|Home | About | Journals | Submit | Contact Us | Français|
Using available communication technologies, clinicians may offer timely support to family caregivers in managing symptoms in patients with advanced cancer at home.
To assess the effects of an online symptom reporting system on caregiver preparedness, physical burden, and negative mood.
A pooled analysis of two randomized trials (NCT00214162 and NCT00365963) was conducted to compare caregiver outcomes at 6 and 12 months after intervention between two randomized, unblinded groups using General Linear Mixed Modeling. Caregivers in one group (Comprehensive Health Enhancement Support System-Only) were given access to an interactive cancer communication system, the Comprehensive Health Enhancement Support System. Those in the other group (Comprehensive Health Enhancement Support System + Clinician Report) received access to Comprehensive Health Enhancement Support System plus an online symptom reporting system called the Clinician Report. Clinicians of patients in the Comprehensive Health Enhancement Support System + Clinician Report group received e-mail alerts notifying them when a symptom distress was reported over a predetermined threshold.
Dyads (n=235) of advanced-stage lung, breast, and prostate cancer patients and their adult caregivers were recruited at five outpatient oncology clinics in the United States.
Caregivers in the Comprehensive Health Enhancement Support System + Clinician Report group reported less negative mood than those in the Comprehensive Health Enhancement Support System-Only group at both 6 months (p=0.009) and 12 months (p=0.004). Groups were not significantly different on caregiver preparedness or physical burden at either time point.
This study provides new evidence that by using an online symptom reporting system, caregivers may experience less emotional distress due to the Clinician Report’s timely communication of caregiving needs in symptom management to clinicians.
In the United States, an estimated 4.6 million people served as family caregivers for cancer patients in 2009.1 Caregivers, or carers, are family or friends who offer unpaid support to patients. As advanced cancer treatments prolong patients’ lives, symptom management becomes the major responsibility of family caregivers.2,3 Cancer symptoms such as pain and fatigue are significantly associated with caregiver physical burden (e.g., sleep disorders and fatigue) and emotional problems (e.g., anxiety and depression).4–6 Physical and emotional challenges are common among cancer caregivers.7 As patient symptoms worsen, caregivers may experience more burdens and less confidence to manage patients’ care.8,9 Prepared and supported caregivers may cope better with adversity10,11 and subsequently experience less burden.12–14 Unfortunately, many caregivers facing advanced cancer are neither prepared nor supported, particularly in managing patient symptoms at home.9,15–17
Clinicians could cooperate with caregivers in managing patient symptoms. However, many clinicians (1) feel pressured to spend less visit time with patients/families, (2) avoid discussions of psychosocial issues, and (3) tend to underestimate patient’s/family’s needs.18–22 To overcome these barriers, nurses in the Study to Understand Prognoses and Preferences for Outcomes and Risks of Treatments (SUPPORT) repeatedly communicated caregivers’ and patients’ concerns, including pain control, to physicians.23 Yet, physician behavior did not change. More recent studies with similar communication approaches reported mixed results—one showed no improvement on caregiver depression24 while the other showed reduced caregiver distress.25 Moreover, clinician-initiated communication is pre-scheduled and may not occur when caregivers and patients need help the most. Through timely communication, clinicians may help caregivers be prepared in managing symptom changes over time.26–30 A more proactive and feasible intervention that ensures timely caregiver-initiated communications may overcome these barriers.
Information technology may help improve the effectiveness, feasibility, and timeliness of clinician communication. Growing evidence demonstrates that electronic patient reported outcome (ePRO) systems are feasible, well-received by both patients and clinicians, and equivalent to original paper-based systems.31–36 Patients typically log into ePRO systems via Internet-enabled computers or smartphones, and complete electronic report forms. Reported outcomes are transmitted to their clinicians or integrated into the electronic medical records. Moderately to significantly improved patient symptoms and quality of life were reported.37–40 Several ePRO system developments are underway in the U.S. and Europe.39,41,42 The ePRO systems may facilitate timely interaction between patients, caregivers, and clinicians. While current ePRO systems are designed for patients, some43–45 argued that caregivers are in a good (possibly better) position to assess patient condition. Systematic development and evaluation of systems that involve caregivers are needed.
