SEARCH

SEARCH BY CITATION

Keywords:

  • quality of life;
  • end-of-life care;
  • bereavement;
  • depression;
  • regrets

Abstract

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. FUNDING SUPPORT
  8. CONFLICT OF INTEREST DISCLOSURES
  9. REFERENCES
  10. Supporting Information

BACKGROUND

The objective of the current study was to determine the best set of predictors of psychological disorders, regrets, health-related quality of life, and mental health function among bereaved caregivers of patients with cancer, thereby identifying promising targets for interventions to improve bereavement adjustment.

METHODS

Coping with Cancer is a longitudinal study of patients with advanced cancer and their informal caregivers who were enrolled from 2002 to 2008. The main outcome measure was bereavement adjustment of 245 caregivers (eg, depression, anxiety, and regrets) 6 months after the loss of the patient. The Structured Clinical Interview of the Diagnostic and Statistical Manual of Mental Disorders determined whether caregivers met the criteria for major depressive disorder or an anxiety disorder. Changes in health-related quality of life and mental health function from baseline to after the patient's death were assessed with the Medical Outcomes Study Short Form (SF-36).

RESULTS

Greater than 50% of the caregivers reported regret about the cancer patient's end-of-life care; better patient quality of death (adjusted odds ratio, 0.77; 95% confidence interval, 0.67-0.88) reduced the risk of bereavement regret. The incidence of major depressive disorder or anxiety among the bereaved caregivers was 12.6% and was less likely for caregivers with better mental health before the loss of the patient (adjusted odds ratio, 0.03; 95% confidence interval, 0.004-0.25). Better patient quality of death also predicted improved caregiver health-related quality of life (adjusted standardized beta, .28; P < .001). The completion of a do-not-resuscitate order was found to be predictive of improved mental health from before the death of the patient to after the death (adjusted standardized beta, .29; P < .001).

CONCLUSIONS

Reducing caregiver distress, encouraging advance care planning by patients, and improving patients' quality of death appear to be promising targets of interventions to improve caregiver bereavement adjustment. Cancer 2014;120:918–925. © 2013 American Cancer Society.


INTRODUCTION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. FUNDING SUPPORT
  8. CONFLICT OF INTEREST DISCLOSURES
  9. REFERENCES
  10. Supporting Information

Bereaved family caregivers are at an elevated risk of psychological distress.[1-4] Prior research has focused on the suddenness of death and intrapersonal and interpersonal factors influencing bereavement adjustment, including feelings of strain, amount of social support, and benefit derived from the act of caregiving.[4, 5] To the best of our knowledge, few studies to date have examined the end-of-life (EOL) experience of patients with advanced cancer to determine what potentially modifiable factors most influence the mental health of bereaved caregivers. Knowing the factors that best predict bereavement mental health among caregivers of patients with advanced cancer will identify targets for interventions to improve surviving caregivers' quality of life.

To the best of our knowledge, to date, research on the bereaved caregivers of patients with cancer has shown that a caregiver's mental health before the patient's death,[4, 6] religiousness,[7] preparedness for the patient's death,[8] the patient's quality of life in the last week of life,[9] use of life-prolonging care within the last week,[9] and site of care at the time of death[10] are predictive of a bereaved caregivers' mental health. We know very little else about the other potentially modifiable factors at the EOL of patients with advanced cancer that might affect bereavement adjustment. Among caregivers of patients with a variety of terminal illnesses, a greater dependence on the patient for emotional and material needs predicts worse bereavement adjustment.[11] In addition, a greater burden among caregivers of patients with dementia, which includes the burden of providing emotional and behavioral support to patients, is associated with feelings of both guilt and relief during bereavement.[12, 13]

In contrast, more is known about factors that improve patients' quality of life near death. Quality of life near death is better for patients who use fewer life-prolonging care measures at the EOL, including fewer intensive care unit (ICU) stays,[11, 14] and those who use more hospice and palliative care.[11, 14, 15] It is better for patients who participate in religious prayer and meditation and who receive spiritual support from medical staff.[16] Patient psychosocial characteristics (peacefulness and lack of worry)[14] and therapeutic alliance with health care providers (trust, empathy, and shared goals of care) also predict better global ratings of patients' quality of life near death by their professional and familial caregivers. Given that the patient's quality of life near death predicts the surviving caregivers' bereavement adjustment, factors that directly affect patient quality of life in the patient's last week of life may also indirectly influence caregiver bereavement adjustment.

