The Parkinson's Disease Sleep Scale–2 (PDSS‐2): Validation of the Spanish Version and Its Relationship With a Roommate‐Based Version

ABSTRACT

Background

Because of the prevalence and impact of sleep disorders in Parkinson's disease (PD), valid instruments for their evaluation and monitoring are necessary. However, some nocturnal sleep disorders may go unnoticed by patients themselves.

Objectives

To validate a pan‐Spanish version of the Parkinson's Disease Sleep Scale Version 2 (PDSS‐2) and to test the relationships between the PDSS‐2 and a PDSS‐2 roommate version.

Methods

PD patients (n = 399) from seven Spanish‐speaking countries were included. In addition to the tested PDSS‐2 scales, valid measures for sleep disorders and both motor and nonmotor manifestations were applied. Acceptability, dimensionality, reliability, precision, and construct validity were explored, as well as discrepancies and agreement between the PDSS‐2 and the roommate version.

Results

PDSS‐2 showed negligible floor and ceiling effects. Four factors (57% of the variance) were identified. Reliability parameters were satisfactory: alpha = 0.84; item homogeneity coefficient = 0.27; corrected item total correlation = 0.28 to 0.61; and test‐retest reliability (average kappa = 0.70; intraclass correlation coefficient [ICC] = 0.83). The standard error of measurement was 5.84, and correlations with other scales assessing nocturnal sleep were high (r_S = 0.62–0.56). In comparison to the patient‐based total score, the by proxy total score showed no significant difference, high correlation (r_S = 0.70), and acceptable agreement (ICC = 0.69), but there were discrepancies in two or more points in 18% of item scores.

Conclusions

The Spanish version of the PDSS‐2 has shown satisfactory clinimetric attributes. Acceptability and precision data are presented for the first time. The PDSS‐2 roommate version could be useful to complement the patient‐based evaluation, but additional studies are needed.

Sleep disorders like insomnia, excessive daytime sleepiness, rapid eye movement behavior disorder, sleep apnea, and RLS are very frequent in patients with Parkinson's disease (PD). Over the past few decades, prevalence of these disorders in PD has been estimated to be between 22% and 98%, depending on the study.1, 2, 3, 4, 5, 6 Moreover, prevalence of sleep disorders is ≥2 times higher in PD patients than in control populations, except for nocturia.7, 8, 9

These sleep disorders may be present from PD's prodromal phase,10, 11 may increase their presence and severity over time,12 and may be highly prevalent in the disease's advanced stages.13 Furthermore, sleep disorders in PD have been correlated with increased disease‐related disability14 and significant quality‐of‐life deterioration.6, 15, 16, 17, 18

As a result, validated instruments are necessary to diagnose and monitor these varied sleep disorders. One of these instruments, the Parkinson's Disease Sleep Scale version 2 (PDSS‐2),19 is a patient‐completed clinical rating scale that assesses the frequency of sleep disturbances over the past week. Each question is scored between 0 (“never”) and 4 (“very often”), and a total score is calculated by summing a patient's responses to each of the 15 questions (minimum 0 to maximum 60). Showing satisfactory psychometric properties, the scale has been validated in English (original version),19 Italian,20 and Hungarian versions.21 However, a Spanish version of the PDSS‐2 has not been validated yet.

Furthermore, because some disorders occurring during the nocturnal sleep period are not noticed by patients themselves, a PDSS‐2 version adapted for roommates was created to obtain a proxy assessment of the sleep disorders noticed by an observer.

The objectives of the present study were to: (1) determine the clinimetric properties of the PDSS‐2 Spanish Version (PDSS‐2); (2) add validation data never before explored; (3) compare the results with those of the two previous PDSS‐2 validation studies;19, 20, 21 and (4) determine the relationship between the patient‐ and an original caregiver/roommate‐based versions of the PDSS‐2.

Materials and Methods

Design

This was an international, observational, and cross‐sectional study.

