-
Research designs
- Experimental
- Observational
- Meta-analyses
-
Observational Study Design
Study design where the investigators collect, record, and analyze data on subjects as they naturally divide themselves by potentially significant variables. Human intervention is not par of the process.
-
Types of Observational Studies
-
Examples of descriptive studies
- Case report (case series)
- Cross sectional
-
Studies that focus in describing what exists
Descriptive Studies
-
Studies that focus on characterizing the occurrence of the condition or problem by person, place or time variables
Descriptive Studies
-
What is the weakest observational study design?
Case report
-
Detailed description of interesting characteristics observed in an individual patient. Unusual and rare conditions
Case report
-
Purpose of a case report
Introduce the audience to potentially new information about a possible association between a drug and a clinical outcome. Cannot test hypotheses.
-
Extension of the case report
Case series
-
Describes the characteristics of a group or cluster of individuals with the same disease or symptoms
Case series
-
Advantages of case series studies
- easy to write
- Observation may be useful for researchers designing a study to evaluate causes or explanations of the observations
-
Disadvantages of case series studies
- Susceptible to bias (subject selection, characteristics observed)
- Cannot reach conclusions
-
Analysis of data on a group of subjects at one point in time
- Cross-sectional, aka prevalence study, survey or epidemiologic study
- What is happening right now?
-
Disadvantages of cross-sectional studies
Only provides a "snapshot in time" of a disease or process. Can result in misleading info
-
Advantages of cross-sectional studies
- Relatively quick and inexpensive
- Useful for determining the status quo of a disease condition and evaluate diagnostic procedures
-
Examples of Analytical Studies
- Cross sectional
- Ecologic
- Multi-level analysis
- Case control
- Cohort
- Hybrid
-
Ecologic Analysis aka Aggregate Analysis
- Analyses that focus in comparisons of groups
- The groups are derived from ecological units
- The analyses combine existing datasets on these large populations
-
Compare groups, identify association between risk factors and prevalence of disease.
- Ecologic Analysis
- It is based on secondary data
- Groups, NOT individuals
-
Individual level analyses that incorporate some ecologic measurements
Help to assess if the individual's health is shaped by group or population characteristics
Multi-level analyses
-
Comparison of cases of the disease being investigated with a comparable group of controls
Case control
-
Group of individuals who are followed over time
Cohort
-
Randomization
- Flip a coin
- Random number generators
- Permuted block
- Stratification
-
Main limitation of Self-Controlled studies
Hawthorne effect: people change their behavior and sometimes improve simply because they received special attention by being in a study and not because of the intervention
-
Study that is a combination of concurrent and self-control studies
Crossover studies
-
Experiment where the unit of analysis are groups of people or the community.
Community trials
-
Group of individuals from different disciplines who evaluate protocols for clinical studies to asses risks to research participants and benefits to society
Institutional Review Board
-
Duties of IRB
- Protection of human subjects
- informed consent
- documentation
-
Nuremberg Code 1947 major contribution
Voluntary consent
-
Development of phisician's guidelines to perform medical research in human subjects and clinical research
Declaration of Helsinki 1964
-
Belmont Report 1979
Developed the ethical principles and guidelines for protection of human subjects participating on biomedical and behavioral research.
-
Basic ethical principles present in Belmont Report
- Respect for person
- Beneficence
- Justice
-
Respect for persons
- Acknowledgement of authority
- Protection of those with diminished authority
-
Beneficence
- Do not harm
- Maximize possible benefits and minimize possible harms
-
Justice
To each person an equal share
-
Considered the most important document in the history of ethics
Nuremberg Code 1947
-
Belmont Report 1979 applications
- Informed consent
- Assessment of risks and benefits
- Selection of subjects
-
IRB responsibilities
- Accomplish with federal laws for ethical research
- Guarantee that: human subjects should not be placed undue risk & Subjects should give uncoerced informed consent to their participation in the research
- Documentation
-
Requirements for IRB approval
- Minimal risk to subjects
- Expected risk/benefit ratio is reasonable
- Use of equitable subject selection technique
- Written informed consent
- Data monitoring system to ensure safety
- Confidentiality guaranteed
- Additional safeguards for vulnerable populations
-
Types of informed consent forms
- Adult/General Consent Forms
- Child Assent Form (Age 7-12)
- Adolescent Assent Form (Ages 13-17)
- Parent or guardian Consent Form (<18)
-
Examples of dissemination of research results:
- Abstracts
- Journal article
- Poster presentation
- Podium presentation
- Newsletter
- Patent
- Dissertations and theses
- Web postings
-
Extent to which an instrument measures what we wish it to measure
Validity
-
Accuracy & precision of the measurement procedure
Realibility
-
Its is a label for an abstract idea
Concept
-
Objectives of meta-analyses
- Increase statistical power by ↑ sample size
- Resolve uncertainty when reports don’t agree
- Improve estimates of effect size (the degree to which phenomenon is present in population)
- Answer questions not posed at beginning of studies
-
What are the hybrid studies?
- Nested Case-Control
- Nested Case-Cohort
- Panel Study
- Repeated Surveys
-
Why to use matched pair design in case control?
- To control confounding and selection bias
- Increase similarity of cases and controls
- Can increase power
-
The primary objective of this study is to compare the occurrence of symptoms, disease and death in the exposed and unexposed groups.
Cohort
-
Types of populations in cohort studies:
- Open: enter the study at various time after it started
- Closed: nobody is added after it started
- Fixed: irrevocable event. individual's exposure does not change over time
-
Studies that combine the results of two or more individual studies into one large study
Meta-analyses
-
Involves a case-control within a prospective cohort study
Hybrid > Nested Case-Control
-
Combines cross-sectional and prospective cohort study
Panel Study
-
Successive cross-sectional studies are performed over time on the same study population, but each sample is selected independently.
Repeated surveys
-
Limitations of meta-analyses
- Heterogeneity can affect reliability
- Publication bias
- Biases of individuals studies are combined
- Controversy over whether unpublished data should be included
-
A systematic error in the design of the study that leads or conduct to an erroneous association between the exposure and disease
Bias
-
Examples of Bias
- Publication bias
- Selection bias
- Performance
- Detection
- Attririon
-
Kind of bias where the subject can no longer be located or, they can no longer want to participate
- Attrition bias
- Can decrease power
-
Bias in which the nature & direction of the study results influence its likelihood of being published.
Publication bias
-
Blinding can help minimize which form of bias?
Performance bias: subjects or providers change their behavior based on what group they’re in.
-
How can attrition bias be minimized?
Collection of contact info and data that helps locate people as study progresses, maintaining regular contact, sending letters to last known address w/ “address correction request”, checking directories, internet, & contact persons identified @ start of study.
-
Define I2 Statistics:
- The percentage of variation across studies that cannot be attributed to chance
- Low heterogeneity : <25%
- Moderate 25-75%
- High >75%
-
Refers to the repeated performance of meta-analysesin a chronological fashion
Cumulative Meta-analysis
|
|