The current study uses data from the Coping with Cancer (CwC1) study that was designed to examine how the EOL experiences of patients with advanced cancer and their primary informal caregivers affect the surviving caregivers' bereavement adjustment. The objective of the current study was to derive the most parsimonious set of modifiable factors that have the greatest influence on survivors' bereavement adjustment. We hypothesized that mental health prior to the patient's death, emotional dependence on deceased patients, and the patient's quality of death would be among the most important direct predictors of a bereaved caregiver's psychological distress and regrets about the patient's EOL experiences. The effects of receipt of life-prolonging and hospice care in the last week of life, therapeutic alliance between patients and oncologists, spiritual support of patients by the medical community, and the patients' degree of peacefulness and lack of worry on caregiver bereavement adjustment would be mediated through their effect on patients' quality of death.

MATERIALS AND METHODS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. FUNDING SUPPORT
  8. CONFLICT OF INTEREST DISCLOSURES
  9. REFERENCES
  10. Supporting Information

Study Sample

From 2002 through 2008, patients and caregivers were recruited from outpatient clinics in Connecticut, Massachusetts, New Hampshire, New York, and Texas. This study was approved by the Institutional Review Board of each site, and patients and caregivers provided written informed consent. Patients in CwC1 were aged ≥ 20 years and met diagnostic criteria for advanced cancer (including distant metastases and disease that was refractory to first-line chemotherapy) but were well enough to participate in an interview. All patients in CwC1 identified an unpaid or informal caregiver who provided the majority of their care; these caregivers also were included in the study. Patients and caregivers who did not speak English or Spanish or those who had dementia (Short Portable Mental Status Questionnaire score ≥ 6[17]) or delirium were excluded.

Patients and caregivers participated in in-person baseline interviews, and patients were followed from baseline until death (median, 4 months). Caregivers were interviewed 6 months after the patient's death, and additional data were gathered through a postmortem interview 2 to 4 weeks after the patient's death. Postmortem interviews were completed by the clinician or caregiver deemed by study staff to be most knowledgeable about the patient's death.

Of the 361 caregivers who completed the follow-up and baseline interviews, 245 (68%) answered questions about bereavement adjustment. Reasons for nonresponse to bereavement adjustment questions were not collected, but perceived respondent fatigue is one possible reason for the missing responses. Caregivers with and without outcome data were largely similar, but those with outcome data were more likely to be from the Northeast region of the United States (57.14% vs 37.07%; P < .001) and to have cared for a patient who died in hospice (13.47% vs 4.31%; P = .008) compared with caregivers without outcome data. Caregivers with outcome data also had slightly worse baseline mental health (0.67 vs 0.73; P = .01) and energy (0.56 vs 0.62; P = .02) scores on the Medical Outcomes Study Short Form (SF-36)[18] than those without outcome data (range, 0-1; higher scores indicate better function).

Measures

Caregiver bereavement adjustment

At baseline and follow-up, the Structured Clinical Interview for the 4th edition of the Diagnostic and Statistical Manual of Mental Disorders (DSM-IV) Axis I Disorders (SCID)[19] was administered to caregivers to determine whether they met diagnostic criteria for major depressive disorder (MDD), generalized anxiety disorder, panic disorder, and posttraumatic stress disorder (PTSD). We measured new onset of depression or anxiety between the baseline and follow-up interviews. At follow-up, caregivers were asked to rate their regret on 3 aspects of a patient's EOL on a scale from 1 (no regrets) to 5 (extremely regretful): patient's manner of death, care provided by the caregiver just before the patient's death, and care provided by health care providers just before the patient's death. Because answers were skewed toward no regrets, we combined the 3 questions into 1 dichotomous indicator of any regret (1 indicates some regrets and 0 indicates no regrets). The third and fourth measures of caregiver bereavement adjustment were taken from the SF-36; changes in caregiver total scores (health-related quality of life) and in the Mental Health subscale from baseline to follow-up were recorded.

