Epi practicals - 14 April 2021

Orrell et al Adherence is not a barrier of ART S Africa AIDS 2003

1. What is the rationale for the study? Is it clear and sensible?

Researchers intended to determine which kind of factors influence adherence of HIV-infected African people to ART
It is clear and sensible

2. What are the aim and objectives of the study? Are they clear and unambiguous?

Their aim is to find out predictors of decreased adherence and virologic treatment failure.
They are clear and unambiguous.

👉 Was there a clearly identified primary outcome

This study's primary outcome is to determine predictors which produce adherence less than 95% and HIV-RNA copies more than 400/mL.
The primary outcome is clearly identified.

3. What is the study design, exposure of interest and outcome of interest?

Design is prospective cohort study within phase 3 randomized controlled trial.
Exposures are following:
- age
- home language (English or not)
- dosing (three times a day)
- baseline viral load
- mode of ART (two NRTI)
- number of tablets (more than 10 a day)
- food restrictions
- adherence less than 95% (against virological failure)
Outcomes are adherece less than 95% and HIV-RNA more than 400copies/mL.

4. What is the target population and how was it selected? How representative is the population?

Target population is indigent African HIV-infected people.
Sample is selected from Cape Town AIDS Cohort, who are the patients of U of Cape Town HIV Clinic and almost indigent.
Sample is moderately representative of population because it is definitely indigent but chosen only from South Africa.

5. How are exposure variables and outcomes defined?

As mentioned in answer for question 3.

6. Was there potential for misclassification of the outcome?

It might have potential for misclassification of the outcome, because low adherence might influence virologic failure in one direction, which means outcome itself might turn to be an exposure.

7. Was there potential for misclassification of the exposure?

I don't know details of ART at that time but guess dosing, mode of ART and number of tablets might confound each other in a certain extent.

8. What outcome measure is used and is this appropriate for this type of study?

They measured adherence by tablet-returns but it may potentially miss another mode of adherence such as disposal of tablets at home.
Measuring RNA copies is considered appropriate.

9. What other sources of bias are likely?

In terms of adherence, selection/response bias exists that participants of clinical trial may be more health-conscious or be in lower socio-economical status and influence adherence of taking medicine.

10. Discuss the statistic

👉 did the analysis match the stated objectives?

Researchers did multivariate analysis by Logistic regression and selected variables to include in the multivariate analysis by each of univariate analyses of variables. But at least I myself couldn't find on how many variables they did univariate analyses. If they did on more than 20 variables and set probability of risk of chance at 5%, variables they included in multivariate analysis might already have alpha-error.

👉 was a sample size calculation reported?al analysis strategy.

I couldn't find sample size calculation in the article. This cohort study was derived from phase 3 clinical trial and it might be difficult to independently calculate appropriate sample size in the phase 3.

11. Discuss the tables (and figures if relevant). What is presented in each? Are the results presented clearly?

👉 Are there any data not presented that you think should have been?

12. Is their discussion of the results appropriate, and do they draw appropriate conclusions which can be justified based on the results presented?

👉 do they discuss the strengths and limitations and potential sources of bias of the study? – do you agree with the authors?

Theron Feasibility accuracy clinical effect XpertTB primary care Africa RCT Lancet 2014