What is a diagnosis? Part 2
Last updated: November 23, 2025
In What is a diagnosis? Part 1, I explained that diagnoses can generally be divided into biomedical diagnoses supported by observable features that can be independently verified, and syndromic diagnoses based on specific collections of symptoms without independent observable features. Neither category has a complete, consistent taxonomy (system of classification).
This post examines the purpose of a diagnosis and then considers diagnoses from a systemic perspective. This analysis demonstrates that there is usually no single cause, and if a primary cause of an illness exists, it can be within any of the systems involved. Furthermore, sometimes the interaction between systems at different levels or components within a system is dysfunctional, leading to the condition without any one part being abnormal.
Table of Contents
Introduction
In my first post related to diagnosis, I pointed out that many challenges are connected to the seemingly simple label given to a person presenting with a health issue. These challenges stem from variations in diagnostic criteria and the absence of a standard classification system. I concluded that a multi-axial approach might be required, moving away from reliance on a single label for a person’s illness.
Before exploring that idea, we must clarify the purpose of a diagnosis. I suggested that diagnoses are used for many reasons, and this may contribute to the difficulties encountered.
What is a diagnosis for?
A diagnosis is an understanding of a person’s illness [OED], and it is axiomatic that this must be beneficial; without knowledge, one cannot make rational decisions.
When a diagnosis is made, its main aim is to enhance patient care by providing better advice on prognosis and management. Therefore, the process of agreeing that a diagnosis groups together patients with similar prognostic and management features serves to validate the label for those purposes.
Very few, if any, diagnoses have been validated for other purposes, and most will be invalid when applied for different reasons.
Broadly speaking, achieving a diagnosis may benefit the patient because it may improve:
- Confirming that their problems fall within the scope of healthcare services (See: my post on disease, illness, sickness, and disability)
- The explanation of what is happening to them
- Prediction of the possible future
- Advice on how to optimise their future
- Giving specific treatments to
- Cure (reverse or remove) the underlying cause
- Slow or stop further progression
- Alleviate painful or distressing experiences arising from the illness
Moreover, the patient will gain secondary benefits such as:
- Reduced anxiety about the future
- Access to societal resources providing care and support
- The opportunity to reduce the risk of further episodes
Having a diagnosis for the individual may have benefits for society, including family members, such as:
- Researching into the presumed disease or condition, finding causes and cures
- Improving public health epidemiologic understanding and potentially reducing incidence or prevalence
- Improving healthcare service design.
- Advising others, for example, when the condition is genetic or infectious
However, diagnoses are used for many other reasons, such as:
- Establishing access to healthcare; problems unrelated to ill health are not within the scope of healthcare
- Determining payment to a healthcare provider
- Calculating the premium for a health insurance policy
- Establishing a right to many socially funded financial benefits, such as sick pay, payment for long-term institutional care, and disability allowance
- Giving access to many other supporting resources, such as educational support, vocational retraining, assistive technology, and personal care.
- Determining access to specific healthcare resources such as stroke services, acquired brain injury service, and Long Covid services.
The primary purpose, which is to establish a person’s eligibility to access healthcare generally, is entirely valid because healthcare is only available to someone with a health condition, and, regardless of usage, a diagnosis implies that one has a health condition. There is, of course, an active debate about whether some conditions represent sickness and therefore warrant access to healthcare. A diagnosis is invalid for all other purposes and risks causing harm.
Malady and system malfunction.
For the rest of this post, I will refer to the person as having a malady. As I have discussed in a post on malady and another on disease, illness, sickness, and disability, malady is a flexible and adaptable term that describes anyone who feels unwell without making any further assumptions.
A diagnosis is usually presented as an unchangeable fact without further details. Medical letters often state that the person has a stroke, cancer, or inflammatory arthritis. Words like mild or severe are sometimes used without explanation. They lack enough detail to serve as a useful summary of a person’s illness.
To address this shortcoming, I will now evaluate diagnosis holistically, starting with the individual who may be unwell. In another post on biopsychosocial disequilibrium, I suggest that any significant or persistent disturbance of the balance within a person maintained by homeostatic or allostatic means will cause them to feel unwell or develop a malady. This principle is recognised for physiological parameters but also applies to social, psychological, emotional, and other parameters.
One way to understand this is to think of the person as the centre of many interacting systems at different levels. If one or more systems start to malfunction, the person will feel an unpleasant sensation, which draws attention to the malfunction. The relevant systems are those in the biopsychosocial model, with a significant expansion of the biological area. They are illustrated in the Mind Map below, which shows how they influence a person’s capacity to respond to demands.
I will discuss each system briefly.
Biochemical system.
The body includes millions of molecules, many hugely complex, all interacting; this is a biochemical system, and its main controlling component is the genes.
