Study design and setting
The AGe-FIT was a prospective, randomized, controlled and assessor-blinded, single-center trial involving two parallel groups. Data collection took place between February 2011 and December 2013. Participants were enrolled continuously during the whole year of 2011. Outcomes were assessed at baseline, 12 and 24 months after inclusion date. The study was conducted in Norrköping, a municipality in southern Sweden with 130,000 residents. In 2010, about 8% of residents were aged ≥ 75 years. The county council of Östergötland provided tax-funded healthcare to all residents at 10 primary care centers and a hospital with 300 beds and 24-h admittance for emergencies. The Geriatric Department at the hospital had a memory clinic and an orthogeriatric ward. The study protocol and results on the primary outcome (number of hospitalizations), other health-related outcomes and cost effectiveness have previously been published [14,15,16,17].
Participants and randomization
Community-dwelling persons aged ≥ 75 years, who had been hospitalized three or more times in the previous year, and had three or more current medical diagnoses according to the International Classification of Diseases, 10th Revision (ICD-10), were eligible for study participation. Potential participants (n = 837) were identified using a population-based administrative database maintained by the county council. Randomization to the intervention group (IG) and control group (CG) was performed using a random list generated by SPSS software (PASW Statistics 18; IBM Corporation, Chicago, IL, USA). A total of 382 (45%) participants were included in the IG (n = 208) and CG (n = 174).
Frailty assessments
Assessments at baseline and 24 months were conducted by blinded research nurses or an occupational therapist, not involved in the care of study participants. Frailty was measured using the criteria from the Cardiovascular Health Study (CHS): weight loss, weakness, exhaustion, slowness, and low activity level [18]. Measurements for slowness and low activity were measured differently from the original CHS criteria. At baseline, unintentional weight loss was assessed by participants’ estimation of weight loss in the previous year, and current weight was measured using a portable beam scale. The percentage of weight loss was calculated as estimated weight loss / estimated weight one year previously. At 24 months, current weight was measured and the proportion of weight loss was calculated using the value obtained at 12 months: (weight at 12 months – weight at 24 months)/weight at 12 months. A weight loss of ≥ 5% was the cutoff value for frailty. Weakness was measured in kilograms using a Jamar handheld dynamometer; the greatest value of two attempts with the dominant hand was used. The cutoff values were defined according to the weakest quintile in the study conducted by Fried et al. [18] Exhaustion was assessed by two statements from the Center for Epidemiological Studies Depression Scale (CES-D) (“I felt that everything I did was an effort” and “I could not get going”). Response options ranged from 0 to 3 (0 = rarely or none of the time, 1 = some or little of the time, 2 = moderately or much of the time, 3 = most or almost all the time) [19]. A response of alternative 2 or 3 to either statement (or both statements) fulfilled the frailty criterion. Slowness was evaluated using gait speed measurement over a distance of 4 m. Participants were instructed to walk at their normal pace, with a walking aid if used in daily life, and a time ≥ 5 s was categorized as slow. This cutoff value is recommended by the International Academy for Nutrition and Ageing (IANA) [20]. Participants’ activity levels were assessed using the International Physical Activity Questionnaire–Short Form (IPAQ-SF), a validated self-report instrument that summates the duration (minutes) and frequency (days) of different kinds of activity, leading to the categorization of activity level as high, moderate, or low [21, 22]. A low activity level fulfilled the frailty criterion. Participants were classified as robust (no criterion fulfilled), pre-frail (one to two criteria fulfilled), or frail (three or more criteria fulfilled), regardless of missing data.
Intervention and control procedures
Participants in the IG received the CGA-based intervention in addition to usual care. CGA-based care was provided at an ambulatory geriatric unit (AGU) opened specially for the AGe-FIT, in a real-life, non-academic setting. Patients were assessed according to a standardized procedure [16]. Initially, a nurse and a social worker went home to each participant and administered a survey of health, functional status and need for social care. A pharmacist collected information on compliance with the use of prescribed and non-prescription drugs by telephone. This information was conveyed to the physician, who consulted patients at their homes or at the AGU as part of the initial CGA-based evaluation. All information gathered was discussed at the following interdisciplinary team meeting; two such meetings were held per week. Decisions regarding interventions were made at these meetings, often involving the need for additional assessments, for example, by a physiotherapist, occupational therapist, and/or dietician. When needed, participants were referred to specialized medical care. Personalized care and follow-up plans were created and revised when required, and all participants were offered annual medical evaluations. The tailored interventions were often of preventive nature, such as reduction of polypharmacy, advice on exercise or diet, provision of adaptive equipment, and increased social support. Participants could reach nurses at the AGU directly by telephone during office hours.
Participants in the control group received medical and social care as usual. Healthcare was provided by the acute care hospital, 10 primary care centers, and at home. In most cases, it was initiated by the patients or their relatives.
Sample size
Sample size calculation for the AGe-FIT study was based on a two-tailed significance level of 5%, power of 80%, and an expected detectable effect over 24 months of a 20% reduction in primary outcome mean hospital admission rate, from five to four admissions per year. The result of a minimum of 142 subjects per arm was increased to 200, considering an estimated 40% attrition rate over the study period due to death, withdrawal, and relocation to nursing homes.
Statistical analyses
Pearson’s chi-squared test was used to compare categorical data between the IG and CG. Adjusted residuals (z scores) were calculated for each cell in the contingency tables and then converted into p values. The independent-samples t test was used for comparison of means of continuous variables. The significance level was set to p ≤ 0.05 for all analyses. Participants with measurements for frailty at baseline and at 24 months were included in the analysis. At 24 months, frail participants were combined with deceased participants, to minimize the risk of mortality bias and to describe the most negative outcomes. Analyses were also performed where frail and deceased participants were analyzed as separate groups. The analysis was made accordingly to the intention-to-treat (ITT) principle. Data were stored and analyzed using SPSS Statistics 22 (IBM Corporation, Armonk, NY, USA).