ABSTRACT NUMBER: ESOC2026A800 MULTICENTRE EVALUATION OF IN-HOSPITAL STROKE CARE PROCESSES USING RANDOM INTERCEPT AND RANDOM SLOPE-BASED INTERRUPTED TIME SERIES MODELS: THE IMPETUS STUDY

Abstract Background and aims High-quality in-hospital stroke-care facilitates improved patient-outcomes, yet variation exists in the implementation evidence-based practices. We evaluated temporal changes in in-hospital stroke-care process-indicators following a multicentre intervention of uniform stroke-care pathway, accounting for centre-level heterogeneity in baseline performance and trends. Methods Effect of implementing an evaluation-treatment-package at 23 hospitals (10,193 stroke-patients) was assessed through a prospective-quasi-experimental-design using mixed-effects interrupted-time-series-analysis on weekly-pooled hospital-level data. Outcomes were modelled as proportions using binomial mixed-effects segmented regression with fixed-effects for time, intervention, and post-intervention-slope. Center was included as a random-intercept, and random-slope for time, to account for heterogeneity in baseline-performance, and secular-trends across centres. Results Across 131 weeks, baseline adherence was high for assessing GCS, blood-pressure, positioning, swallow, rehabilitation-advice, sugar, but low for DVT, NIHSS, CTA, MRA, HbA1c, noting-of-complications, with significant between-centre variability. GCS had significant pre-intervention trend. Significant immediate post-intervention increase found for NIHSS (OR 1.40, 95%CI: 1.02,1.91), Holter (1.99, 95%CI: 1.22,3.24), HbA1C (1.66, 95%CI: 1.21,2.26) without a sustained change in slope. Significant change in post-intervention-trend seen for Rehabilitation-advice (8.82, 95%CI: 4.20,18.50) blood-pressure-assessment (140.0, 95%CI: 28.58,686.06), swallow-assessment (3.08, 95%CI: 1.10,8.64), DVT-assessment (4.66, 95%CI: 1.77,12.31) indicating accelerated improvement over time. Positioning-care, noting-complications, showed no significant intervention-effects. Large standard-deviation of random-slope implied centres improved or deteriorated at markedly different rates. Conclusions It is possible for uniform stroke-care intervention to achieve sustained post-intervention improvements in many in-hospital stroke-care processes, conditional on centre-level heterogeneity. However, some indicators failed to sustain post-intervention improvement despite achieving an immediate improvement which requires centre-level assessment of reasons. Conflict of interest Partha Haldar: nothing to disclose; Rohit Bhatia: nothing to disclose; Shweta Gupta: nothing to disclose; Vasantha Padma: nothing to disclose; Mamta Singh: nothing to disclose; Inder Puri: nothing to disclose; Vijay Sardana: nothing to disclose; Alok Verma: nothing to disclose; Samhita Panda: nothing to disclose; Biman Kanti Ray: nothing to disclose Table 1 - belongs to Results Figure 1 - belongs to Results Figure 2 - belongs to Conclusions