A Prospective Evaluation of an Outpatient Assessment of Postural Instability to Predict Risk of Falls in Patients with Parkinson's Disease Presenting for Deep Brain Stimulation

View the entire article with references and supplemental information on the Wiley Online Library. 
Note:  Reference links embedded in the article below will also take you to the article on the Wiley Online Library.

Return to Table of Contents

Authors: Nicholas J. Brandmeir MD, Cheryl L. Brandmeir MSPT, Kristine Kuzma MSN, CNS and James McInerney MD

Article first published online:   27 NOV 2015 | DOI: 10.1002/mdc3.12257



Postural instability (PI) and falls, major causes of morbidity in patients with PD, are often overlooked. DBS is a mainstay therapy for Parkinson's disease (PD) and has been purported to both worsen and improve PI. An effective PI evaluation that can predict fall risk in patients with PD presenting for DBS is needed.


Forty-nine consecutive patients with PD were enrolled. Self-reported falls were the gold standard. Tests evaluated were the Berg Balance Scale (BBS), Timed-Up-and-Go (TUG), Pull Test, and Biodex Balance System Sway Index on firm (SI-FIRM) and soft (SI-SOFT) surfaces.


The best single tests for fall risk were the BBS and SI-FIRM, each with sensitivities of 79% and specificities of 60% and 65%, respectively. When the evaluation was combined into a composite measure requiring four positive tests out of five, the sensitivity was 72% and specificity was 80%.


A simple, efficient outpatient physical therapy assessment is effective in diagnosing fall risk in patients with PD.

Idiopathic Parkinson's disease (PD) is a common illness that is marked by both motor and nonmotor symptoms. The motor symptoms of PD can be divided into those that respond to levodopa and those that do not.[1] These non-levodopa-responsive motor symptoms are often characterized chiefly by postural instability (PI) and gait disturbance. The other common, nonmotor symptoms of PD, such as dementia, dysphagia, and autonomic dysfunction, are also not responsive to l-dopa.[1] These non-l-dopa-responsive symptoms are the major causes of morbidity, mortality, and institutionalization in patients with PD, with a major contribution coming from falls, often secondary to postural instability.[1, 2]

Several studies assessing PI in PD and its response to DBS have demonstrated no injury to postural stability from the procedure, but also no improvement.[3-5] These same studies did demonstrate statistically significant improvements in some laboratory measures of different components of postural stability, but no study has translated these laboratory findings into clinical benefits.[3, 4] A single influential study demonstrated that gait worsened in the first 0 to 6 and 6 to 12 months post-STN-DBS.[6] A review of postural stability in PD in the setting of DBS demonstrated the conflicting evidence on the effects of DBS on postural stability in this patient population.[7] Other studies have shown improvements in PI secondary to administration of l-dopa and/or decrease in patients' postural stability when in the off stage of l-dopa therapy.[8] Because of these varied reported findings, it would be useful to have a standard assessment of PI in patients with PD that could determine baseline PI and the response of the patient's PI to changes in therapy, including STN-DBS and how these changes in PI relate to fall risk. Before any test of PI could be used to follow changes through time or before and after surgery, it would need to be validated in its ability to predict fall risk in this patient group.

Although there is a standardized, well-studied, and disseminated instrument for assessing the l-dopa-responsive symptoms of PD, the UPDRS; it has many limitations, including requiring specialized training for a professional to administer, requiring a large amount of patient time/effort, only one measure evaluating postural instability (the “pull” or “posterior tug” test), and little clinical utility in predicting morbidity, especially morbidity associated with falls.[9] Furthermore, new treatments are emerging for addressing PI in patients with PD.[10, 11] Especially intriguing among these is stimulation of the pedunculopontine nucleus (PPN) for improvement of PI.[12] To date, conclusive data showing an improvement for this treatment have not been forthcoming, but that may be because of the lack of an effective tool for evaluating PI in patients with PD.[13]

Past studies that have evaluated PI and gait in patients with PD and DBS have relied on complex, laboratory-based kinematic assessments.[5, 14] Other studies focused on wearable technology to improve the diagnosis of PI.[15] Though accurate, these assessments are difficult to generalize because of the significant expense and expertise required for their use. Other studies have validated some highly reliable, outpatient assessments in PD, but these patient populations were restricted to a geriatric population,[16] or specifically excluded patients that were pursuing or had undergone surgical therapy for their PD.[17] Both of these studies were able to demonstrate that a battery of simple outpatient physical therapy tests could be useful in diagnosing falls and fall risk among patients with PD.[16, 17] Furthermore, these studies were limited by their reliance on multiple self-reported falls as a gold standard.