Potential predictors of caregivers' bereavement adjustment
Caregiver baseline function

Baseline scores from the SF-36 were used as potential predictors of depression/anxiety incidence and caregiver regrets. Subscales (Self-Rated Health, Physical Function, Social Function, Role Limitation, Mental Health, Pain, Energy, and Health Change From Previous Year; range, 0-1) and weighted (range, 0-36) and unweighted (range, 0-8, higher scores indicate better function) summaries of subscales were considered to be predictors.[18]

Patient's impression of therapeutic alliance

Patients were asked at baseline whether their physician viewed them as a whole person and respected them, and whether they respected and trusted their physician and felt comfortable asking the physician questions about their care.[20] We considered these items individually and in an aggregate measure (a “yes” answer to each item).

Advance care planning

Patients were asked at baseline whether they had discussed their EOL plans with a physician, and whether they had a do-not-resuscitate (DNR) order, living will, or health care proxy.

Patient's mental health at baseline

The SCID assessed patients for MDD, generalized anxiety disorder, panic disorder, and PTSD. Patients were asked how nervous/worried they were in the 2 days before the interview (0 indicates not at all to 10 indicates extremely [McGill Quality of Life Questionnaire])[21] and whether they felt a “deep sense of inner peace or harmony” most days.[22]

Patient's terminal illness acknowledgment

At baseline, patients were asked whether they considered themselves “terminally ill.”[14]

Baseline caregiver religiousness

The importance of religion to the caregiver (very vs somewhat or not very important) and the caregiver's religious coping (Brief RCOPE[23]) and religious service attendance were assessed.

Spiritual coping and care of patient

At baseline, patients were asked whether religious beliefs or activities were their main method of coping with their illness, whether they had received spiritual care services within the clinic or hospital, and whether community-based clergy had visited them.

Life-prolonging care

Receipt of life-prolonging care (chemotherapy, ventilation, feeding tube, or resuscitation) and ICU stays within the last week of life were documented.

Hospice

Any hospice use and length of hospice stay were documented.

Location of death

The location of death was recorded (ICU, hospital [outside of ICU], nursing home, inpatient hospice, or home).

Quality of life in week before death/quality of death

During the postmortem interview, the overall quality of death was assessed with the question “In your opinion, how would you rate the overall quality of the patient's death/last week of life?” (0 indicates worst possible, 10 indicates best possible).

Specific questions regarding distress in the week before death were also asked (“In your opinion, just before the death of the patient [his/her last week, or when you last saw the patient], how would you rate his/her level of ‘psychological distress? and ’… physical distress?” (0 indicates none, 10 indicates extremely distressed).[14] Two aggregate measures were constructed: a sum of answers and the average response to the overall quality and reverse-scored distress questions. These questions were asked of the caregiver, but if the caregiver was uninformed or unavailable, the clinician who last cared for the patient was interviewed. Caregivers' and patients' quality-of-life measurements at baseline were highly correlated,[14] suggesting that caregivers were able to accurately assess patients' quality of death. Because the overall quality of death was rated lower by caregivers than by clinicians (5.72 vs 6.83; P = .01), rater identity was added as a control variable in models using quality-of-death ratings.

Burden of providing emotional support

At baseline, caregivers were asked how much time they spent providing emotional support to the patient (with 1 indicating little or no time to 4 indicating a great deal of time) and how difficult providing this support was (with 1 indicating little or no difficulty to 4 indicating a great deal of difficulty). The answers to these 2 questions were multiplied to create a measure of burden of providing emotional support (range, 1-16, with higher numbers indicating a greater burden).

Caregiver sociodemographic characteristics

Caregiver sociodemographic characteristics included age, length of time spent caregiving, sex, self-reported race/ethnicity (white, black, Asian, Hispanic, other), religious affiliation (Catholic, Protestant, other/none), and interview site (Northeast vs Southwest). We included race and religious affiliation as proxies for cultural attitudes toward death that might impact bereavement adjustment.[24] We also included the caregiver relationship to the patient (spouse vs other).

