Intensive follow-up for women with breast cancer: review of clinical, economic and patient’s preference domains through evidence to decision framework

The research question was addressed by means of a systematic review of the literature on the evidence of health outcomes related to the alternative strategies – intensive and less intensive follow-up. An operational definition was used for intensive follow-up, where intensive was defined in comparison with a less intensive follow-up schedule or a patient-initiated approach. The review protocol is available upon request. Standard Cochrane Collaboration methods were followed [19]. For the evaluation of the importance of the outcomes and for the assessment of the quality of evidence, the GRADE system was used.

Research question: the clinical question was structured following the PICO (Patient, Intervention, Comparison, Outcomes) format:

Critical outcomes included mortality due to breast cancer, breast cancer recurrences and breast cancer specific survival. Quality of life and satisfaction were considered important outcomes.

Following the WHO Handbook for Guidelines Development [20] as guidance, existing relevant systematic reviews of observational and experimental evidence were included as a source of individual studies; additional individual studies were searched, to update the body of evidence. Temporal or language restrictions were not applied. Studies in which the effects of follow-up intensity were not assessed, or when the outcomes were out of the scope of the clinical question, were excluded.

Systematic reviews were identified by introducing a combination of controlled vocabulary and search terms (e.g., follow-up, breast neoplasms, mortality, recurrences, quality of life, satisfaction, cost, healthcare resources, survival) in The Cochrane Database of Systematic Reviews (2015, issue 11), The Database of Abstracts of Reviews of Effects (DARE), and PubMed limiting the search to the subset “systematic [sb]”.

Original studies were searched in MEDLINE (through PubMed; from 1946 to January 2016), EMBASE (through Ovid; from 1980 to November 2015), PDQ, McMaster Health Systems Evidence, CENTRAL, and NHS EED (through The Cochrane Library; January 2016).

The complete search algorithms designed for each database, the hits retrieved, and the reasons for exclusion are presented in Additional file 1 and Fig. 1a.

One reviewer screened the search results based on title and abstract. This process was subjected to a quality process, by reviewing 20% of the references by a second reviewer. Two reviewers independently confirmed eligibility, based on the full text of the relevant articles. In case of disagreement between reviewers the inclusion of studies was determined by consensus.

Data extraction was conducted by one reviewer. As quality control, another reviewer went through 20% of the data for accuracy.

The assessment was carried out by one reviewer. As quality control, another reviewer went through 20% of the data for accuracy. For each study, the risk of bias was rated for each domain as low, high or unclear risk of bias.

Odds ratios (OR), risk ratios (RR) and hazard ratios (HR) were extracted, with their 95% confidence intervals (CI). If available, only adjusted effect measures were collected. Data from any estimation of effect provided (percentages, means, medians) were also collected.

The quality of evidence per outcome was rated from high to very low considering the standard GRADE domains: risk of bias, imprecision, inconsistency, indirectness and publication bias [21, 22].

A pooled analysis was conducted applying the inverse-variance method under the random-effects model [23]; the analysis was performed through the Software Review Manager v. 5.3. Heterogeneity was assessed using the I² statistic.

The report of results of the meta-analysis adhered to the guidelines articulated in the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement [24].

A review about women’s values and preferences for intensive follow-up versus non-intensive follow-up after breast cancer treatment was undertaken.

In a first stage, after conducting the systematic search of the literature, the screening of references was carried out, initially prioritising the identification of systematic reviews. In a second stage, individual studies were retrieved (e.g., qualitative studies, surveys, utility elicitation studies). Studies in English, French, German, Greek, and Spanish, carried out in the Organisation for Economic Co-operation and Development (OECD) Region, were included. Included studies were:

Studies assessing only women’s knowledge, views, perceptions, attitudes and expectations regarding follow-up strategies after breast cancer treatment were excluded; similarly, studies assessing barriers to follow-up strategies after breast cancer treatment were not included.

A search strategy was designed to identify relevant studies in MEDLINE (accessed through Ovid). For systematic reviews, there were no time restrictions. For primary studies, only studies published after 2006 were included. The complete search strategy can be found in Additional file 2.

One reviewer screened the search results based on the title and abstract. Two reviewers independently confirmed eligibility based on the full text of the relevant articles. In case of disagreement between researchers, the inclusion of studies was decided by consensus (Fig. 1b).

One reviewer extracted the main characteristics of the included studies and their findings in a tabulated format. A second reviewer checked the extracted data for accuracy.

