Activation for Health Management in Colorectal Cancer Survivors and Their Family Caregivers

Abstract

Activation, the state of possessing the skills, knowledge, and confidence to manage one’s own health, is associated with positive self-management behaviors in individuals with chronic illness. Little is known about its role in cancer survivorship. The aims of this study were to describe activation in patients with colorectal cancer and their family caregivers, examine the relationship between patient and caregiver activation, and determine if activation is related to symptom distress, depression, anxiety, fatigue, physical activity, and work productivity. Using a longitudinal, correlational design, a convenience sample of 62 patients and 42 family caregivers completed surveys during postoperative hospitalization, and at 6 weeks and 4 months post-op. Activation scores for both patients and caregivers were stable over time, were not correlated, and were at the third level of activation. Linear mixed effects models revealed that negative emotions were associated with less patient activation and lower caregiver self-efficacy for caring for oneself.

The recent Institute of Medicine report, Delivering High-Quality Cancer Care: Charting a New Course for a System in Crisis, underscored the importance of patient-centered care to improve communication and assure high-quality health outcomes for patients with cancer and their families (Institute of Medicine, 2013). A key component of patient-centered care is enabling and fostering activation in patients and families so that they can participate fully and proactively in their care. Activation refers to the state of possessing the skills, knowledge, and confidence to manage one’s own health (Hibbard, Mahoney, Stock, & Tusler, 2007). It is especially critical during the transition to post-treatment survivorship, when self-management tasks shift for both patients and their family caregivers from acute treatment of the disease to engagement in follow-up care, managing persistent treatment side effects, learning about signs of recurrent disease and late effects, resuming social roles, coping with emotions, and changing and maintaining behaviors to enhance physical and emotional wellness (Given, Sherwood, & Given, 2011; McCorkle et al., 2011). Successful adoption of these tasks, which have significant implications for physical and emotional health during survivorship, requires sufficient activation. Yet, little is known about the role of activation along the cancer care trajectory and specifically, the transition to post-treatment survivorship.

Activation

Activation, as conceptualized by Hibbard and Green (2013), can be thought of as a precursor to the more broad concepts of patient engagement and self-management. It refers to the individual’s willingness and ability to be an active participant in his/her health management. Increases in activation have been shown to be associated with positive self-management behaviors and outcomes in patients with diabetes, hypertension, heart disease, lung disease, arthritis, and HIV infection (Hibbard et al., 2007; Marshall et al., 2013; Rask et al., 2009). The importance of patient activation to economic outcomes has been supported, with lower levels of activation both associated with and predictive of higher health care costs in an analysis of more than 33,000 patients (Hibbard, Greene, & Overton, 2013). In another study, lower activation scores in 695 hospitalized adult medical patients were associated with a two-fold risk of hospital post-discharge utilization within 30 days compared to patients who were more activated (Mitchell et al., 2014). Thus, activation has potential implications for not only achieving good outcomes in patients with cancer, but also in reducing cancer care costs.

Activation is essential for colorectal cancer survivors as studies have stressed the importance of health promotion behaviors, such as physical activity, in reducing cancer-specific mortality in colorectal cancer survivors (Meyerhardt et al., 2006; Meyerhardt et al., 2009). Individuals living with colorectal cancer make up the third largest group of cancer survivors in the United States, comprising approximately 8% of the 14.5 million cancer survivors (American Cancer Society, 2014). These long term survivors will be vulnerable for a myriad of persistent physical side effects and psychosocial challenges, including anxiety and depression that negatively affect health-related quality of life (HRQOL) (Hornbrook et al., 2011; Houldin & Lewis, 2006). Colorectal cancer survivors are at increased risk for a second primary colon cancer (Raj, Taylor, Wray, Stamos, & Zell, 2011) and health problems related to other chronic illnesses may emerge as up to 80% of colon cancer patients report having at least one comorbid condition, most commonly hypertension, arthritis, and diabetes (Ramsey, Berry, Moinpour, Giedzinska, & Andersen, 2002).

