Burden of disease of people with dementia in nursing homes measured by IPOS-Dem: a stepped-wedge cluster randomised controlled trial

Auteur Frank SPICHIGER
Directeur /trice Philipp Larkin
Co-directeur(s) /trice(s) Andrea Koppitz (Heds-so Fribourg)
Résumé de la thèse


People with dementia (PWD) are at high risk of serious health-related suffering with limited access to any palliative care input to address their complex needs (Knaul et al., 2018; Sleeman et al., 2019). As stated by the Swiss federal office for public health (2012), complexity is a consequence of multidimensional needs influencing the individuals’ health which hinder prognosis. Accordingly, diagnosis, prognosis, response to treatment and palliation in PWD is challenging (Livingston et al., 2017). Hence un- and undertreated medical symptoms, unmet needs and undisclosed wishes and choices contribute to unnecessary, avoidable burdens and behavioural changes that are detrimental to the quality of life of PWD and their families (Deuschl & Maier, 2016; Husebo et al., 2011; Marvardi et al., 2005; Shim et al., 2016).

During their disease trajectory, a majority of PWD eventually moves into nursing homes (Livingston et al., 2017). Consequently, 65% of Swiss nursing home residents live with dementia (Ecoplan, 2019).


Assessment of medical symptoms, undisclosed wishes, choices and needs in PWD in nursing homes is difficult because of impaired cognitive abilities to report medical symptoms and corresponding needs (Lichtner et al., 2014). In nursing homes, miscommunication with PWD and among staff may lead to gaps in medical symptom detection and recognition (Ellis-Smith et al., 2018; Krumm et al., 2014). Current routine assessment in nursing homes (e.g. RAI, BESA, PLAISIR) tends to discriminate against timely, clinically relevant observations on PWD (Boyd et al., 2014; Koppitz et al., 2016). About 69% of the nursing staff in Swiss nursing homes have no tertiary nursing education (Obsan, 2009) and are therefore not allowed to assess their observations in routine assessments (Vettori et al., 2017) leading to eliciting critical concerns for the PWD.

Furthermore, routine assessment instruments used in nursing homes omit family perspectives and by extension, their disclosure and deliberation regarding the person’s prior and current wishes and choices (Vettori et al., 2017). Items on family involvement are absent, although the importance of recognising family caregivers has been established as a priority (Hasson et al., 2020). To recognise unmet needs in PWD, a structured systematical symptom and needs assessment may help (Ellis-Smith et al., 2016, 2018).

The integrated palliative outcome scale-dementia (IPOS-Dem) was developed to recognise needs and symptoms in PWD (Cicely Saunders Institute (CSI), 2018). The assessment instrument stems from a comprehensive family of patient-reported outcome measures for palliative care (Cicely Saunders Institute (CSI), 2015). Its holistic approach may help to inform the overview of PWD burden of disease at the point of care. We conceptualise the burden of disease as the subjective symptoms and unmet needs burdened by dementia and comorbidities (Isfeld-Kiely & Balakumar, 2015). Consequently, the burden of disease indicates “non-fatal health outcomes”, e.g. medical symptoms and unmet needs caused by dementia and comorbidities (Isfeld-Kiely & Balakumar, 2015). To reduce the burden of disease in PWD, it is not only necessary to embed a systematic assessment in nursing home routines. Other requirements are a shared definition of values and clinical measures following the assessment findings (Haan et al., 2018; Kupeli et al., 2016).

To our knowledge, this will be the first Swiss nursing home study to provide long term effects of palliative care outcome measurement combining staff and relative’s assessment as well as applying interventions in PWD. The study may encourage usage of patient-reported outcome measures as well as family input across the staff grade mix in nursing homes.

Aims & Hypothesis

We aim to evaluate the effectiveness of the IPOS-Dem in identifying and alleviating the burden of disease of PWD in nursing homes by staff and relatives.

Thus, we investigate the research question: Does the introduction of IPOS-Dem for application through staff and family in nursing homes and following case discussions significantly reduce the burden of disease as measured by the IPOS-Dem sum score?

Null Hypothesis:

There is no difference in PWD burden of disease, as measured by the IPOS-Dem sum score after having been exposed to the intervention.

