To our knowledge, this is the first study describing a care situation as a CAS and analysing the elements explaining complexity, not only in palliative care but in health care in general. The criteria contributing to complexity of palliative care situations from the professionals´ perspective could be allocated to three systems of the overall CAS of a palliative care situation: the system patient, the social system and the system team as well as to environmental factors. The developed conceptual framework reflects the holistic approach of palliative care and highlights that elements, such as symptoms, persons or certain family relations, cannot be understood independently and separated from the overall system of the palliative care situation.
It could be argued that the results merely mirror the domains of care (physical, psychological, social and spiritual) incorporated by the holistic model of palliative care, and that knowledge of this model might even have limited the participant’s answers to these domains. The findings are certainly shaped by the domain-based understanding of palliative care. They are, however, not limited to those. The experts not only described system elements and their relationships associated with these domains, but also additional aspects of complexity, such as dynamics and interactions of these elements as well as environmental factors and team aspects. The findings suggest that the existing domain-based model of palliative care does not comprehensively describe complexity of a care situation, since it does not incorporate these additional aspects of complexity.
The understanding of the palliative care situation as a CAS supports and supplements findings from other studies on complexity and palliative care. On the patient level, Pask’s et al. findings of applying Bronfenbrenner’s Ecological Systems Theory to the complexity of patients’ and families’ needs also show that there is more to complexity in palliative care than the physical, psychological, spiritual and social dimension . They identified additional components of complexity, such as dynamics, relationships, influence on the societal and organisational level, which agree with the conceptual framework presented in this paper. The results from Tuca et al. indicate that interactions between the variables included in their study predicted complexity better than the sole variables . The Spanish research group suggests complexity to be a multidimensional construct complying with complexity theory. In terms of CAS, Ciemins et al. pointed out that it is supportive for the work of the multiprofessional team to comprehend patients, families, teams and organisations as CAS . CAS has been suggested as an appropriate conceptual framework to understand team processes and support team development [32, 33]. Defining the palliative care situation as a CAS provides a systemic view in which the patient and his or her relatives are still central elements, but in addition, the team assumes a position within the care situation. Besides, it merges various hierarchical levels and enables the understanding of lower hierarchical level agents such as symptoms acting and interacting with elements of higher hierarchical level, such as the team. The application of CAS theory supports a better understanding, building the theoretical foundation upon which to develop a situation sensitive method of problem-solving – not only in the palliative care context. The findings of this study depict the CAS of a specific problem and show how other problems of health care can also be framed by systems thinking.
Using CAS framework to influence system behaviour
Some of the identified system elements and environmental factors do not refer to the patient but are imposed by the organisation and management of care. Structural and process characteristics on the level of the team, the care organisation or the health care system influence the system behaviour. Acknowledging the effect of structural and process characteristics on the complexity of care situations enables the development of strategies to influence the system behaviour and outcomes by reshaping these characteristics, for example by setting appropriate incentives in payment for care. Changes regarding the timing of integrating palliative care in the care trajectory may for instance have an impact on the continuity of care, enable easier transitions for patients and carers and thus result in an increased quality of care [34, 35]. In consequence, this could potentially decrease the complexity of a care situation. The specification of quality criteria for care facilities on a structural level, such as the number of team members and professions within the team, enables the creation of a constant on the structural level. This would enable the evaluation and comparison of the complexity of a care situation independent of differences on the organisational level.
Pype et al. pointed out that in social systems such as a palliative care team, the agents’ internalized rules are subject to change . Considering the CAS of a palliative care situation, this also applies to other social agents involved: the patient and individuals in the social system. In social agents, the internalized emotional, cognitive and social rules are not static and are subject to change, if e.g. a person reflects on those rules and consciously changes them or if rules are dictated and changed by the environment . For example, legal specifications, documentation requirements, or funding structures provide rules, which the team follows. If external rules change (e.g. a legislative change), the team will adapt its behaviour which will in turn have an impact on the overall CAS of a palliative care situation. While Pype et al. focus on how this understanding may influence team behaviour and can be used for team development purposes, the team’s integration in the overall CAS of a palliative care situation suggests that these changes will also have an impact on other system elements in the realm of the patient and social system and therefore the overall care situation.
Looking at different system elements (and environmental factors) referring to structure may help to discover potential for change and improvement of quality of care.
