CONTEXT: The COVID-19 pandemic has placed a tremendous stress on healthcare systems and caused reorganization. As the pandemic intensifies, identifying the profile of patients with COVID-19 was primordial in order to predict negative outcomes and organize healthcare resources. Age is associated with COVID-19's mortality, but for obvious ethical reasons, chronological age cannot be the sole criterion for predicting negative outcomes. OBJECTIVE: The objective of this study was to determine the relationship between frailty index (FI) and length of hospital stay, and death in a non-COVID population of patients aged 75 years old and above. METHODS AND DESIGN: A retrospective, analytical, single-centered observational study was performed in the geriatric short-stay accommodation unit at Guadeloupe University Hospital. For this study, 158 patients who were at least 75 years old were recruited from November 2020 to May 2021. FI was calculated as the number of deficits in a participant divided by the total number of deficits considered (the cut-off of FI is.25 in outpatient). Multivariate logistics regression analyses were conducted to assess the association between frailty and death, and length of stay. RESULTS: The average age of the participants was 85.7 ± 6.74 with a range of 75-104. Twenty-four of the patients died during hospitalization. FI was only significantly associated with mortality even after adjustment for age and gender (HR 26.3, 95% CI 1.7-413.4, P = 0.021). The association was stronger in the highest tertile of the FI (age- and gender-adjusted HR 4.6, 95% CI 1.39-15.11, P = 0.01). There was no significant interaction between FI and length of stay. CONCLUSION: Our study shows an association between FI (in terms of age-related deficit accumulation) and mortality in a non-COVID geriatric short-stay unit in Guadeloupe. The FI seems to have a lower capacity to catch events such as length of stay in this very complex population. Further research studies have to be conducted for better understanding and investigation of our findings.