Background: Hospitalized patients on chronic inhaled therapy suffer high rates of therapeutic failure. Current approaches often overlook patient heterogeneity, treating failure as a uniform problem. We hypothesized that clinical inertia, a key driver of failure, is not a monolithic entity but is governed by specific, non-overlapping factors. Methods: In this unicentric, observational cohort study of 499 hospitalized adults on chronic inhaled therapy, we used unsupervised clustering to identify patient phenotypes. Multivariable logistic regression was used to model predictors of critical inhaler errors and three distinct forms of clinical inertia: Therapeutic Class (TCI), Device-Level (DLI), and Adherence-Related (ARI). Results: Inhaler misuse was driven by objective capability—deficient knowledge (aOR 6.03, 95% CI 2.88–12.64) and low inspiratory flow (aOR 3.11, 95% CI 1.06–9.12)—while patient-reported adherence was not a significant independent predictor. Crucially, the three forms of clinical inertia were governed by distinct, non-overlapping predictors: TCI was predicted by high therapeutic potency (aOR 7.80, 95% CI 3.65–16.64), DLI by a failure in the clinical process (lack of patient training, aOR 3.49, 95% CI 1.21–10.03), and ARI by regimen complexity (aOR 0.06, 95% CI 0.02–0.25). Post-discharge mortality (21.6% overall; 25.8% in Cluster 1 vs. 18.3% in Cluster 2) was independently predicted by objective risk factors, including older age (aOR 1.51, 95% CI 1.20–1.89) and an unassessed inspiratory flow (aOR 2.44, 95% CI 1.19–5.03). Two underlying patient phenotypes were identified—an “Unassessed/Older” (n = 225) and an “Assessed/Younger” (n = 274)—which represented distinct in-hospital care pathways but did not independently predict mortality after multivariate adjustment. Conclusions: Therapeutic failure in hospitalized patients is a predictable outcome driven by distinct, non-overlapping factors. This study deconstructs this failure by identifying the specific, actionable drivers of inhaler misuse (patient capability) and the three forms of clinical inertia (therapeutic potency, failures in the care process, and regimen complexity). These processes occur within two distinct patient phenotypes that represent different in-hospital care pathways. Our findings provide a new framework to move beyond generic interventions toward a more precise, evidence-based approach to inhaled therapy.

Background: Inhaler device changes at hospital discharge should address patient capacity yet often reflect routine. We evaluated the appropriateness of these decisions and their impact on clinical outcomes. Methods: In this prospective observational study (N = 480), we assessed patient technical capacity using a composite of critical errors, inspiratory flow, adherence, and knowledge. We stratified patients into ‘Need-Positive’ and ‘Need-Negative’ cohorts to quantify patterns of clinical inertia and over-adjustment. Multivariable models identified predictors of decision-making and associations with 30-day outcomes. Results: Device changes were primarily determined by the pre-admission device class (spacers: aOR 0.52; 95% CI 0.28–0.96; p = 0.037) and by the patient’s treatment pathway rather than by clinical need. This disconnect generated two types of errors: 38.3% of Need-Positive patients (n = 214) experienced clinical inertia (no corrective action), while 36.8% of Need-Negative patients (n = 266) underwent over-adjustment (unnecessary switching). Inertia perpetuated errors in patients with need, whereas over-adjustment was associated with the emergence of new errors in patients without need. Successful mismatch resolution was associated with a significantly lower 30-day readmission rate (12.1% vs. 32.5%; OR 0.48; 95% CI 0.26–0.88; p = 0.017). Conclusions: Discharge prescribing is driven more by habit than by objective assessment, leading to widespread missed opportunities for correction. Implementing evidence-based protocols to identify and resolve patient–device mismatches may represent a high-impact strategy to reduce readmissions and associated healthcare use.