Understanding the influence of different proxy perspectives in explaining the difference between self-rated and proxy-rated quality of life in people living with dementia: a systematic literature review and meta-analysis

Quality of life (QoL) has become an important outcome for research and practice but obtaining reliable and valid estimates remains a challenge in people living with dementia [1]. According to the Diagnostic and Statistical Manual of Mental Disorders (DSM-5) criteria [2], dementia, termed as Major Neurocognitive Disorder (MND), involves a significant decline in at least one cognitive domain (executive function, complex attention, language, learning, memory, perceptual-motor, or social cognition), where the decline represents a change from a patient's prior level of cognitive ability, is persistent and progressive over time, is not associated exclusively with an episode of delirium, and reduces a person’s ability to perform everyday activities. Since dementia is one of the most pressing challenges for healthcare systems nowadays [3], it is critical to study its impact on QoL. The World Health Organization defines the concept of QoL as “individuals' perceptions of their position in life in the context of the culture and value systems in which they live and in relation to their goals, expectations, standards, and concerns” [4]. It is a broad ranging concept incorporating in a complex way the persons' physical health, psychological state, level of independence, social relationships, personal beliefs, and their relationships to salient features of the environment.

Although there is evidence that people with mild to moderate dementia can reliably rate their own QoL [5], as the disease progresses, there is typically a decline in memory, attention, judgment, insight, and communication that may compromise self-reporting of QoL [6]. Additionally, behavioral symptoms, such as agitation, and affective symptoms, such as depression, may present another challenge in obtaining self-reported QoL ratings due to emotional shifts and unwillingness to complete the assessment [7]. Although QoL is subjective and should ideally be assessed from an individual’s own perspective [8], the decline in cognitive function emphasizes the need for proxy-reporting by family members, health professionals, or care staff who are asked to report on behalf of the person with dementia. However, proxy-reports are not substitutable for self-reports from people with dementia, as they offer supplementary insights, reflecting the perceptions and viewpoints of people surrounding the person with dementia [9].

Previous research has consistently highlighted a disagreement between self-rated and proxy-rated QoL in people living with dementia, with proxies generally providing lower ratings (indicating poorer QoL) compared with person’s own ratings [8, 10‐12]. Impairment in cognition associated with greater dementia severity has been found to be associated with larger difference between self-rating and proxy-rating obtained from family caregivers, as it becomes increasingly difficult for severely cognitively impaired individuals to respond to questions that require contemplation, introspection, and sustained attention [13, 14]. Moreover, non-cognitive factors, such as awareness of disease and depressive symptoms play an important role when comparing QoL ratings between individuals with dementia and their proxies [15]. Qualitative evidence has also shown that people with dementia tend to compare themselves with their peers, whereas carers make comparisons with how the person used to be in the past [9]. The disagreement between self-reported QoL and carer proxy-rated QoL could be modulated by some personal, cognitive or relational factors, for example, the type of relationship or the frequency of contact maintained, person’s cognitive status, carer’s own feeling about dementia, carer’s mood, and perceived burden of caregiving [14, 16]. Disagreement may also arise from the person with dementia’s problems to communicate symptoms, and proxies’ inability to recognize certain symptoms, like pain [17], or be impacted by the amount of time spent with the person with dementia [18]. This may also prevent proxies to rate accurately certain domains of QoL, with previous evidence showing higher level of agreement for observable domains, such as mobility, compared with less observable domains like emotional wellbeing [8]. Finally, agreement also depends on the type of proxy (i.e., informal/family carers or professional staff) and the nature of their relationship, for instance, proxy QoL scores provided by formal carers tend to be higher (reflecting better QoL) compared to the scores supplied by family members [19, 20]. Staff members might associate residents’ QoL with the quality of care delivered or the stage of their cognitive impairment, whereas relatives often focus on comparison with the person’s QoL when they were younger, lived in their own home and did not have dementia [20].