Genetic disorders arise from malfunctions within the highly complex network of interactions among thousands of molecules, mostly large organic molecules. Although only a small percentage of diagnoses depend on genetic data, this figure is likely to rise. I will use Huntington’s disease as an example.
Huntington’s disease results from a mutation in the gene that encodes the protein Huntingtin. The mutation causes the bases cytosine, adenine, and guanine (CAG) to repeat within the gene. The effects depend on the number of repeats. We now understand many other aspects of the genetic changes and their effects.
Clinically, 27 or fewer repetitions have no impact; between 27 and 35 repetitions, clinical features may become evident, and the risk can be passed on to children; and with 36 or more repetitions, the disorder is usually clinically detectable. Although the abnormal gene is present from birth, the illness can manifest at any age, from early adolescence to well beyond 50. The typical symptoms include movement disorder, cognitive decline, and psychiatric disturbances, but the manifestations differ considerably between patients, especially in the early stages. There are also some less common presentations.
Consequently, although a genetic abnormality may seem precise, there can be significant variability. As a diagnosis, it will rarely offer much relevant information about the individual’s problems beyond the understanding that they will likely worsen.
Cellular system.
Cellular diagnosis is the most common level of definitive diagnosis; within the body, some cells exhibit detectable changes in their structure and function. This field includes all cancers, numerous vascular diseases and their consequences, and certain viral infections. It is the quintessential form of diagnosis. Examples include myocardial infarction, cerebral glioma, and chronic lymphatic leukaemia.
The strength of this level of diagnosis lies in the certainty provided by the diagnostic abnormality observed in the individual patient, such as through biopsy and tissue examination.
In practice, diagnosis is often made indirectly through imaging techniques, clinical features, and other information, including associated biomarkers such as abnormal proteins in the blood. When a diagnosis is made indirectly, there is often uncertainty unless a definitive direct examination is carried out, but this doubt may not be obvious. For example, a cerebral tumour may appear on imaging similar to a local infective process.
A cellular diagnosis generally guides treatment choices and prognosis. On the other hand, as with all diagnoses, they do not provide information about the patient, their problems, or other relevant factors.
In other words, a diagnosis at the cellular level will be reasonably conclusive if based on a biopsy, but otherwise remains uncertain. Either way, it offers little or no information about the patient.
Organ system.
Many diagnoses are based on the level of an organ or organ system; much of the International Classification of Diseases (ICD) uses organs as the classificatory framework. Examples include acute kidney failure, thyrotoxicosis, and peripheral neuropathy. In some cases, the underlying cause is also known, such as coronary heart disease underlying heart failure, but it is the disturbed organ function that is important.
These diagnoses are usually based on altered physiological parameters that often underlie the primary and defining symptoms. Almost all diagnoses result from recognising patterns of symptoms and signs, which then lead to a test; for example, shortness of breath and difficulty breathing when lying down suggest heart failure.
A second type of organ diagnosis occurs when another cause of bodily damage incidentally impacts an organ. Stroke serves as a good example. The primary or initiating cause is vascular, but it can then affect the brain, kidney, heart, or any other part of the body, including a limb.
Therefore, focusing solely on the symptomatic malfunction of an organ system can be misleading, as there may be a more fundamental underlying cause. The main sign of a broader dysfunction could appear as more localised organ issues. Osteoporotic bone fractures and trauma exemplify similar challenges in distinguishing between manifestation and cause.
Conversely, there is usually little uncertainty in diagnosing the organ’s dysfunction. In many cases, management centres on the organ, and identifying the cause is not always necessary. The focus may legitimately be on the organ failure itself rather than the cause.
Body-level.
As Bertillon recognised in 1893, the causes of death cannot all be classified by organs (or cells). He started with “general diseases”, mainly infections.
A more effective approach is to perceive the body’s malfunction as a whole, a system made up of multiple interacting organ systems. These systems can malfunction in different ways. An external agent, such as a virus, might disrupt bodily functions. The failure of a single organ system can have widespread effects, as seen in many autoimmune disorders. In theory, a disease can occur even if no individual organ system malfunctions independently, because complex interactions might fail to stay in balance; this analysis provides an alternative explanation for some autoimmune disorders.
The body also encompasses the person and all their cognitive, emotional, and other higher-order processes, and dysfunction within this system is considered next.
The key aspect of the body-level system is that it determines the person’s behavioural capacity—the functions they can undertake if necessary. This may also be called their capabilities.
Person-level.
Person-level malfunctions are particularly fascinating because their existence is indisputable, yet they are not linked to any specific or singular cause. The biopsychosocial model of illness predicts that person-level illness will occur, whereas the biomedical model neither predicts nor explains person-level diagnoses.
