Health Care Delivery
Original Contribution
Randomized Controlled Trial of Shared Care for Patients With Cancer Involving General Practitioners and Cancer Specialists
The University of Western Australia; Harry Perkins Institute of Medical Research, The University of Western Australia; School of Primary, Aboriginal and Rural Health Care, The University of Western Australia; School of Medicine and Pharmacology, The University of Western Australia, Crawley; Royal Perth Hospital, Perth; General Practice and Primary Care Academic Centre, University of Melbourne, Carlton; Sir Charles Gairdner Hospital; Haematology Care Centre, Sir Charles Gairdner Hospital, Nedlands, Australia; and University of Oxford, Oxford, United Kingdom
Abstract Purpose: We aimed to determine whether a shared care model (SCM) during chemotherapy treatment improved emotional wellbeing, empowerment, and prevalence of symptoms for people being treated for cancer.
Methods: People receiving chemotherapy for hematologic, breast, ovarian, or colorectal malignancies at two cancer centers were randomly assigned to receive SCM or standard care. The SCM involved a patient-held record, a project coordinator, routine contact between the patient and general practitioner/primary care physician, and primary care physician education. Participants completed the Hospital Anxiety and Depression Scale, the Mini-Mental Adjustment to Cancer, and an empowerment questionnaire before, in the middle of, and on completion of chemotherapy. The presence and severity of adverse effects of chemotherapy were recorded by patients in a symptom diary.
Introduction Australian cancer survival rates are high,1 with 5-year relative survival of 66%.2 Most Australians diagnosed with cancer are referred by their primary care physician (PCP) for definitive diagnosis and treatment by cancer specialists, with little ongoing PCP involvement.3 Specialists manage cancer-specific treatment, and often also coordinate care and provide support, palliation and follow-up. However, oncologists are costly and too few in most Australian public health systems to adequately care for the increasing number of patients,4 and there are concerns about fragmented care5,6 This traditional model fails to meet patient needs in relation to satisfaction, quality of life, empowerment, or morbidity.3 Many patients and caregivers report frustration, dissatisfaction, and inability to access appropriate and timely specialist care.8-10 Consequently, policymakers seek models of care that are cost effective and demonstrably improve psychosocial outcomes.11 Australia has a robust PCP workforce, with approximately 107 PCPs per 100,000 people,12 who can safely provide supportive and follow-up care for patients with cancer.13 InvolveCopyright © 2015 by American Society of Clinical Oncology
Results: Ninety-seven eligible participants were randomly allocated, less than half the intended recruitment. There were no significant differences between the groups for empowerment, symptom prevalence, or Mini-Mental Adjustment to Cancer scores. The proportion with clinical anxiety (Hospital Anxiety and Depression Scale anxiety score of ⱖ 11) decreased over time in both groups (P ⫽ .013) but decreased more in the intervention group (P ⫽ .002). Depression was unchanged over time. Conclusion: Our study was limited by low recruitment and predominance of patients with breast cancer, and was underpowered for the main analyses. Results should therefore be interpreted with caution. Little benefit was seen for SCM in the majority of domains including empowerment, symptom prevalence, and psychological adjustment to cancer. The SCM showed efficacy in clinically anxious patients. Such interventions may be better implemented by using a targeted approach to identify at-need subgroups.
ment of PCPs with specialists in cancer management has resulted in improved continuity of care, greater teamwork, proactive (not reactive) care, and improved patient support.14 Patients, however, report that PCPs are not always well informed about treatments and management of adverse effects.8-10 PCPs are also frustrated about limited communication from specialists and loss of involvement in ongoing care. Contact is usually limited to PCP referrals, specialists’ letters, and discharge summaries. These are frequently late and lacking information on the overall treatment plan and potential treatment adverse effects.15 At some point after treatment is completed, patients return to the care of their PCP, but timing is variable and can occur with limited information about diagnosis, treatment, and survivorship care. Most patients with cancer receive their chemotherapy in outpatient settings, and treatment toxicities commonly develop away from the treatment center, so greater PCP involvement may reduce the severity of toxicities and adverse effects through earlier intervention.16 PCPs may increase the level of psychosocial support for patients during their treatment and in the follow-up phase, which is especially beneficial to patients in
S E P T E M B E R 2015
•
jop.ascopubs.org
349
Information downloaded from jop.ascopubs.org and provided by at Monash University on November 22, 2015 from 130.194.20.173 Copyright © 2015 American Society of Clinical Oncology. All rights reserved.
By Claire E. Johnson, RN, CM, PhD, Christobel M. Saunders, MBBS, FRCS, FRACS, Michael Phillips, Jon D. Emery, DPhil, MBBCh, Anna K. Nowak, MBBS, PhD, FRACP, Kate Overheu, Alison M. Ward, PhD, and David J.L. Joske, MBBS, FRACP, FRCPA
Johnson et al
Methods Ethics approval was obtained from the Human Research Ethics Committees at Sir Charles Gairdner and Royal Perth Hospitals (Ref Nos 2005-037 and RA-06/010A). The study was registered with the Australian New Zealand Clinical Trials Registry (ACTRN 12606000022561). 350
JOURNAL
OF
ONCOLOGY PRACTICE
•
Participants Patient, specialist, and PCP triads were recruited to the study. Written informed consent was obtained from all participants. Hematologists and medical oncologists treating patients at two hospitals in Western Australia were invited to participate. Patients cared for by six consenting specialists were invited to participate. All aspects of the project were explained by the chief investigator (D.J.L.J.). Patients receiving their first round of chemotherapy in participating outpatient clinics were invited to participate. Eligible patients were ⬎ 16 years of age, able to understand and complete trial documentation, and had biopsy-proven malignancy. Four patient streams were recruited: (1) patients with nonHodgkin lymphoma receiving cyclophosphamide, vincristine, doxorubicin, and oral prednisolone; (2) women with stages I and II breast cancer offered four to six cycles of adjuvant chemotherapy; (3) women with stages Ic, II (a-c) and III (a-c) ovarian cancer who had undergone surgical debulking and were offered 6 to 8 cycles of adjuvant chemotherapy with carboplatin and paclitaxel; and (4) patients with cancer of the colon, Dukes B or C, who had undergone surgical resection and offered 6 months chemotherapy with infusional fluorouracil, leucovorin, and oxaliplatin; fluorouracil/leucovorin; or capecitabine. PCPs of consenting patients were contacted by the coordinator to invite participation. To prevent contamination, individual PCPs could have only one patient on study. Intervention group. The SCM involved a PHR, a project coordinator to assist with patient care and information, and PCP educational resource (“upskilling”) packages detailing anticipated adverse effects of each treatment regimen and actions to be taken (Appendix 2). Patients were asked to see their PCP after each chemotherapy treatment. Control group. Usual care was provided by specialists and their associated PCP. PCPs received a letter from the specialist team after each visit, discharge summaries, and telephone communication, where appropriate. Communication frequency and content for standard care was not prescribed by the study. All aspects of the study were explained to consenting patients and PCPs by the coordinator after group allocation was revealed. A comparison of the elements of the intervention and control is shown in Appendix 2.
Study Outcomes The Hospital Anxiety and Depression Scale (HADS)21 measured patient anxiety and depression. HADS scores for each subscale (HADS-A [anxiety] and HADS-D [depression]) were classified as low (0 to 7), borderline clinical (8 to 10), or clinically significant (11 to 21). Empowerment was measured by the Patient Empowerment Scale.22 Coping was measured using the Mini-Mental Adjustment to Cancer Scale (Mini-Mac).23 Each was administered before chemotherapy, midway through treatment, and on completion of treatment. All participants were asked to complete a daily symptom diary to record the presence and severity of six key adverse effects of chemotherapy (tiredness, nausea, sore mouth, poor appetite, numbness and fever,
V O L . 11, I S S U E 5
Copyright © 2015 by American Society of Clinical Oncology
Information downloaded from jop.ascopubs.org and provided by at Monash University on November 22, 2015 from 130.194.20.173 Copyright © 2015 American Society of Clinical Oncology. All rights reserved.
rural areas. Improved co-operation and communication between primary and specialist providers would facilitate timely information, care, and support for patients.8-10 Shared care is an organizational model involving both PCPs and specialists in a formal, explicit manner. Potential advantages include improved quality of life and psychological well-being, reduced hospitalizations, and better medication compliance. Shared care may also overcome difficulties in traditional hospital-based models of cancer care, particularly when patients live in remote areas.17 Shared care models (SCMs) ensure that patients and caregivers have access to both hospital- and community-based supportive care throughout treatment and along the continuum to cure or palliation.18 Cost reductions through fewer hospital attendances without reduction in quality of care have also been reported.17 Patient-held records (PHRs) are an important component of many SCMs. The PHR is a formal, structured record retained by the patient containing patient and health care information. Clinicians and patients can add to the PHR, ensuring better continuity and quality of care19 Randomized controlled trials of PHRs alone have not shown significant improvements in communication, quality of life, or health outcomes for people with cancer.19,20 Reports of multicomponent SCMs that include PHRs in chronic diseases and a systematic review of seven trials of PHRs in patients with cancer show equivocal impact on the patient journey.17,19 However, there are few reports of complex SCMs that include a combination of PHRs and other interventions such as PCP training, opportunities for enhanced communication between health professionals, and access to a patient coordinator for people receiving cancer treatment. Our comprehensive cancer SCM was designed to improve information sharing and PCP knowledge and involvement, and was pilot tested in the hematology department at a tertiary cancer center in Perth, Western Australia (Appendix 1, online only). Based on the outcomes of the pilot, we developed a conceptual framework and refined the SCM (Appendix 2, online only). We evaluated the SCM in patients with hematologic, breast, ovarian, or colorectal malignancies receiving chemotherapy, and compared the model to standard care. Our primary hypothesis was that by using our SCM, people being treated with chemotherapy would have higher levels of emotional well-being and empowerment, lower levels of anxiety and distress, and fewer adverse effects compared with people receiving standard care.