Adjustment to colorectal cancer occurs within the context of the family (Northouse, Mood, Templin, Mellon, & George, 2000). Family caregivers of individuals with newly-diagnosed colorectal cancer often describe balancing the pervasive disruption in their lives with the need to keep a positive attitude and normal family routines (Houldin, 2007). Although the research describing the experiences of caregivers in the survivorship phase is limited, studies have noted that emotional distress continues to be a problem during the first year after colorectal surgery, especially for female caregivers (Northouse et al., 2000; Tuinstra et al., 2004) and spouse caregivers report having significantly more emotional distress and less social support than patients (Northouse et al., 2000). There are numerous research reports in the general cancer caregiving literature of the complex problems and burdens of caregiving (Stenberg, Ruland, & Miaskowski, 2010). Family caregivers play a substantial role in supporting and managing patients with cancer and there is a growing awareness amongst oncology clinicians of the importance of enhancing the health and well-being of caregivers through improved assessment, education, support and training of caregivers. Yet, a study of 677 caregivers of patients with colorectal or lung cancer found that half of the caregivers felt inadequately trained for clinical care tasks (van Ryn et al., 2011).

Caregiver activation has two components. First, caregivers must have the skills, knowledge, and confidence for assuming the role of caregiving during all phases of the cancer care trajectory. Second, an often overlooked issue by health professionals is that caregivers must be activated for managing their own physical and emotional health during times when their focus is generally on the needs and health of their family members. A key component of being activated to manage one’s health is having self-efficacy (confidence) to perform specific self-care behaviors. Self-efficacy is considered a core determinant of behavior and refers to an individual’s belief that he or she is capable of performing behavior in a particular situation to produce a specific outcome (Bandura, 1977). Managing feelings, coping with stress, practicing healthy behaviors related to exercise and nutrition, attending to personal health issues, seeking support, and engaging in leisure activities are self-care aspects of caregiving that were found to be inversely related to caregiver stress and burden (Merluzzi, Philip, Vachon, & Heitzmann, 2011). Caregivers who attend to their own physical and emotional health during the trajectory of cancer care are likely to experience less stress. Furthermore, prepared, confident, and healthy caregivers may have a direct influence on the physical, functional, and emotional well-being of the patient (Belgacem et al., 2013; Berry, Elliott, Grant, Edwards, & Fine, 2012; Lau et al., 2010). Yet, there is little research of the concept of activation in family caregivers.

Purpose

This study was undertaken to: (a) describe activation levels in patients with colorectal cancer and their family caregivers, (b) describe the relationship between caregiver and patient activation, and (c) examine the relationship between activation and variables of symptom distress, depression, anxiety, fatigue, physical activity, and work productivity. The specific research questions were:

1. What are activation levels in patients and caregivers?

2. Do activation levels change over time in patients and their caregivers?

3. What is the relationship between patient and caregiver activation levels?

4. Is patient activation related to patient variables of symptom distress, depression, anxiety, fatigue, physical activity, and work productivity?

5. Is caregiver activation related to caregiver variables of depression, anxiety, fatigue, physical activity, and work productivity?

6. What demographic and health-related variables are related to activation in patients and caregivers?

Methods

Study Design, Sample, and Setting

A descriptive, correlational, longitudinal design was used to examine activation in patients with colorectal cancer and their family caregivers. The study was approved by the institutional review board at University Hospitals of Cleveland (UHCMC IRB number 11-11-10C). From April 2012 to November 2013, a convenience sample of patients with colon cancer and their family caregivers was obtained from surgical oncology inpatient division at the Seidman Cancer Center, part of the Case Comprehensive Cancer Center. Ambulatory adult patients with a diagnosis of stage I, II, or III colon cancer, treated with surgery, with or without chemotherapy were included. Patients having stage IV disease or a recurrence were excluded. Adult caregivers, who were the spouse, sibling, or child of patient were included. Caregivers did not need to be living with the patient.

Potential participants were identified in conjunction with the surgical nurse practitioners during a weekly review of patients on the division. Patients were approached in the hospital during their postoperative stay. After consent was obtained, the patient was asked to identify a family caregiver who was a close and consistent support since diagnosis. The designated family caregiver was then invited to participate and consent was obtained either in-person or via telephone. Data were collected from both patients and caregivers at three time points. Baseline surveys were completed with the investigator during the postoperative stay. On occasions when the caregiver was not present in the hospital, the baseline data were collected via telephone. Data collection at time point two (six weeks postop) and three (four months postop) occurred via mail.

