Survivorship care for cancer patients in primary versus secondary care: a systematic review

Background Cancer survivorship care is traditionally performed in secondary care. Primary care is often involved in cancer management and could therefore play a more prominent role. Purpose To assess outcomes of cancer survivorship care in primary versus secondary care. Methods A systematic search of MEDLINE and EMBASE was performed. All original studies on cancer survivorship care in primary versus secondary care were included. A narrative synthesis was used for three distinctive outcomes: (1) clinical, (2) patient-reported, and (3) costs. Results Sixteen studies were included: 7 randomized trials and 9 observational studies. Meta-analyses were not feasible due to heterogeneity. Most studies reported on solid tumors, like breast (N = 7) and colorectal cancers (N = 3). Clinical outcomes were reported by 10 studies, patient-reported by 11, and costs by 4. No important differences were found on clinical and patient-reported outcomes when comparing primary- with secondary-based care. Some differences were seen relating to the content and quality of survivorship care, such as guideline adherence and follow-up tests, but there was no favorite strategy. Survivorship care in primary care was associated with lower societal costs. Conclusions Overall, cancer survivorship care in primary care had similar effects on clinical and patient-reported outcomes compared with secondary care, while resulting in lower costs. Implications for cancer survivors Survivorship care in primary care seems feasible. However, since the design and outcomes of studies differed, conclusive evidence for the equivalence of survivorship care in primary care is still lacking. Ongoing studies will help provide better insights.


Background
To date, the number of patients with incident cancer and cancer survivors is increasing, due to an aging population and the improvements in cancer screening, diagnosis, and treatment. Cancer survival has increased to over 60% between 2010 and 2020, as previously predicted by The Dutch Cancer Society [1]. In numerous countries worldwide, including the Netherlands, patients treated for cancer are initially included in a secondary care-based follow-up program, mainly focusing on the early detection of recurrences and treatment of symptoms caused by the cancer or its treatment. However, survivorship care for cancer encompasses not only detection of recurrences but also the attention to rehabilitation (psychological and social support, integration in society and secondary prevention) as addressed by the Institute of Medicine (IOM) back in 2006 [2].
Following completion of cancer treatment, many patients experience unmet needs and symptoms [3] and primary care is often involved in management of these needs and symptoms, especially for the older population with comorbidities [4][5][6][7][8][9]. A more general approach is therefore likely to be favorable to patient outcomes [10]. Traditional core values of primary care, such as the continuity and coordination of care, can lend themselves for the improvement of cancer survivorship, but the role of primary care may vary depending on context and setting [10,11].
Several reviews have been published on alternative survivorship care strategies, such as GP-, PCP-, nurse-led, patient-initiated, and shared care [12][13][14][15][16]. However, none have focused exclusively on survivorship care by physicians working in primary care. The aim of this systematic review is to provide an overview of the outcomes of survivorship care in primary-compared with secondary-based care.

Study design and search strategy
In February 2020, a systematic search was performed in MEDLINE and EMBASE to identify original studies on cancer survivorship care. General terms for survivorship care, including follow-up and aftercare, were used. In addition to the MEDLINE and EMBASE search, reference checking was performed to identify possible other relevant publications. (See Appendix 1 for the search strategy.)

Eligibility, selection, and data extraction
Original studies comparing cancer survivorship care in primary to secondary-based care were included. As health care systems differ around the globe, generalist professions providing primary-or community-based care, such as a general practitioner (GP), primary care physicians (PCPs), and family physicians (FPs), were included in this review. Studies reporting on patients of any age who were (curatively) treated for any type or stage of cancer were eligible. No restrictions were made on the type of outcomes. Economic evaluations of cancer survivorship care programs were also considered for inclusion. Studies on shared care and patient or physician preferences for survivorship care were excluded from this review.
All studies were screened on title and abstract by two independent researchers (JV and TW). Subsequently, complete texts were read to ensure inclusion criteria, and data were extracted. Data extraction was performed by one researcher (JV) based on a predefined data format. Disagreement between the two researchers on study selection and data extraction was resolved by discussion or, if necessary, by consulting a third independent researcher (KvA).

