The intervention: The UHCE approach
A general template of the UHCE approach was developed by systematically reviewing the literature to identify evidence based interventions and validated assessment instruments for frailty, fall risk and polypharmacy (see www.uhce.eu). Furthermore, focus groups and interviews with main stakeholders (older persons, health and social care professionals, caregivers and policy makers) were held to identify their demands and preferences regarding the UHCE template, which led to the decision to address loneliness as a separate health problem, in addition to frailty, fall risk and polypharmacy.
UHCE starts with a frailty assessment and an assessment of fall risks, polypharmacy and loneliness in order to identify priorities for prevention and care of the older persons participating (Fig. 1). We use validated instruments that are practical and commonly used in a primary care setting. Fall risk is measured following a validated protocol developed by the Dutch safety research institute . Polypharmacy is measured following the common definition of using of five or more different medicines , in addition we measure whether persons have difficulty to take the medicines as prescribed . Loneliness is measured with the social subscale of the Tilburg Frailty Indicator . Frailty is measured with the Tilburg Frailty indicator, which was made and validated for use in primary care and has been extensively researched [24, 25].
The results of the assessments are discussed with the older person, a person in charge of care coordination (nurse practitioner or other) and a physician (Fig. 1). As a result of this shared decision-making process, a decision on a care plan is made and each participant is referred to evidence based “care pathways” (interventions) that are described in the UHCE template and adapted to the context of each of the five participating cities. The main UHCE care pathways are: multifactorial fall prevention actions (which include physical exercise groups, home hazard identification or other actions based on the judgement of a physician), actions addressing polypharmacy (which include appropriate prescribing and adherence action or other actions based on the judgement of a physician), actions addressing loneliness (which include support groups, social activities, or other actions based on the judgement of a physician) and frailty action (which include group based exercise programs or other actions based on the judgement of a physician).
The care coordinator (a nurse, social worker or trained physician assistant) coordinates and monitors the progress of each individual care plan under the supervision of the physician (Fig. 1). Follow-up visits are scheduled if needed. The care coordinator monitors the compliance to the care plan. The general UHCE template is adjusted in accordance with national standards and the local setting of the five participating cities.
Design, setting and procedures
The evaluation of UHCE has a specific pre-post controlled design . Intervention and control sites (general practitioner; GP practices or primary health centres; PHC) are chosen based on their location in distinct neighbourhoods in the participating cities. Older persons in the catchment area of an intervention site receive an invitation by their physician to join the study in the area where the UHCE approach is applied. Older persons in the catchment area of a control site receive an invitation by their physician to join the study in the area where ‘usual care’ is applied. The study is performed in accordance with the capacity, organizational and contextual factors of each of the five participating cities, as described below.
Greater Manchester is a metropolitan county in North West England, with a population of 2.7 million persons . Participants are recruited through GPs of individual GP practices, the intervention practices are located in Tameside and Glossop districts and the control practices are located in South Manchester. Assessments are taken at the participant’s home by a trained assistant. The results are assessed by researchers and clinicians before being provided to the participant’s GP. The participant’s GP is responsible for decisions on care in collaboration with health and social care staff at the GP practice.
Pallini is a suburban town and a municipality situated in the eastern part of the greater Athens area, with a population of 54,415 persons . Participants are recruited through the Municipal Health and Social Services. In Pallini, as only city, participants are randomized (by using a random numbers table) into the intervention group (UHCE approach) and the control group (‘usual care’). Assessments are taken at three community centres and the Municipal Health Centre by trained health staff. A health professional or a social worker is the care coordinator, and a physician is responsible for decisions on care in collaboration with the nurse.
Rijeka is a port city located at the most western part of the Republic of Croatia and has a population of 128,384 persons . Participants are recruited through individual GP practices, intervention practices are located in the Western part of Rijeka and control practices are located in the Eastern part of Rijeka. Assessments are taken at the participant’s home by community nurses, who act as care coordinators. The participant’s GP, supported by the nurse is responsible for decisions on care.
Rotterdam is a port city in the Netherlands province of South Holland, with a population of 638,714 persons . Participants are recruited through their GP based in PHC, the intervention PHC is located in Ommoord neighbourhood and the control PHCs are located in the Oosterflank and Zevenkamp neighbourhoods. Assessments are taken at the participant’s home by a trained assistant. Results are then provided to a geriatric nurse, who is the care coordinator, in collaboration with the GP. The participant’s GP, supported by the nurse is responsible for decisions on care.
