Predictors of clinically significant quality of life impairment in Parkinson’s disease

Quality of life (QOL) plays an important role in independent living in Parkinson’s disease (PD) patients, being crucial to know what factors impact QoL throughout the course of the disease. Here we identified predictors of QoL impairment in PD patients from a Spanish cohort. PD patients recruited from 35 centers of Spain from the COPPADIS cohort from January 2016, to November 2017, were followed up during 2 years. Health-related QoL (HRQoL) and global QoL (GQoL) were assessed with the 39-item Parkinson’s disease Questionnaire (PDQ-39) and the EUROHIS-QOL 8-item index (EUROHIS-QOL8), respectively, at baseline (V0) and at 24 months ± 1 month (V2). Clinically significant QoL impairment was defined as presenting an increase (PDQ-39SI) or decrement (EUROHIS-QOL8) at V2 ≥ 10% of the score at baseline (V0). A comparison with a control group was conducted for GQoL. GQoL did not change significantly in PD patients (N = 507; p = 0.686) or in the control group (N = 119; p = 0.192). The mean PDQ-39SI was significantly increased in PD patients (62.7 ± 8.5 years old; 58.8% males; N = 500) by 21.6% (from 16.7 ± 13 to 20.3 ± 16.4; p < 0.0001) at V2. Ninety-three patients (18.6%) presented a clinically significant HRQoL impairment at V2. To be younger (OR = 0.896; 95% CI 0.829–0.968; p = 0.006), to be a female (OR = 4.181; 95% CI 1.422–12.290; p = 0.009), and to have a greater increase in BDI-II (Beck Depression Inventory-II) (OR = 1.139; 95% CI 1.053–1.231; p = 0.001) and NMSS (Non-Motor Symptoms Scale) (OR = 1.052; 95% CI 1.027–1.113; p < 0.0001) total scores from V0 to V2 were associated with clinically significant HRQoL impairment at the 2-year follow-up (Hosmer–Lemeshow test, p = 0.665; R2 = 0.655). An increase in ≥5 and ≥10 points of BDI-II and NMSS total score at V2 multiplied the probability of presenting clinically significant HRQoL impairment by 5 (OR = 5.453; 95% CI 1.663–17.876; p = 0.005) and 8 (OR = 8.217; 95% CI, 2.975–22.696; p = 0.002), respectively. In conclusion, age, gender, mood, and non-motor impairment were associated with clinically significant HRQoL impairment after the 2-year follow-up in PD patients.

regard to how the QoL of PD changes throughout the course of the disease, there is much less information [18][19][20][21][22][23] and prospective longitudinal studies are needed. In clinical practice, it is important to know what factors worsen PD patients' QoL with the intention to carry out effective interventions. Known information limited by factors from the studies such as the sample size, the differences between scales used for assessing QoL, the different types of QoL assessed, being a non-multicenter study, the absence of a control group, and/or the lack of a global evaluation including different aspects that could impact on QoL [18][19][20][21][22][23] . In addition, the impact of some complications on QoL in advanced PD has been analyzed before 24,25 . However, it is not clear what the significance of shortterm changes in QoL is in early PD patients or what factors contribute to it when an extensive assessment considering motor and NMS is performed 26 . It is remarkable that NMS occur not only in advanced but also in the early stages of PD. Some symptoms, for example, olfactory deficit, constipation, rapid-eyemovement sleep behavior disorder, and depression, can even precede the appearance of motor symptoms by many years 1 . By the contrary, others such as psychosis or dementia are not present. The first years are conditioned by the acceptance of the diagnosis, but in general, the patient has greater autonomy. In this context, it is essential to know what influences the changes in the PD patient QoL perception with the intention of being able to act as soon as possible.
The aim of the present study was to (1) analyze the change in HRQoL and GQoL in PD patients from the COPPADIS cohort after the 2-year follow-up, (2) to compare with a control group, and (3) to identify predictors of clinically significant QoL impairment in the PD group. Finally, a subanalysis was conducted in a subgroup of patients with early PD (≤5 years of disease duration).
The change in the score of other scales from V0 to V2 in PD patients and controls is shown in Table 1.
Predictors of the change in the PDQ-39SI from V0 to V2 Finally, similar results were observed in both groups, the whole cohort and the early PD subgroup, when a linear regression model was considered (PDQ-39SI change from V0 to V2 as dependent variable) (Supplementary Table 2). To be a female (β = 0.17; p < 0.0001) and change in UPDRS-III (β = 0.23; p < 0.0001), FOGQ (β = 0.20; p < 0.0001), and NMSS (β = 0.37; p < 0.0001) scores provided the highest contribution to the model (adjusted R-squared 0.45) in the whole cohort. In early PD patients, the variables associated with HRQoL change at the 2-year follow-up were the same (Supplementary Table 2). When the ADLS score was included in the model, the results were similar but with the ADLS as an independent variable associated with HRQoL change too (β =

