Bibliographic record
Abstract
Optimal anemia management strategies for end-stage kidney disease patients treated with hemodialysis are unknown, with controversies over how best to utilize erythropoiesis-stimulating agents (ESA) and intravenous iron to support hemoglobin levels and minimize adverse events. With large randomized trials rare in nephrology, it is thus crucial that research questions are clearly defined, study designs are appropriately selected, and analytic techniques are properly implemented when using observational data. The three aims of this dissertation attempt to address current controversies in anemia management using innovative statistical methods, leveraging data from the Dialysis Outcomes and Practice Patterns Study (DOPPS), an international prospective cohort study of hemodialysis patients. Aim 1 focused on anemia management during the transition period to hemodialysis. Among patients who initiated hemodialysis with hemoglobin < 10 g/dL, the highest (vs. low) doses of ESA and intravenous iron were each associated with elevated mortality. To assess the impact of pre-dialysis anemia treatment, a seemingly counterintuitive design – restricting to patients who achieved target hemoglobin (>=10 g/dL) four months later – was used to limit inclusion of patients whose low hemoglobin at hemodialysis initiation was likely confounded by poor health status. Even in this subset, anemia at hemodialysis initiation was common and associated with elevated mortality. A more proactive approach to anemia management prior to end-stage kidney disease may thus avoid aggressive correction of hemoglobin levels during the early dialysis period and improve survival. Aim 2 focused on how hemoglobin response to ESA therapy may be blunted by inflammation. Hemoglobin and ESA doses were compared over the 3 months before and after detection of new inflammation, defined as an acute C-reactive protein increase from < =5 to >10 mg/L. Confounding due to baseline characteristics, whether measured (age, sex, comorbidity history) or unmeasured (genetic or environmental factors), was avoided by this longitudinal self-matched design. Patients experiencing new inflammation had both higher ESA doses and lower hemoglobin (vs. pre-inflammation levels), supporting the hypothesis that inflammation increases resistance to ESA treatment. Quicker recognition of new inflammation in hemodialysis patients could help identify the cause of worsening anemia and guide ESA and intravenous iron dosing decisions more proactively. Aim 3 focused on applying the parametric g-formula, an extension of standardization to longitudinal data, to replicate a randomized trial using observational data. DOPPS data were used to compare iron supplementation strategies, with the goal of mimicking the recently published PIVOTAL randomized trial. Comparing the proactive high-dose vs. reactive low-dose strategy, 1-year mortality risk was 20% greater under the parametric g-formula simulation, but similar in the PIVOTAL trial. Simulated differences for all secondary outcomes were directionally consistent but of lesser magnitude than in the PIVOTAL trial. Success in mimicking the PIVOTAL trial was mixed, and potential explanations for the divergent results include model misspecification and/or differences in the study populations. This example illustrates the potential of the parametric g-formula to evaluate many variations of complex interventions across different populations, which could prove enormously informative in the age of big data. This dissertation outlines critical gaps in the literature on anemia management in hemodialysis patients, and describes three studies that utilize innovative designs and complex statistical analyses to address these gaps. These studies attempt to advance both the optimization of anemia management strategies in hemodialysis patients and the use of causal inference principles to guide epidemiologic research using observational data.
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 | 0.003 | 0.016 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.001 | 0.001 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.001 | 0.001 |
| Open science | 0.000 | 0.001 |
| Research integrity | 0.000 | 0.000 |
| Insufficient payload (model declined to judge) | 0.000 | 0.000 |
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".