Impact of Relationship and Communication Variables on Ambulatory Blood Pressure in Advanced Cancer Caregivers

Abstract

Background

Cancer impacts both patients and their family caregivers. Evidence suggests that caregiving stress, including the strain of taking on a new role, can elevate the risk of numerous health conditions, including high blood pressure (BP). However, the caregiver’s psychosocial experiences, including their interpersonal relationship with the patient, may buffer some of the negative physiological consequences of caregiving.

Purpose

To examine the influence of psychosocial contextual variables on caregiver ambulatory BP.

Methods

Participants were 81 spouse–caregivers of patients with advanced gastrointestinal or thoracic cancer. For an entire day at home with the patient, caregivers wore an ambulatory BP monitor that took readings at random intervals. Immediately after each BP reading, caregivers reported on physical circumstances (e.g., posture, activity) and psychosocial experiences since the last BP measurement, including affect, caregiver and patient disclosure, and role perceptions (i.e., feeling more like a spouse vs. caregiver). Multilevel modeling was used to examine concurrent and lagged effects of psychosocial variables on systolic and diastolic BP, controlling for momentary posture, activity, negative affect, and time.

Results

Feeling more like a caregiver (vs. spouse) was associated with lower systolic BP at the same time point. Patient disclosure to the caregiver since the previous BP reading was associated with higher diastolic BP. No lagged effects were statistically significant.

Conclusions

Caregivers’ psychosocial experiences can have immediate physiological effects. Future research should examine possible cognitive and behavioral mechanisms of these effects, as well as longer-term effects of caregiver role perceptions and patient disclosure on caregiver psychological and physical health.

In advanced cancer spouse caregivers, feeling more like a spouse (vs. caregiver) and perceiving more patient self-disclosure were associated with increased caregiver ambulatory blood pressure

Background

Cancer impacts both patients and their family caregivers [1, 2]. Family caregivers perform a wide variety of tasks in their role, including assisting with activities of daily living and instrumental activities of daily living; supporting the patient’s medical care, helping with decision-making; communicating with providers, family, or friends about the patient’s condition; and providing emotional support to the patient. Caring for someone with advanced, life-limiting cancer can be particularly demanding, as caregiving tasks typically become increasingly complex as the patient’s disease progresses [3] and the physical and psychological symptom burden increases. Additionally, advanced cancer caregivers often face questions and concerns about the patient’s end of life. Unsurprisingly, past work has shown that caregivers of patients with advanced cancer may be at especially high risk of poor outcomes like distress and low quality of life [4, 5].

Although many find benefit in providing care [6–8], serving as a family caregiver can be highly stressful; these tasks can take up significant amounts of time and energy [9], and caregivers often are not given the information or support needed to feel confident and prepared to perform care responsibilities [10]. Caregivers must also cope with their own emotions surrounding the diagnosis [2], including adapting to role changes in their relationship with patients as they shift from spouse, child, or friend to caregiver, with spouses often experiencing particular difficulty adjusting to these changes [11]. In fact, independent of the amount of care needed by the patient, the “sheer restructuring of an important and established relationship” [12](p. 55) like a marriage is a stressful experience in itself [12, 13]. However, past research has found that a strong couple identity (i.e., feeling like a part of a couple) may mediate the stress process for spouse–caregivers and result in better mental health outcomes [14].

The stress process model of family caregiving [15] holds that caregiving stressors can impact caregivers’ psychological and physiological well-being. This model is supported by a large body of evidence suggesting that caregivers are at elevated risk of mental and physical health issues including anxiety, depression, immune dysfunction, and increased blood pressure [4, 9, 16–20]. The finding that caregivers are more likely to have high blood pressure (BP) than their non-caregiver peers is particularly noteworthy, as chronically elevated BP also confers increased risk of developing numerous secondary conditions such as atherosclerosis, heart attack, heart failure, stroke, aneurysm, kidney damage, vision loss, sexual dysfunction, metabolic syndrome, cognitive impairment, and dementia [21, 22]. The sequelae of these negative physical health outcomes can further impact the caregiver’s well-being and, ultimately, their ability to care for the patient. Thus far, the impact of caregiving on physical health—and hypertension/cardiovascular disease specifically—has been well-documented in the dementia literature (e.g., [23, 24]), but much less work has focused on the cancer context [25], where caregiving is often shorter-term, more intense, and more variable as the patient goes through different phases of disease and treatment [26].

