Findings from this investigation indicate that patients admitted for inpatient hospital rehabilitation were able to predict their discharge health-related quality of life on the EQ-5D instrument with a moderate level of accuracy in each of the five broad domains. Patients’ health-related quality of life improved in all domains over the duration of their stay. Patients did not systematically overestimate or underestimate their discharge utility score derived from the individual domain responses. However, there was greater variability between anticipated and actual discharge summary scores for the poorer cognition group than the better cognition group. The small observed mean difference in EQ-VAS (2.3 points mean overestimation on the 100 point scale) is unlikely to represent a clinically meaningful difference[50–52].
There was no clear pattern of difference in predicting individual item responses across the individual health-related quality of life domains between the poorer and better cognition groups. The domain kappa scores and exact matches were comparable across cognition groupings and across domains. This may be attributable in part to the limited response options at discharge; where most respondents utilised only the two higher response options. However, the wider LOA among the lower cognition group for the EQ-5D utility index and EQ-VAS indicated that patients in the better cognition group had a smaller error margin than their peers in the lower cognition category.
Comparisons to previous research are difficult given the scarcity of empirical evidence on this topic. This research provides the first empirical evidence indicating that patients undergoing in-hospital rehabilitation have, at worst, moderately accurate expectations of their discharge health-related quality of life. This adds to the weight of foundational evidence supporting joint goal setting and patient centered models of care in rehabilitation contexts for older adults[53–56]. Patients who are well informed about prognosis, the impact of treatment and their future health-related quality of life are more likely to make informed treatment choices to target priority areas where meaningful improvements can be made[53–56]. This may also facilitate participation in therapies and adherence to treatment protocols[53–58]. In contrast, patients who overestimate their discharge health-related quality of life may become anxious, depressed or lose motivation as they fall short of their expectations.
An important consideration when interpreting implications from this study’s findings is that reports of patients’ anticipated health-related quality of life may act as a self-fulfilling prophecy. Those patients who felt helpless and anticipated poor levels of physical functioning, pain and depression may have been less likely to participate in therapies and other treatments. Similarly, patients with a positive outlook and high levels of self-efficacy may have maximized their rehabilitation outcome through active participation during their rehabilitation stay. However, it is not possible to draw strong conclusions in this regard from this observational study design as the degree to which this postulation was true amongst this sample remains uncertain.
Including a comparison between health-professional expectations and patient expectations of discharge health-related quality of life may be a worthwhile undertaking as a future research direction. The notion of patient expectations acting as a self-fulfilling prophecy would be supported if patients who anticipated a poorer outcome then their therapists, did actually achieve a poorer outcome in comparison to those where the patient and health professionals were in agreement. However, investigation of the influence of health-professional expectations on their patients’ expectation for discharge health-related quality of life would also be worthy of consideration. It is plausible that health professional expectations may act as a self fulfilling prophecy if patients considered to have greater potential to improve were provided with additional therapies, treatments or other resources.
It is also possible that patients in this study anticipated the level of functioning that would be required to be discharged safely back into the community and simply reported how they anticipated their health-related quality of life would be if they were to able to function at that level. They may have taken this heuristic response approach by surmising they would not be discharged until that level of functioning had been achieved. This would not necessary have been an undesirable outcome or changed the implications of these findings for patient centered models of care in rehabilitation settings where a common goal of hospital rehabilitation is to prepare patients’ for discharge. Accurate expectations held by patients regarding the level of functioning required for discharge may allow patients and health professionals to target interventions to priority areas of functioning required for successful community living.
Another factor worthy of consideration is whether patients recalled their anticipated discharge health-related quality of life responses and intentionally repeated the same responses at the discharge assessment. The investigators do not believe this occurred for four primary reasons. First, patients completed a wide range of routine assessments from multiple health professional disciplines during the first 72 hours of their admission to the participating rehabilitation unit. The large number of items assessed in this period offered natural protection against recalling their response to the six specific anticipated EQ-5D items. Second, some level of cognitive impairment is present among many patients in this older clinical group. This is evident in the MMSE scores, which indicated a large proportion of patients (including those in the ‘better’ cognition group) were likely to have some difficulty with memory and other rudimentary cognitive functions. Third, the long length of time between assessments (median 6 week length of stay) also provided natural protection against recalling responses from the initial assessment. Fourth, prior research has indicated that patients from comparable clinical groups do not give much consideration to health state scales when reporting their health-related quality of life and do not accurately recall responses to health-related quality of life reports completed at earlier assessments[4, 5, 60].
A number of caveats should be considered when interpreting findings from this investigation. The EQ-5D is a straightforward instrument with limited response options. In this study the 3-level multiple choice EQ-5D was used. This was a logical choice of instrument for this style of investigation where the objective was to examine a patient reported outcome capturing generic health-related quality of life information. Nonetheless correct prediction of the broad response categories did not require a detailed understanding of their discharge health state (e.g. no problems versus some problems walking around). This is likely to have contributed to a higher level of agreement than that which may have been observed if a more detailed prediction was required. It is also noteworthy that alternative instruments with different psychometric properties may have resulted in more (or less) accurate predictions depending on the qualities of the instrument (response options, sensitivity to change etc.).
There are several factors limiting the extent to which these findings can be generalized. First, all participants were from a single tertiary hospital. Patients from other hospitals or geographical locations may not have responded in the same way. Second, a single generic health-related quality of life instrument was used. Additionally, patients beginning the subacute rehabilitation phase of their recovery are likely to have already been provided with substantial information and advice about their prognosis. Patients in acute hospital care or community based settings may not have the same level of accuracy in anticipating their future health-related quality of life as the sample in this investigation.
A priority for future research following this investigation includes examining patients’ expectations across the continuum of care. This could potentially reveal valuable information regarding the role and timing of health education in joint decision making and patient-centered models of care. The nature of health information and focus of advice is likely to contrast across acute, subacute and community settings. This investigation has also exposed several opportunities for methodological improvement when undertaking future investigations of this nature among older adults. These opportunities include collecting a wider range of patient demographic clinical information that may influence ability to predict future health states. This may include recording patients’ level of education, evaluating patient depression or anxiety levels, determining the amount and content of health education already delivered to patients prior to study commencement and a potential comparison to health professionals accuracy in predicting patients’ future health-related quality of life.
On a broader note, it would also be valuable for future investigations to consider how positive or negative findings regarding patients’ preferences and expectations for their recovery should impact models of service delivery and individual treatment choices. There are many complex ethical considerations that could arise from this line of enquiry. For example, how should health-professionals with a duty of care to their patients respond if inaccurate patient expectations of disease progression (or potential recovery) result in a declination of evidenced based treatments to pursue an unadvisable course of action? How would this response differ depending on the potential severity of outcome or impact on third party dependents, such as children? Many issues in this sphere may initially seem straight forward in the context of patients being central decision makers in their care. Similarly, additional ethical complexity may be exposed if health-professionals do not have some degree of accuracy in anticipating future health-states. To this end, future research should investigate whether health professionals have the ability to predict patients’ future health-related quality of life in a variety of contexts, given that patients are likely to formulate their own expectations after taking into account the opinion of their treating health professionals. Expectations held by health-professionals are likely to directly influence therapies and other treatment options offered to patients.