Prevalence and Cumulative Incidence of Caregiver-Reported Aggression in Advanced Parkinson Disease and Related Disorders

Abstract

Objective

To estimate the point prevalence and cumulative incidence of caregiver-reported aggressive behaviors among people living with advanced Parkinson disease and related disorders (PDRDs) and secondarily examine variables associated with aggression.

Methods

Caregivers from a clinical trial of outpatient palliative care for PDRD were surveyed about patient aggression at baseline and every 3 months over 12 months. Baseline responses were used for point prevalence. Cumulative incidence was calculated using responses from caregivers with no reported baseline aggression and available data at all other time points. Measures of disease severity, quality of life, mood, and caregiver burden were included in correlation and relative risk models, adjusting for age, sex, and diagnosis.

Results

Of 170 caregivers, 31 (18.2%) reported physical aggression, and 18 (10.6%) reported sexual aggression. Twelve-month cumulative incidence for physical and sexual aggression was 21.1% (23/109) and 16.0% (19/119), respectively. Physical aggression cumulative incidence was associated with patient depression (r = 0.37), patient-perceived quality of life (r = −0.26), caregiver burden (r = 0.26), caregiver-perceived patient quality of life (r = −0.26), and caregiver anxiety (r = 0.20). Age, sex, cognitive impairment, and dementia were not associated with aggression. No variables were associated with cumulative sexual aggression.

Conclusion

There was a high prevalence and incidence of aggression in our PDRD cohort. This is an understudied issue in PDRD, and our findings highlight the need for increased awareness among neurologists. Providers should consider assessing for aggression when discussing neuropsychiatric symptoms or screening for caregiver burden. Future research should examine the relationship between aggression and patient and caregiver health outcomes.

Trial Registration Information

Clinical trial registration number: NCT02533921.

Individuals with Parkinson disease and related disorders (PDRDs) are prone to changes in cognition and behavior. As many as 20%–40% of patients with Parkinson disease (PD) experience neuropsychiatric and behavioral disturbances such as depression, apathy, fatigue, hallucinations, psychosis, and impulse control disorders (ICDs).^1–3 Behavioral issues confer an increased risk of hospitalization, higher mortality,^4,5 and are a major driver of caregiver burden.^5,6

In other neurodegenerative conditions, like Alzheimer disease and frontotemporal dementia, behavioral changes can cooccur with physical or sexually aggressive behaviors toward others.^7–9 The prevalence of aggression, including verbal abuse, physical violence, and unwanted sexual advances, among community-dwelling patients living with Alzheimer disease is estimated to be as high as 20%.⁷ There are emerging data to suggest that patients living with PDRD may have a similarly high frequency of aggression, with 1 study of 52 caregivers reporting physical aggression in 65% of patients and sexual aggression in 27%.¹⁰ The underlying cause was thought to be multifactorial and related to cognitive impairment, medication side effects, mood disturbances, and patients' perceptions of disease. Another estimated a wide range of prevalence for physical violence within PD patient–caregiver relationships, with violence observed in 44%–72% of dyads.¹¹ Despite these reports, there remains a dearth of descriptive studies that estimate the prevalence or incidence of aggression in PDRD.

The objective of this study was to estimate the point prevalence and 12-month cumulative incidence of caregiver-reported aggression within a large PDRD cohort, while secondarily examining patient and caregiver characteristics associated with aggression.

Methods

Study Design

This study is a secondary analysis leveraging data from a randomized, comparative effectiveness trial of integrated, outpatient palliative care for individuals with PDRD vs standard care.¹² Participants were recruited from specialized movement disorder programs, community-based neurology outpatient practices, patient support groups, and the Michael J. Fox Foundation trial finder. Participants were enrolled from November 1, 2015, to September 30, 2017. A primary, informal caregiver was enrolled alongside patients as a patient-caregiver participant dyad. The presence of a caregiver was not required for enrollment. Dyads were evaluated at 3 academic medical centers: the University of Alberta, University of Colorado Denver (UCD), and University of California San Francisco (UCSF). Main findings from the larger clinical trial are published elsewhere.¹² Study data were collected and managed using Research Electronic Data Capture electronic data capture tools hosted at the Colorado Clinical & Translational Sciences Institute.