Valencia is a port city located on the Southeastern coast of Spain, it has a population of 800,666 persons . Participants are recruited through the GPs of the intervention PHC in the Nou Moles neighbourhood and the control PHC in the El Botanic neighbourhood. Assessments are taken at the participant’s home by a trained assistant who supports case management by the GP. The participant’s GP is responsible for decisions on care supported by the nurse and social workers of the health centres.
Study population and eligibility to participate in the study
We aim to include 250 participants in both the intervention group and 250 participants in the control group in each city. In total, 1250 participants are included in the intervention group and 1250 participants in the control group. In each city, the target population consists of persons living independently, aged 75 years or more, who are, according to their GP, expected to be able to participate in the study for at least 6 months. Persons are not eligible to participate if they are not able to comprehend the information provided in the local language or if they are not able to cognitively evaluate the risks and benefits of participation and are not expected to be able to make an informed decision regarding participation in the study, according to their GP or physician. If possible, the participant is invited to designate an informal caregiver to support him or her, such as the partner, a child, sibling, friend or neighbour.
Data-collection and measures
Data collection is done with the use of a questionnaire; which includes the UHCE assessment (described above), outcome and other measures. Two non-invasive measurements (hand-grip strength and mid-upper arm circumference) are additionally performed and written down in the questionnaire. These data are collected at both baseline (T0) and after 12 months (T1). The instruments used for the outcome measures are described in the measurements section. The instruments and items for which no validated translations are available are translated forward and backward by translators. Forward- and back-translations are discussed by the study team and translation is adapted when needed.
Both general outcome health measures and specific outcome health measures applicable to each care pathway are applied: healthy life styles, fall risk, appropriate medication use, loneliness, frailty, level of independence and health-related quality of life. Healthy life style is measured with one item on physical activity, two items on smoking, and three items of the AUDIT-C  on high-risk alcohol use. Fall risk is measured by two items on (the number of) falls in the previous year, a single item asking whether or not the person is afraid of falling, and fear of falling while performing several daily activities as measured by the 7-item Falls Efficacy Scale International (FES-I) short version . Appropriate medication use is measured with 10 items of the Medication risk questionnaire (MRQ-10) , a tool developed for use by older persons to identify who is at increased risk of potentially experiencing a medication-related problem. Loneliness is measured with the short 6-item version of the Jong Gierveld loneliness scale , which measures the degree of what one wants and what one has in terms of interpersonal affection and intimacy. Frailty is measured with the 15-item Tilburg Frailty indicator [24, 25], that includes questions on physical, psychological and social components of frailty. Physical frailty is additionally measured with the SHARE-Frailty instrument which is an instrument that was developed and validated in a European population [35, 36], SHARE-frailty includes hand-grip strength measurement, and physical frailty is also measured with a measurement of the mid-upper arm circumference , a measure for malnutrition. Level of independence is measured with the Groningen activity restriction scale , that includes 18 items on independence of activities of daily living (ADL) and instrumental activities of daily living (IADL) and additionally with the one-item Global Activity Limitation Index (GALI) [39, 40]. Health-related quality of life is measured with the 12-item short-form (SF-12) [41, 42] and the full 5-item mental health scale of the SF-36 .
Additionally to health measures, use of care is measured with four questions regarding the use of doctor appointments, household work, help caring (such as washing or dressing) and hospital admissions.
Various socio-demographic characteristics are measured: age, gender, country of birth, educational level, income, marital status, employment situation, household composition and religion. Additionally, several questions on the participant’s general health are asked: self-reported height and weight, use of walking or other aids, whether they ever have been diagnosed with any of fifteen listed health conditions. Any additional remarks can be left in an open box at the end of the questionnaire.
A process evaluation is used to monitor program implementation and help to understand the relationship between the delivery of specific UHCE approach elements and program outcomes. We based our design of the process evaluation on the theoretical framework for public health interventions as developed by Steckler and Linnan [44, 45]. The following elements are included and outlined below: reach, dose delivered and received, fidelity, satisfaction, and context.
This process element aims at measuring the proportion of the intended target population that is reached by the care approach. In UHCE we calculate which proportion of the participants that we contacted participate in the UHCE approach. If possible reasons for refusal are reported.
