Case series:
A study on a group of patients with a known disease, where we simply track them to follow their outcomes.
This usually shows up as a distractor answer choice on USMLE, but I’ve seen it as a correct answer for following a group of patients with West Nile virus.
- In other words, we’ve got 20 patients with West Nile virus. –> We follow them to see how they progress in terms of symptoms, morbidity, mortality, etc.
Can’t be used to assess causality of disease.
Their main benefit is that they can sometimes help generate hypotheses about diseases. For instance, after following 20 patients with West Nile virus and seeing 2 develop myocarditis, we could possibly hypothesize that there’s increased risk of myocarditis from West Nile virus, even though we can’t conclude this in the slightest. It could be mere conjecture / hypothesis.
Case-series are also considered very low-level on the evidence hierarchy:
Systematic review:
A big/comprehensive summary that pools together all available evidence on a topic, usually including lots of RCTs.
For example, it might include all 30 RCTs on a topic, e.g., LASIK.
This is very powerful and the peak of the evidence hierarchy pyramid because it includes all of the available evidence on a topic.
Where it can get confusing is, meta-analyses can be performed within systematic reviews so that the systematic review’s conclusions are more powerful. Meta-analyses pool together RCTs as well, but they are very rarely stand-alone.
Meta-analyses help see significance within small pools of RCTs when those individual RCTs may not have shown significance.
For example, within our systematic review of 30 RCTs on LASIK, we do one meta-analysis on 6 of those RCTs and a second meta-analysis on 4 other RCTs. Each of these meta-analyses has refined our data from those small pools of RCTs, where alone the RCTs may not have shown significance, but now as part of meta-analyses we do see significance.
These analyses are part of the greater systematic review. In other words, these meta-analyses are not separate from the systematic review and “incorporated” into it. They are a tool simply used within our systematic review as part of how we analyze all of the available evidence/RCTs.
That being said, meta-analyses are a tool to help increase power within a systematic review by increasing sample size and reducing random error.
Another example:
We compile all available evidence on a topic, e.g., 40 known RCTs on Zika virus. Within our systematic review, we might conduct one or more meta-analyses, where the latter pool together RCTs to refine the data. So within our 40-RCT systematic review, we conduct one meta-analysis on 7 RCTs, where individually these RCTs didn’t show significance, but when their data is pooled as part of a meta-analysis, we do see significance. This data isn’t “incorporated” into the systematic review as though the meta-analysis is an external study. It is already part of the systematic review.
So systematic reviews may contain one or more meta-analyses. Meta-analyses aren’t separate from systematic reviews.
Meta-analyses can’t contain systematic reviews.
Crossover study:
A type of RCT where a group of patients will receive both interventions (control and treatment) in sequential order, with a period in between known as a washout.
- A group of 500 patients receives Drug X for 2 months –> 2-week washout period/break –> Drug Y or placebo for 2 months.
- Another group of 500 patients receives Drug Y or placebo for 2 months –> 2-week washout period/break –> Drug X for 2 months.
Crossover studies can have more than 2 interventions. For example, if we have 3 different interventions, we’d have 2 washout periods, e.g., Drug X –> Washout 1 –> Placebo –> Washout 2 –> Drug Z.
Placebos are still considered “interventions” in clinical research.
One point about crossover studies for USMLE is that each patient “serves as their own control,” which means that each patient’s response to a drug is compared to their response to the placebo/control. This helps eliminate differences between subjects, which means less variability in data –> more power for n= fewer number of patients.
Essentially, power can be increased in crossover studies since patients serve as their own controls and standard error is reduced. But this doesn’t necessarily mean that the overall power of the study is high. It still depends on things like sample size, effect size, and variability.
When you get one of these Qs on USMLE, it’s quite easy to be honest. They’ll just describe a scenario such as the above, and the answer is literally just “crossover study.”
—