Statistical Analysis

The goal of the current study was to determine the “best” models (ie, most parsimonious set of predictors that provides the best fit to the data) for the prediction of bereavement adjustment. We constructed 2 parsimonious models for each of the 4 bereavement adjustment outcomes: a theoretical model based on information criterion comparisons and an automated model constructed with least-angle regression to ensure that results obtained in the theoretical model were not due to inherent biases of the authors. Logistic regression was used to model the incidence of meeting the criteria for MDD or an anxiety disorder and regrets after the patient's death, and linear regression was used to model changes in health-related quality of life and mental health. Analyses were conducted using SAS statistical software (version 9.2; SAS Institute Inc, Cary, NC). For each outcome, we compared the fit and similarity of predictors across theoretical and automated models with and without adjustment for caregiver sociodemographic characteristics.

To create theoretical models, variables were separated into the conceptual categories listed above, and regression models of individual variables and combinations of variables within each category were constructed. We selected the best model(s) from each category according to its Akaike information criterion, corrected for small sample size (AICc).[25-27] We then compared the fit of models comprising ≥ 1 of the best categorical submodels. More detail regarding the theoretical model construction is available in the online supporting information.

To construct automated models, we used least-angle regression (SAS PROC GLMSELECT; SAS Institute Inc.).[28, 29] We considered for inclusion all variables with a P value < .20 in unadjusted regression analyses that were not highly collinear. We used the AICc to evaluate candidate models.

RESULTS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. FUNDING SUPPORT
  8. CONFLICT OF INTEREST DISCLOSURES
  9. REFERENCES
  10. Supporting Information

Sample Characteristics

Of the caregivers, 12.6% (27 of 214 caregivers) met criteria for depression or anxiety, and 54.5% (126 of 231 caregivers) experienced at least some regret after a patient's death (Table 1). Caregiver health-related quality of life declined on average from baseline to follow-up (mean change, −0.47; standard deviation [SD], 5.03 [n = 211 caregivers]). Mental health ratings declined for 44.9% of caregivers (mean decrease for caregivers with worsened mental health, −0.15; SD, 0.12 [n = 102 caregivers]).

Table 1. Caregiver Characteristics (n=245)
VariablesNo. (%) or Mean (SD)
  1. Abbreviation: SD, standard deviation.

  2. a

    Percentages may not add to 100% due to missing values.

  3. b

    Percentage of the 214 individuals with nonmissing relationship data.

  4. c

    Distribution among caregivers with nonmissing values for each outcome.

  5. d

    Possible scores of −36 to 36; positive score indicates improved function over time.

  6. e

    Possible scores of −1 to 1; positive score indicates improved function over time.

Sociodemographic data 
Age at baseline interview, y51.96 (13.50) (range, 20-86)
Female sex186 (75.92)
Race/ethnicitya 
White159 (64.90)
Black42 (17.14)
Asian10 (4.08)
Hispanic33 (13.47)
Religious affiliation 
Catholic94 (38.37)
Protestant41 (16.73)
Other96 (39.18)
None14 (5.71)
Northeast interview site140 (57.14)
Length of time caregiving at wave 1, mo29.41 (64.15) (range, 1-600)
Caregiver was patient's spouseb121 (56.5%)
Outcomesc 
Incidence of depression/anxiety disorder at wave 2 (n=214)27 (12.62%)
Any regret following the patient's death (n=231)126 (54.55%)
Change in health-related quality of life from wave 1 to wave 2 (n=211)d−0.47 (5.03) (range −19.75 to 11.50)
Change in mental health rating from wave 1 to wave 2 (n=227)e0.00 (0.18) (range, −0.44 to 0.60)
Change in mental health rating from wave 1 to wave 2 among those with worsened function over time (n=102)e−0.15 (0.12) (range −0.44 to 0.00)

Adjusted Analyses: Best Models for Each Outcome

Depression/anxiety onset

Both the best theoretical and automated models for depression/anxiety onset included only baseline caregiver mental health. Better baseline mental health predicted a lower incidence of depression/anxiety after controlling for sociodemographic characteristics (adjusted odds ratio [AOR], 0.03; 95% confidence interval [95% CI], 0.004-0.25) (Table 2).

