Improving Documentation of Impulse Control Disorders at a Movement Disorder Program During the COVID-19 Pandemic

Abstract

Background and Objectives

Impulse control disorders (ICD) are a group of behaviors in Parkinson disease (PD), (compulsive buying, gambling, binge eating, craving sweets, and hypersexuality) that occur in up to 20% of individuals with PD, sometimes with devastating results. We sought to determine the rate of ICD screening based on 2020 quality measures for PD care by the American Academy of Neurology.

Methods

We conducted a quality improvement project to document and improve physician ICD screening in a tertiary movement disorder program. Serial medical records were reviewed for 5 weeks before and 13 weeks after an educational session and documentation tool deployments in 2020. Inclusion criteria included the following: idiopathic PD, PD dementia (PDD), or dementia with Lewy bodies (DLB). Individual encounters for 109 patients preintervention and 276 patients postintervention were reviewed.

Results

There was no difference between the preintervention and postintervention (pre-IG vs post-IG, respectively) in terms of age, male to female ratio, proportion of patients with PD, PDD, or DLB, duration of diagnosis, or levodopa equivalents. There was a shift to increased ICD queries above the median for the study period (28.8%) for 7 consecutive weeks in post-IG. The frequency of ICD diagnosis was not different from pre-IG to post-IG (95% confidence interval, 0–32.6 vs 2.7–13.4%, p = 0.444).

Discussion

ICD queries immediately after ICD education and dissemination of documentation tools increased. Both preintervention and postintervention groups were similar in demographic and clinical characteristics. This program was instituted at the height of wave 2 of the COVID-19 pandemic in Alberta during staff redeployment and 100% shift to telemedicine ambulatory care. Our results demonstrate that amid a crisis, quality improvement can still be effective with education and provision of tools for clinicians.

Introduction

Impulse control disorders (ICD) is an umbrella term describing impaired inhibition and judgment that result in compulsive buying, gambling, sexual expression, and other behaviors. They occur in up to 20% of individuals affected by Parkinson disease (PD),¹ which is itself a common neurologic disease. Furthermore, a longitudinal study of a group of patients with PD in whom a large majority were exposed to dopamine agonist (DA) therapy showed a cumulative five-year incidence of 46.1% who developed impulse control disorder symptoms.² Dopamine agonist therapy has classically been identified as the major risk factor of ICD; however, that same longitudinal study showed a 5-year cumulative incidence of ICD of 12.4% in patients never exposed to DA. Male patients with early-onset PD seem to be at particularly high risk, while depression, alcohol consumption, and smoking habit are also associated.³

Routine assessment of patients with PD through the use of the standard nonmotor symptoms questionnaire (NMSQ)^4,5 does not currently include screening for ICD,⁶ although they can have devastating consequences. Furthermore, the American Academy of Neurology has recently identified ICD as sufficiently important to warrant a recommendation for annual screening independent of mood disorders and psychosis in the 2020 Parkinson Disease Quality Measurement Set Update.⁷

We studied the effect of an education program and documentation tool for querying ICD in patients with parkinsonism due to Lewy body pathology (PD, PD dementia, and dementia with Lewy bodies). In this study, we report the results of a single round of plan-do-study-act cycle in which we compared the rate of ICD queries in 2 independent patient populations before and after our intervention. Our report conforms to the Revised Standards for Quality Improvement Reporting Excellence (SQUIRE 2.0) guidelines.⁸ We hypothesized that clinicians in our movement disorder group did not routinely query ICD due to it not appearing on the standard NMSQ⁴ and that education and tools to simplify documentation in the electronic medical record (EMR) would improve performance on this quality measure.

Methods

Standard Protocol Approvals, Registrations, and Patient Consents

Ethical approval was granted by the Health Research Ethics Board at the University of Alberta in September 2020 (Pro00094369). The need for direct patient consent was waived by the Health Research Ethics Board.

Structure of the Clinic

The Movement Disorder Program at the University of Alberta is a multidisciplinary clinic devoted to the care of adult patients with a wide range of movement disorder presentations. Physicians in the clinic are neurologists with subspecialty experience in providing care for patients with movement disorders. On a week-to-week basis, the volume of patients seen depends on the number of physicians available. The availability of physicians is affected by service obligations to the hospital neurology services that typically run 1 week at a time and other academic and administrative commitments.

