The UPDRS-8: A Brief Clinical Assessment Scale for Parkinson's Disease

ABSTRACT

We evaluated a brief, 8-item version of the Unified Parkinson's Disease Rating Scale (UPDRS) using two existing patient databases. One database included 1,445 PD patients. Spearman correlation between UPDRS-8 motor scores and full UPDRS motor scores was .765 (p < .001). Correlation between total UPDRS-8 scores and full UPDRS total scores (parts I–III) was .798 (p < .001). Correlation between total UPDRS-8 scores and total 39-item PD questionnaire (PDQ-39) scores was .629 (p < .001). In 177 patients undergoing deep brain stimulation (DBS), UPDRS-8 motor scores were similarly significantly sensitive to change as full UPDRS motor scores in assessing change from the medication OFF state to the medication ON state at baseline and from the medication OFF state at baseline to the medication OFF/stimulation ON state 1 year post-DBS. The UPDRS-8 focuses on items that are most relevant for clinical decision making. In this study, the UPDRS-8 exhibited good correlation with the full UPDRS and the PDQ-39. We therefore believe that it can provide a useful, rapid assessment of PD patients in clinical practice. Whether it might be useful in clinical trials depends on demonstrating that it is also sensitive to relatively small changes in clinical status.

KEYWORDS: :

• assessment
• clinical trial
• Parkinson's disease
• scale
• treatment
• Unified Parkinson's Disease Rating Scale (UPDRS)
• UPDRS-8

INTRODUCTION

The Unified Parkinson's Disease Rating Scale (UPDRS) [1] has been one of the principal tools used to evaluate Parkinson's disease (PD) severity in clinical trials for many years, and is still commonly used today. It has not been widely used in clinical practice, largely because it takes approximately 17 minutes to complete [2]. The recently developed Movement Disorder Society UPDRS (MDS-UPDRS) was designed to be more comprehensive than the original UPDRS, adds additional nonmotor elements, and was designed to take approximately 30 minutes to complete [3].

In some situations, a very brief PD rating scale that can be completed quickly may be desirable. For example, in a pharmacokinetic–pharmacodynamic study that requires many ratings over a short period of time, a brief but sensitive motor assessment may be useful. In clinical practice, a brief rating scale that captures the most critical elements on which management decisions are made may be more practical than the full UPDRS or the MDS-UPDRS.

Several prior studies have found that many items of the original UPDRS are highly redundant [2, 4]. We therefore investigated whether a few selected items from the original UPDRS could constitute a very brief, valid, and sensitive PD rating scale that could be rapidly completed.

METHODS

We selected items from the original UPDRS to be evaluated as a rapid assessment scale. The items were selected a priori based on our understanding of key features of PD, clinical experience, and clinical trial participation, with emphasis on items we believe to be most relevant for clinical decision making. The two main objectives were to have a very brief motor assessment that could potentially be repeated many times at relatively frequent intervals and to have a brief “total” assessment that would include these motor items and add key historical items that would provide a broader reflection of patient status.

The original UPDRS includes 42 items. Thirty-five of these are rated on a 5-point (0–4) scale, including 4 items in Section I (mentation, mood, and behavior), 13 items in Section II (activities of daily living [ADL]), 14 items in Section III (motor examination), and 4 items in Section IV (complications of therapy). Section IV also contains 7 binary yes/no items.

By consensus, we selected 8 items from the original UPDRS to comprise the UPDRS-8. All are rated on a 5-point (0–4) scale. The items selected are delineated below.

Nonmotor:

1. Cognition (UPDRS item 1)

2. Mood (UPDRS item 3)

Motor:

1. Rest tremor upper extremity—right, left (UPDRS item 20)

2. Finger taps—right, left (UPDRS item 23)

3. Gait (UPDRS item 29)

Motor complications:

