Is the EuroQol a Valid Measure of Health-Related Quality of Life After Stroke?
Background and Purpose The EuroQol measures aspects of quality of life that are highly relevant to stroke patients. It is short and simple and many stroke patients can complete the form without help. However, its validity has not been adequately assessed after stroke. We therefore assessed its concurrent and discriminant validity in a group of prospectively studied stroke survivors.
Methods We assessed the validity of the EuroQol in a series of 152 patients with stroke who were all visited by a study nurse. The nurse gave the patients the EuroQol, the Frenchay Activities Index, a visual analogue pain scale, and the Hospital Anxiety and Depression Scale in the form of questionnaires to be self-completed where possible. The nurse interviewed the patient directly to assess disability using the Office of Population Censuses and Surveys Disability scale and Barthel Index.
Results The nurse assessed 152 patients; of these 92 were able to complete the EuroQol without help, the remaining 60 could only be assessed by interview. The EuroQol had reasonable concurrent validity; median scores on the relevant standard instruments varied significantly (and in the appropriate direction) for groups defined by their response to the relevant EuroQol domain. The EuroQol had reasonable discriminant validity since the responses enabled separation between patients with differing stroke syndromes and stroke severities. Accuracy for predicting outcome after stroke was good for both self-completed and interview-completed questionnaires.
Conclusions The EuroQol appears to have acceptable concurrent and discriminant validity for the measurement of health-related quality of life after stroke. It may be administered by either a questionnaire for self-completion in patients with mild to moderate stroke or by interview in patients with significant motor deficits.
Randomized trials of treatment for acute stroke have primarily used measures of impairment and disability to assess the effect of treatment on outcome.1 These measures, however, do not take into account difficulties with psychological and social functioning that can pose major problems for patients after stroke. These outcomes are important as they may be more relevant to patients and might not parallel physical outcomes in either qualitative or quantitative terms.2
There are several valid instruments for the measurement of single aspects of psychological, social, or physical outcome after stroke.3 4 5 6 However, the use of a series of such assessments may subject patients to an unacceptable burden and so reduce the overall frequency of response. The EuroQol is a generic instrument for the measurement of health-related quality of life (HRQoL), see “Appendix.”7 It provides a simple descriptive profile of health in five dimensions (mobility, self-care, social, pain, and psychological), each with three levels. The patient’s health state can therefore be classified into one of 243 (35) theoretically possible health states, each of which has been assigned a utility (ie, value to the patient).8 These utilities were assigned by a group of stroke-free individuals and therefore probably require further validation, but they might allow the EuroQol instrument to be used for the economic evaluation of health-care interventions and also the relative cost-utility of treatments for stroke compared with interventions for other diseases (eg, cardiac transplantation). The EuroQol also includes a visual analogue scale on which patients rate their own health between 0 and 100, thereby providing an overall numeric estimate of their health-related quality of life.
The EuroQol measures aspects of quality of life that are highly relevant to stroke patients. It is short and simple enough that many stroke patients (despite cognitive, motor, and sensory deficits) can complete the form without help (a higher proportion can complete the EuroQol than the Short Form-36).9 However, although the EuroQol is a valid assessment of HRQoL in the general population,10 its validity has not been adequately assessed after stroke. We therefore assessed certain aspects of its validity by comparing it with a variety of widely used and previously validated instruments in a group of prospectively studied stroke patients.
Validity is the degree to which an instrument measures what it is intended to measure. There are at least four aspects to validity.11 Face validity involves a subjective assessment of whether or not the instrument measures what it is intended to measure. Content validity is a subjective assessment of how well the domain of interest is sampled. We did not examine these aspects of validity quantitatively in the current study. Neither did we assess criterion validity, which involves comparison of the results of the instrument under study against those of a gold standard, because there is no gold standard for the measurement of HRQoL. We therefore focused on the concurrent (convergent) validity and compared outcome in each domain of the EuroQol with an assessment of function in the same domain using a standard instrument (see below). We examined the discriminant validity of the EuroQol in two ways: by assessing the relationships between the individual domains of the EuroQol and also by examining whether it could distinguish groups of patients with different types and severities of stroke (and therefore likely to have different HRQoL outcomes).
