Why Do Doctors Care About This?

Complexity does not push medicine outside science. It makes medicine probabilistic, contextual, and iterative. The physician draws on anatomy, physiology, pathology, evidence, guidelines, statistics, examination, the patient's own account, and ongoing monitoring to act under uncertainty. The aim is never to command the body with perfect precision. It is to influence a dynamic system in a helpful direction, watch how it responds, and adjust.

A few features of complex clinical systems explain why this is unavoidable. One presentation can spring from many mechanisms, as breathlessness can come from obstructed airways, poor gas exchange, anaemia, heart failure, deconditioning, anxiety, a clot, an infection, an acid disturbance, weak breathing muscles, or a drug. One mechanism can produce many effects, as failing kidneys disturb fluid balance, blood pressure, potassium, acid handling, red cell production, bone metabolism, drug clearance, and cardiovascular risk all at once. And treating one mechanism can move another, as diuresis eases congestion while shifting kidney perfusion and electrolytes, or steroids calm inflammation while raising glucose and infection risk. These are not edge cases. They are the ordinary structure of clinical problems.

The molecular picture underneath supports the same view, since the effect of a genetic or molecular abnormality depends on its place in the wider network of interacting components, and most disease reflects interacting processes rather than one isolated single-gene fault. The bedside picture is multimorbidity, where a person's conditions and treatments interact, where those interactions shape quality of life, and where treatment burden, adverse events, unplanned care, and poor coordination all bear on outcomes. Good practice here begins by establishing the burden of disease, the burden of treatment, and the patient's own goals, values, and priorities, then builds an individualised plan from those rather than from any single guideline.

This is why doctors so often manage in cycles rather than single decisions. They start or change a treatment, monitor the symptoms and the measurements, check for adverse effects, and adapt. Type 2 diabetes shows the pattern well, since its physiology spans insulin resistance, the pancreas compensating and later faltering, fat tissue effects, inflammation, incretin biology, the kidney's handling of glucose, and vascular risk, while its management may run across nutrition, activity, weight, several drug classes, blood pressure and lipid control, smoking cessation, and the monitoring of kidneys, eyes, and feet. The doctor is asking not merely whether a treatment "works" in general, but whether it suits this patient's mechanism, risk profile, other conditions, other medicines, capacity, values, and likely trajectory.

Several distinctions organise this kind of reasoning. Evidence-based medicine is not algorithmic medicine, since evidence has to be interpreted for a patient and a guideline supplies structured knowledge without removing judgement. Efficacy is not effectiveness, the first asking whether something can work under study conditions, the second whether it works in the real world of adherence, comorbidity, access, cost, and competing priorities. A treatment target is not a patient outcome, because a glucose or cholesterol number may matter while the person also cares about symptoms, function, independence, cognition, side effects, cost, and burden. Risk reduction is not symptom relief, since some treatments are felt quickly and others are taken purely to lower the future probability of an event, and in multimorbidity those future-facing treatments can themselves become a heavy burden. And protocolised urgency is not the opposite of individualised adjustment, since even a condition demanding rapid structured care, like sepsis, still builds in reassessment, de-escalation of antibiotics, source control, and follow-up.

Complexity is the reason behind so many of the contextual questions doctors ask. A symptom cannot be read without its time course, triggers, severity, the patient's medications, comorbidities, age, exposures, prior illnesses, functional baseline, social support, and goals of care, because the same symptom in two people can imply entirely different physiology. It is also why monitoring sits at the centre of care rather than at its margins. The blood pressure drug that needs repeat pressure and kidney checks, the diabetes drug that needs glucose, weight, kidney, and tolerability monitoring, the diuretic that needs weight, pressure, kidney function, and a panel of salts tracked, all of these are gathering feedback from a living system, not completing paperwork.

The same logic is why shared decision-making belongs to real medicine rather than sitting beside it as a courtesy. Care in multimorbidity has to centre the person's needs, priorities, lifestyle, and goals, weighing the benefits and risks of single-condition recommendations against quality of life, treatment burden, adverse events, and the strain of fragmented care. This is not a soft addition to the clinical work. Adherence, feasibility, burden, goals, and values all feed back into physiological outcomes, which means a patient's preferences are not separate from their physiology at all.