Measures

Activation. The Patient Activation Measure (PAM) is a 13-item survey that assesses the patient’s skills, knowledge, and confidence for self-management (Hibbard, Stockard, Mahoney, & Tusler, 2004). The Caregiver Patient Activation Measure (CPAM) is a newly developed 13-item survey that assesses the caregiver’s skills, knowledge, and confidence for caregiving (Craig Swanson, Insignia Health, LLC, personal communication, June 16, 2011). For both surveys, individuals respond to statements using a 4-point Likert-type scale with the options strongly disagree, disagree, agree, strongly agree, and not applicable. Scores are summed to provide a raw score, which are then converted to an “activation score” using a scale provided by the developers of the PAM. Activation scores range from 0 – 100, with higher scores indicating greater activation. The activation scores can also be categorized into an activation level of 1 to 4. These are described as: starting to take a role (Level 1), building knowledge and confidence (Level 2), taking action (Level 3) and maintain behavior (Level 4).

Reliability and construct validity of the PAM were established in a large sample of multimorbid adults (Skolasky et al., 2011). Reliability of the CPAM has been supported with a reported Cronbach’s alpha of .87 (B. Mahoney, personal communication, July 25, 2013). In the present study, the PAM demonstrated high internal consistency, with Cronbach alphas of .90, .87 and .92 at time points one, two, and three, respectively. The CPAM had alphas of .84, .81, and .82 at the respective time points.

Caregiver self-efficacy for performing behaviors related to their own physical and emotional health while caring for the care recipient, was evaluated with five items from the “Caring for Oneself” subscale of the Caregiver Inventory (Merluzzi et al., 2011). These items are rated on a nine-point scale, ranging from not at all confident (1) to totally confident (9). Scores are summed, with higher scores indicating greater confidence. Reliability and validity of the Caregiver Inventory was supported in 133 caregivers of terminally ill individuals, with a Cronbach alpha of .88 for the Caring for Oneself subscale (Merluzzi et al., 2011). In this study, the Cronbach’s alphas were .86, .90, and .93 at time points one, two, and three, respectively.

Depression/Anxiety/Fatigue. Symptoms of depression, anxiety, and fatigue were evaluated in both patients and caregivers using three short form surveys from the Patient-Reported Outcomes Measurement Information System (PROMIS; Cella et al., 2010). For depression and anxiety, individuals were asked to rate on five-point scale if they experienced a particular feeling or mood during the past seven days, ranging from never (1) to always (5). For fatigue, the frequency and impact of fatigue on physical, mental, and social activities during the past seven days was rated on a 5-point Likert scale, ranging from never (1) to always (5). The anxiety and fatigue surveys consists of seven items; the depression survey has eight items. For each survey, the responses are summed to produce a total raw score, which is then converted to T-score using a table provided at PROMIS assessment center website. Higher T-scores represent greater amounts of the symptom.

Symptom Distress. The Memorial Symptom Assessment Scale Short Form (MSAS-SF) was used to assess symptom distress in patients (Chang, Hwang, Feuerman, Kasimis, & Thaler, 2000). Distress associated with 28 physical symptoms and the frequency of four psychological symptoms are rated on Likert scales. Scoring of the MSAS-SF is based on three subscales: the Global Distress Index (GDI), the Physical Symptom subscale (PHYS), and the Psychological Symptom subscale (PSYCH). A total MSAS-SF score is calculated as the average of the 32 symptoms. The possible total score range is 0 to 4 with higher scores indicating higher distress. Reliability and validity of the MSAA-SF has been supported in a study of 299 cancer patients with mixed diagnoses (Chang et al., 2000). In this study, the Cronbach’s alpha for the total scale were .84, .89, and .91 at time points one, two, and three, respectively.