DISCUSSION
In this longitudinal follow-up study, we report that there is a significant HRQoL impairment in PD patients in the short-term and that impairment in the motor status during the OFF state (UPDRS-III), increased gait problems (FOGQ), and increased NMS burden contribute to it. Specifically, mood impairment and NMS burden increase were independent factors associated with clinically significant HRQoL impairment at the 2-year follow-up, which one was present in about every 5 patients. Moreover, the results indicate that it will be especially important to be vigilant about clinically significant HRQoL impairment in women and younger patients. After a 2-year follow-up, PD patients from the COPPADIS cohort demonstrated impairment in motor function (H&Y, UPDRS-III, UPDRS-IV, FOGQ). The increase of motor impairments measured with the UPDRS were in agreement with other studies 27,28 . Also, significant changes in NMS were observed in the NMS burden as a whole, pain, fatigue, and cognition, but not in controls. These results aligned with previous longitudinal studies indicating that the severity of NMS in PD tends to become progressively worse with the course of the disease and also indicate that non-motor evaluation is complementary to measuring PD progression 19,26,[29][30][31][32] . With respect to the QoL, although more than a half of PD patients presented a PDQ-39SI score at the 2-year follow-up higher than at baseline, only 18.6% presented HRQoL impairment as clinically significant. In a previous study with 707 PD patients followed prospectively for the 2-year as well, 17% worsened clinically while 584 were rated as stable 29 . The results can be vary due to the definition of QoL impairment as clinically significant 31,33,34 . Based on the postal reply of 728 PD patients, Peto et al. 34 determined that 1.6 points worsening on a PDQ-39SI is the minimal clinically important difference threshold. More recently, Horváth et al. 31 considered the most optimal estimates threshold for PDQ-39-SI in + 4.22 points for detecting minimal clinically important worsening. However, there is no "gold standard" methodology of estimating the minimal important difference and as the degree of improvement is conditioned by the baseline score; therefore, the use of a percentage might be more appropriate 35,36 . Patients appear to be able to detect changes of 7-10% on QoL instruments or pain scales 36 . In our case, the minimal important difference was considered as an increase of 10% or more in the PDQ-39SI score 33,[35][36][37] . Ten percent of the mean score of the PDQ-39SI in our study represents 1.6 points; therefore, similar to the proposal of Peto et al. 34 . However, in a patient with a higher baseline PDQ-39SI score, for example, 50 points, the minimal clinically significant worsening change should be 5 points. Hence, in less than 1 in 3 patients who had an increase in the PDQ-39 score, this was considered clinically significant. In any case, it seems clear that even in a relatively short follow-up period, patients with PD experience a significant decrease in HRQoL 21,29 . However, as Reuther et al. 22 reported in 145 PD patients after a 12month follow-up, there doesn't seem to be a significant change in QoL generic scales. For assessing the NMS as a whole, we used the NMSS. To date, this scale has been used in more than 100 clinical studies and trials and it has shown to be capable of detecting longitudinal changes in NMS, where studies have shown differential changes over time of several of the NMSS domains 32,38,39 . Moreover, it has been demonstrated a consistent and strong correlations between NMSS burden and HRQoL measures 32,[40][41][42] . In our study, a very clear difference in the change of NMS burden was observed between patients with and without clinically significant HRQoL impairment. Changes in all domains of the NMSS scale correlated with QoL changes. Similarly, previous studies observed a correlation between NMS burden assessed with the NMSS and QoL changes over time 19 . Moreover, in our analysis, NMS burden progression was an independent factor related to HRQoL impairment. Prakash et al. observed for the first time that nonmotor problems provided a better prediction of the change of QoL in 227 PD patients over a 2-year follow-up period 19 . However, they did not provide the variance value of the model, many factors potentially affecting QoL were not included, and what they considered was the baseline NMSS score. On the contrary, in this study we wanted to analyze in detail what changes in many aspects of the disease observed after the 2-year follow-up contributed to a worsening in the patients´QoL. So, several variables were included, the results of the model represented 70% of the variance when HRQoL changes were considered, and the changes in all variables were adjusted to the scores at baseline. To our best knowledge, this is the first longitudinalprospective study analyzing in such detail which are the predictors of QoL impairment in a large sample of PD patients. Reinforcing the idea that the progression of NMS is pivotal to the worsening of the QoL throughout the evolution of the disease, improvements of NMS were associated with improved QoL in advanced parkinsonian patients during 2-year treatment with levodopa-carbidopa intestinal gel infusion therapy 43 . In line with this, Erro et al. observed that NMS significantly affected QoL in PD, demonstrating that this was especially the case when patients were in their honeymoon period (during which time the side effects of the disease aren't too disabling and there is a response to medications) 44 . In the subgroup of early PD patients from our study, the change in the NMSS total score at 2-years was one of the most significant contributors to HRQoL impairment.
