Emergence and evolution of social self-management of Parkinson’s disease: study protocol for a 3-year prospective cohort study

Extensive research has shown that social engagement and supportive social networks protect the health and well-being of older adults [10‐12]. Studies are small and rare that address how families manage their social lives and relationships while living and aging with PD and its progressive disability and complicated medical management [13‐15]. Longitudinal studies of PD typically do not examine the dynamics and management of social life [16‐19]. These studies do however provide initial evidence that people with PD become lonelier, more emotionally vulnerable and more socially isolated as the disease progresses, and show that changes in disability and socio-emotional well-being can be detected over a 3- to 4-year period. Family caregiving in PD can be emotionally and physically exhausting, yet some families report emotionally rich and rewarding lives [14, 20‐22]. Families living with a person who has greater capacity to express spontaneous warmth and gratitude may have a more enriched interpersonal and social life than families living with someone who is less capable of displaying positive feelings. No research, to the best of our knowledge, has addressed this possibility or the outcomes of expressive masking on physical and mental health in PD.

Most research on the management of daily living with PD has focused on self-management of physical symptoms by following practices such as doing exercises, taking medication, or modifying activities to conserve energy or prevent falling [23, 24]. Yet our preliminary studies of daily living with PD have demonstrated that social concerns are a primary focus for individuals with PD and their care partners, that disease symptoms and social aspects of self-care are inter-related and mutually influential, and that social networks and environment are critical factors in navigating daily life with PD [25, 26]. Expressive disability creates additional needs to manage stigma, public understanding and perceptions [27, 28].

The current project proposes that social self-management of chronic disease is a valuable quality of life indicator. We define social self-management as the practices and experiences that ensure personal social comfort while supporting mental and physical well-being. Articulating this model will guide research to identify social factors that are deleterious to or protective of quality of life when living with chronic disease. PD offers a model for studying the effect of physical disease on the social self-management of daily life when physical symptoms affect fundamental social capacities. Table 1 shows our definition of social self-management and the contrasting and physically-focused definition of “looking after one’s health” in the World Health Organization’s International Classification of Functioning, Disability & Health (ICF) [29]. Our definition brings to the forefront the missing social link in the management of health.

To better understand the social lives of people with PD, it is necessary to translate motor impairment in the face, body and voice into social participation outcomes. A social ecological analysis of facial, bodily and vocal behavior produces this translation [30‐32]. Healthy engagement in social life occurs in relation to bodily capacities and resources in the social environment of informal and formal social networks. Figure 1 shows that social self-management is a social ecological model that integrates each of three major ICF components of health together as a conceptual unit. Our longitudinal study of the daily lives of individuals with PD and their care partners is designed to validate the construct of social self-management as an evolving ecological system.

The objective of this project is to understand the social self-management systems and trajectories of people living with PD. The specific aims are as follows:

This prospective cohort study design tracks general patterns of the sample as a whole as well as variation among individual trajectories in social self-management. The tracking creates evidence necessary to develop social life interventions that address the typical issues faced by people living with the disease as well as individual variations in needs. Over a 3-year period, the study follows 120 individuals with PD and their associated primary care partner. Individuals with PD are included whether or not they have an identified care partner.

There are seven full assessments of approximately two hours duration and in-person, one at baseline and one every six months thereafter, with six in a clinical research lab and one in the home. Between the 6-month full assessments, there is an additional brief telephone call (15 to 30 minutes), totaling seven phone calls over three years. Our innovative design of 14 assessments over the three years aims at detecting social or health triggers that affect individual participant’s trajectories [33]. Frequent points of contact, while also building statistical power, increases the possibility of capturing remembered details of daily life that send health patterns into non-linear trajectories, such as anniversaries, retirement, bad colds or significant family events [34, 35]. Figure 2 shows the study timeline.

Recruitment, consent and data collection protocol have been approved by the Social, Behavioral & Educational Institutional Review Board of Tufts University Medford Campus (protocol # 1212038), and the Boston University Medical Campus (BUMC) Institutional Review Board (protocol # H-32114).

Each of the seven in-person assessment sessions is designed to take a maximum of two hours to complete. At each session, initially, participants with PD and care partners are interviewed separately with parallel questionnaires to assess each of their daily life activities, health and quality of life. After separate interviews, the two are interviewed together about their combined social self-management. Participants with PD are asked to take their medication approximately 45 minutes before their assessment session in order to be “on”, that is, moving and functioning at maximal capacity during their session. Upon arrival they are asked about the timing and effectiveness of their medication on that day as well as the severity of their movement symptoms. Updates on medication changes and background information are collected such as, marital status, living arrangement, occupational status, and notable mental and physical health or life events in the recent past or since the previous visit.

