Introduction

While breast cancer incidence has grown over time—up to 2.3 million diagnoses in 2020 worldwide [1]—mortality rates have declined [2, 3]. Consequently, this results in a large number of breast cancer survivors in follow-up care, consisting of two parts: surveillance and aftercare. Surveillance aims to detect asymptomatic locoregional recurrences (LRR) or second primary breast cancers (SPBC) using mammograms and physical examination. The ultimate aim is to curatively treat patients. The ultimate aim is to curatively treat patients. As distant recurrences are in most cases not curable, and the early detection of distant recurrences does not improve prognosis, surveillance does not actively aim to detect these. Aftercare aims to detect diagnosis- or treatment-related side effects and subsequently use interventions to reduce these and improve quality of life (QoL).

Although in some countries guidelines advise to personalise follow-up [4, 5], current guidelines in, for example, the Netherlands and Belgium still advise similar surveillance schedules for all patients [6, 7]. This is probably due to a lack of clinical evidence that adapting surveillance schedules according to risk profiles is effective. However, it is widely known that differences in individual characteristics largely influence LRR and SPBC risks [8]. Furthermore, about 50% of LRRs and 25% of SPBCs are detected by patients themselves due to symptoms, outside of scheduled surveillance visits [9]. Moreover, patients’ beliefs and expectations of surveillance are often not realistic, including the incorrect assumption that surveillance also aims to detect distant metastases and that breast cancer cannot recur in between scheduled visits [10]. Importantly, overall LRR and SPBC risks are low and largely differ among individual patients [11, 12]. This is expected to lead to unnecessary surveillance visits for many patients and perhaps too little visits for specific high-risk patients.

While surveillance often consists of a one-size-fits-all approach, a large variation in aftercare is present, as this is often arranged according to both clinicians’ and patients’ preferences. It depends on the hospital which health care provider is involved (e.g. surgeon, specialised nurse [13]) and whether they make use of prescheduled consultations. Health care providers are also looking for appropriate tools they can use to personalise aftercare [13]. In addition, there are no guidelines on the specific contents of aftercare plans, resulting in unmet supportive care needs regarding fear of cancer recurrence, daily activity and sexual and psychological well-being [14].

Both these arguments and the fact that an increasing number of breast cancer survivors will receive follow-up care, lead to an increasing belief that surveillance and aftercare should be personalised [15,16,17]. However, clear evidence is needed about the effectiveness of both personalised surveillance and aftercare. The aim of this review was to identify all studies published from 2010 that investigated the effectiveness of personalised surveillance and/or aftercare in curatively treated nonmetastatic breast cancer patients.

Methods

This review’s protocol has been registered and made available in PROSPERO [18] (CRD42022375770). The Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) 2020 checklist [19] was used for transparent reporting (Online Resource 1–2).

Search strategy

Databases of PubMed, Scopus (including Medline and keywords of Embase) and Cochrane were searched for relevant publications between January 1st, 2010, and November 10th, 2022. Reference lists of relevant reviews were consulted. Studies published before 2010 were excluded, as we expected these to be less relevant for current clinical practice. The full search strategy is shown in Online Resource 3.

Eligibility criteria

The study population concerned nonmetastatic breast cancer patients ≥ 18 years, starting follow-up after completion of curative treatment (surgery and, if applicable, radiotherapy and/or chemotherapy). Patients could still be treated with endocrine therapy or targeted therapy. We included all intervention-control studies studying personalised surveillance and/or aftercare, including any intervention tailored to a patient’s individual characteristics and designed to be used for the entire follow-up period. Studies that investigated the effectiveness of short-term dietary, physical interventions, cognitive behavioural therapy or psychoeducational interventions, which were not part of a larger intervention designed for use during the entire follow-up period, were therefore excluded. We included studies on all outcomes, except for diagnostic accuracy, feasibility or patient experiences of the intervention only, without evaluating effectiveness.