Patients

According to international criteria,22 consecutive patients with diagnosis of PD by a neurologist specialized in movement disorders were included in the study. Additional inclusion criteria were: (1) to have a caregiver or roommate able to complete the PDSS‐2 proxy version and (2) the ability to provide informed consent.

Patients who did not meet any of the following criteria were excluded from the study: (1) those with any acute, severe, or concomitant condition that could interfere with the assessment of PD, including dementia of any degree but not mild cognitive impairment; or (2) patients with PD treatment modifications over the past 2 weeks (e.g., levodopa, dopamine agonists, and IMAO_B) or those in treatments that can affect sleep (e.g., benzodiazepines) or psychiatric condition (e.g., fluoxetine).

The sample was recruited from eleven centers in seven countries: Argentina, Colombia, Cuba, Ecuador, Mexico, Spain, and the United States (Florida). An a priori 20:1 patient‐to‐item ratio was considered a satisfactory sample size to conduct the analysis even under exigent conditions for factor and other validity analyses.23, 24 In addition, a minimal recruitment of 30 patients per center was established, but a greater recruitment was recommended for countries in which only one center was participating.

Ethical Issues

The study was accepted by the Ethics Committee for Research at the Carlos III Institute of Health and by the corresponding local committees in each participant site. Patients and caregivers signed the informed consent before participating in the study.

Assessments

Information regarding sex, age, education, and years of diagnosis and treatment were collected.

The MAPI Research Trust (https://eprovide.mapi-trust.org), the entity that manages the copyright of the PDSS‐2, kindly provided a Spanish cross‐cultural adaptation of the PDSS‐2 for the present study. Researchers from the participant centers reviewed the scale to check whether it was understandable by patients in their respective countries, and a final “Spanish version” was agreed upon after corrections (an expression and three words affecting four items). This was the version used for the present validation study. The research team also prepared a “PDSS‐2 by proxy” version that referred to the patient, but was answered by the caregiver/roommate.

In addition to the PDSS‐2, the following instruments were used to assess each patient: H & Y staging;25 Scales for Outcomes in Parkinson's Disease–Motor (SCOPA‐Motor);26 Scales for Outcomes in Parkinson's Disease–Sleep (SCOPA‐Sleep),27 which includes a question for global evaluation of the quality of nocturnal sleep (see Supplementary Material S1) and is recommended by the Movement Disorder Society Task Force;28 Non‐Motor Symptoms Scale (NMSS);29 Clinical Impression of Severity Index for PD (CISI‐PD);30 Visual Analogue Scales for pain severity and frequency (VAS Pain);31 and Hospital Anxiety and Depression Scale (HADS),32 which has been recommended for use in PD.33 Details about these assessments are provided into the Supplementary Material of this article.

l‐dopa equivalent daily dose (LEDD) was calculated according to Tomlinson et al.34

Procedure

The field work of the study was carried out from July 2017 to June 2018. For patients who had any problem impeding direct responses to the questionnaires, the help of a person other than the doctor and the caregiver (e.g., fellow, nurse) was allowed. A second application of the PDSS‐2 (retest) in the strict range of 7 to 14 days after the first evaluation was offered to all included patients except to those who needed help for completing the assessments the first time. The completed retest scale could be sent out to the respective researchers through post‐mail, fax, e‐mail, phone application, etc.

Statistical Analysis

Anonymized data were sent to be analyzed at the National Center of Epidemiology at the Carlos III Institute of Health in Madrid, Spain. Nonparametric tests were deemed appropriate for this analysis because of the ordinal nature of the PDSS‐2 data and their non‐normal distribution (Shapiro‐Francia test).

