Investigating the effectiveness of care delivery at an acute geriatric community hospital for older adults in the Netherlands: a protocol for a prospective controlled observational study

Introduction Hospital admission in older adults with multiple chronic conditions is associated with unwanted outcomes like readmission, institutionalisation, functional decline and mortality. Providing acute care in the community and integrating effective components of care models might lead to a reduction in negative outcomes. Recently, the first geriatrician-led Acute Geriatric Community Hospital (AGCH) was introduced in the Netherlands. Care at the AGCH is focused on the treatment of acute diseases, comprehensive geriatric assessment, setting patient-led goals, early rehabilitation and streamlined transitions of care. Methods and analysis This prospective cohort study will investigate the effectiveness of care delivery at the AGCH on patient outcomes by comparing AGCH patients to two historic cohorts of hospitalised patients. Propensity score matching will correct for potential population differences. The primary outcome is the 3-month unplanned readmission rate. Secondary outcomes include functional decline, institutionalisation, healthcare utilisation, occurrence of delirium or falls, health-related quality of life, mortality and patient satisfaction. Measurements will be conducted at admission, discharge and 1, 3 and 6 months after discharge. Furthermore, an economic evaluation and qualitative process evaluation to assess facilitators and barriers to implementation are planned. Ethics and dissemination The study will be conducted according to the Declaration of Helsinki. The Medical Ethics Research Committee confirmed that the Medical Research Involving Human Subjects Act did not apply to this research project and official approval was not required. The findings of this study will be disseminated through public lectures, scientific conferences and journal publications. Furthermore, the findings of this study will aid in the implementation and financing of this concept (inter)nationally. Trial registration number NL7896; Pre-results.

Historical control groups 218 The first control group from the Transitional Care Bridge Study consists of 674 patients that were  The primary outcome measure is the 3-month unplanned readmission rate to the AGCH or hospital.
237 Secondary outcomes measured at 1,3 and 6 months will include: 238 1) ADL-functioning as defined by the KATZ-ADL scale. 33 239 2) Healthcare utilization, including institutionalization in a long-term care facility. Data collection 246 Eligible patients and/or legal representatives will be contacted and informed about the study 247 procedures after which written informed consent is obtained. Inclusion and interviewing of patients 248 is conducted by an onsite researcher. Routine data on functioning and risk assessments are collected 249 by a trained registered nurse and physiotherapist as part of the CGA for each patient. 35 Table 1 gives 250 an overview of measurement of primary and secondary outcomes over time. Measurements during 251 admission are at H1 which is within 48 hours after admission and H2 which is within 48 hours before 252 discharge. Follow-up is completed by telephone at 1,3 and 6 months after discharge (P1, P3 and P6). 258 Chronic conditions. The number and severity of chronic conditions will be assessed using the Charlson 259 Comorbidity Index. 36 This index is commonly used to indicate the risk of mortality; each condition is 260 scored 1, 2, 3 or 6 points, with a higher total number of points indicating a greater risk at death.

261
Polypharmacy. Polypharmacy will be assessed by counting the number of individual drugs that are 262 chronically prescribed to a participant, in which a number of 5 or more drugs is considered 263 polypharmacy.

264
Mortality. This will assessed during follow-up, where possible by reporting from patients electronic 265 files, otherwise from registries of the general practitioner. 289 Health-Related Quality of Life. This will be measured by determining Health-Related Quality of Life 290 using the EuroQoL-5D (EQ-5D). The EQ-5D is a broadly used and validated instrument for measuring 291 generic health-related quality of life. 34 It consists of 5 dimensions: 1) mobility 2) self-care 3) usual 292 activities 4) pain/discomfort 5) anxiety/depression. We will use the EQ-5D-3L which has three 293 options: no problems, some problems or severe problems. In addition, the following questions will 294 be used to assess self-reported quality of life: 1) How would you rate your quality of life in general ? functioning. This included assessment of performance in basic ADL-KATZ-6), as in instrumental ADL 305 (KATZ-9). 44, 45 We measure KATZ-score both currently (at admission), as two weeks before admission, 306 reflecting pre-morbid level of functioning. (I)ADL-functional is also included in follow-up 307 measurements.