Physical Activity. The International Physical Activity Questionnaire – short form (IPAQ-SF; IPAQ, 2005) was used to determine a global estimate of physical activity in patients and caregivers. This 7-item self-report survey assesses the time spent in walking, moderate-intensity activities, and vigorous-intensity activities during the past week. Using the IPAQ-SF scoring protocol, a metabolic equivalent task (MET) value for each activity is multiplied by the time spent per week to calculate the MET minutes per week for each of the three categories of physical activity (walking, moderate-intensity, and vigorous-intensity). The total score of MET minutes per week is the summation of the three categories and is converted to a categorical variable of low, moderate or high physical activity. The IPAQ-SF also assesses sedentary behavior by asking the participant to report the time spent sitting per weekday during the last seven days. Reliability and validity of the IPAQ-SF was initially established in a large international study of adults and was found to be comparable to other self-report measures (Craig et al., 2003).

Work Productivity. The impact of cancer or caregiving on absenteeism, presenteeism, and non-work activity impairment was measured using the six-item Work Productivity and Activity Impairment Questionnaire (WPAI; Reilly, Zbrozek, & Dukes, 1993). Using a one week recall period, employed patients and caregivers answer questions to determine the hours missed from work, hours actually worked, and the extent to which cancer or caregiving affected his/her productivity while working. Impairment related to non-work activities is measured with a single item. The WPAI yields four scores, reported as percentages, for the impact of cancer or caregiving on (a) absenteeism or work time missed, (b) presenteeism or reduced productivity while at work, (c) overall work impairment, and (d) activity impairment or reduced productivity for regular, daily, non-work activities. Validity of the WPAI has been supported in both caregiver and patient samples and sufficient test-retest reliability has been reported (Giovannetti, Wolff, Frick, & Boult, 2009; Reilly et al., 1993).

Demographic and other variables. Demographic information included age, race, gender, marital status, employment status, education, income, and living arrangement. Cancer type, stage of cancer, months since diagnosis, type of treatments received to date, concurrent treatments, was collected from the patient’s medical record. Patients and caregivers rated their own current physical health status and health-related quality of life (HRQOL) using 5-point scales ranging from excellent (1) to poor (5).

Analyses

Descriptive statistics were used to describe demographic and medical characteristics. The distribution and time trend of activation levels over the study period in patients and caregivers were explored with descriptive statistics and box plots. The relationships between the dependent variables (patient activation, caregiver activation, and caregiver self-efficacy) and independent variables of fatigue, depression, anxiety, physical activity, symptom distress, work impairment, health status, and quality of life were assessed using standard linear mixed effects (LME) models with random intercept and slope. The random effects in the mixed models accommodate the influence and/or heterogeneity of each individual on the repeated observations that is not captured by the observed covariates. The inference and estimation of the LME models is based on the likelihood function. In the LME models, we assume that errors are normally distributed with a mean of zero and a homoscedastic variance. The 2 by 1 vectors of random effects are assumed to be independent of the covariates, and have bivariate normal distribution with zero and unstructured covariance matrix. All LME models for the dependent variables included fixed effects (covariates) such as time of the study and a missing data categorical variable (missing or not). The LME models for the dependent variable, caregiver activation, also included fixed effects of caregiver age and gender. The analyses were performed using statistical software Stata 11.0. Alpha was set at .05 level of significance.

Results

Sample Characteristics

A total of 129 patients were screened and 87 patient-caregiver dyads were invited to participate in the study. Of these, 63 patients and 44 caregivers consented, leading to enrollment rates of 72 and 51 percent, respectively. However, data from one patient-caregiver dyad were excluded from analysis because a non-cancer diagnosis was determined pathologically after enrollment. Another caregiver dropped out of the study after signing consent, but before completing any surveys. Therefore, the final sample consisted of 62 patients and 42 caregivers.

During the course of the study, 16 (25.81%) patients and 7 (16.7%) caregivers dropped out by not returning surveys or responding to calls from the research team. When comparing patient dropouts to patients who completed the study, statistically significant differences were found only for marital status, χ² (1, N = 62) = 4.82, p = .028. Patients, who were non-respondents to follow-up surveys were more likely to be single. Caregivers who dropped out of the study differed statistically from those who remained in the study on marital status, χ² (1, N = 42) = 5.60, p = .018. Caregiver non-respondents were more likely to be married. There were no statistical differences found on baseline scores for activation, fatigue, depression, anxiety, or symptom distress between non-respondents and respondents for both patients and caregivers.