Much of the work on the impact of caregiving on BP has relied on lab-/clinic-based measures of BP or a personal medical history of hypertension (e.g., [19, 27]), and this work has generally found that caregiving is associated with higher blood pressure. However, empirical research suggests that these measurements are not always accurate (e.g., white coat effect/white coat hypertension, masked hypertension) [28]. Further, numerous studies have found that ambulatory BP (ABP) measurements—BP measurements taken at regular intervals throughout a typical day—are more strongly associated with cardiovascular morbidity and mortality than are clinic-based measurements. Interestingly, ABP has been found to predict cardiovascular outcomes even when controlling for the explanatory power of clinic-based BP [29–33]. The increased prognostic value of ABP relative to clinic/lab-based BP may be due to its higher reliability (i.e., derived from a greater number of within-person readings) and ecological validity [34]. Because measurements are taken repeatedly over a longer period (generally over many hours), ambulatory assessments also offer the opportunity to capture the effects of moment-to-moment psychosocial and behavioral factors on an individual’s BP [35].

Research on caregivers that has made use of ABP measurement has not consistently taken full advantage of the longitudinal, within-person nature of the data; studies on the relationship between psychosocial experiences and BP often collapse BP measurements into averages (e.g., over 24 hr, daytime, during a specific task) that are used as the primary outcome (e.g., [36, 37]). As a result, these studies largely focus on between-person associations (i.e., effects that are measured or compared across a set of individuals). However, within-person associations (i.e., effects measured or compared across time for a given individual) are also informative for evaluating psychological theories and testing causal hypotheses, and effects that exist at the between-person level do not necessarily resemble those at the within-person level [38, 39]. Therefore, additional research that specifically examines the within-person effects of psychosocial experiences on BP is needed.

One potential contributing factor to a person’s BP is their social relationships; numerous studies have found support for the idea that relationship and communication factors can influence BP. In particular, relationships that are more supportive and intimate—i.e., relationships with more disclosure of thoughts and feelings between partners and high perceived responsiveness to disclosure [40, 41]—have consistently been associated with lower BP, whereas less social support and more relationship conflict have been associated with increased BP [42–45]. Similarly, in experimental work, providing social support (compared with a control task) has been linked to lower BP during and after a stressor [46]. Note that most of these studies have focused on between-person relationships. A smaller literature has investigated the within-person effects of momentary or daily interpersonal processes on BP (e.g., [47]), but to the authors’ knowledge none have done so in cancer caregivers.

Present Study

It is important to investigate the relationships between caregivers’ psychosocial experiences and BP at the within-person level to fully understand the health effects of caregiving for patients with advanced cancer. Additionally, much of the past work in this area has been conducted in a laboratory setting, raising questions regarding external validity and the applicability of findings to caregivers’ daily lives or clinical care. The present study aimed to address these methodological limitations. Our primary goal was to examine the within-person effects of relationship and communication variables on ambulatory BP in spouse–caregivers of advanced cancer patients, measured during a typical day at home with the patient.

First, we expected that the caregiver’s perceptions of their relationship to the patient (i.e., relationship role as caregiver vs. spouse) would influence their BP. Specifically, we hypothesized that when a caregiver reports feeling more like a caregiver (vs. more like a spouse), they would have higher BP (H1). We also expected that a more open and supportive relationship between caregiver and patient would be associated with lower BP. Specifically, when the caregiver perceived more self-disclosure from the patient (H2), when the caregiver self-disclosed more to the patient (H3), and when the caregiver perceived the patient as being more open and responsive to the caregiver’s self-disclosure (H4), the caregiver would experience lower BP.

Methods

The current study is a secondary analysis of data from a larger observational study of patients with advanced cancer and their spouse–caregivers. For additional details about recruitment and data collection procedures, see [48] and [49]. All study procedures were conducted in accordance with ethical standards and were approved by the Institutional Review Board, and informed consent was obtained from all participants included in the study.

Participants

Participants in the current study were spouse–caregivers of patients with advanced cancer—specifically, stage III or IV pancreatic, esophageal, gastric, gallbladder, colorectal, hepatocellular, bile duct, or non-small cell lung cancer. Caregivers were eligible to participate if they (i) were the patient’s spouse or romantic partner, (ii) cohabited with the patient, (iii) self-identified as a caregiver to the patient, (iv) were at least 18 years old, and (v) were able to communicate in English.