308
(Im)mobility. We will assess mobility by reviewing three questions that are in the admission 309 assessment regarding: 1) use of a walking aid (from , and from the CGA: 2) being able to 310 walk outside of the house for five minutes (two weeks before and currently) and 3) performing, and   and study recruitment at the AGCH and will provide study discontinuation rates at 1, 3 and 6 months 372 follow-up. 30 We will describe demographic, clinical and prognostic characteristics of the study 373 participants at baseline. The number of participants with missing data will be collected and described 374 alongside our variables to check for the pattern of missingness. Inversely-weighted propensity scores 375 will be used to control for any imbalances between the treatment groups. 54 Propensity scores will be 376 calculated using generalized booted methods. Balance and overlap of propensity scores distribution 377 will be assessed. Propensity score weights for the estimation of the average treatment effect will be 378 created using all covariates where groups differed on baseline or that were associated with 90-day 379 readmission rate. As this is a repeated measures design, we will assume equal weighting for all 380 measurements. 55 381 All hypotheses will be tested using two-tailed-significance level of 0.05. Descriptive analyses will be 382 performed to examine participant's characteristics. Differences in changes over time in outcomes will 383 be compared between groups using multilevel models. All models will include a main effect of 384 treatment group, a linear term for time and an interaction between time and treatment group.

385
Models will be checked with residual and appropriate goodness-of-fit statistics.

387
Economic evaluation 388 A healthcare and societal perspective is planned for this economic evaluation. 56 The evaluation from 389 the healthcare perspective will only include direct medical costs accrued in the six months after the 390 admission to the AGCH. Propensity scores will also be used in the economic evaluation. Missing data 391 will be imputed using multiple imputation chained equations if necessary, for cost and effect data.

Other information
Funding 22 Give the source of funding and the role of the funders for the present study and, if applicable, for the original study on which the present article is based *Give information separately for exposed and unexposed groups. recruit for 18 to 24 months. The first three months of data collection will consist of a piloting phase 209 to assess the feasibility of data-collection and follow-up. In addition, a qualitative process evaluation

343
Hospital readmission. This outcome will be assessed during follow-up. Follow-up will consist of three 344 telephone interviews at 1, 3 and 6 months after discharge. Readmission will be both assessed by 345 interview as by checking care data from an aggregated database of expense claims of various 346 healthcare insurers. Data that will be collected are: number of readmissions, total days of 347 readmission, reasons for readmission and whether the readmission was planned or unplanned.

348
Emergency department (ED) visits. ED visits will be assessed during follow-up and checked in 349 insurance data. We will record the number of separate ED visits. to follow-up, we will aim to include a total of 567 (= 515*1.10) patients from the AGCH. Planned statistical analysis 374 The complete participant flow diagram will show a summary of admissions and study recruitment at 375 the AGCH and will provide study discontinuation rates at 1, 3 and 6 months follow-up. 30 We will 376 describe demographic, clinical and prognostic characteristics of the study participants at baseline.

377
The number of participants with missing data will be collected and described alongside our variables 378 to check for the pattern of missingness. Inversely-weighted propensity scores will be used to control 379 for any imbalances between the treatment groups. 55 Propensity scores will be calculated using 380 generalized booted methods. Balance and overlap of propensity scores distribution will be assessed.

381
Propensity score weights for the estimation of the average treatment effect will be created using all 382 covariates where groups differed on baseline or that were associated with 90-day readmission rate.

383
As this is a repeated measures design, we will assume equal weighting for all measurements. 56 384 All hypotheses will be tested using two-tailed-significance level of 0.05. All secondary outcomes will 385 be adjusted for multiple testing using a Hochberg method. 57,58 Descriptive analyses will be performed 386 to examine participant's characteristics. Differences in changes over time in outcomes will be 387 compared between groups using multilevel models. All models will include a main effect of 388 treatment group, a linear term for time and an interaction between time and treatment group.

389
Models will be checked with residual and appropriate goodness-of-fit statistics.

391
Economic evaluation 392 A healthcare and societal perspective is planned for this economic evaluation. The evaluation from 393 the healthcare perspective will only include direct medical costs accrued in the six months after the 394 admission to the AGCH. Direct medical cost will only include costs that are funded through the Dutch 395 healthcare system. The evaluation from a societal perspective will include an estimation of the cost 396 of informal care. Costs will be based on the reference prices found in the Dutch Manual for Costing 397 studies and will be set for final year of data collection (2020 or 2021). According to this guideline Process evaluation and patient experience 409 We plan to use a qualitative study design to describe barriers and facilitators to implementation of 410 the AGCH-concept and describe experiences of patients and healthcare professionals with the AGCH.