In addition to descriptive statistics regarding demographic and historical data, the following analyses were carried out:

1. Statistics regarding data quality and acceptability—namely, percentage of missing data, mean, median, standard deviation (SD), skewness, range, and floor and ceiling effects—were determined for each of the PDSS‐2 items as well as for its total score. Concerning data quality, satisfactory data sets should contain a maximum of 5% of missing data.35 For acceptability, a maximum floor or ceiling effect of 15%36 and skewness values between –1 and + 1 are suggested.37

2. To evaluate scaling assumptions, roughly equivalent item SDs and corrected item‐total correlations ≥0.30, representing item convergent validity, are recommended.38

3. Dimensionality was investigated by means of exploratory factor analyses (principal component factor technique; orthogonal and oblique promax rotations). Sample adequacy was also determined using Bartlett's sphericity test (acceptable for P < 0.05) and the Kaiser‐Meyer‐Olkin statistic (adequate if >0.60).

4. For internal consistency, Cronbach's alpha >0.7039 and an item homogeneity coefficient ≥0.1540 were considered satisfactory, in addition to the aforementioned corrected item‐total correlation.

5. Test‐retest reliability was explored using kappa index with quadratic weights for items and intraclass correlation coefficient (ICC; one‐way, random effect) for total scores.

6. Precision of PDSS‐2, as measured with the standard error of measurement (SEM), was calculated as $\mathit{SEM}={\mathit{SD}}_p*\sqrt{(1-r_{\mathit{xx}}})$, where SD_p represents the pooled SD of the two applications (${\mathit{SD}}_p=\sqrt{\frac{{\mathit{SD}}_1^2+{\mathit{SD}}_2^2}{2}}$) and r_xx is the ICC of the test‐retest analysis.41, 42, 43 A SEM value <1/2 SD_p was considered acceptable.

7. Convergent validity between the PDSS‐2 total score and the other assessments of sleep disorders utilized in this study was determined using Spearman rank‐correlation coefficients, which were deemed moderate if they ranged between 0.30 and 0.59 or high if they were ≥ 0.60.44 High correlation coefficients were hypothesized between the PDSS‐2, SCOPA‐Sleep, and the NMSS Sleep/fatigue domain.

8. Discriminant validity between the groups formed by sex, age, PD duration, H & Y, and nocturnal sleep quality from the SCOPA‐Sleep was analyzed using the Mann‐Whitney U or Kruskal‐Wallis tests, which were evaluated using the magnitude of the difference and P values.45

9. Comparison between PDSS‐2 scores of patients and caregivers/roommates was carried out with the Wilcoxon test, and the association between both scores was elicited using Spearman's correlation coefficient. Agreement of item and total scores was tested by means of the weighted kappa and ICC, respectively.

Using IBM SPSS (Version 24; IBM Corp., Armonk, NY) and Stata software (version 15; StataCorp LP, College Station, TX), the data analysis was conducted.

Results

The main characteristics of our sample of 399 patients, which was were 56.89% males, and the sample's distributions on the applied assessments are shown in the Table 1. Surgery for PD was undergone by 4.26% of the sample: 13 had some form of DBS, 1 unilateral palidotomy, and 3 unilateral subthalamotomy. All H & Y stages were present, and the median H & Y stage was 2 with an interquartile range of 1 to 3.

Table 1.

Sample characteristics

	Mean	Median	SD	Range
Age, years	67.81	69	10.82	35 to 92
Years of education	11.70	12	5.51	0 to 35a
Age at onset of PD	58.21	59	11.43	20 to 91
Age at diagnosis	59.59	60	11.14	24 to 92
Age at starting treatment	59.85	61	11.07	28 to 92
Disease duration	9.60	9	6.06	0 to 35
Treatment duration	7.97	7	5.78	0 to 32
LEDD	858.76	800	431.39	0 to 2,250
SCOPA‐Motor Total score	23.70	22	13.13	1 to 61
Non‐Motor Symptoms Scale Total score	52.01	38	44.45	0 to 222
Clinical Impression of Severity Index for PD	8.73	8	4.89	0 to 24
Visual Analogue Scale for Pain	15.56	4	21.97	0 to 100
HADS‐Depression	6.83	6	4.72	0 to 19
HADS‐Anxiety	6.31	6	4.13	0 to 19
SCOPA‐Sleep Nocturnal sleep	4.91	4	3.63	0 to 15
SCOPA‐Sleep Daytime sleepiness	4.30	3	3.85	0 to 18

^a Three patients did not have formal education, but completed programs for late literacy. On the other extreme, 2 patients were university professors with extended continuous education.

