Biostatistics - Study Design

Biostatistics - Study Design

Edited and Delivered by: Jitendra Belani, PhD, MS, RPh, CDCES, MBA
Created by: Jackie Wasynczuk, PharmD

Learning Objectives

  • At the end of this session students will be able to:

    • Evaluate the importance of a specific research question in clinical research and clinical decision making.

    • Update the known framework used to design research questions.

    • Summarize how to match the research question to a study design.

    • Explain the potential biases in key study designs.

Understanding Study Design

  • Answering Research Questions and Practicing Evidence-Based Medicine (EBM):

    • Clinical studies that are well designed provide valuable evidence for making patient care decisions.

    • When practicing EBM, clinicians integrate their clinical experience with the best available and most relevant research.

    • For a study to be considered the "best and most relevant," it should have a sound design and methodology.

  • Importance of Study Design in EBM:

    • For optimal patient care decisions, an understanding of study design and research methodology is essential.

    • Best Research Evidence, Clinical Expertise, and Patient Values work together in making informed decisions.

What is a Study Design?

  • Definition of Study Design:

    • A plan that allows researchers to collect data that can be used to evaluate a study hypothesis.

    • A plan that allows researchers to evaluate the relationship between an independent variable (Intervention) and one or more dependent variables (outcome).

What is a Research Question?

  • A researcher asks a very specific question and tests a specific hypothesis.

  • Broad questions are usually broken into smaller, testable hypotheses or questions.

  • Often termed as an objective or aim, referring to it as a question helps focus the hypothesis and how to find an answer.

  • PICO Format:

    • P (Population) - specific group of interest.

    • I (Intervention) - what is being tested.

    • C (Comparison) - what it is compared against.

    • O (Outcome) - the outcome measured.

How to Focus Your Question?

  • Conduct a brief literature search for previous evidence (consider using library services).

  • Discuss with colleagues to refine the question.

  • Narrow down the question based on:

    • Time

    • Place

    • Group affected

  • Clarify what answer is expected to be found.

What Makes a Good Question?

  • Specificity/Focus:

    • PICO-TS/PICOTS Format:

    • P - Population

    • I - Intervention

    • C - Comparison

    • O - Outcome

    • T - Time frame (duration of follow-up, timing of outcome assessment)

    • S - Optimal study design to answer the question.

PICOTS: The “T” — Time

  • Importance of Time Frame:

    • Time frame defines when and for how long the intervention or exposure is evaluated.

    • Why Time Matters:

    • Outcomes may change over time; distinction between short-term and long-term effects is critical.

    • Determines clinical relevance and feasibility of study outcomes.

    • Influences study design and interpretation of results.

  • Examples of “T” in Pharmacy Research:

    • 30 days (e.g., hospital readmission).

    • 6 months (e.g., blood pressure control).

    • 12 months (e.g., A1C reduction).

    • Time to event (e.g., myocardial infarction).

    • Duration of exposure (e.g., opioid use ≥90 days).

Types of Questions

  • Two Types of Questions:

    • Foreground Questions:

    • Specific questions regarding a patient or population, often comparing two treatment approaches or two diagnostic tests.

    • Example: “Does metformin improve glycemic control in a 50-year-old patient with Type 2 Diabetes and mild renal impairment?”

    • Example: “Is the combination of albuterol and ipratropium more effective than albuterol alone in reducing symptoms during exacerbations in a 60-year-old patient with COPD?”

    • Background Questions:

    • Address general knowledge about a disease, condition, or process.

    • Types: “What is…?” “Why do…?” and “How does…?”

    • Examples: “What are the clinical manifestations of menopause?” “What causes migraines?”

Type of Question and Study Design

  • The type of question dictates the appropriate study design:

    • Diagnosis:

    • Seeks to select and evaluate diagnostic tests.

    • Best study design: Cross-sectional or prospective, blind comparison to a gold standard.

    • Therapy:

    • Seeks to determine if a treatment is effective.

    • Study designs: Systematic review, randomized controlled trial (RCT), or cohort study.

    • Prognosis:

    • Seeks to understand a patient’s likely clinical course over time.

    • Study designs: Cohort study, case control, and case series.

    • Harm/Etiology:

    • Seeks to determine the cause of a disease.

    • Best study designs: Cohort, case control, or case series studies.