Before the onset of the COVID-19 pandemic, most visits were in-person assessments. Nursing staff were responsible for taking patients to their rooms, completing a medication reconciliation, ascertaining chief concerns for the physician to address, and reviewing nonmotor symptoms. This was performed with the standard NMSQ.^4,5 After the onset of the pandemic, nursing staff were redeployed and were no longer able to participate in reviewing nonmotor symptoms. In parallel, encounters changed to 100% virtual or telephone delivery.

Documentation Tool

To facilitate documentation of nonmotor symptoms and documentation of ICD, we added the ICD item based on the MDS-UPDRS item 1.6, which has previously been validated.⁹ The derivative questionnaire is based on the standard NMSQ with ICD added as a stand-alone line item (Figure 1). This was provided as a smart phrase in the EMR ( Epic Systems Corporation, Verona, WI) to facilitate documentation of nonmotor symptoms by a physician during a virtual or telephone encounter.

Figure 1. Documentation Tool Provided for Impulse Control Disorder Queries.

Documentation tool derived from the standard nonmotor symptoms questionnaire that was provided to physicians to facilitate documentation of impulse control disorder queries. ICD = impulse control disorder; RBD = rapid eye movement sleep behavior disorder.

Quality Improvement Initiative

We undertook a retrospective chart review for a 5-week period (September 8, 2020 through October 6, 2020). The beginning of this period was chosen when the volume of follow-up appointments with providers in our movement disorder program had returned to normal after a 100% transition to telemedicine, which includes telephone visits and virtual visits. For a two-month period, all ambulatory activity stopped due to the COVID-19 pandemic, and thereafter, volumes gradually recovered to those handled prepandemic. The end of the preintervention period was the day before the educational intervention described in the following paragraph.

The education intervention was developed to be reviewed in less than 1 hour in a single session to raise awareness about ICD queries and our forthcoming research project after the initial chart review. The educational module focused on the prevalence, impact, and treatment of ICD in addition to the planned quality improvement and the proposed documentation tool (Figure 1). This was provided by email to all physicians working in the movement disorder program on October 7, 2020. A follow-up email with the attached tools and instructions on accessing them in the EMR was sent 6 weeks later.

Starting the week of October 12, 2020, we reviewed 13 weeks of consecutive patients who collectively made up the postintervention group. 13 weeks was chosen because the intention was to review 12 weeks of charts to make up roughly a 3-month period or 1 quarter of a year; however, the winter holidays meant that there were no appointments the week of December 28, 2020, through January 1, 2021, and there were reduced volumes the weeks of December 21 to 25, 2020, and January 4 to 8, 2021. The postintervention period therefore ended on January 15, 2021.

Inclusion criteria were an established diagnosis of PD, PD dementia, or Lewy body dementia and 1 encounter only during either the preintervention or postintervention period (to avoid undercounting queries because the quality measure stipulates annual review). We performed a visual review of the progress note contents and used keyword searches for “impulse,” “control,” and “disorder” to ascertain the ICD query rates. We coded the query as a yes or no binary. In postintervention cases where ICD had been queried, we noted the method by which ICD was documented either by the provided tool or an alternative. We also documented whether an ICD query led to a diagnosis of ICD, which we coded as a yes or no binary.

We excluded patients with encounters coded for deep brain stimulator assessments because these appointments occur very frequently in the initial period after the electrode(s) are implanted. We therefore could not reliably identify standard follow-up encounters for patients with deep brain stimulator implants without extensively reviewing individual charts, which would have exceeded the scope of privilege granted by our ethics application. We excluded encounters for levodopa intestinal gel pump titrations because these typically did not involve a standard assessment of motor and nonmotor symptoms; however, we did not otherwise exclude patients with this formulation of levodopa delivery. We excluded patients encounters coded for dystonia because it was not clear whether these always encompassed a complete clinical follow-up beyond repeat botulinum toxin injections.