It is clear from previous work that despite the success of DBS in treating some motor symptoms of PD, significant morbidity remains that may prevent patients from realizing the full benefits of their surgery.[18, 19] PI is one area of PD where DBS may not effectively alleviate symptoms, but the actual effects of STN-DBS on PI have not been adequately explored in a prospective fashion. Because of this, it is necessary to have a reliable, simple, and affordable to way to evaluate patients for PI before surgery. Potentially, an effective measure of PI would allow adjunctive therapies to be maximized and allow their response to surgery and other therapies to be followed objectively in both academic and community settings.

The aim of this study was to generate and validate an outpatient assessment of PI as it relates to fall risk for patients with PD being evaluated for STN DBS.

Patients and Methods

Fifty-six consecutive patients with PD presenting for DBS were observed by a physical therapist for a preoperative evaluation to assess the cardinal features of their PD as well as their postural stability. Patients were accrued from August 2013 through February 2015. Fall data were collected for 49 of 56 patients. The 7 patients without fall data were excluded from this analysis. Subject descriptive statistics are displayed in Table 1.

Table 1. Descriptive characteristics of the patient cohort

  All Patients

Medians with ranges shown

N 49
Age, years 68 (45–83)
Years with PD 7 (1.5–20)
Gender 64% male
UPDRS 3 42.5 (13–103)
UPDRS 4 7 (0–18)

The reference standards for this study were patient or family reports of falls within the last year. Any falls reported within the last year were considered as positive for the reference standard.

All tests were carried out by one licensed physical therapist with additional training in movement disorders and postural instability while patients were in the on medication state. The tests utilized for the cardinal features of PD were the UPDRS parts III and IV. Part III of the UPDRS is a clinician administered motor evaluation whereas part IV is concerned with motor complications of medication. The five tests utilized for assessing postural stability in these patients were the Timed-Up-and-Go (TUG),[8] the Berg Balance Scale (BBS),[20] item 3.12 from the UPDRS III (the Pull Test),[21] and two measurements from the Biodex Stability System (Biodex, Inc., Shirley, NY).[22] The Biodex is a circular platform-based system that determines balance perturbations by detecting changes in the center of pressure imparted to the platform by the patient. The Biodex measurements included were the Sway Index with the patient's eyes closed on both stable (SI-FIRM) and unstable (SI-SOFT) surfaces. Sway Indices are automated, objective, numerical scores of postural stability based on manufacturer determined standards. The BBS is a standard test used to evaluate patients with falls.[23] The TUG is also a standard test used to assess falls in the elderly population and has been used to evaluate patients with PD in the on and off medication states as well as to separate them reliably from age-matched controls.[8] The Biodex system has been used to measure postural stability in many patient populations and center of pressure systems with eyes open, eyes closed on stable and soft surfaces have been used with good reliability to measure postural stability and changes over time in elderly and fall-prone populations.[22, 24, 25] Some patients were unable to complete the Biodex test without maximum assistance from the physical therapist to avoid a fall. For these patients, a maximum score of 18.0 was assigned regardless of the actual automated score. The SI-FIRM was considered positive for values >1.5. The SI-SOFT was considered positive with values >2.9. These cutoffs were chosen because they were the cut-off values that provided the highest value when adding sensitivity and specificity of the test together. According to published standards, the BBS was considered positive if it was <46 and the TUG was positive if patients took longer than 14 seconds.[20, 26] The outpatient examination as a whole was evaluated by considering the predictive value of a patient being positive on three of the five individual tests evaluated and positive on four of the five tests evaluated.

Statistical Analysis

Statistical analysis was done with Microsoft Excel (Microsoft, Inc., Redmond, WA). Sensitivity, specificity, and likelihood ratios (LRs) were generated using standard 2 × 2 tables. Test results were reported as LR because LR gives the change in probability of a diagnosis given a positive or negative test. Thus, they give the clinician an intuitive and easy-to-use interpretation of the diagnostic accuracy of each test derived from the more traditional, but less clinically applicable, sensitivity and specificity.

Study data were collected and managed using REDCap electronic data capture tools hosted at the institution. REDCap (Research Electronic Data Capture) is a secure, Web-based application designed to support data capture for research studies.[27]

This research was reviewed and approved by the institutional review board and informed consent was obtained from each participant.


Sensitivity, specificity, positive LR, and negative LR tests detecting falls are shown in Table 2.

Table 2. Diagnostic accuracy of tests in detecting self-reported falls

  Sensitivity (%) Specificity (%) LR+ LR

Sensitivity, specificity, positive (LR+) and negative (LR) likelihood ratios shown. 3+ = patients with 3 of 5 positive tests; 4+ = patients with 4 of 5 positive tests

BBS 79 60 1.98 0.34
TUG 45 90 4.48 0.61
Pull Test 79 50 1.59 0.41
SI-FIRM 79 65 2.27 0.32
SI-SOFT 83 20 1.03 0.86
3+ 79 65 2.26 0.32
4+ 72 80 3.62 0.34

For the 29 patients reporting falls at their initial evaluation, 23 had a positive SI-FIRM, 24 a positive SI-SOFT, 23 a positive BBS, 13 a positive TUG, and 23 a positive Pull Test. For those not reporting falls, 7 had a positive SI-FIRM, 16 a positive SI-SOFT, 8 a positive BBS, 2 a positive TUG, and 10 a positive Pull Test.