An RCT of Shared Care for Patients With Cancer
Assessed for eligibility (N = 153)
Randomly allocated (n = 100)
Excluded Did not meet inclusion criteria Refused to participate Other reasons Reason: patients missed before start of chemotherapy
Allocated to intervention (n = 52) Received allocated intervention (n = 52) Did not receive allocated (n = 0) intervention
Did not complete study Reasons Lost to follow-up Patients changed chemotherapy Patients chose to withdraw from study
Did not complete study Reasons Patients ceased chemotherapy Patients chose to withdraw from study
(n = 5) (n = 1) (n = 2) (n = 2)
Exited from study (n = 42) Excluded from analysis (n = 1) Reason: patient already had chemotherapy in past, therefore should not have been recruited Analyzed (n = 41 + 5 withdrawn = 46)
(n = 4) (n = 1) (n = 3)
Exited from study (n = 48) Excluded from analysis (n = 1) Reason: patient’s GP was already in the trial, therefore patient should not have been recruited Analyzed (n = 47 + 4 withdrawn = 51)
Figure 1. Cancer shared care trial recruitment flow chart.
identified by clinicians and in the literature24) on a scale of 0 to 10 (0 ⫽ no symptoms, 10 ⫽ the worst I can imagine). Patients could also record other symptoms. On completion of the study, patients, PCPs, and specialists were asked to complete a questionnaire exploring perceptions of the usefulness of the SCM.
Sample Size A sample size of 240 (120 patients per group) was calculated to provide 80% power to detect a reduction from 30% to 15% in the proportion of people showing borderline clinical anxiety or depression on the HADS with an alpha of .05.
Random Allocation Random allocation was performed externally using an opaque sealed envelope method.25 The sample was stratified by cancer type. The baseline questionnaire was performed triple blinded to knowledge of allocation. To implement the intervention, medical oncologists, the coordinator, PCPs, and patients were subsequently made aware of the patient’s allocation.
Statistical Analyses Descriptive methods used arithmetic means with 95% CIs for normally distributed continuous variables and geometric means for skewed and log-normal distributions. Percentages were used to describe categorical and ordinal variables. Associations between continuous dependent variables were assessed using Spearman’s rho statistic because the assumption of normality was rarely satisfied. Copyright © 2015 by American Society of Clinical Oncology
Between-groups comparisons used the likelihood ratio 2 test for categorical variables and longitudinal mixed and random-effects models.26 In mixed models, time was treated as a random effect and randomization group was a fixed effect. Alpha ⬍ .05 was regarded as statistically significant. All analyses were conducted using Stata version 12 (Stata Corporation, College Station, TX).
Results Recruitment was open for 2 years, but stopped early as a result of slow accrual. Forty-six (47%) eligible participants were randomly allocated to the control group and 51 (53%) to the intervention. Nine subjects dropped out of the control group, and seven dropped out of the intervention group (20% and 14%, respectively; P ⫽ .41). Seven subjects left after cycle 1 of their chemotherapy and a further nine after cycle 3 (Figure 1).
Participant Characteristics Most participants were female (n ⫽ 83; 86%), had a diagnosis of breast cancer (n ⫽ 74; 76%), and were from the metropolitan area (n ⫽ 76; 78%). There were no differences between the groups in age, sex, type and stage of cancer, and geographical location, or in measures of psychological well-being or physical symptoms at baseline (Table 1).
Compliance All participants in the intervention group discussed the project with the coordinator and were given the PHR. Four (10%)
S E P T E M B E R 2015
•
jop.ascopubs.org
351
Information downloaded from jop.ascopubs.org and provided by at Monash University on November 22, 2015 from 130.194.20.173 Copyright © 2015 American Society of Clinical Oncology. All rights reserved.
Allocated to control (n = 48) Received allocated intervention (n = 47) Did not receive allocated (n = 1) intervention; reason: patient did not have chemotherapy
(n = 53) (n = 21) (n = 23) (n = 9)
Johnson et al
Table 1. Characteristics of Participants in the Control and Intervention Groups at Baseline Overall Sample Characteristic
No.
%
Control No.
%
Intervention No.
%
Age, years Mean
54.7
Range
28-77
Randomization
53.2
Lower: 53.3, upper: 59.6
Lower: 50.2, upper: 56.2
46
47.4
51
52.6
Sex Female
83
85.6
41
89.1
42
82.4
Male
14
14.4
5
10.9
9
17.7
74
76.3
37
80.4
37
72.6
Cancer type Breast, female Colorectal
2
2.1
1
2.2
1
2.0
Lymphoma
15
15.5
6
13.0
9
17.7
6
6.2
2
4.4
4
7.8
I
15
16.5
5
12.5
10
19.6
II
50
55.0
25
62.5
25
49.0
III
23
25.3
10
25.0
13
25.5
IV
3
3.3
0
0.0
3
5.9
Metropolitan
76
78.4
36
78.3
40
78.4
Rural
21
21.7
10
21.7
11
21.6
None
46
47.9
23
51.1
23
45.1
Borderline†
25
26.0
11
24.4
14
27.5
Clinical‡
25
26.0
11
24.4
14
27.5
None
79
82.3
37
82.2
42
82.4
Borderline†
10
10.4
3
6.7
7
13.7
7
7.3
5
11.1
2
3.9
36
44.4
15
38.5
21
50.0
6
8.7
3
9.1
3
8.3
60
62.5
18
46.2
42
73.7
6
8.0
3
8.3
3
7.7
75
80.7
33
84.6
42
77.8
0
0
0
0
0
0
Ovarian Stage*
Residence
HADS anxiety
HADS depression
Clinical‡ Symptoms Loss of appetite Fever Nausea Numbness Fatigue Mouth ulcers
NOTE. N does not equal 97 for all variables as a result of participant withdrawals. Abbreviations: CL, confidence limit; HADS, Hospital Anxiety and Depression Scale. * Stage is a composite of American Joint Committee on Cancer and Ann Arbor staging for different cancer types. Stage is unknown for six patients in the control group. † Borderline clinical anxiety or depression ⫽ HADS score of 8-10 ‡ Clinically significant anxiety or depression ⫽ HADS score of ⱖ 11.
contacted the coordinator outside of regular clinic visits. More than half (56%) of intervention patients saw their PCP during treatment, compared with 46% of control patients. The average number of visits to the PCP was also greater for the intervention patients (2.79 v 1.61; P ⬍ .001). 352
JOURNAL
OF
ONCOLOGY PRACTICE
•
At baseline, 26% of all participants had clinically significant anxiety levels (ie, HADS-A ⱖ 1121), and another 26% had borderline anxiety levels (HADS-A 8 to 10). Fewer participants were depressed as measured by HADS scores: 7.3% were classified as having clinically significant depression and 10.4% as having borderline clinical depression. Mean anxiety and depression scores were 7.05 and 3.12, respectively. There was a 10% reduction in HADS-A over the duration of enrolment for all patients (P ⬍ .001) but no significant reduction in HADS-D (P ⫽ .137). There was a statistically significant interaction between time and group for HADS-D on longitudinal analysis. This showed an increase in the control group and a slight reduction for the intervention group (P ⫽ .040; Table 2). We found a statistically significant interaction between time and group for clinically anxious patients. The proportion of such patients decreased over time in both groups (P ⫽ .013); the decrease was greater in the intervention group (14 patients [27.5%] at baseline to three patients [6.8%] after the sixth cycle of chemotherapy; P ⫽ .002) versus the control group (11 patients [24.4%] at baseline to five patients [13.9%] after the sixth cycle of chemotherapy; P ⫽ .014). Thus, the intervention resolved clinically significant anxiety, to subclinical levels, more frequently than was seen in the controls. No group difference was detected in the proportion of patients clinically depressed over time, that is: five patients (11.0%) to four patients (8.7%) in the control group (P ⫽ .96) versus two patients (3.9%) to three patients (6.8%) in the intervention (P ⫽ .86).
Psychological Outcomes: Mini-MAC and Empowerment On univariable analysis, no differences were observed between the intervention and control group mean scores for psychological variables (Mini-MAC and Empowerment) at any of the three time points.
