Design and setting
A single-blind, three-armed, cluster-randomized controlled trial with a one-year follow-up is being conducted (see Figure ). Recruitment was performed in three primary care networks with almost 70 practices in Utrecht, the Netherlands.
Selection of patients is performed by the U-PRIM system, a software application that is installed in all participating general practices. Exploring the electronic medical records (EMRs) in each general practice, U-PRIM will screen for three inclusion criteria in patients aged 60 years or older:
- Multimorbidity (defined as a frailty index score of ≥ 0.20; see the 'U-PRIM intervention' section)
- Polypharmacy (defined as the chronic use of five or more different medications) [20
- Care gap in primary care of three or more years (defined as not having consulted the GP in the past three years, except for the yearly influenza vaccination).
Terminally ill patients or patients living in an elderly home or nursing home are excluded. Reasons for exclusion are registered on the general practice level.
At the start of the inclusion period, U-PRIM automatically generates a list of frail patients of 60 years and older in every participating practice. Using the U-PRIM software, data extractions from the electronic medical records (EMRs) in the practices are uploaded to an external server area. Here, reports on frail patients are generated and delivered back to the general practice. To guarantee patient privacy, U-PRIM software encodes the personal data by means of a third trusted party procedure, so personal data are only disclosed to the general practice personnel.
Eligible patients are listed in the first U-PRIM report. These patients are approached by their GP with a patient information letter and informed consent form for participation in the U-PROFIT trial. In addition, patients are asked if they have an informal caregiver. If so, the caregiver is also invited to participate in the study to investigate caregiver burden.
In the practices in the control group, a similar U-PRIM report with potentially frail patients is generated, but this report is not visible to the GP.
The U-PROFIT trial is approved by the Institutional Review Board of the University Medical Center Utrecht (UMCU) with protocol ID 10-149/O and registered in the Netherlands Trial Register: NTR2288.
Randomization and blinding
The participating general practices are randomly allocated to one of the two intervention groups (A or B) or the control group (C) by cluster randomization on the general practice level (see flowchart). Practices in group A are allocated to the U-PRIM intervention, those in group B to the U-PRIM plus U-CARE intervention and the practices in group C formed the control group. Within the 58 participating general practices, clusters are created because some general practices are working closely together at the same location. Before randomization, clusters are stratified according to the expected number of frail older people in the general practice. The cluster size is estimated based on the number of invitations for the yearly influenza vaccination per practice.
A modified informed consent procedure is used to maintain a single-blind design; the so- called "consent to postponed information" [21
]. With this procedure, a valid assessment of subjective outcomes can be obtained in a trial even if the patients cannot be blinded to the intervention. Additionally, selection bias and dropout in the control group can be reduced. In the U-PROFIT trial, patients were not informed as to which intervention group their general practice was allocated until the end of the follow-up period.
Blinding of the GPs and practice nurses
Blinding the GPs and their practice nurses is not possible in this study because they are part of the intervention.
Blinding the investigators
Because the investigators need to directly communicate with the general practices about the study, it is not possible to blind the investigators. However, during data analysis, investigators will be blinded to the data. When the data analysis is completed, this information will be disclosed to the investigators.
Two interventions are being tested in the U-PROFIT trial: 1. Screening and Monitoring of frailty (U-PRIM) and 2. Nurse-led multidisciplinary intervention program (U-CARE).
Intervention 1: U-PRIM
The U-PRIM software application is an electronic monitoring system aiming at identification of older patients at increased risk of frailty in routine health care data. The software is based on periodic screening for relevant risk factors in the EMRs of the general practice.
U-PRIM screens for three core risk factors in patients aged 60 years or older. These are also the eligibility criteria of the U-PROFIT trial as described earlier (multimorbidity, polypharmacy and a care gap).