Sample characteristics are displayed in Table 1. The mean age was 65.11 years for patients and 55.05 years for family caregivers. The typical family caregiver was female, White, and married. Most caregivers (71.43%) were in a spousal or partner relationship with the patient. The typical patient was male, White, and married. The mean time since diagnosis was 97.97 days (SD = 87.22, median = 57). In addition, approximately 60% of both patients (n = 37) and caregivers (n= 25) reported having two or more comorbid conditions with hypertension and arthritis being most common.

Table 1.

Sample Characteristics

	Patient	Family Caregiver
	N = 62	N = 42
Characteristic	n (%)	n (%)
Gender
Female	27 (43.55)	33 (78.57)
Male	35 (56.45)	9 (21.43)
Race
White	53 (85.48)	37 (88.10)
African American	8 (12.90)	5 (11.90)
Asian	1 (1.61)
Marital Status
Married	41 (66.13)	36 (85.71)
Single	21 (33.87)	6 (14.29)
Employment Status
Employed	31 (50.00)	28 (66.67)
Not employed	31 (50.00)	14 (33.34)
Annual Household Income ($)a
20,000 or less	9 (14.52)	2 (4.88)
21,000 – 49,999	19 (30.65)	14 (34.15)
50,000 or greater	31 (50.00)	25 (60.98)
Education
High school or less	36 (58.07)	26 (61.90)
College or more	26 (41.93)	16 (38.10)
Type of Cancer
Colon	37 (59.68)
Rectal	25 (40.32)
Cancer Stage
I	13 (20.97)
II	22 (35.48)
III	27 (43.55)
Type of Surgery
Laproscopic resection	49 (79.03)
Open resection	9 (14.52)
Re-anastomosis	4 (6.45)
Chemotherapy
Received neoadjuvant chemotherapy	22 (35.48)
Planned future adjuvant therapy	23 (37.10)

^aThree patients had missing data for income.

As shown in Table 2, the PROMIS measure T-scores for depression, anxiety and fatigue were within one standard deviation of the mean. Patients had, on average, low scores for symptom distress. The mean percentage of overall work productivity loss was highest at time one, which was the post-operative period. At subsequent time points, overall productivity loss ranged from 20.16% to 26.22% for both patients and caregivers. The percentage of patients reporting low physical activity ranged from 59.3% (n = 35) during the immediate post-operative period, to 46.7% (n = 14) and 38.7% (n = 12) at six weeks and four months post-op, respectively. Fewer caregivers reported low physical activity, with 45.5% (n = 15) at post-op, 12.9% (n = 8) at six weeks, and 1.6% (n = 1) at four months.

Table 2.

Descriptive Statistics for Patient and Caregiver Activation and Independent Variables over the Study Period

	Baseline (T1)	6 Weeks (T2)	4 Months (T3)
	N = 62 Pts	N = 41 Pts	N = 46 Pts
	N = 42 Cgs	N = 32 Cgs	N = 34 Cgs
Variable	M (SD)	M (SD)	M (SD)
Patient Activationa	64.63 (15.44)	60.24 (13.39)	64.73 (16.35)
Caregiver Activationa	66.33 (13.26)	59.03 (13.34)	62.11 (13.64)
Caregiver Self-efficacyb	31.74 (8.30)	30.38 (9.03)	31.54 (9.10)
Depressionc
Patient	47.45 (7.54)	47.02 (9.53)	47.80 (8.95)
Caregiver	50.31 (7.07)	51.46 (7.31)	49.46 (8.29)
Anxietyc
Patient	55.22 (8.76)	48.71 (9.83)	47.11 (9.75)
Caregiver	55.70 (7.93)	53.24 (8.65)	51.11 (9.55)
Fatiguec
Patient	50.80 (8.39)	49.95 (6.72)	51.38 (6.68)
Caregiver	50.44 (6.25)	49.20 (5.38)	50.33 (6.68)
Patient Symptom Distressd	0.72 (0.44)	0.57 (0.47)	0.65 (.51)
Overall Work Impairmente
Patienf	70.54 (28.97)	20.16 (29.11)	21.49 (32.60)
Caregiverg	53.43 (32.53)	24.27 (30.52)	26.22 (23.76)

Note. Pts = patients; Cgs = caregivers.

^aPossible range of scores for patient and caregiver activation is 0 – 100.

^bPossible range of scores for caregiver self-efficacy is 5.0 – 45.00.

