In this scoping review, we identified 70 articles addressing how customization approaches are delivered in health care. Our results indicate the absence of a common strategy for delivering customized care based on a conceptual basis. First, while the benefit of a customized delivery model is often advanced in recent years (the majority of papers have been published in the past four years), there is little conceptual basis that supports current efforts (the reference to a conceptual basis is only mentioned in about half of the studies). In this case, studies refer to a variety of models and theories (e.g., individualized care or consumer-centered care; patient-centered care or precision medicine; social marketing theories or models equity-oriented health care and “essential care”) (75) (39). The theory of the “MC” is for its part little cited (24, 26, 76).
Second, the current efforts appear to be carried out in parallel, according to different logics of action. By logic of action, we refer to the definition of Strauss et al. (77) in which specific participants reflect common motivation for acting (i.e. categorizing patient characteristics and customized interventional responses) (77). Depending on the focus given to one patient characteristic, or to the different association between them, four distinct logics of action can be defined: 1. Clinical; 2. Population health; 3. Patient-centered care and 4. Financial. Segmenting patients according to their clinical characteristics is “natural” in health care, and flows from the fact each patient is a single case. This clinical logic has recently been updated with the emergence of personalized and precision medicine, as highlighted by different articles (50, 78). Beside this clinical logic, a second logic orients customization development to the psychosocial needs of the patient. In this instance, customization efforts have generally the goal to reduce inequalities to access and outcomes across populations. A good example of this comes from Ford-Gilboe et al. (40), who demonstrated that providing more equity-oriented health care improved health outcomes for people living in marginal conditions. The responsibility for structuring population health customization is most often borne by public health professionals, academics engaged with communities, local decision-makers, and social workers (79, 80). The third logic, patient-centered care, describes how healthcare organizations respond to patient demands and preferences. In this context, care-giving is seen as a “service” on behalf of, and at the direction of patients. This logic is often referred to as “patient-centered care”, but in some cases is viewed through the notion of individualized care or marketing theories (42). When patients are targeted by the costs their processes of care generate, they fall into a fourth financial logic. In this case, customization strategies have a common goal to target higher resource users, and to propose a customized intervention specifically to this subpopulation, applying it in various conditions. This financial logic generally refers to managers of health organizations and health systems who wish to rationalize costs. These four customization logics are not separate or independent, as evidenced by the combinations between clinical, psychosocial and service needs. The most common combinations observed in our review are segmentation analyzes associating clinical, psychosocial and service needs (n = 19), and clinical and psychosocial needs (n = 11). However, each logic reflects a potentially strong force that guides the division of work in health care. As such, they may create a set of “stovepipes” where participant groups develop a specific point of view of their organizational and professional goals, and similarly, the structure of their work follows this view, potentially undermining the promise of a common care customized delivery model. For example, the enthusiasm for personalized and precision medicine contributes to the spread of individually-focused clinical practices. This may, on its own, lead to quality improvements, and thereby advance the health of an individual population (81). But this type of customization could be viewed by public health leaders as “premature” or inappropriately individualistic, thus missing out on the potential offered by broader applications of customization (82).
Given the nascent and heterogeneous state of the knowledge in this area, this is an important time to reflect on the definition and use of theoretical frameworks for building a theory-based customized care delivery model. We highlight three areas for researchers and managers involved interested in this field to move in this direction.
A first area of research could focus on the content of the segmentation analysis and the customized intervention steps in order to define a better conceptual basis. A first step consists in clarifying and increasing the characteristics of the patients taken into account during the segmentation analysis (65). If we set aside the cost criterion which embodies another logic (i.e. the measurement of efficiency), many criteria are used to cover the needs, demands and preferences of patients. Beyond clarifying the concept of clinical and social needs (41), different articles stress the role of other patient characteristics: patient behaviors (e.g. the impact of negative personality traits in care delivery (35, 69), and preferences or demands expressed on subjects unrelated to health, but related to certain aspects of daily life, when encountering a health issue (e.g. lifestyle changes to prevent childhood obesity) (61). They suggest that the usual characteristics of age, sex, clinical condition and education, may not predict how patients tolerate the adaptation of particular interventions (68), and call for opening up of patient characteristics used during segmentation analysis, and the type of data collected (67, 83). In this consideration of a greater number of patient characteristics, modern profiling methods based on the processing of big data, may also have a role (44). They can help develop forms of segmentation that identify the needs and demands of each patient (84). This trend, if it is confirmed, could influence the conceptual basis of any care customization delivery model. More than a "MC" model (from mass to customization), it is a model of "singularity on a large scale" (from singularity to mass) that acknowledges the uniqueness of each patient by capturing a variety of needs and demands, which would serve as a conceptual basis.