Finally, complexity explains a move that can look counterintuitive: doctors sometimes reduce treatment rather than add to it. Deprescribing, watchful waiting, rehabilitation, lifestyle support, reconciling a medication list, and honest goals-of-care conversations are active clinical decisions in their own right. In some complex patients, lowering the burden improves safety, function, and quality of life more than any further intervention would.

A few cautions to keep the framing honest. Doctors influence, monitor, and adapt; they do not steer a patient's system with certainty, since outcomes stay probabilistic. Guidelines do not uniformly ignore complexity, given that some explicitly address it. Personalised medicine does not always mean genomics, since personalisation also takes in comorbidity, kidney function, frailty, prior response, social context, cost, and feasibility. And more treatment is not reliably better, because in a complex patient it can mean more benefit, more harm, more monitoring burden, or some uneasy mixture of the three.

1. Core thesis

Doctors care about complexity because clinical care is not performed on isolated mechanisms under laboratory conditions. It is performed on adaptive human beings whose bodies contain interacting diseases, treatments, behaviours, environments, histories, preferences, and physiological reserves. A doctor may know the mechanism of a drug, the expected effect of a treatment, and the evidence from clinical trials, but the final clinical question is still: what is likely to happen in this person, in this state, with these risks, constraints, and goals?

Complexity does not make medicine unscientific. It makes medicine probabilistic, contextual, and iterative. The physician uses anatomy, physiology, pathology, evidence, guidelines, statistics, clinical examination, patient narrative, and monitoring to make decisions under uncertainty. The aim is not to command the body with perfect precision. The aim is to influence a dynamic system in a beneficial direction, observe the response, and adjust the plan.

2. Scientific synthesis

A complex clinical system has several features. First, multiple mechanisms may contribute to one presentation. Breathlessness may involve airway obstruction, impaired gas exchange, anaemia, heart failure, deconditioning, anxiety physiology, pulmonary embolism, infection, acidosis, neuromuscular weakness, or medication effects. Second, one mechanism may produce multiple effects. Kidney dysfunction may alter fluid balance, blood pressure, potassium, acidbase status, anaemia, bone-mineral metabolism, medication clearance, and cardiovascular risk. Third, treatment of one mechanism may alter another. Diuresis can reduce congestion while changing renal perfusion and electrolytes. Steroids can reduce inflammation while increasing glucose and infection risk. Sedatives can reduce agitation while impairing airway protection and respiratory drive.

Network medicine provides one molecular foundation for this view. Barabási, Gulbahce, and Loscalzo argue that the effect of a molecular abnormality depends on its network context, because cellular components exert functions through interactions with other components. They describe disease phenotypes as reflecting interacting pathobiological processes in complex networks, rather than isolated single-gene abnormalities in most cases.

Multimorbidity provides the bedside foundation. NICE states that people with multimorbidity may require an approach that accounts for how their conditions and treatments interact, how these interactions affect quality of life, and how treatment burden, adverse events, unplanned care, and coordination problems shape outcomes. It recommends establishing disease burden, treatment burden, patient goals, values, priorities, and individualised management plans.

This is why doctors often manage patients through cycles rather than single decisions. A clinician starts or changes treatment, monitors symptoms and measurements, checks for adverse effects, and adapts. In type 2 diabetes, for example, the underlying physiology includes insulin resistance, beta-cell compensation and dysfunction, adipose tissue effects, inflammation, incretin biology, renal glucose reabsorption, and vascular risk. Treatment may involve nutrition, physical activity, weight management, metformin, GLP-1 receptor agonists, SGLT2 inhibitors, insulin, blood pressure control, lipid management, smoking cessation, kidney monitoring, eye screening, and foot care.

The doctor therefore cares not only whether a treatment works in a general sense. They care whether it is suitable for this patients mechanism, risk profile, comorbidities, other medicines, capacity, values, and expected trajectory.

