Rehabilitation usually promotes itself as holistic, considering the patient as a whole and being patient-centred. Using the biopsychosocial model should indeed enforce a holistic, patient-centred approach. (here) Nevertheless, there are counter-forces at play, forces that we sometimes encourage. The primary countervailing power is a desire to categorise, classify, and develop small specialise treatment programmes. For example, services offer cognitive rehabilitation programmes, upper limb programmes, and many more separate programmes. How does that fit in with a holistic approach? Why is this happening? Does it matter? I will discuss these issues.
A view from biomedical physicians
In the American Medical Association (JAMA) Journal, Maitra and Verghese recently discussed the medical imperative ‘to make a diagnosis’. (here) They contrasted the medical perspective, establishing a biological, structural or functional, cause for a patient’s illness based on a biomedical model of illness, discussed here, to the patient’s perspective, as explained by Kleinman. (here) The patient wants an explanation and uses an explanatory model of illness.
The contrast between cause and explanation may appear a subtle distinction, but it is accurate and necessary. You may tell someone that the cause of her stroke is hypertension and smoking leading to vascular disease and cerebral infarction. She may reasonably point out that (a) she has had those factors for years without having a stroke, and (b) she has friends with the same factors who never have a stroke. Moreover, why did she have her stroke at ten past five on Friday afternoon, two days after arguing with her daughter?
They review the increased ability between 1700 and 1900 to localise diseases within the body at the expense of considering illness. They refer to illness thus: “Yet, in all cultures, illness is fleshy, full of matter and corporal experience; it takes place at the bodily level.“
They then go on to point out that “Disease is not the stable, natural category many clinicians have come to rely on but a “social discourse open to intervention,” [I think they more accurately mean that illness is a social construct, rather than disease.] They emphasise, crucially, that “the biomedical label cannot encompass subjectivity and lived experience.“
They finish by stating: “Diagnosis is not the destination. The goal (and the glory) is in getting to know the unique person experiencing illness by fully understanding their biographies to appreciate the nature of the disease that affects them, and then using individualised treatment and the art of caring to bring them comfort.”
This viewpoint highlights three points:
- The drive to make a medical diagnosis still centres on identifying pathology within the body
- The patient, in contrast, wants a fuller explanation of their individual experience
- Their interpretation of a ‘proper diagnosis’ is similar to the process of formulation within rehabilitation. (here)
In summary, over the years, healthcare has moved from considering the person, with no actual knowledge of what the illness was, to a position where the diagnostic process identifies with reasonable certainty a specific structural or functional abnormality, disease or pathology, within the body. Almost complete uncertainty has been replaced by reassuring confidence, albeit t the cost of overlooking the patient.
Biopsychosocial formulation and uncertainty
The biopsychosocial model of illness is covered on several pages on this site, starting here. The features I want to stress here are that:
- it is a systems approach, including many different factors and the overall model includes at least nine subsytems
- it inevitably includes great uncertainty both about
- explanation and causes, and about
- prognosis and prediction of the effect of an intervention
- it is nevertheless the best model we have.
Mathematical complexity is the critical characteristic of any system where three or more factors have mutual and/or multiple and/or non-linear relationships. This complexity means that the overall effect of changing one element cannot be predicted.
For example, introducing targets for maximum waiting times in emergency departments was supposed to increase throughput and reduce waits. One consequence was making ambulances wait outside until the department could take the patient in and transfer them within the four-hour limit. The waiting time increased.
This unpredictability manifests as uncertainty. When analysing a situation, one can decide the most likely contributing factors, but one cannot be sure. More importantly, one can suggest interventions to change one or more aspects, but one cannot be sure that the intervention will have the desired effect. Furthermore, it is difficult to predict adverse effects that might arise.
This unpredictability is also often associated with instability. A situation may be breaking down but may remain stable as other systems flex and adapt to retain stability until, suddenly, everything fails and stability is lost.
Therefore a rehabilitation clinician must, first and foremost, recognise that every conclusion and every prediction has an element of uncertainty. It is helpful to consider prediction in terms of the upper and lower probability limits where this is possible. This technique emphasises to others how great the uncertainty is.
The clinician also needs to help others understand that absolute certainty is not possible in any area of healthcare. Instead, every decision is based upon probabilities, and realistic estimates should be used. This lack of confidence is often unsettling for patients and families, and members of the rehabilitation teams.