From Research Question to Proposal

  • Key questions to consider when developing a proposal:

    • Who am I collecting information from?

    • What kinds of information do I need?

    • How much information will I need?

    • How will I use the information?

    • How will I minimize chance/bias/confounding?

    • How will I collect the information ethically?

    • Sample size considerations: Consult a statistician for assistance.

Clinical Question + Study Design

  • Best Research Design by Question Type:

    • Therapy (Treatment): RCT.

    • Prevention: RCT or Prospective Study.

    • Diagnosis: RCT or Cohort Study.

    • Prognosis: Cohort Study and/or Case-Control Series.

    • Etiology (Causation)/Harm: Cohort Study.

    • Meaning: Qualitative Study.

    • Cost: Economic Analysis.

Hierarchy of Evidence

  • Hierarchy of Evidence Definition:

    • The higher up a methodology is ranked, the more robust and closer to objective truth it is assumed to be.

    • Evidence-based medicine posits that not all evidence is created equal; a pyramid illustrates this hierarchy.

  • As you ascend the pyramid, the potential for bias decreases, alongside the publications of studies:

    • Analytic Descriptive

    • All Studies

    • Qualitative Experimental

    • Observational analytic

    • Randomized (parallel group)

    • Randomized (cross-over)

    • Cohort study

    • Cross-sectional (analytic)

    • Case-control study.

Key Study Designs: Overview

  • Not exhaustive but crucial:

    • For each study, consider:

    • What was the aim of the study?

    • When were the outcomes determined?

      • Some time after the exposure or intervention.

      • At the same time as the exposure or intervention.

      • Before the exposure was determined.

    • Differentiate between exposure assigned vs. exposure not assigned.

    • Types of evidence include Case Reports, cross-sectional studies, Case Series, and Surveys.

Assignment 2: Question 1

  • Task: Rank the hierarchy of evidence:

    • 1. Meta-Analysis

    • 2. Systematic Reviews

    • 3. Randomized Controlled Trials

    • 4. Cohort Studies

    • 5. Case Control Studies

    • 6. Case Series/Case Reports

    • 7. Animal Studies/Laboratory Studies.

Descriptive vs Analytic Research

  • Descriptive Research:

    • Objective: To describe the current state of a variable or condition.

    • Focus: Describing a population, phenomenon, or event.

    • Data Analysis: Minimal, mainly focused on observing, recording, and describing data.

    • Outcome: Detailed description of the subject.

    • Examples:

    • Determining the prevalence of diabetes in a population.

    • Reporting the number of COVID-19 cases in a region.

  • Analytical Research:

    • Objective: To understand or explain why and how certain phenomena occur.

    • Focus: Explores associations and tests hypotheses to infer causality or identify risk factors.

    • Data Analysis: In-depth, focused on understanding relationships by analyzing and interpreting data.

    • Outcome: Insights, explanations, and an understanding of causal relationships.

    • Examples:

    • Investigating whether smoking increases the risk of lung cancer.

    • Assessing the effectiveness of a new vaccine in reducing disease incidence.

Methodology Comparison

Aspect

Descriptive Research

Analytical Research

Objective

To describe

To understand

Focus

Describe population

Explore associations

Data Analysis

Minimal

In-depth

Outcome

Detailed description

Insights and explanations

Comparative Group

None

Yes!

Hypothesis

Generates hypothesis

Requires and tests a hypothesis

Experimental vs Observational Designs

  • Observational Studies:

    • Arise from ethical and cost restrictions of experimental studies.

    • Investigator does not assign exposure status; rather, it relies on the subjects' self-selection into treatment groups.

    • Investigators observe the relationship between drug exposure (independent variable) and the outcome of interest (dependent variable).

    • No randomization into intervention and control groups; data collected through interviews, surveys, or medical charts.

    • Less valid than experimental designs but more feasible for rare outcomes.

  • Experimental Studies:

    • Provide higher quality evidence where investigators manipulate conditions by assigning treatment groups.

    • Ethical considerations must always prioritize the participant’s welfare.

    • Require randomization into intervention and control groups, addressing confounding and biases directly.