The individual assessment of each item of the PDSS‐2 is summarized in Table 2. There were 19 item scores missing (0.32% of the 5,985 obtainable), most of them (18 of 19) from item 7 (“Distressing hallucinations at night”). Imputation for missing values was not performed; therefore, there were 380 (95.2%) full computable total scores.

Table 2.

Acceptability data of the PDSS‐2

Item	Mean	SD	Median	Floor Effecta	Ceiling Effecta	Skewness
1. Sleep during the last week	1.32	1.17	1	29.07	6.27	0.67
2. Difficulties falling asleep	0.97	1.13	1	45.36	4.76	1.08
3. Difficulties staying asleep	1.22	1.27	1	38.10	8.52	0.81
4. Restlessness of legs or arms at night	0.81	1.09	0	54.64	3.76	1.31
5. Urge to move your legs or arms	0.76	1.03	0	54.89	2.26	1.33
6. Distressing dreams at night	0.90	1.11	0	50.88	3.51	1.07
7. Distressing hallucinations at night	0.65	1.14	0	68.50	5.25	1.76
8. Get up at night to pass urine	1.92	1.36	2	18.30	18.55	0.14
9. Uncomfortable because of immobility	1.35	1.34	1	38.35	10.03	0.60
10. Pain in arms or legs	0.93	1.08	1	48.62	2.76	0.91
11. Muscle cramps in your arms or legs	0.84	1.07	0	51.88	2.76	1.15
12. Painful posturing in the morning	0.71	1.06	0	60.90	2.01	1.36
13. Tremor on waking	0.72	1.07	0	59.90	3.26	1.50
14. Tired/sleepy in the morning	1.28	1.23	1	34.42	6.78	0.67
15. Snoring or difficulties in breathing	0.76	1.14	0	61.90	2.01	1.28
PDSS‐2 Total Score	15.02	9.84	13	0.26	0.26	1.01

^a Expressed in percentage (%).

The entire possible range of scores (0–4) was observed on each item, but the total score had an empirical maximum of 51 whereas the theoretical maximum was 60. Moreover, significant floor effects were observed for every item of the PDSS‐2, whereas only item 8 (“Get up at night to pass urine”) showed a slight ceiling effect. Nonetheless, the PDSS‐2 total score showed negligible floor and ceiling effects. All skewness values were inside or very near the standard range and item SDs ranged from 1.03 to 1.27 (Table 2). The corrected item‐total correlations were above the criterion, except for item 15 (“Snoring or difficulties in breathing”), which was slightly below (r_S = 0.28; Table 3).

Table 3.

Corrected item‐total correlations, exploratory factor analysis, and test‐retest reliability

		Factor Analysis*
		Factors
Item	Item‐Total Correlation	1	2	3	4	Weighted Kappa
1. Sleep during the last week	0.46		0.86			0.63
2. Difficulties falling asleep	0.57		0.48			0.62
3. Difficulties staying asleep	0.50		0.89			0.67
4. Restlessness of legs or arms at night	0.57	0.77				0.69
5. Urge to move your legs or arms	0.61	0.81				0.67
6. Distressing dreams at night	0.46			0.40		0.69
7. Distressing hallucinations at night	0.32			0.81		0.74
8. Get up at night to pass urine	0.39		0.52			0.82
9. Uncomfortable because of immobility	0.58		0.31			0.80
10. Pain in arms or legs	0.54	0.80				0.69
11. Muscle cramps in your arms or legs	0.49	0.67				0.69
12. Painful posturing in the morning	0.57	0.76				0.74
13. Tremor on waking	0.34				0.80	0.66
14. Tired/sleepy in the morning	0.40				0.71	0.72
15. Snoring or difficulties in breathing	0.28			0.79		0.64

^* Principal component factor, promax rotation. Only loads >0.30 are displayed.

