Bo Strangert (RD17)

On a design-scientific approach 

to healthcare planning 

A review (RD16) considered why it is difficult to infer complex regulatory mechanisms by merely collecting and analyzing empirical data. One question was how to explain the regulation of a negative effect E on an individual due to a disturbance D. The reasoning showed that the standard of representing the connection between D and E by a mediating link T, according to a composite function [e = g(f(d))] runs into difficulties when there are interactions between a large set of variables and also incomplete data. A strict inductive way of construing regulatory concepts seems not suitable.

At first sight, this conclusion seems not to be congruent with the common paradigm in clinical healthcare that involves identifying specific symptoms (E) of disturbances (D), which can be diagnosed (explained) either as defective inherent life functions (T) or as negative causal effects of external influences on the system (e.g. virus infections). 

Decomposing complex phenomena by specialization 

through analysis and synthesis 

That paradigm has been extremely successful for an increasing number of illnesses and damages, where critical T-links or external causal agents can be diagnosed and treated. In general terms, it means that it is possible to decompose the complex bodily system into separate subsystems whose malfunctioning can be treated and often prevented, too. The analytical progress is closely connected to the existence of physical laws and advanced scientific methods.

Obviously, the development of medical special functions facilitates both population studies and clinical practice. For example, a specific disturbance of an individual can be quantitatively compared to population indices of relevant symptoms but also examined qualitatively regarding individual-specific causal or contributing factors. As already stated, it is a consequence of the progress of medical science to analyze many complexities of life functions. Additional aspects on both progress and hindrances for the transformation of complex systems into complicated ones are discussed in Applied Research Methods for Development Projects, section 7.

Possible limits of paradigms for analysis and decomposition

of complex systems

However, the present discussion focuses on the limits of such transformations and how to cope with this challenge. Of course, some hindrances can be temporary but solvable by present analytical tools. For example, some fragmented relations in Figure 1, RD16, can be regarded as incomplete but possible to elaborate and reformulate in categories, indicating specific illnesses or damages for defined populations of individuals. Nevertheless, other hindrances may be impossible to formulate sufficiently well by equation systems or logical representations.

Some areas of psychological and social nature require more advanced approaches to master complexity is. To a certain extent that may depend on the relative immaturity of their respective academic disciplines. Moreover, the needs for theorizing and experimentation about complexity issues prevail in all cases that have to manage multiple interactions, characterized as uncertain and dynamic, often with nontransparent layers of processing. Many such cases have outcome spaces that are impossible to predict or calculate in practice.

The present object of exploration, that is, the design of support systems for long-term inpatients, is in principle a task of innumerable requirements and regards. Just consider typical requirements of a ”holistic view” of patients (i.e. not neglecting psychic and social aspects), or of an individualized treatment plan, or all patients’ right to self-determination. Add the complications that each patient has unique personal preferences, experiences, and capabilities, and probably risks degeneration of crucial life functions due to age or dementia, poor social and physical healthcare contexts, and so forth.

On top of that, any designer of support systems must also consider which socio-cultural, organizational and economic conditions must be congruent with the stated goals of healthcare.

The idea of the design task as consisting of a set of interactive and diverse conditions will certainly force most designers to lean towards existing practices and try to remake them to master possible deficiencies. A plausible strategy is then to formulate a refined set of requirement specifications based on customary categories of activities. The functional integration of specific means-ends structures will probably follow linear assumptions, ultimately resulting in tendencies towards sub-optimizing certain means and ends. A plausible guess is that such strategies lead to little genuine progress, despite recurrent attempts.

In summary, a readymade design plan is neither logically suitable nor practical. Probably, it would involve unrealistic assumptions about analyzability and integrability regarding complex matters.  Furthermore, its construction has to comply with strong demands from end-users not to change the existing practices too much. If these restrictions seem plausible but important to overcome, then some theoretical and methodical prerequisites have to be altered.

An alternative control-theoretic approach

Theoretical adjustment. An alternative start of development is to presuppose a simple but open regulatory concept structure to steer both the continuing concept development (CD) and the following empirical design work. Briefly, in the present case, it means to create a general conceptual frame that has important implications for the relations between all relevant components of the healthcare practice. Thus, it is contrary to an approach that takes concrete empirical observations and structures as given and a base for revision of particulars. And it is also contrary to the common underlying theoretical assumption of linear decomposition and synthesis.

Applied to the foregoing example of the relation between disturbances D and effects E, it could mean that an abstract regulatory concept R has to be defined in addition to an intermediating system link T. The regulation R can exercise its influence by feedback (error-controlled regulation) or by feedforward (planning makes proactive action possible). Figure 1 shows a simple graphical illustration of a preliminary concept structure, adapted from Ashby’s (1956,  Ch.11) abstract machine model. It is explained in RD15.

Figure 1. Example of an initial regulatory model 

   (adapted from Ashby, 1956).