Procedure

Caregivers were approached for participation during one of the patient’s regularly scheduled appointments in the gastrointestinal or thoracic oncology clinics at Moffitt Cancer Center. After providing informed consent, caregivers completed a baseline questionnaire that included demographic information. Dyads were scheduled to complete the observational portion of the study in their home on a day that the patient and caregiver were home most of the day; participation typically started in the late morning (between approximately 10:30 AM and noon) and ended about 10 hr later (median = 9.78 hr; range = 1.35 – 16.00 hr). As part of this observation, caregivers were equipped with an ambulatory BP monitor (SunTech Medical^® Oscar 2™). Following procedures developed in previous research [50], the BP monitor was programmed to initiate a BP reading at random intervals during this period, averaging 20 min apart. The caregiver was asked to complete a brief self-report questionnaire (diary) immediately after each reading.

Diary Measures

Primary predictors: relationship and communication variables.

Four relationship and communication variables from the diary were examined as focal predictors. Participants were asked several questions with regard to their interactions with the patient since the previous BP reading. On a Likert-type scale ranging from 1 (very little) to 5 (a great deal), caregivers reported how much they had self-disclosed to the patient (caregiver self-disclosure) and how much they perceived the patient had disclosed to them (patient self-disclosure) since the previous BP reading, and the degree to which they felt understood, validated, and accepted by the patient (perceived responsiveness) since the previous reading [50]. Finally, caregivers reported on their current perceptions of their role in relation to the patient (role perceptions)—that is, the degree to which they felt more like a spouse versus more like a caregiver (at the time of the questionnaire). Responses were given on a Likert-type scale ranging from 1 to 5, with a 1 indicating that they felt completely like a spouse, a 5 indicating that they felt completely like a caregiver, and a 3 indicating that they felt equally like a spouse and a caregiver.

Control variables: physical circumstances and mood.

Caregivers reported on their physical circumstances at the time of each BP reading, including posture (i.e., lying, sitting, or standing), activity level (i.e., not active, a little active, moderately active, or strenuously active), temperature (i.e., too cold, comfortable, or too hot), location (i.e., at home, at work, in transit, or other) and talking (i.e., not talking, dealing with a hassle, interacting with spouse, or other talk). They also reported whether (i.e., yes/no) they had exercised since the previous reading and whether (i.e., yes/no) they had consumed nicotine, caffeine, alcohol, or a meal since the previous reading. Additionally, caregivers reported momentary negative affect using five modified items drawn from the Positive and Negative Affect Schedule (PANAS) [51]; responses were given on a Likert-type scale ranging from 1 (not at all) to 5 (very much), with higher scores indicative of greater levels of negative affect. These items all have established associations with ABP (e.g., [52–56]), and as such, were considered for inclusion in analyses as control variables.

Data Preparation and Analysis

Data preparation.

Before conducting analyses, systolic and diastolic BP data were cleaned. Values for which the BP monitor also recorded an error code during measurement and impossible/implausible BP values (per prior research, defined as < 70 and > 250 for systolic and < 45 and > 150 for diastolic) were removed from the data set and treated as missing values [28, 52, 57].

Within-person bivariate correlations between all potential control variables and systolic and diastolic BP were first examined. To increase model parsimony and avoid model non-convergence, control variables were only entered as time-varying covariates in the main model if the zero-order within-person correlation with BP exceeded r > .10.

Data analysis.

To test the focal hypotheses, multilevel structural equation modeling was conducted in Mplus Version 8 [58]. Multilevel modeling was used to accommodate dependency within each participant’s repeated measures and allowed for examination of within- and between-person effects of relationship and communication variables on BP. Structural equation modeling allowed for the simultaneous modeling of both systolic and diastolic BP outcomes. Bayesian estimation with non-informative priors was used to facilitate both the estimation of lagged effects and overall model convergence, given substantial model complexity and current sample size [58, 59]. Because non-informative priors are used, the results obtained have similar substantive interpretations to those obtained from typical maximum likelihood-based approaches. The use of Bayesian estimation in Mplus allowed for use of latent (vs. observed) centering of lagged predictors. This is the preferred approach, as it substantively reduces the bias introduced when traditional observed centering is used (Nickel’s bias; see [60]).