Participants

A total 210 patients and 170 caregivers were enrolled in the larger clinical trial. Eligibility for enrollment in the larger study included English-speaking fluency, age ≥40 years, diagnosis of probable PD, or another PDRD diagnosis, including multiple systems atrophy, dementia with Lewy bodies, corticobasal syndrome, or progressive supranuclear palsy,¹³ and demonstrable palliative care needs reflected in the Palliative Care Needs Assessment Tool (PC-NAT). The PC-NAT screens for issues amenable to a palliative care approach, such as social factors (e.g., presence of caregivers), disease severity, and symptom burden, as well as specific red flags conferring a poor prognosis including persistent hallucinations, falls, aspiration, and/or Hoehn and Yahr score >3.¹⁴ Screening and detection of aggressive behavior is not included on the PC-NAT. Participants were excluded if they had urgent palliative care needs, were unable to commit to study procedures, had another illness requiring urgent palliative care referral, or were already receiving palliative care services. Caregivers were identified by patients' response to the question “Could you please tell us the one person who helps you the most with your PD outside of clinic?” or through caregiver self-referral. One caregiver per participant was enrolled when available. Following enrollment, baseline patient and caregiver characteristics were collected, including age, disease duration, duration of caregiving, sex, race/ethnicity, education level, marital status, income, patients' medications, deep brain stimulation status, measures of disease severity, and caregiver burden.

Outcome Measures of Caregiver-Reported Patient Aggression

Our coprimary outcomes were baseline point prevalence and 12-month cumulative incidence of caregiver-reported physical or sexual aggression. Aggression was measured by caregiver responses to 2 questions: (1) “Do you feel that your relative is physically aggressive towards you or others?” and (2) “Do you feel that your relative is sexually aggressive towards you or others?” Responses ranged from “never,” “rarely,” “sometimes,” “quite frequently,” to “nearly always.” The presence of aggression was determined if caregivers gave any response other than “never.” Caregivers completed surveys privately, and responses were not shared with patients. Study personnel were available to intervene in situations where there was a danger for physical harm or self-harm to participants.

Data collection occurred at baseline and every 3 months over 12 months for participants allocated to the treatment arm and every 6 months for those randomized to standard care. To reduce the possibility of selection bias or treatment effect, only baseline caregiver responses were used to calculate point prevalence. Cumulative incidence was measured using cumulative responses at all other time points over 12 months for caregivers who had answered “never” at baseline, comparing the time from baseline to either the first incident of reported aggression or to the last observation.

Patient Variables

Baseline variables included in correlation and risk estimate analyses included patient demographics such as age, sex, race/ethnicity, study site, treatment arm allocation, marital status, and living situation relative to the caregiver, education, and income. Disease characteristics were also analyzed including disease duration, diagnosis, the use of dopamine agonists, total daily levodopa dose, presence of dementia, study site, the Edmonton Symptom Assessment Scale–Revised for Parkinson's Disease (ESAS-PD) scale for symptom burden,¹⁵ the Unified Parkinson's Disease Rating Scale motor subscale (UPDRS-III) and Hoehn and Yahr scale for disease progression,^16,17 and the Montreal Cognitive Assessment (MOCA) for cognitive impairment.¹⁸ Executive domain items included in the MOCA (Trail Making Test, Part B; phonemic fluency; and verbal abstraction) were analyzed separately. Mood disturbances, including depression and anxiety, were measured using the Hospital Anxiety and Depression Scale (HADS).¹⁹ Patient grief was measured using the Prolonged Grief 12-item (PG-12) questionnaire.²⁰ Measures of quality of life and health-related quality of life (HRQoL) were also included, such as the Quality of Life in Alzheimer's Disease (QoL-AD) scale where lower scores are indicative of poorer quality of life²¹ and the 39-item Parkinson's Disease Questionnaire (PDQ-39) with higher scores suggestive of worse HRQoL.²² PDQ-39 subscales of disease stigma, social support, and communication were individually included in the analysis.

