Introduction

The number of cancer survivors worldwide is steadily increasing. It has been usual practice for survivorship care to be specialist-led, usually within the cancer center where primary treatment was delivered [1]. Because of the growing prevalence of cancer survivors and their specific care needs, this traditional model of specialist and cancer center survivorship care is no longer sustainable [2].

Several different models of survivorship care have been proposed [3]. There is evidence that survivorship care can be provided by specialists, primary care physicians (PCP), nurses, or by sharing the care among a multidisciplinary team [4–6].

Cancer organizations worldwide [7] promote the use of survivorship care plans (SCPs). SCPs are purported to facilitate a seamless transition from active cancer care to survivorship care, specifically as it relates to the transition to the primary care setting [8]. Although many articles outline the potential benefits of SCPs, research on SCPs to date has been largely theoretical [8–10]. A smaller number of studies have evaluated the impact of SCPs on processes of care, health service, and patient-reported outcomes [11–13]. We conducted the first randomized clinical trial (RCT) evaluating a SCP. The objective of the RCT was to determine if a SCP for breast cancer survivors ready for transition from oncologist care to care with their own PCP improved outcomes [14]. We hypothesized that patient-reported outcomes would be positively affected by the implementation of a SCP. By delivering the SCP as teaching strategy, patients will be involved in the problem solving process, and they might be able to manage their psychosocial needs and have a sense of control over one’s life. Moreover, we hypothesized that a SCP would increase recommended medical surveillance test rates and decrease rates of non-recommended testing. Results of the first planned analysis with outcomes measured at 12 months post-randomization showed that the implementation of a SCP achieved no statistically significant improvement on any of the patient-reported or health service outcomes [20] and was not cost-effective [15]. Here, we report the final results of the trial with health service and patient-reported outcomes measured at 24 months. By evaluating health service and patient-reported outcomes, repeated observations are needed to uncover the influence of incidental behaviors or occasionally circumstances.

Methods

The setting of the RCT was specialized cancer centers throughout Canada. Details of the trial protocol and 12-month outcomes are reported elsewhere [14]. The study was approved by the ethics committees of each participating center.

Study population

Four hundred and eight patients were enrolled through nine specialized cancer centers, representing a 64 % participation rate. Eligible patients were those with early-stage breast cancer under active routine follow-up who were without recurrence or new primary cancer, having completed primary adjuvant treatment at least 3 months before entry into the study, and had a PCP to provide follow-up care. Patients were excluded if they had persistent complications of primary treatment, were previously enrolled in a study that required continued oncology follow-up, were actively observed for another primary cancer, or had a PCP who already had a patient enrolled in the study.

Design and procedures

Eligible patients were randomly assigned to PCP-led routine follow-up care without a SCP (control group) or with a SCP (intervention group) in a 1:1 ratio. Patients in both groups had a standard discharge visit with their oncologist and were informed that responsibility for continued routine follow-up care was now transferred to their PCP, and all PCPs were sent a discharge letter. In the intervention group, patients also received a 30-min educational session with a nurse, a comprehensive SCP [8] consisting the prescribed key elements including a personalized treatment summary, a patient’s version of the Canadian follow-up guidelines [16], and a resource kit tailored to the patient’s needs on available supportive care resources. These documents were also sent to the patient’s PCP together with the full clinical practice guideline on follow-up care [17], a summary of the guideline, and a follow-up visit reminder table. Center-specific stratified randomization was performed along with stratification for time after diagnosis: <24 months after diagnosis (n = 180) and ≥24 months after diagnosis (n = 228). Patients completed questionnaires at baseline, 3, 6, 12, 18, and 24 months, and brief telephone interviews at 9, 15, and 21 months.

Outcomes

Health service outcomes

  1. 1.

    Adherence to the guideline on follow-up care was assessed in each questionnaire and telephone interview. Patients were asked to record the frequency and types of follow-up visits to the PCP, the oncologist, and the frequency and types of diagnostic and screening measurements.

  2. 2.

    Patient knowledge of which physician was primarily responsible for follow-up care was assessed at baseline, which was the last follow-up appointment with the oncologist, and at each questionnaire.

  3. 3.

    Reasons for post-discharge cancer center visits were assessed in each questionnaire and telephone interview.

Patient-reported outcomes

Levels of cancer-specific distress, psychological distress, health-related quality of life, patient satisfaction, and continuity and coordination of care were assessed by the Impact of Event Scale (IES) [18, 19], the Profile of Mood States questionnaire (POMS) [20], the Physical and Mental Component SF-36 summary scales (PCS, MCS) [21, 22], the Medical Outcomes Study-Patient Satisfaction Questionnaire (MOS-PSQ) [23], and the Continuity and Coordination of Care Questionnaire (CCCQ) [24–26].

