Although Parkinson's disease is most prominently identified with physical symptoms such as tremors and akinesia, this disease has a substantial impact beyond motor impairment and physical disability with an on overall reduction in all health-related quality of life dimensions including social and emotional well-being. To date, the relatively limited application of existing tools for the measurement of health-related quality of life (HRQoL) has made it difficult to compare the loss of HRQoL in PD to that experienced by individuals with other chronic conditions . Using a health utilities "index" approach we found a substantial reduction in HRQoL in a cohort of individuals attending a PD specialty clinic, similar to other reports [16, 31–34]. However, we also found that diminished HRQoL as measured by changes in health utilities was closely correlated with changes in scores on a PD-specific disease severity measure, the UPDRS.
HUI and UPDRS
We are aware of at least one other prior effort to map health utilities onto UPDRS scores ; Siderowf and colleagues identified agreement between overall UPDRS scores and the HUI-II, as well as other utility-based instruments. Our mean utility estimate (0.42) is lower than that reported by Siderowf et al. (with a mean utility of 0.74), and this may reflect the fact that our cohort was assembled at a clinic to which patients were referred due to complexities of medical management, and could also reflect a different profile of co-morbid conditions in the two populations. It may also, in part, reflect the fact that HUI-III includes domains (such as vision and hearing) that are not included in HUI-II, and which may be sources of diminished global quality of life in the age group at greatest risk of PD.
In comparison to the Siderowf study, our study further refined the relationship between health utilities and UPDRS scores. Perhaps surprisingly, we found that these reductions were most strongly correlated with the self-care component of the UPDRS (UPDRS-II), rather than the UPDRS-III motor sub-score. This finding serves as an important reminder that loss of independence may be an important source of morbidity in individuals with PD. As we demonstrated in regression analyses (Figure 1), the correlation between UPDRS-II and HUI scores was so substantial that it may be possible to generate approaches whereby existing disease-specific scores can be transformed into health utility estimates, for the purposes of comparing the health burden associated with PD to that seen in other chronic medical conditions, and in order to utilize HRQoL as the outcome of interest in economic evaluations of novel therapies for PD.
Predictors of Low Baseline HRQoL
Other important predictors of low baseline HRQoL in this study included reductions in S+E disability scores, and higher axial sub-scores (PIGD). Though health-related quality of life and self-care ability in PD are inextricably linked to severity of motor dysfunction, the relationship between motor impairment and reduction in health-related quality of life may be complex and indirect, as demonstrated by our failure to find an independent relationship between UPDRS motor III sub-scores and HUI, after controlling for UPDRS-II and other scores. These results are consistent with previous findings that motor impairment in and of itself does not reduce health-related quality of life but the functional consequences of poor motor function including loss of self-care capabilities, inability to ambulate and loss of independence and its emotional consequences that may provide the link between physical impairment and low HRQoL [16, 35].
We failed to find an association between either cognitive impairment or evidence of depression and low HRQoL, similar to one other study . However this lack of association may reflect the fact that our study population was relatively intact cognitively (mean MMSE = 27.5/30). Nonetheless, it is also well-recognized that the MMSE is insensitive to capturing early cognitive decline in PD patients  and therefore we may not have identified individuals with subtle cognitive changes. Alternatively, it is possible that the mild cognitive changes in this cohort were insufficient to contribute substantially to low HRQoL.
Six prior longitudinal studies have evaluated HRQoL in PD. The first, based on a community-based cohort, found no relationship between any baseline clinical characteristics and reduction in HRQoL . Another study  using both disease specific measures (PDQL and PDQ-39) and a generic utilities-based instrument (EQ-5D) did not identify change in HRQoL over time using the EQ-5 D. However, low disease-specific quality of life scores in general were predicted by depression, motor complications, cognitive impairment, and gait instability. The lack of change in the EQ-5 D was attributed to short follow-up time (12 months); the authors also postulated that the EQ-5 D was not sufficiently sensitive to pick up the subtle changes that may have occurred over only 1 year. A third study, by Fitzpatrick and colleagues , identified correlation between a generic HRQoL measure (SF-36) and a disease-specific HRQoL measure (the PDQ-39) (neither of them scaled nor preference-based) and also identified correlation between these measures in change over time , similar to the findings reported here.
Forsaa et al.  prospectively followed patients for 4 to 8 years, with HRQoL measured using the Nottingham Health Profile (NHP), a validated generic instrument. This study found that the greatest predictor of reduction in HRQoL was decline in physical mobility (as captured in part by worse S+E scores and higher H+Y scores), though depression and sleep disturbance were also important contributing factors; Contrary to our findings, UPDRS-II sub-score was not found to predict reduction in HRQoL.
Marras et al. also evaluated predictors of diminished HRQoL  using a large cohort from the DATATOP database. HRQoL was evaluated using the physical component sub-score (PCS) and mental component sub-score (MCS) of the SF-36, a generic HRQoL scale. Depression and self-rated cognitive function predicted low PCS; low MCS was predicted by older age and S+E disability scores at baseline. HRQoL and PIGD sub-scores declined in parallel over time. As in our study, these authors suggested that physical impairments associated with PD did not directly reduce health-related quality of life. Rather, lower health-related quality of life reflected diminished ability to perform ADLs, with increased dependence on others. Most recently, Brown and colleagues evaluated the relative performance of SF-36 and PD-specific quality of life instruments in predicting change in criterion indices of disease severity and quality of life (measured with a visual analogue scale); disease-specific measures outperformed generic measures in explaining variance in criterion indices, though SF-36 was more responsive to change over time .
Change Over Time
Health utility estimates and most indices of PD severity were relatively stable over the course of our study, which may reflect the relatively short duration of study, and perhaps also the fact that notwithstanding the decline in status expected with a degenerative disease like PD, at least some subjects may have experienced improved health-related quality of life as a result of optimized medical management following referral to the PADRECC. Changes in utility were correlated with changes in multiple PD-specific measures, though our ability to document relationships between changes in health-related quality of life and changes in UPDRS-II scores were limited by the fact that repeated UPDRS-II scores were available in only a small subset of subjects.
This study had several important limitations. Our failure to identify a link between depression and low HRQoL contrasts with the results of other studies [15, 38–45] and could reflect misclassification of depression, which was based on records of physician diagnosis or prescription of antidepressant medication, rather than through standardized prospective assessment. Studies that have identified associations between depression and low HRQoL have generally confirmed depression using validated mood assessment instruments. As such, our failure to find an association between depression and HRQoL in patients with PD should be interpreted with caution.
Other limitations of this study relate to the generalizability of findings in a cohort of male U.S. veterans: our findings may not be generalizable to non-veterans or to women, as they were not represented in our cohort. Previous epidemiological surveys have suggested gender differences in PD; Men have been described to have earlier symptom onset , increased incidence of cognitive impairment , increased risk of pathological gambling  and decreased rates of depression . Women have cited greater disability and lower health-related quality of life in comparison to men with PD . Finally, as discussed above, we had a limited ability to assess changes in UPDRS-II scores over time as these measurements were repeated infrequently.