^cThe PROMIS measures for depression, anxiety, and fatigue yield a standardized T-score with a mean of 50, based on the U. S. population, and a standard deviation of 10.

^dPossible range of scores for patient symptom distress is 0 – 4.00.

^eOverall work impairment due to cancer or caregiving is reported as a percentage.

^fn = 27 at time one; n = 10 at time two; n = 20 at time three.

^gn = 28 at time one; n = 19 at time two; n = 21 at time three

Activation

At each time point mean scores for patient and caregiver activation were within parameters for the third level of activation, indicating some confidence in managing their health and a readiness to make behavior changes. The majority of patients (80.9%) reported level three or four activation during the post-operative period, but this dropped to 62.5% and 67.7% at six weeks and four months, respectively. For caregivers, 77.4% reported level three or four activation initially, with 63.4% and 70.5% at six weeks and four months, respectively. Scores for caregiver self-efficacy for caring for oneself were mid-range at all time points.

The LME model, when adjusting for missing data, indicated that study time point was not related to patient activation levels (regression coefficient = −.14, z = 0.14, p = .889), caregiver activation levels (regression coefficient = −1.99, z = −1.44, p = .149), or caregiver self-efficacy for managing one’s health (regression coefficient = .39, z = 0.53, p = .597). Activation levels did not differ statistically over time.

When adjusting the LME models for time and missing data, there was not a statistically significant association between patient activation for health management and caregiver activation for the caregiving role across time points (regression coefficient = −0.10, z = −1.03, p = .302). Also, there was not a statistically significant relationship between patient activation and caregiver self-efficacy for managing one’s health (regression coefficient = −0.02, z = 0.12, p = .904). When adjusting the LME model for time, missing data, caregiver age and caregiver gender, there was not a statistically significant relationship between caregiver activation for the caregiver role and caregiver self-efficacy for managing one’s health (regression coefficient = 0.18, z = 1.07, p = .284).

Factors Related to Activation

Variables associated with patient activation, caregiver activation, and caregiver self-efficacy are shown in Table 3. Across all time points, lower scores for patient activation were significantly associated with higher scores for fatigue and depression, greater work impairment, poor perceived health status, and poor perceived quality of life. Single marital status and living alone were associated with greater patient activation.

Table 3.

Associations among Dependent and Independent Variables Estimated by Linear Mixed Effects (LME) Models

	Dependent Variable
	Patient Activation			Caregiver Activation			Caregiver Self-Efficacy
Independent Variablea	Regression Coefficient	95% CI	p	Regression Coefficient	95% CI	p	Regression Coefficient	95% CI	p
Age	−0.06	−0.36, 0.24	.686	−0.33	−0.63, −0.04	.028	0.12	−0.09, 0.32	.257
Gender	−2.47	−9.43, 4.48	.486	−8.65	−16.35, −0.95	.028	−2.82	−8.26, 2.62	.310
Race	0.68	−9.13, 10.48	.893	−5.44	−15.69, 4.82	.299	9.30	2.74, 15.86	.005
Marital status	10.99	3.97, 18.01	.002	2.13	−8.95, 13.22	.706	2.22	−4.72, 9.15	.531
Living alone	5.81	1.41, 10.21	.010	−4.99	−13.89, 3.90	.271	4.46	−0.84, 9.76	.099
Cancer type	3.15	−3.82, 10.12	.376	4.06	−2.45, 10.57	.221	−4.66	−9.04. −0.28	.037
Fatigue	−0.004	−0.006, −0.002	.000	0.28	−.0.11, 0.68	.164	−0.18	−0.41, 0.05	.131
Depression	−0.53	−0.82, −0.25	.000	−0.07	−0.41, 0.27	.688	−0.29	−0.48, −0.10	.003
Anxiety	−0.18	−0.42, 0.07	.158	0.09	−0.20, 0.38	.538	−0.20	−0.36, −0.03	.021
Physical activity
Low	--	--	--	--	--	--	--	--	--
Moderate	4.01	−2.22, 10.24	.207	4.00	−3.17, 11.18	.274	−1.38	−5.84, 3.08	.544
High	2.29	−4.30, 8.87	.496	3.99	−3.68, 11.66	.308	5.51	0.85, 10.17	.021
Symptom distress	−4.13	−9.24, 0.98	.113	2.44	−3.00, 7.87	.380	−3.32	−6.58, −0.05	.046
Work impairment	−0.10	−0.19, −0.02	.014	0.008	−0.08, 0.10	.856	0.0004	−0.0002, 0.001	.187
Health status	−5.64	−8.56, −2.72	.000	−0.85	−3.85, 2.15	.577	−0.42	−2.23, 1.38	.648
Quality of life	−5.02	−7.77, −2.33	.000	−0.41	−1.37, 0.55	.402	0.25	−0.25, 0.75	.329

