Evaluation of Clinical Utility of the Personal KinetiGraph ® in the Management of Parkinson Disease

INTRODUCTION: Parkinson’s disease (PD) is a disorder characterized by complex motor and non-motor symptoms that can be difficult for patients to accurately communicate. Wearable technologies portend improvements in assessment and monitoring of these symptoms, with their clinical utility cur-rently being evaluated in routine clinical care. OBJECTIVE: To evaluate the clinical utility of the Personal KinetiGraph® (PKG®) Movement Recording System in the routine clinical care of persons with PD (PWP). METHODS: Clinically stable, non-demented PWP presented for two routine clinic visits that included: medication review, symptom review, neurological examination including the Movement Disorder Society Unified Parkinson’s Disease Rating Scale (MDS-UPDRS) III/IV, and completion of a clinical management plan by a movement disorder specialist prior to review of the PKG report. After reviewing the PKG report, the clinician completed a modified clinical management plan taking into consideration the findings of the PKG. This was re-peated at a second visit to evaluate various outcome measures following PKG-enhanced management. RESULTS: The PKG improved the assessment of PD symptoms and the response to treatment, while increasing patient activity levels and compliance. Clinical management plans enhanced by PKG led to different recommendations in 29.4% of cases compared with standard of care due to higher rates of bradykinesia, dyskinesia, tremor, and fluctuations identified by PKG. Using the PKG in the clinical management plan led to a change in medications in 75% (21/28) of patients and both a statistically significant difference and a


Introduction
Parkinson's disease (PD) is a neurological disorder characterized by progressive loss of motor control and a host of non-motor symptoms including cognitive impairment, depression, and impulse control behaviors [1]. The prevalence of PD in the United States increases with patient age and the minimum case burden is estimated to be 930,000 by 2020 increasing to over 1.2 million cases by 2030 [2]. Medicare claims data reveal persons with PD (PWP) have increased health care utilization and spending with an annual, all-cause, per-patient cost of $55,033, a healthcare burden double that of the non-PD cohort [3]. PWP require various types of healthcare services with outpatient physician office visits being the mainstay of clinical management. In terms of overall office visits and healthcare utilization for any medical reason, PWP attended over 9 office visits per patient year [4]. Furthermore, a recent report commissioned by the Michael J. Fox Foundation showed a $51.9 billion per year total cost of PD, a substantial increase over previous estimates [5].
PD is a complex disorder with varying symptom frequency, severity and coexistence of motor and non-motor symptoms for each PD patient. Patients report challenges with articulating symptoms and treatment response because symptoms may be unpredictable, difficult to recognize, and vary within a given day and from day-to-day. Patients report that important goals for improving their care include reductions in "off" time, better symptom control, and less side effects [6].
Clinical management of PWP is similarly complex, symptom-based and requires individualization to address functional status and quality of life. Optimal management relies on patient/caregiver-reported assessments and clinic-based evaluations during physician office visits that commonly occur every 3 -6 months, with frequency increasing as symptoms dictate. The patient's condition during a physician office visit may not reflect the range of or most troubling symptoms experienced during routine activities of daily living [7]. Therefore, an opportunity exists to obtain more thorough and accurate information clinicians can be used to evaluate patient symptomology, response to therapy, and monitor disease progression over time to support delivery of patient-centered care [8].
To address this unmet need, wearable technologies offering continuous objective measurement (COM) platforms have emerged [9] [10]. These new technologies provide clinicians with information about a patient's movement throughout the day during activities of daily living and variation in movement

Objectives
The primary objective of this study was to evaluate the clinical utility of the PKG System when used in routine clinical care of PD patients.  [28]. A REDCap database was utilized for data collection and management [29]. Quality control audits were carried out to ensure data accuracy and integrity.

Design and Procedures
No restrictions were placed on the types of treatments that could be prescribed or changed by the MDS given the pragmatic nature of the study and the time elapsed between visits was per the standard of care. To reduce the potential for bias, at each visit, the routine clinic visit and clinical management plan were completed by the study MDS prior to reviewing the PKG report. Additionally, Visit 1 UPDRS data was not reviewed by the study MDS prior to completing the UPDRS for Visit 2.

Information Provided by the PKG
The PKG is a 4-page report depicting patient movement graphically and numerically, and includes the following information for each day of the wear period and summarized for all wear period days ( Figure 1): • A: Validated dyskinesia score (DKS) representative of hyperkinetic movements plotted against time of day for individual days over the full recording period [14] • B: Validated bradykinesia score (BKS) representative of the degree of slowed movements plotted against time of day for individual days over the full recording period [14] • C: Record of the patient's self-reported acknowledgement of taking levodopa relative to the patient's prescribed medication times Use of the median BKS and DKS scores helped to characterize whether patient symptoms were considered to be "controlled" or "uncontrolled" in relation to target ranges that were defined based on normal reference ranges and modified according to expert opinion [14] [21] [27] [30] [31].