Caregiver Variables

Several baseline caregiver variables were included in the correlation and risk estimate analyses including demographics, study site, mood, perceptions of patients' quality of life, and caregiver strain. Demographic variables included duration of caregiving, marital status, and living situation relative to the patient, age, sex, race/ethnicity, education, and income. Caregiver mood including depression and anxiety was measured using the HADS.¹⁹ Caregiver perception of patients' quality of life was assessed using the QoL-AD-Caregiver scale, with lower scores indicative of worse quality of life.²¹ The QoL-AD-Caregiver is validated for use in measuring caregiver perceptions of patient quality of life in PD.^21,23 Finally, caregiver burden was assessed using the Zarit Burden Interview (ZBI) scale, with higher scores indicative of greater caretaking strain.²⁴

Statistical Analysis

Descriptive statistics were used for patient and caregiver variables and outcomes to estimate frequencies, means, and SDs. Physical and sexual aggression were defined by dichotomizing the physical and sexual aggression questionnaire items into “never” or any other answer aside from “never.” Cumulative physical and sexual aggression were defined as any reported aggression across all time points in the study. Only patient-caregiver dyads with aggression item responses at all time points were considered for cumulative aggression to account for missing data. Proportions were calculated for physical and sexual aggression both at baseline and cumulatively.

Associations between patient and caregiver characteristics and caregiver-reported physical and sexual aggression was analyzed using 2-way tables, relative risk models, and Spearman correlation models, adjusting for age, sex, and diagnosis. The Benjamini-Hochberg procedure was used to control for the false discovery rate. Statistical significance was determined by a p value <0.05. Statistical analyses were performed using SAS software (version 9.4; SAS Institute, Cary, NC).

Standard Protocol Approvals, Registrations, and Patient Consents

This study was approved by institutional review boards at all sites, and the clinical trial was registered on ClinicalTrials.gov (NCT02533921). All participants provided informed consent or, if they lacked capacity to consent, provided assent with informed consent obtained from a legally authorized representative such as a medical proxy.

Data Availability

The authors confirm that the data supporting the findings of this study are available in previously published supplementary materials.¹² Any data that support the findings of this study that are not included in the aforementioned materials are available from the corresponding author (Z.A.M.) on request.

Results

Baseline patient and caregiver characteristics are summarized in Table 1. Patients were disproportionately male (70.0%), White (92.4%), and educated with at least some college education or greater (85.2%). Across all participants, there was quantifiable cognitive impairment at the time of enrollment (mean MOCA score = 23.6, SD = 5.0), with 31.8% meeting the clinical criteria for dementia at baseline. The majority of patient-caregiver dyads were married (90.0%) and shared the same household (90.6%). Caregiving duration was greater than half of the reported disease duration, defined as the time since diagnosis (5.4 years of caregiving vs 9.5 years of disease duration).

Table 1.

Baseline Patient and Caregiver Characteristics

Point Prevalence

A total 170 caregivers responded to the aggression questionnaire items at baseline. The point prevalence of physical and sexual aggression at baseline was 18.2% (31 of 170 caregivers) and 10.6% (18 of 170 caregivers), respectively. There were relative differences between study sites with a lower baseline prevalence of sexual aggression at the University of Alberta (5.3%) compared with UCD (15.3%) and UCSF (11.1%).

Cumulative Incidence

Among caregivers with data at all other time points over 12 months (N = 135), 109 (80.7%) reported no physical aggression at baseline, and 119 (88.1%) reported no sexual aggression. The 12-month cumulative incidence of aggressive behaviors in this subgroup was 21.1% (23 of 109 caregivers) for physical aggression and 16.0% (19 of 119 caregivers) for sexual aggression, with an exposure time of 100.5 years for each and an annual rate of 0.23 (95% confidence interval [CI] 0.15–0.34) and 0.17 (95% CI 0.11–0.27), respectively. Thirty-five caregivers had missing data at one or more time points across the study.