The system is more sophisticated than the others, including most cognitive, emotional, and other non-sensorimotor cerebral functions, as well as the physical and social contexts and personal factors. Its complexity increases the likelihood of malfunction.
The main aspect of the person is that they generate, either directly or in response to external factors, the demands that must be met.
The hypothesis is straightforward. Just as we experience negative emotions when our physiological levels go outside normal ranges, similarly, if the balance among the many emotional, motivating, cognitive, and social factors that constitute the person shifts beyond acceptable limits, they are likely to feel unpleasant sensations. I discussed this hypothesis in my recent post on biopsychosocial disequilibrium.
The resulting group of conditions arising from malfunction within person systems is broad, and some are likely secondary to an as-yet-unproven malfunction in a lower-level system; for example, there might be a significant association between Attention Deficit Hyperactivity Disorder and a gene.
A common feature of most conditions in this category is the absence of a clear physical abnormality that explains their symptoms. Examples include functional disorders, such as chronic low back pain and chronic fatigue syndrome, psychological disorders like anxiety and depression, and behavioural disorders such as alcoholism. Psychotic disorders like schizophrenia may also fall into this category, but genetic or other factors could significantly increase a person’s risk of psychosis.
A second common feature is that, without exception, many factors across all domains of the biopsychosocial model are linked to the nature and development of person-system maladies; the extent to which they cause, precipitate, or increase the risk of these conditions is largely unknown.
Although they are merely descriptive labels categorising a patient’s condition, diagnostic labels are often also seen as implying a cause. Before I use this overview of systems to propose a way forward, I will discuss issues related to identified factors connected to a condition and the distinction between association and conclusions about cause and risk.
Association, causation, and risk.
In most Western societies, there is a strong expectation that every malady has an identifiable, usually single cause. Some people accept that certain factors may increase the likelihood of a condition or trigger a person into a malady.
Nonetheless, most people believe that all maladies are associated with a bodily abnormality, a disease. They will assert that a cause must exist for the 25% of conditions with no evident causative bodily cause. They will suggest that the cause has not been found or, occasionally, that no one believes the suggested cause. This is unfortunate as it may close off well-evidenced treatments.
In some conditions, there is an association between a condition and a bodily abnormality. At its simplest, the associated sign may be a direct consequence of the condition, such as multiple scars on someone who self-harms or liver cirrhosis in someone who drinks too much alcohol.
Other associations may represent more subtle consequences of the malady rather than a cause. An example is shown below for Chronic Fatigue Syndrome, undoubtedly a severe, disabling malady for some people, but one where some people believe strongly that there is a single underlying cause.
Chronic Fatigue Syndrome
In 1999, Gudrun Lange et al found that Brain MRI abnormalities exist in a subset of patients with chronic fatigue syndrome in a group of 18 patients who have no pre-existing illness, including psychiatric conditions, but had developed “a DSM III-R Axis-I psychiatric disorder occurring since their CFS diagnosis.” This seemed evidence of a biological cause.
However, Floris de Lange and colleagues undertook a randomised controlled trial of cognitive behavioural therapy in 22 people with chronic fatigue syndrome. At the end of treatment, cortical grey matter volume increased in the treated group, and cognitive speed improved.
This finding was supported by a recent trial in people with chronic fatigue syndrome conducted by Marieke van der Schaaf and colleagues, who examined the central effects of cognitive behavioural therapy. The therapy decreased pain and fatigue, and it resulted in detectable changes in neural connectivity among the premotor, somatomotor, and default mode networks. Other studies predicted these changes, and they were not observed in control patients.
As further evidence, another study on 89 women with chronic fatigue syndrome found that variation in grey matter volume was closely associated with variation in the experience of pain and reliably associated with Chronic Fatigue Syndrome.
The association between a factor and a malady may arise from an increase in risk arising from the factor without it being a cause. People who purchase alcoholic drinks are more likely to drink alcohol, but the purchasing does not cause them to drink. People who live in poverty in the United States or the United Kingdom are more likely to be overweight and to die young, but poverty is not a direct cause of obesity or death.
Individuals who go through major adverse events in childhood face a higher risk of developing functional disorders. The probable connection is through a diminished ability in their allostatic control systems, leading to biopsychosocial imbalance.
Person-level summary.
All person-level diagnoses are syndromes, meaning “a group of symptoms which consistently occur together, or a condition characterized by a set of associated symptoms:”. [OED] There is no additional independent observation to validate the diagnosis. Instead, validation depends on evidence that people who fulfil the selection criteria (i.e., have the agreed symptom complex) have a similar natural history or response to treatment and are discernibly different from people with alternative diagnoses.