Using CAS framework for differentiation of patients’ needs
CAS theory not only offers a comprehensive conceptual framework for problem solving in palliative care. It can also be used to support the development of a systematic approach to differentiate patients according to their need for general and specialized care. The CAS of a palliative care situation provides potential criteria for the classification of complexity. Since the emphasis of CAS is on relations between elements, criteria included in a classification need to account for that. In fact, a classification such as the diagnosis related groups (DRG) system in health care taking only diagnoses and procedures into account is too reductionist to meet the multifaceted nature and relations of the palliative care situation. Therefore, it is unfit to mirror complexity and resource needs. The development of a model or classification of complexity certainly requires the reduction and simplification of information to make it measurable. This holds two major challenges: 1) Not all elements identified to add to complexity are measurable. Elements such as the patient’s personality, prior experience with the health care system or a difficult underlying family situation may have a major impact on the system behaviour but cannot be assessed easily and accordingly cannot be included in the modelling. 2) The large number of elements and relations needs to be reduced to a manageable number for assessment which still describes a situation comprehensively.
The in-depth understanding of interdependences may help to find alternative ways of incorporating system elements which cannot be measured or whose measurement would be too resource-intensive. The knowledge of their influence on other system elements allows involving them indirectly in a classification. Accordingly, the understanding of interdependences can be used to reduce the number of variables without oversimplifying information.
According to complexity science, the degree of complexity depends on the number of system elements, such as symptoms and social agents, environmental factors, and the quality of the relations with each other. Statistical modelling methods need to account for that. Methods arising from the traditional reductionist paradigm of science aiming for principles which follow the assumption of linear relations are not appropriate to deal with complex problems since they strongly reduce and oversimplify information [18, 19, 37, 38]. An appropriate method needs to reflect relationships and build on multivariable analysis methods such as applied in the development of the Australian palliative care classification .
Three of the four factors used in the Australian classification – functional status, problem severity and age – are also represented in the elements of the CAS and could be used as a starting point for a German classification. Phase of illness as the factor predicting resource use best in the Australian studies was not directly identified in our data. The concept of “phase of illness” could, however, be understood as a result of the presence of and interactions between the identified elements and factors.
The use of attractors in modelling a patient classification
With the idea of attractors, CAS theory offers an additional approach to assess complexity of care situations. Attractors are states which the system will adopt over the course of time and through the system behaviour. The system behaviour is the result of interacting agents.
The data in our study did not provide any states which could be interpreted as attractors of the CAS of a palliative care situation. However, attractors are defined by the problem and by the system tailored to the problem. For example, in the psycho-spiritual subsystem, “coping with disease” was acknowledged as the process of the subsystem’s self-organisation. It could be argued that the stages of coping with the disease can be understood as the attractors of the subsystem. On the higher level of the system patient, phase of illness, as proposed by Masso et al., and used in the Australian AN-SNAP classification [10, 39], could be defined as attractor. While the disease progresses, the patient will change between these phases: stable, unstable, deteriorating and dying. Hence, phases of illness are states which will be adopted by the patients, independent from the disease, symptoms, social situation, etc. The phases refer to the patient as well as the carers and reflect the concept of the unit of care inherent to palliative care. The description of the phases includes several references to the carers’ situation and how it may influence the care situation . Furthermore, phases of illness do not follow a predefined order. Patients and care situations can move between phases in any direction . The patient and the respective care situation will always be in one of the phases or in transition between two phases. This complies with the concept of attractors. Hence, the CAS concept of attractors enables the inclusion of a measurable variable reflecting several elements and relations, in this case phase of illness, into the concept and the classification of complexity with respect to patients in need for palliative care.
Since attractors are a construct, it is not possible to determine which agents and relations are covered by them. Phase of illness refers to the patient and the social system. The system team and environmental factors are not considered in the concept. Accordingly, the use of phase of illness as the sole predictor for resource use would not be appropriate since it entails the risk of excluding relevant system elements, environmental factors, and relations.