Statistical Analysis
An accrual size of up to 40 patients was empirically selected to allow for potential drop-outs. Given the intended design of this study, relevant summary descriptive statistics are reported for all outcome measures of interest at each visit and we did not correct for multiple comparisons. Levodopa-equivalent dose (LED) was calculated for each PD medication dose using a standardized protocol [33] because the amount of levodopa included in each PD medication varies. MDS-UPDRS III and IV scores were calculated using previously published and validated procedures [26]. A paired t-test was used for comparisons of continuous measures between study visits, and Wilcoxon signed rank test for ordinal data. Subgroup analyses were not carried out due to small sample sizes. All P values are two-sided, with statistical significance evaluated at the 0.05 alpha level. All analyses were performed in SPSS version 26 (Armonk, NY: IBM Corp).

Study Population and Demographics
A total of 35 PWPs fulfilling inclusion/exclusion criteria provided informed consent and were enrolled into the study between 6/2/2016 and 2/28/2017 with 28 completing both study visits and were included in the final analyses. Seven patients were withdrawn from the study prior to completion by the investigator. One patient admitted to using a walker after enrollment (exclusion criterion), one patient changed his own medication schedule during a PKG Watch wear period, one was non-compliant with PKG wear, one PKG led to inadequate data collection and the patient was unable to repeat, one patient became bedridden due to an unrelated illness, and two were lost to follow up. Table 1 outlines the demographics and baseline characteristics of the 28 patients who completed the study.

Clinical Management Optimization
The addition of the PKG to the clinical assessment revealed a higher degree of symptom severity than was noted by the clinical history alone in 18 patients (64%) at Visit 1 and 8 patients (29%) at Visit 2, resulting in 21 (75%) and 9 (32%) clinical management plan changes at Visit 1 and Visit 2, respectively (Table 3). These medication changes included one or more of the following: add a new medication, stop a mediation, increase a medication dose, decrease a medication dose or adjust dose timing.

Case Study
A 60-year old man with 1-year history of slowed movements, rest tremor and rigidity was diagnosed with PD. Carbidopa/levodopa 1 tablet every 4-hours (QID) had been initiated and produced subtle improvements in tremor and bradykinesia according to the patient who was unable to discern shifts in on-off status. Work-up included a normal brain MRI and positive DaTscan demonstrating reduced dopamine transporter binding on the right more than the left. During visit 1, the patient reported symptoms of bradykinesia, freezing and tremor, with onset of levodopa effect in ~1 hour and no clear wearing off which he reported were "unpredictable". MDS-UPDRS showed moderate bilateral rigidity, bradykinesia, mild gait dysfunction, slight freezing, moderate postural instability, moderate global bradykinesia, and slight rest tremor of left leg. Based on the history and exam, the patient appeared undermedicated and the management plan included increasing the frequency of levodopa to every 3-hours (5/day). After reviewing the PKG results, the bradykinesia and tremors were confirmed along with medication compliance, though no dose-response was observed over the 6-day wear suggesting the patient was globally bradykinetic and always "off". This was confirmed by the high median BKS of 35.1 (severe bradykinesia), low FDS of 6.1 (minimal fluctuations), and PTT of 4.7% (moderate tremor). Based on the PKG, the patient was considered severely undermedicated and the plan was to increase to 1-tab every three hours for 1-week, then increase again to 2-tabs every 3 hours as tolerated after 2-weeks. The patient returned approximately 3-months later for follow up again noting onset of levodopa effect in 1-hour, no fluctuations, and now reported peak-dose dyskinesia on 2-tabs taken 5-times daily. The patient also reported daytime somnolence. MDS-UPDRS now showed mild rigidity, slight bradykinesia, mild gait dysfunction, slight freezing, moderate postural instability, and no tremor or dyskinesia. The clinical management plan included a change to carbidopa/levodopa 2 tabs taken every 4-hours to reduce symptoms of dyskinesia that the patient reported were occurring > 75% of the day and causing at least a mild impact on his daily activities. A review of the PKG report showed improved bradykinesia (BKS = 29.8), evidence of fluctuations (FDS = 8.1), no dyskinesia (DKS = 2.9), and improvement in tremors (PTT = 2.4%) (Figure 2).
The case illustrates a routine challenge encountered in the clinical management of PD, including the over-reliance on the medical history, the limitations of a brief examination without ecologically valid corroborating evidence, and the complexity of PD symptoms the further exacerbate these challenges. In this case, both the standard of care and the additional PKG findings correctly identified the patient to be bradykinetic, though the quantitative nature of the PKG emphasized the severity of the symptoms and led to a potentially more aggressive levodopa titration schedule. On follow up, the standard of care and PKG management plans diverged due to the reliance on the patient's history of peak-dose bothersome dyskinesias that were not seen on the PKG. Clinicians often encounter similar scenarios whereby a potentially unnecessary dose adjustment is carried out due to patient report that subsequently leads to a clinical decline.  Having the corroborating objective data from the PKG helped the clinician to avoid a scenario that may have led to an additional office visit or an adverse event.

