Discussion

Proxies’ perceived most useful perspective when reporting on different HRQOL-related domains varied modestly across and within conditions. One explanation for this heterogeneity could be that carers may have interpreted the question differently. Proxies provided different reasons for perceiving a perspective or perspectives as most helpful, including person-centred, proxy-centred and communication-centred reasons. Regardless of the perspective or the domain, familiarity with the person emerged as a shared rationale.

This variation shows that there is no one- size-fits-all approach. The number of carers who selected a perspective as most useful varied based on both the condition of the person for whom they were caring and the type of information the carer reported. In MS, especially in early and mid-stages, proxies may not need to imagine the person’s internal experience as people with MS could self-report. Cognitive impairment in MS tends to be slower, domain-specific, and less global than in dementia, although differentiating the impairment at severe levels is difficult [49]. People with MS may be able to reliably self-report even with cognitive symptoms [50]. In this context, proxies may lean toward the proxy-proxy perspective, relying on their own observations to supplement the person’s input and provide additional information. The variation on most useful perspective based on type of information reported may have implication for multidimensional HRQOL measure design.

The reasons for perceived usefulness of different perspectives in our study align with the themes identified by Caiels et al. [21]. Caiels et al. found that proxies’ preferences are influenced by what they consider beneficial for the person, what aligns with their own understanding, and what they believe facilitates better communication. Similarly, Lobchuk et al. [10] classified carers’ responses to perspective-taking prompts into categories such as patient-oriented and caregiver-oriented thoughts, which align with our findings of person-centred and proxy-centred reasons, respectively. The third category, communication-centred in our study reflects carers’ consideration of how best to convey information to HCPs. Lobchuk et al. similarly highlighted communication as relevant to carers’ interpretation of patients’ experiences, although their study focused on perspective-taking and communication difficulties rather than on how carers communicated information to HCPs [10]. These underlying rationales could be considered when designing or interpreting HRQOL or other patient-reported outcome measures. An important point, however, to consider is that the choice of perspective would be determined by the study purpose. For example, in clinical trials where proxy responses may be used to substitute for missing patient-reported outcomes, the proxy-person perspective may be used if it provides closer agreement with self-report as the aim is to reflect the patient’s own viewpoint [13]. This approach supports substitutability and maintains consistency with self-report measures. In contrast, if the goal is to provide an additional, complementary viewpoint, the proxy-proxy perspective may be more suitable. Clear guidance on which perspective to adopt may therefore be tailored to the purpose of measurement. Along with this, as explained in the conceptual model by Pickard and Knight, researchers should be aware that proxies could understand and respond to questions differently due to the perspective used, and this has implications for measurement validity and reliability. Clarifying why the researcher wanted a proxy respondent to report from a certain proxy perspective could help the respondents to follow the instructions [13].

A critical issue is whether proxies understand and follow the perspective provided in the instructions. Previous studies [21, 51] found that proxies may unintentionally switch perspectives unless explicitly reminded, pointing to the importance of clear instructions. As proxies may try to respond from the perspective or mix of perspectives that they find useful when answering questions about someone else’s health, clearly communicating the intended perspectives to proxies may help reduce misinterpretations and support adherence to instructions [16].

Our findings show that “familiarity with the person” was a rationale for both proxy perspective choices aligns with and extends prior research. For example, carers have emphasised that proxies should know the person well [21]. Similarly, the Health and Retirement Study [52] identified familiarity as an important characteristic of proxy respondents. Prior research [2] has suggested that familiarity can be operationalised through factors such as the intensity and frequency of contact, time spent together, and the quality of communication and relationship closeness. Relationship closeness, in particular, has been shown to reduce judgment error in proxy reports [53], and proxies with better knowledge of the person tend to provide responses that more closely align with self-reports [54]. Our findings show that the use of familiarity as a justification for perspective choice may vary by context. For the proxy-person perspective, familiarity was mentioned by proxies in the context of knowing the person to represent their lived experience. For the proxy-proxy perspective, familiarity may be used to justify providing an objective or complementary viewpoint. However, there is an underlying commonality that familiarity is an important justification provided by carers, regardless of the perspective selected as most useful for communicating with HCPs.

Implications

These findings may have practical implications for improving proxy-reported measurement, including measurement of HRQOL or related domains. However, further investigation will be required before reaching conclusions about the generalisability and applicability of these findings to those contexts. The evidence from this exploratory study may help inform such future work. Our findings do not provide support for the suggestion that one specific perspective should be chosen for all conditions or domains. One possible approach could be designing measures that include both perspectives, an approach that is taken for ASCOT [17], which has been shown to be feasible and improve acceptability [16]. However, there are potential implementation challenges, such as determining which perspective to prioritise for decision-making and reporting if responses differ across perspectives. In any case, it is important to document which perspective was used, for example by asking proxies how or if they adhered to the specified perspective(s). The recommendation for documenting perspectives has previously been suggested by the International Society of Quality of Life Proxy Task Force [1]. Our findings about the importance of familiarity as a reason for perspective choice suggests that this should be captured as well when proxy reporting is used.

Limitations and strength

This study’s limitations include the use of self-reported online data from self-selected participants, which may introduce biases such as self-selection, social desirability, and voluntary response bias. In addition, the study sample size was relatively small. Web-based data collection excluded individuals without internet access. Future studies with more diverse samples may generate additional or different insights. We used free text online qualitative surveys instead of qualitative focus group or interview, which limited the opportunity to clarify concepts to respondents or to probe responses in depth. At the same time, the online survey approach supports collecting information from diverse populations [42–44], makes surveys accessible to hard-to-reach participants [46], and facilitates anonymous [46] and rapid data collection. Future research may be able to add to this work by examining if respondents have fixed views of ‘useful’ perspectives or if they change their responses regarding ‘useful’ perspectives when probed.

The current study focusses on MS symptoms in general, and symptoms may vary in terms of observability, as well as pain and physical function in dementia. Future studies could consider other aspects of HRQOL such as independence, control, social participation, and emotional function. Although this study did not assess the perceived most useful proxy perspective in the context of completing a specific HRQOL instrument, its findings could inform future work focusing on specific HRQOL or PRO instruments. Additionally, only current carers were included in the MS survey, and the classification of current and former carer was left open for respondents’ interpretation in the dementia survey. We did not include this in the analysis. We acknowledge that asking carers about past situations could add to the complexity of proxy reporting. Future studies could consider the effects of asking carers about past events and the effect of recall period on proxy reporting. The current study was a preliminary online survey with a small sample size. Further studies with other qualitative techniques may be helpful. Lastly, entry codes and passwords were not used to ensure data quality in both surveys.

Nonetheless, this preliminary study had several strengths. In particular, study materials were reviewed by people with lived experience of being a carer. The study also provides additional evidence in an area in which evidence is scarce, helping to build the groundwork that can support improved measure design and administration of proxy-reported measures when HRQOL or related domains are assessed.