Data analysis
As we intended a broad and conclusive review, no restrictions were made on the type of patient, outcomes, or methodology, which resulted in substantial heterogeneity of studies. Therefore, meta-analyses were not feasible and a narrative synthesis was used.
Outcomes were grouped into three distinctive categories: (1) clinical outcomes as measured by medical records (including survival, serious clinical events, and documented followup care), (2) patient-reported outcomes as measured by patient questionnaires and interviews (including quality of life, symptoms, patient satisfaction, and self-reported receipt of survivorship care), and (3) costs of survivorship care programs (including societal and patient costs).

Quality assessment
A risk of bias analysis was performed for all included studies according to the designated quality assessment tools as advised by the Cochrane collaboration. The consort instrument was used for randomized clinical trials [17], and the ROBINS-I for (non-randomized) observational studies [18].

Study selection
The systematic search retrieved 1766 original studies (Fig. 1). Reference checking did not identify any additional studies. After title and abstract screening, full text of 42 studies was reviewed. Based on the predefined eligibility criteria, 16 studies were included in this review. Figure 1 illustrates the selection process.

Quality assessment
Risk of bias assessment revealed low risk of bias in 10 studies, intermediate in 3, and high risk of bias in 3 out of 16 studies (see Appendix Table 5). Risk of bias was often inherent to the design of the study, including selection, misclassification, recall, and interviewer bias. Table 1 shows the baseline characteristics of the included studies. Seven randomized controlled trials (RCTs) were included in this review [19][20][21][22][23][24][25]. Three studies of Grunfeld et al. were based on the same RCT, but reported on separate outcomes [20][21][22]. Other included studies were based on a type of observational study [26][27][28][29][30][31][32][33][34]. Most studies reported on patients with solid tumors, such as breast and colorectal cancers. The number of patients ranged from 98 in a retrospective cohort study [26] to 5009 in a quasi-experimental observational study [29]. Six studies reported on physicians working in primary care, of which two studies did not further specify the provider type [31,32]. The length of follow-up ranged from 1 to 15 years.

Clinical outcomes
Ten studies reported on clinical outcomes (see Table 2). No important differences were seen in survival between follow-up strategies after 3 up to 15 years of follow-up [30,33,34]. Follow-up in secondary care was associated with shorter relapse-free survival (RFS) and higher likelihood of receiving palliative treatment with chemotherapy (58% versus 34%, p = 0.03) in pancreatic cancer patients in a cohort study, in part because patients in secondary care had more advanced primary tumors [33]. Eight studies examined the occurrence of serious clinical events. No differences were seen relating to the number (and time of diagnosis) of recurrences and metastases [19,20,[23][24][25][26], deaths [23,25,29], or other clinical events [23,24,26] between primary and secondary care-based follow-up.
Documented follow-up care, as measured by adherence to medical guidelines and follow-up tests, was assessed by two RCT's. Murchie et al. [24] found that 98.1% of patients in primary-based care were seen according to guidelines versus 80.9% of patients in secondary-based care (p = 0.020). In the second study, patients in primary care were more likely to have one or more fecal blood tests (rate ratio 2.4, CI 1.4-4.44, p = 0.003), whereas patients in secondary care were more likely to have ultrasounds and colonoscopies, although it remained unclear whether or not this was done in accordance with follow-up guidelines [25].
High levels of patient satisfaction and perception of care were found for survivorship care in both primary-and secondary-based care [22,[24][25][26][27]32]. Using an adapted validated questionnaire, higher levels of patient satisfaction were found in primary care-based groups in two RCTs (9 out of 15 aspects by Grunfeld et al. [22] and 6 out of 15 by Murchie et al. [24]). In contrast, a questionnaire administered in an observational study [26] showed greater satisfaction in all 6 dimensions for breast cancer patients in secondary-based care (p < 0.05).
Five observational studies examined self-reported receipt of survivorship care by means of questionnaires and interviews. Disparate results were seen among primary-and secondary-based care, but there was no evidence for a more favorable strategy based on these results. Two studies showed a lower adherence to recommended periodic clinical examinations for breast cancer patients by physicians working in primary care (approximately 80% versus 90% in secondary care, p < 0.05) [31,32]. In another study, patients in primary care were more likely to receive examination as is recommended by national guidelines (58% versus 36%, p = 0.004) [27]. No differences were seen in patient self-reported mammogram frequency [28,31,32]. Maly et al. [28] found a higher uptake of preventive tests, including Pap smear (AOR 2.90, CI 1.05-8.04, p = 0.040) and colonoscopy (AOR 2.99, CI 1.5-8.51, p = 0.041), among underserved breast cancer patients in primary care. Physicians in primary care helped more often with lifestyle improvements for colorectal cancer patients [27], but this was not the case among breast cancer patients [31].

