Skip to main content

Treatments, interventions or tests are rarely exactly the same between studies. For example:

Drug studies: dosage, timing, length of treatment may all vary somewhat between studies.

Interventions: the same general surgery may differ somewhat in technique between studies. Some interventions may include multiple components and may combine components in different ways across different studies.

Diagnostic accuracy studies: test manufacturers, analytic and pre-analytic procedures may vary between studies.

If you were thinking ahead when we were talking about using generic suggestions for arms, you may have wondered, “How can we capture the detail on differences between arms?”

This is where the Arm (or Diagnostic Test) Detail tab comes in. Capturing details about differences between arms is going to be vital in the analysis phase because you are capturing information that may explain a substantial amount of the heterogeneity in a meta-analysis. Essentially, you want to use the Arm Details tab to not merely capture details on the different arms, you want to capture differences among studies in how the treatments, interventions, tests or exposures were carried out or measured.

It bears repeating: capturing this detail may prove to be a huge benefit in the analysis phase.

What Details Should We Capture in the Arms or Diagnostic Test Details Tab?

This depends heavily on your purpose and questions. It also assumes you have an understanding of what differences might actually make a difference in the results of different studies.

Below, we offer some suggestions for types of Arm Detail questions based on different types of systematic review questions.

  • Dosage (how much? how long? how often?)
  • Technique
  • Provider (e.g., physician, physical therapist, nurse, dietitian, etc.)
  • Mode of treatment (e.g., with counseling: face to face? in person?)
  • Intervention format (e.g., individual sessions, group sessions, mixed?)
  • If multi-component interventions, what are the different components (e.g., drug, dietary, physical activity, counseling, etc.)?
  • Techniques for providing the intervention (e.g., with physical activity: walking versus aerobic exercise versus strength training, etc.)
  • If intervention chains (i.e., sequenced practices), which components in the intervention chain are included (e.g., pre-analytic procedures –> organism identification –> organism susceptibility –> stewardship or communication to treating clinician –> treatment technique)?
  • Dosage (how much? how long? how often?)
  • Provider (e.g., physician, physical therapist, nurse, dietitian, etc.)
  • Mode of treatment (e.g., with counseling: face to face? in person?)
  • Intervention format (e.g., individual sessions, group sessions, mixed?)
Diagnostic Accuracy
  • Test manufacturer
  • Pre-analytic procedures (e.g., for diagnosis of C diff, was there a requirement for unformed stools prior to the test request?)
  • Differences in analytic technique
  • Different methods of measuring the exposure (e.g., different methods of assessing dietary intake)
  • Other potential confounding variables (i.e., other sources of exposure or factors that could mitigate or modify the result)

This list is not exhaustive, nor need every one of the suggested questions be asked in the different types of projects. The key, however, is to plan ahead before you begin building out your data extraction template for which details are likely to be important (which is to say, may perhaps affect the differences in results between subjects and studies).

A couple of suggestions:

If you are working as part of a team and have some knowledge of the topical area: plan to have a team discussion to identify Arm Details questions that should be extracted.

If you are working alone and do not have a deep understanding of the topical area: during the full-text review phase of study selection, keep notes on differences you see among studies on the characteristics listed above. Reading and re-reading the articles may help you find patterns that you’ll want to capture at the data extraction phase.

One last thing: the suggestions mentioned earlier in this resource for creating structured (versus free text) questions is particularly important for Arms Details. Using structured questions in Arm Details will allow you to easily sort and filter studies by their commonalities and differences. This will make it much easier to identify factors or covariates for subgroup meta-analyses.