Another important factor is mood. Like in other studies, the mean score of BDI didn't change over time 18,22 , suggesting that depression-type frequency does not appear to change over time in PD 45 . Cross-sectional studies have reported the clear contribution of depression or a worse mood to a poorer QoL in PD patients 3,12,13 . In fact, it was observed in the COPPADIS baseline cross-sectional analysis 14 . However, to our knowledge, this is the first time that mood worsening is identified as an independent factor associated with clinically significant HRQoL impairment in PD patients. This subgroup of patients (N=93) presented a mean increase in the BDI-II score of 4.5 points at the 2-year follow-up and specifically, an increase in ≥5 points multiplied by 5 the probability of presenting a clinically significant HRQoL impairment, independent of other factors. Reuther et al. 22 identified depression as the strongest predictor for reduced HRQoL in 145 PD patients after 1-year follow-up. However, we identified the change in the score of the BDI-II as a predictor of clinically significant HRQoL impairment after adjustment to BDI-II score at baseline. From a practical point of view, our findings suggest an important role of the neurologist being alert to a possible worsening of mood, as well as greater NMS burden, in patients with PD throughout the evolution of the disease since this is what impacts on the patient's QoL. Knowing what impacts on the QoL and contributes to its worsening, depending on the variable, intervention measures with the intention of correcting them can be proposed 46 . Studies demonstrating a QoL improvement correlated with mood and NMS burden improvement have been published 47 . With regards of the results observed here, it should be necessary to be alert about mood and NMS burden changes over time, especially in younger patients and females. A mildly significant gender difference in disability and QoL reporting has been noted, with women citing greater disability and reduced QoL 48,49 . Depression and fatigue were the major causes of low HRQoL in women even in the early phases of PD 50 . To attenuate this sex difference in disease experience, psychological distress screening and management (particularly targeting females) should be considered as part of PD clinical care 23 . Moreover, QoL, as measured on the PDQ-39, is significantly worse in youngonset PD patients than in older-onset PD patients, and youngonset PD patients also experience loss of employment, disruption of family life, greater perceived stigmatization, and depression than do older-onset PD patients 51,52 .
The most important limitation of this study is the fact that information about follow-up was recorded only in 524 patients of 695 initially included in the study (75.5%). Of them, data for the PDQ-39, PQ-10, and EUROHIS-QOL8 at baseline and at V2 was available in 500, 503, and 507 PD patients, respectively. Thirty-eight patients (5.5%) dropped out of the study (1 death; 2 with change in diagnosis; 35 other reasons) at the 2-year follow-up and 132 (19%) were not assessed. However, this is a limitation observed in other prospective studies. Of 7507 PD patients, follow-up data was available only for 4680 participants (62.3%) 53 . In the study of Antonini et al. 29 , 707 PD patients from 1142 initially included (61.9%) were evaluable at 24 months. An important second limitation is that PD patients older than 75 years old were excluded from participation by COPPADIS study protocol 14 , which leads to an early PD bias in this cohort. For some variables, the information was not collected in all cases. Moreover, this is a multicenter mono-country study, being the ideal for this type of studies the participation of patients from different parts of the world, so the results should be considered with caution when extrapolating them to the general PD population (i.e., race, country healthcare, etc.). By the contrary, strengths of our study include a very complete assessment, the large sample size, a prospective longitudinal follow-up design, the fact that this analysis was "a priori" planned as one objective of the multicenter COPPADIS project 16 , and the extensive clinical and demographic information recorded.
The findings of this study have important implications in daily clinical practice. In a disorder like PD in which one there is no a cure, treatment is symptomatic and the aim is to improve the patient's QoL. This is complex because many factors influence QoL in PD. Furthermore, PD is a complex disorder with many manifestations and with a great variability in its progression among patients. Regarding this study observations, some important points should be considered in daily clinical practice. First, a complete assessment of the patient with PD periodically including motor status, NMS, QoL and disability should be the ideal practice. Second, NMS progression contributes significantly to a QoL worsening and it is crucial its evaluation. Very interestingly, we reported very recently that PD patients from the COPPADIS cohort with a lower H&Y stage but a greater global NMS burden may have a worse QoL than patients with a higher H&Y stage but lower global NMS 54 . Third, mood is another key factor to consider whenever we evaluate the patient in clinical practice. Fourth, we have to keep in mind that mood impairment and global NMS progression predict a patient´s QoL worsening. Finally, we should be especially careful in all of the above in the case of a female patient and in young patients.
A problem in clinical practice is the lack of time to evaluate the patient. For the PD patient, to bring adequately covered questionnaires to the consultation, for example with the help of nursing staff, or even in the future with mobile applications that transfer the data to the patient's medical record, it could be a possibility that facilitates the complete and comprehensive assessment. In general, it is something that is not done today, and proof of this is the alarming lack of literature about the global progression of the disease including NMS in large cohorts of patients. More studies with large PD cohorts and long-term follow-up are required. Our aim with the COPPADIS cohort is to follow for 5 years 55 . Collecting data from different cohorts and making comparisons would also be of great interest.
In conclusion, the present study observes HRQoL impairment in PD patients in a short 2-year follow-up, even in early PD patients, but not the GQoL. A younger age, to be a female, and mood and NMS burden impairment were associated with clinically significant HRQoL impairment after the 2-year followup. The progression of NMS is pivotal in the worsening of the QoL throughout the evolution of the disease in PD, and it is necessary to keep in mind to ask for mood or NMS changes, especially in females and young patients.

