Every day, NHS ‘detectives’ hunt for missing or inconsistent codes that cost hospitals millions. Our Lead Pharmacist, Pete Hughes, explains why automation is the only fix.
Every day, across every NHS hospital, armies of “detectives” get to work. These clinical coding teams and specialist reimbursement pharmacists go through the records of patients that have recently been discharged, and review all case notes and clinical documentation.
These teams aren’t looking for criminals - they’re hunting data. Missing, inconsistent, or misapplied codes can cost hospitals millions.
By the time the detectives get into patients’ records, diagnoses, surgical procedures, allergies, medications, and interventions have already been documented. It’s a mish-mash of data - the same term/piece of information might appear as free text, a dropdown term, or one of several SNOMED codes.
Similarly, a prescription might have been recorded as a SNOMED code in the patient record. But it could be one of many possible codes for various forms, brands, storage, and intended administration routes. In fact, the exact same thing happening multiple times may have been recorded in several different ways in the same patient record.
The information is usually all technically valid, but it’s inconsistent. This variability degrades its usefulness and in some cases - safety. For example, I’ve seen a patient’s insulin regimen recorded with ‘NovoRapid,’ ‘fast-acting insulin,’ and a note saying ‘uses pens at home’. None lined up with the actual prescribed strength or dose. The system couldn’t match the records, so it didn’t flag anything when a prescriber tried to order the wrong insulin type and dose on admission.
In the end, it didn’t escalate into a crisis because a resident doctor realised the numbers looked off - not because the system did its job.
That one inconsistency shows how fragile data integrity can be when every record speaks a slightly different language. Data needs to be not only technically present, but structurally useful.
For many new, high-cost drugs, standardisation and correct coding is important for the purposes of reimbursement, continuity of care and monitoring outcomes. Hospitals need to report to NHS England for this mechanism to work properly, and they can’t do that without proper coding.
You might be starting to see where these examples are going, and the problem these detective teams are setting out to tackle - standardising and ‘correcting’ clinical information into code.
Yet even with all the technology we could have at our disposal, this laborious job takes a human eye - that’s what I want to look at in this article.
Although NHS treatment is free at the point of use, it of course has a cost somewhere.
For hospitals, many drug costs are reclaimed from NHS England through complex mechanisms that rely on coded clinical data. Some of the more expensive medicines - like Ustekinumab, which, for certain brands, is currently more than £2,000 per dose - are funded centrally and require perfectly codified reporting to ensure the money is reimbursed.
The detectives’ role is to find missing or incorrectly recorded care information and add or translate it into correct, standardised, machine-readable codes that can be matched to specific costs centres at NHS England. And the clock is ticking. The success of the matching process directly determines the hospital's financial reimbursement for care it has provided. Extrapolated across the millions of procedures and drugs each hospital will see each year, misplaced codes can cost millions of pounds and divert highly trained staff away from patient care.
This is why data integrity is such a huge part of operational efficiency, and one that doesn’t get a fraction of the attention it deserves. After all, if we have a solution to the problem (detectives), why look at it again? What if the solution is duplicative, labour intensive, prone to errors and delays, and pulls staff and resources away from patient care - like the process described above? How about when you consider that one study found that more than 50% of cases contained a coding error, requiring further involvement from clinicians to correct it? Then reassessing it surely becomes an imperative.
It’s obvious that cleaning and properly codifying clinical data is a crucial step in streamlining the operations behind prescribing safety and reimbursement. But doing it manually is madness. The answer to much of the problem lies in automation.
To do this, we need a Clinical Decision Support (CDS) system that integrates deeply into the workflow and solves this problem at the point of prescribing. Systems like Touchdose can not only recommend the right dosage of the right drug for each individual patient and their profile, but also automatically codifies this data and writes it back into the patient's record with the correct SNOMED and dm+d code.
Clean, consistent, safe, reimbursable data - instead of a fragmented mess.
This codification is entirely passive, and eliminates the need for detectives to trawl through every record and enter data manually in one fell swoop. It even eliminates the need for clinicians to enter prescribing decisions into the patient record themselves, ensuring the patient record has the necessary data integrity for downstream uses - from a reduction in errors like prescribing penicillin to an allergic patient, to swift financial reimbursement.
By supporting the data integrity of the patient record, these systems directly solve the financial reimbursement challenge. They can automatically assign the correct SNOMED codes for high-cost drugs, eliminating the manual detective work, accelerating hospital cash flow, and ensuring hospitals receive the money they are rightfully owed.
It makes other operational tasks easier too - like patient co-horting, which is important for risk stratification, identifying patients who may align with a trust campaign (for example, biosimilar switching), or spotting cohorts who need proactive clinical reviews before their condition deteriorates and requires avoidable treatment.
I wish I’d had this when I was a pharmacist - and one of the detectives - myself, as manually collating and codifying clinical information was a headache. There was also always the fear that missing something could lead to a financial loss for the department, or worse risk adverse consequences to patients.
Finally, automated coding and clean data future-proof the NHS for the next generation of patient care: the rise of personalised medicine and complex new treatments like genomic medicines.
These bring an increased risk of errors due to highly variable, complex dosing. A system like Touchdose, which provides programmatic dosing and relies on clean, codified data, can prevent these mistakes and support the safe, compliant and cost-effective delivery of these advanced therapies.
One of the central conflicts of data integrity in the NHS is that it is being secured by detective work, not design. Every day, highly trained clinical coding teams and specialist pharmacists are diverted from patient care to recover millions of pounds that could be lost due to inconsistent data - from identifying if a drug has been given, to what it's been given for. From free text allergy shorthand to missing codes for high cost drugs.
Clinical safety and financial stability both rely on the same thing: perfectly codified, machine-readable data.
To achieve this without wasting more time, effort, and money on things that could be done at the touch of a button - we must shift from detection of errors to prevention.
Next-generation CDS systems like Dosium's Touchdose embed data integrity directly into the clinical workflow. By performing passive codification and write-back at the moment of prescribing, they eliminate manual entry errors, simultaneously ensuring patient safety alerts function correctly and securing every vital pound in each reimbursement claim.
The imperative for every NHS organisation is clear: implement systems that codify previously fragmented clinical data, for the benefits of:
It’s time to retire the detective model and build integrity into the system itself.
If you’re tackling these challenges in your organisation, our team would love to explore how automation could help.
