Thresholds are not static
Clinical thresholds are treated as fixed policy. They are not. They drift. A guideline might be reviewed every five years, but the practical thresholds we use change far more frequently. This drift is a major, under-appreciated patient safety risk.
See how this translates to practice: Explore our Clinical governance features, visit the Patient Safety Hub, or review Clinical Safety & Assurance for enterprise rollout.
I have seen this in every department. A consultant mentions a new study in a morning meeting. A pharmacist flags an updated drug bulletin. An email circulates from a specialty network. Each event subtly shifts the local standard of care. The written policy remains untouched, but the operational threshold has already moved.
The referral threshold creep
Referral thresholds are particularly fluid. A classic example is the HbA1c for diabetes referral. The guideline might state "consider referral if HbA1c > 86 mmol/mol despite dual therapy."
Last year, the endocrine team was under immense pressure. Informally, the message became: "We're only taking referrals for Hba1c > 90 mmol/mol unless there are complications." This was never a written directive. It was communicated through corridor conversations and reply-all emails. For six months, that was the de facto threshold.
Then, a new consultant joined, concerned about high-risk patients slipping through the net. The informal word changed: "Please refer anyone > 80 mmol/mol with poor adherence." The policy document still said 86. This creates a postcode lottery within the same trust, dependent on which consultant you spoke to last.
Prescribing thresholds and antimicrobial stewardship
Antimicrobial prescribing is a minefield of shifting thresholds. The trust's antibiotic policy is a hefty PDF, but its real-world application is dynamic.
Take the threshold for switching from oral to IV antibiotics in a community-acquired pneumonia. The policy might say "if CRP > 100 mg/L." But during a winter bed crisis, the microbiology team, seeing rising resistance patterns, pushes for a lower threshold. An email goes out: "We advise IV therapy for CRP > 80 mg/L in high-risk patients." Suddenly, the threshold has dropped by 20%.
Conversely, during a shortage of piperacillin-tazobactam, the threshold for its use skyrockets. What was once first-line for a severe UTI now becomes a drug of last resort. This happens overnight. The policy document is obsolete the moment the drug shortage is announced. Clinicians are left relying on memory and word-of-mouth, a clear safety vulnerability.
Monitoring thresholds in chronic disease
In chronic disease management, monitoring thresholds drift with new evidence and local audit results. The threshold for intensifying antihypertensive therapy is a prime example.
The NICE guideline provides a framework, but local implementation varies. After a trust-wide audit showed poor control in patients with CKD, the renal team advocated for a tighter grip. The informal teaching became: "Aim for <130/80 in CKD patients, not the standard <140/90." This more aggressive target was adopted across the medical wards before any formal policy update.
Similarly, the INR target for a mechanical mitral valve is typically 3.0-4.0. However, after a cluster of bleeds, the anticoagulation clinic might temporarily tighten the range to 2.5-3.5 for elderly patients. This reactive adjustment is a sensible local safety measure, but it exists outside the formal protocol, creating ambiguity and potential for error at handover.
The safety risks of undocumented drift
This constant, undocumented drift poses several concrete risks. First, it creates dangerous inconsistency. A registrar on a night shift, following the written policy, might make a decision that contradicts the new, unwritten daytime standard. This leads to conflict and patient harm.
Second, it undermines training. Junior doctors learn by following guidelines. If the real-world thresholds are different, they are set up to fail. I have had to correct foundation doctors for "over-referring" based on the published policy, because the actual threshold had shifted. This is unfair and confusing for them.
Third, it makes audit meaningless. Auditing practice against a static policy that no longer reflects real-world care generates useless data. It shows "non-compliance" where there is actually adaptive, sensible clinical practice. This wastes governance resources and misses the real opportunities for improvement.
Numerical examples of drift
The magnitude of drift can be significant. Here are real examples I have encountered:
- Neutropenic Sepsis: Policy: admit if Neutrophils < 1.0. De facto threshold during winter pressures: < 0.8.
- Blood Transfusion: Policy: transfuse if Hb < 80 g/L. De facto threshold post-TRISS trial dissemination: < 70 g/L in stable patients.
- Head Injury CT: Policy: GCS <15 at 2 hours. De facto threshold in a busy ED: GCS <14 or any focal neurology.
- GFR & Metformin: Policy: stop if eGFR < 30. De facto threshold after a drug safety alert: review risk/benefit if eGFR dips < 40.
These are not minor adjustments. They represent a 20-30% change in the numerical trigger for action. Operating with the wrong number has direct consequences for patient outcomes.
Why policies lag behind practice
Formal policy updates are slow for good reason. They require committee approval, legal review, and widespread communication. This process can take 12-18 months. Clinical evidence and operational pressures change weekly.
The result is a gap. The written policy is the official, defensible position. The de facto threshold is the practical, often safer, current standard. We live in this gap. It is where most clinical decisions are actually made.
This is not about clinicians being rogue. It is about the system being unable to keep its official documentation agile enough to match the pace of clinical practice. The problem is systemic.
The burden on individual clinicians
The responsibility for knowing the current threshold falls on individual clinicians. We are expected to be aware of the latest journal club discussion, the new consultant's preference, the email from a specialty lead. This is an unreasonable memory burden.
I have kept personal lists of "current thresholds" in my phone's notes app. This is a common but flawed workaround. My list might differ from my colleague's. It is a fragmented, unreliable system that increases cognitive load and risk.
A potential solution: dynamic threshold management
The solution is not to try and update the main policy more often. That is impractical. The solution is to decouple the dynamic thresholds from the static policy document.
We need a single, easily accessible source of truth for the current numerical thresholds. This should be a live resource, updated rapidly by clinical leads in response to new evidence or operational needs. It must be trusted and authoritative.
This is where a dedicated threshold look-up function becomes critical. Such a tool allows for the rapid dissemination and standardisation of threshold changes across the entire clinical workforce. A junior doctor on a night shift can check the current transfusion threshold instantly, rather than relying on hearsay.
The goal is to make the de facto standard the documented standard. This closes the safety gap between policy and practice. It ensures that everyone is working from the same numbers, reducing variation and protecting patients.
For this to work, the resource must be comprehensive and well-maintained. It requires clinical ownership and governance oversight to ensure changes are evidence-based and properly communicated.
Conclusion
Threshold drift is a reality of hospital medicine. It is a sign of a responsive, evidence-aware clinical culture, but it becomes a safety hazard when it remains informal and undocumented. The misalignment between policy and practice is a systems issue, not an individual failing.
Addressing it requires acknowledging that thresholds are dynamic. We must create systems that can keep pace with clinical practice. Centralising this information in a searchable, up-to-date format is a pragmatic step towards safer, more consistent care. The focus should be on creating a definitive threshold index that clinicians can trust implicitly, ending the reliance on fragmented, informal communication.