Hi Kristin,

Thanks for your questions - I'll have a go at answering each of them in turn:

1. Pretest/posttest studies. You need to have one of the following:

• Total N; difference in means (pre-post); the standard deviation of the difference in means (pre-post); correlation.
• The size of the group and the t value for the difference in means and the correlation (often imputed!).
• The size of the group and the p value for the difference in means and the correlation (again, often imputed).

If you have the above, it's possible to calculate a standardised mean difference. The ability to do this is in EPPI-Reviewer 3 at the moment and I can add to ER4 any time - so can get this in this week if you need it. The statistic that is often not reported is the correlation, so reviewers either have to guesstimate, ask the authors or leave the outcome out altogether. If you impute a correlation, it's usual to do a sensitivity analysis, seeing whether estimating different values changes your results.

2. Studies with no comparison group: as above, you can enter these manually, or I can put the necessary code online soon. (I know you know this isn't the most robust study design!)

3. Outcome classifications are codes that you specify in your code set to be 'outcome classifications'. These codes then appear on the right-hand-side of the outcome entry screen for selection against specified outcomes. They are useful if you want to categorise outcomes when, for example, you have the same outcome assessed with different measurement tools, or in different subgroups. They then appear in the 'single' reports so that they can be used as covariates in analyses conducted outside EPPI-Reviewer. We have some documentation and screenshots on their use, so these should make their way into the manual soon. In the meantime, I can email you what we have on this.

4. I'm not sure what the outcomes options pop-up is? All your outcomes appear the right number of times in their various listings as far as I can see.

5. In terms of organisation, you can apply codes at the review level and at the outcome level. You probably want to have some categories that apply to both (broad categories of interventions, outcomes, comparisons, for example), and some (the outcome classifications, mentioned above) that give you a more fine-grained approach to categorising outcomes.

I'm happy to talk all this through of course!

Best wishes, James.

Hi Kristin,

Re the outcomes, you should have one outcome record per comparison, so if your comparison is pre-post, there should be one record. We can go through this tomorrow if you want? We can have a look at the data you have available and see what the best way forward is.

Re outcome classifications, we'll get some text into the manual this week. Essentially, codes are descriptors that are usually applied to studies (e.g. 'This study had a population with characteristic x'). However, sometimes the characteristic doesn't apply to the study, but to the outcome. For example, some studies I know of that promoted fruit & veg intake among children measured their success in terms of portions per day as well as calorific intake. When a study has done both, you'd want to be able to tell the difference between their measurement tools; this is what classifications for outcomes are for, as they apply to outcomes, rather than the study as a whole. (We could code the study as a whole as using both portions per day and calorific intake, but that doesn't help us distinguish the outcomes.)

Does this help? James.