LONGITUDINAL EXUDATE BURDEN AND DELAYED HEALING IN VENOUS LEG ULCERS: A REAL-WORLD ELECTRONIC HEALTH RECORD ANALYSIS
Author(s)
Peter Gloviczki, PhD, MD1, Yun Han, PhD2, Vladica M. Velickovic, PhD, MD3, Grace Gahlon, PhD2, Cornelia Santoso, PhD4, Shelby Corman, MS, PharmD5, David G. Armstrong, PhD, MD6, Jan Kottner, PhD7, Kenneth Rothman, PhD8.
1Mayo Clinic, Phoenix, AZ, USA, 2Precision AQ, New York, NY, USA, 3Head of Evidence Generation, Hartmann group, Neu-Ulm, Germany, 4Hartmann group, Heidenheim, Germany, 5Precision AQ, Gladstone, NJ, USA, 6Keck School of Medicine of University of Southern California, Los Angeles, CA, USA, 7Charité-Universitätsmedizin, Berlin, Germany, 8Boston University School of Public Health, Boston, MA, USA.
1Mayo Clinic, Phoenix, AZ, USA, 2Precision AQ, New York, NY, USA, 3Head of Evidence Generation, Hartmann group, Neu-Ulm, Germany, 4Hartmann group, Heidenheim, Germany, 5Precision AQ, Gladstone, NJ, USA, 6Keck School of Medicine of University of Southern California, Los Angeles, CA, USA, 7Charité-Universitätsmedizin, Berlin, Germany, 8Boston University School of Public Health, Boston, MA, USA.
OBJECTIVES: To quantify the association between longitudinal wound exudate burden and complete healing of venous leg ulcers (VLUs), using outcomes relevant to prognosis and resource planning.
METHODS: De-identified electronic health record data from US outpatient wound centers (2002-2020) were analyzed retrospectively. Eligible VLUs had an initial area of 2.0-100.0 cm² and follow-up after the index visit. Clinician-recorded exudate was categorized at each visit as none, mild, moderate, or heavy, modeled as a time-varying exposure. Complete healing was identified from documentation of closure, resolution, or treatment completion. Marginal structural models were applied using stabilized inverse probability-of- treatment and censoring weights. Weighted Royston-Parmar flexible parametric survival models estimated hazard ratios (HRs) for healing, counterfactual healing probabilities, and restricted mean survival time (RMST) representing days unhealed.
RESULTS: The analysis included 18,435 VLUs from 8,184 patients (mean age, 67.8 years; 48.1% male). Compared with no exudate, the hazard of healing declined progressively with mild (HR, 0.11; 95% CI, 0.09-0.15), moderate (HR, 0.08; 0.06-0.10), and heavy exudate (HR, 0.02; 0.01-0.04). Corresponding RMST values were 26, 218, 287, and 343 unhealed days, representing 192, 260, and 317 additional days unhealed for mild, moderate, and heavy exudate. Estimated 52-week healing probabilities were approximately 100.0% (no exudate), 58.4% (mild), 31.2% (moderate), and 8.7% (heavy). Median observed follow-up was 45 days; therefore, 52-week estimates required model extrapolation. The outcome definition included treatment completion, which may not reflect closure.
CONCLUSIONS: Greater exudate burden was associated with slower VLU healing and longer time unhealed. This exudate gradient association may partly reflect wound chronicity. Routine exudate assessment may improve prognostic stratification and inform resource-use and economic models. Because absent exudate is atypical in active VLUs and late estimates were heavily model-dependent, absolute contrasts should be interpreted cautiously. Prospective studies should combine standardized exudate measurement with resource-use and cost data.
METHODS: De-identified electronic health record data from US outpatient wound centers (2002-2020) were analyzed retrospectively. Eligible VLUs had an initial area of 2.0-100.0 cm² and follow-up after the index visit. Clinician-recorded exudate was categorized at each visit as none, mild, moderate, or heavy, modeled as a time-varying exposure. Complete healing was identified from documentation of closure, resolution, or treatment completion. Marginal structural models were applied using stabilized inverse probability-of- treatment and censoring weights. Weighted Royston-Parmar flexible parametric survival models estimated hazard ratios (HRs) for healing, counterfactual healing probabilities, and restricted mean survival time (RMST) representing days unhealed.
RESULTS: The analysis included 18,435 VLUs from 8,184 patients (mean age, 67.8 years; 48.1% male). Compared with no exudate, the hazard of healing declined progressively with mild (HR, 0.11; 95% CI, 0.09-0.15), moderate (HR, 0.08; 0.06-0.10), and heavy exudate (HR, 0.02; 0.01-0.04). Corresponding RMST values were 26, 218, 287, and 343 unhealed days, representing 192, 260, and 317 additional days unhealed for mild, moderate, and heavy exudate. Estimated 52-week healing probabilities were approximately 100.0% (no exudate), 58.4% (mild), 31.2% (moderate), and 8.7% (heavy). Median observed follow-up was 45 days; therefore, 52-week estimates required model extrapolation. The outcome definition included treatment completion, which may not reflect closure.
CONCLUSIONS: Greater exudate burden was associated with slower VLU healing and longer time unhealed. This exudate gradient association may partly reflect wound chronicity. Routine exudate assessment may improve prognostic stratification and inform resource-use and economic models. Because absent exudate is atypical in active VLUs and late estimates were heavily model-dependent, absolute contrasts should be interpreted cautiously. Prospective studies should combine standardized exudate measurement with resource-use and cost data.
Conference/Value in Health Info
2026-11, ISPOR Europe 2026, Vienna, Austria
Value in Health, Volume 29, Issue 12S
Code
RWD5
Topic
Clinical Outcomes, Epidemiology & Public Health, Real World Data & Information Systems
Topic Subcategory
Health & Insurance Records Systems
Disease
Cardiovascular Disorders (including MI, Stroke, Circulatory), Injury & Trauma