In summary, a holistic approach to illness, achieved using the biopsychosocial model of illness, will inevitably carry with it uncertainty. There will usually be multiple causal factors, not one unique and complete cause. There will usually be many possible interventions, but the effect of each will be unpredictable, and some interventions may need to be undertaken together to have an impact. The simple certainties associated with the biomedical analysis are replaced with a vapour where some things are reasonably conspicuous, but much is hidden.
Association and prediction.
Research is a process that reduces uncertainty; that is a reasonable definition, and research into rehabilitation does help. There is almost too much research into the associations between impairments and disabilities, with some quite complex modelling. It has more recently expanded to include contextual factors too.
For a few conditions, there is sound research evidence about prognosis. Stroke is probably the best example. Unfortunately for most disabling conditions, the only reasonably obvious predictor is global severity; the more impaired or disabled a person is at one point, the more disabled they are likely to be at some later point.
However, the greatest weakness of research into concurrent and predictive associations is that the links are statistical. They apply at a population level. They are often not applicable to a specific patient. Even for people after stroke, the evidence may fail to predict outcomes accurately in half or more of all cases. Clinical experience allows people to make some predictions, though how accurate they are is an open question.
More importantly, there is almost no good evidence to predict who will benefit from any specific intervention. There is strong evidence that input from a multi-professional rehabilitation team improves a patient’s outcome compared with not having such input. There is also strong evidence to support some generic interventions such as exercise and practice of activities. (here)
However, all that can be said is that a person receiving rehabilitation has a greater probability of improving than a person who does not. There is no evidence to show who is more or less likely to benefit. There is no evidence to identify people who will or will not benefit. (here)
This inability to predict individual benefit is no different from the situation for almost all healthcare interventions – taking drugs to control hypertension and reduce the risk of vascular events, taking an antibiotic for a urinary tract infection, having an operation to remove a tumour. In some cases, the size of the probability of success may make success seem inevitable, but it is not. In most cases, the ability to predict is similar to that in rehabilitation.
In summary, for many conditions, one can know at a population level what factors have significant influence upon some clinical feature, and one can predict at a population level what the clinical situation might be like a few months later. These predictions are rarely robust and are often very weak. Most importantly, they cannot be used at the level of an individual patient. At best, when the association is strong at a population level, they may alter probability a bit.
Criteria and categorisation of patients
Despite the almost complete absence of evidence, commissioners and the managers of services usually require a healthcare service to set out selection criteria. Many rehabilitation teams have complied. It is probably more comfortable to have a set of criteria, even if they have no evidence to support them than it is to have to consider each patient separately. When measures are derived from evidence, the evidence concerns the natural history and does not concern who will benefit.
At the same time, even without pressure from commissioners, we categorise patients. All the time. Without evidence that the categories are appropriate for the purpose. Without considering the potential risks.
Examples of categorisation
Prolonged disorders of consciousness
When prolonged unconsciousness first became a recognised problem in 1976, it was described as the “vegetative state”. This term became equated to a clinical state of being unaware (of self or environment) when the law needed a basis for withdrawing active medical treatment in 1992. Further clinical experience, and legal arguments, led to the recognition that awareness was a spectrum, and the minimally conscious state was born. This state itself became subdivided into a plus and minus pair of states. Yet, when the evidence is reviewed, there are no definite boundaries. (here) Indeed, it is challenging to define unequivocally when a prolonged disorder of consciousness ends.
This lack of specificity would not matter much, except that both clinicians and relatives attach great significance to a definite ‘diagnosis’ of the state of consciousness. This specificity cannot be achieved, as the well-known but erroneously interpreted comment that “40% of diagnoses are incorrect” shows. (here) A patient would base any decision about continuing life-sustaining treatment on their experience, prognosis, values, beliefs, and priorities. A patient would not base a decision on the label attached.
Suitable for (or ready for) rehabilitation.
This categorisation is frequently used. There is no evidence a clinician can use to make a decision. Any criteria used are likely to lead both to people who will not benefit from rehabilitation being taken on and, more importantly, people who would benefit from rehabilitation being excluded. (here) Decisions are ultimately clinical judgments based on individual facts. Different people make different decisions.