Symptom Outcomes Total compliance with completion of symptom diaries was 26% (control group) and 28% (intervention group). Partial compliance (at least one entry missing) was 38% and 45%, respectively, and the remainder did not complete any entries (36% and 27%, respectively). When reported, symptoms and adverse effects of chemotherapy were generally of low intensity (1 or 2 out of 10), with the intervention group scoring slightly more symptoms. The majority of participants reported feeling fatigued, about half reported nausea, and fewer than half reported other symptoms. The number of symptoms reported during any cycle of chemotherapy showed no difference between the two groups (incidence rate ratio ⫽ 1.23; 95% CI, 0.77 to 1.97; P ⫽ .39). When adjusted for covariates, including the cycle of chemotherapy, day of the cycle, and cancer type, participants in the intervention group were more likely to report loss of appetite
V O L . 11, I S S U E 5
Copyright © 2015 by American Society of Clinical Oncology
Information downloaded from jop.ascopubs.org and provided by at Monash University on November 22, 2015 from 130.194.20.173 Copyright © 2015 American Society of Clinical Oncology. All rights reserved.
95% CL
56.5
Psychological Outcomes: HADS
An RCT of Shared Care for Patients With Cancer
Table 2. Mean Psychological Well-Being Scores Over Time, Control Versus Intervention Group Control Group Outcome Measure
Geometric Mean
LCL 95%
Intervention Group UCL 95%
Geometric Mean
LCL 95%
UCL 95%
Anxiety score (HADS) Baseline
.80 7.91
6.50
9.32
8.05
6.71
9.40
6.44
5.40
7.48
6.65
5.64
7.66
5.24
4.26
6.22
5.49
4.54
6.43
3.66
2.92
4.40
4.09
3.31
4.86
Depression score (HADS)
.04
Mid-study
4.16
3.39
4.92
4.06
3.36
4.76
After cycle 6
4.72
3.72
5.72
4.04
3.25
4.83
Anxious preoccupation (Mini-MAC) Baseline
.92 19.5
18.1
20.8
19.8
18.5
21.1
Mid-study
18.5
17.2
19.8
18.8
17.6
20.0
After cycle 6
17.5
16.1
19.0
17.8
16.5
19.2
Cognitive avoidance (Mini-MAC) Baseline
.66 10.1
9.51
10.7
9.45
8.89
10.0
Mid-study
10.2
9.66
10.7
9.61
9.15
10.1
After cycle 6
10.2
9.55
10.9
9.77
9.18
10.4
Fighting spirit (Mini-MAC) Baseline
.35 20.8
19.9
21.8
20.7
19.8
21.6
Mid-study
20.4
19.5
21.3
20.6
19.8
21.4
After cycle 6
20.0
19.0
21.0
20.5
19.5
21.4
15.2
13.9
16.5
15.1
13.9
16.3
Hopelessness/helplessness (Mini-MAC) Baseline
.40
Mid-study
15.3
14.1
16.4
14.8
13.7
15.8
After cycle 6
15.3
13.9
16.7
14.4
13.2
15.6
12.1
10.4
13.7
12.1
10.4
13.7
Patient empowerment scale Baseline
.47
Mid-study
11.6
10.0
13.2
11.6
10.0
13.2
After cycle 6
13.0
11.7
14.3
13.0
11.7
14.3
Abbreviations: HADS, Hospital Anxiety and Depression Scale; Mini-MAC, Mini-Mental Adjustment to Cancer; LCL, lower confidence limit; UCL, upper confidence limit.
(P ⫽ .004) and nausea (P ⫽ .012; Appendix Table A1, online only). A subgroup analysis of patients with breast cancer showed similar results to the sample as a whole only for the main outcomes (Appendix 3, online only).
tected in perceptions of care between the intervention and control groups.
Emergency Department/Hospital Admissions
Previous experience with PHRs in cancer populations has been equivocal.19 Models of shared care for people with other chronic conditions have reported few differences in mental health, physical outcomes, and psychosocial outcomes between intervention and control groups.17 Our SCM consisted of a PHR, upskilling resources for PCPs, routine contact between the patient and their PCP, and the opportunity for the PCP to liaise with the project coordinator. Overall, our results are consistent with previous findings, showing little impact on psychological variables as measured by HADS,21 the Patient Empowerment Scale,22 and MiniMAC23 instruments, nor on self-reported symptoms. Indeed, the slight increase in reported symptomatology in the intervention group could suggest increased focus by patients on their symptoms. The intervention had no significant effect on patients’ self-reported mood or on overall levels of anxiety, em-
We recorded emergency department presentations and hospital admissions. There was no difference between the intervention and control groups (Appendix 4, online only).
Participants’ Perceptions of Care and the SCM Participants’ responses to the SCM are reported in Appendix 4 (online only). One of five cancer specialists who completed the exit survey thought the PHR was useful, and all had reservations about the usefulness of other aspects of the SCM. The majority of the PCPs in the intervention (88%) reported that they found the PHR useful and the majority of open comments about the intervention were positive. There were no differences detected in perceptions of care between PCPs in the intervention and the control groups. Among patients, there were no differences deCopyright © 2015 by American Society of Clinical Oncology
Discussion
S E P T E M B E R 2015
•
jop.ascopubs.org
353
Information downloaded from jop.ascopubs.org and provided by at Monash University on November 22, 2015 from 130.194.20.173 Copyright © 2015 American Society of Clinical Oncology. All rights reserved.
Mid-study After cycle 6
Baseline
P for Comparison of Groups Over Time
Johnson et al
Limitations Caution should be exercised in generalizing our results because of potential limitations. First, the sample size was small and was mostly women (85%) with breast cancer (75%). The predominance of patients with breast cancer related to the higher volume of such cases at the trial sites and the late involvement of colorectal cancer services. Early closure meant our target sample size was not met. Factors contributing to slow recruitment included changes in clinical practice (increased treatment of patients with cancer in rural and regional centers) and trial fatigue due to concurrent chemotherapy trials, in colorectal cancer particularly. The intended statistical power of the study was not achieved. Second, it is unclear to what extent participants experienced physical symptoms or adverse effects of treatment because compliance with daily symptom diaries was low. Third, oncology specialists’ commitment to the intervention was limited, with some patients reporting underuse of the PHR by their specialist. This was compounded by other concurrent trials and workload pressures. 354
JOURNAL
OF
ONCOLOGY PRACTICE
•
Barriers to successful SCM implementation included patient trial fatigue, limited engagement with the PHR by patients and specialists, and lack of specialist buy-in to the model. Alternative methods of information sharing are being studied, such as case conferences,30 faxed treatment regimes,31 structured information packs,32 and electronic health records.33 Reported benefits include improved sharing of information, increased collaboration, better preparation of PCPs to provide care, more coordinated care, and higher PCP satisfaction with the secondary provider. However, translation of these interventions into routine practice is slow, with many programs dependent on charismatic leadership rather than systematic implementation.34
Summary In summary, we found a lack of even a trend for improved psychosocial functioning, empowerment, or symptom control with our SCM, and we urge caution in its wide application. Indeed, this study adds further evidence regarding the limited usefulness of the PHR during chemotherapy in patients with cancer.19 Our findings support data suggesting that shared cared interventions are better targeted toward patient groups identified as being most likely to benefit,35 (eg, patients who are clinically anxious). A comprehensive SCM may be most valuable when implemented with distress or anxiety screening. However, because we found that the outcomes of the SCM were not inferior to those of usual care, the use of shared care may be beneficial if such a model was more cost efficient and accessibility for patients was improved. Further research is warranted to explore these possibilities. Acknowledgment Supported by the National Health and Medical Research Council, Australia Grant No. 353678. Previously presented in part at Change Champions: Innovations in the Management of Cancer Services, Melbourne, Australia, March 2006; and Achieving Excellence in WA Health Conference, Perth, Australia, October 2006. We acknowledge the patients and health professionals who participated in this study. We also thank Nick De Klerk, Andy Redfern, and Caroline Bulsara for their contributions. Authors’ Disclosures of Potential Conflicts of Interest The authors indicated no potential conflicts of interest.
Author Contributions Conception and design: Christobel M. Saunders, Jon D. Emery, Kate Overheu, Alison M. Ward, David J.L. Joske Administrative support: Kate Overheu Collection and assembly of data: Anna K. Nowak, Kate Overheu, David J.L. Joske Data analysis and interpretation: Claire E. Johnson, Christobel M. Saunders, Michael Phillips, Jon D. Emery, Anna K. Nowak, Alison M. Ward, David J.L. Joske Manuscript writing: All authors Final approval of manuscript: All authors Corresponding author: Claire E. Johnson, RN, CM, PhD, School of Surgery, University of Western Australia, M507, 35 Stirling Hwy, Crawley 6009 Australia; e-mail: [email protected].
DOI: 10.1200/JOP.2014.001569; published online ahead of print at jop.ascopubs.org on March 24, 2015.