Costs
Survivorship care in primary care was associated with lower societal and patient costs in all four studies that performed cost  analyses (see Table 4) [19,21,26,29]. The main cost driver in all studies was the mean cost per visit, including organizational and physician costs.

Discussion
In this review, similar effects on clinical and patient-reported outcomes were seen for survivorship care in primary-compared with secondary-based care. Although the evidence should be interpreted with caution, survivorship care in primary care seems feasible and results in lower costs.

Comparison with existing literature
A recent Cochrane review found little to no effects on predefined outcomes for RCTs comparing non-specialist (e.g., PCP-led, nurse-led, patient-initiated, and shared care) to specialist-led follow-up [12]. The certainty of evidence was generally low due to the limited amount of RCTs. Similarly to the Cochrane review, this review found no important differences in survivorship care between primary and secondary care relating to clinical (survival and recurrences) and patient-reported outcomes (quality of life and symptoms). This review has identified additional outcomes in comparison with the Cochrane review relating to the content and quality of survivorship care. The content of survivorship care is examined by both documented follow-up care and self-reported receipt of survivorship care. Some differences were seen in these outcomes, especially relating to the adherence to guidelines and follow-up tests, but the results showed no favorite strategy. It remains unclear whether or not these differences may affect other outcomes, such as detection of recurrences and survival. Showing differences in these types of outcomes requires great numbers of patients and considerable follow-up time among often older patients with comorbidities, making this a challenging undertaking. This review has examined patient's perceptions and satisfaction with care as indicators for the quality of survivorship care. High levels of quality of care were found for survivorship care in both primary-and secondary-based care. Two out of three RCTs showed higher levels of patient satisfaction with primary-based care, illustrating its feasibility [22,24]. The aggregation of these results provides us with the indication that survivorship care in primary care is similar to care by a specialist. Moreover, survivorship care in primary care led to lower costs in all studies that performed cost-analyses.

