
  • 5.
    Importance of descriptivestatistics • Descriptive statistics provide the foundational data necessary to understand and replicate the analysis. • They allow researchers to verify calculations and support the transparency of reported findings. • Numerators and Denominators: Essential for interpreting percentages (e.g., risk ratios, odds ratios, or hazard ratios). • Group Comparisons: Clearly state the sample sizes for each group to ensure reproducibility and comparability.
  • 6.
    Reporting Guidelines Randomized trialCONSORT Observational study STROBE Systematic review PRISMA Study protocol SPIRIT, PRISMA-P Prediction model study TRIPOD Case report CARE Clinical practice guideline AGREE, RIGHT Qualitative research SRQR, COREQ Animal preclinical study ARRIVE Quality improvement study SQUIRE Economic evaluation CHEERS
  • 7.
    Essential Components ofMethods Section • Study Design: Objectives, patient population, and inclusion/exclusion criteria. For clinical trials, include parameters such as Type I error, study power, and primary endpoint effect size. • Statistical Analysis Plan: Pre-determined before analysis begins. Data Collection methods Specify significance level (e.g., p < 0.05) and any adjustments for multiple comparisons.
  • 8.
    Study Area Description •Description of the hospital or lab (e.g., type, size, specialty, and services provided). • Characteristics of the patient population (e.g., demographics, common diagnoses, or conditions relevant to the study). • Sample source (e.g., blood, urine, imaging data) and criteria for inclusion/exclusion. • Study timeframe and location (e.g., data collected from January to June 2024 at XYZ Diagnostic Lab).
  • 9.
    Variables in StatisticalAnalysis • Types of Variables: – Independent: Intervention (e.g., drug dosage levels). – Dependent: Outcome measured (e.g., systolic blood pressure reduction). – Extraneous: Patient's age, gender, or pre-existing conditions. • Example: – "We used logistic regression to assess the relationship between smoking (independent variable) and chronic cough (dependent variable), adjusting for age and BMI."
  • 10.
    Statistical Tests andAssumptions • Common Tests: – Independent t-test: Compare two groups (e.g., treated vs. placebo). – ANOVA: Compare multiple groups (e.g., dosage levels). • Assumptions: – Normality tested with Shapiro-Wilk test. – Homogeneity of variances checked via Levene’s test.
  • 11.
    Handling Confounders • Adjustment: –Adjust for confounders such as BMI and smoking using multivariable regression. • Study-Specific Criteria: – Example: Hypertension defined as systolic BP ≥ 140 mmHg.
  • 12.
    Handling Missing Data Methodsto Specify: Imputation techniques. Exclusion criteria and reasons for exclusions. • Example: – "Missing data on smoking (n=86) were excluded, leaving a final cohort of 702 patients."
  • 13.
    • Slide 8:Advanced Analytical Techniques • Numerical Methods: – Normalize skewed data using logarithmic transformation. • Graphical Methods: – Box plots for visualizing distribution. – Scatterplots for examining correlations. • Post hoc Analysis: • Tukey’s HSD for pairwise comparisons
  • 14.
    Reporting Confidence andSignificance Levels • Confidence Intervals: – Example: "The drug reduced systolic BP by 8 mmHg (95% CI: 5–11 mmHg, P=0.03)." • Significance Thresholds: – Results considered significant if p < 0.05.
  • 15.
    • Slide 10:Avoiding Misinterpretations • P Values: – Do not state: "The drug was ineffective" (P ≥ 0.05). – Instead: "No evidence of a drug effect was observed." • Trends: – Avoid calling P > 0.05 a "trend."
  • 16.
    • Slide 11:Statistical Software • Transparency in Tools: – Report software and version used. – Example: "Analyses were conducted using SPSS v28 and R v4.2.1." • Collaboration: – Include details of statistical review by biostatisticians.
  • 17.
    • Slide 12:Reporting Cohort Selection • Details to Include: – Dates (e.g., diagnosed March 2013 to December 2017). – Inclusion criteria: Newly diagnosed lung cancer. – Exclusions: Prior surgery (n=43), missing data (n=86). – Final cohort size: 702 patients.
  • 18.
    • Slide 13:Reporting Study Questions • Clarity: – List each primary question and the associated statistical approach. – Justify the choice of methods. • Example: – "Sample size of 120 per group calculated to detect a 20% difference with 80% power."
  • 19.
    • Slide 14:Reporting Results • Numerators and Denominators: – Example: "20 out of 50 patients (40%) showed improvement." • Effect Sizes: – "Symptom scores reduced by 15 points (95% CI: 7– 13, P < 0.001)."
