Evaluation of the WHO criteria for antiretroviral treatment failure among adults in South Africa: authors' reply
Bibliographic record
Abstract
We thank Schuelter-Trevisol et al. [1] for their interest in our paper [2]. Our aim was specifically to assess the performance of clinical and CD4 cell count criteria for antiretroviral treatment (ART) failure as defined in the WHO guidelines, which is important given that the criteria were developed based on expert opinion rather than evidence. Schuelter-Trevisol et al.[1] question our conclusions, based on concerns about exclusions from the analysis. Individuals excluded from this analysis fall into two main groups: first, those who were lost to follow up by 12 months and, second, those excluded because of missing laboratory data. The majority of those lost to follow up will no longer be on ART and are very likely to fulfil virological criteria for treatment failure. Our analysis was based on the scenario of a clinician assessing a patient attending 12 months after starting first-line ART and having to decide whether to continue first-line treatment or switch to second line. Individuals who are lost to follow up before 12 months are a very important group (which we are addressing in separate study), but the clinical dilemma of whether to switch to second-line therapy at the 12-month visit does not arise, which is why we think it is preferable to exclude them from this analysis. The second group excluded were, necessarily, 97 individuals with missing laboratory data at 12 months. We have compared these 97 excluded individuals with the 324 included in the study with respect to baseline factors found, in a previously published analysis, to be associated in this population with virological outcome at 12 months [3]. Comparing those excluded to those included; median age was 40.2 years in each group (P = 0.91), median weight was 65 kg in each group (P = 0.96), median CD4 cell count was 164 versus 154 cells/μl (P = 0.41) and viral load was 43 766 versus 47 503 copies/ml (P = 0.41) (P-values calculated using the Wilcoxon rank sum test in each case). Thus, those excluded from the study due to missing data were very similar to those included, and these exclusions are unlikely to have affected the reported prevalence of virological failure. The prevalence of virological failure among those retained in care in our study is similar to that quoted in other routine programmes and is rather higher than reported from many early programmes from resource-constrained settings [4]. Thus, the positive predictive value for the WHO criteria would be expected to be even lower in these settings with better outcomes, making our recommendations all the more relevant. We would expect the prevalence of virologically defined treatment failure to be higher at later time points. We plan to investigate this when we have additional data from individuals with longer duration of follow up. Our results are consistent with data from the recent studies in Canada [5], Botswana [6] and Thailand [7] cited in our paper [2] and a study in Malawi [8]. On the basis of this accumulating evidence, we stand by our recommendation that individuals fulfilling clinical or CD4 cell count criteria for treatment failure should have HIV viral load measured before switching to second-line therapy. Our recommendations are consistent with a proposal by Colebunders et al. [9] with respect to a model using clinical and simple laboratory evidence along with HIV viral load in selected cases to make decisions on switching to second-line treatment in resource-limited settings. The question of whether sex is associated with ART adherence was addressed in a literature review carried out by Ammassari et al. [10]. No association between sex and adherence was found in 10 out of 11 studies in which the question was addressed. In addition, the studies cited above [5–8] from cohorts, including a higher proportion of women, reached conclusions similar to ours.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.010 | 0.076 |
| Meta-epidemiology (narrow) | 0.001 | 0.001 |
| Meta-epidemiology (broad) | 0.002 | 0.001 |
| Bibliometrics | 0.001 | 0.001 |
| Science and technology studies | 0.001 | 0.002 |
| Scholarly communication | 0.002 | 0.005 |
| Open science | 0.003 | 0.002 |
| Research integrity | 0.013 | 0.020 |
| Insufficient payload (model declined to judge) | 0.004 | 0.002 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".