Statistical methods

The sample size calculation and the statistical methods for the primary (the change in IES total score) and secondary endpoints of patient-reported outcomes (POMS, PCS, MCS, MOS-PSQ, and CCCQ questionnaires) have been described in detail previously [14]. A linear mixed model was used to determine intervention group effects on primary and secondary patient-reported outcomes due to time from diagnosis. Mean change scores from baseline and between-group differences were calculated for all the patient-reported outcomes. Score changes of >1 point from baseline (5 % of the scale breadth) were considered as potentially clinically meaningful [27]. The adherence to the guideline was calculated according to the procedures of Hutchison et al. [23]. The number of recommended maneuvers (i.e. clinical examination and breast imaging—mammograms, breast magnetic resonance imaging (MRI), or breast ultrasound) and the number of non-recommended maneuvers (i.e. routine imaging with bone scan, computed tomography (CT) scan, chest X-ray, abdominal ultrasound, or MRI scan) were calculated based on the clinical practice guideline [17]. Since the Canadian clinical guideline does not recommend any particular frequency of clinical breast examinations, the ASCO guideline was used [28]. The adherence score for each patient was calculated as the number of recommended maneuvers minus the number of the non-recommended maneuvers over 24 months. A multivariable logistic regression analysis was performed to determine the association of factors prognostic for adherence to the guideline (defined as at least two breast imaging tests over the 24 months). Forward stepwise selection was used to identify factors prognostic for adherence to guidelines. The frequency and reasons for post-discharge cancer center visits were compared between groups using the Fisher’s exact test. The proportion of patients who correctly identified their PCP as primary responsible for follow-up care was calculated using the number of patients reporting at least one physician responsible in the denominator. The difference between groups was estimated by using the continuity-corrected Wilson score statistic. All statistical analyses were performed with SAS version 9.1 or R version 2.7.1.

Results

Patients

Of 408 women enrolled in the trial, 337 were followed ≥18 months after randomization and are included in this analysis (CONSORT diagram; Supplemental File 1). These patients had longer time from primary treatment to study randomization (p = 0.041), were more likely to be married/cohabiting (p = 0.006), and more likely to be from Quebec or Western Canada (p = 0.026) than patients who were not followed for at least 18 months (Supplemental File 2). Baseline characteristics were balanced between the control and intervention groups (Table 1).

Table 1 Baseline characteristics by group
Full size table

Health service outcomes

Adherence to guidelines

Over the 24 months, 230 patients had at least two breast imaging tests: 113 (68.9 %) in the intervention and 117 (67.6 %) in the control group (p = 0.82). In contrast, 45 (13.5 %) women had no breast imaging tests: 19 (11.6 %) in the intervention group and 26 (15 %) in the control group (p = 0.42). No difference between the control and intervention group was observed in the total adherence score (median = 3, p = 0.43) (Table 2). Education and tumor grade had a substantial proportion of patients with missing data. Neither was significant in the univariate model or the initial multivariate model based on patients with complete data. Hence, these two factors were excluded, and the forward selection process was performed using only the remaining variables. In the multivariable analyses, geographic region (p < 0.001), time from completion of primary treatment (p = 0.013), and SF-36 mental component score (p = 0.044) were prognostic for adherence with women from Quebec, those with <2 years from completion of primary treatment and those with higher SF-36 mental component scores all showing better adherence scores (Table 3).

Table 2 Adherence to guidelines
Full size table
Table 3 Potential prognostic factors for adherence to guidelines
Full size table

Patient knowledge of physician primarily responsible for follow-up

At 24 months, no difference between the intervention and control groups was detected for correctly identifying their PCP as primarily responsible for follow-up; 94.0 vs. 88.4 %, p = 0.10 and 91.7 vs. 92.0 %, p = 1.0, respectively (data not shown).

Post-discharge cancer center visits

There were no differences in the frequency or reasons for post-discharge cancer center visits between groups. Overall, ≤3 % of patients returned to oncologist care for routine follow-up (Supplemental File 3).

Patient-reported outcomes

No differences between groups were observed in change from baseline to 24 months for cancer-specific distress, psychological distress, health-related quality of life, patient satisfaction, or continuity and coordination of care (Fig. 1). Within strata fluctuations (diagnosed <24 or ≥24 months) were small, varying between +0.31 and −0.24 (of 20 points) in the intervention group and −0.21 and +0.24 in the control group. The net between-group difference was +0.52 in patients diagnosed at <24 months and −0.48 in patients diagnosed ≥24 months and did not come to the one-point threshold.