Inclusion and data extraction

Two reviewers (JvH, MvM) independently screened and judged all identified studies on title and abstract. In case of doubt or disagreement, the study was included for full-text analysis. Both reviewers independently read the full text and decided on definite inclusion. Discrepancies were extensively discussed and resolved. Co-authors were consulted if necessary. Final data extraction was performed by one reviewer (MvM), and the second reviewer (JvH) was consulted in case of doubts. In case of missing or unclear information on interventions, the study’s first author was consulted. Data on population, intervention, control and outcomes were extracted. As multiple types of interventions and outcomes were studied, the data is presented descriptively.

Risk of bias

To assess risk of bias (ROB) of included studies, the Cochrane risk of bias tool for randomised (RoB2) [20] or nonrandomised studies (ROBINS-I) [21] were used, whenever applicable. ROB assessment was independently performed by two reviewers, and discrepancies were discussed.

Results

The search strategy yielded 3708 publications. Sixty-four publications were deemed eligible for full-text analysis. After full-text reading, 16 were included for data extraction. Reasons for the exclusion of the other 48 studies were (1) no intervention designed for the entire follow-up, (2) no control group, (3) no effectiveness measured (feasibility studies) and (4) intervention not personalised. Examples of excluded studies are Wallner et al. [22], Haq et al. [23] and Admiraal et al. [24], because they studied feasibility only, did not use a control group, and studied a short-term psychoeducational intervention (so did not personalise the entire follow-up period), respectively. The entire selection is visualised in Fig. 1. The 16 studies finally included are summarised in Table 1. Fifteen studies presented results of randomised controlled trials (RCTs) and one study of a pretest–posttest design. Three studies concerned the same RCT [25,26,27] but with different outcome assessments (longer follow-up or cost-effectiveness analyses), so were considered different studies in further analysis.

Fig. 1
figure 1

Flow chart of inclusion of publications. *Including Medline and keywords of Embase. **Including studies focusing on short-term dietary, cognitive behavioural, physical or psychoeducational interventions and interventions applied during active treatment

Full size image
Table 1 Detailed overview of included publications with the most important study characteristics
Full size table

Populations

All studies included nonmetastatic breast cancer patients who completed primary treatment, except for adjuvant targeted or endocrine therapy. Two studies focused on very specific populations. The first included Latina breast cancer patients with a deficit in either cancer screening (PAP smear or colonoscopy) or a positive comorbidity screening [35]. The second only included hormonal receptor-positive breast cancer patients ≥ 50 years old undergoing endocrine therapy [36]. One study included multiple cancer types [37].

Interventions

Interventions could consist of the use of personalised aftercare plans or a combined surveillance and aftercare plans, either or not supplemented with (1) educational or counselling sessions, (2) active support by patient navigators and/or (3) monitoring patient-reported outcomes (PROs).

Personalised surveillance

Although nine out of 16 studies included general recommended surveillance guidelines in aftercare plans [25,26,27,28,29, 31, 32, 34, 39], only one study evaluated a form of personalised surveillance [31]: all patients were dissuaded from scheduling routine follow-up visits and received suggestions for visits that could be postponed based on their aftercare plan (including personal tumour and treatment history).

Personalised aftercare

Aftercare often concerned a plan containing general information supplemented with personalised information. Of all 16 studies, 12 incorporated personalised treatment summaries [6, 25,26,27,28,29, 31,32,33,34, 38, 39], six contained lists of available supportive care resources [25,26,27, 29, 31, 39], four incorporated PRO measures (PROMS) [30, 34, 36, 37], two contained reminders on next follow-up dates [6, 38], one contained a general guide for people treated for cancer [28] and one contained, additional to a complete overview of individual patient-, tumour- and treatment-related details, information on the medical care team, potential side effects, dates of recent visits, current medications, barriers, upcoming appointments and libraries with further information [39].