Dose delivered and received
Dose delivered measures whether the anticipated care is offered to the participant and dose received measures the extent in which participants actively engage in the care that is offered. For this purpose, data on the delivery of the UHCE approach and reasons for non-participation in care are recorded by the care coordinator. A 1-page logbook is kept for each participant that includes the different stages of the UHCE approach: 1. assessment, 2 shared decision making, and 3. care pathways and monitoring. At least after 6 months, the care coordinator records (if needed telephone contact with the participant or responsible health care provider is sought) whether key elements of the UHCE approach are delivered. In case of loss to follow- up at the T1 measurement, reasons are recorded by the research staff, given they are provided by the participant.
Fidelity and satisfaction
We aim to measure the extent to which the UHCE approach is implemented as planned (fidelity) and the satisfaction of main stakeholders with the UHCE approach. In the T1 questionnaire, three items measure the general satisfaction with professional, social and self-care in the past 12 months, 4 items measure the satisfaction with specific UHCE elements (assessment, shared-decision making and care-pathways) and a final question rates the whole UHCE approach on a scale from 1 to 10. To gain more in-depth knowledge on how the UHCE approach is carried out and which barriers are encountered, focus groups and semi-structured interviews with older persons, caregivers, and social and health professionals involved in UHCE are held 12 months after inclusion of the last group of participants. In each city we organize 1 focus group with 6–8 older persons and caregivers (e.g. family, friends) and 1 group with 6–8 social and health care professionals. All professionals involved with UHCE are invited to participate in the focus groups. The focus group discussions are recorded and translated in English.
As the UHCE approach is implemented in five diverse settings its’ success depends on the context in which it is implemented. With the use of structured forms we make an inventory of relevant contextual factors of each city in which the UHCE approach is embedded; type and experience of health staff, setting, resources and interventions available or newly developed.
In each of the five cities, 250 participants are included in the intervention group and 250 participants in the control group. Assuming a 20% loss to follow-up between T0 and T1 due to mortality, rehousing or impossibility to participate, we expect to receive complete data of 2000 participants at follow up, equally divided over the intervention group and the control group. We assume equal standard deviations in the intervention group and the control group, alpha of 0.05 and power of 0.80. Given 5 participating cities with each an intervention and control group, we applied a correction factor to account for the cluster design, assuming an average cluster size of 200 older citizens (2000/10) and an intra-class correlation coefficient of 0.02. For this expected sample size and assumptions, with regard to the continuous outcome measures, a difference of 0.25 SD (standard deviation) between the intervention and the control group can be detected at follow-up. For example, regarding the health-related quality of life as measured by the Physical and Mental Component Summary Scale scores of the SF-12, a difference of 2.9 points can be detected for the Physical Component Summary Scale (SD = 11.4) and 3.0 points for the Mental Component Summary Scale (SD = 11.9) .
Data management and analysis
Data from all cities is combined and data-management and analysis is done at Erasmus MC. Paper questionnaires are scanned and automatically transferred into electronic files. Paper participant logbooks are entered into an electronic data-entry form by research staff. Electronic data is checked for missing or incorrect data. If an error is present in the electronic data, scans of the paper questionnaires and logbooks are consulted. If needed responsible staff are contacted for clarification.
Descriptive statistics are used to summarize characteristics of participants in each city and in the total study population. Participant socio-demographic characteristics (age, gender, income, educational level) are compared at baseline between the intervention and control group of each city and in the total study population. A multilevel modelling approach is used to examine differences in the outcome measures between the intervention and control group, taking into account the clustering effects at the city-level. Multilevel linear regression analyses are conducted for the continuous outcome variables with group (intervention or control) as independent variable and baseline values and potential confounders as covariates. Multilevel logistic regression is performed for dichotomous outcome variables. Subgroup analyses are conducted by means of formal interaction tests for intervention and those variables which are likely to influence the effect of the intervention itself: gender, age and educational level. In addition, subgroup analyses are done for subgroups of individuals with an indication for specific care pathways (frailty, fall risk, polypharmacy and loneliness), comparing participants with this indication in control and intervention groups. In addition, the above mentioned analyses are repeated for each city separately.
All qualitative data (interviews and focus groups) are recorded, transcribed and translated to English. Thematic analysis of data is done using a pre-defined coding framework which is developed through discussion and consensus among the research team .
We have set up an Advisory Board with experts from six EU countries. The role of the Advisory Board is to provide a critical perspective throughout the project. The scientific project results are disseminated by the project team through publications in scientific journals and conferences. To further disseminate the knowledge to all stakeholders we use the project website (www.uhce.eu). The European Local Inclusion and Social Action Network (ELISAN) is one of the partners of the UHCE project and aids the dissemination of project results to all stakeholders via social media.