METHODS
PD patients and controls who were recruited from January 2016 to November 2017 (baseline visit; V0) and evaluated again at the 2-year follow-up (V2) from 35 centers of Spain from the COPPADIS cohort 56 , were included in the study. Methodology about COPPADIS-2015 has been previously published 57 . This is a multicenter, observational, longitudinalprospective, 5-year follow-up study designed for analyzing disease progression in a Spanish population of PD patients. Specifically, 17 objectives were proposed in the protocol 55 . Even though the recruitment period ended in October 2017, the prospective follow-up phase is ongoing. Patients, caregivers (patient´s primary caregiver), and controls (subjects without PD and any other severe and disabling concomitant disorder) were included 55 . Annual visits from V0 (baseline) to V5 (60 moths ± 3 months) are conducted to the patients and at V0, V2, V4, and V5 to the controls and caregivers. All patients included were diagnosed according to UK PD Brain Bank criteria 57 . Exclusion criteria 55 were: non-PD parkinsonism, dementia criteria (Mini Mental State Examination [MMSE] ≥ 26), age < 18 or >75 years, inability to read or understand the questionnaires, to be receiving any advanced therapy (continuous infusion of levodopa or apomorphine, and/ or with deep brain stimulation), and presence of comorbidity, sequelae, or any disorder that could interfere with the assessment.
Information on sociodemographic aspects, factors related to PD, comorbidity, and treatment were collected. V0 and V2 evaluations included 55 : (1) 70 . In patients with motor fluctuations, the motor assessment was made during the OFF state (without medication in the last 12 h) and during the ON state. On the other hand, the assessment was only conducted without medication in patients without motor fluctuations. The same evaluation as for the patients, except for the motor assessment, was conducted in control subjects at V0 and at V2 (2 years ± 1 month). Three scales were used to assess QoL at V0 and at V2 28 : (1) the 39-item Parkinson's disease Questionnaire (PDQ-39) 71 , (2) a rating of global perceived QoL (PQ-10) on a scale from 0 (worst) to 10 (best) 13 , and (3) the EUROHIS-QOL 8-item index (EUROHIS-QOL8) 72 . The PDQ-39 is a PDspecific questionnaire that assesses the patients' HRQoL. There are 39 items grouped into 8 domains: (1) Mobility (items 1 to 10); (2) Activities of daily living (items 11 to 16); (3) Emotional well-being (items 17 to 22); (4) Stigma (items 23 to 26); (5) Social support (items 27 to 29); (6) Cognition (items 30 to 33); (7) Communication (items 34 to 36); (8) Pain and discomfort (items 37 to 39). For each item, the score may range from 0 (never) to 4 (always). The symptoms refer to the 4 weeks prior to assessment. Domain total scores are expressed as a percentage of the corresponding maximum possible score and a Summary Index is obtained as average of the domain scores. The EUROHIS-QOL8 is an 8-item GQoL questionnaire (quality of life, health status, energy, autonomy for activities of daily living, self-esteem, social relationships, economic capacity, and habitat) derived from the WHOQOL-BREF. For each item, the score ranges from 0 (not at all) to 5 (completely). The total score is expressed as the mean of the individual scores. A higher score indicates a better QoL. In controls, only the PQ-10 and the EUROHIS-QOL8 were assessed. Clinically significant HRQoL impairment was defined as presenting an increase in PDQ-39SI score at V2 ≥ 10% of score at baseline (V0) whereas GQoL impairment as presenting a decrement in PQ-10 and/or EUROHIS-QOL8 score at V2 ≤ 10% of score at baseline (V0) 33 . Taking into account that in the COPPADIS cohort the range of disease duration varies from <1 year to 30 years and based on the general response to treatment and progression of symptoms in PD and considering a recent publication of this same cohort 73 , patients with ≤5 years of disease duration were considered as early PD patients.