The quality of evidence was rated with GRADE. In the case of qualitative research, the Confidence in the Evidence from Reviews of Qualitative research (CERQual) approach was used [25].

A review about the economic evidence for intensive follow-up versus non-intensive follow-up after breast cancer treatment was carried out.

Screening of literature and study selection was done in a step–by- step approach. Firstly, the search focused on studies that addressed economic aspects directly related to the PICO question. Then, recent European cost-effectiveness or cost-utility analyses related to the PICO question were looked for. Only studies in English were included.

Search strategies were designed to identify relevant studies in MEDLINE (through Ovid, January 2016) and in the NHS Economic Evaluation Database (through The Cochrane Library, January 2016). The complete search strategies are included in Additional file 3. Study design filters were applied to retrieve relevant studies. The selection process is presented in Fig. 1c.

Main characteristics of included studies were described in a tabulated format, including the following data: author and publication year, country, type of economic analysis, perspective of the analysis, time horizon and discounting, relevant outcomes and costs included, sources of information (baseline outcomes, relative intervention effects, resource use and costs), Quality Adjusted Life Years (QALY), Incremental Cost Effectiveness Ratio (ICER), sensitivity analysis and conflict of interest.

The quality of evidence for the resource requirements was rated according to GRADE [26]. The NICE methodology checklist for economic evaluations [27] was used to assess the risk of bias and decide whether to include the studies. Included studies were of low risk of bias and were considered applicable to the European context.

To summarise the evidence, and in accordance to the GRADE methodology [28] and the interactive Evidence to Decision framework guidance [29], an EtD Framework was developed. The authors covered the role of the panel with respect to the EtD framework.

Five systematic reviews were included for the evaluation of health outcomes [30‐34]. These systematic reviews were used as a source to identify primary studies. Eight papers, referring to six randomised clinical trials for a total of 3534 randomised women [35‐42], were retrieved and included. These studies are summarised in Table 1.

The included studies had different definitions for intensive follow-up. In four studies, intensive follow-up referred to a greater number of diagnostic tests compared to regular follow-up [35, 36, 40, 42], while in two studies it referred to more frequent visits without modification in the number of diagnostic tests [37, 39]. Three studies compared an intensive versus a standard follow-up [35, 36, 42], while the other three compared a low-intensity patient-initiated versus a standard follow-up [37, 39, 40]. Five studies specified that patients (including the non-intensive follow-up group) underwent an annual mammography [35‐38, 42]. No studies provided information about specific breast cancer mortality or survival. Among all studies, only the one carried out by the GIVIO group [35] reported the expected 5-year relative mortality reduction used for the calculation of sample size, i.e. 20% reduction; this threshold may be considered as the clinically significant mortality reduction expected.

Results and pooled analysis are provided when possible. Quantitative estimates are available only for the following outcomes, presented in Fig. 2:

10-year overall mortality: one trial [38], that compared intensive vs. standard follow-up in 1243 women, found a RR = 1.05 (95%CI: 0.90 to 1.22). The quality of evidence was high because non-blinding was not considered a cause of risk of bias for this outcome.

5-year overall mortality: three trials [35, 36, 40], on 3035 patients, that compared intensive vs. standard follow-up found a RR 1.00 (95%CI: 0.86 to 1.16; I² = 0%). The 5-year mortality RR lower 95%CI did not reach the expected value for clinical significance either (vs. an expected 20% mortality reduction, as per GIVIO group outcome). The quality of evidence was high because non-blinding was not considered a cause of risk of bias for this outcome.

5-year breast cancer recurrences (any loco-regional and distant): three trials [35, 36, 40], on 3035 patients, that compared intensive and standard follow-up, resulted in a RR = 1.08 (95%CI: 0.89 to 1.30; I² = 60%). The quality of evidence was moderate because it was downgraded for risk of bias (the outcome assessment was not blinded). From the clinical point of view, when the patient is diagnosed with recurrence, there will be an initiation of new treatments or change in the treatment, so we considered that non-blinding is not an issue in this case.

Breast cancer recurrences at any time: five trials [35, 36, 39, 40, 42], on 3217 women, found a RR = 1.10 (95%CI: 0.95 to 1.27; I² = 23%, when comparing intensive vs. standard follow-up. The quality of evidence was moderate because it was downgraded for risk of bias (the outcome assessment was not blinded). As discussed above, non-blinding was not considered as an issue.

