The aim of this study is to compare the effect of 12 weeks group-based exercise program enhanced with 24 weeks health empowerment program (HEP) to an extended usual care condition (the HEP program alone) on physical function, disability, health-related quality of life (HQoL) in home-dwelling older adults at risk for disability. Interventions were implemented into existing health care pathways and added to routine preventive programs using a two-armed randomized intervention design with multiple sites.

Study design: a two-armed multi-faceted exercise intervention study with two phases: 12 weeks intensive phase followed by 12 weeks maintenance phase.

Intensive phase

• Active condition (exercise + health-empowerment): twice-weekly exercise + 4 health-empowerment sessions
• Comparison condition (health-empowerment) 8 health-empowerment sessions

Maintenance phase

• Active condition (exercise + health-empowerment): 4 health-empowerment sessions
• Comparison condition (health-empowerment): 4 health-empowerment sessions

Sampling method Danish home-dwelling older adults entitled to a nationally regulated preventive home-visit according to the Danish social act, living in three provincial municipalities in the Southern region of Denmark (Odense, Slagelse and Esbjerg) were invited to participate in the WIPP screening as an integrated part of the home-visit. The screening resulted in a risk-profile for functional loss and disability, on which eligibility for interventions was based. The interventions studied here were offered by the municipalities in line with existing services during the project period. In order to enable proper evaluations, citizens who volunteered to participate where randomly allocated to either of the two conditions by sealed randomization procedures. Citizens were informed about this prior to agreeing to participate. Towards the end of the project phase, exceptions from this procedure took place, to accommodate project interests. Tracking of allocation procedure is possible in the dataset. Subject recruitment and allocation, data collection and management as well as interventions were all run by the health care providers (i.e. municipalities).

Registry procedures and other quality factors On-site data collection (self-report and objective assessments) was led by the municipalities. The raw-data was registered in either paper- or digital format depending on the technical prerequisites (access to portable digital equipment and internet connection) of each site.

Paper-format data was subsequently digitalized by municipality staff.

Quality assurance of the validation and registry procedures primarily consisted of three elements:

1. Written instructions on how to fill out the individual parts of the data registration were available during the on-site registration, regardless of whether this was done electronically or in paper format. The information was also available during the final registration into the database.
2. The written instructions were further qualified by a verbal introduction on how to use the data registry platform and how to fill out the individual parts correctly.
3. To assure standardization and validation of data registry, as well the paper-format as the digital databases passed through consecutive cycles of: development (academia and engineer), pilot testing (municipalities), feedback/evaluation (municipalities and academia) and correction (municipalities and academia), until a final product was ready.

The software platform REDCap Cloud (cloud-based data management platform) was used to setup and administer the databases ensuring the required level of data security (GDPR).

The platform allowed the database to be set up with predefined rules of range as well as connections between related data fields. For continuous outcomes on scales that could in principle be infinite (ex. time to complete 10-meter walking distance), predefined rules in the database where based on pilot testing and or qualified by normative data on similar populations when possible. For ordinal and categorical outcomes as well as outcomes restricted to a given range (ex. self-rated health on a VAS scale) the predefined data-base rules were set according to these given restrictions.

The team setting up the database consisted of:

• One primary coder (profession: Engineer)
• Five Testers/Data collector managers from the different data collector sites (municipalities)
• One scientific advisor The outcome data in this database is not routinely collected elsewhere. Therefore, only few items, and only of descriptive character (e.g. sex and age) exist to our knowledge in other databases. Due to GDPR-restrictions these items were not compared to external data sources.

As external validation is not possible for this dataset, an internal verification procedure is developed and will be executed before running the analysis for this study.

1. The dataset consists of multiple waves of data-collection. Descriptive items that are considered time constant (e.g. sex, age, group-allocation, subject identification number) will be compared for inconsistencies across waves.
2. Initially, descriptive analysis to detect outlying data were performed on all outcome data for each wave separately. Each case will be investigated to identify potential erroneous data.
3. Variables with delta-values (i.e. changes from one test-time to another) will be generated for each outcome, and the procedure from point 2 will be followed
4. Spot checks will be done across related variables. For example, measured gait speed will be compared to a composite test-score in which gait speed is an important element. If gait speed is very high and the score of the composite test is very low the case will be investigated in order to identify which data is erroneous.
5. The database has dedicated a section for comments to the data made by either the test-personnel or the person digitalizing the data. Data will be qualified against these comments before being included in the analysis
6. If erroneous data is identified in step 1-5 or in case of doubt about data correctness, data will be marked as missing.

The step-wise verification process will be registered in the database material as Stata .do and Stata .log files, to ensure back-tracking and replicability of the procedure.

The main data dictionary was built in REDCap Cloud on the first language for the different sites (Danish or German).

As the database was later converted into .dta file format, an expansion of the data dictionary was initiated with added information and renaming of certain variables. The main language of the .dta database was then changed to English. All information on both translation, conversion and expansion of the dictionary is stored and available from Stata .do and .log files.

The REDCap Cloud database had a dedicated section for the recruitment pathway with a box to register time of data collection.

All steps in data preparation (changes) and analysis activities are registered both by syntax (.do) and output (.log)

Intervention instructors and the test-personnel were all instructed to identify and register adverse events. The registration was systematically done in all phases of the intervention.

The project had a prespecified number of participants it aimed at recruiting according to the funding requirements.

The database was fitted with options for registering reasons for missing data both for each individual section (i.e. self-report and objective data) and for each wave.

If erroneous data is identified in step 1-5 or in case of doubt, the data will be marked as missing.

Main intervention effects on the primary outcome SPPB and on secondary outcomes of muscle function, physical function, self-reported disability and quality of life as well as hypothesized effect modifiers: self-efficacy, outcome expectancies and barrier management will be estimated using repeated measures mixed models.

To identify significant covariates, adjusted models will be fitted using a step-wise approach.

Inclusion Criteria:

Subjects were eligible if at least one of following criteria was met:

1. Reduced physical function: SPPB score ≤9
2. Fatigability: Pittsburg Fatigability score of 15 or above (≥15)*
3. Low physical activity: Physically active ≤1day per week while sitting ≥8 hours on a normal day
4. Falls: ≥2 fall over the past year
5. Have pain: Brief Pain Inventory interference score ≥20

Exclusion Criteria:

1. High functioning (Composite SPPB score >10)

2. Too physically active (Physically active ≥3 days/week while sitting down <5 hours during a normal day)

3. Deadly or critical illness (Cancer, severe heart failure)

4. Recent surgery that is expected to affect the intensity of exercise and limit activity#

5. Recent fractures that is expected to affect the intensity of exercise and limit activity#

6. Chronic pain that prevents regular exercise

7. Reduced cognitive function (Dementia, Alzheimer's)

   • If the subject has completed a rehabilitation process in relation with an operation or fracture, the citizen is free to attend.

Additional precautions:

1. Indication of reduced cognitive function: When Six-Item Cognitive Impairment Screener score <4, subjects are referred to experts for further assessment and determination of cognitive function.
2. When unintended weight loss and underweight is detected by the nutritional assessment scheme (score =2), citizens are referred to medical doctor or dietician to recover this, prior to being enrolled in the study.

Home-dwelling older adults entitled to a yearly preventive home-visit according to the Danish social act, living in three provincial municipalities in Denmark (Odense, Slagelse and Esbjerg)