Over the past two decades, many patient-reported assessment tools have been developed to identify common problems experienced in cancer survivors following curative treatment [1–3]. These approaches have allowed for improved identification and development of corresponding interventions across many different problem areas [2, 3]. One such patient-reported tool is the Cancer Survivor Profile (CSPro) [4]. The CSPro was developed to detect a multidimensional range of symptoms, function-related challenges, lifestyle behaviors, financial strain, and difficulty with skills helpful in obtaining quality health care  for in breast cancer survivors (BCS) following primary oncology treatment and beyond.

The original goal of the CSPro was to provide a practical, valid, reliable, and relatively rapid assessment tool for the valid detection of a wide range of problem areas of breast cancer survivors. The selection of problem areas was based on a comprehensive review of both the epidemiological and qualitative literature [4, 5], resulting in identification of 18 distinct problems. Corresponding measures were carefully selected and found to have sound measurement features in BCS. Factor analysis and confirmatory factor analysis indicated the resulting CSPro measured problems across the following domains: health care–seeking skills (healthcare competence, health information, patient-provider communication. health information, information acquisition), symptoms (fear of cancer recurrence, poor body image, pain, fatigue, depressive symptoms, anxiety), function (cognitive, social, sleep, work, and sexual function), health behavior (low levels of physical activity and unhealthy diet), and economic strain [4]. The CSPro has been translated into Chinese and this version has also been rigorously validated [6].

The CSPro was comprehensive by design. It was intended for scales measuring different problems to be relatively independent of each other and facilitate identification of specific challenges across several many problem areas. While this may serve a helpful function (i.e., intervention targeting), the comprehensiveness of 18 distinct problem areas presents its own challenges. For example, when using the CSPro as a tool in a clinical setting, BCS reported difficulty with the number of individual areas, noting that it was confusing/overwhelming to pay attention to all these problems simultaneously [7]. Similarly, this level of comprehensiveness may have the unintended consequence of limiting investment of finite clinical and research resources (i.e., forcing reduced investment in time across 18 domains vs a greater focus on a select few). Because of these reasons, it was thought that reducing this diverse set of problems into more manageable groupings may be easier to understand and implement.  serve a helpful function. Specifically, this type of integration optimization may provide a simpler way to conceptualize and more efficiently and effectively manage these multiple problems. It may also help identify common underlying mechanisms in the future. The current study sought to achieve this simplification through cluster analysis.

The use of cluster analysis in “symptom” science [8] has generated an improved understanding of potential mechanisms that underlie many symptoms experienced by BCS. The use of clustering techniques to better understand how several individual problem areas (i.e., not limited to symptoms) might be related or nested into a few clusters [14] may also help identify common mechanisms of a multidimensional array of problems. It was reasoned that by following a similar methodology, a diverse set of problems in addition to symptoms, as measured by a tool such as the CSPro, may be reduced into more manageable groupings.

During the original development of the CSPro, it was observed that, despite independence of scales, there was some shared variance across problem areas [4]. Therefore, it was assumed that perhaps a more parsimonious set of problem areas might empirically emerge and assist in identifying common underlying pathways in the future. This effort might in turn provide a more efficient way to classify and manage the diverse set of the problems that can be observed in BCS.

Hierarchical cluster analysis indicated that the broad array of problem areas that can be experienced by BCS fell into two clusters: (1) cases that report higher levels of functional limitations and (2) cases with multiple elevated problem areas, or a pattern characterized by lower levels of health-seeking skills, higher symptom burden, unhealthy lifestyle factors, and financial strain. As expected, the cluster experiencing the greater number of problems also reported a lower quality of life, providing a validation of the clustering of a two problem grouping in BCS. These findings were noted in over one thousand cases with diverse breast cancer pathology, as per medical records, in relatively young Chinese woman diagnosed and treated for breast cancer.