An ePRO system, the Clinician Report (CR), was designed specifically for caregivers caring for advanced-stage cancer patients, built upon a non-commercial, web-based system, the Comprehensive Health Enhancement Support System (CHESS),46,47 which already offered patients a means of tracking their symptoms. Current CHESS modules include caregiver and patient tracking of patient status and caregiver reporting of his/her own status, via a ‘Check-in’ service. The CR delivers this information to the clinicians. With the CR’s alert function, concerns about patient symptoms and caregiving needs may be more likely to reach the clinicians’ attention, potentially leading them to address these issues in a timely manner. Caregivers, with immediate support from clinicians, may be better prepared to manage patient symptoms. As the literature supports, uncontrolled cancer symptoms and side effects lead to increased caregiving physical burden and negative mood. Conversely, well-managed patient symptoms and side effects may reduce caregiving physical burden and negative mood. We examined whether caregivers with access to the CR are better prepared, and experience less physical burden and less negative mood than those without.
Between September 2004 and April 2007, 235 patients with advanced-stage breast, prostate, or lung cancer and their caregivers (family/friends) were jointly recruited. Eligible breast cancer patients were women with metastatic, recurrent or metastatic inflammatory breast cancer, or a chest wall recurrence following mastectomy. Prostate cancer patients were eligible if they had hormone refractory or metastatic prostate cancer. Eligible lung cancer patients included those in stage IIIA, IIIB, or IV. Depending on disease statuses, patients were receiving standard care including curative or palliative treatment. Patients may or may not have had a hospitalization during the course of the treatment, but our intervention was targeted to the outpatient setting. Eligible caregivers were at least 18 years old and were identified by patients as their primary source of physical, emotional, and/or financial support. Caregivers were predominantly female (64.2%) and an average of 56 years old; caregivers’ average educational attainment fell between ‘some college course’ and ‘associate degree’ (Table 1). In total, 55.8% of patients were female. Patients’ average age was 63 years old. Most caregivers (69.3%) were spouse/partners. At baseline, caregivers were fairly comfortable with using the Internet, with the average between ‘a medium amount’ and ‘quite a bit’ of comfort.
Recruitment sites were five cancer centers in the Northeastern, Midwestern, and Southwestern United States. At an outpatient clinic visit, the physician or nurse invited the patient and caregiver (if present) to learn about the project. A site enrollment coordinator then talked with interested people about the study and the consent process. Both the patient and a caregiver had to agree to be in the study. To boost enrollment a one-time invitation was mailed to 561 breast and prostate cancer patients in the State of Wisconsin, identified through Medicaid records. Participants were recruited into one of two randomized trials. One trial enrolled breast and prostate cancer patients and caregivers, the other lung cancer patients. Both studies originally were designed with three randomization groups: (1) Usual-Care (the breast/prostate cancer study) or Internet-Only (the lung study), (2) CHESS-Only, and (3) CHESS + CR. Due to low recruitment rates for both studies, each study was changed to include only two groups. The breast/prostate cancer study retained the CHESS-Only and the CHESS + CR groups. The lung cancer study retained the Internet-Only and the CHESS + CR groups. The 5 breast/prostate cancer dyads in the Usual-Care group and the 37 lung cancer dyads in the CHESS-Only group continued the intervention. This analysis excludes the Usual-Care and the Internet-Only groups but includes the 37 lung cancer dyads in the CHESS-Only group. We sequentially selected a subsample of 37 lung cancer dyads in the CHESS + CR group to match the number of those in the CHESS-Only group from the same recruitment period. Caregiver, patients, and their clinicians were not blinded regarding group assignments. The Internet-Only group in the lung cancer study was not included in this analysis. Approximately 95 caregivers per group were needed to achieve 0.80 power with two-tailed alpha of 0.05 in a planned analysis to detect an effect size of 0.35σ between CHESS-Only and CHESS + CR groups at the two different time periods. This paper reports on a total of 117 dyads in the CHESS-Only group and 118 in the CHESS + CR group across the two studies (Figure 1).
