Thresholds are clinical decision points
In daily practice, a threshold is the point at which the pre-test probability of disease, or the potential benefit of an intervention, becomes sufficient to trigger action. It’s the moment you decide to refer, prescribe, scan, or admit. These are not abstract concepts; they are the operational definitions of clinical guidelines.
For example, the NICE CG95 chest pain guideline sets a pre-test probability threshold of 10% for investigating stable chest pain. Below that, no further investigation is recommended. This is a clear, numerical threshold. The negligence claim does not arise from the guideline itself, but from the application—or misapplication—of that specific percentage to an individual patient.
Why thresholds are the epicentre of negligence
Negligence hinges on breach of duty. Breach is rarely about a complete failure to act. It is almost always about acting, or failing to act, at the wrong threshold. The court’s question is: did this doctor’s decision-making fall outside the range of what a responsible body of medical opinion would deem acceptable? That range is defined by thresholds.
A classic example is the threshold for referral for suspected cancer. A GP sees a 55-year-old with three weeks of dyspepsia. The risk of cancer is low, but not zero. The threshold for a 2-week-wait referral might be a risk score of 3% or above. If the GP estimates the risk at 2% and does not refer, and the patient is later diagnosed with oesophageal cancer, the claim will focus entirely on whether that 2% estimation was reasonable. The dispute is over a single percentage point.
The evidential weight of a documented threshold
Your defence in such a scenario rests on evidence that your threshold was sound. If your clinical system merely records "dyspepsia? cause?" the court has little to work with. But if your note includes "NICE DG12 risk score calculated: 1.5% (below 3% referral threshold). Advice re PPI and safety netting given," you have created a robust, contemporaneous record of your threshold-based reasoning.
This moves the argument from "you should have referred" to "was your application of the recognised threshold reasonable?" The latter is a much stronger position to defend. The absence of a documented threshold calculation is often the single greatest weakness in a clinical negligence defence.
Thresholds in practice: referral and prescribing
Thresholds are dynamic and context-dependent. What is safe in one setting may be negligent in another.
Referral threshold escalation
Consider a patient with lower back pain. The initial threshold for specialist referral is high: red flags must be present. A junior doctor in the Emergency Department discharges a patient with mechanical back pain and no red flags. This is standard practice.
Now, that same patient re-presents to their GP one week later. The pain is unchanged, but there is now a single, vague reference to possible bilateral leg tingling in the patient's history. The threshold for referral has now lowered. The patient has re-presented, and a new, albeit subtle, symptom is present. The GP who applies the same high threshold as the ED and again discharges the patient is now on much thinner ice. The clinical context has changed, and the acceptable threshold for action has shifted accordingly. The negligence claim alleges the GP failed to recognise this shift.
Prescribing threshold drift
Prescribing decisions are a minefield of implicit thresholds. The threshold for initiating a high-risk medication like anticoagulation in atrial fibrillation is well-defined by tools like CHA₂DS₂-VASc. However, the threshold for *reviewing* or *stopping* that prescription is often dangerously vague.
A patient on apixaban for AF has a minor GI bleed. It is managed conservatively in the ED. The discharging doctor’s note says "continue apixaban as per cardiology." Six months later, the patient has a major GI bleed requiring massive transfusion.
The claim will argue that the minor bleed was a significant event that should have triggered an immediate re-evaluation of the risk-benefit threshold for anticoagulation. The defence of "continuing as per specialty" is weak. The threshold for harm had been met; the duty was on the discharging doctor to recognise that the original indication threshold was no longer valid without a fresh assessment. This is threshold drift in action—failing to adjust your decision point when the clinical picture evolves.
The impact of investigations on thresholds
Diagnostic tests exist to change probability and thus cross thresholds. Misinterpreting their impact is a common source of error.
A patient has a 40% pre-test probability of PE. A D-dimer comes back elevated. In this high-risk scenario, the D-dimer is meaningless—the threshold for further imaging (CTPA) was already crossed by the clinical assessment alone. The doctor who sees the elevated D-dimer and *then* decides to scan has made a cognitive error, but the patient gets the correct test.
The negligence occurs when the reverse happens. A patient has a 5% pre-test probability of PE. The doctor, perhaps erring on the side of caution, orders a D-dimer. It is elevated. They now feel compelled to order a CTPA, which is negative. The patient suffers contrast-induced nephropathy.
The claim here is for the harm caused by an unnecessary test. The root cause was a failure to understand that the D-dimer should not be used in a low-probability scenario because a positive result is almost certainly false. The doctor allowed a test result to inappropriately lower the action threshold, leading to harm. The breach was not in performing the CTPA, but in the flawed reasoning that led to it.
The silent threshold: when not to test
The most defensible threshold is often the one for inaction. Unnecessary investigation is a form of harm. A robust clinical note that states "Probability of pathology < 1%, below the testing threshold. Risks of false positive and over-investigation outweigh minimal benefits" demonstrates advanced, safe clinical reasoning. It shows an understanding that thresholds protect patients from the system as well as from disease.
Operationalising thresholds in your practice
Awareness is not enough. Thresholds must be integrated into your workflow to be defensible. This means moving from implicit, gut-feeling decisions to explicit, recorded ones.
The most effective method is to use validated risk calculators at the point of care. For a suspected PE, actually calculate the Wells Score. For AF anticoagulation, calculate the CHA₂DS₂-VASc and HAS-BLED scores simultaneously. Document the scores and the resultant decision. This creates an auditable trail.
For conditions without a numerical score, define your own threshold explicitly in the notes. "Decision: Not referring to neurology at this time. Threshold for referral would be objective weakness or progressive symptoms. Currently absent." This states your clinical rule and confirms the patient does not meet it.
Leveraging a dedicated threshold look-up tool within your clinical system can standardise this process. Instead of relying on memory, you can quickly access the specific percentage or criteria for action, ensuring consistency and providing immediate evidence of your adherence to accepted practice.
Conclusion: thresholds as a safety net for clinicians
Clinical negligence is not about dramatic errors in technique. It is about subtle errors in judgment at critical decision points. Those points are thresholds. Your defence rests on your ability to demonstrate that your judgment at those points was within the bounds of reasonable practice.
The only way to prove this is through contemporaneous documentation that captures your threshold-based reasoning. Making your thresholds explicit—through calculated scores, reference to guideline percentages, or clear clinical criteria—transforms your notes from a simple record of events into a powerful shield against allegations of negligence.
For a comprehensive repository of established clinical thresholds across specialties, the clinical thresholds index serves as a key clinical governance resource.