A primary goal of this study was to determine whether PI (as measured by the diagnostic tests reported above) could be correlated with self-reported fall risk. In this regard, the BBS and SI-FIRM tests were the best individual tests with similar positive and negative LRs. Meaningful comparison of these results to earlier studies of these tests is difficult given that the patient population in this study is different from those included in previous studies (no age limit, undergoing DBS, and so on).[8, 20, 22] Even so, these tests were easily outperformed by the evaluation as a whole. The best performing “test” was the combination of the entire evaluation into one. If a patient had a positive result in any four of our five tests, then they were 3.62 times more likely to report falls than if not. Conversely, if a patient failed to meet this threshold, they were only 0.34 times as likely to report falls. Interestingly, the standard element of the UPDRS III for assessing PI (element 3.12, the Pull Test) was one of the worst performing tests. These results show that the entirety of the evaluation (i.e., examining multiple tests results together) is superior to relying on any one test of PI, including commonly used standards such as the Pull Test, the TUG, and the BBS. This result demonstrates the importance of performing a comprehensive evaluation for PI in this patient population rather than relying on a single test.

Even though the testing reported on here was comprehensive, it was also efficient and fast. The postural elements of the physical therapy evaluation usually take under 30 minutes. The entire evaluation with UPDRS III and IV and the postural evaluation can almost always be accomplished in under 60 minutes in a standard outpatient physical therapy clinic. We feel that this evaluation offers an excellent compromise between the highly reproducible and accurate, but inconvenient and expensive, laboratory-based gait and postural assessments and the fast, but less accurate, office-based tests (such as the Pull Test).

Previous studies have suggested that a physical exam and history are inadequate for accurately assessing falls and have advocated more cumbersome measures.[15] This study demonstrates that an efficient physical therapy evaluation that can be performed at most outpatient physical therapy offices in the community is effective at using measurements of PI with objective thresholds to predict fall risk. Furthermore, this study demonstrates that the comprehensive evaluation is superior to each single test.

One limitation of this study stems from its reliance on patient self-report of falls as the gold standard. This is the best available standard in this patient population and has been used multiple times and has been shown to correlate with actual PI.[16, 17, 28] Given the recall bias and inaccuracy inherent in patient self-reports, it may be that the objective measures reported here are more accurate at predicting fall risk and associated morbidity than self-reported falls. This study was not designed to address that question, but the data here represent a step toward better measurement of PI because it relates to falls in patients with PD. A further limitation of this study is this specific patient population. There is clinical value in measuring postural instability in patients with mild PD, and our study does not address that problem given that all our patients had at least moderate disease. Furthermore, although we are interested in the use of this metric to follow the development of PI in patients with PD over time, those studies remain to be completed, and it has not been validated for that use and it may not be precise enough to detect those small differences.

A strength of this study is that all diagnostic tests were applied equally to every patient regardless of the presence of the gold-standard diagnosis. Also, all tests employed in this study are objective, easily learned, and have a large body of clinical evidence in support of them for this indication. Another strength is that whereas other studies have examined some of these tests in isolation on patients with PD, or on institutionalized patients to evaluate falls and fall risk[8, 24, 25] or evaluated patients with PD excluded from surgical therapy,[16, 17] or with detailed kinematic equipment,[5, 14] none have combined multiple, simple, outpatient tests into one standardized evaluation in this patient population. This is important, because as new treatments become available and patients with PD continue to seek treatment, accurate assessment of the baseline characteristics of these patients and their response to treatment is paramount. Furthermore, as the population ages and the burden of neurodegenerative disorders increases, effective, community-based treatment will be required. The physical therapy evaluation described here can fill this need.

Knowing that the most effective treatments for PD do not significantly improve PI, patients with PI should be directed toward therapy targeted at this problem specifically to give them a better opportunity for maximum improvement from medical and surgical therapies. For most patients, this would mean a course of physical therapy directed at improving posture and gait because this is the most effective known intervention aimed at the non-l-dopa-responsive symptoms of PD.[29, 30] Finally, as new therapies emerge to specifically address PI in patients with PD,[10, 12] it is critical to have an objective, easy-to-teach, and reliable evaluation, such as the one described in this study, for these symptoms given that they often go undetected, underappreciated, and untreated by the current standard treatments in PD.

Leave a Comment

The following required items were not provided or are in the wrong format. Please provide the required responses and submit again:

Comment Title
Comment: 1000 characters
  [[put error message here]]