3
$\begingroup$

I am trying to figure out the name for a design study in the following scenario. Imagine a hospital where patients with a certain condition (say bacteria infection) are treated with a certain medication (say antibiotic A). At some point in time, there was a decision to change the antibiotic used to treat the condition to another medication (Antibiotic B). A sample were drawn from time1 (with use of Antibiotic A) and another sample from time2 (after change to Antibiotic B). The samples can be considered independent and differences can be investigated using either Independent samples t-test or chi-square. I am not sure what type of study design this is:

The following are my thoughts so far:

  1. Causal-Comparative or Ex post facto [but since I can change the treatment (antibiotic), I am not sure this is right].
  2. Quasi-experimental - My doubt with this is that this is an observational study and the change in practice (antibiotics) was not due to experimentation but change in operation.
  3. A combination of retrospective and prospective study as stated here, however, the page and every example of this that I saw seems to suggest that this design is used on THE SAME cohort.

I saw similar design here but there was no mention of the study design. All assistance are welcome.

Improve this question
$\endgroup$

1 Answer 1

Reset to default
4
$\begingroup$

I would consider this to be a natural experiment. While sometimes the separation into groups is geographic (see, e.g., https://en.wikipedia.org/wiki/Natural_experiment#History), other times the separation into groups is temporal (e.g., https://en.wikipedia.org/wiki/Natural_experiment#Smoking_ban) or otherwise. You could also see this as a quasi-experiment (or not) depending on how you define that term (compare https://en.wikipedia.org/wiki/Quasi-experiment#Ethics and https://en.wikipedia.org/wiki/Quasi-experiment#Advantages).

Improve this answer
$\endgroup$
6
  • 1
    $\begingroup$ This is a useful framing: one advantage of "naming" study designs is to think about their common sources of bias. Here the gold standard design would be a randomised controlled trial (RCT) for the two treatments; having a before-after design that is separated on treatment opens up questions of confounding and influence from broader contextual factors (e.g. admission procedures for in-patient vs. community treatment) $\endgroup$
    – James Stanley
    Commented Nov 11, 2020 at 2:01
  • $\begingroup$ @JamesStanley, I appreciate your contribution. I do agree that from a statistical point of view, there are concerns for confounding and bias, however, I am not focusing of the statistical ramifications at this moment. I am more focused on the design, which I have seen in a few study set-ups particularly healthcare but haven't been able to pinpoint its appropriate study design name. $\endgroup$
    – adejames2000
    Commented Nov 11, 2020 at 15:42
  • 1
    $\begingroup$ @adejames2000 This paper discusses quasi-experimental designs in the evaluation of interventions for infectious disease. academic.oup.com/cid/article/38/11/1586/285372 $\endgroup$
    – Diana Petitti
    Commented Nov 11, 2020 at 16:36
  • $\begingroup$ @adejames2000 -- that's all good, I appreciate my comment didn't address your direct question (hence a comment rather than an answer). My comment point was that most of the utility in naming a study design is in what this tells us as researchers/readers about the pros/cons of that design and how to deal with them (i.e. the design name is shorthand for what to consider when drawing conclusions based on that evidence) $\endgroup$
    – James Stanley
    Commented Nov 11, 2020 at 22:16
  • 1
    $\begingroup$ @JamesStanley That is right, naming the study design help to put a lot of things in better perspective. $\endgroup$
    – adejames2000
    Commented Nov 12, 2020 at 21:26

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.