Patients and methods

All patients diagnosed with probable Parkinson's disease in a community based study on the prevalence of parkinsonism were included. The methods of patient ascertainment and study design were reported previously.11 In brief, the records of 15 general practices in the area of London were screened for patients with a suspicion of Parkinson's disease or parkinsonism, for patients with tremor with onset after the age of 50, and for patients who had ever received antiparkinsonian drugs. Participation rate was high (84%). Among all 202 patients seen, 124 were diagnosed with probable Parkinson's disease according to the UK Brain Bank criteria.12 A structured interview and a complete neurological examination including the Hoehn and Yahr scale (range 0-–5; the highest disease severity indicated by 5),13 the Schwab and England disability scale (range 0 to 100%; the greatest independence indicated by 100%),14 the motor section of the unified Parkinson's disease rating scale (UPDRS; range 0–56; the highest disease severity indicated by 56)15 and a mini mental state examination (MMSE; range 0–30; the best cognitive performance indicated by 30)16 were performed by a neurologist on the same day. The structured interview included questions about clinical features such as presence/absence of falls, forgetfulness, on-off fluctuations, dyskinesias, and hallucinations. All eligible patients were asked to complete the EQ-5D, the PDQ-39, the SF-36, and the Beck depression inventory, and a question on whether their general health in comparison with 1 year ago was improved, much the same, or worse, and to return them in a stamped, addressed envelope.

QoL INSTRUMENTS

The EQ-5D

The EQ-5D comprises five questions on mobility, self care, pain, usual activities, and psychological status with three possible answers for each item (1=no problem, 2=moderate problem, 3=severe problem; see appendix).7 8 A summary index with a maximum score of 1 can be derived from these five dimensions by conversion with a table of scores. The maximum score of 1 indicates the best health state, by contrast with the scores of individual questions, where higher scores indicate more severe or frequent problems. In addition, there is a visual analogue scale (VAS) to indicate the general health status with 100 indicating the best health status.

The PDQ-39

The PDQ-392 comprises 39 questions each with five different answer options, from which eight subscores (mobility, activities of daily living, emotional wellbeing, stigma, social support, cognition, communication, and bodily discomfort) and one summary index can be calculated. The maximum score of 100 indicates the worst level of problem.

The SF-36

The SF-3617 comprises 36 questions. Eight different subscores (physical and social functioning, physical and emotional role limitations, mental health, energy, pain, and general health perceptions), and a physical and mental summary score can be derived from these. The maximum score of 100 indicates the best possible health state.

Feasibility—Feasibility was determined by the number of missing responses in each instrument.

Validity—The validity of an instrument is the extent to which it measures what it claims to measure. Different types of validity can be assessed. Criterion validity is the correlation of a scale with some other measure of the disorder under study, ideally a “gold standard” which has been used and accepted in the field. In this study, we examined the correlations between the EQ-5D and other measures assessing similar aspects of wellbeing. Construct validation is an ongoing process whereby hypothetical “constructs” are tested. One such hypothesis might be that higher disease severity would correlate with lower scores indicating worse QOL.18 We tested the a priori hypothesis that patients with more severe Parkinson's disease, those with longer duration of disease and those with complications of disease and treatment would have worse QoL than those with early disease without these complications.

STATISTICAL ANALYSIS

The subscores and summary scores were calculated according to the respective scoring algorithms. We selected non-parametric tests for comparisons because of the ordinal nature of the scales. Group differences were analysed by the Mann-Whitney and Kruskal-Wallis tests and associations between scales with Spearman rank correlations. Statistical comparisons were performed between all variables that may hypothetically influence QoL scores, such as the duration and severity of disease, the presence of complications of advancing disease or treatment, or those that have been reported to be associated with reduced QoL, such as age or socioeconomic group.19 The validity of the EQ-5D was examined in several ways: firstly, by measuring the association between the EQ-5D summary index, the subscores, and the visual analogue score with the clinical ratings of disease severity in Parkinson's disease; secondly, by examining its correlation with the generic (SF-36) and the disease-specific (PDQ-39) measures of QoL; thirdly, by determining whether the EQ-5D successfully differentiated between patients with and without features of advanced disease and complications of its treatment, such as those with or without falls, dyskinesias, and dementia. Because we performed many comparisons we applied the Bonferroni correction. Given that we had 108 comparisons in the first part of the analysis, 133 in the second part of the analysis, and 30 in the third part, for an overall level of significance of α=0.05, the corrected p values for consideration in individual comparisons become p=0.0005, p=0.00038, and p=0.0017 respectively. All statistical analyses were performed with SPSS for Windows.20

Results

PROPERTIES OF THE EQ-5D

The median EQ-5D score was 0.62 (minimum −0.59 and maximum 1.0) with an interquartile range of 0.29. To examine whether the EQ-5D adequately reflects patient diversity, we inspected the distribution of the EQ-5D scores using histogram plots, which disclosed a similar distribution for the EQ-5D, SF-36 physical summary, Schwab and England and Hoehn and Yahr scores, with most patients in the low to moderate ranges of disease severity, disability, and impairment of QoL score.