Table 2. Best Theoretical and Best Automated Models for Bereavement Adjustment Outcomes, With and Without Adjustment for Sociodemographic Characteristicsa
 TheoreticalAutomated
VariableUnadjusted OR (95% CI)PAdjusted OR (95% CI)PUnadjusted OR (95% CI)PAdjusted OR (95% CI)P
  1. Abbreviations: 95% CI, 95% confidence interval; AICc, Akaike information criterion with a second-order correction for small sample size; EOL, end of life; OR, odds ratio; SF-36, Medical Outcomes Study Short Form.

  2. a

    Sociodemographic characteristics of caregivers included sex (except for incidence of depression/anxiety, in which only females received new diagnoses), age at baseline interview, white versus nonwhite race, religious affiliation (Catholic, Protestant, other/none), length of time caregiving in months, and Northeast vs Southwest clinic site.

  3. b

    Range of 0 to 1, with a higher score indicating better function.

  4. c

    Range of 0 to 30, sum of answers to questions about quality of life in the week before death. Higher scores indicate greater quality of life.

  5. d

    Range of 0 to 10, average of answers to question about quality of life in the week before death. Higher scores indicate greater quality of life.

Incidence of Depression/Anxiety (n=201)
SF-36 Mental Healthb0.03 (0.004-0.20)<.0010.03 (0.004-0.25).0010.03 (0.004-0.20)<.0010.03 (0.004-0.25).001
AICc137.36   137.36   
Any Regret After Patient's Death (n=190)
Patient had EOL discussion with physician2.10 (1.06-4.19).031.82 (0.89-3.75).10 
Quality of death: sumc0.93 (0.89-0.97)<.0010.92 (0.88-0.96)<.001
Quality of death: average scored0.79 (0.69-0.90)<.0010.77 (0.67-0.88)<.001
SF-36 Pain scoreb0.17 (0.04-0.77).020.09 (0.02-0.50).006
AICc227.98   233.91   
Any regret

The best theoretical model for regret had a slightly better fit than the best automated model (AICc, 227.98 vs 233.91) (Table 2). After adjusting for sociodemographic characteristics, a higher average quality of death (AOR, 0.77; 95% CI, 0.67-0.88) and less caregiver pain (AOR, 0.09; 95% CI, 0.02-0.50) were found to predict a lower likelihood of regret. The best automated model also indicated that a better quality of death predicted less regret. Relationships between quality of death and regret were not sensitive to quality of death rater identity (data not shown).

Change in health-related quality of life

Change in health-related quality of life was better explained by the theoretical model (AICc, 579.14) than the automated model (AICc, 585.49) (Table 3). After adjusting for sociodemographic characteristics, improved health-related quality of life was predicted by a higher overall quality of death (adjusted standardized beta, .28; P < .001) and caregiver baseline depression/anxiety. However, the relationships with depression and anxiety disappeared after including the burden of providing emotional support and interaction terms between burden of providing emotional support and patient PTSD and caregiver depression/anxiety (data not shown). After adjustment for these confounders, only better quality of death was found to be predictive of improved health-related quality of life. The automated model also indicated that quality of death predicted health-related quality of life. This relationship persisted after controlling for rater identity (data not shown).

Table 3. Best Theoretical and Best Automated Model for Change in Health-Related Quality of Life and Change in Mental Health, With and Without Adjustment for Sociodemographic Characteristicsa
 TheoreticalAutomated
VariableUnadjusted βbPAdjusted βPUnadjusted βPAdjusted βP
  1. Abbreviations: AICc, Akaike information criterion with a second-order correction for small sample size; DNR, do not resuscitate; EOL, end of life; PTSD, posttraumatic stress disorder.

  2. a

    Sociodemographic characteristics of caregivers included sex, age at baseline interview, white versus nonwhite race, religious affiliation (Catholic, Protestant, other/none), length of time caregiving in months, and Northeast vs Southwest clinic site.