The frailty index concept is used as an indicator of multimorbidity [23
]. The frailty index uses 50 so-called 'health deficits': symptoms, signs, diseases, social problems and functional impairments, all routinely encoded in the EMR using International Classification of Primary Care (ICPC) codes (see Additional file 1
). In the choice of the deficits, we followed previously published guidelines for the construction of a frailty index [24
U-PRIM assesses the number of deficits in each individual. The frailty index score expresses the number of deficits present as a proportion of the total number of deficits [25
]. Thus, a patent with 15 deficits has a frailty index score of 0.30 (15/50). For this study, multimorbidity based on the frailty index alone is defined as a frailty index score of ≥ 0.20 [26
The U-PRIM software screens the medication list for chronic drug use, using anatomical therapeutic chemical (ATC) codes. Chronic use is present when the medication was prescribed at least three times in the past year, with at least one prescription in the last six months. Polypharmacy is in this study is defined as 5 or more different drugs in chronic use in the past year [20
The period that patients are out of sight of their GP is assessed to include possible care avoiders prone to self-neglect, for example patients with dementia, psychiatric conditions or alcohol abuse [27
]. For this study, a "care gap" is defined as a period of at least 3 years without GP consultation, excluding the annual influenza vaccination.
The U-PRIM procedure
In the U-PROFIT trial, the periodic U-PRIM frailty screening of the trial population takes place every three months in intervention groups A and B. This results in a U-PRIM report for each general practice with a selection of older patients at high risk of adverse health outcomes. Patients are prioritized by means of the frailty index score, with possibilities to prioritize according to polypharmacy or care gap. For an example of a U-PRIM report, see Additional file 2
The report will be passed on to the GP in intervention groups A and B. In group A, GPs are asked to act upon the U-PRIM report in accordance with current available guidelines and best practices and to carry out interventions among the frail elderly patients if needed. In group B, all patients selected by U-PRIM will receive the additional steps of the U-CARE program (see intervention 2). In every participating practice in group A and B, a staff member is responsible for generating the reports with the U-PRIM computer program and for distributing the report among the care providers involved. These contact persons received protocolized, one-on-one guidance with the first U-PRIM report, with an explanation of the software application and suggestions on how to implement the report in daily clinical practice.
Intervention 2: U-CARE program
U-CARE is a nurse-led, multidisciplinary intervention program to be used in frail patients selected by U-PRIM. Specially trained, registered practice nurses provide structured and integrated care based on a patients' needs approach.
U-CARE is developed by a multidisciplinary team consisting of researchers and practitioners in nursing and primary care medicine. Three experienced practice nurses, a panel of experts and a panel of older people are involved to validate the content.
The program consists of three steps. The first step is a frailty assessment for patients at risk. The second step is a comprehensive geriatric assessment (CGA) at home of frail patients. The third step is a tailor-made care plan with evidence-based interventions developed by the practice nurse. Details of the development and the content of the program are described elsewhere [Bleijenberg et al: Development of a nurse-led multidisciplinary intervention program (U-CARE) to prevent functional decline in frail older people in primary care, Unpublished].
Step 1. Frailty assessment
The level of frailty in patients at risk selected by U-PRIM will be further explored with the Groningen Frailty Indicator questionnaire (GFI). The GFI is a 15-item validated questionnaire that assesses frailty from a functional ADL/IADL perspective on four domains: physical, cognitive, social and psychological [28
]. Scores on each item are zero or one, and the total score ranges from 0 (not frail) to 15 (severely frail). We chose a score of 4 or higher as the relevant cut-off [29
] for the selection of patients that should be visited for a comprehensive geriatric assessment. The GFI has shown high internal consistency and construct validity [30
]. This questionnaire will be sent to all patients selected by U-PRIM.
The INTERMED for the Elderly (IM-E) [31
] and the Groningen Wellbeing Indicator (GWI) are additional assessments included in U-CARE to enable a multidimensional approach and to measure patients' needs and complexity of care among frail patients on the GFI.