Fig. 1
figure 1

Trial outcomes: change scores over time since baseline. Red line, survivorship care plan (SCP) group, black line, no SCP. CCCQ Continuity and Coordination of Care Questionnaire, IES Impact of Events Scale, POMS Profile of Mood States Questionnaire, PSQ Patient Satisfaction Questionnaire, PCS, MCS Physical and Mental Component SF-36 summary

Full size image

Discussion

We previously reported the 12-month results of a RCT evaluating a SCP for breast cancer patients [14]. Here, we report the results extended to 24 months including the outcome of adherence to guidelines, which has not been reported previously. The 24-month assessments were needed to uncover the influence of incidental behaviors or occasionally circumstances on health service and patient-reported outcomes. Moreover, the 24-month assessments represent the long-term benefit and sustainability of the implementation of a SCP. Neither the 12- nor the 24-month analysis of this RCT supports the hypothesis that the implementation of a SCP improves adherence to guidelines, improves patient knowledge regarding which physician is primary responsible for follow-up, reduces the number of post-discharge cancer center visits, or improves patient-reported outcomes over the control condition.

Adherence to guidelines and, more particularly, the adherence to recommended breast imaging tests were evaluated in this study. The hypothesis was that a SCP would increase recommended medical surveillance test rates and decrease rates of non-recommended testing [29]. However, in this study, we found no difference in the number of recommended and non-recommended surveillance tests between both groups. Although most women in our study population had regular follow-up visits with their PCP and regular breast imaging tests during 24-month follow-up, approximately 15 % of them had fewer than recommended breast imaging. This finding is consistent with the results of our previous population-based study showing a similar magnitude of underuse of surveillance breast imaging in breast cancer survivors, despite over 80 % being under oncologist-led follow-up care [1]. While there is no compelling evidence to support the frequency and intensity of follow-up visits in breast cancer patients, there is evidence to support routine breast imaging [30, 31]. Adherence rates were observed to be higher in patients from Quebec, patients within 2 years of primary treatment, and in patients with better SF-36 mental component scores.

A limitation of this study is that the adherence score is based on patient self-report. Although self-reports have been shown to be accurate [32–34], the accuracy of the data or the objectivity of patients’ responses cannot be guaranteed. Our data do not allow us to confirm whether tests were ordered for surveillance or for the evaluation of symptoms. Accordingly, there might be some misclassification of tests as routine when they were in fact diagnostic. However, this is likely to be balanced between the intervention and control group. Moreover, questions were framed so that women could establish if the test was ordered as routine or diagnostic.

At baseline, despite being under active oncologist-led follow-up care, one third of patients considered their PCP to be the health care professional primarily responsible for their breast cancer follow-up care. This is consistent with earlier studies [35] which showed a lack of clarity in the patient’s mind regarding responsibility of the involved health care providers for routine follow-up care, potentially resulting in an underuse, overuse, or inefficient use of follow-up care resources [1, 36]. However, at 24 months post-discharge, 92 % of patients in both groups were able to correctly identify their PCP as primarily responsible for follow-up. Additionally, ≤3 % of patients visited their oncologist for routine follow-up after transfer to the PCP, indicating that both patients and PCPs maintained their willingness to adhere to PCP-led follow-up. Similarly, there was no evidence that psychosocial variables, patient satisfaction, or continuity and coordination of care improved with the implementation of a SCP in these women. However, our study results do not inform as to whether SCPs could be beneficial for patients with high levels of psychological distress or whether the SCP will be implemented in other cancer groups. Patients in this study had relatively low scores of distress at baseline, and it is possible that breast cancer patients are relatively better informed in comparison to other cancer groups. Moreover, the IES may not have been sensitive enough to capture intervention effects in patients beyond 10 years after primary treatment. However, results were analyzed by strata, and there was no clinically meaningful effect observed related to time from diagnosis. Although, our study results are similar to those found in a randomized controlled trial of a clinic-based survivorship intervention following adjuvant therapy in breast cancer [13]. While, there were differences in study design between both studies; the patient-reported outcome measures used in these studies were similar.

In this study, patients were referred from a specialized oncology setting to a primary care setting. These study results are generalizable to practice settings where the oncologist has a “treatment ending” visit and a setting where PCPs are available and willing to accept cancer survivors. Additionally, this study does not inform as to whether SCPs could be beneficial for improvement of PCP knowledge about survivorship care, which remains an important topic for further research.

Conclusion

The implementation of a SCP in the transition of care did not contribute to improvements in adherence to guidelines, to health service, or to patient-reported outcomes. In both intervention and control groups, there was a high level of adherence to follow-up guidelines and a small number of post-discharge cancer center visits, indicating that a standard discharge visit with the oncologist appears to achieve similar objectives as the SCP. The suggestions that a SCP may address a number of deficits in the complexity of survivorship care planning are based on a theoretic approach. Currently, several institutes are implementing survivorship care plans in clinical practice. However, results of this RCT should raise questions concerning the utility of a SCP in this patient population. The essential elements necessary for survivorship care and determination of which patients are likely to benefit should be further investigated before a costly [15] spread of implementing survivorship care plans in clinical practice.