Decisions will inevitably be wrong, probably quite frequently. The process of selecting patients should be structured around this fact. For example, one might offer a trial of rehabilitation intervention with a plan of what to do in the event of failure to benefit. Alternatively, one might arrange to review a patient not offered rehabilitation at a later date.
This move towards categorising patients without any evidential basis is unfortunate because it makes decisions less suitable; patients suffer. I will give two examples.
In 1986, about four days after starting as a consultant, I was asked to review a patient in the neurology department “to advise because he has been rejected for rehabilitation”. On arrival, I looked at the notes to see why he had been rejected. The consultant had written, more-or-less exactly: “Thank you for asking me to see Mr XX who has had a stroke. He is an ideal candidate for rehabilitation. Unfortunately he is over 65 years old, and we cannot take him.“
He was about a month over 65 years. Funders imposed no rule. It was purely that the consultant had, in the distant past, decided that the Rivermead Rehabilitation Centre was a national service and that it would not take people over the age of 65 years because they were old and could not benefit.
Two years later, I was running a ‘Young Disabled Unit’, Ritchie Russell House. The staff told me they did not believe in the so-called medical model,/ and they made decisions centred on the patient, implying that decisions were not based on the diagnosis. Imagine my surprise when I read out a letter referring a 45-year-old man with Huntington’s disease.
As soon as I gave his diagnosis said “Oh, stop. We do not take people with Huntington’s disease.” This was, apparently, written into a constitution (this was an NHS service) and was justified by saying that “they are psychiatric patients”. Needless to say, we did see him. They had categorised suitability by diagnosis, not by need.
Categorisation of services
A further unfortunate type of categorisation occurs. Services are categorised, again with little consideration of the consequences. The obvious example in the UK are the Level I, level II services etc. These require the patients to be categorised so that they only attend the ‘correct’ service, and so that they move out of a service when their state no longer meets the criteria.
Less obvious, and perhaps more pernicious, is the self-imposed categorisation of services. This leads to the disintegration of care.
In 2002, the Rivermead Rehabilitation Centre and Ritchie Russell House were merged together with Mary Marlborough Lodge Specialist Disability Services (an expert service is Assistive Technology), the Wheelchair services, and the Prosthetics service to form the Oxford Centre of Enablement. One of the reasons put forward was that it would offer a “one-stop shop” with “seamless services” for our patients.
Within a few days, it was obvious that this would not occur. Asking the Specialist Disability Service (located in one corridor) to see a ward in-patient (at the end of the corridor) required a new referral, going on a waiting list, a ‘home visit’ (to the ward), and getting funding permission.
Later on the services developed specialist clinics and services such as a spasticity clinic, and an orthotics clinic and again inpatients were fitting into these services rather than being seen when needed, despite the fact that the same people were doing the work. At least one driver of this behaviour was the management. Specific services asked for and received specific contracts which set specific criteria and processes. I do not know what has happened since I stopped working there in 2016, but I suspect the situation is similar.
The Queen Square Upper Limb Neurorehabilitation Programme is an example. (here) I will start by emphasising that I support wholeheartedly their research, and I would support any other specialist service doing research in isolation from a patient’s local services. I also support them in providing a service as one way of attracting people with a problem that they are researching.
On the other hand, I question the sense in someone being referred by a General Practitioner or other rehabilitation service to have one very specific, focused intervention. The affected arm will not be the patient’s only problem, and other problems might be more easily treated. In most cases, the gains made will be modest, and will not make much impact on all the other difficulties.
There are two risks. It is probable that the patient’s expectations will far exceed what is possible, despite the great care that is doubtless given to ensure a realistic expectation. It is also probable that, without careful follow-up recommendations and encouragement, some of the gains will be lost. I am sure that telephone or virtual follow-up occurs, but follow-up in a local service would be more effective.
The more important, and much more frequent problem arises when a rehabilitation develops specific ‘programmes’ focused on a common problem or frequently given intervention.
An example of this is the provision of wheelchairs. Wheelchair services are all separate from any of the many services providing rehabilitation to patients needing a wheelchair. The wheelchair service has very specific criteria, and will often not provide a wheelchair. The wheelchair service is grossly underfunded, leading to long delays, and often a restriction on what can be provided. The wheelchair service rarely if ever ensures proper training in the use of wheelchairs. It is not integrated into rehabilitation services.