The Mplus Bayesian approach relies on Markov chain Monte Carlo with Gibbs sampling and, as noted above, can be expected to result in estimates similar to those from maximum likelihood estimation. Ten thousand iterations were used to achieve stable results. As with maximum likelihood estimation, this approach provides valid inferences assuming data are missing at random (a reasonable assumption in the case of longitudinal data) [61]. To determine statistical significance consistent with the maximum likelihood interpretative framework, the Bayesian approach provides 95% credibility intervals (CIs) for parameter estimates. CIs that do not include zero are statistically significant at the p < .05 level. The potential scale reduction factor (PSR) was examined as an indicator of model fit, with values < 1.1 indicating adequate fit and convergence [62, 63].

Latent separation of within- and between-person variability in the psychosocial predictor variables and BP outcomes was used to examine pure within- and between-person effects [64]. Physical circumstances (specifically posture and activity) and mood (negative affect)—assessed via self-report concurrently with the relationship and communication predictor variables—were included as time-varying covariates. Time in minutes was also included as a time-varying within-person covariate (additional explanation follows below). These four time-varying covariates (posture, activity, negative affect, and time) were person-mean centered and only included in the within-person portion of the model. Random intercepts for diastolic and systolic BP were estimated to allow for person-to-person variability in average levels over the monitoring period. Unfortunately, random slope effects for the focal predictors were unable to be estimated due to model convergence problems.

The first multilevel structural equation model examined concurrent effects—that is, the effects of the relationship and communication predictors and time-varying covariates (posture, activity, negative affect, time) measured at time point t on the systolic and diastolic BP readings obtained at the same time point t. In this model, time was represented as minutes elapsed since the first reading of the monitoring period (with zero representing each caregiver’s first reading). The between-person effects of the relationship and communication predictors on systolic and diastolic BP were also estimated. These between-person effects were examined to determine whether caregivers’ average levels of role perceptions, caregiver and patient self-disclosure, and patient responsiveness over the course of the monitoring period were linked to higher levels of average BP during the same period. Note that time-varying covariates are not included in the between-person part of the model.

The second model examined time-lagged effects—that is, the effects of the relationship and communication predictors measured at time t - 1 on the systolic and diastolic reading obtained at the following time point t. In this model, the time variable represented minutes elapsed since the previous BP reading (t - (t - 1)), with zero representing the average time between consecutive readings for each caregiver. This approach was used to account for the unequal intervals between BP readings and the possibility that lagged effects of relationship and communication variables on BP may differ systematically by the length of lag between consecutive time points [65]. Lagged (by t - 1) versions of all relationship, communication, and BP variables were created within the Mplus model (using the LAGGED command). Time-varying covariates (posture, activity, negative affect, time) remained unlagged at time t. In the within-person part of the model, systolic and diastolic BP at time t were regressed on the relationship and communication predictors at time t - 1 and the time-varying covariates at time t. In addition, to control for autoregressive (AR(1)) effects, each relationship and communication predictor and BP outcome at time t was regressed on itself at time t - 1. For parsimony, at the between-person level, only random intercepts and covariances between average levels of the relationship and communication variables and average BP were estimated (rather than regressive effects, which are estimated in the concurrent model described above).

Results

Sample Characteristics

A total of 2127 observations from 81 spouse–caregivers were included in analyses. An observation was defined as one BP reading plus its associated diary data, or one BP reading with no associated diary data (e.g., if the caregiver forgot to complete the diary after a cuff inflation). See Table 1 for sample demographic characteristics and Table 2 for descriptive statistics for key variables used in analyses.

Table 1.

Caregiver Demographic Characteristics (N = 81)

	M	SD
Age (years)	66.90	9.18
Relationship length (years)	34.63	15.65
	n	%
Gender
Male	22	27.2
Female	59	72.8
Ethnicity
Hispanic/Latinx	3	3.7
Non-Hispanic/Latinx	78	96.3
Race
American Indian/Alaska native	1	1.2
Black/African-American	5	6.2
White/Caucasian	75	92.6
Marital status
Unmarried	1	1.2
Married	80	98.8
Education
7–11 years	3	3.7
High school graduate or equivalent	11	13.6
Some college or vocational school	25	30.9
College graduate (4 years)	15	18.5
Some graduate or professional school	7	8.6
Graduate or professional degree	20	24.7
Employment status
Not currently employed	64	79.0
Employed part-time	2	2.5
Employed full-time	14	17.3
Missing	1	1.2
Annual household income
$10,000–$24,999	5	6.2
$25,000–$39,999	10	12.3
$40,000–$49,999	8	9.9
$50,000–$74,999	26	32.1
$75,000 or more	30	37.0
Missing	2	2.5

Table 2.