Five studies additionally included a 30-min to 1-h educational session with a nurse and/or nutritionist [25,26,27,28, 33]. This sometimes consisted of an explanation that follow-up care was now the responsibility of the primary care physician [25,26,27], and mostly included additional information on the contents of a general aftercare plan. One study included motivational interviewing to engage patients in the development of a patient-owned aftercare plan [32].

One study included active support by patient navigators providing patients personalised assistance including phone calls, home visits, transportation assistance and care coordination, and help with practical things [35]. In three studies, patients regularly met/called with a patient navigator, nurse practitioner or a physician assistant at predefined intervals [31, 34, 35].

Two studies integrated PROMs evaluating symptoms, followed by telephone consults [30] or visits with a nurse or physician assistant [34]. One study had no mandatory consultations, but regularly collected PROs that were used as screening and dialogue tool [36]. Two studies included web-based applications, one supporting cancer survivors in self-management by monitoring symptoms and QoL, providing feedback and personalised supportive care options [37], and one included tailored information on treatment and side effects and included push notifications to remind women to take medicine (e.g. endocrine therapy), participate in a module or seek remedies for side effects [6].

Controls

Most studies included routine follow-up care as control [6, 25,26,27,28,29,30,31,32,33, 35, 36, 39], which could differ per hospital and between studies. In one study, this routine follow-up care consisted of outpatient clinic visits which were based on patient and clinician experiences, which could be seen as personalised surveillance. However, this study specifically focused on the use of online health questionnaires that were monitored by a nurse practitioner (intervention) vs. no monitoring (control) and did not describe the potential effectiveness of personalised surveillance [30].

Two studies included elements of the intervention in the control group, like a guide for cancer survivorship [28] or a fact sheet with contact information of patient navigators [35]. One study used similar intervention and control conditions except for personalised treatment summaries and recommended the next surveillance due dates, which were only provided in the intervention group [38].

Outcomes

A detailed overview of all studied outcomes can be found in Table 1. Below, a summary of the most studied outcome categories is given, while Fig. 2 shows all of the specific evaluated outcomes.

Fig. 2
figure 2

Overview of all outcome(s) (categories) in the 16 included studies. The solid lines indicate the number of studies that analysed a specific outcome category, and the patterned lines indicate the number of times the specific outcome is analysed. The latter numbers do not add up to the numbers in the solid lines, because in one study, multiple outcomes could have been analysed

Full size image

QoL/depression/cancer worry/well-being/impact of cancer

Ten studies evaluated one of these outcomes. Seven studies evaluated QoL [26, 27, 31,32,33, 35, 39], and four found a significant positive effect of personalised aftercare on QoL [6, 32, 35, 37]. One of these evaluated active support by patient navigators (differences on subscales ranging from 1.6 to 8.1, all significant except for well-being after 6 months, based on FACT-B/FACT-G questionnaires) [35], one evaluated a web-based eHealth application supporting self-management (summary score difference of 2.3 after 6 months using EORTC QLQ-C30) [37], one evaluated provision of tailored information using a web-based application (summary score difference 6.9 after 12 months, based using WHOQOL-BREF) [6] and one evaluated a coaching encounter to engage patients in the development of a patient-owned aftercare plan (proportion of clinically meaningful improvement in physical role 55 vs.18%, bodily pain 47 vs. 24% and emotional role 42 vs.21% for intervention and control group, respectively, based on SF-36). The latter also found a small significant improvement in depressive symptoms (mean difference of − 1.6 in the intervention group between baseline and 3 months, based on PHQ-9) [32]. Three studies evaluated distress/worries [6, 28, 36], and one found a significant decrease in fear of recurrence (mean difference of − 1.6 after 12 months, based on cancer worry scale) after access to a web-based aftercare plan (high ROB, see the ‘Risk of bias’ section) [6], and one found a significant improvement in health worry after three (mean scores of 2.7 vs. 2.3, respectively, based on ASC), but not after 6 months, for patients who received a personal educational meeting, compared to the control group. They did not find any difference between intervention and control on physical and functional well-being and impact of cancer [28].