411
We will conduct semi-structured interviews with various stakeholders, such as geriatricians, nurses, 412 physiotherapists and hospital administrators. These interviews will concern the implementation of 413 the AGCH concept. In addition, semi-structured interviews with patients and informal caregivers will 414 be conducted in order to describe the patient experience and satisfaction with this new form of care.

415
A representative sample of patients and/or caregivers who participate in the prospective cohort 416 study will approached and invited to be interviewed shortly after the discharge from the AGCH.

417
Stakeholders and healthcare professionals will be selected by a researcher and will be invited for an 418 interview to discuss their experience and opinion on the AGCH. Interviews will be typed verbatim 419 and analysed by two researchers independently, using thematic analysis. 62 In our analysis of barriers 420 and facilitators to implementation, we will describe these factors at different levels: micro 421 (healthcare professionals), meso (care organizations) and macro level (legal and financial 422 framework). 63 The findings will be summarized in matrices with facilitators and barriers at different The complex acute medical needs of older patients require the delivery of specialized geriatric care.

82
-This study will be the first to evaluate an acute geriatric community hospital in the Netherlands 83 on both patient-reported and economic outcomes.

84
-Patients, informal caregivers and professionals were involved in the design and implementation 85 of the Acute Geriatric Community Hospital (AGCH).

86
-A process evaluation is planned to describe the experience of various stakeholders with this new 87 concept and reveal barriers and facilitators to its implementation.

88
-A limitation of this study is the use of two historic cohorts as the control population, which may 89 result in baseline differences between the control and intervention population.  The AGCH opened in July 2018. It serves the south-eastern part of Amsterdam and its surrounding 180 areas (an area with approximately 147 500 inhabitants). 28 The AGCH is a 23-bed facility within a 181 skilled nursing facility. The hospital has 24-hour geriatric and nursing assistance. Physiotherapy and 182 routine laboratory testing are available during the workweek and simple X-ray is available once a 183 week. The population that is eligible for admission to the AGCH are patients with a combination of an Outcomes 253 The primary outcome measure is the 3-month unplanned readmission rate to the AGCH or hospital.

254
Secondary outcomes measured at one, three and six months will include:

347
Hospital readmission. This outcome will be assessed during follow-up. Follow-up will consist of three 348 telephone interviews at one, three and six months after discharge. Readmission will be both assessed 349 during the follow-up interviews and by checking care data from an aggregated database of expense at the AGCH will yield 80% power to detect an absolute difference of 8% in the readmission rate 376 (which is a 25% reduction in the relative risk) using a two-sided test with an alpha of 0.05. 54 As we 377 expect 10% loss to follow-up, we aim to include a total of 567 (= 515*1.10) patients from the AGCH.

379
Planned statistical analyses 380 The complete participant flow diagram will show a summary of admissions and study recruitment at 381 the AGCH and will provide study discontinuation rates at one, three and six months follow-up. 30 We 382 will describe the demographic, clinical and prognostic characteristics of the study participants at 383 baseline. The number of participants with missing data will be collected and described alongside our 384 variables to check for the pattern of missingness. Inversely weighted propensity scores will be used 385 to control for any imbalances between the treatment groups. 55 Propensity scores will be calculated 386 using generalized booted methods. Balance and overlap of propensity score distribution will be 387 assessed. Propensity score weights for the estimation of the average treatment effect will be created 388 using all covariates where groups differed at baseline or that were associated with the 90-day 389 readmission rate. As this is a repeated measures design, we will assume equal weighting for all 390 measurements. 56 391 All hypotheses will be tested using a two-tailed significance level of 0.05. All secondary outcomes will 392 be adjusted for multiple testing using a Hochberg method. 57,58 Descriptive analyses will be performed 393 to examine the participants' characteristics. Differences in changes over time in outcomes will be 394 compared between groups using multilevel models. All models will include a main effect of 395 treatment group, a linear term for time and an interaction between time and treatment group.