V O L . 11, I S S U E 5
Copyright © 2015 by American Society of Clinical Oncology
Information downloaded from jop.ascopubs.org and provided by at Monash University on November 22, 2015 from 130.194.20.173 Copyright © 2015 American Society of Clinical Oncology. All rights reserved.
powerment, or depression. This may reflect a wider issue that such measures are not the most appropriate or sensitive to detect benefits or harms of alternative models of cancer care. The low level of engagement with the PHRs and contact with the coordinator by both patients and clinicians is consistent with other trials. In a systematic review of PHRs in cancer care, Gysels et al19 documented similar difficulties with the implementation of a PHR to those we encountered. We also found the PHR to be ineffective at improving communication between health care providers, or recognizing unmet needs of patients. Some 26% of our sample were clinically anxious at the beginning of treatment (HADS-A scores ⱖ 1121), higher than reported in previous research with patients with cancer or the general population. In a study of 1,529 patients with cancer and 2,037 people representative of the German population, 21% of patients with cancer were clinically anxious compared with 6% in the population sample.27 Women were more anxious than men for most cancer types. In two studies of women with breast cancer, 16% had clinically significant anxiety in hospital postsurgery,28 and 18% were anxious at diagnosis.29 In both, the proportion of patients with anxiety decreased with time, although the rate of decrease varied considerably. The higher rate of anxiety at baseline in our study may reflect that the majority of patient participants were women with breast cancer and that assessment was done immediately before chemotherapy. As with previous research, anxiety levels in our participants decreased over time, but participants with elevated anxiety scores in the intervention group showed greater resolution to normal levels than those in the control group. Analyzing the symptom diaries, participants in the intervention group reported some symptoms more frequently than participants in the control group. However, we stress that these were reported at very low levels and are unlikely to be clinically meaningful. Therefore, regular visits to the PCP postchemotherapy may not be warranted.
An RCT of Shared Care for Patients With Cancer
References 1. Coleman MP, Forman D, Bryant H, et al: Cancer survival in Australia, Canada, Denmark, Norway, Sweden, and the UK, 1995-2007 (the International Cancer Benchmarking Partnership): An analysis of population-based cancer registry data. Lancet 377:127-138, 2011 2. Australian Institute of Health and Welfare: Cancer Survival and Prevalence in Australia: Period Estimates From 1982 to 2010. Cancer Series No. 69 (Cat. No. CAN 65 2012). Canberra, Australia, Australian Institute of Health and Welfare 3. Clinical Oncological Society of Australia, The Cancer Council Australia, and National Cancer Control Initiative: Optimising Cancer Care in Australia. Melbourne, Australia, National Cancer Control Initiative, 2003
5. Royal College of Physicians and Royal College of Radiologists: Cancer Patients in Crisis: Responding to Urgent Needs. Report of a Working Party. London, United Kingdom, Royal College of Physicians, 2012 6. The Senate Community Affairs References Committee: The cancer journey: Informing choice. Report on the inquiry into services and treatment options for persons with cancer, 2005. Canberra, Australia, Commonwealth of Australia, pp 125-136
19. Gysels M, Richardson A, Higginson IJ: Does the patient-held record improve continuity and related outcomes in cancer care: A systematic review. Health Expect 10:75-91, 2007 20. Ko H, Turner T, Jones C, et al: Patient-held medical records for patients with chronic disease: A systematic review. Qual Saf Health Care 19:1-7, 2010 21. Zigmond AS, Snaith RP: The hospital anxiety and depression scale. Acta Psychiatr Scand 67:361-370, 1983 22. Bulsara C, Style I, Ward AM, et al: The psychometrics of developing the patient empowerment scale. J Psychosoc Oncol 24:1-16, 2006 23. Watson M, Homewood J: Mental Adjustment to Cancer Scale User’s Manual. Sutton, United Kingdom, The Royal Marsden Hospital, 2008 24. Shapiro CL, Recht A: Side effects of adjuvant treatment of breast cancer. N Engl J Med 344:1997-2008, 2001 25. Egger M, Smith G, Altman D (eds): Systematic Reviews in Health Care Meta-analysis in Context. London, United Kingdom, BMJ Books, 2001
7. Reference deleted
26. Fitzmaurice G, Laird N, Ware J: Applied Longitudinal Analysis. Hoboken, NJ, John Wiley, 2004
8. King M, Jones L, Richardson A, et al: The relationship between patients’ experiences of continuity of cancer care and health outcomes: A mixed methods study. Br J Cancer 98:529-536, 2008
27. Hinz A, Krauss O, Hauss JP, et al: Anxiety and depression in cancer patients compared with the general population. Eur J Cancer Care 19:522-529, 2010
9. Kleeberg UR, Feyer P, Günther W, et al: Patient satisfaction in outpatient cancer care: A prospective survey using The PASQOC questionnaire. Support Care Cancer 16:947-954, 2008
28. Mehnert A, Koch U: Prevalence of acute and post-traumatic stress disorder and comorbid mental disorders in breast cancer patients during primary cancer care: A prospective study. Psycho-oncology 16:181-188, 2007
10. Sandoval GA, Brown AD, Sullivan T, et al: Factors that influence cancer patients’ overall perceptions of the quality of care. Int J Qual Health Care 18:266274, 2006
29. Stafford L, Judd F, Gibson P, et al: Screening for depression and anxiety in women with breast and gynaecologic cancer: Course and prevalence of morbidity over 12 months. Psycho-oncology 22:2071-2078, 2013
11. National Health and Hospitals Reform Commission: A Healthier Future for All Australians - Final Report of the National Health and Hospitals Reform Commission. 2009. Canberra, Australia, Commonwealth of Australia, 2009, p 107
30. Mitchell G, Cherry M, Kennedy R, et al: General practitioner, specialist providers case conferences in palliative care—Lessons learned from 56 case conferences. Australian Family Physician, 34(5): 389-392, 2005
12. Australian Institute of Health and Welfare: What types of medical practitioners are there? www.aihw.gov.au/workforce/medical/types-of-medical-practitioners/ 13. Emery JD, Shaw K, Williams B: The role of primary care in early detection and follow-up of cancer. Nat Rev Clin Oncol 11:38-48, 2014
31. Jefford M, Baravelli C, Dudgeon P, et al: Tailored chemotherapy information faxed to general practitioners improves confidence in managing adverse effects and satisfaction with shared care: Results from a randomized controlled trial. J Clin Oncol 26:2272-2277, 2008
14. Murray SA, Boyd K, Campbell C, et al: Implementing a service users’ framework for cancer care in primary care: An action research study. Fam Pract 25:7885, 2008
32. Kousgaard KR, Nielsen JD, Olesen F, et al: General practitioner assessment of structured oncological information accompanying newly referred cancer patients. Scand J Prim Health Care 21:110-114, 2003
15. McConnell D, Butow PN, Tattersall MH: Improving the letters we write: An exploration of doctor-doctor communication in cancer care. Br J Cancer 80:427437, 1999
33. Schoen C, Osborn R, Squires D: A survey of primary care doctors in ten countries shows progress in use of health information technology, less in other areas. Health Aff 31:2805-2816, 2012
16. Roorda C, de Bock GH, van der Veen WJ, et al: Role of the general practitioner during the active breast cancer treatment phase: An analysis of health care use. Support Care Cancer 20:705-714, 2012
34. Gardiner C, Gott M, Ingleton C: Factors supporting good partnership working between generalist and specialist palliative care services: A systematic review. Br J Gen Pract 62:e353-e362, 2012
17. Smith SM, Allwright S, O’Dowd T: Does sharing care across the primary “specialty interface improve outcomes in chronic disease? A systematic review. Am J Manage Care 14:213-224, 2008
35. Nielsen JD, Palshof T, Mainz J, et al: Randomised controlled trial of a shared care programme for newly referred cancer patients: Bridging the gap between general practice and hospital. Qual Saf Health Care 12:263-272, 2003
Copyright © 2015 by American Society of Clinical Oncology
S E P T E M B E R 2015
•
jop.ascopubs.org
355
Information downloaded from jop.ascopubs.org and provided by at Monash University on November 22, 2015 from 130.194.20.173 Copyright © 2015 American Society of Clinical Oncology. All rights reserved.
4. Blinman PL, Grimison P, Barton MB, et al: The shortage of medical oncologists: The Australian Medical Oncologist Workforce Study. Med J Aust 196:58-61, 2012
18. Joske D, McGrath P: Implementing the holistic philosophy in cancer care: New directions for the Sir Charles Gairdner Hospital. Austral-Asian J Cancer 2:33-38, 2002
Johnson et al
Appendix
Table A1. Frequency of Participants Reporting Any Level of Symptoms by Group Across All Cycles: Results of a Longitudinal Negative Binomial Regression Analysis Symptom
Control %
Intervention %
IRR*
LCL 95%
Appetite loss
34.2
49.2
2.46
1.36
Fever
UCL 95% 4.55 10.8
P (H0: IRR ⴝ 1) .004
11.1
2.17
0.436
42.4
50.6
1.74
1.13
2.68
.012
.34
Numbness
23.0
24.8
1.83
0.868
3.86
.11
Fatigue
86.4
85.5
1.28
0.993
1.65
.057
Mouth ulcers
37.4
39.0
1.38
0.780
2.44
.27
Abbreviations: IRR, incidence rate ratio; LCL, lower confidence limit; UCL, upper confidence limit. * The IRR estimate is adjusted for cycle of chemotherapy, day of cycle, cancer type, place of residence, sex, and stage for each symptom if the P value for the covariate was greater than .1.
JOURNAL
OF
ONCOLOGY PRACTICE
•
V O L . 11, I S S U E 5
Copyright © 2015 by American Society of Clinical Oncology
Information downloaded from jop.ascopubs.org and provided by at Monash University on November 22, 2015 from 130.194.20.173 Copyright © 2015 American Society of Clinical Oncology. All rights reserved.