Through exploratory factor analysis, it was determined that the PDSS‐2 contained four relevant factors explaining 57% of the variance (Kaiser‐Meyer‐Olkin = 0.85; Bartlett test of sphericity, P < 0.001): Factor 1 included RLS and nocturnal pain (items 4, 5, 10, 11, and 12); factor 2, insomnia (items 1, 2, 3, 8, and 9); factor 3, dreams, hallucinations, and sleep apnea (items 6, 7, and 15); and factor 4 experiences on waking (items 13 and 14; Table 3). Factor 1 was strongly correlated with factor 2 (r_S = 0.52) and moderately correlated with factors 3 (0.36) and 4 (0.35); however, the other factors only showed weak correlations among themselves.

Interitem correlation ranged from 0.05 (items 8 and 15) to 0.65 (items 4 and 5). Item homogeneity coefficient for the PDSS‐2 was 0.27, and Cronbach's alpha was 0.84. The test‐retest (n = 302) showed high percentages of agreement (93.38–97.31%), strong weighted kappa values (0.62–0.82; average, 0.70) for the items (Table 3), and high ICC (0.83) for the overall score. The SEM was 5.84 (0.5 SD_p = 7.12).

A strong correlation was found between the PDSS‐2 and the SCOPA‐Sleep NS (r_S = 0.62), and moderate‐to‐high correlations were shown with the NMSS domain 2 (sleep/fatigue, 0.56), NMSS total score without sleep/fatigue (0.52), and HADS‐Anxiety (0.54). Correlations with other measures and variables were low or moderate (Table 4).

Table 4.

Convergent validity and correlates of the PDSS‐2

	Spearman's Rho
SCOPA‐Sleep Nocturnal sleep	0.62
SCOPA‐Sleep Daytime sleepiness	0.44
H & Y staging	0.30
SCOPA‐Motor Examination	0.24
SCOPA‐Motor Activities of daily living	0.37
SCOPA‐Motor Complications	0.34
SCOPA‐Motor Total score	0.34
Non‐Motor Symptoms scale–Sleep/Fatigue	0.56
Non‐Motor Symptoms scale Total scorea	0.52
Clinical Impression of Severity Index	0.39
Visual Analog Scales for pain	0.46
Hospital Anxiety and Depression Scale‐Anxiety	0.54
Hospital Anxiety and Depression Scale‐Depression	0.41
Ageb	0.01
PD durationc	0.16
Duration of treatment for PD	0.18
LEDD	0.22

^a Without the domain Sleep/Fatigue.

All coefficients, P < 0.001, except (^b), not significant, and (^c), P < 0.01.

PDSS‐2 total score showed no statistical differences between sex or age group (<60, 60–74, and ≥75 years), but it significantly increased with H & Y stage (Kruskal‐Wallis test, P = 0.0001) and quality of nocturnal sleep according to the specific SCOPA‐Sleep question (Mann‐Whitney U test, P = 0.0001).

The PDSS‐2 proxy version was completed by 329 caregivers. When comparing patient and caregiver PDSS‐2 assessments, it was evident that almost half of the item scores were not coincident, with differences ranging from 1 point (in 26.9% of the cases on average) to 4 points (1.7% on average; Table 5). Patient‐caregiver differences were statistically significant for seven items, especially items 6, 9, and 10 (P < 0.0001). However, patient‐caregiver differences for the total score (15.32 ± 10.17 vs. 16.33 ± 10.66) were not significant (P = 0.44), and a strong correlation between patient and caregiver total scores was found (r_S = 0.70). Concerning agreement, kappa values were 0.29 to 0.61 for items (Table 5), and the ICC for the total scores was 0.69.

Table 5.