Satisfaction of women with the type of follow-up: satisfaction was measured as reassurance (capacity of the type of follow-up to clear patients’ doubts or fears) in two studies [37, 39] on 245 patients, and as convenience (suitability of the follow-up to the woman’s life circumstances) in one study [39] on 61 women. The results on reassurance were in favour of intensive follow-up (RR 1.28, 95%CI: 1.07 to 1.54; I² = 90%). The overall quality of evidence was very low due to risk of bias (the studies were not blinded) and imprecision (the number of events is small). The results on convenience favoured non-intensive follow-up (RR 0.04, 95%CI: 0.01 to 0.31). The overall quality of evidence was low because evidence needed to be downgraded for risk of bias (the outcome assessment was not blinded) and imprecision (the number of events is very small).

For each studied outcome, the evidence profile is reported in Table 2.

For women’s values and preferences, three European studies were included [37, 43, 44] (Table 3).

Gulliford et al. [37] compared experiences of 193 patients with breast cancer, randomised into a group with a conventional schedule of clinic visits, and a group of less intensive follow-up. Both cohorts received identical mammography and were invited to call for immediate appointments if they detected symptoms. Stemmler et al. [43] conducted a surveillance study in a population of women with breast cancer; among the respondents, most (59%) belonged to an organised self-help group. Kimman et al. [44] conducted a multicentre discrete-choice experiment survey to measure the strength of preferences for several characteristics of breast cancer follow-up. The results of these three studies were inconsistent: in the first study, women appeared to prefer non-intensive follow-up schedules, while in the other two the preferences favoured intensive schedules. However, important variability was present among studies and within studies. There was low confidence in the evidence due to risk of bias and inconsistency.

The results of the review indicated that most of the regularly scheduled follow-up visits used further extensive laboratory and imaging procedures exceeding the quantity of examinations recommended in most of the current follow-up guidelines.

Four studies [41, 42, 45, 46] assessed resources used, costs and cost-effectiveness of intensive follow-up strategies.

Robertson et al. [46] conducted a cost-utility analysis in the UK and provided estimated costs (in 2008 value) for different mammographic surveillance regimens in women after breast cancer surgery. By assuming the cost of a mammography and of a clinical follow-up visit to be 71 and 110 €, respectively, in a cohort of 10,000 UK women with a mean age of 57, total costs varied from 3.27 million € (mammographic surveillance every 2 years) to 16.8 million € (yearly mammographic and clinical follow-up) for a 10-year surveillance period. The study used a Markov model and found that the most cost-effective strategy was surveillance with mammography alone, provided every 12 months since the incremental cost-effectiveness ratio (ICER) for this strategy compared to no surveillance was € 6051 per QALY gained.

A cost-utility evaluation conducted in The Netherlands [45] analysed data (costs in 2008 value) on 299 patients randomised into four groups: (1) hospital follow-up; (2) nurse-led telephone follow-up; (3) hospital follow-up plus a short educational group programme (EGP); and (4) nurse-led telephone follow-up plus EGP. Hospital follow-up plus EGP had an ICER of 236 € per QALY compared to the next best alternative nurse-led telephone follow-up plus EGP. The other two strategies were dominated (higher costs and fewer QALYs). The authors concluded that nurse-led telephone follow-up combined with a short EGP could be a cost-effective option. However, they did not estimate the ICER of this strategy compared to standard follow-up. Furthermore, the time horizon of the study (one year) was clearly insufficient to evaluate the cost-effectiveness of compared alternatives.

The study of Oltra et al. [42] found that an intensive follow-up characterised by multiple laboratory and imaging tests triples average costs of the standard clinical follow-up without differences in early detection of relapses during the three years of follow-up. The study of Kokko et al. [41] found that the most expensive strategy doubled the costs of the cheapest one without important differences in breast cancer recurrences among them.

The quality of the evidence on economic evidence was moderate due to indirectness. The study [56] considered in the evidence to decision framework was conducted in the UK, and the results may not be applicable to other European countries.

The EtD framework was applied to conclude the assessment. The research question is summarised in Table 4, while Table 5 represents the assessment, carried out in its 12 domains: among others, the certainty of evidence (e.g., no statistically significant differences in mortality) between different types of follow-up), the important uncertainty and variability in women’s values, and the cost-effectiveness of the intervention (which favours non-intensive schedules) are crucial elements in drawing conclusions.

Finally, Table 6 reports the conclusions summarised by the authors, in the form of a suggestion to perform breast cancer follow-up once a year with a mammography visit, as opposed to other types of regimens.