Baseline Correlates of Aggression

Several variables were associated with caregiver-reported physical and sexually aggressive behaviors at baseline. The greatest associations with physical aggression included patient depression (HADS-Depression, r = 0.23, p = 0.002), caregiver burden (ZBI, r = 0.22, p = 0.01), and caregiver depression (HADS-Depression, r = 0.21, p = 0.01). Physical aggression was also associated with HRQoL (PDQ-39, r = 0.20, p = 0.01) and caregiver anxiety (HADS-Anxiety, r = 0.19, p = 0.01). Total daily levodopa dose was associated with both physical (r = 0.18, p = 0.007) and sexual aggression (r = 0.19, p = 0.003). Other variables associated with sexual aggression included caregiver burden (ZBI, r = 0.18, p = 0.02), disease duration (r = 0.16, p = 0.04), and symptom burden (ESAS-PD, r = 0.16, p = 0.04). PDQ-39 subscales related to mobility (r = 0.22, p = 0.004) and disease stigma (r = 0.20, p = 0.01) were associated with physical aggression. Both physical and sexual aggression were associated with the PDQ-39 social support subscale (physical, r = 0.18, p = 0.02; sexual, r = 0.17, p = 0.03) and communication subscale (physical, r = 0.18, p = 0.03; sexual, r = 0.17, p = 0.03). There were no associations between either type of aggression and measures of cognitive impairment, the presence of dementia, executive function domain items on the MOCA, and the use of dopamine agonists.

Correlates of Cumulative Aggression

Several baseline variables were associated with caregiver-reported cumulative aggression over 12 months (see Table 2). Cumulative physical aggression was associated with patient depression (risk estimate 1.17, 95% CI 1.10–1.24), caregiver depression (risk estimate 1.11, 95% CI 1.02–1.21), caregiver anxiety (risk estimate 1.10, 95% CI 1.02–1.18), patient-reported quality of life (risk estimate 0.93, 95% CI 0.89–0.97), caregiver perceived patient quality of life (risk estimate 0.92, 95% CI 0.88–0.97), patient grief (risk estimate 1.04, 95% CI 1.01–1.06), and caregiver burden (risk estimate 1.05, 95% CI 1.02–1.09). There were no variables associated with cumulative sexual aggression after correction for the false discovery rate. There was no association between treatment group allocation (palliative care vs standard care) and aggression at 12 months (risk estimate 1.03, 95% CI 0.61–1.76, p = 0.89).

Table 2.

Baseline Variables Associated With Cumulative Physical Aggression at All Other Time Points After Controlling for False Discovery Rate

Discussion

We observed a high point prevalence of caregiver-reported physical and sexual aggression and a similarly high 12-month incidence with physical aggression being reported at higher rates than sexual aggression. This suggests that caregivers experience patient aggression at comparable frequencies to other conditions associated with behavioral changes, like Alzheimer disease or frontotemporal dementia.^7–9,25 Aggression was associated with medications, patient mood, symptom burden, patient perceptions of disease, and quality of life. Caregiver mood and caregiving burden were also associated with aggression, likely reflecting the interplay between these behaviors and caregiver well-being. Caregivers of persons living with advanced PDRD already feel the stresses of caretaking, and it seems that aggressive behaviors may be intensifying this strain.²⁶

Depression among patients was associated with caregiver-reported aggression, both at baseline and through the duration of the study. This builds on previous descriptions of the relationship between PD and mood, where more than a quarter of patients experience depressive symptoms and nearly a third develop an anxiety disorder during the disease.²⁷ However, the relationship between mood and aggression is complex and likely involves psychosocial factors. Mood disturbances can influence patients' perceptions and beliefs about their illness, with increasing depressive symptom burden correlating with a greater sense of disability.²⁸ Patient perceptions of social support can also affect mental and physical health in PDRD, with fewer intimate relationships correlating with greater depression, anxiety, and stress.²⁹ Our findings suggest that these perceptions are associated with specific changes in behavior, namely aggression, for those individuals living with advanced disease.