Step 2. Comprehensive Geriatric Assessment at home (CGA)
For those patients identified as being frail, a CGA at home is conducted by a registered practice nurse. During this home visit, the practice nurse focuses on patients' health problems and needs in a structured manner based on the outcome of the frailty assessment. Based on the literature and their prevalence [32
], ten health problems in older patients with additional assessments are included in the CGA (see Additional file 3
Step 3. Tailor-made care plan
In collaboration with the GP, the practice nurse will prepare a tailor-made care plan based on the outcome of step 2. This tailor-made care plan consists of interventions derived from evidence-based care plans developed by the research team, practice nurses and experts. For all ten health problems assessed in the CGA, separate evidence-based care plans are developed. The use of the care plan ensures uniformity among practice nurses in tailoring and delivering interventions per health problem. Flowcharts with suggested (nursing) interventions per health problem are developed as a practical tool and will help to guide the practice nurses through a structured process of decision making.
All practice nurses will receive an extended U-CARE training program that consists of 5 weeks of 4 hours of lessons in class and 4 hours of self-study. During this training program, the included frailty assessments, the content of the CGA and the evidence-based care plans will be discussed. The U-CARE training program is set up in collaboration with the University of Applied Science Utrecht in the Netherlands.
One month prior to the start of the trial, all GPs and registered practice nurses from intervention group are participating in a training session of 4 hours in which the content of U-CARE program is explained and discussed. Additionally, a workshop about collaboration between GP's and practice nurses is set up.
Outcomes and measurements
The primary outcome of the U-PROFIT trial is the level of Activities of Daily Living (ADL) as measured with the Katz ADL index score [35
]. The Katz index measures independence of ADL on six items (bathing, dressing, toileting, transferring, eating and the use of incontinence materials). The score ranges from 0 (total independence) to 6 (total dependence), and it is widely used to assess activities of daily living [36
]. Baseline ADL functioning (T0) will be compared with ADL functioning after six months (T1) and one year of follow-up (T2). The questionnaire will be filled in by the patient or a proxy relative.
Secondary outcome parameters will be measured at the same time as the primary outcome parameter (T0-T1-T2).
1. Quality of life will be measured with RAND-36 and EuroQol (EQ-5D) [37
3. Number of nursing home admissions
4. Number of emergency department and out-of-hours GP surgery visits
5. Caregiver burden measured with Self-Rated Burden (VAS) and Carer-Qol [39
Additional data collection
Routine health care data will be extracted from the EMRs of the participating practices. Socio-demographic data, such as age, gender, educational level, ethnicity, marital status and living situation, will be gathered at baseline. General practice characteristics, such as size, percentage of older people, working experiences and geographical location of the general practice, will also be gathered.
To understand the different components, their interaction and the applicability of the U-CARE program, a feasibility study will be conducted among doctors and practice nurses of intervention group B. Furthermore, interventions delivered by the practice nurse or other health care providers will be registered to gain insight into targeted interventions that are performed by the practice nurses.
The U-PRIM system will be evaluated on psychometric properties, prognostic value for adverse health outcomes and in concordance with the GFI, and the system will be refined following a user demands study.
In addition, qualitative data on patients' satisfaction with the U-CARE program will be qualitatively assessed. In the end, various data will be collected to perform a cost-effectiveness analysis, e.g., data on workload of the GP and practice nurses and time registration.
Sample size calculation
At present, a valid estimation of the variance in the KATZ ADL results within and between general practices cannot be given because these data are not available for Dutch populations. For that reason, a formal power analysis for the cluster-randomized trial is not possible. Therefore, it is also not feasible in this study to take into account a potential cluster effect. In line with Faber et al., [40
] we assume that any randomization effect per practice will be absent. Furthermore, we assume that with an expected number of at least 5000 frail older people included, relevant effects can be found on the outcome between the clusters because the power of a trial increases if the number of clusters, subjects, or repeated measures within a subject increases.
The data will be analyzed using SPSS version 17.0. An 'intention to treat' analysis will be carried out to assess the differences between the intervention groups and the control group regarding ADL functional status. The variations in outcome between the groups will be calculated using mixed linear model analysis. Regression analyses and (co)variation analyses will be carried out when relevant to correct for baseline differences between older people in the three groups. Survival analysis using a Cox regression model with Kaplan-Meier survival curves will be used on mortality and admission into nursing homes. As social economic status (SES), gender, age and education are assumed to be potential effect modifiers, subgroup analysis will be applied where relevant. We will also take the working experience of the participating GPs and practice nurses into account in separate analyses.