All patients with acute stroke (first or recurrent) who attend our hospital, as either inpatients or outpatients, are prospectively identified and assessed by experienced stroke physicians. These assessments include examination of the patients’ clinical status as well as an estimation by the clinician of the patients’ prestroke level of functioning with the Oxford Handicap Scale.12 Details of eligible patients are included in our stroke register. For convenience, we selected patients from the register who had survived at least 3 months after their stroke and who lived within an approximate 10-mile radius of the hospital (determined by scrutiny of their postal codes). The NHS central registry Office of Population Censuses and Surveys notified us of the death of any patients enrolled in the register. We excluded 12 patients whose vital status could not be confirmed by the Office of Population Censuses and Surveys (OPCS) on the day the sampling frame was assembled. For the current study, we aimed to examine the validity of the EuroQol in consecutive surviving patients from two distinct time periods. As a pilot study, we examined the validity of the EuroQol in a group of patients who might be considered longer term survivors. Of 98 consecutive patients registered between October 1, 1990 and May 18, 1991, 36 were still alive at the start of the present study. Twenty-eight of this group (78% of survivors) were able to be contacted and willing to participate. Three hundred and forty five patients were assessed and registered at our hospital between May 31, 1993 and the April 20, 1995. Of these, 193 were alive at the start of the current study. One hundred and twenty four of these patients (64% of surviving patients) were able to be contacted and willing to be interviewed. Both groups of patients will be considered together for the purpose of this report.
All patients were visited by a research nurse (F.W.) in their places of residence. The nurse administered the modified simple questions,5 EuroQol, Frenchay Activities Index,4 Hospital Anxiety and Depression Scale (HADS),13 and a visual analogue pain scale as questionnaires for self-completion when possible. When patients could not complete these questionnaires by themselves, they were administered by interview. The modified single questions and EuroQol were always administered first, to limit interaction with the subsequent questionnaires. The nurse assessed the Barthel Index and the OPCS disability scores by direct questioning at the end of the interview.6 14 The OPCS disability instrument includes a communication subscale that we used to identify patients who had significant problems with communication.
We initially assessed the concurrent validity by calculating the median score and interquartile range for the appropriate unidimensional instrument for each level in the corresponding domain of the EuroQol. We used the Kruskal-Wallis one-way ANOVA to assess the significance of the differences between these distributions. We also assessed the convergence between each domain of the EuroQol and the relevant standard instrument with the Spearman rank method. For the assessment of discriminant validity we determined the outcomes of the EuroQol for patients with different stroke syndromes (defined at baseline by clinical examination).15 We tested the discriminant validity further by examining responses to the EuroQol for patients with differing stroke severities. We based our assessment of stroke severity on their predicted prognosis at baseline. This predicted prognosis was calculated using a validated prognostic model designed to predict each patients’ probability of being alive and independent at 6 months.16 The variables in this model were age, preexisting disability, marital status, verbal component of the Glasgow coma score, the ability to lift both arms against gravity, and the ability to walk without help from another person.16 We also estimated the correlation between the domains of the EuroQol using the Spearman rank method with a two-tailed test of significance to determine the degree by which the different domains of the EuroQol discriminate between the different constructs of HRQoL. We determined the independent explanatory factors of overall HRQoL (based on the visual analogue scale) by multiple linear regression, using a method of forward selection of variables. We examined whether the assumptions of linearity and homogeneity of variance held for the data by plotting the distributions of the residuals. All analyses were performed using Access 2.0 software (Microsoft Corporation) and the statistical software package “SPSS for Windows” (Release 6.1, SPSS Inc).
One hundred fifty two patients participated in the study. Of these, 92 patients (61%) were able to complete the questionnaires without help; the remaining 60 patients (39%) could only be assessed by interview. Using the OPCS communication subscale (a score of over five implies “very difficult for strangers to understand or worse”), the interviewer rated six of the interviewed patients as having significant difficulties in communication. We excluded the data on these six patients from the analyses.
The patients were assessed at a median interval of 72 weeks after the onset of the index stroke (interquartile range: 43 to 104 weeks). Thirty-six percent of patients had not been admitted to the hospital for treatment during the acute phases of their strokes. Patients are normally only admitted to our hospital during the acute phase if inpatient nursing or rehabilitation is required. The characteristics of the patients recorded at the time of registration after their index stroke are shown in Table 1⇓, and their functional status at the time of assessment is shown in Table 2⇓. About one third of patients reported dependency in activities of daily living, one third were independent but had persisting problems, and the remaining third were independent and had no stroke-related problems. Patients who could not complete the EuroQol questionnaire themselves had significantly worse functional ability than those who could (P<.0001).
The median scores (and interquartile ranges) determined using the standard instruments are shown for groups defined by their responses to the mobility, self-care, social, pain, and psychological functioning domains of the EuroQol (Table 3⇓). The median scores determined using the standard instruments were ordered appropriately (increasing dysfunction reported with the EuroQol was associated with lower scores on the standard instruments) and differed significantly from each other (P=.0002 in all domains). We also examined convergence between each domain of the EuroQol and the standard instrument with the Spearman rank method (Table 4⇓). Bivariate correlations were moderately good for all domains except psychological functioning with the EuroQol and the depression subscale of the HADS. However, the patient estimates of overall HRQoL correlated most closely with this subscale of the HADS (Table 4⇓).