Descriptive Statistics for Key Study Variables (N = 2127 observations)

	n	Grand M	SD	Min	Max
Systolic BP	1878	143.99	23.77	56	231
Diastolic BP	1878	78.90	16.73	30	200
Patient self-disclosure	742	2.53	1.26	1	5
Caregiver self-disclosure	743	2.39	1.19	1	5
Perceived responsiveness	752	3.62	1.01	1	5
Role perceptions*	1342	1.78	1.25	1	5

Note. *Higher scores indicate feeling more like a caregiver than a spouse.

Concurrent Effects

Within-person effects.

Concurrent, within-person effects were modeled to evaluate the impact of relationship and communication variables reported on at time t on BP measured at the same time point (t). Feeling more like a caregiver than a spouse was significantly associated with lower systolic BP at the same time point (B = −1.70, 95% CI = −3.03, −0.37), controlling for posture, activity, time, and negative affect. This indicates that when participants reported a one-unit change toward feeling more like a caregiver (and feeling less like a spouse) compared with what is typical for them, they had, on average, almost a two-unit lower systolic BP reading at that same time. Role perceptions were not related to diastolic BP (B = −0.47, 95% CI = −1.55, 0.61). Caregiver self-disclosure to the patient since the previous BP reading was associated with neither systolic (B = −0.96, 95% CI = −2.99, 1.19) nor diastolic BP (B = −1.78, 95% CI = −3.46, 0.32). Similarly, perceived responsiveness since the previous BP reading was associated with neither systolic (B = −1.30, 95% CI = −3.24, 0.67) nor diastolic BP (B = −1.18, 95% CI = −2.83, 0.50). Patient disclosure to the caregiver since the previous BP reading was unrelated to systolic BP (B = 0.63, 95% CI = −1.28, 2.45). However, patient disclosure was significantly associated with higher concurrent diastolic BP (B = 1.97, 95% CI = 0.01, 3.52), indicating that when a caregiver reported that the patient disclosed one-unit more than what is typical for him/her, the caregiver had a two-unit higher diastolic BP reading in that same period.

Between-person effects.

Concurrent between-person effects were modeled to evaluate the impact of average levels of relationship and communication variables over the study period on average BP over the same period. In this between-person part of the model, average caregiver role perceptions were neither related to average systolic (B = −2.45, 95% CI = −6.38, 1.44) nor diastolic (B = −0.27, 95% CI = −2.62, 2.10) BP. Interestingly, caregivers who reported more self-disclosure to the patient on average over the monitoring period also had significantly higher systolic BP over the same period (B = 14.08, 95% CI = 2.91, 24.56), with no significant differences in their average diastolic BP (B = 6.70, 95% CI = −0.53, 13.09). Average patient self-disclosure to the caregiver over the monitoring period was not significantly predictive of average systolic (B = −2.13, 95% CI = −12.59, 9.30) or diastolic (B = −0.95, 95% CI = −7.76, 6.40) BP. Average perceived responsiveness was also not significantly related to average systolic (B = −0.56, 95% CI = −6.82, 5.65) or diastolic (B = 0.55, 95% CI = −3.19, 4.23) BP.

Lagged Effects

Within-person effects.

Lagged within-person effects were modeled to evaluate the impact of relationship and communication variables reported on at time t - 1 on BP measured at the following time point (t). Neither caregiver role perceptions, caregiver self-disclosure, spouse self-disclosure, nor responsiveness at the previous time point emerged as statistically significant predictors of systolic of diastolic BP at the following time point (controlling for systolic or diastolic BP, respectively, at the previous time point, as well as posture, activity, and time elapsed since the prior reading). It may be worth noting that the effect of caregiver self-disclosure at t - 1 on diastolic BP at t approached significance (B = 1.53, 95% CI = −0.07, 3.07, two-tailed p = .060), though this should be interpreted cautiously, pending replication.

Conclusions

The aim of this study was to assess the influence of contextual relationship and communication factors on the ambulatory BP of advanced cancer caregivers, with a particular emphasis on within-person associations. Results indicate that a caregiver’s moment-to-moment experience of their relationship and communication with the patient may have an immediate physiological impact on the caregiver’s BP (i.e., concurrent within-person effects). However, this effect may only last a few minutes, as suggested by the absence of significant lagged effects. Additionally, examination of between-person effects revealed that higher average caregiver self-disclosure to the patient was associated with higher average caregiver BP. These findings were somewhat unexpected and do not support our hypotheses.