Differences and agreement between patients and caregivers on the PDSS‐2

	Difference in Points*
	(% of Patients)
	0	1	2	3	4	P Value**	Weighted Kappa
1. Sleep during the last week	54.3	29.9	9.5	4.9	1.4	0.41	0.52
2. Difficulties falling asleep	53.2	31.0	10.1	3.7	2.0	0.79	0.48
3. Difficulties staying asleep	51.7	30.5	12.6	3.5	1.7	0.31	0.57
4. Restlessness of legs or arms at night	53.7	28.2	11.5	4.6	2.0	0.39	0.41
5. Urge to move your legs or arms	57.5	28.2	10.4	2.8	1.1	0.12	0.49
6. Distressing dreams at night	44.0	32.5	15.8	6.3	1.4	<0.0001	0.37
7. Distressing hallucinations at night	66.7	18.2	10.6	2.7	1.8	0.95	0.56
8. Get up at night to pass urine	52.9	27.9	13.2	4.9	1.1	0.03	0.61
9. Uncomfortable because of immobility	55.2	25.9	12.6	4.3	2.0	<0.0001	0.58
10. Pain in arms or legs	57.5	20.4	17.0	3.7	1.4	<0.0001	0.42
11. Muscle cramps in your arms or legs	57.5	31.0	8.6	2.3	0.6	0.01	0.58
12. Painful posturing in the morning	59.2	27.3	10.1	2.3	1.1	0.45	0.52
13. Tremor on waking	58.9	22.7	12.6	3.7	2.0	0.01	0.45
14. Tired/sleepy in the morning	53.6	26.5	13.3	5.5	1.2	0.003	0.52
15. Snoring or difficulties in breathing	51.2	23.9	16.9	5.5	2.5	0.16	0.29
Average %	55.1	26.9	12.3	4.0	1.7	—	—

^* Difference in points represents the absolute value of the difference between the PDSS‐2 self‐completed version and the proxy version (no difference equals = 0 points, while the maximum difference is = 4 points).

Discussion

The present study seeks to validate the Spanish version of the PDSS‐2 while confirming the results of earlier studies19, 20, 21 and exploring some clinimetric properties not previously tested. In addition, the scores of a PDSS‐2 roommate‐based version were compared to those of the patient‐based PDSS‐2.

PDSS‐2 data quality was considered acceptable because missing items represented <5% of the collected data, which was mostly attributed to item 7 (“Distressing hallucinations at night”) because it was responsible for 18 of the 19 missing item scores. Not mentioned19, 20 or not found21 in previous studies, this finding may be the result of a lack of question clarity or of concealment attributed to shame, stigma, or another reason.

Concerning acceptability, there were high floor effects for all items, which are related to the apparent absence of symptoms in the corresponding proportion of patients. Although it would not be realistic to expect that most symptoms would be ubiquitous throughout the sample, other factors might be eliciting a negative response (e.g., unawareness or shame). Importantly, ceiling effect was only found in item 8 (nocturia), and it was mild in magnitude. PDSS‐2 total score showed no floor or ceiling effects and acceptable skewness. Kovács et al.21 also found high floor effects in two of the three subscales embedded in the PDSS‐2, no floor effect for the total score, negligible ceiling effect, and satisfactory skewness for all these scores. Therefore, it was concluded that the PDSS‐2 possesses adequate acceptability (Table 2). Parameters for testing scale assumptions were also acceptable, indicating that item scores can be directly summed to obtain the total score.

Factor analysis revealed that four factors likely exist within the PDSS‐2: namely, RLS and nocturnal pain; insomnia; dreams, hallucinations and sleep apnea; and experiences on waking (Table 3). However, previous studies have recognized one to five factors in this rating scale.19, 20, 21 The original and the Hungarian validation studies found coincident solutions (one or three factors) and built three subscales based on their results: motor problems at night, PD symptoms at night, and disturbed sleep.19, 20, 21 Given the variability in findings, we do not propose any grouping of items in subscales at present. Confirmatory factor analysis could help determine the factorial structure of the PDSS‐2 in the future.