Note. LME models for patient activation and caregiver self-efficacy were adjusted for missing data and time point. LME models for caregiver activation were adjusted for missing data, time point, caregiver age, and caregiver gender. Binary variables include: gender, race (White, Not White), employment (yes, no), living alone (yes, no), cancer type (colon, rectal).

^aPatient variables were evaluated for patient activation. Caregiver variables were evaluated for caregiver activation and self-efficacy, except for patient symptom distress.

Greater activation for assuming the caregiver role was significantly associated with younger age and female gender. After controlling for caregiver age and gender, activation for assuming the caregiver role was not significantly associated with caregiver variables of depression, anxiety, fatigue, physical activity, and work productivity.

Lower scores for caregiver self-efficacy for caring for oneself were associated with higher scores for depression and anxiety, higher patient symptom distress, and lower levels of physical activity. Non-White race, and colon cancer diagnosis were significantly associated greater caregiver self-efficacy for caring for oneself.

The following parameters did not have a statistically significant association with any of the dependent variables: ethnicity, employment, education status, socioeconomic status, relationship to care recipient, and cancer stage.

Discussion

This study contributes to the growing body of literature on activation by describing both patient and caregiver activation over three time points during the post-colorectal cancer surgical transition. The findings underscore the importance of the role of patient activation along the cancer trajectory as it was significantly and directly related to QOL and health status. This finding is similar to longitudinal and cross-sectional studies of patient activation in adults with chronic conditions (Hibbard et al., 2007; Mosen, Schmittdiel, Hibbard, Sobel, Remmers, & Bellows, 2007). Patients with higher activation generally practice healthful behaviors, which in turn lead to positive health outcomes that may ultimately impact quality of life.

The lack of relationship between caregiver activation for the caregiving role and QOL and health status was not an unexpected finding because, conceptually, a caregiver who is activated for the caregiving role will perform behaviors that are directed to the well-being of the care recipient, rather than one’s own. A possible explanation for the lack of relationship between caregiver self-efficacy for managing one’s own health and QOL and health status is that the measures of caregiver self-efficacy, health status, and QOL were too limited in scope to measure the array of behaviors that may be associated with broad concepts of quality of life and health status.

The majority of patients in this study were highly activated to manage their health and their activation levels were stable over four months. This finding is consistent with Jerofke and colleagues (2014), who noted high levels of activation and no statistical changes in activation from pre-discharge to 6-week post-discharge in cancer and cardiac surgical patients. The high levels of activation suggest that there is an opportunity at the post-surgical transition to include health promotion interventions at discharge. However, activation is a dynamic concept and may fluctuate overtime in individuals with chronic disease (Chubak et al., 2012). Future studies are needed to describe changes in individual trajectories of activation through cancer treatment and into survivorship.

In this study, lower patient activation and caregiver self-efficacy for managing one’s health were associated with higher levels of depression and anxiety. The findings confirm those by other researchers that negative emotions are a barrier to activation (Gerber et al., 2011; Goodworth et al., 2014; Hibbard et al., 2007; Marshall et al., 2013). According to Hibbard and Mahoney (Hibbard & Mahoney, 2010) positive emotions enhance creativity, coping, resilience, problem-solving ability, and receptivity to new information. When positive emotions are combined with gaining confidence and experiencing successes, an individual internalizes one’s ability to self-manage into the self-concept. Negative emotions, failures at managing one’s health, and low confidence create a cycle that leads to low activation. Psychological distress is prevalent in both cancer patients and caregivers (Pitceathly & Maguire, 2003; Zabora, BrintzenhofeSzoc, Curbow, Hooker, & Piantadosi, 2001). If managing one’s health is predominantly a psychodynamic phenomenon, it is critical to screen patients and family caregivers for anxiety, depression, and distress, as these impede activation. Interventions aimed at reducing negative emotions and increasing positive emotions will likely impact activation, and will, in turn, increase effective self-management.