The observed association between aggression and PDQ-39 subscales measuring patients' perceptions of disease stigma and communication with others provides additional insight into the intersecting relationship between social support, mood, and behavior. People living with PD often feel misunderstood by others, leading to shame, embarrassment, and isolation.³⁰ The presence of depression or communication barriers can intensify this sense of stigma and amplify the experience of social isolation.³⁰ Aggressive behaviors in this context may be representative of patients' attempts to externalize frustrations with depression, coping with illness, or managing social isolation. To address these issues likely requires an integrated model to PDRD care, merging medical management with social services and disease-tailored behavioral health interventions, like cognitive behavioral therapy and mindfulness-based approaches.³¹

Along these lines, patients' grief was associated with physical aggression over the course of our study. Patients living with PDRD may experience grief related to the sense of loss, which accompanies advancing disease, changing perceptions of functional ability, lost years of living, loss or changes in personal identity, and reductions in quality of life. Poor perceptions of social support, disease stigma, or communication barriers may accentuate this sense of loss, especially when patients feel unable to talk about grief with their loved ones. Possible targets for addressing grief in PDRD may include incorporating interventions like grief counseling and support services into comprehensive care.

The majority of caregivers in our cohort were women and spouses to patients. To this end, aggression may be representative of marital discord or intimate partner violence. The Centers for Disease Control and Prevention's National Intimate Partner and Sexual Violence Survey (2015) estimates that in the United States, the lifetime prevalence of contact sexual violence and physical violence by an intimate partner toward women to be 18.3% and 30.6%, respectively, whereas violence toward men occurs less frequently, 8.2% sexual violence and 31.0% for physical violence.³² However, there are discrepancies between determinants of violence in the general population and associated characteristics of patients and caregivers in our cohort. We were surprised to see that patient and caregiver sex was not associated with aggression, indicating that the pattern of violence among PDRD patient–caregiver dyads may not follow trends observed in the general population. This suggests that factors associated with intimate partner violence in the community may not be the same as those in the PDRD patient–caregiver relationship.

We observed an association between total daily levodopa dosing and both physical and sexual aggression. Higher doses of dopaminergic medications, like levodopa and dopamine agonists, are well-known contributors to behavioral changes in PD, conferring an increased risk of ICDs with hypersexuality being 1 known manifestation.³³ ICD-associated behaviors, and subsyndromic ICDs, present unique clinical dilemmas such as requiring reductions in dopaminergic medications or, in severe cases, initiate the use of antipsychotics, both of which carry an increased risk for unmasking or exacerbating motor symptoms.³⁴ Despite 22.9% of patients in our study being prescribed a dopamine agonist, there was no association between the use of this class of medication and caregiver-reported aggression. Although changes in sexual behavior are a known manifestation of ICD, we would expect dopamine agonists to be larger drivers of this type of aggression if it were primarily due to ICD.³³ Unfortunately, our study was limited in that we did not track ICDs among patients. Future prospective studies on aggression in PDRD should include monitoring for the presence or development of ICDs. The lack of association with dopamine agonists (especially sexual aggression) also needs to be examined further.

Despite the well-described link between PD and frontal lobe dysfunction, we found no association between executive domain items on the MOCA and caregiver-reported aggression.^35,36 In PD, this form of cognitive impairment can lead to a loss of emotional detection and decreased empathy, which can translate to higher caregiver burden.³⁶ Detection of these cognitive and metacognitive deficits can be challenging for even experienced clinicians and is likely underrecognized in patients who do not otherwise meet the threshold for a diagnosis of dementia.³⁷ However, the lack of association does not rule out a relationship between executive dysfunction and aggression, as our study included a limited cognitive battery with relatively small sample.

