Group Response Analysis: Clinically Interpretable Longitudinal Responder Analysis Methods Developed Using FDA Data
Bibliographic record
Abstract
1 Abstract Responder analyses for the evaluation of randomized clinical trial (RCT) data have become more common in the recent past, since they can provide the medical community with results that are more directly applicable to clinical care. For pain studies, the predominant responder analysis compares the change in the individual participants’ pain level at baseline to their value at the end of the study period and uses a predetermined clinically important change cut-off value to define a response. While useful, this method substantially reduces the efficiency of the RCT by dichotomizing the results and is limited to comparing the baseline to the end of the study only. In this paper, we introduce a novel approach to the patient response over time with a focus on single dose post-operative studies. This technique provides graphical presentations and statistical approaches to understand the onset of any specified level of response, the maximum proportion of patients with a response at any point in time, and the duration of that response over time. In addition, each outcome can be summarized to examine the result across all possible cut-off points for clinically important differences (CID). We accomplish this by introducing three interrelated, longitudinal efficacy statistics: ROOT, GRO, and GROOT. The response outcome over time (ROOT) estimates the total proportion of a study period an individual patient spends as a responder. The group response outcome (GRO) estimates the instantaneous proportion of responders at all time points across the study period. The group response outcome over time (GROOT) summarizes total efficacy in a cohort, and can be calculated as the area under the GRO curve, or as the mean ROOT; they are identical. This novel method provides a clinically interpretable responder analysis over the full period of the study and, by using every data point across time, mitigates the loss of statistical power typically associated with dichotomized responder outcomes. Group response analysis is based upon repeated assessments of categorical or continuous measures categorizing each participant’s status as a treatment responder or non-responder at every timepoint based on the prespecified clinically important difference. Both the visual and statistical comparison of any two or more curves provide a comparison of the overall efficacy, which can be statistically tested using a standard asymptotic hypothesis test (such as Wald (Johnson & Romer, 2016)). The method allows for an integrated evaluation of three main components of drug efficacy: the proportion of participants achieving a CID over time (effect), the time to achieve that response (onset), and the length of the response (duration). In this paper, we present the group response analysis methodology and then illustrate it using data from a placebo-controlled randomized clinical trial (RCT) for postoperative pain after third molar extraction treated with meloxicam and ibuprofen as an active comparator (Christensen et al., 2018). Our approach yields similar effect sizes as the sum of pain intensity differences (SPID) commonly used for pain study analyses while providing superior clinical interpretability and a more complete evaluation of drug therapies beyond just efficacy. We propose that this method can be used as a primary or secondary analysis of pain RCTs to answer the question of the patient response to treatment and provide suitable data to compare efficacies across treatment groups.
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 | 0.234 | 0.472 |
| Meta-epidemiology (narrow) | 0.002 | 0.002 |
| Meta-epidemiology (broad) | 0.003 | 0.005 |
| Bibliometrics | 0.008 | 0.005 |
| Science and technology studies | 0.001 | 0.002 |
| Scholarly communication | 0.003 | 0.003 |
| Open science | 0.004 | 0.004 |
| Research integrity | 0.003 | 0.006 |
| Insufficient payload (model declined to judge) | 0.019 | 0.003 |
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".