Satisfaction with care/self-efficacy/patient activation

Ten studies evaluated forms of satisfaction, self-efficacy/self-management or patient activation [26,27,28, 31, 32, 34, 36,37,38,39]. One found a nonsignificant trend towards improvement of self-efficacy and self-management after a coaching encounter to engage patients in a patient-owned aftercare plan [32].

Symptom reporting/health care use

Six studies evaluated symptom(s) (reporting) or outcomes related to health care use [26, 27, 30, 31, 36, 39]. One reported a significant positive effect of symptom monitoring using online questionnaires in between standard surveillance visits on symptom reporting (mean of 7.4 vs 3.2 new or changed symptoms within 18 months, respectively), but not on health care resource use [30]. This study had a high ROB (see the ‘Risk of bias’ section). Another study found a significantly lower number of consultations in the intervention group—where PROs were collected and used as screening and dialogue tools—compared to the control group (2.1 vs 4.3 within 2 years, respectively) [36]. All other studies focusing on symptoms, type and/or frequency of care use did not find any significant or clinically relevant differences between intervention and control groups.

Adherence to treatment/guidelines

Four studies evaluated treatment/guideline adherence [27, 31, 33, 36], focusing on adherence to recommended visits or adherence to use of endocrine therapy. None of the studies found significant or clinically relevant differences between intervention and control groups.

Survivor knowledge

Three studies evaluated survivor knowledge [29, 34, 38], of which two found a significant positive effect of the intervention [34, 38]. One of these evaluated the effect of an individualised aftercare plan but hypothesised that the effect was more related to repeated administration of the survey than receipt of the aftercare plan [34]. The other—which was the only pretest–posttest study included in this review—showed that patients who received a personalised survivorship care plan reported greater perceived knowledge, but that the standardised plans resulted in a significant increase in perceived knowledge from pre to post [38]. Importantly, both of these studies were considered high ROB.

Care coordination

Three studies evaluated care coordination [26, 27, 32], but none found significant differences between intervention and control groups. Two of these reported on the same RCT, but with different follow-up times [26, 27]. One study reported a trend towards a positive effect with a mean score of 47.4 vs 35.1 on ‘discussion of survivorship care with primary care physician’ for intervention and control group, respectively [32].

Unmet needs

Two studies evaluated unmet needs [6, 36], which were measured by either the Patient Experiences Questionnaire [36] (including open questions on certain procedures that were not offered or concerns that were not discussed with care providers) or the Cancer Survivor Unmet Needs-Chinese Scale [6] (including questions on communication, information, physical/psychological, medical care and communication needs). The latter found a significant decrease in unmet needs after access to a web-based personalised aftercare plan, compared to the control group (mean difference of − 3.6 after 6 months using CaSUN-C, high ROB) [6].

Other

Two studies (based on the same RCT) evaluated patients’ awareness of which physician was primarily responsible for follow-up care, as follow-up care was transferred to primary care [26, 27]. Both did not find any significant or relevant effect of a personalised aftercare plan plus an educational session. One study evaluated physician implementation of treatment summaries and a personalised aftercare plan (score based on the number of needs addressed by physicians), and found a significant positive effect (mean difference of 16 (scale of 1–100) after 12 months) [33]. Finally, one study evaluated the cost-effectiveness (based on the same RCT as two other studies that did not report any intervention effects [26, 27]) of a personalised aftercare plan plus an educational session and concluded it was not cost-effective [25].

Risk of bias

Of all 16 studies, one study (6.3%) was classified as low [28], four (25%) as high ROB [6, 30, 34, 38] and 11 (68.8%) with concerns [25,26,27, 29, 31,32,33, 35,36,37, 39]. The three studies based on one RCT were all rated as with concerns [25,26,27]. There were some discrepancies between reviewers which could primarily be explained by different interpretations of signalling questions of domains two and four of RoB2 [20]. This regarded mostly discrepancies between low ROB or having concerns. After careful discussion, the most stringent outcomes were used for the final assessment (Fig. 3, Table 1).

