Where Do Things Go Wrong?

Across the preceding packets, a single theme has been gathering: the body's responses are usually helpful and occasionally treacherous, and the difference is often a matter of context, persistence, or timing. This packet names the specific patterns by which a complex system goes wrong, because complexity is only useful if it points to something concrete. Used loosely, it explains everything and therefore nothing. Used precisely, it identifies recognisable failure modes: maladaptive compensation, harmful feedback, the crossing of a threshold, propagation through a network, the weight of treatment, the exhaustion of reserve, and the late recognition of deterioration that was building all along.

The first pattern is maladaptive compensation, where a response that rescues the body acutely becomes harmful when it persists. Type 2 diabetes is the clearest case. When tissues resist insulin, the pancreas initially compensates by secreting more, holding glucose in range for a time. But the compensation has a shelf life, and as the insulin-producing cells falter, the glucose climbs and stays high, with fat tissue, inflammation, incretin biology, glucagon, and the kidney's glucose handling all contributing to the broader state. The body bought time, and then ran out of it.

The second pattern is systemic dysregulation, where the harmful state belongs to the whole system rather than any organ. Sepsis is the paradigm, defined as life-threatening organ dysfunction caused by the body's dysregulated response to an infection, with septic shock its more dangerous subset. This is worth stating precisely, because it is not simply "runaway inflammation" and it is not single-organ failure. It is a whole-system state pulling in infection, immunity, vascular tone, the vessel lining, perfusion, metabolism, clotting, and multiple organs at once.

The third pattern is reserve exhaustion, which is the story of frailty. A frail patient carries reduced physiological reserve across several systems and a heightened vulnerability to stress, so that a urinary infection, a sedating tablet, a bout of dehydration, poor sleep, constipation, or simply a hospital admission can precipitate delirium, a fall, functional decline, or kidney injury. The insult is small; the consequence is large, because there was so little spare capacity to absorb it.

The fourth pattern is treatment-system overload, where the care plan itself becomes destabilising. The work of managing future risk can mount into a substantial burden in multimorbidity, and single-condition guidance is typically drawn from people who lacked multiple conditions and long medication lists. When treatment interactions, patient preferences, quality of life, adverse events, and coordination are left out of the reckoning, the accumulated plan can do physiological and practical harm of its own. The fifth pattern is network propagation, where a disturbance in one molecular component spreads through its connections, which is part of why complex diseases resist single-target explanations and why a drug's effect can depend on the network context it lands in.

Several distinctions turn these patterns into usable judgement. Adaptation and maladaptation are the same responses in different contexts, since fluid retention, inflammation, a faster heart rate, insulin secretion, and vasoconstriction can each help in one situation and harm in another. Stability and fragility can look identical from outside, because a patient may appear stable precisely because compensation is working hard, and that stability is fragile if the reserve behind it is nearly spent. Gradual change differs from threshold change, since some deterioration creeps while other deterioration arrives suddenly as a system tips over into septic shock, decompensated heart failure, delirium, or respiratory failure. A side effect differs from a cascade, the first being one treatment's direct consequence, the second being what happens when that consequence triggers further interventions and complications. And undertreatment and overtreatment are both real harms, since a complex patient can be hurt by failing to treat a dangerous mechanism and equally by treating too many targets with no regard for burden, reserve, and interaction.

For clinicians, recognising these patterns changes management. In sepsis, treatment must begin urgently and yet be reassessed continually, addressing the infection, the perfusion, the blood pressure, the metabolic stress, the organ dysfunction, the antibiotic exposure, and the anatomical source together rather than leaning on any single intervention. In chronic care, complexity-related failure tends to surface as treatment burden, non-adherence, adverse drug events, falls, kidney injury, low blood sugar, dizziness, confusion, or declining function, which is why it helps to flag the patients most likely to benefit from a multimorbidity-focused approach: those struggling with their treatments or daily activities, those under several services, those with frailty or falls, those repeatedly seeking unplanned care, those on many regular medicines. In frailty, small changes deserve to be taken seriously, since a minor stressor can exceed reserve, though this is emphatically not an argument against active treatment but for treatment that weighs baseline function, goals, proportionality, delirium risk, nutrition, mobility, and support. And in diabetes and the other chronic metabolic diseases, complexity is why a single biomarker is never the whole disease, with vascular risk, kidney function, weight, pressure, lipids, hypoglycaemia risk, access, nutrition, activity, mental health, and sleep all part of the picture alongside the glucose.