1. Off time (UPDRS item 39)

2. Dyskinesia duration (UPDRS item 32)

3. Dyskinesia disability (UPDRS item 33)

This scale was evaluated using the University of Kansas Medical Center (KUMC) PD patient database. Data from all new patient evaluations conducted by one of the authors (RP) at the KUMC Parkinson's Disease Center from August 7, 2002, through September 1, 2010, who had a complete UPDRS rating, a 39-item PD questionnaire (PDQ-39) [5] assessment, and a PD diagnosis with a clinical certainty of 90%–100% were included. Three analyses were conducted. For the first analysis, Spearman correlations were used to evaluate (a) the correlation between the UPDRS-8 motor subscale (items 3–5) and the full UPDRS motor subscale, (b) the correlation between the UPDRS-8 (nonmotor plus motor subscales, items 1–5) and the total full UPDRS (parts I–III), and (c) the correlation between the total UPDRS-8 (items 1–8) and the total full UPDRS (parts I–III). We then repeated these analyses for patient subgroups based on Hoehn–Yahr score at the time of the assessment (1, 1.5 or 2, 2.5 or 3, and ≥3.5) to evaluate correlations across a range of PD severity. For the second analysis, we evaluated correlations between the UPDRS-8 (motor [items 3–5], nonmotor + motor [items 1–5], total [nonmotor + motor + complications, items 1–8]) and PDQ-39 scores, and for comparison, evaluated correlations between the full UPDRS (parts I–III, total [parts I + II + III]) and PDQ-39 scores. For the third analysis, we evaluated the UPDRS-8 in patients who had undergone bilateral subthalamic nucleus (STN) deep brain stimulation (DBS) surgery, and for whom there were full UPDRS data available at baseline and 1 year postsurgery. In this group, using baseline data, we evaluated correlations between UPDRS-8 motor scores (ON and OFF) and full UPDRS motor scores (ON and OFF). We then evaluated the ability of the UPDRS-8 motor subscale to detect change from the medication OFF to the medication ON state at baseline and from baseline (medication OFF state) to 1 year postsurgery (medication OFF/stimulation ON). We also evaluated these same changes using the full UPDRS motor subscale. Changes from OFF to ON and from baseline to 1 year were analyzed using Wilcoxon signed rank comparisons for nonparametric related samples.

RESULTS

Analysis 1: Correlations Between the UPDRS-8 and the Full UPDRS

The KUMC new patient database included 1,445 unique PD patients meeting eligibility criteria. Mean age (SD) was 67.3 (11.0) (range = 24.5–94.2) years, and 914 (63.3%) were male. Mean disease duration was 6.2 (5.9) years, and mean full UPDRS scores were as follows: part I (mentation/mood/behavior) = 2.6 (2.0), part II (ADL) = 12.9 (6.8), part III (motor) = 25.6 (9.5), and total = 41.2 (16.2).

Spearman correlations between UPDRS-8 scores and full UPDRS scores are presented in Table 1. For the whole dataset, the correlation between UPDRS-8 motor scores and full UPDRS motor scores was .765 (p < .001). The correlation between UPDRS-8 nonmotor + motor scores and total full UPDRS scores (parts I–III) was .765 (p < .001), and the correlation between total UPDRS-8 and total full UPDRS scores (parts I–III) was .798. Correlations calculated according to Hoehn–Yahr stage ranged from .501 to .747 and were all highly significant. Correlations between UPDRS-8 motor scores and full UPDRS motor scores, and between UPDRS-8 nonmotor + motor scores and total full UPDRS scores increased with increasing Hoehn–Yahr stage.

TABLE 1.  Spearman correlations between UPDRS-8 and full UPDRS scores

Population	UPDRS-8 motor scores (items 3–5) vs. full UPDRS motor scores (part III)	UPDRS-8 nonmotor + motor scores (items 1–5) vs. full UPDRS scores (parts I–III)	UPDRS-8 total scores (items 1–8) vs. full UPDRS scores (parts I–III)
Total, n = 1445	.765*	.765*	.798*
H-Y stage 1, n = 345	.615*	.622*	.623*
H-Y stage 1.5 or 2, n = 491	.715*	.640*	.650*
H-Y stage 2.5 or 3, n = 426	.731*	.680*	.564*
H-Y stage ≥ 3.5, n = 183	.747*	.695*	.501*

Note: H-Y, Hoehn–Yahr.

*p < .001.