Statistical analysis

Statistics were calculated by the authors using IBM SPSS Statistics version 28.0.0.0. Demographic and clinical characteristics were compared using standard statistical tests for normally distributed data. Categorial variables were compared using Pearson Chi-square tests. Continuous variables were compared using independent samples T tests. Levodopa equivalents were calculated based on the formula outlined in a previous systematic review.¹⁰ For categorical variables, proportion or frequency is displayed. For continuous variables, mean and SD are displayed. A run chart is displayed with week-to-week frequency of ICD queries on the y-axis and study week on the x-axis to allow statistical analysis of the quality improvement initiative.

Data Availability

Anonymized data for individual patient encounters from which analyses were derived will be made available to qualified investigators upon request.

Results

From September 8, 2020, through October 6, 2020, there were 109 individual patients with encounters in our movement disorder program who met the inclusion criteria and therefore made up the preintervention group (pre-IG). From October 12, 2020, through January 15, 2021, there were 276 individual patients with encounters in our movement disorder program who met the criteria outlined and therefore made up the post-IG . Demographic and clinical characteristics are summarized in Table.

Table.

Demographic and Clinical Characteristics of the Study Groups

	Preintervention	Postintervention	p Value
Period	September 8, 2020, to October 6, 2020	October 12, 2020, to January 15, 2021	—
Number	109	276	—
Sex, M/F	60.6%/39.4%	59.1%/40.9%	0.788
Age (y)	71.8 ± 9.7	70.5 ± 10.0	0.255
Diagnosis, PD/PDD/DLB	89.9%/8.3%/1.8%	86.6%/10.5%/2.9%	0.657
Duration of diagnosis (y)	10.6 ± 6.4	9.9 ± 6.0	0.305
LEDD (mg)	932 ± 570	948 ± 545	0.792
Levodopa RR	93.6%	94.2%	0.816
Levodopa SR	27.5%	36.0%	0.830
Dopamine agonist	14.7%	7.6%	0.034
COMTI	12.8%	16.7%	0.352
MAOB	4.6%	3.6%	0.660
NMDAR	13.8%	13.0%	0.851
Antidepressant	28.4%	39.1%	0.049
Neuroleptic	11.0%	7.2%	0.228

Abbreviations: COMTI = catechol-O-methyltransferase inhibitor; DLB = dementia with Lewy bodies; F = female; LEDD = levodopa equivalent daily dose; M = male; MAOB = monoamine oxidase B inhibitor; NMDAR = NMDA receptor antagonist; PD = Parkinson disease; PDD = Parkinson disease dementia; RR = rapid release; SR = sustained release.

Demographic and clinical characteristics of the preintervention and postintervention groups. Note the preintervention period ended on October 6, 2020, as the educational intervention occurred October 7, 2020.

There were no differences in sex distribution ratio; mean age; proportion of patients with PD, PDD, or DLB; mean duration of diagnosis; levodopa equivalent daily dose; or proportion of patients prescribed rapid-release levodopa, sustained-release levodopa, catechol-O-methyl transferase inhibitors, monoamine oxidase B inhibitors, NMDA receptor antagonists, or neuroleptics (Table). More patients in pre-IG were prescribed a dopamine agonist, while more patients in post-IG were prescribed an antidepressant.

Analysis of the week-to-week frequency of ICD queries in a run chart demonstrates a shift to more frequent documentation in post-IG (Figure 2). For 7 consecutive weeks from week 12 through week 18, the frequency of ICD documentation was greater than the median frequency of documentation (28.8%). In post-IG, 54.8% of the queries used the documentation tool provided, while 45.2% documented the query in the body of the note separate from the provided tool. The proportion of queries that resulted in a positive diagnosis of ICD was not different between pre-IG and post-IG (observed proportions of 14.2% vs 8.1% and 95% confidence intervals of 0–32.6 vs 2.7–13.4, p = 0.444).

Figure 2. Comparison of Week-to-Week Frequency of Impulse Control Disorder Queries.

Run chart demonstrating a comparison of week-to-week frequency of impulse control disorder queries in the preintervention (white, left of vertical gray line) and postintervention (black, right of vertical gray line) periods. Median frequency is demonstrated as a horizontal line (dashed gray). ICD = impulse control disorder.