A few cautions to keep this honest. Small causes do not always produce huge effects; they do so near a threshold, while many small inputs are simply buffered away. Illness is not unpredictable, since a great deal of risk is predictable in probabilistic terms; what complex physiology limits is exact prediction. Treatment cascades are not always mistakes, given that some sequential interventions are entirely appropriate; the problem is unrecognised cascade harm, avoidable escalation, and the failure to reassess. And complexity does not explain depression, obesity, pain, or fatigue on its own, since it earns its keep only when tied to identifiable interacting contributors rather than invoked as a label.

C1.1.6-clinical-2 — Where Do Things Go Wrong?

1. Core thesis

In complex adaptive physiology, things often go wrong when adaptive mechanisms become maladaptive, when feedback loops amplify instability, when reserve is exhausted, when several modest problems combine, or when treatment of one problem destabilises another. Disease is not always the failure of one isolated part. It may be the result of an interacting system entering a harmful state.

This packet should show that complexity has clinical consequences. It explains why acute illness can deteriorate rapidly, why chronic disease can progress despite apparently sensible treatment, why frail patients may collapse after small insults, why polypharmacy can generate unexpected harm, why single-disease guidelines can conflict, and why patients with similar diagnoses may follow different trajectories.

The scientific aim is to make complexity usable. Complexity should not be used as a vague explanation for everything. It should help identify specific failure patterns: maladaptive compensation, harmful feedback, threshold crossing, network propagation, treatment burden, loss of reserve, and delayed recognition of system-level deterioration.

2. Scientific synthesis

One major failure pattern is maladaptive compensation. A physiological response that is useful acutely can become harmful when persistent. In type 2 diabetes, insulin resistance is initially countered by increased insulin production to maintain glucose homeostasis, but over time beta-cell dysfunction can reduce the ability to compensate, producing persistent hyperglycaemia. The same source notes that adipose tissue, inflammatory mechanisms, adipokine dysregulation, incretin biology, glucagon, renal glucose reabsorption, and gut microbiota may contribute to the broader pathophysiology.

A second failure pattern is systemic dysregulation. Sepsis is defined by Sepsis-3 as life-threatening organ dysfunction caused by a dysregulated host response to infection. Septic shock is a subset involving circulatory, cellular, and metabolic abnormalities associated with increased mortality risk. This is not a single-organ failure. It is a whole-system state involving infection, immunity, vascular tone, endothelial function, perfusion, metabolism, coagulation, organs, and treatment response.

A third failure pattern is reserve exhaustion. Frailty reflects decreased physiological reserve and increased vulnerability to stressors. It is associated with adverse outcomes including mortality, nursing home admission, falls, and delirium, and it reflects multisystem dysfunction rather than one disease alone. In a frail patient, a urinary infection, sedating medication, dehydration, pain, sleep disruption, constipation, or hospitalisation can produce delirium, falls, functional decline, kidney injury, or institutionalisation.

A fourth failure pattern is treatment-system overload. NICE notes that risk-factor management can become a major treatment burden in multimorbidity and that single-condition guidance is often based on people without multimorbidity and fewer medicines. It recommends considering treatment interactions, patient preferences, benefits and risks, quality of life, adverse events, unplanned care, and coordination. When these factors are ignored, the care plan itself can become physiologically and practically destabilising.

A fifth failure pattern is network propagation. In network medicine, a perturbation in one molecular component may affect other components through network links, and disease phenotypes can reflect interacting pathobiological processes. This helps explain why complex diseases often resist single-target explanations and why drug effects may depend on network context.

3. Key distinctions

The first distinction is adaptation vs maladaptation. Fluid retention, inflammation, tachycardia, insulin secretion, vasoconstriction, and stress responses can be useful in one context and harmful in another.