  • 20.
    • Slide 15:Hypothesis Testing and CIs • Correct Use: – Use CIs for estimation, not hypothesis testing. • Example: – Event rates: 70% vs. 50%, P = 0.066, but OR 95% CI excludes 1.
  • 21.
    • Slide 17:Transparency Example • Methods Clarity: – "We used an independent samples t-test to compare mean BP between Group A (n=50) and Group B (n=48), significance set at 0.05." • CI Reporting: – "The odds ratio was 2.5 (95% CI: 1.8–3.5)."
  • 22.
    • Slide 18:Clinical Relevance vs. Statistical Significance • Example: – "A marker may predict outcomes statistically but may not warrant clinical decision-making."
  • 23.
    • Slide 19:Bayesian Analyses • Details: – Provide sufficient descriptions in an appendix. – Allow interested readers to replicate the methods.
  • 24.
    • Slide 20:Variable Selection Techniques • Steps to Include: – Initial variables, thresholds, and criteria. – Example: "Stepwise selection with p < 0.05 criteria."
  • 25.
    • Slide 21:Good Practices in Reporting • Pre-Analysis Planning: – Document decisions before starting analysis. • Example: – "Adjusted for age, gender, and BMI in multivariable regression."
  • 26.
    • Slide 22:Sample Size Calculations • Clinical Trials: – "120 participants per group calculated to detect a 20% difference with 80% power." • Confidence: – Justify assumptions for sample size estimates.
  • 27.
    • Slide 23:Reporting of Interim Analyses • Details: – Planned analyses, stopping rules, and adjustments. • Example: – "Interim analysis at 50% accrual for safety outcomes."
  • 28.
    • Slide 24:Reporting Limitations • Acknowledge: – Potential biases and limitations in methods. – Example: "Missing data on 10% of participants may bias results."
  • 29.
    Study Description • Typeof facility: Tertiary care hospital specializing in cardiology. • Location: Urban hospital in Springfield with 500-bed capacity. • Key services: Advanced diagnostic imaging and laboratory facilities. Patient Population: • Demographics: Adults aged 30-65 years undergoing hypertension screening. • Inclusion criteria: Patients referred for echocardiography. • Exclusion criteria: Patients with incomplete clinical data.
  • 30.
    Study Description Sample Collectionand Handling: • Source: Blood samples drawn in fasting state and processed within 2 hours. • Imaging data: Collected using standardized protocols. Workflow Overview: • Include a schematic diagram of the patient flow and sample processing steps to enhance clarity.
  • 31.
    Sampling Design • Structuredapproach: • Controls used. • Variables measured. • Replicates. • Data format: • Quantitative measures (e.g., counts, percentages).
  • 32.
    Data Collection Protocol •Logical sequence: • Pre-experiment handling and care. • Procedures followed. • Quantitative details: • Amounts, durations, timings.
  • 33.
    Study Design Considerations •- Hypotheses tested. • - Controls used. • - Variables measured. • - Replication details. • - Subheadings for multiple studies.
  • 34.
    Example - StudyDesign • - **Hypothesis:** Impact of nutrient levels on plant growth. • - **Variables:** Plant height, leaf size. • - **Controls:** Consistent water supply. • - **Replication:** 5 replicates per group.
  • 35.
    Data Analysis Overview •Summarize how data were processed: • Descriptive statistics (e.g., mean, SD). • Statistical analyses (e.g., t-tests, ANOVA). • Use of subheadings to group analyses.
  • 36.
    Statistical Software • -Common tools: • - SPSS, R, Python, etc. • - Include specific version used. • - Example: • - "Data were analyzed using R version 4.2.2."
  • 37.
    Data Summarization • -Reporting measures: • - Means ± SD or SEM. • - Percentages or proportions. • - Example: • - "Average height was reported as mean ± SD."
  • 38.
    Data Transformations • -Examples: • - Log transformation for normality. • - Square root transformation for variance stabilization.
  • 39.
    Statistical Tests • -Examples: • - Paired t-test: Compare pre/post-treatment. • - One-way ANOVA: Compare group means.
  • 40.
    Significance Threshold • -Standard significance level: • - Alpha = 0.05. • - Example: • - "Results with p < 0.05 were considered statistically significant."
  • 41.
    Common Challenges • -Wordiness: • - Combine related actions into concise sentences. • - Ambiguity: • - Use clear designators (e.g., Group A vs. Group B).