There are many other examples. Services develop separate programmes such as a fatigue management service for people with multiple sclerosis, and possibly a separate one for people with stroke, and there will already be a chronic fatigue syndrome service. But there will be many other patients with fatigue associated with traumatic brain injury, anoxic brain injury, Parkinson’s disease etc.
These specific programmes are often commissioned, often at the request of the provider. Commissioning a service leads to a requirement for inclusion and exclusion criteria, defined periods of treatment etc. Furthermore, the programme is run outside a person’s general rehabilitation. A patient may be having a variety of therapies for their problems, coordinated as a whole, but attending the programme is a separate, non-integrated process.
This approach also results in other patients, with identical problems, not receiving a service. All the staff with specific expertise in, say, fatigue work in the programme and are less available for all other patients.
The end result of the development of specific programmes within a service is the fragmentation of a patient’s rehabilitation and a return to the approach that more biomedical services are turning away from. Healthcare services, for many years, developed a production line approach where the task, diagnosing and managing a person’s illness was broken down into specialities divided by organ systems (e.g. cardiology, rheumatology) and process (e.g. radiology, haematology, surgery). We risk fragmenting a patient’s rehabilitation into a series of weakly connected programmes, losing sight of the patient as a whole.
Summary of the argument
Between about 1950 and 2000, healthcare processes became increasingly industrialised. The professionals working in healthcare and the services provided became more and more specialised in smaller and smaller parts of the illness. For many years now, it has been agreed that the General Practitioner was perhaps the only person who considered the patients as a whole. The failure to train and fund enough general practitioners (and, indeed, enough doctors and all other professions) has meant that even general practitioners now have less ability to consider the patient as a whole. This fragmentation was captured of care was captured in a recent viewpoint article. (here)
Rehabilitation started as a speciality with a reasonably holistic view of the patient, based on the biopsychosocial model of illness. However, a combination of the uncertainties associated with the biopsychosocial model of illness and the pressures from commissioners to define precisely the service or services being provided has led to increasing fragmentation of services, even when the service is a unitary organisation. Added to this has been a tendency to provide specific expert rehabilitation services for specific, relatively common problems.
This fragmentation has been based on the categorisation of patients and their problems, and of services, with criteria being used to define the categories. Many of these criteria have no substantial evidence to support them. Others have some evidence to support them, usually as factors predicting prognosis.
The criteria derived from studies on prognosis suffer from two substantial, I would say fatal, flaws. First, the evidence relates to predicting outcomes in general. It does not derive from studies investigating what factors predict how well someone will respond to therapy or other intervention. (here) Second, the ability of the factors known to predict outcome (not benefit) is generally weak, rarely accounting for more than 30% of the variance. In other words, they are statistically associated in large groups but have little ability to predict outcomes in an individual.
There is a further substantial problem. Any test used to predict an outcome, and this includes predicting the presence of pathology, for example, will inevitably have a proportion of false positives and false negatives. Importantly, the proportion of incorrect results is crucially dependent on the prior probability. What this means is that a test applied to one population cannot be used in a population with a significantly different frequency of the key criteria or a significantly different frequency of the expected outcome.
This statistical fact generally means that in order to be 95% certain of identifying all those who are positive, well over 50% of those selected will not actually be positive. Alternatively, if the aim is only to allow 5% of those selected to be negative, then well over 50% of those who should have been selected will be excluded. This is illustrated here.
No one has compared the alternative, clinical judgment, against criteria when it comes to selecting for rehabilitation in general, or for a specific intervention, but it is unlikely to be worse and is likely to be better.
Rehabilitation services should avoid categorising patients unless there is good evidence to support any criteria used for the purpose. The evidence must come from studies investigating criteria for that purpose. If any such criteria are discovered, it is vital that the sensitivity and specificity of the criteria are established. These studies must be carried out on populations similar to those where the criteria are to be used.
Rehabilitation should also stop asking commissioners to commission specific rehabilitation programmes because doing so leads to a fragmentation of a patient’s care and is not patient-centred. It is reasonable to treat patients in groups if that is efficient and effective. It is also reasonable to offer patients a formal, semi-structured programme of treatment, provided the specific details such as duration, intensity, and stepping rules are tailored to the patient’s needs and not to some predetermined criterion.
Rehabilitation should prioritise being patient-centred (here) and tailoring rehabilitation to the patient at all times.