Data analysis
Data were processed using SPSS 20.0 for Windows. For comparisons between patients and controls, the Student's t-test, Mann-Whitney U test, Chi-square test, or Fisher test were used as appropriate (distribution for variables was verified by one-sample Kolmogorov-Smirnov test). The Wilcoxon-signed rank test was performed to test whether the mean differences of the PDQ-39SI, PQ-10, and EUROHIS-QOL8 scores and the individual PDQ-39SI and EUROHIS-QOL8 domain scores between the two visits (V0 and V2) were significant. This test and/or the marginal homogeneity test were applied for other scales for analyzing the change from V0 to V2. Spearman's or Pearson's correlation coefficient, as appropriate, were used for analyzing the relationship between continuous variables. Correlations were considered weak for coefficient values ≤0.29, moderate for values between 0.30 and 0.59, and strong for values ≥0.60.
Clinically significant QoL impairment was expressed as a percentage and it was only calculated if the change between scores (PDQ-39SI; PQ-10; EUROHIS-QOL8) from V0 to V2 was significant. For determining predictive factors of QoL impairment, a logistic regression model (QoL impairment as dependent variable) was performed. The model was well-planned, as recommended by best-practice methods 74 , in which known and presumably predictor variables affecting QoL changes (dependent variable) were included: change from V0 to V2 in levodopa equivalent daily dose (LEDD) 75 , UPDRS-III-OFF (motor severity), UPDRS-IV (motor complications), FOGQ, NMSS (NMS burden), PD-CRS (cognition), BDI-II (mood), and NPI (neuropsychiatric symptoms). The model was adjusted to baseline QoL and age, gender, disease duration, comorbidity (total number of nonantiparkinsonian medications as surrogate marker 14 ), and the score of the rest of the variables at baseline (LEDD, UPDRS-III-OFF, UPDRS-IV, FOGQ, NMSS, PD-CRS, and NPI). Disability (ADLS) was not included in the model because this is consequence of symptoms, but since it is related to QoL, in a second model ADLS at baseline and change in ADLS from V0 to V2 were included. Hosmer-Lemeshow test was applied and adjusted R-squared was calculated for all analyses. Finally, multiple linear regressions were performed with "change in QoL" as dependent variable but only for variables (PDQ-39SI; PQ-10; EUROHIS-QOL8) changing significantly from V0 to V2. The independent variables included were the same as in the binary model. The p-value was considered significant when it was <0.05.

Standard protocol approvals, registrations, and patient consents
The Comité de Ética de la Investigación Clínica de Galicia from Spain (2014/ 534; 02/DEC/2014) approval was obtained. Written informed consents from all participants in this study were obtained before the start of the study. COPPADIS-2015 was classified by the AEMPS as a Post-authorization Prospective Follow-up study with the code COH-PAK-2014-01.

Reporting summary
Further information on research design is available in the Nature Research Reporting Summary linked to this article.

DATA AVAILABILITY
The data that support the findings of this study are available from the corresponding author upon reasonable request.

CODE AVAILABILITY
No computer coding was used in the completion of the current manuscript.