6.8
Nausea
Original Contribution
Randomized Controlled Trial of Shared Care for Patients With Cancer Involving General Practitioners and Cancer Specialists
The University of Western Australia; Harry Perkins Institute of Medical Research, The University of Western Australia; School of Primary, Aboriginal and Rural Health Care, The University of Western Australia; School of Medicine and Pharmacology, The University of Western Australia, Crawley; Royal Perth Hospital, Perth; General Practice and Primary Care Academic Centre, University of Melbourne, Carlton; Sir Charles Gairdner Hospital; Haematology Care Centre, Sir Charles Gairdner Hospital, Nedlands, Australia; and University of Oxford, Oxford, United Kingdom
Abstract Purpose: We aimed to determine whether a shared care model (SCM) during chemotherapy treatment improved emotional wellbeing, empowerment, and prevalence of symptoms for people being treated for cancer.
Methods: People receiving chemotherapy for hematologic, breast, ovarian, or colorectal malignancies at two cancer centers were randomly assigned to receive SCM or standard care. The SCM involved a patient-held record, a project coordinator, routine contact between the patient and general practitioner/primary care physician, and primary care physician education. Participants completed the Hospital Anxiety and Depression Scale, the Mini-Mental Adjustment to Cancer, and an empowerment questionnaire before, in the middle of, and on completion of chemotherapy. The presence and severity of adverse effects of chemotherapy were recorded by patients in a symptom diary.
Introduction Australian cancer survival rates are high,1 with 5-year relative survival of 66%.2 Most Australians diagnosed with cancer are referred by their primary care physician (PCP) for definitive diagnosis and treatment by cancer specialists, with little ongoing PCP involvement.3 Specialists manage cancer-specific treatment, and often also coordinate care and provide support, palliation and follow-up. However, oncologists are costly and too few in most Australian public health systems to adequately care for the increasing number of patients,4 and there are concerns about fragmented care5,6 This traditional model fails to meet patient needs in relation to satisfaction, quality of life, empowerment, or morbidity.3 Many patients and caregivers report frustration, dissatisfaction, and inability to access appropriate and timely specialist care.8-10 Consequently, policymakers seek models of care that are cost effective and demonstrably improve psychosocial outcomes.11 Australia has a robust PCP workforce, with approximately 107 PCPs per 100,000 people,12 who can safely provide supportive and follow-up care for patients with cancer.13 InvolveCopyright © 2015 by American Society of Clinical Oncology
Results: Ninety-seven eligible participants were randomly allocated, less than half the intended recruitment. There were no significant differences between the groups for empowerment, symptom prevalence, or Mini-Mental Adjustment to Cancer scores. The proportion with clinical anxiety (Hospital Anxiety and Depression Scale anxiety score of ⱖ 11) decreased over time in both groups (P ⫽ .013) but decreased more in the intervention group (P ⫽ .002). Depression was unchanged over time. Conclusion: Our study was limited by low recruitment and predominance of patients with breast cancer, and was underpowered for the main analyses. Results should therefore be interpreted with caution. Little benefit was seen for SCM in the majority of domains including empowerment, symptom prevalence, and psychological adjustment to cancer. The SCM showed efficacy in clinically anxious patients. Such interventions may be better implemented by using a targeted approach to identify at-need subgroups.
ment of PCPs with specialists in cancer management has resulted in improved continuity of care, greater teamwork, proactive (not reactive) care, and improved patient support.14 Patients, however, report that PCPs are not always well informed about treatments and management of adverse effects.8-10 PCPs are also frustrated about limited communication from specialists and loss of involvement in ongoing care. Contact is usually limited to PCP referrals, specialists’ letters, and discharge summaries. These are frequently late and lacking information on the overall treatment plan and potential treatment adverse effects.15 At some point after treatment is completed, patients return to the care of their PCP, but timing is variable and can occur with limited information about diagnosis, treatment, and survivorship care. Most patients with cancer receive their chemotherapy in outpatient settings, and treatment toxicities commonly develop away from the treatment center, so greater PCP involvement may reduce the severity of toxicities and adverse effects through earlier intervention.16 PCPs may increase the level of psychosocial support for patients during their treatment and in the follow-up phase, which is especially beneficial to patients in
S E P T E M B E R 2015
•
jop.ascopubs.org
349
Information downloaded from jop.ascopubs.org and provided by at Monash University on November 22, 2015 from 130.194.20.173 Copyright © 2015 American Society of Clinical Oncology. All rights reserved.
By Claire E. Johnson, RN, CM, PhD, Christobel M. Saunders, MBBS, FRCS, FRACS, Michael Phillips, Jon D. Emery, DPhil, MBBCh, Anna K. Nowak, MBBS, PhD, FRACP, Kate Overheu, Alison M. Ward, PhD, and David J.L. Joske, MBBS, FRACP, FRCPA
Johnson et al
Methods Ethics approval was obtained from the Human Research Ethics Committees at Sir Charles Gairdner and Royal Perth Hospitals (Ref Nos 2005-037 and RA-06/010A). The study was registered with the Australian New Zealand Clinical Trials Registry (ACTRN 12606000022561). 350
JOURNAL
OF
ONCOLOGY PRACTICE
•
Participants Patient, specialist, and PCP triads were recruited to the study. Written informed consent was obtained from all participants. Hematologists and medical oncologists treating patients at two hospitals in Western Australia were invited to participate. Patients cared for by six consenting specialists were invited to participate. All aspects of the project were explained by the chief investigator (D.J.L.J.). Patients receiving their first round of chemotherapy in participating outpatient clinics were invited to participate. Eligible patients were ⬎ 16 years of age, able to understand and complete trial documentation, and had biopsy-proven malignancy. Four patient streams were recruited: (1) patients with nonHodgkin lymphoma receiving cyclophosphamide, vincristine, doxorubicin, and oral prednisolone; (2) women with stages I and II breast cancer offered four to six cycles of adjuvant chemotherapy; (3) women with stages Ic, II (a-c) and III (a-c) ovarian cancer who had undergone surgical debulking and were offered 6 to 8 cycles of adjuvant chemotherapy with carboplatin and paclitaxel; and (4) patients with cancer of the colon, Dukes B or C, who had undergone surgical resection and offered 6 months chemotherapy with infusional fluorouracil, leucovorin, and oxaliplatin; fluorouracil/leucovorin; or capecitabine. PCPs of consenting patients were contacted by the coordinator to invite participation. To prevent contamination, individual PCPs could have only one patient on study. Intervention group. The SCM involved a PHR, a project coordinator to assist with patient care and information, and PCP educational resource (“upskilling”) packages detailing anticipated adverse effects of each treatment regimen and actions to be taken (Appendix 2). Patients were asked to see their PCP after each chemotherapy treatment. Control group. Usual care was provided by specialists and their associated PCP. PCPs received a letter from the specialist team after each visit, discharge summaries, and telephone communication, where appropriate. Communication frequency and content for standard care was not prescribed by the study. All aspects of the study were explained to consenting patients and PCPs by the coordinator after group allocation was revealed. A comparison of the elements of the intervention and control is shown in Appendix 2.
Study Outcomes The Hospital Anxiety and Depression Scale (HADS)21 measured patient anxiety and depression. HADS scores for each subscale (HADS-A [anxiety] and HADS-D [depression]) were classified as low (0 to 7), borderline clinical (8 to 10), or clinically significant (11 to 21). Empowerment was measured by the Patient Empowerment Scale.22 Coping was measured using the Mini-Mental Adjustment to Cancer Scale (Mini-Mac).23 Each was administered before chemotherapy, midway through treatment, and on completion of treatment. All participants were asked to complete a daily symptom diary to record the presence and severity of six key adverse effects of chemotherapy (tiredness, nausea, sore mouth, poor appetite, numbness and fever,
V O L . 11, I S S U E 5
Copyright © 2015 by American Society of Clinical Oncology
Information downloaded from jop.ascopubs.org and provided by at Monash University on November 22, 2015 from 130.194.20.173 Copyright © 2015 American Society of Clinical Oncology. All rights reserved.
rural areas. Improved co-operation and communication between primary and specialist providers would facilitate timely information, care, and support for patients.8-10 Shared care is an organizational model involving both PCPs and specialists in a formal, explicit manner. Potential advantages include improved quality of life and psychological well-being, reduced hospitalizations, and better medication compliance. Shared care may also overcome difficulties in traditional hospital-based models of cancer care, particularly when patients live in remote areas.17 Shared care models (SCMs) ensure that patients and caregivers have access to both hospital- and community-based supportive care throughout treatment and along the continuum to cure or palliation.18 Cost reductions through fewer hospital attendances without reduction in quality of care have also been reported.17 Patient-held records (PHRs) are an important component of many SCMs. The PHR is a formal, structured record retained by the patient containing patient and health care information. Clinicians and patients can add to the PHR, ensuring better continuity and quality of care19 Randomized controlled trials of PHRs alone have not shown significant improvements in communication, quality of life, or health outcomes for people with cancer.19,20 Reports of multicomponent SCMs that include PHRs in chronic diseases and a systematic review of seven trials of PHRs in patients with cancer show equivocal impact on the patient journey.17,19 However, there are few reports of complex SCMs that include a combination of PHRs and other interventions such as PCP training, opportunities for enhanced communication between health professionals, and access to a patient coordinator for people receiving cancer treatment. Our comprehensive cancer SCM was designed to improve information sharing and PCP knowledge and involvement, and was pilot tested in the hematology department at a tertiary cancer center in Perth, Western Australia (Appendix 1, online only). Based on the outcomes of the pilot, we developed a conceptual framework and refined the SCM (Appendix 2, online only). We evaluated the SCM in patients with hematologic, breast, ovarian, or colorectal malignancies receiving chemotherapy, and compared the model to standard care. Our primary hypothesis was that by using our SCM, people being treated with chemotherapy would have higher levels of emotional well-being and empowerment, lower levels of anxiety and distress, and fewer adverse effects compared with people receiving standard care.