The diminished expressive capacity that occurs in PD is a motor problem that can be confounded with the motor symptoms of depression, apathy or cognitive impairment [38]. At initial screening for this study, individuals with dementia are screened from participation (i.e., ≥ 26 on the MMSE) yet dementia may develop over the course of the study. We monitor dementia and basic cognitive functioning of both participants at every visit with the Montreal Cognitive Assessment [39]. The Geriatric Depression Scale [40] is administered to both participants at every visit to monitor depression throughout the study period. We do not exclude participants or data based on these measures. Rather we will use the measures for covariate analysis and to group individuals for blocked analyses.

The order of assessments is designed to maximize rapport and coherence of the sequence of questions, while minimizing fatigue effects on performance assessments. Unless noted as otherwise, assessments are administered to both the participant with PD and the care partner. Variables and assessment procedures are described below in order of priority to our overall study objective and summarized in Table 2.

Measures described here focus on assessing participants’ experience and participation in social activities and their self-management of social life – which fall under the Activity & Participation construct of the ICF -- and its relationship to their motor and non-motor symptoms— which fall under the Body Function construct of the ICF. These measures are administered at every visit.

Measures described here assess participants’ social environments, specifically informal and formal social network composition, contact frequency, social exchange of support, and the physical environments of social participation—which fall under the Environment construct of the ICF.

These measures assess social self-management outcomes. The social comfort measures assess emotional well-being derived from one’s social networks and are associated with the Environment construct of the ICF. The health and well-being measures assess health quality of life, disease severity and impairments and are associated with the Body Function construct of the ICF.

Calls are scheduled to interview the participant with PD and the care partner separately. The primary objective of these calls is to help inform findings relative to shifts in social trajectories. Participants have the opportunity to elect an alternate means of assessment, such as a mailed questionnaire, if a telephone assessment is perceived as burdensome. We and others [60] have found that telephone assessments that occur between scheduled in-clinic visits are welcomed by people with PD, who typically do not find them burdensome, and experience them as comforting, positive attention. The 15- to 30-min audiotaped protocol involves the following assessments:

The Interpersonal Communication Rating Protocol: Individual Expressive Behavior (Parkinson’s Disease Version) (ICRP-IEB) [61] is used as the primary measure of the expressive capacity of participants with PD in videotaped discussions about management of social activities at baseline and 6-month follow-ups. This rating protocol employs a “thin slice” method, which links discrete behaviors (e.g., an upturned lip, a movement of the limb) into socially meaningful units (e.g., smiling, happiness, dominance) that are closely aligned with individuals’ social life outcomes (e.g. health or work success) [62‐64]. The primary method is to extract short segments (thin slices) from a video or audiotaped social interaction, and have raters draw behavioral or social conclusions from the segments. For this project, 60-second clips are extracted from the videotapes at two standardized time points during the interview: first when the participant is asked to describe a frustrating activity, and again when asked to describe an enjoyable activity [46]. 60-second clips have been found to yield optimal accuracy-to-slice length ratio for making judgments of behavior [65]. Using the methods described in the ICRP-IEB manual, trained research assistants view the clips separately and rate the quality, intensity and frequency of expressive behavior on 20 discrete actions (e.g., smiling, gesturing, bodily movement, vocal tone) and PD symptoms (tremors and postural slouch) that observers use as cues to form judgments about a target individual’s emotions, thoughts, social motives and personality. Our previous studies provide evidence that expressive capacity in PD can be measured reliably and validly at this social level of analysis [46, 66]. Expressive behavior composite scores are formed based on principal component analyses.

The data analysis assumes 120 participants with PD and seven full assessment time points over three years; dropout per year at 15%; intra-class correlation (ICC) at 0.82, based on prior data [24]; and type I error at alpha = 0.05. We have 80% power to detect a mean change over time of 0.40 of a standard deviation in the PDQ-39 outcome per year; 80% power to detect a correlation between continuous variables of 0.23; and 80% power to detect a difference of 0.50 standard deviations in the outcome scale between genders (Aim 3). Previous studies of similar outcomes have found a sample size of 120 to be adequate for demonstrating statistically significant effects [19, 24].