The formal approach can be roughly summarized in a few points:

1. The core task in healthcare planning is defined as a set of superordinate general control processes (R*)and not in terms of a set of support functions for specific D-T-E connections.

2. Understanding a patient’s self-regulation R_p is a basic precondition for this type of healthcare planning. This task should consider all relevant individual characteristics and their possible interactions. The regulation C is possible superordinate self-regulation (e.g. a patient’s expressed will).

3. The selection and priorities of support (R_t) should be based on explicit principles of self-determination and participation for the patient as well as on ”medical correctness”. This point also implies a suitable trade-off between specific physical and more fuzzy psychic-social support operations.

4. The healthcare plan must account for the patient’s possible incapability of self-control (R_p and C), for example in cases of dementia, psychiatric or terminal illness.

5. Points 2-4 imply a composite function of the two regulatory functions, R* = (R_p °  R_t). The composite function of regulations changes dynamically both as a result of their discrete mechanisms and by the interaction between them. This is characteristic of a self-organizing system.

6. The aim is to maximize the variety of (R_p ° R_t) relative to the variety of disturbances D for a single patient. This is a prerequisite for individual healthcare planning. However, the planning procedure should also be adaptive to diverse patients. Therefore, it must have a general and flexible form, too.

Hence, the perspective of this control approach is upon action, that is, a perspective that emphasizes the treatment aspect of healthcare. Diagnosis is a basis for decision-making about the follow-up of treatment (through feedback), aimed at continuous improvement of individual healthcare.

Consequently, treatment of complex disturbances for complex individuals is regarded as search and decision processes. The solutions can involve diverse means which still may be equifinal.

In a previous case study, healthcare planning was formally conceptualized as a finite state model. Figure 2 illustrates it as a flowchart with synchronized functions. The nodes symbolize decision points. The first operation is to compile an initial healthcare plan or revise an old one. The two associated loops represent a collection of information from the patient and various diagnostic special operations, respectively. The results form the basis for diagnostic judgment of the patient’s capability of self-regulation R_p and need for healthcare R_t. This is followed by the formulation of goals and a treatment plan; the patient should participate in this process if possible, given the state of health.

The treatment stage is the core part of the healthcare process. It includes a set of recurrent (often daily) loops and subloops for generating and testing specific care operations within the limits of the goals and treatment plan. Given that the healthcare situation is complex, there is a need for inventive actions on a broad scale and corresponding monitoring and assessment of effects. Sometimes the process needs to be conducted by trial-and-error. 

The healthcare plan is evaluated and revised either regularly or as a consequence of significant changes in states of health.

Figure 2. Flowchart of healthcare planning 

(from a previous study, RD15, M6, M7).

Some previous contributions emphasized how important it can be to choose a suitable theoretical perspective at the start of a project (e.g. RD8, RD11, RD13). When the project goal or task is complex, as in the present case study, there is a considerable risk of decomposing it into part solutions that may be relevant as such but not possible to integrate adequately into a desired complete solution. An example is a practice of prioritizing physical objectives and common methods at the expense of interacting contextual and psychosocial aspects. 

However, it is difficult to change an accustomed theoretical perspective or common practice. Theoretically, there is an abundance of common analytical methods with underlying assumptions of linear composition which are incompatible with the need of analyzing and controlling interactions. In complex cases, more advanced control-theoretic models and an open systems view are necessary from the very beginning of a project. The choice of a general regulatory concept in the case study is an example.

A corresponding but operational difficulty appears at work when old practices have to be exchanged by new ones. The crucial resistance to change is mostly socio-psychologically based. Therefore, the control model in the case study had to be carefully operationalized by a strict but comprehensible procedure of healthcare planning. All this preliminary case study work was only the beginning of a change process that required action research and development for quite a long period.

Although the choice of an open systems view with a control-theoretic stance is a relevant first step, it must be closely followed up with adequate theoretical and procedural specifications. In the case study, it meant that the patient’s regulatory system, as well as the supporting healthcare regulation, had to be further specified to conform to the requirements of both qualitative individual healthcare and general applicability to a population of diverse patients. Thus, the continued design process was, and should always be an evolutionary endeavor.

References

Ashby, R. (1956). Introduction to cybernetics. London: Chapman & Hall.

A2. Applied research methods for development projects (Section 7).

M6. Argument för modellteoretisk ansats vid utveckling av vård och omsorg.

M7. Om metodiska konsekvenser av val av perspektiv: En designvetenskaplig tillämpning inom vård och omsorg. 

RD8. Unclear reasons behind diverse perspectives in initial project plans.

RD11. Initial project planning: Theoretical considerations.

RD13. Perspectives on projects for R&D: some issues to be contemplated.

RD15. Model-theoretical reasoning to give structure to complex realities:

A case of project planning in healthcare.

RD 16. Case-based reasoning and qualitative modeling for representation of generality and individuality.