FEASIBILITY

Ninety seven of the 124 patients returned the questionnaires, which constitutes a response rate of 78%. Their mean age was 73 (SD 11.3), ranging from 36 to 94 years with a mean disease duration of 5.8 (SD 4.9) ranging from 0.2 to 25 years, and a mean onset age of 67.6 (SD 12.5) ranging from 23.7 to 93.5 years. There were 50 men and 47 women who responded, mean Hoehn and Yahr stage was 2.4 (SD 0.9) ranging from l to 5), and mean MMSE was 27.5 (out of 30; SD 3.6). There were 20 (4.1%) missing responses on the EQ-5D and six (6.2%) patients had not completed the VAS. On the PDQ-3 9, 180 answers were missing (4.8%), and on the SF-36 286 answers (8.2%) were not completed. There was no statistical difference between responders and non-responders for age, disease duration, sex, Hoehn and Yahr stage, Schwab and England score, or MMSE score.

VALIDITY

Correlation with ratings of disease severity (table1)

Given the large number of comparisons, we decided to apply the Bonferroni correction. For this part of the analysis, correlations were considered significant when p<0.0005.

EQ-5D summary index—As expected, lower EQ-5D summary index scores correlated significantly with greater disease severity as measured by Hoehn and Yahr stages (n=77,r=−0.68, p<0.0001; table 1 and 2), with lower independence scores, as measured with the Schwab and England disability scale (n=77, r=0.66, p<0.0001) and with higher BDI scores (n=77, r=−0.63, p<0.0001; table 1). Most dimensions of QoL, as measured in the subscores, also correlated significantly with both disease scales but, as expected, in the opposite direction. Higher scores on the motor section of the UPDRS correlated with lower EQ-5D summary scores, but lower MMSE scores or longer disease duration did not correlate significantly with lower EQ-5D summary or sub-scores.

Visual analogue scale—The correlation of the VAS with measures of disease severity were in the same direction but less strong than those obtained for the EQ-5D summary index. The VAS correlated significantly with the degree of independence quantified with the Schwab and England scale (n=91, r=0.42, p<0.0001) and negatively with the depression score as measured by the BDI (n=9 1, r=-0.63, p<0.0001), but not with the motor section of the UPDRS and the Hoehn and Yahr scale.

Correlation with other qol scales (table3)

According to the Bonferroni method, for this part of the analysis correlations were considered significant when p<0.00038. The EQ-5D summary index correlated strongly and negatively with the PDQ-39 summary index (n=73, r=−0.75, p<0.0001) and most PDQ-39 subscores, and positively with the SF-36 physical summary score and most subscores (table 3). As expected, the strongest associations were between the EQ-5D and PDQ-39 or SF-36 subscores of both instruments that measure similar aspects of QoL, indicating good convergent validity. The VAS of the EQ-5D correlated significantly with all the SF-36 (positively) and PDQ-39 scores (negatively) with correlations ranging from 0.40 to 0.62, except the “social functioning” subscore of the PDQ-39, and the “emotional role” subscore of the SF-36, which did not correlate significantly with any subscale of the EQ-5D.

Sensitivity to clinical features of Parkinson's disease

According to the Bonferroni method, in this part of the analysis correlations were considered significant when p<0.0011.

EQ-5D summary index and visual analogue scale—Using a cut off score of 9 to distinguish mild from moderate/severe self reported depression as measured on the BDI, the EQ-5D summary index was significantly different for patients with and without depression (p<0.0001).21 Patients with cognitive deficits as indicated by an MMSE less than 25 had significantly lower EQ-5D scores than the group without cognitive deficits (p<0.0001). The features of the illness which had a significant impact on EQ-5D scores were presence or absence of falls (p<0.0001) and postural instability (p<0.0001), and on-off fluctuations and experience of hallucinations tended to influence EQ-5D scores (p=0.046 and p=0.007, respectively). In addition, patients who thought that their general health had deteriorated over the past year had significantly worse EQ-5D scores than those whose health had subjectively remained much the same (p<0.0001). By contrast, age (being older or younger than 70 years) had no significant impact on EQ-5D scores. There was no significant difference in EQ-5D scores between those patients who considered themselves improved compared with 1 year before and those who had not, but this may have been due to the few patients (n=6) who had noticed an improvement. Presence or absence of dyskinesias, gait impairment, incontinence, orthostatic symptoms, insomnia, visual disturbances, pain, speech, or swallowing impairment or a family history of Parkinson's disease did not differentiate significantly between patients with higher or lower QoL scores, as measured on any of the QoL instruments.

The visual analogue scale was less sensitive to these disease related features and only significantly differentiated groups with or without depression (p<0.0001), postural instability (p<0.0001), and deterioration of general health over the past year (p<0.0001), and tended to differentiate patients with falls and those with cognitive deficits (table 2).