The second distinction is stability vs fragility. A patient can appear stable because compensatory systems are working. That stability may be fragile if reserve is low or if compensation is close to its limit.

The third distinction is threshold vs gradual change. Some deterioration is gradual. Other deterioration occurs when a system crosses a threshold: septic shock, decompensated heart failure, hyperosmolar crisis, delirium, falls, acute kidney injury, or respiratory failure.

The fourth distinction is side effect vs cascade. A side effect may be a direct adverse consequence of one treatment. A cascade occurs when an adverse effect triggers additional interventions, complications, or functional decline.

The fifth distinction is undertreatment vs overtreatment. In complex patients, harm can come from failing to treat a dangerous mechanism, but also from treating too many targets without regard to burden, reserve, and interactions.

4. Clinical relevance

Doctors need to recognise complexity-related failure because it changes management. In sepsis, treatment must begin urgently, but the patient’s response must be reassessed repeatedly. The Surviving Sepsis Campaign recommends immediate treatment and resuscitation for sepsis or septic shock, lactate measurement in suspected sepsis, fluid resuscitation for sepsis-induced hypoperfusion or septic shock, vasopressors such as norepinephrine when needed, daily assessment for antimicrobial de-escalation, and rapid identification or exclusion of an anatomical diagnosis requiring source control.

This is complexity-informed acute care. It does not rely on one intervention. It addresses infection, perfusion, blood pressure, metabolism, organ dysfunction, antimicrobial exposure, and anatomical source.

In chronic care, complexity-related failure often appears as treatment burden, non-adherence, adverse drug events, falls, kidney injury, hypoglycaemia, dizziness, confusion, or worsening function. NICE recommends identifying people who may benefit from a multimorbidity-focused approach when they have difficulty managing treatments or daily activities, receive care from multiple services, have frailty or falls, seek unplanned care, or are prescribed multiple regular medicines.

In frailty, clinicians must treat small changes seriously. A minor stressor can exceed reserve. This does not mean frail patients should not receive active treatment. It means treatment should account for baseline function, goals, proportionality, adverse effects, rehabilitation potential, delirium risk, nutrition, mobility, medications, and support.

In diabetes and other chronic metabolic diseases, complexity explains why a single biomarker is not the whole disease. Glucose matters, but vascular risk, kidney disease, weight, blood pressure, lipids, hypoglycaemia risk, medication access, nutrition, activity, mental health, sleep, and complications also matter.

5. Examples worth keeping

Sepsis: Keep as the acute whole-system failure example. It demonstrates dysregulated host response, organ dysfunction, perfusion failure, metabolic disturbance, and the need for multi-component care.

Frailty: Keep as the reserve exhaustion example. It shows why small insults can have large consequences in low-reserve systems.

Type 2 diabetes: Keep as the maladaptive-compensation example. It shows how compensation can maintain function for a time before progressive dysfunction appears.

Multimorbidity and polypharmacy: Keep as the treatment-system complexity example. It shows how evidence-based treatments can accumulate into burden and risk when not integrated.

Heart failure decompensation: Useful, but it will be better developed later in organ-specific and management sections.

6. Claims to revise, qualify, or avoid

Avoid saying small causes always produce huge effects in complex systems. Small inputs may produce large effects near thresholds, but many small inputs are buffered.

Avoid saying illness is unpredictable. Many risks are predictable in probabilistic terms. The correct claim is that complex physiology often limits exact prediction.

Avoid saying treatment cascades are always mistakes. Some sequential interventions are appropriate. The problem is unrecognised cascade harm, avoidable escalation, or failure to reassess.

Avoid saying frail patients are “house of cards” in the Synthetic Draft. The scientific language is reduced physiological reserve, vulnerability to stressors, and increased risk of adverse outcomes.

Avoid saying sepsis is simply “runaway inflammation.” Sepsis-3 deliberately defines it as organ dysfunction caused by a dysregulated host response to infection, not merely excess inflammation.

Avoid saying “complexity” explains depression, obesity, pain, or fatigue without specifying mechanisms. Complexity must be connected to identifiable interacting contributors.