Table 4. Association between the clusters and outcomes.
| Outcomes | Post-stroke rehabilitation clusters | p-value | p-value | ||
|---|---|---|---|---|---|
| Lower Complexity with Caregiver (N = 169) | Moderate Complexity without Caregiver (N = 101) | Higher Complexity with Caregiver (N = 114) | |||
| ANCOVA (adjusted by sex) | LINEAR REGRESSION | ||||
| Functional improvement | 21.6±29.0 | 18.2±25.4 | 8.6±18.6 | <0.001a,b | 0.007a,b |
| Relative functional gain | 0.4±0.6 | 0.40±0.8 | 0.2±0.4 | 0.033a | 0.156a,b |
| Length of stay | 58.02±43.1 | 68.7±40.8 | 60.5±52.6 | 0.189 | 0.361c |
| Rehabilitation efficiency | 0.47±1.3 | 0.4±0.8 | 0.1±0.6 | 0.005a | 0.064a,b |
| CHI-SQUARE (linear trend) | LOGISTIC REGRESSION | ||||
| New Institutionalization | 17.8 (28) | 27.2 (25) | 34.6 (36) | 0.008a | 0.144 |
Values are report as mean±SD or percentages (N) for continuous or dichotomous outcomes, respectively. Functional improvement was calculated as BI at discharge minus BI at admission; Relative functional gain was calculated as Functional improvement divided by (pre-stroke BI minus BI at admission); Rehabilitation efficiency was calculated as Functional improvement divided by Length of stay. ANCOVA models were adjusted by sex; multivariable regression models (linear regression for all the outcomes but logistic regression for new institutionalization) were adjusted by sex, type of stroke, dysphagia, beginning of rehabilitation in the acute hospital. Differences according to post-hoc Bonferroni analysis after ANCOVA model, and contrasts between clusters in logistic regression models showed:
a Difference between “Lower Complexity with Caregiver” and “Higher Complexity with Caregiver”, p <0.05.
b Difference between “Moderate Complexity without Caregiver” and “Higher Complexity with Caregiver”, p <0.05.
c Difference between “Lower Complexity with Caregiver” and “Moderate Complexity without Caregiver”, p <0.005.