An RCT of Shared Care for Patients With Cancer
Assessed for eligibility (N = 153)
Randomly allocated (n = 100)
Excluded Did not meet inclusion criteria Refused to participate Other reasons Reason: patients missed before start of chemotherapy
Allocated to intervention (n = 52) Received allocated intervention (n = 52) Did not receive allocated (n = 0) intervention
Did not complete study Reasons Lost to follow-up Patients changed chemotherapy Patients chose to withdraw from study
Did not complete study Reasons Patients ceased chemotherapy Patients chose to withdraw from study
(n = 5) (n = 1) (n = 2) (n = 2)
Exited from study (n = 42) Excluded from analysis (n = 1) Reason: patient already had chemotherapy in past, therefore should not have been recruited Analyzed (n = 41 + 5 withdrawn = 46)
(n = 4) (n = 1) (n = 3)
Exited from study (n = 48) Excluded from analysis (n = 1) Reason: patient’s GP was already in the trial, therefore patient should not have been recruited Analyzed (n = 47 + 4 withdrawn = 51)
Figure 1. Cancer shared care trial recruitment flow chart.
identified by clinicians and in the literature24) on a scale of 0 to 10 (0 ⫽ no symptoms, 10 ⫽ the worst I can imagine). Patients could also record other symptoms. On completion of the study, patients, PCPs, and specialists were asked to complete a questionnaire exploring perceptions of the usefulness of the SCM.
Sample Size A sample size of 240 (120 patients per group) was calculated to provide 80% power to detect a reduction from 30% to 15% in the proportion of people showing borderline clinical anxiety or depression on the HADS with an alpha of .05.
Random Allocation Random allocation was performed externally using an opaque sealed envelope method.25 The sample was stratified by cancer type. The baseline questionnaire was performed triple blinded to knowledge of allocation. To implement the intervention, medical oncologists, the coordinator, PCPs, and patients were subsequently made aware of the patient’s allocation.
Statistical Analyses Descriptive methods used arithmetic means with 95% CIs for normally distributed continuous variables and geometric means for skewed and log-normal distributions. Percentages were used to describe categorical and ordinal variables. Associations between continuous dependent variables were assessed using Spearman’s rho statistic because the assumption of normality was rarely satisfied. Copyright © 2015 by American Society of Clinical Oncology
Between-groups comparisons used the likelihood ratio 2 test for categorical variables and longitudinal mixed and random-effects models.26 In mixed models, time was treated as a random effect and randomization group was a fixed effect. Alpha ⬍ .05 was regarded as statistically significant. All analyses were conducted using Stata version 12 (Stata Corporation, College Station, TX).
Results Recruitment was open for 2 years, but stopped early as a result of slow accrual. Forty-six (47%) eligible participants were randomly allocated to the control group and 51 (53%) to the intervention. Nine subjects dropped out of the control group, and seven dropped out of the intervention group (20% and 14%, respectively; P ⫽ .41). Seven subjects left after cycle 1 of their chemotherapy and a further nine after cycle 3 (Figure 1).
Participant Characteristics Most participants were female (n ⫽ 83; 86%), had a diagnosis of breast cancer (n ⫽ 74; 76%), and were from the metropolitan area (n ⫽ 76; 78%). There were no differences between the groups in age, sex, type and stage of cancer, and geographical location, or in measures of psychological well-being or physical symptoms at baseline (Table 1).
Compliance All participants in the intervention group discussed the project with the coordinator and were given the PHR. Four (10%)
S E P T E M B E R 2015
•
jop.ascopubs.org
351
Information downloaded from jop.ascopubs.org and provided by at Monash University on November 22, 2015 from 130.194.20.173 Copyright © 2015 American Society of Clinical Oncology. All rights reserved.
Allocated to control (n = 48) Received allocated intervention (n = 47) Did not receive allocated (n = 1) intervention; reason: patient did not have chemotherapy
(n = 53) (n = 21) (n = 23) (n = 9)
Johnson et al
Table 1. Characteristics of Participants in the Control and Intervention Groups at Baseline Overall Sample Characteristic
No.
%
Control No.
%
Intervention No.
%
Age, years Mean
54.7
Range
28-77
Randomization
53.2
Lower: 53.3, upper: 59.6
Lower: 50.2, upper: 56.2
46
47.4
51
52.6
Sex Female
83
85.6
41
89.1
42
82.4
Male
14
14.4
5
10.9
9
17.7
74
76.3
37
80.4
37
72.6
Cancer type Breast, female Colorectal
2
2.1
1
2.2
1
2.0
Lymphoma
15
15.5
6
13.0
9
17.7
6
6.2
2
4.4
4
7.8
I
15
16.5
5
12.5
10
19.6
II
50
55.0
25
62.5
25
49.0
III
23
25.3
10
25.0
13
25.5
IV
3
3.3
0
0.0
3
5.9
Metropolitan
76
78.4
36
78.3
40
78.4
Rural
21
21.7
10
21.7
11
21.6
None
46
47.9
23
51.1
23
45.1
Borderline†
25
26.0
11
24.4
14
27.5
Clinical‡
25
26.0
11
24.4
14
27.5
None
79
82.3
37
82.2
42
82.4
Borderline†
10
10.4
3
6.7
7
13.7
7
7.3
5
11.1
2
3.9
36
44.4
15
38.5
21
50.0
6
8.7
3
9.1
3
8.3
60
62.5
18
46.2
42
73.7
6
8.0
3
8.3
3
7.7
75
80.7
33
84.6
42
77.8
0
0
0
0
0
0
Ovarian Stage*
Residence
HADS anxiety
HADS depression
Clinical‡ Symptoms Loss of appetite Fever Nausea Numbness Fatigue Mouth ulcers
NOTE. N does not equal 97 for all variables as a result of participant withdrawals. Abbreviations: CL, confidence limit; HADS, Hospital Anxiety and Depression Scale. * Stage is a composite of American Joint Committee on Cancer and Ann Arbor staging for different cancer types. Stage is unknown for six patients in the control group. † Borderline clinical anxiety or depression ⫽ HADS score of 8-10 ‡ Clinically significant anxiety or depression ⫽ HADS score of ⱖ 11.
contacted the coordinator outside of regular clinic visits. More than half (56%) of intervention patients saw their PCP during treatment, compared with 46% of control patients. The average number of visits to the PCP was also greater for the intervention patients (2.79 v 1.61; P ⬍ .001). 352
JOURNAL
OF
ONCOLOGY PRACTICE
•
At baseline, 26% of all participants had clinically significant anxiety levels (ie, HADS-A ⱖ 1121), and another 26% had borderline anxiety levels (HADS-A 8 to 10). Fewer participants were depressed as measured by HADS scores: 7.3% were classified as having clinically significant depression and 10.4% as having borderline clinical depression. Mean anxiety and depression scores were 7.05 and 3.12, respectively. There was a 10% reduction in HADS-A over the duration of enrolment for all patients (P ⬍ .001) but no significant reduction in HADS-D (P ⫽ .137). There was a statistically significant interaction between time and group for HADS-D on longitudinal analysis. This showed an increase in the control group and a slight reduction for the intervention group (P ⫽ .040; Table 2). We found a statistically significant interaction between time and group for clinically anxious patients. The proportion of such patients decreased over time in both groups (P ⫽ .013); the decrease was greater in the intervention group (14 patients [27.5%] at baseline to three patients [6.8%] after the sixth cycle of chemotherapy; P ⫽ .002) versus the control group (11 patients [24.4%] at baseline to five patients [13.9%] after the sixth cycle of chemotherapy; P ⫽ .014). Thus, the intervention resolved clinically significant anxiety, to subclinical levels, more frequently than was seen in the controls. No group difference was detected in the proportion of patients clinically depressed over time, that is: five patients (11.0%) to four patients (8.7%) in the control group (P ⫽ .96) versus two patients (3.9%) to three patients (6.8%) in the intervention (P ⫽ .86).
Psychological Outcomes: Mini-MAC and Empowerment On univariable analysis, no differences were observed between the intervention and control group mean scores for psychological variables (Mini-MAC and Empowerment) at any of the three time points.