The pattern of outcomes reported by the questionnaire component of the EuroQol for each of the major stroke syndromes (defined by clinical examination at baseline) are shown in Fig 1⇓. For all domains, the worst outcomes were observed in patients with the most extensive cortical strokes (total anterior circulation strokes). The best outcomes were observed in patients with posterior circulation strokes. A similar pattern was observed for the numeric estimates of overall HRQoL. We also assessed discriminant validity in patients ordered by baseline stroke severity (lower tertile, the predicted probability of good outcome was 0.006 to 0.339; middle tertile, the predicted probability of good outcome was 0.339 to 0.530; upper tertile, the predicted probability of good outcome was 0.530 to 0.908). With the exception of the psychological functioning domain, better predicted prognosis was associated with better reported health status at follow-up, see Fig 2⇓. Patients with the highest predicted probability of good outcome at baseline had significantly higher reported overall HRQoL (mean score 73/100 for patients in top tertile versus 62/100 for patients in the middle tertile, and 62/100 for patients in the lower tertile, P<.05).
The bivariate correlations between each of the individual domains of the EuroQol instrument are shown in Table 5⇓. Mobility correlates best with social functioning (Spearman rank correlation coefficient, .56) and least well with the psychological outcome (Spearman rank correlation coefficient, .28).
The significant independent explanatory factors of the patient estimates of overall HRQoL are shown in Table 6⇓. An examination of the residuals confirmed that the assumptions of linearity and homogeneity had been met.
We have investigated some aspects of the validity of the EuroQol in stroke patients by comparing responses to the EuroQol with those to validated instruments for the assessment of mobility, self-care, social functioning, pain, and psychological functioning. We found that concurrent validity was good; patients who reported problems on the EuroQol also reported dysfunction with the relevant standard instrument for that domain. The EuroQol was valid both when patients could complete questionnaires themselves and among more severely affected patients who could only be assessed by interview.
The discriminant validity of the EuroQol was demonstrated by outcome profiles that distinguished between the major stroke syndromes. de Haan and coworkers have reported that lesion location and stroke severity influence HRQoL measured 6months after stroke with the Sickness Impact Profile17 ; in particular, patients with infratentorial strokes had better health status outcomes than patients with supratentorial strokes. Like de Haan et al we found that HRQoL was worst in all domains for patients with total anterior circulation strokes; and best in patients with posterior circulation strokes. Outcomes were similar in patients with partial anterior circulation strokes and with lacunar strokes, which is consistent with the epidemiological data that have suggested that partial anterior circulation strokes and lacunar strokes have very similar prognoses.15 The different stroke syndromes studied are defined by the extent rather than the severity of the neurological deficit, so we also looked to see if health status was different for patients with differing degrees of stroke severity. Patients with less severe strokes (ie, good predicted prognosis) had better outcomes as assessed by the EuroQol. The convergent relationships between the mobility, self-care, and social functioning domains of the EuroQol further support its construct validity, because functioning in all these domains is closely related to the patients’ physical functioning. We found good discriminant validity because mobility, self-care, and social functioning correlated much better with each other than with the domains assessing pain and psychological functioning.
It is difficult to assess the validity of the numeric estimates of overall HRQoL because this domain is difficult to define and is highly subjective. We could not assess its concurrent convergent validity because we could not identify other validated instruments that claimed to measure a similar outcome. However, the visual analogue scale for estimating overall HRQoL does at least appear to have discriminant validity, since we found differences in the mean estimates of overall HRQoL across the different stroke syndromes and severities. These follow a trend similar to that seen in the health outcome profiles; the lowest reported mean HRQoL was observed in patients suffering total anterior circulation strokes, and the highest mean estimate of HRQoL was reported by patients who had suffered posterior circulation strokes. These observed trends were not statistically significant. Linear multiple regression modeling revealed that the statistically significant independent explanatory factors of good overall HRQoL were the absence of depression on the HADS depression subscale, good social functioning on the Frenchay Activities Index, and the absence of pain on the Visual Analogue pain scale. These three variables explained approximately 38% of the variability in the estimates of overall HRQoL. The relationship between psychological, social functioning, and pain and the estimates of overall HRQoL in our study, provides strong support for the validity of the measurement of overall HRQoL by the EuroQol, because psychological outcome has been reported to be as important as physical disability in determining quality of life after stroke.18 Kwa and colleagues,19 using a similar visual analogue scale to investigate quality of life after stroke, reported that dependency (measured with the Rankin score), infarct volume, and aphasia were significant independent predictors of quality of life. However, only 22% of the total variation in the quality of life scores was explained by their model. In our model, physical functioning was not a significant independent explanatory factor of overall HRQoL after stroke. This was probably not because physical functioning was unimportant to the patients, but merely a reflection of the close relationship between a patient’s physical and social functioning (see Table 5⇑).