Internal consistency was acceptable, with a Cronbach's α coefficient value of 0.84 that is well above the standard value and slightly higher or almost coincident with other studies’ values (0.73–0.86).19, 20, 21 All the corrected item‐total correlations were above the standard value 0.30, except for item 15 (“Snoring or difficulties breathing”; Table 3). In the previous studies, item‐total correlation values were mildly lower than in the present one, but item 15 was also the lowest in the study by Arnaldi et al.19, 20

Test‐retest reliability of PDSS‐2 was optimal considering agreement values, as was found previously (Table 3).19, 20, 46 Suggesting that the PDSS‐2 score has adequate reliability and sensitivity (precision) in repeated measurements, the SEM value (5.84), lower than the criterion of <0.5 pooled SD, was deemed satisfactory.41, 47 Calling for future verification of these values, this SEM value was higher than the minimal clinically important difference (MCID) found by Horváth et al. (2.07 for worsening; –3.44 for improvement).48 However, the MCID value is susceptible to variation in different studies, whereas the SEM remains almost constant across samples and should be smaller than the MCID.49, 50

Demonstrating a close relationship with the SCOPA‐Sleep NS, the PDSS‐2 showed strong correlations with this scale, as noted in a previous study,51 whereas the association with a less‐specific domain (sleep/fatigue of the NMSS), issues of daytime sleep (SCOPA‐Sleep DS), and other aspects of PD were looser (Table 4). Notably, the anxiety score (HADS‐Anxiety) reached a moderate‐to‐high association with the PDSS‐2 in our study. An anxiety measure was not included in the previous studies, and therefore we cannot compare this result; however, the correlation with depression in the present study was moderate and almost coincident with the study by Arnaldi et al., although the depression scales applied were different.20 In the previous studies and the present one, the associations with measures of motor disorder and global severity of PD varied from weak to strong.19, 20, 21

According to the results, the PDSS‐2 total score was closely associated with nonmotor symptoms and moderately associated with pain.21 Calling for intervention, pain and anxiety are known determinants of sleep quality in PD52 and may synergize with sleep problems to impair quality of life of PD patients.53

As shown previously, PDSS‐2 total score significantly increased as H & Y stage did.19, 21 Its total score was also clearly different between the groups of good and bad sleepers according to the ad‐hoc item of the SCOPA‐Sleep. From the convergent and discriminant validity analyses, it can be concluded that the PDSS‐2 possesses satisfactory construct validity.

A synopsis of the four validation studies performed up until now appears in Table 6.

Table 6.

Data from the validation studies of the PDSS‐2

	Trenkwalder et al., 2011	Arnaldi et al., 2016	Kovács et al., 2016	Present Study, 2018
Sample size (n)	113	123	537	399
Range of item scores covered	All, 0 to 4	—	—	All, 0 to 4
Data quality–missing values	—	—	0%	4.8%
Floor effect–total score	—	—	1.70	0.26
Ceiling effect–total score	—	—	0.20	0.26
Skewness–total score	—	—	0.57	1.01
Cronbach's alpha	0.73	0.77	0.86	0.84
Interitem correlation	—	—	—	0.05 to 0.65
Item homogeneity index	—	—	0.49	0.27
Item‐total correlation	0.19 to 0.59	0.07 to 0.63	—a	0.28 to 0.61
Factor analysis–factors (% variance)	1 (22.8), or 3 (42.7)	5 (59.4%)	1 (28.9), or 3 (38.9)	4 (57.0)
Test‐retest kappaw values (or ICC)	0.44 to 0.76	—	—	0.62 to 0.82
Test‐retest ICC	0.80	0.94	—	0.83
Precision (SEM value)	3.98	2.14	2.46	5.84
Convergent validity (with measures for nocturnal sleep)	>0.50	—	>0.60	>0.55
Internal validity	—	—	0.52 to 0.69	—
Discriminative validity	Yes	—	Yes	Yes
Bed partners/roommates	Interview	—		PDSS‐2 by proxy

^a Available for the three embedded PDSS‐2 domains identified by factor analysis, with the following ranges: Motor symptoms = 0.39 to 0.68; PD symptoms = 0.34 to 0.65; Disturbed sleep = 0.36 to 0.53.