Our results showed that intensive follow-up, compared with less intensive follow-up including more frequent diagnostic test or visits, does not have beneficial effects on 5- or 10 -year overall mortality or recurrences in women with breast cancer. This finding was consistent between the studies included, and the quality of the evidence was moderate. Among the included studies, two randomised trials showed that intensive follow-up appeared to increase reassurance in patents (data on 250 women; RR 1.28, 95% CI from 1.07 to 1.54) [37, 39]. However, the quality of the studies was downgraded due to the inconsistency of studies. The cost of different regimens of follow-up is variable, with more intensive regimes being more expensive but without increases in health benefits; thus less intensive regimes are favoured. From one cost-utility analysis [46], an annual visit with mammography results in moderate costs, can be considered cost-effective compared to no surveillance, and is likely to be feasible.

The European Society of Medical Oncology (ESMO) Guidelines on breast cancer recommend regular visits every 3–4 months for the first 2 years after treatment (and gradually decreasing thereafter) in addition to an annual mammography [47]. American Cancer Society/American Society of Clinical Oncology Guidelines [8] also recommend detailed cancer-related history and physical examination every 3 to 6 months for the first 3 years after primary therapy (and thereafter decreasing) in addition to a yearly mammography. Their recommendations would fall under the definition of a “less intensive follow-up” that, in the majority of the studies included in our review, would include at least a clinical visit and mammography once a year. However, intensive follow-up is still also quite common in clinical practice [48‐51] and represents a significant workload for radiotherapy, surgery and oncology professionals [10], in addition to being a costly process.

A recently published systematic review on the effects of breast cancer follow-up showed that standard approaches are as effective as intensive ones; moreover, no differences in quality of life were documented [1]. While considering the health outcomes, including mortality and recurrences, our results confirm the already reported results.

Many of the studies included in our review were carried out in previous decades, and their results might be slightly outdated, given the recent substantial changes in breast cancer care [52]; However, our study also took into account further perspectives, by including also women’s preferences and values, and economic aspects, as adopted in the GRADE approach. The reviews on women’s values and preferences and economic evidence were, however, limited to English and for the last ten years and Medline only, and results would have been more robust if such reviews were carried out with a broader scope. Moreover, the suggestion for less intensive follow-up was built by using the EtD: this is a new approach in the clinical oncology field, but has been previously used already in breast cancer screening [53], colon cancer screening [54], as well as in other contexts [55]. The EtD explicitly takes into account factors related, among others, to the quality of evidence, desirable and undesirable effects, values, resources and feasibility, that altogether constitute a comprehensive approach to a decision-making exercise. The suggestion reported in this paper was made by a multidisciplinary group of authors, but it should not be considered as a recommendation from a guideline panel.

The main expectation from an intensified follow-up from a women’s perspective was reassurance and increased sense of security. This finding raises the need to better inform women on the lack of evidence of effect of intensive follow-up on clinical outcomes of mortality and recurrences. However, it needs to be considered that the follow-up visit may also have additional aims than detection of recurrence, such as motivating women to continue endocrine treatment during the follow-up period, providing information about long-term adverse effects of treatment, and helping in their management, as well as providing psychosocial support [56]. These other aims of follow-up are very important in the light of the high prevalence of e.g., depression (varying from 9.4% to 66.1%), and anxiety (varying from 17.9% to 33.3%) among breast cancer survivors [12]. These additional aspects should not be neglected and they should be better explored while evaluating the effects of different follow-up strategies. Hence, further well-designed studies should be performed. There is a need to balance and prioritise these different outcomes, including also additional patient-centred endpoints described above, as well as including undesirable effects of more frequent investigations. Moreover, organisational aspects related to the coordination of follow-up activities (i.e. nurse-led and GP-led activities, etc.) are only analysed in few studies [57] and should be better explored, as they may impact on the acceptance of the protocol by women, healthcare providers, etc. as well as on costs and feasibility.

From the clinical point of view, annual mammography is well justified to detect potential new primary or local recurrences. On the contrary more intensive follow-up schedules including additional diagnostic tests, such as breast MRI, liver ultrasound or bone scans could result in large costs without sufficient evidence regarding their benefits or harms.

In summary, based on these findings, less intensive follow-up could be recommended, although the exact format of the follow-up visit would need to be further clarified, as the studies used quite different follow-up schedules and tests. The treatment of breast cancer has become increasingly individualised [58] as the risk of breast cancer recurrences is very variable and is related, among other variables, to genetic predisposition of individual women, breast cancer characteristics and its treatment. Therefore, also the follow-up should be individualised based on the risk estimates, and on women’s perceptions and values. A “one size fits all” approach may not be relevant.