Research could also help structure the content of a customized intervention. Several articles underlined the importance of coordination and communication efforts in customizing interventions, in particular by emphasizing the role of structural integration or developing multidisciplinary teams (61, 62, 63). Other articles have stressed the importance of personalizing professional-patient relationships (85) while some others have stressed the key-role of modular packages (26). The variety of actions listed during customization interventions calls for a more precise content, and how to apply them to the different stages of the patient pathway, within hospitals, but also during transitional care and primary care.
A second area of research relates to the added-value of a care customization delivery model. Our review reveals some positive impacts of care customization on quality, patient experience and costs. However, if we assume that more customization brings better outcomes and more satisfying experiences, the impact on costs requires more investigations. Investments for increased customization may bring additional costs at the different steps: investment in new methods of segmentation analysis (even if analytical algorithms can facilitate it) (86); new services (e.g. new therapies, “concierge” systems, home services) (87); or new forms of coordination between different healthcare professionals, managers, and social workers. Equally, several studies reported savings generated by customization, by reducing unnecessary hospitalization (54, 62), treatment costs (73), and duplication (88). Other studies (8, 51) highlighted that an earlier and more focused identification of “complex” patients and/or high user can help health care organizations design more appropriate and efficient organizational responses. Chaudhuri and Lillrank (76) also argued that a customized strategy applied to high volumes of similar patients, and could be economic by implementing common standards of care. However, these data are sparse, and require more research of any care customization interventions to give evidence of their added-value, and in particular, of their economic viability (89).
A third last area of research could explore how to align the logics of action that underlie the efforts of care customization. While a single approach to customization of care is probably unrealistic, the compartmentalization of different logics may limit the impact of current customization efforts, resulting in additional costs. Researchers can help unravel the logics of action that support such developments, and find ways to facilitate their alignment, as in the case of customizing "high-need, high-cost patients" (8).
There are several limitations to this scoping exercise. Firstly, customization of care is rapidly growing and changing. There may be recent or more current efforts in developing interventions that have not yet appeared in the literature, or during our search, and as such, we may have inadvertently missed recently published articles. We excluded articles that addressed care customization, but not in routine healthcare delivery contexts (i.e. in clinical trials where they are not applied in real life contexts, fundamental research, or innovations at early development stages). As such, it is likely we underestimated the attributes of recent care customization strategies. The number of excluded clinical trials involving personalized medicine suggests these represent a dominant area of care customization in the mid-term.
Secondly, although we conducted an extensive literature review using a wide variety of terms capturing customization relevant articles, it is possible some articles may not have been identified. Our decision to include only studies that reported care customization may have excluded studies that addressed patient characteristics analysis in a customized effort, but were not labeled as such, or were not accurately represented in the abstract. To limit the impact of this selection process, we studied bibliographies of selected articles, and added six more relevant articles. We also selected only articles identifying customization strategies (i.e. segmentation analysis that led to a customized intervention) as the driver of this review, and we excluded segmentation initiatives that may have added insights (e.g. patient-profiling questionnaires and machine learning methods). In some cases, it was difficult to assess if segmentation analyses were accompanied by a customized intervention, or to circumscribe the notion of segmentation analysis itself, by comparison with stratification methods. To limit the incorrect exclusion of some articles, investigators discussed these cases. Last, we did not include “grey” literature; including newsletters, professional association or institutional news, and publicity publications. As care delivery systems and its analysis are not delineated research elements in medicine, this absence may have represented a bias that overlooked recently developed customized strategies, but not yet published. However, the analytical review of bibliographies in each selected articles did not uncover any more relevant information. Third, it is likely our selection process omitted specific research on one of the four customization approach steps, potentially missing important studies. For example, the definition of a “modular package” requires a better understanding of how the range of care and services could be combined into packages, and be pre-grouped as studies have shown (26, 27). The same was also true for related issues such as care customization from regulatory and ethical perspectives (e.g. individual privacy, segmentation and discrimination of sub-populations based on ethnicity). These elements represent interesting research perspective for the future, and can help improve the conceptual of a customized care delivery model.