Although many of our patients were studied more than 1 year after their strokes, the study cohort included a good mix of “dependent,” “independent but not fully recovered,” and “fully recovered” patients. Very few patients (<10%) were unable to complete the EuroQol, either by themselves or by interview. This study population was ideal, because the measurement of quality of life outcomes is only meaningful in patients who can either complete assessments without help or communicate their views to an interviewer; the validity of proxy-completed EuroQol questionnaires is unclear and requires further study. Furthermore, it is probably appropriate not to test the validity of the EuroQol until at least 1 year after the index stroke because HRQoL may not be stable before then.20
Our data support the EuroQol as a useful measure of health status after stroke. Its simplicity is a definite advantage, because many stroke survivors find more complex instruments difficult to complete without help. Moreover, simple categorical data of the type generated by the EuroQol can convey a surprisingly large amount of information. Simple measures of this type are particularly well suited for use in large randomized controlled trials, audits, and screening projects.5 21 The data generated by the EuroQol may be presented in a variety of formats, including as a health status profile, patient-derived estimates of overall HRQoL, or as utilities generated from the health status profiles. Presenting EuroQol-derived outcomes as health status profiles (eg, Patient JD: Mobility—some problems, Self-care—unable to wash/dress, Activities—unable, Anxiety/depression—not, Pain—none) is likely to be immediately comprehensible to doctors, patients, and relatives. In contrast, the clinical significance of a reported change of x points in the overall HRQoL estimates or health state utility is less clear.
Thus in summary, the EuroQol appears to be a reasonably valid measure that can be administered as either a questionnaire for self-completion in patients with mild to moderate stroke or by interview in patients with significant neurological problems. If further studies confirm its reliability and feasibility in survivors of acute stroke, it could be usefully applied in a variety of ways, ranging from routine clinical screening of patients for psychosocial problems after stroke to the measurement of outcome in large randomized controlled trials and audit studies.
The EuroQol Questionnaire
Paul Dorman is supported by a UK Medical Research Council Training Fellowship. Jim Slattery, Martin Dennis, and Peter Sandercock are supported by a grant from the UK Medical Research Council. The study was also supported by a grant from Glaxo-Wellcome Plc.
- Received January 13, 1997.
- Revision received April 22, 1997.
- Accepted May 30, 1997.
- Copyright © 1997 by American Heart Association
Thorngren M, Westling B, Norrving B. Outcome after stroke in patients discharged to independent living. Stroke. 1990;21:236-240.
Wellwood I, Dennis M, Warlow CP. A comparison of the Barthel Index and the OPCS Disability Instrument used to measure outcome after acute stroke. Age Ageing. 1995;24:54-57.
Dolan P, Gudex C, Kind P, Williams A. A social tariff for EuroQol: results from a UK general population survey. York, Pa: University of York; 1995.
Dorman P, Slattery JM, Farrell B, Dennis MS, Sandercock PAG, and the United Kingdom Collaborators in the International Stroke Trial. A randomised comparison of the EuroQol and SF-36 after stroke. BMJ. 1997;315:461.
Hobart JC, Lamping DL, Thompson AJ. Evaluating neurological outcome measures: the bare essentials. J Neurol Neurosurg Psychiatry. 1996;60:127-130. Editorial.
Mahoney F, Barthel D. Functional evaluation: the Barthel Index. Md Med J. 1965;14:61-65.
Counsell C, McDowall M, Slattery J, Dennis M, in collaboration with the SEPIVAC and Perth Community Stroke Registries. Prediction of functional outcome following stroke using a validated model. Cerebovasc Dis. 1996;6(suppl 2):50. Abstract.
de Haan RJ, Limburg M, van der Meulen JHP, Jacobs HM, Aaronson NK. Quality of life after stroke: impact of stroke type and lesion location. Stroke. 1995;26:402-408.
Ahlsio B, Britton M, Murray V, Theorell T. Disablement and quality of life after stroke. Stroke. 1984;15:886-890.
Astrom M, Asplund K, Astrom T. Psychosocial function and life satisfaction after stroke. Stroke. 1992;23:527-531.