What has been not been fully examined to date is the role of different proxy perspectives employed in QoL questionnaires in explaining disagreement between self-rated and proxy-rated scores in people with dementia. Pickard et al. (2005) have proposed a conceptual framework for proxy assessments that distinguish between the proxy-patient perspective (i.e., asking proxies to assess the patient’s QoL as they think the patient would respond) or proxy-proxy perspective (i.e., asking proxies to provide their own perspective on the patient’s QoL) [21]. In this context, the intra-proxy gap describes the differences between proxy-patient and proxy-proxy perspective, whereas the inter-rater gap is the difference between self-report and proxy-report [21].

Existing generic and dementia-specific QoL instruments specify the perspective explicitly in their instructions or imply the perspective indirectly in their wording. For example, the instructions of the Dementia Quality of Life Measure (DEMQOL) asks proxies to give the answer they think their relative would give (i.e., proxy-patient perspective) [22], whereas the family version of the Quality of Life in Alzheimer’s Disease (QOL-AD) instructs the proxies to rate their relative’s current situation as they (the proxy) see it (i.e., proxy-proxy perspective) [7]. Some instruments, like the EQ-5D measures, have two proxy versions for each respective perspective [23, 24]. The Adult Social Care Outcome Toolkit (ASCOT) proxy version, on the other hand, asks proxies to complete the questions from both perspectives, from their own opinion and how they think the person would answer [25].

QoL scores generated using different perspectives are expected to differ, with qualitative evidence showing that carers rate the person with dementia’s QoL lower (worse) when instructed to comment from their own perspective than from the perspective of the person with dementia [26]. However, to our knowledge, no previous review has fully synthesized existing evidence in this area. Therefore, we aimed to undertake a systematic literature review to examine the role of different proxy-assessment perspectives in explaining differences between self-rated and proxy-rated QoL in people living with dementia. The review was conducted under the hypothesis that the difference in QoL estimates will be larger when adopting the proxy-proxy perspective compared with proxy-patient perspective.

The review was registered with the International Prospective Register of Systematic Reviews (CRD42022333542) and followed the Preferred Reporting Items System for Systematic Reviews and Meta-Analysis (PRISMA) guidelines (see Appendix 1) [27].

The PRISMA diagram in Fig. 1 shows that after the abstract and full-text screening, 38 studies from the database search and 58 studies from the Roydhouse et al. (2022) review were included in this review—a total of 96 studies. A list of all studies included and their characteristics can be found in Appendix 3.

While previous studies highlighted a disagreement between self-rated and proxy-rated QoL in people living with dementia, this review, for the first time, assessed the role of different proxy perspectives in explaining the inter-rater gap. Our findings align with the baseline hypothesis and indicate that QoL scores reported from the proxy-patient perspective are closer to self-reported QoL scores than the proxy-proxy perspective, suggesting that the proxy perspective does impact the inter-rater gap and should not be ignored. This finding was observed across different analyses conducted in this review (i.e., pooled raw mean difference, SMD, ICC analysis), which also confirms the results of two previous primary studies that adopted both proxy perspectives within the same study design [5, 38]. Our findings emphasize the need for transparency in reporting the proxy perspective used in future studies, as it can impact results and interpretation. This was also noted by the recent ISPOR Proxy Task Force that developed a checklist of considerations when using proxy-reporting [39]. While consistency in proxy-reports is desirable, it is crucial to acknowledge that each proxy perspective holds significance in future research, depending on study objectives. It is evident that both proxy perspectives offer distinct insights—one encapsulating the perspectives of people with dementia, and the other reflecting the viewpoints of proxies. Therefore, in situations where self-report is unattainable due to advanced disease severity and the person’s perspective on their own QoL assessment is sought, it is recommended to use the proxy-patient perspective. Conversely, if the objective of future research is to encompass the viewpoints of proxies, opting for the proxy-proxy perspective is advisable. However, it is important to note that proxies may deviate from instructed perspectives, requiring future qualitative research to examine the adherence to proxy perspectives. Additionally, others have argued that proxy-reports should not substitute self-reports, and only serve as supplementary sources alongside patient self-reports whenever possible [9].

