Study design and participants
This was a prospective observational study conducted at the Geriatrics and Palliative Care Department of Osona (Barcelona, Spain) between July 2018 and September 2019. This department provides care to a catchment population of 156,000 inhabitants through a domiciliary care service (hospital at home unit) and two IC hospitals (University Hospital of Santa Creu [Vic, Spain], and Hospital Sant Jaume [Manlleu, Spain]; 170 hospital beds in total), which include a palliative care unit, a rehabilitation unit, a psychogeriatric unit, and a mixed unit. Approximately half of the patients are transferred from the reference acute hospital, and the other half comes from regional hospital emergency rooms and/or from home upon the general practitioner’s request.
All individuals aged ≥ 75 years admitted to the IC facilities within the study period were consecutively offered to enrol in the study. Younger patients were also included if they met the criteria of chronic complex patients (PCC, Pacient Crònic Complex), advanced chronic disease (MACA, Malaltia Crònica Avançada), according to Catalan Health Department criteria . Individuals who could not have been followed up at home 30 days post-IC discharge for geographical reasons were excluded from the study.
The study protocol was approved by the Ethics Committee of the University Hospital of Vic (2,018,958/PR189), and all participants provided written informed consent to participate in the study. The study results are reported according to the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines .
Variables and data sources
The primary outcome was the change between different degrees of frailty as assessed with a frailty index, which is a sensitive tool for measuring frailty change with good ability to predict mortality and other adverse health outcomes [21,22,23]. For this study, we chose the Frail-VIG index, which consists of 22 dichotomous questions that allow identifying 25 deficits from various domains (including socioeconomic status, estimated on the basis of the social history and/or other relevant available information) and has been shown to be a reliable, feasible, and valid tool to assess the degree of frailty in hospitalized older people . The Frail-VIG index also has a good discriminative capacity for the degree of frailty and a high predictive ability for mortality [25,26,27,28].
The degree of frailty was assessed at four time points: baseline status at 30 days before admission to the IC facility (Frail-VIG0), within the first 48 h after admission (Frail-VIG1), at discharge (Frail-VIG2), and at 30 days post-discharge (Frail-VIG3). Frail-VIG scores were obtained by hospital health professionals (doctors and nurses), who were already trained in the use of the Frail-VIG index as it is used in routine clinical practice in the local setting. Frail-VIG0 was retrospectively assessed by anamnesis of the patient and/or main caregiver, who reported on the patient’s status approximately one month before IC admission. The Frail-VIG3 score was obtained by research nurses during home visits. Two of the original Frail-VIG index items were tailored to the study by removing references to temporality (the final version and changes introduced to the Frail-VIG questionnaire are provided in the Supplementary Appendix).
For the comparative analysis between frailty groups at baseline, patients were grouped into four categories, as described elsewhere: [26, 27] no frailty (Frail-VIG index score < 0.2); mild frailty (Frail-VIG index score 0.2–0.35); moderate frailty (Frail-VIG index score 0.36–0.5), and advanced frailty (Frail-VIG index score > 0.5).
Sociodemographic variables included age, sex, and current place of residence. Clinical variables included diagnoses and clinical conditions on IC admission, assessed according to the items of the Frail-VIG index [24, 25] as described in the Supplementary Appendix, date of discharge or death, and all variables included within the Frail-VIG index. We also collected resource use information, including the IC admission unit (i.e., palliative care unit, rehabilitation unit, psychogeriatric unit, mixed unit, and domiciliary care unit), and the length of stay in IC. The instrumental activities of daily living (IADL) were selected from the Lawton-Brody scale considering the items with lower risk of gender bias, as described previously .
Categorical variables were described as frequency and percentage over available data, whereas quantitative variables were described as the mean and the standard deviation (SD), without imputing missing data. The association between clinical and demographic characteristics of patients in each frailty degree were assessed using the chi-square test for categorical variables and t-test for quantitative and categorical variables. The transition probabilities between the different levels of frailty/death corresponded to the proportion of patients experiencing a given transition. Finally, the relationship between frailty and the risk of mortality was assessed with a Cox proportional hazards model using the frailty measure at baseline and IC admission. For all hypothesis tests, the significance threshold was set at a two-sided alpha value of 0.05. Descriptive analyses and comparisons regarding demographic and clinical variables were computed using SPSS (Version 28.0. Armonk, NY: IBM Corp.), whereas survival analyses were performed using the Survival and msSurv (multi-state models) packages from the R project . Transitions between frailty levels were plotted using gplot2 and ggalluvial packages from the R project.