TABLE 2.  Correlations between UPDRS-8 scores and PDQ-39 total scores, and full UPDRS scores and PDQ-39 total scores

	UPDRS-8			Full UPDRS
	Motor	Nonmotor + motor	Total (nonmotor + motor + complications)	Mood, mentation, behavior	ADL	Motor	Total
Correlation with PDQ-39 total score	.352*	.530*	.629*	.656*	.740*	.549*	.713*

*p < .001.

Analysis 2: Correlations Between the UPDRS-8, the Full UPDRS, and the PDQ-39

For the total population described above (n = 1,445), mean PDQ-39 total score was 30.4 (19.0). The correlations between UPDRS-8 (motor, nonmotor + motor, total [nonmotor + motor + complications] scores, full UPDRS (parts I–III and total [I + II + III]) scores, and total PDQ-39 scores are presented in Table 2. The correlation between total UPDRS-8 scores and total PDQ-39 scores was .629 (p < .001) and the correlation between total full UPDRS scores and total PDQ-39 scores was .713 (p < .001).

Analysis 3: UPDRS-8 Ability to Detect Change

There were 177 patients who had undergone STN DBS surgery and who had full UPDRS scoring at baseline and 1 year postsurgery. Mean age was 60.8 (9.5) years, 121 were male (68.4%), and disease duration was 12.1 (4.8) years. At baseline, mean full UPDRS motor OFF score was 41.6 (8.7) and ON score was 23.1 (8.2). Mean UPDRS-8 motor OFF score was 7.3 (2.4) and ON score was 3.4 (1.8).

For this population at baseline, correlation between UPDRS-8 motor OFF scores and full UPDRS motor OFF scores was .758 (p < .001). The correlation between UPDRS-8 motor ON scores and full UPDRS motor ON scores was .722 (p < .001). There was a significant difference between ON and OFF scores at baseline for both the UPDRS-8 motor scale and the full UPDRS motor scale (p < .001).

Mean full UPDRS medication OFF baseline motor scores improved from 41.6 (8.7) to 24.8 (9.6) in the medication OFF/stimulation ON state 1 year post-DBS (40.3%, p < .001). Mean UPDRS-8 medication OFF baseline motor scores improved from 7.3 (2.4) to 4.1 (1.9) in the medication OFF/stimulation ON state 1 year post-DBS (43.8%, p < .001).

DISCUSSION

There was good correlation between UPDRS-8 and full UPDRS scores across a wide range of PD severities. In addition, the UPDRS-8 exhibited good correlation with quality of life scores as assessed by the PDQ-39. We also found that UPDRS-8 motor scores were sensitive to distinguish baseline ON and OFF status in patients undergoing DBS, and sensitive to detect improvement from the baseline medication OFF state to the 1 year post-DBS medication OFF/stimulation ON state.

The items selected for inclusion in the UPDRS-8 reflect many of the key features of PD on which clinical management decisions are based, including bradykinesia, tremor, gait, OFF time, dyskinesia, mood, and cognition. Our results suggest that the UPDRS-8 may be a useful tool to evaluate and monitor patients in clinical practice. The full UPDRS is generally not used in routine clinical practice because of the time it takes to complete. The UPDRS-8 should require substantially less time, can help evaluate many of a patient's key PD features, and correlates with quality of life. We estimate that the UPDRS-8 should take 2–3 minutes to complete for most patients.

Our results also suggest that the UPDRS-8 shows potential as a rating scale for clinical trials, particularly if many evaluations are required in a short period of time, such as in a pharmacokinetic-pharmacodynamic study. However, before we can make such a recommendation, we will need to evaluate the UPDRS-8 sensitivity to change with interventions of mild efficacy, and demonstrate that it can detect changes at or close to the minimal clinically important change [6].

The UPDRS-8 has important limitations in that it can only directly evaluate aspects of PD that are assessed by its individual items. It will not detect some PD features or medication effects that may be clinically relevant, such as sleepiness, fatigue, constipation, nausea, or impulse control disorders. These symptoms are assessed by the MDS-UPDRS, and some are assessed by the original UPDRS. This is the trade-off required to substantially reduce the time burden of the PD assessment. Different versions of the UPDRS may be more appropriate for different purposes.