From both a clinical and theoretical perspectives, it is possible to observe clear subgrouping of certain concerns following cancer treatment in BCS. The hierarchical clustering in the current study provides empirical support for this type of subgrouping, indicating that the several problem areas in BCS can fall into two distinct groups or clusters. This approach may assist in the identification of potential underlying mechanisms, or common pathways of these clusters of problems. This framework could help optimize the development and application of interventions in BCS, simplifying how clinicians and researchers go about managing these multi-dimensional problems. For example, it could enable identification of a single target area, likely to have downstream benefits for other related problem areas (vs targeting each individual problem area on its own). While it is unclear just how cluster 1 (“functional limitations”) may exert its influence on some pathways, there are some possibilities that can be hypothesized for cluster 2, based on the problems that did cluster together. For example, when a BCS experiences some or many of the problems within cluster 2, it might be possible to improve levels of health care–seeking skills, which may exert a positive effect on symptoms, lifestyle, and/or financial strain. Such a relationship is only speculative at this time and is in need of direct empirical support.

Symptom science has identified patterns or clusters using reported symptoms as its focus [8]. Reports related to common underlying mechanism(s) in the presentation of symptoms, (i.e., sympathetic nervous system reactivity or immune dysfunction [8]) have been suggestive of treatment options for these symptom clusters. The current study extends this concept to multiple diverse problems beyond symptoms, indicating these diverse problems also fall into clusters. While the identification of such mechanisms was not the goal of the present study, it is intriguing that past research on symptom clusters (e.g., [8]) indicates that attempts to cluster a major problem area are possible. Next steps are to determine the mechanisms underlying the two clusters observed in the current investigation. Modifying such underlying mechanisms of these multidimensional problems might similarly improve the understanding and management of these diverse problems.

In fact, a recent investigation using cluster analysis to identify whether patterns of problems (not symptoms only) post-treatment were observed provides support for the potential of the approach used in the present study [17]. These investigators identified a set of problems in BCS that included lifestyle, self-care, emotional coping, social support, sexual health, complementary services, practical help, fear of recurrence, depression, anxiety, pain, and fatigue. Problem areas were observed and classified into four general clusters. These problem areas were named: cluster 1, “low needs”; cluster 2, “mainly physical needs”; cluster 3, “mainly psychological needs”; and cluster 4, “combined physical and psychological needs.” While this study did not measure the exact problem areas using a priori psychometrically developed scales as in the present investigation, the study did illustrate how hierarchical cluster analysis can be reduced to clusters or combinations of several problem areas (i.e., not simply symptoms). This clustering resulted in the ability to form logical groupings that also suggested either no intervention or the general types of interventions used for each observed cluster. The current study was also successful in reducing multiple problem areas into potentially more manageable clusters. Overall, the findings of both studies indicate that it is possible to conceptually reduce the multiple problems reported by BCS into clusters with two broad dimensions.

The difficulty generalizing these findings to countries other than China and to cancer survivors other than those diagnosed and treated for breast cancer with stages I–III is apparent. Also given the cross-sectional nature of the design, it is not possible to determine the causality between the multi-problem cluster and quality of life. While this study used convenience sampling, the sample was relatively large, randomly recruited from multiple cancer hospital sites, and typical of BCS survivors in China [18]. Given these minor limitations, a theoretically robust finding in which two clusters represent eighteen potential problem areas in BCS was observed.

While it is possible that other methods or tools to measure problems reported by BCS might generate different clusters than what was observed in the present study, this study did use a range of problem areas that was created from a careful review of problems reported by BCS following treatment. It was this diverse set of problems, identified by precise measures, that were synthesized into two clusters. While the clinical impact of this clustering remains to be determined, the empirical separation of these problems into two clusters suggests that further exploration of these clusters is justified. Future research should determine the common mechanisms underlying each cluster and the potential clinical efficiency and effectiveness of addressing such mechanisms of the multiple problems that are often observed among breast cancer survivors.