Symptom Outcomes Total compliance with completion of symptom diaries was 26% (control group) and 28% (intervention group). Partial compliance (at least one entry missing) was 38% and 45%, respectively, and the remainder did not complete any entries (36% and 27%, respectively). When reported, symptoms and adverse effects of chemotherapy were generally of low intensity (1 or 2 out of 10), with the intervention group scoring slightly more symptoms. The majority of participants reported feeling fatigued, about half reported nausea, and fewer than half reported other symptoms. The number of symptoms reported during any cycle of chemotherapy showed no difference between the two groups (incidence rate ratio ⫽ 1.23; 95% CI, 0.77 to 1.97; P ⫽ .39). When adjusted for covariates, including the cycle of chemotherapy, day of the cycle, and cancer type, participants in the intervention group were more likely to report loss of appetite
V O L . 11, I S S U E 5
Copyright © 2015 by American Society of Clinical Oncology
Information downloaded from jop.ascopubs.org and provided by at Monash University on November 22, 2015 from 130.194.20.173 Copyright © 2015 American Society of Clinical Oncology. All rights reserved.
95% CL
56.5
Psychological Outcomes: HADS
An RCT of Shared Care for Patients With Cancer
Table 2. Mean Psychological Well-Being Scores Over Time, Control Versus Intervention Group Control Group Outcome Measure
Geometric Mean
LCL 95%
Intervention Group UCL 95%
Geometric Mean
LCL 95%
UCL 95%
Anxiety score (HADS) Baseline
.80 7.91
6.50
9.32
8.05
6.71
9.40
6.44
5.40
7.48
6.65
5.64
7.66
5.24
4.26
6.22
5.49
4.54
6.43
3.66
2.92
4.40
4.09
3.31
4.86
Depression score (HADS)
.04
Mid-study
4.16
3.39
4.92
4.06
3.36
4.76
After cycle 6
4.72
3.72
5.72
4.04
3.25
4.83
Anxious preoccupation (Mini-MAC) Baseline
.92 19.5
18.1
20.8
19.8
18.5
21.1
Mid-study
18.5
17.2
19.8
18.8
17.6
20.0
After cycle 6
17.5
16.1
19.0
17.8
16.5
19.2
Cognitive avoidance (Mini-MAC) Baseline
.66 10.1
9.51
10.7
9.45
8.89
10.0
Mid-study
10.2
9.66
10.7
9.61
9.15
10.1
After cycle 6
10.2
9.55
10.9
9.77
9.18
10.4
Fighting spirit (Mini-MAC) Baseline
.35 20.8
19.9
21.8
20.7
19.8
21.6
Mid-study
20.4
19.5
21.3
20.6
19.8
21.4
After cycle 6
20.0
19.0
21.0
20.5
19.5
21.4
15.2
13.9
16.5
15.1
13.9
16.3
Hopelessness/helplessness (Mini-MAC) Baseline
.40
Mid-study
15.3
14.1
16.4
14.8
13.7
15.8
After cycle 6
15.3
13.9
16.7
14.4
13.2
15.6
12.1
10.4
13.7
12.1
10.4
13.7
Patient empowerment scale Baseline
.47
Mid-study
11.6
10.0
13.2
11.6
10.0
13.2
After cycle 6
13.0
11.7
14.3
13.0
11.7
14.3
Abbreviations: HADS, Hospital Anxiety and Depression Scale; Mini-MAC, Mini-Mental Adjustment to Cancer; LCL, lower confidence limit; UCL, upper confidence limit.
(P ⫽ .004) and nausea (P ⫽ .012; Appendix Table A1, online only). A subgroup analysis of patients with breast cancer showed similar results to the sample as a whole only for the main outcomes (Appendix 3, online only).
tected in perceptions of care between the intervention and control groups.
Emergency Department/Hospital Admissions
Previous experience with PHRs in cancer populations has been equivocal.19 Models of shared care for people with other chronic conditions have reported few differences in mental health, physical outcomes, and psychosocial outcomes between intervention and control groups.17 Our SCM consisted of a PHR, upskilling resources for PCPs, routine contact between the patient and their PCP, and the opportunity for the PCP to liaise with the project coordinator. Overall, our results are consistent with previous findings, showing little impact on psychological variables as measured by HADS,21 the Patient Empowerment Scale,22 and MiniMAC23 instruments, nor on self-reported symptoms. Indeed, the slight increase in reported symptomatology in the intervention group could suggest increased focus by patients on their symptoms. The intervention had no significant effect on patients’ self-reported mood or on overall levels of anxiety, em-
We recorded emergency department presentations and hospital admissions. There was no difference between the intervention and control groups (Appendix 4, online only).
Participants’ Perceptions of Care and the SCM Participants’ responses to the SCM are reported in Appendix 4 (online only). One of five cancer specialists who completed the exit survey thought the PHR was useful, and all had reservations about the usefulness of other aspects of the SCM. The majority of the PCPs in the intervention (88%) reported that they found the PHR useful and the majority of open comments about the intervention were positive. There were no differences detected in perceptions of care between PCPs in the intervention and the control groups. Among patients, there were no differences deCopyright © 2015 by American Society of Clinical Oncology
Discussion
S E P T E M B E R 2015
•
jop.ascopubs.org
353
Information downloaded from jop.ascopubs.org and provided by at Monash University on November 22, 2015 from 130.194.20.173 Copyright © 2015 American Society of Clinical Oncology. All rights reserved.
Mid-study After cycle 6
Baseline
P for Comparison of Groups Over Time
Johnson et al
Limitations Caution should be exercised in generalizing our results because of potential limitations. First, the sample size was small and was mostly women (85%) with breast cancer (75%). The predominance of patients with breast cancer related to the higher volume of such cases at the trial sites and the late involvement of colorectal cancer services. Early closure meant our target sample size was not met. Factors contributing to slow recruitment included changes in clinical practice (increased treatment of patients with cancer in rural and regional centers) and trial fatigue due to concurrent chemotherapy trials, in colorectal cancer particularly. The intended statistical power of the study was not achieved. Second, it is unclear to what extent participants experienced physical symptoms or adverse effects of treatment because compliance with daily symptom diaries was low. Third, oncology specialists’ commitment to the intervention was limited, with some patients reporting underuse of the PHR by their specialist. This was compounded by other concurrent trials and workload pressures. 354
JOURNAL
OF
ONCOLOGY PRACTICE
•
Barriers to successful SCM implementation included patient trial fatigue, limited engagement with the PHR by patients and specialists, and lack of specialist buy-in to the model. Alternative methods of information sharing are being studied, such as case conferences,30 faxed treatment regimes,31 structured information packs,32 and electronic health records.33 Reported benefits include improved sharing of information, increased collaboration, better preparation of PCPs to provide care, more coordinated care, and higher PCP satisfaction with the secondary provider. However, translation of these interventions into routine practice is slow, with many programs dependent on charismatic leadership rather than systematic implementation.34
Summary In summary, we found a lack of even a trend for improved psychosocial functioning, empowerment, or symptom control with our SCM, and we urge caution in its wide application. Indeed, this study adds further evidence regarding the limited usefulness of the PHR during chemotherapy in patients with cancer.19 Our findings support data suggesting that shared cared interventions are better targeted toward patient groups identified as being most likely to benefit,35 (eg, patients who are clinically anxious). A comprehensive SCM may be most valuable when implemented with distress or anxiety screening. However, because we found that the outcomes of the SCM were not inferior to those of usual care, the use of shared care may be beneficial if such a model was more cost efficient and accessibility for patients was improved. Further research is warranted to explore these possibilities. Acknowledgment Supported by the National Health and Medical Research Council, Australia Grant No. 353678. Previously presented in part at Change Champions: Innovations in the Management of Cancer Services, Melbourne, Australia, March 2006; and Achieving Excellence in WA Health Conference, Perth, Australia, October 2006. We acknowledge the patients and health professionals who participated in this study. We also thank Nick De Klerk, Andy Redfern, and Caroline Bulsara for their contributions. Authors’ Disclosures of Potential Conflicts of Interest The authors indicated no potential conflicts of interest.
Author Contributions Conception and design: Christobel M. Saunders, Jon D. Emery, Kate Overheu, Alison M. Ward, David J.L. Joske Administrative support: Kate Overheu Collection and assembly of data: Anna K. Nowak, Kate Overheu, David J.L. Joske Data analysis and interpretation: Claire E. Johnson, Christobel M. Saunders, Michael Phillips, Jon D. Emery, Anna K. Nowak, Alison M. Ward, David J.L. Joske Manuscript writing: All authors Final approval of manuscript: All authors Corresponding author: Claire E. Johnson, RN, CM, PhD, School of Surgery, University of Western Australia, M507, 35 Stirling Hwy, Crawley 6009 Australia; e-mail: [email protected].
DOI: 10.1200/JOP.2014.001569; published online ahead of print at jop.ascopubs.org on March 24, 2015.