Discussion

We conducted a single-center quality improvement study evaluating the effects of a single cycle of the plan-do-study-act (PDSA) framework during the height of the COVID-19 pandemic in Edmonton, Alberta, Canada. After an education program and introduction of a documentation tool, we observed 7 consecutive weeks where the frequency of ICD documentation was greater than the median for the study period. The 2 study groups were generally similar in demographic and clinical characteristics. Dopamine agonists were prescribed more frequently in the pre-IG, which should have led to proportionally more queries in the pre-IG if this traditional risk factor was the only impetus for querying ICD. Antidepressants were prescribed more frequently in the post-IG; however, the importance of this is unclear. While this may represent a greater burden of psychiatric comorbidity in the post-IG, our study purview did not allow us to evaluate this. Furthermore, the higher frequency of dopamine agonist use in the pre-IG was insufficient to affect ICD queries; thus a higher burden of psychiatric comorbidity, if present, does not adequately explain the sustained increase in ICD query frequency in the post-IG.

The education program effected a change in behavior that was not related to the care of patients with COVID-19 or a direct response to the health care conditions created by the pandemic. During our study, ambulatory care for patients with PD had shifted to 100% telemedicine delivery; however, additional support was not provided for this transition because physicians, nurses, and allied health professionals were redeployed to meet positions of need directly related to system challenges due to COVID-19. As a result, the typical workflow in which screening for nonmotor symptoms was undertaken by nurses in our program was severely disrupted and the responsibility for nonmotor symptom screening was delegated to the physicians. The EMR in use during the study was first deployed in early November 2019, so clinicians had only been using the EMR for less than a year during the study beginning.

The challenges of the workflow during the pandemic with a relatively new EMR were compounded by physicians sharing in the distraction, stress, and burnout of health care providers. Indeed, a study of community neurologists providing care during the COVID-19 pandemic in the United States captured many of the challenges identified by the authors, including lack of support for new care models, working on the front line, and trouble with self-care.¹¹ Overall, there is limited data on adherence to quality measures sets in PD to provide a comparison of the relative success of our initiative. A report that examined patients with PD cared for in the Veterans Health Administration system showed adherence rates for various quality metrics were variable, but all were below target.¹² Depression, falls, and orthostatic hypotension were queried less than 40% on an annual basis with various reasons posited by the authors. Our study was a cross-sectional analysis of patients who are typically followed up every 3 to 6 months, so we did not expect 100% adherence to querying ICD on a cross-sectional sample. The cross-sectional nature of our study is a limitation to quantifying overall adherence and durability of adherence. Despite this limitation, we show that a short approachable educational module coupled with deployment of a straightforward documentation tool can result in an increase in documentation of ICD in independent groups of patients with similar demographic and treatment characteristics.

Given that this was a single cycle of PDSA, the names of the clinicians were not recorded for future follow-up and the authors did not ascertain individual reasons for not querying ICD at a specific clinical encounter. One possibility was that ICD had been queried at an earlier clinical encounter. We also hypothesized that the lack of a standardized tool for documentation meant that ICD was not prioritized during a busy clinical encounter. Finally, clinicians may not feel that screening has an impact on patient outcomes because a previous study did not find statistically significant relationships between adherence to the 2010 Parkinson Disease Quality Measures and outcome measures, including death, hospitalizations, and ratings of quality of life.¹³ However, the authors of that study acknowledged their small sample size as a limitation. The deleterious impact of ICD are well known in PD, with reports of greater degrees of functional impairment,¹⁴ worsened emotional well-being independent of depression,¹⁵ and increased caregiver burden.¹⁶ Therefore, in addition to challenges produced by limited sample sizes, a lack of relationship between measures and outcomes may represent incomplete management based on the findings from querying items contained in the quality measures.

As with being unable to comment on the reasons for ICD not being queried at individual encounters, the data we collected do not identify individual providers to ascertain whether the providers working in the program each week influenced the rate of ICD queries. It is likely that the structure of our clinic, as described in the Methods section, had an effect early on because the group of providers each week was not identical. Some providers were likely slower to change habits and prioritize ICD documentation than others. Thus, 2 different groups of providers early on could create different rates of query from week to week. Informal discussion at morning meetings, which occur before each clinic encounter, and the reminder sent out 6 weeks after the initial dissemination likely prompted providers to further increase their documentation rate, and therefore, we saw a more sustained and durable rise over the following weeks as demonstrated from weeks 12 through 18 in Figure 2.