Strengths and limitations
Our review provides additional evidence to previous literature by focusing exclusively on survivorship care by physicians GP general practitioner, FP family physician, PCP primary care physician working in primary care and by including non-randomized studies in the results. By performing a non-restrictive search and selection strategy, two additional outcomes relating to the content and quality of survivorship care have been identified in comparison with the recent Cochrane review. The search strategy, including reference checking, provides a sensitive search result.
There are some limitations that need to be addressed. Inherent to the design of some studies, differences were seen in baseline characteristics. Older patients and patients with prognostic better disease stage were sometimes more likely to receive follow-up in primary care [26,27,[30][31][32][33][34]. Despite adjusting for covariates, these differences might have influenced outcomes. Due to the substantial heterogeneity in outcomes and methodology, no data could be pooled for meta-analyses, hampering the interpretation of results. However, using a narrative synthesis, no important differences were seen relating to clinical and patient-reported outcomes. These results are in line with previous reviews [12][13][14][15][16].  (c) Documented follow-up care Adherence to medical guidelines and follow-up tests [24] Patients who visited a GP were more likely to be seen according to guideline (98.1% versus 80.9%, p = 0.020).
[25] Patients in the GP-group were more likely to visit their physician (1.27 times per quarter versus 0.84 times) and to have one or more FOBTS (rate ratio 2.4, CI 1.4-4.4, p = 0.003). Patients in the surgeon group were more likely to have ultrasounds (rate ratio 0.5, CI 0.3-1.0, p = 0.040) and colonoscopies (rate ratio 0.7, CI 0.5-1.0, p = 0.027). No differences were seen relating to other surveillance tests, including CEA, X-ray and CT-scan.
No differences on any subscale.
No differences on any subscale. [25] SF-12 PCS and MCS scores at baseline, 12 and 24 months.
No differences on any subscale.
No differences on any subscale after adjustment for age and chemotherapy. No differences for other symptoms (arthralgias, hot flashes, memory loss, vaginal dryness, insomnia, paresthesias and depression). (c) Patient satisfaction and perception of care [22] Adapted satisfaction questionnaire at baseline, mid-, and end of trial (Cronbach's alpha = 0.70).
Patients who visited a GP had greater satisfaction on 9 out of 15 aspects (relating to service delivery, consultation and continuity of care).
No differences in communication, coordination, nursing care, office staff and follow-up rating; non-significance remained in multivariate regression. [32] Computer-aided telephone interview on perception of follow-up.
Women who visited an oncologist reported a marginally higher degree of care coordination (81.9% versus 73.1%, OR 1.8, CI 1.0-3.5). (d) Self-reported receipt of survivorship care Adherence to medical guidelines and follow-up tests [27] Patient questionnaire on visits, tests and examinations 2-5 years after diagnosis.
The number of visits in the past year varied by physician specialty (p < 0.001). Patients in the PCP-group were less likely to see a doctor for "follow-up medical tests" (68% versus 89%, p < 0.001) and more likely to receive a physical examination (58% versus 36%, p = 0.004 GP general practitioner, PCP primary care physician

Implications for future practice and research
As the number of cancer survivors is rapidly increasing and resources are limited [12][13][14][15][16], alternative survivorship care strategies for the hospital-based survivorship care are deemed desirable. This review showed that cancer survivorship care in primary care seems feasible and worthwhile to consider. However, the role and capacity of physicians in primary care can vary depending on context and setting [10,11]. Most studies were performed in the UK and Canada in which physicians work as gate-keepers to secondary health care services. In these countries, a publicly funded universal health care system is in place. Other studies were performed in countries such as the US and Spain in which the health care system is both publicly and privately funded, and the role of primary care could be less distinguished. The randomized trials that could be identified, were limited to countries with a universal health care system, so further research is warranted to evaluate whether the results of these trials are also applicable to other health care systems. Furthermore, both clinical and patient-reported outcomes might change over time and could be affected by the length of follow-up. Therefore, to assess durable effects of survivorship care, greater number of patients and considerable follow-up time in these trials would be preferable. Moreover, the impact on the work-load for primary care physicians needs to be evaluated in case of growing numbers of patients in primary care-based cancer survivorship care. (b) Patient costs [19] Cost-and utilization questionnaire filled in by patients at baseline up to 24 months (including travel, out-of-pocket expenses and work loss).
More patients had expenses relating to travel in the hospital-group (£156.9 versus £76.7, p < 0.001). No differences were seen relating to out-of-pocket expenses (p = 0.10) or work loss (p = 0.45).
[21] Cost-questionnaire filled in by patients at baseline up to 18 months (including travel, out-of-pocket expenses, work loss, child support and spent time for an appointment).
GP general practitioner, PCP primary care physician

Conclusion
This review presents a comprehensive overview of survivorship care in primary care. To our opinion, this review has underlined the feasibility of survivorship care in primary care or possibility of some form of cooperative care. However, delivering high-quality survivorship care will also put restraints on primary care. This requires not only sufficient funding but also investments in organization and staff. Further studies with adequate designs are needed.
Author contributions All authors contributed to the study conception.

Compliance with ethical standards
Conflict of interest The authors declare that they have no conflict of interest.
Ethical approval This article does not contain any studies with human participants performed by any of the authors.