Discussion
Continuous objective measurement technologies are of significant interest across the healthcare community for the potential to improve healthcare quality and efficiency as well as serve as a catalyst for accelerating the pace of innovation [36] [37] [38]. These technologies are driving a paradigm shift from sole use of in-clinic subjective measurements to in-home continuous monitoring during activities of daily living, a transformation in health care delivery of particular importance in prevalent, heterogeneous, and chronic disease states like diabetes, cardiovascular disease, and asthma [39] [40] [41] [42].
COM use in the care of PWP is following this growing trend and holds promise for earlier diagnosis, improved accuracy of symptom detection, especially for bradykinesia which is difficult for patients to articulate, enhanced disease monitoring, patient engagement, therapy selection, and treatment optimization [43] [44]. COM technologies have the potential to improve upon assessments done in the artificial, constrained environment of the clinic and provide new insights into the impact of medication intake on PD symptoms. The expansion in time and environment that COMs provide has the potential to better inform the clinician of the true state and range of the patient's symptoms. Moving these technologies from research interests to the routine clinical care armamentarium requires clinical validation and ongoing study of clinical utility. In recent years, early adopters have published consensus statements that define how COM data can augment clinical decision making as in the care of patients who are poor historians or have difficulty articulating symptoms, have excessive daytime sleepiness, and in the optimization of new therapies [30] [31].
In this study, we integrated the PKG System into our busy tertiary care PD clinic and evaluated how augmentation of clinical care with the new information this system provides impacts care delivery. One unexpected finding was the divergence of patient-reported symptoms compared to objective measurements of the PKG System. For example, a patient may recall periods of hypersomnolence while the PKG shows accelerometry patterns consistent with wakefulness or a patient reports he/she is experiencing worsening tremors while the PKG shows dyskinesia and no tremor. A number of possibilities may explain these disparities, including poor quality of sleep, mild cognitive impairment limiting recall, and limitations in the patient or caregiver ability to differentiate the complex phenomenology of various motor symptoms in PD. It is not surprising that patients under-reported their medication non-compliance. More importantly though was the observation that compliance rates improved at Visit 2 with the use of the PKG and the patient's presumed awareness that their compliance was being monitored. A similar and equally important observation occurred with regard to patient activity levels. While seemingly trivial and difficult for physi-cians to accurately assess, the combination of improved medication compliance and increased physical activity in PWP are key factors in the management of PD. These findings also highlight the pitfalls of assuming that patient's consistently under-or over-report their symptoms, with the use of the PKG augmenting the history obtained. An important aspect of this study was the capturing of patient qualitative information regarding the use of COM.
Our findings showed that the use of the PKG System yielded more actionable information than could be obtained by an MDS performing a medical history and neurological examination alone. It is therefore not surprising that PKG-guided management led to a change in treatment in 75% of patients. While it could be argued that improvements in motor-based outcome measures such as the MDS-UPDRS III would improve regardless of the treatment, our findings argue that the type of recommendation depends on the patient and their particular symptom(s). In other words, an increase in levodopa dose across the group may have led to reductions in bradykinesia and tremor scores, while worsening dyskinesia measures. Instead, the use of the PKG System led to significant improvements in the UPDRS-III without exacerbating motor complications as evidenced by a small but non-significant improvement in UPDRS-IV scores. Furthermore, while it makes rational sense that increasing medication leads to positive effects on motor scores, Farzanehfar and colleagues [21] found that 22% of 103 study participants managed with the PKG System were well controlled at the onset of the study and did not require medication adjustment. Unfortunately, limitations in the standard of care prevent both patients and clinicians from accurately predicting whether the PWP is adequately treated.
In addition to enhancing the history and neurological exam findings in PWP, the widening use of COMs such as the PKG System in clinical management will accelerate the debate to define what adequate treatment entails. In the management of diabetes mellitus and hypertension, managing patients to established targets is the standard of care. In the management of PD, these targets are starting to be defined by expert consensus [30] [31] though further studies are needed to demonstrate that treating PWP to particular targets will impact their clinical outcomes.
Given the exploratory nature of this study, some limitations deserve mention.
A key limitation of the study was that patients were not followed through medication optimization; therefore, the two visits captured in this study offered a brief snapshot in the care continuum of these patients. As such, while the clinical outcomes observed here are encouraging, overall clinical outcomes achieved when COM is used to optimize medical management of PD patients could not be fully assessed. Additionally, this study did not have a control group; therefore, we cannot directly attribute results seen here to the PKG System. However, the study aimed to isolate the impact of the new information provided

Conclusion
Based on the data collected in this study, we found the PKG System to be a valuable tool in augmenting clinical management planning and decisions, and when utilized along with a clinical assessment. The device was well received by both physicians and patients, scoring high in survey results as a tool to assess impact of therapy and indicating the device had an overall positive impact on patient care and outcomes. Further research is needed to continue the important work of creating evidence-based guidance for the role of COM in the clinical management of PWP.