3. Key distinctions

The first distinction is evidence-based medicine vs algorithmic medicine. Evidence matters, but evidence must be interpreted for a patient. A guideline gives structured knowledge; it does not remove clinical judgement.

The second distinction is efficacy vs effectiveness. Efficacy asks whether an intervention can work under study conditions. Effectiveness asks whether it works in real-world conditions, with adherence, comorbidity, access, cost, behaviour, monitoring, and competing priorities.

The third distinction is treatment target vs patient outcome. A glucose number, blood pressure, oxygen saturation, creatinine, or cholesterol value may matter, but the patient also cares about symptoms, function, survival, independence, cognition, side effects, cost, burden, and quality of life.

The fourth distinction is risk reduction vs symptom relief. Some treatments are felt quickly. Others are taken to reduce future probability of events. In multimorbidity, treatments aimed at future risk can become burdensome, and NICE specifically advises careful consideration of risk-factor management as a treatment burden.

The fifth distinction is protocolised urgency vs individualised adjustment. Sepsis requires rapid structured response, but even sepsis guidelines include reassessment, lactate interpretation in context, de-escalation of antimicrobials, source control, vasopressor targeting, and post-discharge follow-up.

4. Clinical relevance

Complexity explains why doctors ask so many contextual questions. A symptom cannot be interpreted without time course, triggers, severity, medications, comorbidities, age, pregnancy status, exposures, prior illnesses, functional baseline, social support, and goals of care. The same symptom in two people can imply very different physiology.

Complexity also explains why monitoring is central. A blood pressure medicine may require repeat blood pressure checks and renal function testing. A diabetes medicine may require monitoring of glucose, HbA1c, weight, kidney function, hypoglycaemia risk, cardiovascular status, and tolerability. A diuretic may require monitoring of weight, symptoms, blood pressure, kidney function, sodium, potassium, and magnesium. Monitoring is not merely administrative; it is feedback from the system.

Complexity also explains why shared decision-making matters. NICE recommends that care for people with multimorbidity should focus on the persons individual needs, health priorities, lifestyle, goals, benefits and risks of single-condition recommendations, quality of life, treatment burden, adverse events, unplanned care, and coordination. This is not a soft add-on to real medicine. It is part of treating a complex adaptive human system, because adherence, feasibility, burden, goals, and values influence outcomes.

Finally, complexity explains why doctors sometimes reduce treatment rather than add more. Deprescribing, watchful waiting, rehabilitation, lifestyle support, medication reconciliation, and goals-of-care discussions can be active medical decisions. In some complex patients, reducing burden can improve safety, function, and quality of life.

5. Examples worth keeping

Multimorbidity management: strongest example of complexity-informed care. Keep NICEs focus on interactions, treatment burden, individual goals, adverse events, and care coordination.

Type 2 diabetes: useful because management involves metabolic physiology, vascular risk, behaviour, medications, kidney function, weight, and long-term monitoring.

Sepsis: useful because complexity does not mean slow or vague care. It can require urgent structured treatment plus continuous reassessment.

Frailty: useful because reserve changes response to treatment. Frailty is associated with decreased physiological reserve and increased vulnerability to stressors.

Medication review: useful because it shows that treatment itself becomes part of the system.

6. Claims to revise, qualify, or avoid

Avoid saying doctors can steer a patients system with certainty. They can influence, monitor, and adapt, but outcomes remain probabilistic.

Avoid saying guidelines ignore complexity. Some guidelines, such as NICE multimorbidity guidance, explicitly address complexity.

Avoid saying patient preferences are separate from physiology. Behaviour, adherence, stress, sleep, diet, activity, access, and treatment burden influence physiological outcomes.

Avoid saying personalised medicine always means genomics. Personalisation also includes comorbidity, kidney function, frailty, goals, prior response, social context, cost, and feasibility.

Avoid saying more treatment is better. In complex patients, more treatment can mean more benefit, more harm, more monitoring burden, or some combination.