We also note that although there was good correlation between the total UPDRS-8 and the PDQ-39 (.629), the correlation between the total full UPDRS and the PDQ-39 was somewhat stronger (.713). Moreover, the strongest correlation with the PDQ-39 was with the ADL subsection of the original UPDRS (.740) and this was stronger than the correlation between the PDQ-39 and the total UPDRS-8 (.629), the nonmotor + motor sections of the UPDRS-8 (.530), and the motor UPDRS-8 (.352). However, this may be due to the fact that the PDQ-39 contains a large number of items that assess ADL, including items that directly overlap with the full UPDRS such as dressing, washing, cutting food, and writing. A less biased evaluation may therefore require assessing correlations between these scales and quality of life measures using more general quality of life instruments such as the EuroQoL (EQ-5D) or its visual analog scale [7].

We also note that for Hoehn–Yahr stages 2.5 or 3 and ≥3.5, we observed stronger correlations between UPDRS-8 nonmotor + motor scores and total full UPDRS scores than between total UPDRS-8 and total full UPDRS scores. We suspect this is because total full UPDRS scores do not include items from Section IV (complications of therapy), whereas the full UPDRS-8 includes items on motor complications. Thus, we would expect less correlation between these scores in more advanced patients in whom motor complications are more prevalent.

Other groups have also suggested the possibility of shorter PD assessment scales. Van Hilten et al. [4] evaluated the original UPDRS using factor analysis and found that for the ADL section, three factors accounted for 59.3% of the variance and, for the motor section, three factors accounted for 66.6% of the total variance. They demonstrated that 5 ADL items from the original 13 and 6 motor items from the original 14 could be dropped without loss of reliability or validity. Their analysis did not include items from the original UPDRS Section I (mentation, mood, and behavior) or Section IV (complications of therapy).

Rabey et al. [8] developed the Short Parkinson's Evaluation Scale (SPES) based on the UPDRS, but used a 4-point rating system (0–3) instead of a 5-point system, and modified some of the original instructions. This scale consists of 24 items, including 8 motor items, 8 ADL items, 3 mental items, and 5 complications of therapy items. These investigators demonstrated that the SPES has inter-rater reliability similar to the full UPDRS. The SPES was further modified and designated the SPES/SCOPA (Scales for Outcomes in PD) [9] in a version that consists of 21 items including 10 motor items (8 by examination, 2 by history), 7 ADL items, and 4 motor complications items. Inter-rater reliability, internal consistency, and correlations with other measures of disease severity were similar for the SPES/SCOPA and UPDRS, and there was good correlation between the related subscales of the SPES/SCOPA and the UPDRS. The investigators found that it took about 8 minutes to complete this scale.

In comparison to these other scales, the UPDRS-8 offers several potential advantages. First, by selecting a subset of items from the original UPDRS, we have not changed the instructions or the rating system. Therefore, clinicians should be very familiar with these items, and their individual clinometric properties should not be any different. In addition, our scale contains substantially fewer overall items (8 total, including 2 nonmotor, 3 motor, and 3 motor complications items) and should take substantially less time to complete.

Although we selected UPDRS items based on our clinical experience, data from other studies provide support for our choices. Each of the motor items we selected has a high loading factor in one of the three separate factors determined by van Hilten et al. [4] to account for 66.6% of the total variance in the motor section of the UPDRS (finger taps, Factor 1, 0.82; gait, Factor 2, 0.70; rest tremor, Factor 3, 0.78). In addition, Marinus et al. [9] eliminated ratings of lower extremity tremor, as we did, noting that less than 4% of patients in pooled data from several studies had tremor scores in the legs higher than in the arms, and only 2% of patients had tremor in the legs with no tremor in the arms.

The UPDRS-8 consists of items from the original UPDRS that focus on PD features relevant to clinical decision making. It exhibits good correlation with the full UPDRS and the PDQ-39. Based on these attributes, we believe that it can provide a useful, rapid assessment of PD patients in clinical practice. We have thus far demonstrated that it is sensitive to relatively large changes in clinical status. Whether it might be useful in clinical trials depends on demonstrating that it is also sensitive to relatively small changes.

Declaration of interest: The authors have nothing to disclose related to the content of this manuscript.