Our study found that, in moments when a caregiver felt more like a caregiver than a spouse, they had lower systolic BP. This finding was somewhat surprising, given that caregiving often induces psychological stress, which has a known within-person relationship with increased blood pressure [66]. It is possible that feeling more like a caregiver allows caregivers to shift their attention away from worries about their spouse’s cancer or concerns about the future and toward the actual caregiving tasks that need to be accomplished. This shift in attention may simply help to distract the caregiver from negative thoughts and feelings, but the refocusing on immediate, tangible, and addressable problems may also allow the caregiver to feel more in control of the situation.

The finding that perceiving more patient disclosure was linked with higher diastolic BP was also somewhat unexpected. Self-disclosure within romantic relationships has previously been linked to positive outcomes (e.g., increased feelings of intimacy) for both the discloser and the recipient of the disclosure [41]. Some previous work has found that listening to the stressful experiences of others can increase the listener’s negative affect [67] as well as heart rate and skin response [68]. This is consistent with other work showing that the momentary experience of empathy for another indeed involves sharing in their distress or pain [69]. However, other work focused on marital relationships specifically found that self-disclosure is related to positive outcomes for the recipient of the disclosure [70]. There are other contextual factors that may affect how beneficial self-disclosure is for both parties that were not measured in the present study, including the valence of the patient’s disclosure and their reasons for disclosing to the caregiver; it may be, for example, that patients tended to self-disclose more negative thoughts and feelings, which in turn could be stressful for caregivers to hear and respond to, thus raising their BP. It is also possible that more patient self-disclosure is simply reflective of greater underlying needs or reasons for the patient to disclose (e.g., more stress, worse physical symptoms), and this greater underlying stress is what was driving the observed positive relationship between patient disclosure and caregiver BP. This may also help to explain the finding that caregivers who self-disclosed more on average over the study period tended to have higher systolic BP.

Limitations and Future Directions

Results should be interpreted in light of several limitations. First, BP data for each caregiver were collected over the course of a single day. As noted earlier, this was a day when the patient and caregiver expected to be home together for most of the day to ensure that patient–caregiver interactions would occur during data collection. However, this may not have been a “typical” day for all patients or caregivers, and it is possible that the effects observed in the present study could be different on days when the patient and caregiver do not spend the day together, or for dyads who do not spend that much time together normally. Similarly, effects could differ based on the patient’s needs or the amount of caregiving demands that day. It is also possible that the impact of relationship and communication factors on caregiver BP could change over longer periods of time (e.g., soon after diagnosis vs. approaching end of life).

Additionally, our sample was relatively homogeneous in several ways, including in terms of race, ethnicity, income, and education. Most participants were not currently employed (likely due to retirement or medical/family leave), so it is likely that most dyads spent a substantial of time together on a day-to-day basis, which could have influenced outcomes in the current study. It is unclear how generalizable these findings would be to advanced cancer caregivers who continue to work full- or part-time after the patient’s diagnosis or those belonging to other racial/ethnic or socioeconomic groups. The caregivers in this study were, on average, in their mid-60s and had been in a relationship with the patients for nearly 35 years. This too raises questions about generalizability, as the relationship and communication dynamics of older couples in such long-term relationships may look very different from the dynamics of younger individuals or those in newer relationships. It also remains unclear whether these findings would generalize to other types of patient–caregiver relationships (e.g., adult child caregivers, parent caregivers).

Future research should examine possible cognitive and behavioral mechanisms of these effects. For example, as noted earlier, feeling more like a caregiver may lower stress and BP by allowing caregivers to focus more on the practical tasks of caregiving, whereas feeling more like a spouse may increase stress and BP by causing caregivers to focus more on their spouses’ potential suffering. Additionally, patient disclosure may serve to increase the caregiver’s awareness of the patient’s stress, thus increasing the caregiver’s stress and BP. Future work should also explore longer-term effects of caregivers’ role perceptions and patients’ self-disclosure on caregivers’ psychological and physical health, as well as potential downstream effects on the patients’ care and well-being.

Compliance with Ethical Standards

Conflict of Interest: The authors declare that they have no conflict of interest.

Ethical Approval All procedures performed in studies involving human participants were in accordance with the ethical standards of the institutional and/or national research committee and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards.

Informed Consent: Informed consent was obtained from all individual participants included in the study.