Fig. 3
figure 3

Risk of bias assessment. Upper panel: RoB-2, risk of bias tool for randomised controlled trials; D1, randomisation process; D2, deviations from the intended interventions; D3, missing outcome data; D4, measurement of the outcome; D5, selection of the reported results. Lower panel: ROBINS-E, risk of bias tool for nonrandomised studies; D1, confounding; D2, measurement of the exposure; D3, selection of participants; D4, post-exposure interventions; D5, missing data; D6, measurement of the outcome; D7, selection of the reported results

Full size image

Discussion

In this review, 16 studies were identified that evaluated the effectiveness of a personalised surveillance and/or aftercare plan in non-metastatic breast cancer patients after curative treatment. A wide range of personalised interventions and different outcomes were studied. Only one study examined a form of personalised surveillance, which did not find any significant or relevant effect on the frequency of visits, adherence to guidelines, QoL and satisfaction with care [31]. Most studies evaluating aftercare plans included individual treatment summaries, overviews of standard follow-up guidelines and/or overviews of available supportive care resources. QoL was most frequently studied, and four out of seven studies found a significantly positive effect of a personalised aftercare intervention. However, most of these studies found small absolute effects. Importantly, only one study was considered to have low ROB, and this study did not find any effect of personalised aftercare. A wide range of other outcomes was studied, with conflicting results. Surprisingly, only one study found a significant effect of personalised aftercare on the outcome category of satisfaction with care/self-efficacy/patient activation, which seems counterintuitive. However, as all studies used different personalised interventions and studied different outcomes using different measurement instruments, it is impossible to compare all studies and to draw conclusions on the effectiveness of personalised follow-up. The fact that three studies were based on one RCT did not affect the conclusions of this review.

Many studies emphasise the need for personalised surveillance [40, 41], but in clinical practice, guidelines still recommend a one-size-fits-all approach [8]. This could be due to many care providers overestimating patients’ recurrence risks [42], or because patients are hesitant about less intensive surveillance [43] due to inadequate risk perceptions, fear of recurrence [44] or unrealistic expectations [10], and could explain that only one of the included studies in this review evaluated a form of personalised surveillance. For both patients and clinicians to get insight in personal risks, a risk prediction tool can be used. INFLUENCE estimates risks of LRR, distant metastases and SPBC [45], and is currently integrated in a decision aid that can be used to personalise surveillance schemes [44]. Recently, this model has been updated to INFLUENCE 3.0 (results not yet published) and is being tested in a large multicentre study on the effectiveness of personalised follow-up, where the model (as part of a decision aid) is used to support the decision regarding the most optimal surveillance scheme [46]. Importantly, the model is based on data from patients that already have been treated for breast cancer, and therefore, the model can explicitly not be used for treatment decision-making. As recurrence rates are generally low [11], it is expected that the frequency of follow-up visits can be reduced for many patients resulting in decreased costs and lower burden on health care [47, 48]. A previous study has already shown that patients are open to the use of risk information in decision-making [43].

The large variety in the type of intervention and outcomes in aftercare suggests that there is a high need for personalisation, but that people are searching for the right way to do so. This is supported by results of several studies, showing large variations in the organisation of aftercare, especially regarding timing, frequency and disciplines of involved care providers [13, 49, 50]. Other studies showed that there are several barriers regarding the integration of PROMs in aftercare [51, 52], which was also evaluated in several of the included studies in this review [30, 34, 36, 37]. It has also been described that promoting engagement and adherence to care plans may lower psychological distress or cancer-related barriers [53]. Studies that evaluated motivational interviewing techniques to increase patient engagement indeed showed significant improvements in QoL [32, 35].