Indicating that proxy scores should be taken with caution when assessing subjective manifestations, the concordance between ratings of patients and caregivers was only moderate (Table 5).54, 55 This is particularly evident in the differences between patients and caregivers in the scores of some items, like in item 6 (“Distressing dreams at night”) for which only 44% of the scores agreed, or in the difference of ≥2 points in 18% of the paired data. Details about the direction of the discrepancies are shown in Supplementary Table S1 (Supplementary Material S1). The usefulness of the PDSS‐2 proxy version needs to be established in future studies.

Despite some limitations of the study, such as a sample with most patients in mild and moderate severity stages and the diversity of raters in multicenter studies, we can conclude that the Spanish version of the PDSS‐2 has shown satisfactory feasibility/acceptability, reliability, construct validity, and precision. It can be useful in research and clinical settings to assess overall nocturnal sleep quality, identify potential causes of sleep problems in PD, such as pain, and set priorities for intervention. The proxy evaluation of nocturnal sleep using the PDSS‐2 version adapted to the caregiver/roommate could be a valuable aid to complete the patient‐based assessment, but additional studies are needed to definitively determine its validity.

Author Roles

(1) Research Project: A. Conception, B. Organization, C. Execution; (2) Statistical Analysis: A. Design, B. Execution, C. Review and Critique; (3) Manuscript Preparation: A. Writing of the First Draft, B. Review and Critique.

P.M‐M.: 1A, 1B, 2A, 2B, 2C, 3A

J.B.W.: 2B, 3A

C.R.‐B.: 1B, 2B, 3B

T.A.: 1C, 3B

O.B.: 1C, 3B

V.C.‐A.: 1C, 3B

C.C.: 1C, 3B

I.E.‐B.: 1C, 3B

N.G.: 1C, 3B

L.G.: 1C, 3B

J.U.M.‐M.: 1C, 3B

J.C.M.‐C.: 1C, 3B

I.P.: 1C, 3B

M.S.‐D.: 1C, 3B

C.S.: 1C, 3B

M.R.‐V.: 1C, 3B

F.V.: 1C, 3B

Disclosures

Ethical Compliance Statement: The study was accepted by the Ethics Committee for Research at the Carlos III Institute of Health, Madrid, Spain, and by the corresponding local Committees in each participant site. Patients and caregivers signed the informed consent before participating in the study. We confirm that we have read the Journal's position on issues involved in ethical publication and affirm that this work is consistent with those guidelines.

Funding Sources and Conflicts of Interest: The authors report no sources of funding and no conflicts of interest.

Financial Disclosures for previous 12 months: P.M‐M: Honoraria for lecturing in courses from National School of Public Health (ISCIII) and Editorial Viguera; International Parkinson and Movement Disorder Society for management of the Program on Rating Scales; Air Liquide, Abbvie, Zambon, and HM Hospitales de Madrid for Advisory Board. License fee payments for the King's Parkinson's Disease Pain scale. C.R.‐B.: Support of Abbvie for attending international meetings. O.B.: Advisory Board for Medtronic. I.E.‐B.: Honoraria from UCB, Medtronic, and Teva. J.C.M.‐C.: Research support from Allergan, Zambon, Ipsen, Merz. Honoraria from Britannia, Bial, Zambon for Advisory Board and for speaking from Italfarmaco, UCB, Zambon, Allergan and Abbvie. J.U.M.‐M.: Honoraria from Zambon, UCB, Bial. J.B.W.: Fulbright Grant (stay in Madrid, Spain) C.S.: Honoraria from Mitsubishi Pharma; Advisory Board of Neurocrine, Revance, and Adamas. M.R.‐V.: Honoraria from UCB, Allergan, General Electric, Medtronic, and Abbvie. F.V.: Honoraria from Abbvie, Allergan, UCB, Bial, Ipsen, and Zambon T.A., V.C.‐A., C.C., N.G., I.P., L.G., M.S.‐D.: No disclosures.

Supporting information

Appendix S1: Supplementary Information