Association of a Digital Clinical Decision Support Platform With Early Outcomes After Pediatric Allogeneic Hematopoietic Cell Transplantation: A Three-Era Guideline Implementation Cohort Study.
AI interpretation is pending for this paper.
Open original publication →What the AI sees
Not AI summarized yet.
Research significance
Pending deeper interpretation.
Source abstract
Practice variation in supportive care contributes to potentially preventable morbidity after pediatric allogeneic hematopoietic cell transplantation (HCT). Digital clinical decision-support platforms may improve care standardization, but outcome data after sustained implementation remain limited. To evaluate whether sustained implementation of a digital bedside transplant guideline platform was associated with changes in early clinical outcomes after pediatric allogeneic HCT. We conducted a single-center retrospective cohort study of first allogeneic HCTs across 3 eras: preplatform (2013 to 2018; n = 200), wash-in (2019 to March 2023; n = 111), and sustained-use (April 2023 to 2025; n = 78), defined by stable clinical adoption. Main outcomes of interest were adjusted inpatient length of stay (LOS) and nonrelapse mortality (NRM), with relapse treated as a competing event. LOS was analyzed using gamma regression with log link. NRM was analyzed using cumulative incidence functions and Fine-Gray regression. Models used robust standard errors and adjusted for age, sex, malignant indication, HCT-CI ≥3, conditioning intensity, cord graft, and non-MSD donor. Sensitivity analyses for NRM additionally adjusted for malignant case mix and antithymocyte globulin timing. In the 3-era LOS model, wash-in was unchanged versus preplatform (ratio 1.006; 95% confidence interval [CI], 0.876 to 1.136; P = .927), while sustained-use was 14% shorter (ratio 0.858; 95% CI, 0.749 to 0.966; P = .018). Adjusted mean LOS was 53.32 days (95% CI, 48.82 to 57.82) in the preplatform era, 53.64 days (95% CI, 48.18 to 59.10) during wash-in, and 45.74 days (95% CI, 41.41 to 50.08) during sustained use. NRM cumulative incidence at 24 months was 0.116 in sustained-use versus 0.147 in preplatform (12 months: 0.087 versus 0.105), but 24-month estimates for the sustained-use cohort were interpreted cautiously because only 14 patients remained at risk at 24 months. Fine-Gray models comparing sustained-use versus preplatform showed lower NRM (subdistribution hazard ratio [SHR], 0.259; 95% CI, 0.062 to 1.080; P = .064), strengthening after adjustment for malignant case mix (SHR, 0.212; 95% CI, 0.054 to 0.826; P = .025) and ATG timing (SHR, 0.193; 95% CI, 0.047 to 0.794; P = .023). There was no evidence of increased intensive care unit admission (odds ratio [OR], 0.73; P = .391) or 30-day readmission (OR, 1.30; P = .457). Overall survival did not differ across eras (log-rank P = .743). Sustained adoption of a digital bedside HCT guideline platform was associated with shorter hospital LOS and a consistent signal toward lower NRM without adverse utilization or survival outcomes. The absence of similar improvement during wash-in supports, but does not prove, an implementation-dependent association. Prospective multicenter evaluation with patient-level platform exposure and adherence metrics is warranted.