  3. b

    Beta coefficients are standardized to have a variance of 1.

  4. c

    Range of 0 to 10, with higher scores indicating greater quality.

Change in Health-Related Quality of Life From Wave 1 to Wave 2 (n=181)
Patient had PTSD.15.04.14.07
Caregiver with any assessed mood or anxiety disorder.16.03.15.04
Overall quality of deathc.29<.001.28<.001.26<.001.25<.001
AICc579.14   585.49   
Change in Mental Health From Wave 1 to Wave 2 (n=180)
Patient had EOL discussion with physician−.13.09−.14.07
Patient had DNR order.30<.001.29<.001.23.001.22.005
Patient had panic disorder.19.009.19.01
Patient had PTSD.17.02.15.05
AICc−627.17   −618.82   
Change in mental health rating

Again, the theoretical model for change in mental health fit better than the automated model (AICc, −625.17 to −618.82) (Table 3). After adjusting for sociodemographic characteristics in the theoretical model, patient DNR order (adjusted standardized beta, .29; P < .001) and patient panic disorder (adjusted standardized beta, .19; P = .01) were found to predict improved mental health. As in the health-related quality-of-life model, the effect of patient panic disorder disappeared after accounting for burden of providing emotional support (data not shown). In the automated model, patient DNR completion also predicted improved mental health.

Mediation analysis

We explored whether quality of death would mediate the influence of patient and EOL care characteristics (life-prolonging care, patient spiritual support or worry, site of death) on caregiver bereavement adjustment. The basic assumptions of mediation did not hold[30]; the factors that predicted quality of death in previous work with these data[14] were not found to be significantly associated with caregiver bereavement adjustment (see online supporting information).

Sensitivity analyses

Caregiver relationship was missing for 13% of the sample in the current study, but results were robust to relationship in the subset of caregivers with nonmissing data. Results were also robust to the number of months between caregiver baseline and postbereavement interviews (mean, 13.8; SD, 8.7).

DISCUSSION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. FUNDING SUPPORT
  8. CONFLICT OF INTEREST DISCLOSURES
  9. REFERENCES
  10. Supporting Information

We constructed parsimonious models that predict caregiver bereavement depression/anxiety onset, regret, and changes in health-related quality-of-life and mental health ratings. Better quality of death, better caregiver mental health prior to the patient's death, less caregiver pain, and DNR order completion were found to predict improved caregiver bereavement adjustment; these are modifiable targets for interventions to reduce bereavement distress and regret among caregivers of patients with advanced cancer.

Worse quality of death predicted worsened caregiver health-related quality of life and a greater incidence of caregiver regrets after the patient's death. Factors that directly influence quality of death (eg, life-prolonging care)[14] were not found to be associated with caregiver bereavement adjustment. This suggests that perceived patient suffering is more important than actual EOL care received in terms of bereaved caregivers' mental health. Caregivers value knowing what to expect as a patient's death approaches[31]; educating caregivers on the dying process and on different care options may improve perceptions of care and bereavement adjustment.

Worse mental health prior to the patient's death predicted the onset of depression/anxiety after death. Although the effect of prior mental health on bereavement mental health has been demonstrated in caregivers of patients with a variety of physical illnesses,[2] others have observed more postbereavement depression among spousal caregivers without previous dysphoria.[3] In contrast, health-related quality of life improved for caregivers with baseline depression or anxiety, and mental health ratings improved for caregivers of patients with panic disorder. However, these effects disappeared after controlling for the burden of providing emotional support.[13] Caregiver distress before the patient's death appears to be a prodrome for the onset of depression/anxiety after bereavement. Consequently, the early detection and treatment of caregiver distress may reduce bereavement-related morbidity.

Worse baseline caregiver pain was found to be predictive of a greater incidence of regret after the patient's death. Caregivers with worse pain may have had a better understanding of the pain patients experienced, or they may have tried to balance the patient's care needs with their own pain management, leaving them unable to provide all the care they wanted. Either of these mechanisms may lead to regret. Both poor caregiver physical health and feelings of “underinvestment” in caregiving are associated with worse mental health in caregivers who are currently caring for patients.[32, 33] Relationships between caregiver pain and postbereavement regret should be explored in future analyses.