Both groups received access to the CHESS website, which included information, communication, and coaching resources addressing advanced cancer and caregiving needs (Table 2). At initial login to CHESS and then every 7 days, caregivers and patients completed a Check-in, asking questions about their needs and patient symptoms from the modified Edmonton Symptom Assessment System (ESAS)48 and Eastern Cooperative Oncology Group Performance Status49. They could write questions to be addressed by the clinicians in the next visit. Caregivers reported caregiving burden50 and preparedness51. This Check-in allows users to track patient symptom status, monitoring decline or improvement. In addition for the CHESS + CR group, CHESS included the CR that summarized the information provided by patients and caregivers at Check-in and made it available online to the clinicians (Figure 2). Details of the CR development and initial clinician reactions to its use are described in elsewhere.52 Clinicians could access the CR via CHESS anytime. However, any caregiver- or patient-reported ESAS symptom rated at a threshold of 7 or higher on a 0- to 10-scale automatically generated an email alerting the clinician to review the report immediately. Clinicians also received an e-mail alert to review reports 2 days before a scheduled clinic visit, regardless of the ESAS rating. Therefore, the critical distinction between two groups was—while the CHESS-Only group responded to the same Check-in questions as the CHESS + CR group, for the CHESS-Only group, there was no communication of this information to the clinicians.
Both studies were approved by the Institutional Review Boards at recruitment sites. Oncology clinicians were consented, agreeing to receive reports. After patients and caregivers completed the consent and pretest, dyads were randomly assigned to the CHESS-Only or CHESS + CR group (1:1 ratio) with assignment determined at the University of Wisconsin by a random number generator. Each site used a separate randomization schedule. Randomization was blocked by caregiver-patient relationship (spouse/partner vs. nonspouse/partner) and race (Caucasian vs. non-Caucasian). All participants received a laptop with Internet access if needed plus CHESS login information. For those with Internet access, Internet cost was reimbursed while in the study. Computers and user manuals were mailed to the caregivers. Technical support was available by telephone. CHESS staff provided training on using CHESS via telephone or in clinic. Although caregivers and patients were enrolled as dyads, caregivers were the target population in both studies. Patients were offered access but were not required to use CHESS or complete surveys. Breast and prostate cancer dyads had access to the intervention for 12 months; lung cancer dyads had a 24-month intervention period. This analysis utilized the first 12-month data for all three cancer types. Caregivers in both studies were surveyed every two months after receiving the intervention. Breast and prostate cancer caregivers were paid US$20 for each survey completed, while lung cancer participants were able to keep the laptop after study. The 6- and 12-month posttests were hypothesized in the study protocol as the targeted outcome points because 6 months would allow participants enough time to gain benefits from using the systems while 12 months would allow us to study the lasting effect of the Intervention. Study protocols are available upon request.
Caregiver demographic characteristics were assessed at pretest. Caregivers’ comfort in using the Internet was assessed at pretest with a Likert-type item, ‘How comfortable are you using the Internet?’, with response choices from 0 (not at all) to 4 (extremely). Caregiver-perceived patient symptom distress was measured at pretest and posttest using the modified ESAS48 on a 0- to 10-point scale (10 indicating the highest symptom burden). Based on oncologists’ feedback, we replaced three items in the original ESAS (i.e., activity, drowsy, and wellbeing) with three commonly occurring cancer symptoms (i.e., fatigue, constipation, and diarrhea). The ESAS scale score was calculated by summing scores of these nine symptoms.
The four-item caregiver Preparedness scale, a subscale of Family Care Inventory51, was used to assess preparedness for caregiving tasks. Caregivers were asked to indicate, on a 5-point scale, their agreement with the statement from 0 (strongly disagree) to 4 (strongly agree). These items are as follows: I am confident that I can… (1) take care of the patient’s physical needs, (2) take care of the patient’s emotional needs, (3) find out about and set up services for the patient, and (4) cope with the stress of caregiving. Scale scores are calculated as means across items. The scale has high internal consistency measured using Cronbach’s α (pretest: α =0.79, 6-month posttest: α =0.85, and 12-month posttest: α =0.85).
The four-item caregiver Physical Burden scale, a subscale of the Caregiver Burden Inventory50, was used to assess the impact of caregiving on caregivers’ health. Caregivers were asked to indicate, on a 5-point scale, their agreement with the statement from 0 (strongly disagree) to 4 (strongly agree). These items are as follows: (1) I’m getting enough sleep (reverse coded), (2) my health has suffered, (3) I am physically tired, and (4) caregiving has made me physically sick. Scale scores are calculated as means across items. The scale has high internal consistency (pretest: α =0.78, 6-month posttest: α =0.85, and 12-month posttest: α =0.83).
