Observational Studies
Case Reports/Studies
Definition
Case report is 1 case
Case study usually up to 10 (can be more)
Utility
Used for small sample conditions/risk factors
Identification
Small sample size
No population level assumptions
No specific inclusion/exclusion criteria
Examples
Studying progression of epidermodysplasia verruciformis (EV)
Studying QoL in patients w/ auto brewery
Case Control Study
Definition
Compares patients with a disease to those without and assess for relationship to a specific risk factor
Always retrospective
Utility
Helps better understand association between risk factors and disease
Can demonstrate correlation not causation
Identification
Identifies groups based on presence or absence of disease
Results reported as odds ratio
OR states how likely someone with the condition is to have that risk factor compared to how likely someone with the condition is to not have it
Ex: Case-control study looking at lung cancer patients and smoking, an odds ratio of 3 for lung cancer patients means that the odds of a lung cancer patient being a smoker are 3 times greater than the odds of a lung cancer patient NOT being a smoker
Cohort Study
Definition
Compares patients with a specific risk factor with those without and assess for development of a disease
Can be retrospective, prospective, or ambispective
Retro- looks at patient charts in the past to see how many times something has happened as a result of the risk factor in question
Prospective - follows the cohort for a period of time to see if anything happens in the future
Ambispective - checks if anything has happened yet and then follows the people to see if anything does happen
Utility
Helps identify possible consequences are of suspect health behaviors
Identification
Participants are usually recruited for the presence or absence of a risk factor, which can include demographics or jobs
Demographics: A study that follows a group of African-Americans and Caucasians to see rates of hypertension development is a cohort study. Same with a study that follows a group of people in 20’s and a group in their 50’s for 10 years to see rates of STD acquisition
Job: looking at black lung development in coal miners vs. bankers
Results given in relative risk
Risk that someone with the risk factor develops the condition compared to the risk that someone without the risk factor develops the condition
Ex: In that coal miner example, an RR of 2 means a the risk of a coal miner developing black lung is two times greater than the risk of banker developing black lung
Experimental Studies
Quasi-experimental study
Definition
Compares an intervention to a control but there is NO randomization
RCT w/o the randomization or control
Utility
Allows for testing the impact of an intervention in situations when randomizing people to groups is not possible
Identification
Has an intervention group and a control group but they are NOT randomly assigned
Examples
To see if tutoring helps improve a test scores in a 7th grade algebra class, you offer after school tutoring. Since you can’t randomly assign kids to tutoring, you just compare the scores of the kids who came to tutoring and the scores of the kids who didn’t come. Intervention: Tutoring Test group: kids who came to tutoring Control: kids who didn’t. Why quasi? Not randomized
Randomized Control Trial
Definition
Random assignment (after informed consent, of course) of participants to an intervention or control group
Often single or double-blinded
Utility
Highest standard of experimental study
Random assignment of participants increases study validity and makes the results more reliable
Identification
If there’s a test group and a control and the person running the study assigned groups? It’s an RCT
Examples
Drug Clinical Trial
Phase I: Drug safety and dosing (will it hurt the patient?)
Phase II: Drug efficacy (will it actually help the patient?)
Phase III: Large scale comparison of drug to standard treatment (will it help significantly more than what we doing now?)
Study of Studies
Systematic Review vs Meta analysis
Definitions
Systematic Review: Review of the RCTs/best research available on a topic. Essentially an explanation of what the current literature says and a presentation of the studies examined.
Meta-Analysis: Combine the results of the RCTs studied to provide a combined analysis of the issue.
Differentiating between the two though example: if the study objective is to understand the efficacy of a new targeted chemotherapy drug but you don’t want to read every RCT paper published on it.
Systematic Review: Here are all the RCT published about this new drug and a summary of general trends seen in them. Also here’s a table showing all the papers included in this review and what they each found individually
Meta-Analysis: Here’s a combination of all the r2 values (which would show correlation between the treatment and cure) and the p-values (which show significance of findings) from all those papers. Now instead of just knowing in general terms that the papers say this new drug helps, you can see the mathematical evidence of just how good they say it is.
Evaluating studies of studies
I2 is a measure of heterogeneity à the bigger the i2 the more different and difficult to properly compare the studies are
These tests can be stratified by author/population/theme/study size to control for heterogeneity. For example, if you’re doing a systematic review looking into refugee healthcare utilization, it may make sense to compare studies of Bhutanese refugees with other studies of Bhutanese, because DRC refugee use rates may be completely different for community reasons