There was no difference in cumulative incidence of aggression between patients randomized to the palliative care intervention or standard care groups. This was not an explicit focus of the intervention, and providers in the study did not routinely screen for aggression outside of evaluations with study personnel. There may still be a role for using a palliative care approach to address aggressive behaviors, and it should start with better screening. Multidisciplinary palliative care team members, including social workers, chaplains, nurses, and psychologists, may prove useful in screening, detection, and intervention (e.g., training caregivers in de-escalation techniques). Visits with ancillary providers provide an opportunity outside of the caregiver-physician conversation where aggression may be easier to disclose. These types of visits may provide additional face-to-face time, creation of a safe space for discussion, or use diverse approaches to assessing specific issues driving caregiver burden.

Our study has limitations. First, it is difficult to generalize our findings to the greater PDRD patient population due to a lack of diversity in our cohort. Participants were also more medically and psychosocially complex with demonstrable palliative care needs as a requirement for enrollment in the study. Our findings are best generalized to patients with advanced disease, complex psychosocial issues, mood or behavioral disturbances, or to caregivers experiencing significant caregiving strain. The methods used in this study also limit our ability to reach conclusions regarding causality between aggression and patient and caregiver characteristics. This is especially relevant to our observations on mood disturbances, which could be causative or a direct consequence of experiencing aggression. Finally, although we used methods employed in previous studies,^10,11 the psychometric properties of our 2-item questionnaire remain undefined. Studies that examine the operationalization of this method are needed to examine potential barriers to accurate reporting by caregivers. Despite these limitations, strengths of our study include the manner in which caregivers provided information on aggression, namely that their responses were confidential and not shared with patients. This likely helped increase the rate at which caregivers provided open and honest answers as opposed to verbal disclosure with study personnel. This method should be considered in future studies that aim to produce tools for detecting aggression in PDRD.

Physical and sexual aggression toward caregivers and others is an underrecognized complication of PDRD associated with caregiver burden. Our findings support the need for tools that detect and assess for aggression as well as increased awareness of the issue among medical providers. Neurologists, movement disorder specialists, and palliative care providers should be cognizant of the potential for aggression while incorporating caregivers into routine care. Regular assessments of caregiver burden represent one opportunity for detection, prompting investigation of root causes and identification of solutions through careful discussions with caregivers. Aggressive behavior as a manifestation of disease is a clear issue that can negatively affect caregiver well-being. The high prevalence of aggression in our cohort cannot be ignored and expands on preliminary findings previously published in the literature. Future research should examine which factors are predictive of aggression in PDRD and the influence of these behaviors on patient and caregiver outcomes (e.g., institutionalization and caregiver burden). Future studies would benefit from a mixed-methods design to gain a richer understanding of the caregiver experience.

TAKE-HOME POINTS

• → Physical and sexual aggression are understudied issues in advanced PDRD, occurring at comparable rates to other neurologic conditions associated with behavioral changes, such as Alzheimer disease.

• → Aggression in PDRD is associated with worse patient and caregiver depression and anxiety, greater caregiver burden, and other complex psychosocial issues (e.g., grief and patient perceptions of social support).

• → Increased awareness among medical providers (e.g., primary care providers, neurologists, and movement disorder subspecialists) is needed with a low threshold to inquire about aggression when discussing neuropsychiatric symptoms, behavioral changes, or caregiver burden.

• → Prospective studies are needed to determine the true prevalence of these behaviors in the general PDRD population while examining the influence of aggression on patient and caregiver outcomes.

Appendix. Authors

Study Funding

Research reported in this publication was funded through a Patient-Centered Outcomes Research Institute (PCORI) Project Program Award IHS-1408-20134. The views, statements, and opinions in this publication are solely the responsibility of the authors and do not necessarily represent the views of the Patient-Centered Outcomes Research Institute (PCORI), its Board of Governors or Methodology Committee, and the Development and Informatics Service Center (DISC) (NIH/NCRR Colorado CTSI Grant # UL1 RR025780).

Disclosure

The authors report no disclosures relevant to the manuscript. Full disclosure form information provided by the authors is available with the full text of this article at Neurology.org/cp.