Tamaño de muestra

Bienvenida y Estructura del Curso de Investigación Clínica

Introducción al Curso

• El curso es parte de un diplomado en investigación clínica y salud pública, compuesto por cuatro cursos.

• Los cursos incluyen epidemiología aplicada, bioestadística aplicada, y liderazgo en salud pública.

• Se enfatiza que el cuarto curso se centra en cómo aplicar herramientas de investigación a políticas de salud pública.

Oportunidades Futuras

• Se menciona la posibilidad de ofrecer un segundo curso intensivo en línea en agosto, dependiendo del interés.

Cálculo del Tamaño de Muestra

Lectura del Caso

• Se inicia la discusión sobre el cálculo del tamaño de muestra con un caso práctico relacionado con la enfermedad de Lyme.

• La doctora Jennifer Hoffman planea un ensayo clínico controlado para comparar tratamientos antibióticos.

Decisiones Críticas

• La decisión entre realizar un estudio pequeño (menos de 50 pacientes) o uno más grande (200 pacientes) depende del cálculo adecuado del tamaño de muestra.

Importancia del Cálculo

Pregunta de Investigación

• Se recalca la importancia de formular correctamente la pregunta de investigación utilizando el método PICO: población, intervención, comparación y desenlace.

Validación Interna

• El cálculo del tamaño muestral es esencial para minimizar errores aleatorios que pueden afectar los resultados.

Elementos Clave en el Muestreo

Recordatorio sobre Hipótesis

• Se revisan los pasos para clarificar las pruebas de hipótesis y su relación con el muestreo.

Error Estadístico

• Es importante identificar tipos de error estadístico (tipo I y tipo II), así como parámetros que afectan el tamaño muestral.

Estudio sobre Café y Diabetes

Análisis Crítico

• Un estudio previo encontró una relación entre consumo de café y riesgo reducido de diabetes; se discuten hallazgos significativos según género.

Interpretación Estadística

• En hombres no se encontró asociación significativa; sin embargo, mujeres mostraron reducción significativa en riesgo al aumentar consumo.

Consideraciones Finales sobre Resultados

Implicaciones Prácticas

• Discusión sobre cómo diferentes tamaños muestrales pueden impactar los resultados e implicaciones clínicas.

Errores Comunes

• Se explican errores comunes como tipo I (falsos positivos), enfatizando la necesidad crítica del tamaño adecuado para evitar conclusiones erróneas.

Analyzing Group Differences in Diabetes Rates

Understanding Statistical Power and Sample Size

• The speaker discusses findings that suggest coffee drinkers had similar diabetes rates compared to non-drinkers, despite evidence indicating lower rates among coffee drinkers.

• Acknowledges limitations due to a small sample size which affected the statistical power of the study, making it difficult to draw definitive conclusions.

• Defines statistical power as the ability of a study to detect true differences between groups when they exist, emphasizing its importance in research validity.

Implications of False Negatives

• Highlights the risk of concluding groups are equal when they are not, potentially denying patients effective treatments based on erroneous assumptions about drug efficacy.

Error Types in Hypothesis Testing

• Introduces alpha (α), the probability of committing a Type I error (false positive), which should not exceed 5%.

• Clarifies that alpha is standardized while p-values vary based on study results; emphasizes understanding their distinction for accurate interpretation.

Exploring Type II Errors and Statistical Concepts

Defining Beta and Its Relationship with Power

• Discusses beta (β), representing the probability of committing a Type II error (false negative), highlighting its inverse relationship with statistical power.

Key Elements for Hypothesis Testing

• Outlines essential components for hypothesis testing: significance level, statistical power, anticipated treatment effect, and outcome variability.

Calculating Sample Size in Research Studies

Factors Influencing Sample Size Calculation

• Emphasizes that sample size calculations depend on chosen statistical tests and specific research questions being addressed.

Standardized Values for Significance and Power

• Reiterates standard values: significance level at 0.05 and power at 80%, noting these serve as benchmarks for most studies.

Addressing Multiple Comparisons in Genetic Studies

Challenges with Genetic Research

• Discusses how genetic studies often involve multiple comparisons leading to increased false positives; suggests stricter significance levels like α = 0.01 or α = 0.001 to mitigate this issue.

Estimating Variability from Literature

Importance of Literature Review

• Stresses using existing literature to estimate expected outcome variability rather than conducting small pilot studies which may yield unreliable data.

Fixed Sample Sizes and Their Implications

Limitations When Sample Size is Fixed

• Explains scenarios where researchers face fixed sample sizes due to logistical constraints or limited patient availability; highlights calculating statistical power under these conditions.

Ethical Considerations in Study Design

Ensuring Informative Results

• Warn against publishing results from studies with insufficient power as it can mislead stakeholders; stresses ethical responsibility in reporting findings accurately.

Introduction to Statistical Tests in Medical Research

Overview of Statistical Tests

• The speaker discusses various statistical tests relevant for medical research, including rank tests and chi-square tests.

• Emphasis on the chi-square test for comparing independent proportions, specifically mortality rates between heparin and placebo groups.

• The importance of determining sample size and statistical power before conducting studies is highlighted.

Sample Size Calculation

• Discussion on allocation ratio; typically set at 1:1 for equal distribution between treatment and control groups.

• The need to reference existing studies to estimate expected outcomes (e.g., mortality rates with heparin vs. control).

• Example provided where a small difference in mortality requires a larger sample size (200 participants).

Analyzing Numerical Outcomes

Comparison of Means

• Transition to analyzing numerical outcomes, focusing on recovery times from surgeries as an example.

• Importance of defining means and standard deviations for both experimental and control groups is discussed.

Adjusting Parameters

• Speaker notes potential issues with standard deviation estimates affecting sample size calculations.

• Acknowledgment that incorrect data inputs can lead to misleading results in sample size determination.

Additional Study Designs

Cross-sectional Studies

• Introduction to cross-sectional study designs as simpler alternatives for prevalence estimation.

Practical Application Example

• A hypothetical study assessing tuberculosis prevalence in Mexicali is presented, detailing population estimates and expected prevalence rates.

Sampling Methodology Considerations

Avoiding Bias in Sampling

• Discussion on the importance of random sampling methods to avoid bias when selecting participants for studies.

Conclusion of Module

• Final remarks emphasize the significance of proper design alongside accurate sample size calculations in research methodology.