Finally, DNR order completion predicted improved mental health ratings from baseline to follow-up. Advance care planning is associated with the receipt of care concordant with a patient's wishes34; understanding a patient's goals of care may improve caregiver bereavement adjustment. In addition, DNR orders result in the use of less life-prolonging care at the EOL, which predicts higher quality-of-death ratings.[34] Educating patients and caregivers about situations in which resuscitation is unlikely to be successful[35] and encouraging patients to engage in advance care planning appear to be promising ways to improve both patients' quality of death and caregivers' bereavement adjustment.[36]

Bereavement adjustment begins before a patient's death.[13] The current study identified modifiable targets for caregiver bereavement mental health interventions before a patient dies, an area of psychosocial treatment that to the best of our knowledge has received relatively little attention.[37] Interventions to reduce caregiver distress and pain prior to the patient's death and encourage advance care planning by the patient may help the caregivers of patients with advanced cancer psychologically prepare for the death of a loved one. In addition, interventions to improve the quality of death would help caregivers adjust to bereavement.

Although data collection for this study ended in 2008, the CwC1 data are unique in that they include comprehensive psychosocial assessments of patients and caregivers before the patient's death, a postmortem assessment that had the caregiver evaluate the quality of death, and an assessment of bereaved caregivers 6 months after the loss of the patient. These data were designed to provide novel insights into how aspects of EOL care prospectively affect surviving caregivers' bereavement adjustment. Moreover, the use of life-prolonging measures such as mechanical ventilation at the EOL remains a serious concern,[38] highlighting the relevance of the current study results to current disconcerting practice patterns in the United States (eg, the Choosing Wisely campaign of the ABIM [American Board of Internal Medicine] Foundation, which focuses on the concept “less is more”).

Although the factors we identified were robust to the model construction process, these results need to be replicated in samples not used to create the models. Although a large percentage of the patients in the current study received some hospice care, the receipt of hospice care was not associated with bereavement adjustment in the models, suggesting that the current study results are generalizable to patients receiving either life-prolonging or hospice care. Varying response rates to questions limited our ability to include some variables, such as emotional dependence on patients, in our models. In addition, we did not have measures of emotional or instrumental support received by caregivers in the last weeks of the patients' lives.

Conclusions

Psychological distress and regrets are common among caregivers after the death of a loved one from cancer. In the current study, we identified modifiable targets to ease bereavement adjustment, including reducing caregiver distress and pain, engaging in advance care planning for the patient, and improving the patient's quality of death. Interventions focused on improving the EOL experience for both patients with advanced cancer and their caregivers will reap additional benefits to the surviving caregiver's quality of life.

FUNDING SUPPORT

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. FUNDING SUPPORT
  8. CONFLICT OF INTEREST DISCLOSURES
  9. REFERENCES
  10. Supporting Information

Supported in part by the following grants to Dr. Prigerson: MH63892 from the National Institute of Mental Health, CA106370 and CA156732 from the National Cancer Institute, and the Center for Psychosocial Epidemiology and Outcomes Research at the Dana-Farber Cancer Institute.

CONFLICT OF INTEREST DISCLOSURES

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. FUNDING SUPPORT
  8. CONFLICT OF INTEREST DISCLOSURES
  9. REFERENCES
  10. Supporting Information

Dr. Garrido was supported by CDA 11-201/CDP 12-255, Department of Veterans Affairs, Veterans Health Administration, Office of Research and Development, Health Services Research and Development Service for work related to the current study.

REFERENCES

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. FUNDING SUPPORT
  8. CONFLICT OF INTEREST DISCLOSURES
  9. REFERENCES
  10. Supporting Information

Supporting Information

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. FUNDING SUPPORT
  8. CONFLICT OF INTEREST DISCLOSURES
  9. REFERENCES
  10. Supporting Information

Additional Supporting Information may be found in the online version of this article.

FilenameFormatSizeDescription
cncr28495-sup-0001-suppinfo.pdf203KSupporting Information

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.