Assignment 2: Q2 and Q3 Study Design

  • Scenario: The effects of behavioral-oriented counseling on smoking cessation are studied in hospitalized patients. 2024 smokers from four hospitals were randomly assigned to 3 groups: usual care, minimal intervention, and intensive intervention. Follow-up for 1 year to measure smoking cessation post-intervention.

  • Questions:

    • Is this study experimental, observational, or descriptive?

    • How might you change this study to employ a different design?

Potential for Causality

  • Hierarchy of evidence varies based on the design:

    • Experimental designs like RCT and cohort studies provide higher potential for causality compared to observational and descriptive designs.

Hierarchy of Evidence – Study Design Overview

  • The strength of each study design reflects its ability to reduce bias and increase validity:

    • Experimental > Observational > Descriptive studies.

Case Study Insights

  • Case Study: Describes experience or observations related to the care of a single patient.

  • Case Report: Detailed narrative of an individual’s diagnosis and treatment, potentially indicating an association between an outcome and intervention (like an adverse reaction).

  • Case Series: Group documentation of care experiences from multiple patients with similar conditions, often exploratory rather than conclusive.

Advantages and Disadvantages of Case Reports and Series

  • Advantages:

    • Identifies rare events;

    • Pinpoints delayed adverse drug events;

    • Requires minimal resources;

    • Can generate hypotheses.

  • Disadvantages:

    • No causal inferences can be drawn;

    • Risks of reporting bias and potential for false results;

    • Lacks statistical analysis.

Cohort Studies

  • Definition: Used to determine outcomes associated with exposure variables, classified as exposed and unexposed.

  • Types: Prospective and Retrospective; both perspectives allow the evaluation of the relationship between exposure and outcomes.

  • Advantages: Longitudinal approach capable of measuring disease incidence, studying multiple outcomes, less resource-intensive for common diseases.

  • Disadvantages: More costly and labor-intensive; less efficient for rare outcomes.

Case-Control Studies

  • Nature: Retrospective, seeking potential risk factors of diseases or outcomes.

  • Key Design Features: Subjects defined in terms of presence or absence of the outcome, compared within the same population to identify causative risk factors.

  • Advantages: Efficient for studying rare diseases; can explore multiple exposures; less costly compared to cohort studies.

  • Disadvantages: Susceptible to selection bias; reliance on retrospective exposure assessments; cannot determine incidence rates.

Cross-Sectional Studies

  • Definition: Prevalence studies providing a snapshot of the population at a given time thus allowing estimates of outcomes of interest.

  • Limitations: Not suitable for hypothesis testing but useful for generating hypotheses; cannot imply cause and effect relationships.

  • Advantages: Quick, easy, and inexpensive; gives a broad overview of certain characteristics within a population.

  • Disadvantages: Challenges distinguishing if exposure precedes outcome; vulnerable to bias (selection and recall bias).

Assignments and Questions for Reflection

  • Assignment 2 Q5: Analyze a study evaluating weight loss with dapagliflozin and discuss study design.

  • Assignment 2 Q7: Suggest redesigning a study on flu vaccines into a crossover study.

  • Assignment 2 Q12: Reflection on potential biases affecting a weight loss study compared to placebo treatment.

  • Assignment 2 Q10: Identify the design of a study comparing efficacy of COVID-19 vaccines using veterans' records.

Bias in Study Design

  • Definition: Systematic error that can lead to inaccurate results, potentially overstating or understating treatment effects.

  • Types of Bias:

    • Investigator Bias: Errors in design or analysis leading to favoritism towards certain outcomes.

    • Selection Bias: Preferential enrollment into treatment groups leading to biased outcomes.

    • Performance Bias: Differences in care that can impact the outcomes.

    • Attrition Bias: Unequal dropout rates affecting study conclusions, requires intention-to-treat analysis to address.

Insightful Review of Study Designs

Reflect on how the structure of clinical research impacts decision making, the implications of bias, and the approach to ensuring sound methodologies for robust epidemiological conclusions. Understanding these characteristics equips clinicians and researchers to navigate health inquiries critically and effectively, advocating for best practices in medical evidence.

References

  • Yetley et al., AJ CN. (2016).

  • New England Journal of Medicine studies and articles.

Conclusion

This guide serves as a comprehensive resource for understanding the fundamental concepts and importance of various study designs in clinical research. By addressing biases and establishing clear research questions through appropriate study designs, the evidence generated can significantly impact clinical decision-making.