The challenges are that:
- Symptoms are
- on a spectrum of severity
- difficult to categorise as present or absent
- present briefly in everyone from time to time, not associated with any disorder
- People rarely have all the symptoms associated with a diagnosis
- People with a syndrome often have other symptoms
- outside those identified with the syndromic diagnosis
- and, sometimes, associated with another syndrome
- Symptoms vary over time.
The problem is well illustrated by trying to separate chronic fatigue syndrome, fibromyalgia, chronic pain syndrome, and Long Covid, which all share a similar group of symptoms.
Thus, all these conditions:
- Are without any identifiable or provable agreed single cause
- Have multiple variable symptoms
- Are associated with significant limitations of activities and social participation.
Diagnosis – solutions.
The first step is to acknowledge that the current use of diagnostic labels is incoherent, ineffective, and causing more difficulties than benefits. Continuing to act as if ill-health was always attributable to a single diagnostic label or category delays progress and prolongs the increasing harm.
The next important step is to broaden our model and understanding of health, expanding it from the biomedical model to the holistic biopsychosocial model. The many valuable features of the biomedical approach must be maintained; the aim is to extend, not to dominate or replace it.
Third, we should develop our formulation skills. The biomedical model already uses differential diagnosis early on, briefly recording the clinical reasoning behind subsequent steps. However, computer systems, limited time, and the managerial pressure to create diagnostic labels that generate income all work together to hinder a comprehensive evaluation of a person’s situation.
To achieve this, clinicians, management professionals, and researchers must discuss the balance in clinical records between recording specific items needed for managerial purposes, such as service design, planning, staffing, and funding, and supporting the clinical team’s analysis and decision-making. Researchers may also suggest useful items required to improve public health.
The goal would be to develop a multi-axial summary for a patient’s situation. The level of detail must be proportionate to the complexity and the context.
For example, an initial diagnostic summary in an emergency department might be restricted to major pathologies and essential physiological variables. For a healthy adult attending a family doctor with a recent severe earache, the summary might be restricted to ‘acute bacterial otitis media’.
Nevertheless, a generic detailed preferred format must be developed, evaluated and improved over time. As a starting point, one might build on the psychiatric model, as discussed next.
A multi-axial formulation.
A formulation could be free text or structured. The items below are the main considerations in most circumstances. I have drafted a form that could be used.
The formulation would generally cover:
- Direct causal factors:
These would be unequivocal items that directly damaged the body’s structure and function. They would map reasonably onto much of the ICD, leaving out all the syndromic and ‘not otherwise specified parts. - Precipitating factors:
These would be items that lead directly to the malady. For example, if a woman who witnessed her son being killed presented with functional paralysis or acute withdrawal, the event would be the main ‘cause’. - Predisposing factors:
These are items that may explain why now, why this person? For example, a person who has had a stroke might already have diabetes and coronary artery disease and have recently developed palpitations. Someone presenting with acute depression after retiring might have had long-standing mild depression. - Perpetuating factors:
In someone who has had their malady for some time, these items may explain why progress is not as expected. For example, someone whose accommodation is unsatisfactory and who is in hospital after a stroke may not want to return to a lonely life in unsuitable accommodation when she has developed friends in a rehabilitation centre. Arthritic knee problems may slow or another person’s recovery of walking. - Protecting factors:
These explain why a person’s illness is not as bad as expected or their progress is better than expected. Most unexpectedly good outcomes are inexplicable, but factors such as unusually devoted and competent family members who become actively engaged will often improve outcomes.
These factors can be placed into various domains:
- Internal bodily systems:
malfunction within parts of the body; biochemical, cellular, and organ systems, usually referred to as pathology. This is the domain covered by the ICD. - Body system:
malfunction in the body as a whole, often referred to as impairment of body structure and function. This includes most symptoms, signs, and patient-reported experiences such as pain. - Person-physical context system:
Alteration or limitation of capabilities. This refers to behaviours and equates to disabilities. It is observable. It may also include physical factors influencing outcome. - Person-social context system:
Alteration or limitation in social participation, manifest as altered social roles. It refers to the meaning attributed to behaviours. This domain may also include a comment on social networks. - Personal factors.
These are often crucial, but not classifiable. I have written a post on personal factors.
Conclusion
The era of relying on a single diagnostic label in healthcare should end now. It was developed with biomedical healthcare, initially helping both clinical management and research. Over the past 50 years, our understanding of health and illness has grown, along with the rise of chronic diseases and multimorbidity. We now recognise the strong links between a person’s social and physical environments, their personal characteristics, and the nature and management of their illness. We know that healthcare issues and systems are complex and multifactorial, making it inappropriate to use just one word to describe a person’s situation. Health services should adopt the concept of formulation and be more precise in the data used to manage services, identify needs, and allocate resources. We can no longer pretend that a single diagnostic label allows “the identification of the nature of an illness.”