Background

Proxy reporting involves obtaining information about patients’ health, including their health-related quality of life (HRQOL) or domains related to HRQOL. Proxy reporting is often sought if a patient has cognitive or physical impairments [1]. However, differences between proxy and patient reports are often raised as a concern when trying to interpret proxy-reported data [2, 3], including data about patient HRQOL.

Proxy reports are affected by several factors. This includes patient-related factors (e.g., disease severity, progression, educational status) [4], proxy-related factors (e.g., caregiving burden, age) [2, 4], and the proxy’s relationship with the patient [5–7]. The measure-related factors that can affect proxy reports include the observability of the domain to be reported [2], and the perspective the proxy are asked to adopt while responding [4, 8–10]. Although regulatory bodies such as the US Food and Drug Administration (FDA) [11] and the European Medicines Agency (EMA) [12] often view proxies as only taking one perspective when reporting, the literature describes more than one perspective: “proxy-person” (aligning with the regulatory definitions or view of proxies) and “proxy-proxy” [13]. The proxy-person perspective involves proxies reporting as they believe the patient would, adopting the patient’s viewpoint [14]. To illustrate this, an example question from the proxy-person perspective could be “How much pain do you think the patient would say they are experiencing if they were able to communicate?” In contrast, the proxy-proxy perspective reflects the proxy’s own assessment based on their observations and understanding. The previous question phrased using the proxy-proxy perspective could be “How much pain do you think the patient is experiencing?” Some proxy-reported outcome measures such as the Dementia Quality of Life (DEMQOL) [15] are designed from a proxy-person perspective, while others such as the Adult Social Care Outcomes Toolkit (ASCOT) [16] is designed to ask the proxy to consider both perspectives together [17]. The EQ-5D allows users to opt for one or both versions [18]. EuroQol, which is the developer of the EQ-5D, now provides only the proxy–proxy version, but the proxy‑person version is available by wording adjustment with permission [19].

Regardless of the perspective used, providing clear instructions to proxy respondents is recommended [1, 4]. The research objective should determine which perspective to use [20]. A qualitative study [21] proposed offering separate response options for each perspective to improve clarity and validity. The evidence on which perspective provides closer agreement with patient self-report is mixed. Some studies showed stronger agreement with self-report when proxy-person perspective is used [20, 22–25]. Another study [14] found better concordance with the proxy-proxy perspective for certain sub-scales. However, agreement of proxy-report with self-report is not the only metric used for evaluating proxy-report quality. Because self-reports are not available in the context where proxy reports are needed, such as advanced dementia, comparing proxy-reports with self-report is not always possible. This makes it problematic to rely on concordance as a primary quality indicator. Moreover, there is no evidence whether the proxies adhere to the perspective instructions provided [26].

Although evidence remains limited, some studies have examined how proxies interpret and follow perspective instructions for specific measures. For example, Rand and colleagues [16] explored proxies’ understanding of the proxy-proxy and proxy-person perspectives in the ASCOT-Proxy. Our study extends this literature by investigating interpretation and adherence across a broader range of domains relevant to HRQOL and does not focus on a single instrument. There is limited evidence on which perspective proxies may thought to be most useful to report about others’ health. This study, therefore, sought to identify which perspective proxies found most useful and why. This information may be helpful in guiding future research that can inform the design of future measures.

We aimed to gather this additional evidence in two different clinical contexts. The first was dementia, in which proxy report is widely used due to cognitive decline in the person living with dementia, especially in the later stages of the condition [27]. The second context was multiple sclerosis (MS), a context in which proxy reporting is less common, though not non-existent. In MS, proxy reporting may be used during mood disturbances, physical disability, or when physical impairments hinder self-reporting [28].

Additionally, given previous evidence that the observability of the domain on which the proxy reports may play a role in proxy reporting [2], we examined this across different HRQOL-related domains of varying observability. For example, less visible MS symptoms [29, 30] such as fatigue [31, 32], depression [33], pain and cognitive impairment [34, 35] and visible symptoms such as walking difficulty [29] significantly affect the HRQOL of the person with the condition [36]. Both pain [37] and physical function [38] are key determinants of HRQOL in people with dementia, but also differ in terms of observability. Our objective in this study was to undertake a preliminary study on which perspective proxies consider as most useful while communicating about the health of others to health care professionals (HCPs) and explore their reasons for this choice across these different domains and clinical contexts. The focus was not on the evaluation of a specific measurement tool. Therefore, to orient participants and aim for consistency in how they interpreted the questions, communication with HCPs was used as a guiding context throughout the surveys. Anchoring the questions to communication with HCPs gave a shared reference point that could help participants reflect on real interactions rather than answering hypothetically.