Comparisons with other QoL measures—The correlation of the EQ-5D summary index and its subscores with the clinical scales was very similar to that of the PDQ-39 and the SF-36 summary scores (table 1). The correlation of the VAS with these scales, on the other hand, was weaker than that of the PDQ-39 summary index and SF-36 physical summary, except its correlation with the BDI. However, the association of the VAS of the EQ-5D with the clinical measures was better than the correlation of the mental summary of the SF-36 with these scales. Furthermore, the EQ-5D significantly differentiated between patients with and without particular clinical features such as postural instability, worsening of health over the past year, or depression as well as the PDQ-39 and better than the SF-36 physical score (table 2).

Discussion

We found that the EQ-5D summary index had good feasibility and validity in this community dwelling population of patients with Parkinson's disease. It correlated moderately with the measures of disease severity and disability as well as with depression, reflecting its sensitivity to disease severity coupled with sufficient distinctiveness. It also correlated strongly with the disease specific PDQ-39 summary index and its subscores and, to a lesser degree, with the physical score of the generic health status instrument SF-36. Its sensitivity to the clinical features of Parkinson's disease was similar to that of the disease specific instrument, whereas the SF-36 was less sensitive. These results are similar to those reported for the disease specific QoL instrument, the PDQ-39.2 The validity of the EQ-5D summary index in this sample of patients with Parkinson's disease was high and better than reported for other disorders such as epilepsy,22 or lower limb ischaemia,23 and comparable to that of the disease specific PDQ-39. A possible explanation for this finding could be that the correlation of the PDQ-39 with clinical scales was poorer than in previous studies, which were either clinic based or used a selected sample (members of the Parkinson's disease society).2 However, the correlation of the PDQ-39 and the SF-36 with the Hoehn and Yahr and other clinical scales found in this study was very similar to previously reported results,4 24 confirming their validity in an unselected sample of patients with Parkinson's disease. Therefore, at first sight it may seem surprising that a non-specific generic measure which only comprises five questions with three levels of response option correlated well with the clinical scales. However, it should be noted that these five questions cover five core areas which are particularly important in Parkinson's disease: mobility, self care, usual activities, pain, and anxiety/depression. They represent the aspects of life most affected by Parkinson's disease and its typical features. Other features such as stigma, social support, communication, and cognition are not covered, but the areas most relevant for QoL in Parkinson's disease, as found by Fitzpatrick et al,24 are also those covered in the EQ-5D: difficulty with leisure activities, difficulty looking after the home, problems writing clearly (covered in “usual activities”), difficulties getting around in public places, problems walking half a mile, difficulty carrying shopping bags (covered in “mobility”), problems in doing up buttons or shoelaces (covered in “self care”), worrying about the future (covered in anxiety/depression) and bodily aches and pains (covered in “pain”). The EQ-5D may be a particularly suitable generic instrument for patients with Parkinson's disease as it is a very mobility based questionnaire. It may also be more valid in this patient group than, for example, in patients with epilepsy or headache as there are no disease free intervals which could distort the true extent of the problem when asked for the “health state today”. It may also be more suitable for patients with Parkinson's disease than for patients with congenital or early onset disorders as Parkinson's disease usually starts late in life so that patients can easily compare their current to their previous health state. The VAS part of the EQ-5D, although easy to complete, was not as strongly correlated with clinical measures or sensitive to the presence or absence of clinical features. On the other hand, it correlated highly with depression. The SF-36 was a generally less sensitive measure (lower correlation with clinical scales, and poorer differentiation between patients with and without features of advanced disease) than the EQ-5D, summary index and the PDQ-39, and additionally had a higher percentage of missing answers. Moreover, patients often added comments to its questions as they did not think that these were appropriately reflecting their problems, such as

”I have arthritis and can therefore not walk long distances”.

A shortcoming of this study is that we did not formally assess the EQ-5D's sensitivity to change, which is important if a QoL instrument is to be used in clinical trials to test effectiveness of treatment. This was a cross sectional study with only one evaluation of QoL. However, patients who reported deterioration of their general health over the past year had significantly worse QoL scores than those whose health had not deteriorated or improved, indicating the instrument's ability to reflect change in disease severity. Furthermore, it may be argued that a selection bias towards patients with less severe disease or better cognitive scores in those who completed the questionnaire cannot be excluded. However, disease severity and cognitive scores did not differ between responders and non-responders (all of whom were examined), and the participation rate in the overall community based prevalence study was high (84%). Nevertheless, those who declined participation were significantly older than those who were seen (p<0.001), and such a bias can therefore not be entirely excluded. Finally, the multitude of comparisons makes the study, despite application of the Bonferroni method, an exploratory one, and the results will later have to be confirmed in a subsequent set of data.

We have shown the EQ-5D to be feasible and valid in a community dwelling population of patients with Parkinson's disease. The findings of this study suggest that the EQ-5D summary score differs between patients grouped by certain clinical features, and correlates well with clinical scales, similar to the disease specific PDQ-39 summary score. However, the significance of individual comparisons will have to be confirmed on new sets of patients, and its sen-sitivity to change needs to be assessed in further studies.

• Download figure
• Open in new tab
• Download powerpoint