Alternative Hypothesis:

The burden of disease will be reduced by eight points in the IPOS-Dem sum score compared to baseline measurements in PWD.


This study will be conducted as a stepped wedge design (K. Hemming et al., 2015). For the primary endpoint burden of disease, data will be analysed on an ‘intention to treat’ basis and also according to the protocol to verify the intention to treat results.

We define the burden of disease as measured with the IPOS-Dem sum score. The effect of the intervention on reducing the burden of disease over the study period will be examined by using linear mixed-effects models. This category of models is flexible, supporting various distributions of outcome variables and link functions (Cnaan et al., 1997; Lindstrom & Bates, 1990). Mixed-effect models allow for nested data as well as repeated measures. On the one hand, a mixed-effect model can control for the possible confounding through chronological bias (K. Hemming et al., 2018).

On the other hand, for handling the nested nature of the resulting data, a mixed-effect model is particularly useful (Luke, 2019). The closed cohort model will be composed of a random effect for the nursing homes (i.e. clusters), a random effect for the repeated measurements on the same cohort of individuals, a fixed effect to account for time trends and a fixed effect representing the intervention effect (i.e. treatment). A linear mixed-effect model further enables us to separate effects due to observed and unobserved cohort and cluster characteristics in the trial (Luke, 2019).

From our power calculations (13 clusters, three steps, α = .05, β = .2) and accounting for attrition rate (Chenoweth et al., 2011; Husebø et al., 2015), we will screen 260 PWD (n=260) in order to include 228 PWD into the study to be able to enter 208 PWD into the analysis. For a 17-month study period, we propose to randomise 13 nursing homes in three steps (i.e. at three, six and nine months into the trial duration). In three monthly intervals, four to five nursing homes will administer the intervention at given periods until all nursing homes are applying the intervention. All nursing homes will employ IPOS-Dem at least monthly for the full study duration of 17 months. Depending on randomised allocation, the cohorts will receive the intervention for six to twelve months. Based on previous studies with IPOS-Dem implementation in nursing homes, each cohort will start without receiving intervention for three months minimum (Ellis-Smith et al., 2018).

During the intervention phase, an intervention nurse will visit the nursing homes monthly across rotating shift patterns. They support staff and family members IPOS-Dem observations. The observations will be validated during case conferences in which nursing home staff and family members will be available. The intervention nurse will synthesise case conferences (Boyd et al., 2014; Koppitz et al., 2016; Neylon, 2015). Therefore, we want to encourage regular case conferences focusing on PWD needs, choices and wishes (Goodman et al., 2016) in nursing homes. Case conferences based on shared IPOS-Dem assessments may help elicit medical symptoms, needs of PWD when adopted by the nursing home staff as well as PWDs family members (Ellis-Smith et al., 2018; Hennings & Froggatt, 2016; Müller et al., 2017). The intervention nurse will act as a facilitator in both the IPOS-Dem observations and during the case conferences (Moore et al., 2020). Case conferences are planned to follow an approach similar to moral case deliberation (Haan et al., 2018) which usually employs a systematical approach to deliberate on the concrete case than regular clinical team meetings, propagating a “bottom-up” approach.


This project involves a vulnerable patient population, which requires approval through respective cantonal ethics committees as well as study personnel proficient in geriatrics and palliative care (Hanson et al., 2014) as well as. Because the study does not target intervention with drugs, medical devices or transplants, according to ClinO Art.61, the study risk category is A (Swissethics, 2018). The design chosen has the advantage that all nursing homes receive the intervention. Through the longitudinal nature of data collection, we expect to minimise confounding factors, e.g. correct application of IPOS-Dem or time effects (Haines & Hemming, 2018; Jordan et al., 2015).

For PWD, staff as well as family we expect immediate benefit through establishing situational overview and targeted communication as well as action. We expect targeted symptom management to reduce burden of disease in PWD as well as empowerment of family members involved.


The Stanley Thomas Johnson Fund awarded this study a grant to Prof Dr Andrea Koppitz. We are currently checking my eligibility for an SNFS Doc.CH grant.

Statut au début
Délai administratif de soutenance de thèse 2023