Implications for practice, policy and future research
The approach applied in our analysis will contribute to overcoming the present arbitrariness in the use of the term and the concept of complexity, and thereby lay a foundation for future theoretical modelling and clinical applications. In terms of a necessary operationalisation of complexity, a set of relevant outcome measures needs to be identified which can and should be clinically applied. As shown in the Australian AN-SNAP model such outcome measures can be used for a classification to differentiate patients according to their needs, benchmark palliative care services , and as a basis for a financing model . Our data suggest that, in accordance with the developments in Australia, these outcome measures should cover problem severity, functional status, and potentially phase of illness. The current version of the AN-SNAP classification consists of 30 classes, 21 of which refer to adult patients . Furthermore, classes are divided by in-patient and home care situations, reflecting the relevance of the care setting as acknowledged by the environmental factor “care setting” in our findings.
In Australia the Palliative care problem severity score is used for the classification, measuring pain, other symptoms, psychological and spiritual distress of the patient and carer burden . In Germany, the Integrated Patient Outcome Scale (IPOS) and the Symptom and Problem Checklist of the German Hospice and Palliative Care Evaluation (HOPE) are validated outcome measures well established in clinical care [43,44,45]. Especially IPOS can be considered a suitable instrument to routinely measure factors influencing the complexity of a palliative care situation. Apart from questions regarding the distress caused by physical symptoms, IPOS also covers questions regarding the psychological and spiritual situation of the patient as well as practical problems and carer burden [43, 44]. Also, the IPOS offers a more comprehensive problem assessment than the Palliative care problem severity score, since it explicitly covers other symptoms, such as breathlessness, which have major effect on complexity. Physical impairment, also included in the AN-SNAP classification, can be measured by the Australian Karnofsky Performance Status or the 20-point Modified Barthel Index [46,47,48]. These already established outcome measures offer starting points for the measurment of system elements identified in this study, which could be involved in a classification by scores or categories.
The Australian classification can be considered as a successful example for the development and use of a classification and can be an orientation point for the development of a classification in Germany and other countries. However, as systems thinking suggests, even a successfully used classification cannot be seen independently from its superordinate system. The Australian classification cannot simply be transferred to other countries due to differences in health care systems, organisations and work place culture. Further research is needed in Germany and other countries to enable classifications fitting the respective national system characteristics.
Furthermore, our findings address the demand for a stronger theoretical foundation of health services research. Complex problems cannot be represented adequately by a scientific understanding of linear causalities usually prevailing in medical research. Future research concerning complexity in palliative care may benefit from drawing on the theoretical model of CAS throughout all phases of the research process, including the definition of the research question, the identification, operationalisation and measurement of relevant parameters, and the interpretation of findings. The consideration of the CAS as a theoretical framework may be particularly useful in the development of interventions, and in implementation research, since the anticipation and understanding of complex interactions will be vital for the successful realisation of innovation and change in healthcare. This may also involve a stronger focus on healthcare providers such as teams or individual healthcare professionals as agents in the care system, contributing to the outcome of care, and hence constituting a relevant research variable.
Strengths and limitations
To our knowledge, this is the first study to systematically analyse the definition and operationalisation of complexity in palliative care using the framework of complex adaptive systems.
A particular strength of this study was the relatively large sample including stakeholders with diverse perspectives on palliative care, represented by clinical experts as well as experts with political and economic background. The two sample groups (group a and group b) and the heterogeneity of the experts included regarding the selection criteria (profession, care setting, rural or urban area, geographical region and university affiliation of the centre) were selected to ascertain variation in perspectives and hereby reduce potential bias.
A limitation of this study is that it focused on the professional carers’ perspectives on complexity only. Due to resource limitations patients’ and carers’ perspectives were not included in this analysis and their needs incorporated in the results are based on the professionals’ perspective. Besides, the study was only conducted in one country. However, in the meantime, a study exploring the perspective of patients and carers in the UK has been published. The results confirm our findings and do not show any additional elements not represented in our data .
This paper provides a conceptional framework and a comprehensive understanding for complexity in palliative care. On the level of the individual care situation, the systemic view can help to understand and shape situations and dynamics. On a higher hierarchical level, it can support an understanding and a framework for the development of care structures and concepts.
The framework and the identified system elements can be used as a basis for the development of a classification of complexity in palliative care, drawing on a differentiation of patients according to their care needs. Relevant outcome measures mirroring the identified system elements have to be identified and implemented in clinical practice. The consideration of phases of illness as an attractor may constitute a promising starting point for the operationalisation of complexity in research, clinical practice, and health policy planning. Further elaboration of relevant parameters and suitable methodology to adequately model complexity should be pursued in future research and theory-based deliberation among interdisciplinary experts.