V O L . 11, I S S U E 5
Copyright © 2015 by American Society of Clinical Oncology
Information downloaded from jop.ascopubs.org and provided by at Monash University on November 22, 2015 from 130.194.20.173 Copyright © 2015 American Society of Clinical Oncology. All rights reserved.
powerment, or depression. This may reflect a wider issue that such measures are not the most appropriate or sensitive to detect benefits or harms of alternative models of cancer care. The low level of engagement with the PHRs and contact with the coordinator by both patients and clinicians is consistent with other trials. In a systematic review of PHRs in cancer care, Gysels et al19 documented similar difficulties with the implementation of a PHR to those we encountered. We also found the PHR to be ineffective at improving communication between health care providers, or recognizing unmet needs of patients. Some 26% of our sample were clinically anxious at the beginning of treatment (HADS-A scores ⱖ 1121), higher than reported in previous research with patients with cancer or the general population. In a study of 1,529 patients with cancer and 2,037 people representative of the German population, 21% of patients with cancer were clinically anxious compared with 6% in the population sample.27 Women were more anxious than men for most cancer types. In two studies of women with breast cancer, 16% had clinically significant anxiety in hospital postsurgery,28 and 18% were anxious at diagnosis.29 In both, the proportion of patients with anxiety decreased with time, although the rate of decrease varied considerably. The higher rate of anxiety at baseline in our study may reflect that the majority of patient participants were women with breast cancer and that assessment was done immediately before chemotherapy. As with previous research, anxiety levels in our participants decreased over time, but participants with elevated anxiety scores in the intervention group showed greater resolution to normal levels than those in the control group. Analyzing the symptom diaries, participants in the intervention group reported some symptoms more frequently than participants in the control group. However, we stress that these were reported at very low levels and are unlikely to be clinically meaningful. Therefore, regular visits to the PCP postchemotherapy may not be warranted.
An RCT of Shared Care for Patients With Cancer
References 1. Coleman MP, Forman D, Bryant H, et al: Cancer survival in Australia, Canada, Denmark, Norway, Sweden, and the UK, 1995-2007 (the International Cancer Benchmarking Partnership): An analysis of population-based cancer registry data. Lancet 377:127-138, 2011 2. Australian Institute of Health and Welfare: Cancer Survival and Prevalence in Australia: Period Estimates From 1982 to 2010. Cancer Series No. 69 (Cat. No. CAN 65 2012). Canberra, Australia, Australian Institute of Health and Welfare 3. Clinical Oncological Society of Australia, The Cancer Council Australia, and National Cancer Control Initiative: Optimising Cancer Care in Australia. Melbourne, Australia, National Cancer Control Initiative, 2003
5. Royal College of Physicians and Royal College of Radiologists: Cancer Patients in Crisis: Responding to Urgent Needs. Report of a Working Party. London, United Kingdom, Royal College of Physicians, 2012 6. The Senate Community Affairs References Committee: The cancer journey: Informing choice. Report on the inquiry into services and treatment options for persons with cancer, 2005. Canberra, Australia, Commonwealth of Australia, pp 125-136
19. Gysels M, Richardson A, Higginson IJ: Does the patient-held record improve continuity and related outcomes in cancer care: A systematic review. Health Expect 10:75-91, 2007 20. Ko H, Turner T, Jones C, et al: Patient-held medical records for patients with chronic disease: A systematic review. Qual Saf Health Care 19:1-7, 2010 21. Zigmond AS, Snaith RP: The hospital anxiety and depression scale. Acta Psychiatr Scand 67:361-370, 1983 22. Bulsara C, Style I, Ward AM, et al: The psychometrics of developing the patient empowerment scale. J Psychosoc Oncol 24:1-16, 2006 23. Watson M, Homewood J: Mental Adjustment to Cancer Scale User’s Manual. Sutton, United Kingdom, The Royal Marsden Hospital, 2008 24. Shapiro CL, Recht A: Side effects of adjuvant treatment of breast cancer. N Engl J Med 344:1997-2008, 2001 25. Egger M, Smith G, Altman D (eds): Systematic Reviews in Health Care Meta-analysis in Context. London, United Kingdom, BMJ Books, 2001
7. Reference deleted
26. Fitzmaurice G, Laird N, Ware J: Applied Longitudinal Analysis. Hoboken, NJ, John Wiley, 2004
8. King M, Jones L, Richardson A, et al: The relationship between patients’ experiences of continuity of cancer care and health outcomes: A mixed methods study. Br J Cancer 98:529-536, 2008
27. Hinz A, Krauss O, Hauss JP, et al: Anxiety and depression in cancer patients compared with the general population. Eur J Cancer Care 19:522-529, 2010
9. Kleeberg UR, Feyer P, Günther W, et al: Patient satisfaction in outpatient cancer care: A prospective survey using The PASQOC questionnaire. Support Care Cancer 16:947-954, 2008
28. Mehnert A, Koch U: Prevalence of acute and post-traumatic stress disorder and comorbid mental disorders in breast cancer patients during primary cancer care: A prospective study. Psycho-oncology 16:181-188, 2007
10. Sandoval GA, Brown AD, Sullivan T, et al: Factors that influence cancer patients’ overall perceptions of the quality of care. Int J Qual Health Care 18:266274, 2006
29. Stafford L, Judd F, Gibson P, et al: Screening for depression and anxiety in women with breast and gynaecologic cancer: Course and prevalence of morbidity over 12 months. Psycho-oncology 22:2071-2078, 2013
11. National Health and Hospitals Reform Commission: A Healthier Future for All Australians - Final Report of the National Health and Hospitals Reform Commission. 2009. Canberra, Australia, Commonwealth of Australia, 2009, p 107
30. Mitchell G, Cherry M, Kennedy R, et al: General practitioner, specialist providers case conferences in palliative care—Lessons learned from 56 case conferences. Australian Family Physician, 34(5): 389-392, 2005
12. Australian Institute of Health and Welfare: What types of medical practitioners are there? www.aihw.gov.au/workforce/medical/types-of-medical-practitioners/ 13. Emery JD, Shaw K, Williams B: The role of primary care in early detection and follow-up of cancer. Nat Rev Clin Oncol 11:38-48, 2014
31. Jefford M, Baravelli C, Dudgeon P, et al: Tailored chemotherapy information faxed to general practitioners improves confidence in managing adverse effects and satisfaction with shared care: Results from a randomized controlled trial. J Clin Oncol 26:2272-2277, 2008
14. Murray SA, Boyd K, Campbell C, et al: Implementing a service users’ framework for cancer care in primary care: An action research study. Fam Pract 25:7885, 2008
32. Kousgaard KR, Nielsen JD, Olesen F, et al: General practitioner assessment of structured oncological information accompanying newly referred cancer patients. Scand J Prim Health Care 21:110-114, 2003
15. McConnell D, Butow PN, Tattersall MH: Improving the letters we write: An exploration of doctor-doctor communication in cancer care. Br J Cancer 80:427437, 1999
33. Schoen C, Osborn R, Squires D: A survey of primary care doctors in ten countries shows progress in use of health information technology, less in other areas. Health Aff 31:2805-2816, 2012
16. Roorda C, de Bock GH, van der Veen WJ, et al: Role of the general practitioner during the active breast cancer treatment phase: An analysis of health care use. Support Care Cancer 20:705-714, 2012
34. Gardiner C, Gott M, Ingleton C: Factors supporting good partnership working between generalist and specialist palliative care services: A systematic review. Br J Gen Pract 62:e353-e362, 2012
17. Smith SM, Allwright S, O’Dowd T: Does sharing care across the primary “specialty interface improve outcomes in chronic disease? A systematic review. Am J Manage Care 14:213-224, 2008
35. Nielsen JD, Palshof T, Mainz J, et al: Randomised controlled trial of a shared care programme for newly referred cancer patients: Bridging the gap between general practice and hospital. Qual Saf Health Care 12:263-272, 2003
Copyright © 2015 by American Society of Clinical Oncology
S E P T E M B E R 2015
•
jop.ascopubs.org
355
Information downloaded from jop.ascopubs.org and provided by at Monash University on November 22, 2015 from 130.194.20.173 Copyright © 2015 American Society of Clinical Oncology. All rights reserved.
4. Blinman PL, Grimison P, Barton MB, et al: The shortage of medical oncologists: The Australian Medical Oncologist Workforce Study. Med J Aust 196:58-61, 2012
18. Joske D, McGrath P: Implementing the holistic philosophy in cancer care: New directions for the Sir Charles Gairdner Hospital. Austral-Asian J Cancer 2:33-38, 2002
Johnson et al
Appendix
Table A1. Frequency of Participants Reporting Any Level of Symptoms by Group Across All Cycles: Results of a Longitudinal Negative Binomial Regression Analysis Symptom
Control %
Intervention %
IRR*
LCL 95%
Appetite loss
34.2
49.2
2.46
1.36
Fever
UCL 95% 4.55 10.8
P (H0: IRR ⴝ 1) .004
11.1
2.17
0.436
42.4
50.6
1.74
1.13
2.68
.012
.34
Numbness
23.0
24.8
1.83
0.868
3.86
.11
Fatigue
86.4
85.5
1.28
0.993
1.65
.057
Mouth ulcers
37.4
39.0
1.38
0.780
2.44
.27
Abbreviations: IRR, incidence rate ratio; LCL, lower confidence limit; UCL, upper confidence limit. * The IRR estimate is adjusted for cycle of chemotherapy, day of cycle, cancer type, place of residence, sex, and stage for each symptom if the P value for the covariate was greater than .1.
JOURNAL
OF
ONCOLOGY PRACTICE
•
V O L . 11, I S S U E 5
Copyright © 2015 by American Society of Clinical Oncology
Information downloaded from jop.ascopubs.org and provided by at Monash University on November 22, 2015 from 130.194.20.173 Copyright © 2015 American Society of Clinical Oncology. All rights reserved.
6.8
Nausea