Aftercare is complex and comprises a lot of elements. Ideally, it includes assessment and management of physical and psychosocial effects due to cancer diagnosis and treatment, health promotion and care coordination [54]. In order for patients to get engaged in the management of their own recovery, it is important to empower patients by providing clear information on possible (late) side effects of breast cancer and its treatment—including available self-help and support options—and to give them information on breast awareness (i.e. how to notice potential signs of recurrence in an early state). The relevance of patient empowerment has been acknowledged in literature [55] and has been shown to improve quality of life [56]. In addition, it is crucial for patients to get insight in individual needs. A previous study showed that these individual needs are not always assessed, as only 16.1% asked patients about it [57]. Additionally, many patients have difficulties in expressing their needs [58], and the degree of communication about preferences varies widely between patients with different cultural backgrounds [59]. To support patients to understand their own needs and preferences and to base decisions regarding their health care on it, a patient decision aid or dialogue tool could be used [58, 60], which can form the basis for individual counselling sessions. A prior pilot study showed a newly developed decision aid to have promising effects on shared decision-making, choice evaluation, choice of aftercare and hospital costs, but to substantially increase consultation time [61]. However, one could argue that providing patients with completely individualised aftercare would finally decrease health care use and thus costs. In case a patient timely takes action in case of psychological or physical complaints, or any other concerns, worsening of symptoms and thereby future, more intensive, care use could be prevented. However, this remains to be investigated, as care use might also increase as a result of increased detection of unmet needs. Finally, we can learn from experiences in other cancer types, such as the shared-care survivorship programme for testicular cancer [62] and the Dutch Childhood Oncology Group guideline for follow-up [63].

Strengths and limitations

To our knowledge, this is the first review that included all published intervention-control studies on the effectiveness of personalised follow-up for breast cancer patients. A broad search strategy was used, ensuring a high level of completeness. Title abstract screening and full-text reading were performed independently by two reviewers, which is described to increase the number of relevant studies identified [64]. ROB assessment was also performed by two independent reviewers, which is crucial since ROB judgements can differ substantially between reviewers, especially regarding interpretation on random sequence generation, blinding of participants and personnel and incomplete reporting [65]. The two reviewers extensively discussed discrepancies, and in case a consensus could not be reached, the most stringent judgement was used for final assessment. Data extraction was performed by one reviewer, which could have resulted in higher error rates [66]. However, as the second reviewer had read all publications’ full text, this reviewer could carefully judge the data extraction on completeness. There were two studies [38, 39] included in this review where both reviewers doubted whether only provision of personalised information on treatments, side effects and/or standard surveillance guidelines (without counselling/educational sessions) could really be considered personalised aftercare. To be complete, these papers were included, also to show the inconsistencies in current practice, confirming the belief that one is still searching for the right way to personalise aftercare.

Clinical implications

Fifteen out of 16 studies included in this review solely focus on personalised aftercare, and they all include different types of interventions, studied different outcomes and used different measurement instruments. Besides, in some cases, it could be questioned whether the intervention can be called ‘personalised’. This makes it impossible to draw firm conclusions on the effectiveness of the interventions. First, there is need for a definition of personalised surveillance and aftercare. Ideally, surveillance consists of a decision aid including a prediction tool [45] to jointly discuss personalised surveillance schemes. Besides, personalised aftercare should comprise (1) a patient’s needs assessment (e.g. using PROs), (2) information on potential side effects of cancer (treatment) and available care resources and (3) a personalised aftercare plan, including a diagnosis and treatment summary, decisions on organisation of aftercare (e.g. frequency, involved care providers) and signals to seek care for. A dialogue tool could support the shared decision-making process between care professionals and patients of the development of this personalised aftercare plan. Effectiveness can consequently be measured according to uniform information standards such as the ICHOM initiative [67].

Conclusions and future prospectives

Personalised follow-up varies widely and is not structurally embedded in clinical practice. Therefore, there is still a lack of evidence on its effectiveness. This review shows the current gaps in literature and forms the basis of a large multicentre prospective study on the effectiveness of personalised surveillance and aftercare in breast cancer patients. This prospective study is expected to conquer the problems addressed in this review, and will provide clear evidence on the (cost-)effectiveness of personalised follow-up.