A subset of negative mood items from the Shortened Version Profile of Mood States (SV-POMS)53 was used to assess caregiver negative mood. To reduce participant survey burden, we further shortened the SV-POMS. Items were selected to be representative of the three negative mood sub-scales: (1) Tension-Anxiety (tense, on edge, uneasy, nervous, anxious), (2) Anger-Hostility (annoyed, angry, grouchy, furious, bitter), and (3) Depression-Dejection (discouraged, helpless, hopeless, sad, unhappy, worthless). Using data from a previous CHESS study54, selected items demonstrated reliability with similar overall scale α compared to the original measures. The internal consistency for the depression sub-scale was α = 0.94 for the original SV-POMS and 0.92 for the CHESS version. Caregivers were asked to indicate, on a 5-point scale, the level of negative mood from 0 (not at all) to 4 (extremely). Scale scores are calculated as means across items. The scale has high internal consistency (pretest: α =0.95, 6-month posttest: α =0.96, and 12-month posttest: α =0.96).
All data were doubled-entered. Little’s Missing Completely at Random (MCAR) Test55 showed that the item missingness for all outcome variables was missing completely at random at each time point. Therefore, the scale scores were calculated if at least 75% of items were answered, in order to minimize case deletion due to item missingness.
Some caregivers did not provide data at posttests due to unreturned surveys or officially dropping out. To assess the possible impact of the dropout data, we used pattern mixture modeling56 and determined that no additional adjustments were necessary.
To assess the potential effect of sample characteristics on outcome missingness due to discontinued intervention and lost to follow-up, we compared baseline outcome measures and demographics between caregivers who completed the 6- or 12-month survey versus those who did not. Hypothesis testing was done using General Linear Mixed Model procedure in SPSS 17.57 We examined differences between the CHESS-Only and CHESS + CR group at 6 and 12 months on three outcomes—caregiver preparedness, physical burden, and negative mood. We tested the effects of group, time, and group by time interaction, while controlling for caregiver age, education, gender, caregiver’s comfort of using the Internet, caregiver-reported patient ESAS at 6 and 12 months, and pretest levels of the dependent variables. Group comparisons at 6 and 12 months were conducted using estimated marginal means. Covariates were selected based on their likelihood of influencing caregiver outcomes. Research has shown that age and gender correlate with caregiver distress.58,59 Education and comfort using the Internet may influence a caregiver’s ability to use and benefit from the CR. As mentioned earlier, patient symptom distress is highly correlated to caregiver outcomes.
Tests revealed that caregivers who completed posttests at 6 or 12 months reported lower patient symptom distress at pretest (p=0.04 and 0.03, respectively) than those who did not. However, dependent variables, age, gender, education level, and Internet comfort at baseline were not significantly different between these groups. Tests also showed no significant differences between treatment groups at baseline on demographics, covariate, or dependent variables.
General Linear Mixed Model analyses were used to test for effects on three caregiver outcomes—preparedness, physical burden, and negative mood—at 6- and 12-month follow-up. Compound symmetry covariance structure of the repeated measure was used for a better model fit based on the likelihood test and Bayesian Information Criteria (BIC)60. No main effect of the group assignment was found for caregiver preparedness, F(1, 118.98) = 0.42, p = 0.52, or physical burden, F(1, 107.60) = 0.03, p = 0.86. However, a significant group main effect was found for caregiver negative mood, F(1, 113.74) = 11.04, p = 0.001. Contrast tests comparing groups at each time point showed no significant differences between groups for either preparedness or physical burden, though caregivers in the CHESS + CR group were found to experience less negative mood than those in the CHESS-Only Group at both 6 (p = 0.009) and 12 months (p = 0.004) (Table 3).
This study provided randomized trials of an online symptom reporting system designed to facilitate timely communication between clinicians and caregivers caring for advanced-stage cancer patients. In that sense, CR may increase the opportunity for cooperation between clinicians and caregivers. Through receiving timely alerts and accessing the CR, clinicians may help to manage uncontrolled symptoms and offer support to caregivers immediately. The study found that caregivers with access to the CR reported less negative mood than those without. In an earlier article specifically focused on the CR, we found that high patient symptom distress was improved significantly for those with the CR relative to those without.61 This may explain the effects of negative mood. The CR may enable earlier intervention because clinicians receive automatic e-mail alerts triggered by high patient symptom distress that might otherwise have gone unreported and unattended. With the feeling that patient symptoms are in control, caregivers may experience less emotional distress.9,25,62
However, preparedness and physical burden were not significantly different between those with and without the CR. Both groups can access the same CHESS services which have improved patient self-efficacy in other trials46,54. As such, the CR may not have affected physical burden and preparedness much because the comprehensive resources offered within CHESS may have helped with these outcomes in both groups to some extent. However, another alternative explanation may offer more insight. The CR—as a communication tool—neither tells the clinicians how to better support caregivers nor magically allocates clinicians’ time and resources to do so. Previous ePRO studies showed that patient self-reported outcomes have led to more frequent discussion of symptoms during clinic visits, more symptom issues addressed by clinicians, and majority (84%) of the alerts responded within 24 h.37,38,40 The clinicians, under a tight schedule, may more likely address patient symptom burden than take time helping caregivers cope.15 Future studies should track clinicians’ actual responses to caregivers’ needs, which may help to better understand the findings.