Our patient population was representative of a subspecialty movement disorder practice in terms of demographics. PD dementia and DLB were likely underrepresented relative to the population prevalence due to redistribution of care of patients with dementia to long-term care facilities and specialized geriatric clinics. The demographic effects of PD were therefore heavily weighted in both groups. The mean ages of the study populations were not different and were both older than 70 years, which is consistent with increasing incidence of PD after the age of 50 years¹⁷ Furthermore, there were more male than female patients, which is typical for PD. The mean length of time from diagnosis and levodopa equivalents were not different between groups and were representative of patient populations with advanced disease and relatively high pharmacotherapeutic need. We did not quantify cognitive impairment in this population, although it is likely that there was at least a moderate burden of mild cognitive impairment.¹⁸ The average age and possible burden of cognitive impairment likely contributed to the overall therapeutic distribution heavily weighted toward levodopa formulations. Furthermore, the distribution of pharmacologic choices represents the practice patterns of the movement disorder practice at the University of Alberta in Edmonton, Alberta, Canada, where dopamine agonists are typically avoided due to the known association with ICD. This may represent yet another reason for not screening ICD because clinicians likely believe in a lessened risk of ICD by avoiding the best-known risk factor. This is an area that we intend to explore in future cycles of quality improvement aimed at increasing adherence of the quality measures for PD.

While we were not able to separately study effects of the education intervention from the documentation tool, we show that a brief education intervention and simple tool can have a meaningful impact on increasing documentation of ICD. Our study was conducted at a single center in Canada that represents a mid-size group subspecialty movement disorder practice. We believe our results are generalizable to neurologists and other clinicians working in various practice settings based on our patient population studied and the generalizability of the circumstances in which we undertook this initiative. The challenges added by introducing a quality initiative during the COVID-19 pandemic persist and will continue to add barriers to implementing quality improvement initiatives in the future. However, patients with neurodegenerative conditions also experienced disruptions to care delivery, increased needs, caregiver burnout, and difficulties with telehealth and virtual care delivery systems that have become the standard through the pandemic.¹⁹

In conclusion, for quality improvement initiatives to be successful, they need to be simple and focused, such that neurologists and neurology-adjacent specialties may continue to meet the needs of our vulnerable patients. The results of this quality improvement study show that simple and pragmatic quality improvement projects can succeed. This success can occur despite significant disruptions to workflow through redistribution of staff, modified and expanded duties, and a relatively new EMR. In addition, those tasked with an altered workflow may be experiencing the threat of burnout, increased call burden, and periodic illness of staff members, yet can still alter practice patterns. Our initiative shows that quality improvement can be instituted in times of crisis to continue improving the lives of our patients.

Appendix. Authors

Name	Location	Contribution
Brendan N. Putko	Division of Neurology, Department of Medicine, University of Alberta, Edmonton, Canada	Drafting/revision of the article for content, including medical writing for content; major role in the acquisition of data; study concept or design; and analysis or interpretation of data
Janis M. Miyasaki, MD, MEd	Division of Neurology, Department of Medicine, University of Alberta, Edmonton, Canada	Drafting/revision of the article for content, including medical writing for content; major role in the acquisition of data; study concept or design; and analysis or interpretation of data

Study Funding

The authors report no targeted funding.

Disclosure

J. Miyasaki: PCORI research grant, 2021-2023, Canadian Consortium on Neurodegeneration in Aging 2018-Present,University Hospital Foundation 2016–2025, Canadian Open Parkinson Network, Brain Canada 2018–2023, Up to Date royalty 2021, Oxford University Press, US Ambassador 2022–2023; Vice President American Academy of Neurology, Board of Directors, Parkinson Foundation (uncompensated). Full disclosure form information provided by the authors is available with the full text of this article at Neurology.org/cp.

TAKE-HOME POINTS

• → Quality improvement is possible despite challenging staffing and clinical demands, such as during a pandemic.

• → Providing documentation tools allows clinicians to seamlessly implement change in clinical encounters.

• → Tools should be flexible to reflect the nature of clinical encounters (in-person or telemedicine).