This review considered various QoL instruments, but most instruments adopted one specific proxy perspective, limiting detailed analyses. QoL instruments differ in their scope (generic versus disease-specific) as well as coverage of QoL domains. The QOL-AD, an Alzheimer's Disease-specific measure, was commonly used. Surprisingly, for this measure, the mean differences between self-reported and proxy-reported scores were smaller using the proxy-proxy perspective, contrary to the patterns observed with all other instruments. This may be due to the lack of studies reporting QOL-AD proxy scores from the proxy-patient perspective, as the study by Bosboom et al. (2012) found the opposite [5]. Previous research has also suggested that the inter-rater gap is dependent on the QoL domains and that the risk of bias is greater for more ‘subjective’ (less observable) domains such as emotions, feelings, and moods in comparison with observable, objective areas such as physical domains [8, 40]. However, this review lacks sufficient observations for definitive results on QoL dimensions and their impact on self-proxy differences, emphasizing the need for future research in this area.

With regard to proxy type, there is an observable trend suggesting a wider inter-rater gap when family proxies are employed using the proxy-proxy perspective, in contrast to formal proxies. This variance might be attributed to the use of distinct anchoring points; family proxies tend to assess the individual's QoL in relation to their past self before having dementia, while formal caregivers may draw comparisons with other individuals with dementia under their care [41]. However, the opposite was found when the proxy-patient perspective was used, where family proxies scores seemed to align more closely with self-reported scores, resulting in lower SMD scores. This suggests that family proxies might possess a better ability to empathize with the perspective of the person with dementia compared to formal proxies. Nonetheless, it is important to interpret these findings cautiously, given the relatively small number of observations for formal caregiver reports. Additionally, other factors such as emotional connection, caregiver burden, and caregiver QoL may also impact proxy-reports by family proxies [14, 16] that have not been explored in this review.

Our review found that the SMD between proxy and self-report increased with greater level of dementia severity, contrasting a previous study, which showed that cognitive impairment was not the primary factor that accounted for the differences in the QoL assessments between family proxies and the person with dementia [15]. However, it is noteworthy that different interpretations and classifications were used across studies to define mild, moderate, and severe dementia, which needs to be considered. Most studies used MMSE to define dementia severity levels. Given the MMSE’s role as a standard measure of cognitive function, the study findings are considered generalizable and clinically relevant for people with dementia across different dementia severity levels. When examining the role of the proxy perspective by level of severity, we found that compared with the proxy-proxy perspective, where self-reported scores were greater than proxy-reported scores across all dementia severity levels, the proxy-patient perspective yielded the opposite results, and proxies reported better QoL than people with dementia themselves, except for the severe subgroup. It is possible that in the early stages of dementia, the person with dementia has a greater awareness of increasing deficits, coupled with denial and lack of acceptance, leading to a more critical view of their own QoL than how proxies think they would rate their QoL. However, future studies are warranted, given the small number of observations adopting the proxy-patient perspective in our review.

The heterogeneity observed in the studies included was high, supporting the use of random-effects meta-analysis. This is not surprising given the diverse nature of studies included (i.e., RCTs, cross-sectional studies), differences in the population (i.e., people living in residential care versus community-dwelling people), mixed levels of dementia severity, and differences between instruments. While similar heterogeneity was observed in another review on a similar topic [42], our presentation of findings stratified by proxy type, dementia severity, and living arrangement attempted to account for such differences across studies.

This review demonstrates that the choice of proxy perspective impacts the inter-rater gap. QoL scores from the proxy-patient perspective align more closely with self-reported scores than the proxy-proxy perspective. These findings contribute to the broader literature investigating factors influencing differences in QoL scores between proxies and individuals with dementia. While self-reported QoL is the gold standard, proxy-reports should be viewed as complements rather than substitutes. Both proxy perspectives offer unique insights, yet QoL assessments in people with dementia are complex. The difference in self- and proxy-reports can be influenced by various factors, necessitating further research before presenting definitive results that inform care provision and policy.