The small difference in negative mood (Table 3)—about 0.6 for CHESS + CR group and 0.9 for CHESS-Only group—may draw concerns regarding the clinical relevance of the effect. To address this concern, we compared the results to another study that reported subscales of SV-POMS for a healthy versus a cancer population. The CHESS + CR caregivers’ perceived levels of negative mood were found to be comparable to that reported by the healthy population whereas the CHESS-Only caregivers experienced similar levels as cancer patients undergoing chemotherapy.53 Therefore, the improvement in the CHESS + CR group compared to the CHESS-Only group is statistically significant and may be clinically meaningful as well.
Several limitations are noteworthy here. First, those caregivers reporting lower patient symptom burden at pretest were more likely to complete their 6- and 12-month follow-up surveys. Our results may underestimate the potential impact of CHESS + CR on those who may benefit the most. Secondly, in our study, about 26% (29/110) of CHESS + CR caregivers and 21% (22/107) of CHESS-Only caregivers did not use the Check-in. However, the small amount of nonusers does not offer sufficient power for sub-group analyses. Future studies acquiring larger sample size may be needed to explore further the effects of amount of system use.63 Moreover, the majority of patients in our sample are well-educated Caucasians. This may limit the ability to generalize the results to other ethnic populations. Although a digital divide may still exist64, recent studies have highlighted benefits of information and communication technologies to both the underserved population65,66 and the hospitals providing care to them67. Finally, because CR is integrated in CHESS our study could not assess how CR alone would influence caregiver outcomes.
The needs for quality, empirical, and longitudinal research in the development and testing of interventions to support caregivers have been well documented.7,68,69 Despite the limitations, this study provides new evidence testing an ePRO system designed specifically for cancer caregivers. Moreover, whereas a previous nursing communication study reported outcomes at 10 weeks after intervention25, our study showed a similar but longer (1 year) effect of improved negative mood. The results are likely generalizable to populations facing other cancers because this study included both gender-specific (breast and prostate) and gender-neutral (lung) cancers, as well as cancers with long-term and short-term median survival periods. The timely communication of caregiver reported information—through the CR—seems to help reduce caregiver’s negative mood. Future system development to overcome barriers identified here may better assist clinicians to support caregivers caring for advanced-stage cancer patients.
This research was made possible through grant funding from the National Cancer Institute (1 P50 CA095817-01A1) and National Institute of Nursing Research (RO1 NR008260-01).
ClinicalTrials.gov Identifiers: NCT00214162 and NCT00365963
Conflict of interest
The Authors declare that there is no conflict of interest.
Ming-Yuan Chih, Department of Industrial and Systems Engineering, University of Wisconsin-Madison, Madison, WI, USA; Center for Health Enhancement Systems Studies, University of Wisconsin-Madison, Madison, WI, USA.
Lori L DuBenske, School of Medicine and Public Health, University of Wisconsin-Madison, Madison, WI, USA; Center for Health Enhancement Systems Studies, University of Wisconsin-Madison, Madison, WI, USA.
Robert P Hawkins, School of Journalism and Mass Communication, University of Wisconsin-Madison, Madison, WI, USA; Center for Health Enhancement Systems Studies, University of Wisconsin-Madison, Madison, WI, USA.
Roger L Brown, School of Nursing, University of Wisconsin-Madison, Madison, WI, USA.
Susan K Dinauer, Center for Health Enhancement Systems Studies, University of Wisconsin-Madison, Madison, WI, USA.
James F Cleary, School of Medicine and Public Health, University of Wisconsin-Madison, Madison, WI, USA.
David H Gustafson, Department of Industrial and Systems Engineering, University of Wisconsin-Madison, Madison, WI, USA; Center for Health Enhancement Systems Studies, University of Wisconsin-Madison, Madison, WI, USA.