**Background:** Dysphagia affects 27%–64% of stroke patients and can lead to malnutrition, aspiration pneumonia, longer hospital stays, and increased mortality. Enteral feeding via percutaneous endoscopic gastrostomy (PEG) or nasogastric tube (NGT) is often required, but many patients eventually recover swallowing function. Identifying which patients will regain complete oral intake (COI) is critical for appropriate feeding tube selection. Previous studies were limited by single-center designs or short follow-up periods, and few assessed outcomes at discharge from convalescent rehabilitation wards (CRWs).
**Methods:** This retrospective case-control study used data from the Kaga Regional Cooperation Clinical Pathway for Stroke (KRCCPS) database, which includes 19 acute care hospitals and 11 hospitals with CRWs in Ishikawa Prefecture, Japan. Data were collected over 10 years (January 2011–December 2020). Patients were included if they were dependent on enteral feeding via PEG or NGT upon admission to a CRW. Exclusion criteria included transfer to acute care, death in CRW, epidural/subdural hematoma, dysphagia due to comorbidities, or enteral feeding before stroke. Patients were divided into COI (FOIS ≥4 at discharge) and ICOI groups. Logistic regression with forced-entry variables identified independent predictors of COI. All patients received swallowing training from speech-language-hearing therapists plus physical and occupational therapy (up to 180 min/day).
**Key Results:** Of 5502 patients in the database, 380 were tube-dependent on CRW admission; after exclusions, 347 patients were analyzed. At discharge, 140 achieved COI and 207 had ICOI. The COI group was significantly younger (median 74 vs. 79 years, p<0.01), had more initial strokes (80.7% vs. 67.1%, p<0.01), higher FOIS scores (median 2 vs. 1, p<0.01), higher FIM motor scores (median 13 vs. 13, p<0.01), higher FIM cognitive scores (median 11 vs. 6, p<0.01), higher BMI (median 21.5 vs. 20.0 kg/m², p<0.01), lower PEG rate (26.13% vs. 50%, p<0.05), and shorter acute care stays (median 34 vs. 44 days, p<0.01). Logistic regression identified six independent predictors of COI: younger age (OR 0.96, 95% CI 0.93–0.98, p<0.01), initial stroke (OR 3.05, 95% CI 1.59–5.84, p<0.01), higher FOIS score (OR 3.25, 95% CI 2.04–5.16, p<0.01), higher FIM cognitive score (OR 1.11, 95% CI 1.04–1.17, p<0.01), higher BMI (OR 1.16, 95% CI 1.06–1.28, p<0.01), and shorter acute care stay (OR 0.97, 95% CI 0.96–0.99, p<0.01). FIM motor score and feeding tube type (PEG vs. NGT) were not significant predictors. Notably, 44.58% of NGT patients achieved COI vs. 29.59% of PEG patients.
**Clinical Implications:** This large, multi-center study provides clinicians with evidence-based factors to predict which dysphagic stroke patients are likely to regain oral feeding during convalescent rehabilitation. Younger patients with first-ever strokes, better preserved swallowing and cognitive function, higher BMI, and shorter acute care stays have the best prognosis for COI. For patients predicted to achieve COI, temporary NGT feeding may be sufficient, avoiding unnecessary PEG placement. Conversely, patients with poor prognostic factors may benefit from earlier PEG insertion to prevent complications of prolonged NGT use. The finding that NGT placement itself does not impede swallowing recovery is clinically important, as it supports initial NGT use even when recovery is uncertain. Limitations include the retrospective design, non-randomized tube allocation, and inability to assess sarcopenia or medication effects.