We will perform longitudinal data analysis using the multilevel model for change (also called random coefficient, mixed, or hierarchical model) [70]. Primary outcomes measured at baseline and the semi-annual fixed time points are: 1) retained activities (ACS), 2) proportions of quantitatively content coded words during open-ended responses of social self-management discussions (LIWC), 3) network composition, structure and exchange scores, 4) social network resource utilization scores (modified CIRS), 5) social isolation score, 6) positive interaction with partner score, 7) felt stigma and enacted stigma scores (SSCI), 8) physical and mental health scores (SF-12), 9) health quality of life with PD scores (PDQ-39 or PDQ-8), and 10) motor capacity scores (MDS-UPDRS, Parts III and IV). Outcomes measured at all 14 time points are: 1) physical and mental health scores (SF-12), 2) health quality of life with PD score (PDQ-8), and 3) loneliness score (Social Isolation, NHP). We will explore the temporal dynamics of these outcomes and will attempt to better understand the similarities, differences and factors governing such dynamics. By including random effects, we will explore the population trajectory and the degree of variations of individual trajectories. For outcomes reported by both the person with PD and the care partner, we will be depicting both perspectives, which enable us to assess the synchronicity between the two sources. We will formally evaluate, through statistical tests of interaction terms, whether particular factors affect the ratings of one source more than another.

The multilevel models can accommodate subjects with and without care partners, as well as incomplete data collection, as long as data are missing at random [71]. For sensitivity analyses we will use pattern mixture models to account for the possibility of informative missingness. The models will stratify by pattern of and reason for missing data [72].

For exploratory purposes, we will expand the general multilevel modeling approach to consider potential effects of unanticipated triggers assessed at the individually identified time points [33]. We will re-calibrate the timing of measurements as “time elapsed since an event” or “time preceding an event”. Recalibration will allow us to synchronize triggering events across the cohort, allowing us to determine if there are synchronized changes in one of our outcomes of interest after occurrence of an unanticipated event. For example, there may be an increase in loneliness following hospitalization of a spouse or a decline in social activity following an episode of physical illness.

If a sufficient number of subjects have experienced a similar event, we will explore non-linearities with non-parametric fitting. For example, some social triggers may have a temporary U-shaped trajectory. This description of potential rapid or unexpected changes in the trajectories is highly innovative for research in the social aspects of PD and other chronic degenerative conditions. Adding trigger event monitoring to our research design builds power for repeated measures, and it may illuminate our understanding of disease progression as a social ecological phenomenon.

Attrition can be a problem in longitudinal studies of a disease like PD, due to morbidity, mortality or other uncontrollable factors. Our previous research, which enrolled similar numbers of people with PD, had a relatively low attrition rate despite significant time burden [24], and we expect a similarly low rate in the proposed study. We will collect data on reasons for any attrition, implement rigorous strategies to counteract controllable sources of attrition and test whether our strategies are effective. Retention strategies, such as activities that develop strong rapport with participants, may create unintentional “social intervention” effects on study outcomes [73]. We will apply our strategies with the intent of maximizing retention while minimizing and monitoring unintentional effects on everyday social life.

Other symptoms of PD may be more predictive of social outcomes than expressive masking. We will monitor other potentially stigmatizing symptoms, including tremor, drooling and stooped posture. In addition, we will augment and triangulate our ICRP-IEB measurement of spontaneous expression with a recently developed measure that specifically identifies discrete facial actions of participants with PD during a set of brief standardized emotion role-playing and imitation scenarios [74].

Although existing research is supportive of our hypothesis that gender moderates the effects of expressive capacity on life trajectories, there are many parameters of living with PD that cannot be controlled. Expressive masking may create vulnerability in women, but in general aging women are more likely than aging men to have strong social support networks, which is a possible confound for our study [75]. Recent literature suggests that PD presents and progresses differently in men and women, for example, that women may have a more benign disease than men [76]. Gender research into quality of life domains is in its infancy, thus our study is unusual and does not have precedents to follow. We will be careful to evaluate gender at all steps of the project.

Our long term goal is to guide the development of family-centered and evidence-based interventions aimed at supporting social integration and preventing isolation and loneliness in women and men living with PD. Interventions that support the social lives of people living with PD are likely to have a significant positive effect on both their and their family members’ physical health and socio-emotional well-being. To the best of our knowledge, this project is the first to address whether an individual’s capacity for nonverbal expression has a crucial predictive role in the social self-management of PD.Our intensive repeated measures design provides a model for detecting triggers that rapidly change life trajectories. We expect that the results will contribute solid evidence to guide the development of interventions that directly target protective and risk factors for the social self-management of PD. In addition, this project will provide a model for rigorous examination of social self-management in other chronic health conditions that affect middle-aged and older adults’ social capacity, such as other neurodegenerative disorders, stroke, facial paralysis, and hearing or visual loss.