An important finding in this study was the lack of statistically significant relationships between patient activation for health management, caregiver activation for the caregiving role, and caregiver self-efficacy for managing one’s health. The theoretical underpinnings of activation and caregiving help to explain this finding. First, activation is a measure of one’s self-concept as a self-manager (Hibbard & Mahoney, 2010) and accordingly, the measures used in this study were a measure of activation as an individual, rather than relational or dyadic, construct. Second, family caregiving is a dynamic and complex phenomenon, consisting of the stress process, contextual factors (personal, sociocultural, economic and health care-related), and the cancer trajectory from diagnosis thru survivorship and end-of-life (Fletcher, Miaskowski, Given, & Schumacher, 2012). It is likely, therefore, that caregiver activation for the caregiving role is more closely associated with other factors that were not measured in this study, such as care demands and caregiver appraisal of burden. The clinical significance of this finding is that given the reliance of patients on family caregivers for tangible care and assistance, caregiver activation is important and cannot be predicted by patient activation. A patient who has low activation for managing his/her health, may or may not have a highly activated caregiver. Also, a caregiver who is highly activated for the role of caregiving may or may not have confidence in managing his/her own health. Clinicians should specifically assess activation in both patients and caregivers.

Overall, family caregivers were activated for assuming the caregiver role, but male caregivers were less activated than female caregivers, meaning that they perceived themselves as having less knowledge, skills, and confidence. This gender difference in activation is reflective of previous research describing society, healthcare providers, and women themselves as “positioning” women to be expert, full-time, caregivers with little option to decline the role (Ussher & Sandoval, 2008). Despite embracing the caregiving role, women report greater burden and more unmet needs related to caregiving than men (Perz, Ussher, Butow, & Wain, 2011). In contrast, society positions men as less suited for the caregiving role. Male caregivers often view caregiving as a competency task and may have difficulty with the emotional reactions of cancer caregiving (Ussher & Sandoval, 2008). Assessment of activation for caregiving by clinicians is needed regardless of gender. It is possible that the activation scores for women in this study are inflated and represent societal expectations, rather than reality. Future research is needed to test activation interventions that are tailored to gender differences in how the caregiving role is viewed and experienced.

There are several limitations to this study. First, generalizability is limited due to the small sample size and missing data. The analytical models, therefore, included a limited number of covariates to avoid an overfitting problem. In addition, longitudinal studies suffer from missing data which leads to reduced number of observations per subject. Our study is not an exception from the missing data issue. About twenty-five percent of observations were missing. The linear mixed model analysis, which assume missing data are missing at random, was adjusted for missing data by including a binary missing data indicator variable in the regression model. Inclusion of a missing data indicator variable leads to efficient parameter estimates and hence better statistical inference for the parameters of interest.

A second limitation is that only nine male caregivers participated in the study. It is important to have a more representative sample to more fully understand gender differences in caregiver activation. Third, assessment of whether the patient actually received adjuvant therapy was not conducted after the baseline measures. This is important to track because patients and family caregivers face additional stressors and new self-management tasks during adjuvant therapy, which may influence activation. Fourth, details of the caregiving situation, such as the illness demands, care needs, and specific clinical skills used by the caregiver, were not captured. This limited our ability to make associations between the caregiving context and caregiver activation, and other study variables. Finally, social desirability and self-enhancement biases may have influenced study results because all measures are based on self-report. Objective measurement of outcomes or behaviors associated with patient and caregiver activation, such as adherence to the treatment plan, would be helpful to substantiate self-reports of activation levels.

A strength of this study is exploration of longitudinal relationships between activation and other variables through a significant transition along the cancer continuum. Additional research is needed to explore activation as it relates to specific self-management behaviors pertinent to cancer survivors, such as adherence to therapy and follow-up appointments, health promotion behaviors, and communication with clinicians. Activated patients and family caregivers are central to improving health outcomes during cancer survivorship.